Loss of C. elegans GON-1, an ADAMTS9 Homolog, Decreases Secretion Resulting in Altered Lifespan and Dauer Formation

ADAMTS9 is a metalloprotease that cleaves components of the extracellular matrix and is also implicated in transport from the endoplasmic reticulum (ER) to the Golgi. It has been reported that an ADAMTS9 gene variant is associated with type 2 diabetes. The underlying pathology of type 2 diabetes is insulin resistance and beta-cell dysfunction. However, the molecular mechanisms underlying ADAMTS9 function in beta cells and peripheral tissues are unknown. We show that loss of C. elegans GON-1, an ADAMTS9 homolog, alters lifespan and dauer formation. GON-1 loss impairs secretion of proteins such as insulin orthologs and TGF-beta, and additionally impacts insulin/IGF-1 signaling in peripheral tissues. The function of the GON domain, but not the protease domain, is essential for normal lifespan and dauer formation in these scenarios. We conclude that the GON domain is critical for ADAMTS9/GON-1 function across species, which should help the understanding of type 2 diabetes in humans.


Introduction
Type 2 diabetes is a multifactorial disease characterized by impaired insulin secretion and insulin resistance. The risks for and progression of type 2 diabetes are determined by a combination of genetic and environmental factors. Recently, more than 60 common type 2 diabetes risk variants were identified through genome-wide association studies (GWAS) [1]. It was reported that an ADAMTS9 gene variant found in the 5'-upstream region (rs4607103) is associated with type 2 diabetes [2,3]. Numerous studies have raised the possibility that this ADAMSTS9 gene variant is associated with insulin resistance and beta-cell function. However, the molecular mechanisms underlying how ADAMTS9 affects beta cells and peripheral tissues are unknown. In mice, an ADAMTS9 null allele is lethal in early embryonic stages [4]. The phenotype of the ADAMTS9 gene variant (rs4607103) is milder, perhaps because this allele is a weak reductionof-function ADAMTS9 allele.
Insulin signaling is highly conserved between C. elegans and humans. In C. elegans, the insulin/insulin-like growth factor (IGF) receptor DAF-2 signals through the PI3-kinase [5] signaling cascade, which activates downstream serine/threonine kinases [6]. These kinases negatively regulate FOXO/DAF-16. The insulin signaling pathway has been shown to regulate developmental processes, such as dauer formation, lifespan and behavior [5,[7][8][9]. In the presence of food, worms progress through larval stages to the adult stage, whereas in the absence of food, worms halt development at the dauer larval stage. Because of fundamental similarities in the insulin signaling pathway between worms and humans, genetic and molecular pathways that control metabolism in the worm could be highly informative for analysis of human metabolic diseases such as type 2 diabetes.
ADAMTS9 is a metalloprotease that cleaves components of the extracellular matrix (ECM) [10,11], and it is also implicated in transport from the endoplasmic reticulum (ER) to the Golgi [12]. Downregulation of GON-1, the C. elegans homolog of ADAMTS9, results in the inhibition of protein transport from the ER to the Golgi [12]. Because a variant of ADAMTS9 has been established as a risk factor for type 2 diabetes [1], we investigated how ADAMTS9/ GON-1 is involved in the insulin/IGF-like signaling pathway. We show that both insulin secretion and the insulin/IGF-like signaling pathway are affected by GON-1 depletion and recovered by GON domain expression.

GON-1 is important for insulin secretion
In C. elegans, GON-1 has been reported to be expressed in distal tip cells (DTCs) and body wall muscle cells, and it functions in ECM remodeling and the protein secretory pathway. Depletion of GON-1 results in defects in DTC migration [13] and protein secretion [12]. To determine whether GON-1 is involved in other phenomena, we evaluated whether gon-1 is expressed in other cells in addition to DTCs and body wall muscle cells. gon-1 is expressed in certain additional cell types at the larval and adult stages, including neuronal, intestinal, and excretory cells. gon-1 is also expressed in certain cells at the embryonic stage (Fig 1 and S1 Fig).
To determine whether GON-1 functions in more than insulin secretion, we assessed the effects of the gon-1 mutation on DAF-7, another secretory protein. We investigated whether DAF-7 is normally secreted in the gon-1 mutant background. daf-7 encodes a TGF-β homolog that is secreted from ASI neurons. DAF-7::mCherry was hardly detected in ASI neurons (Fig  6a and 6c), and it accumulated in the coelomocytes of wild-type animals at the adult stage (Fig  6d and 6f). By contrast, the accumulation of mCherry fluorescence was observed in ASI neurons in the gon-1(tm3146) mutant background (Fig 6b and 6c) and DAF-7::mCherry fluorescence intensity was decreased in the coelomocytes of gon-1(tm3146) mutants (Fig 6e and 6f). Thus, GON-1 is necessary for the secretion of at least two major insulin peptides and TGF-β in C. elegans.
In C. elegans, insulins act as either DAF-2 agonists (such as INS-7 and DAF-28) or antagonists (INS-1 and INS-18) [18,19]. We investigated the effect of gon-1 depletion on the secretion of INS-18. Under growth-promoting conditions, INS-18::Venus was weakly detected in head neurons (Fig 7a and 7c), and it accumulated in the coelomocytes of wild-type animals at the adult stage (Fig 7d and 7f). By contrast, the accumulation of Venus fluorescence was observed in head neurons in the gon-1(tm3146) mutant background (Fig 7b and 7c), whereas INS-18:: Venus fluorescence intensity was decreased in the coelomocytes of gon-1(tm3146) mutants  (Fig 7e and 7f). Thus, secretory defects may not be specifically regulated between agonists and antagonists.

GON-1 is required in the cells where insulin acts
To investigate the role of GON-1 in peripheral tissues, we next examined the subcellular localization of DAF-16/FOXO because the localization of FOXO in the nucleus is controlled by insulin signaling. DAF-16 has been shown to translocate from the nucleus to the cytoplasm by activating the insulin/IGF-like signaling pathway [20]. We used a strain harboring the transgene muIs71, an integrated DAF-16a::GFP construct [21], to study DAF-16 localization. As previously reported, DAF-16a::GFP was present in both the nucleus and the cytoplasm of wildtype animals (Fig 8a). By contrast, DAF-16a::GFP was exclusively localized to the nucleus in gon-1(tm3146) mutants (Fig 8c), and this phenotype was rescued by the expression of the GON domain under the control of the gon-1 promoter as described above (Fig 8e). However, DAF-16a::GFP was localized to the nucleus in transgenic animals expressing either unc-119p::gon-1 (sig_GON) (Fig 8g) or myo-3p::gon-1(sig_GON) (Fig 8i and 8m) in the gon-1(tm3146) mutant background. We inferred that DAF-16a::GFP localization requires both insulin secretion from neurons and signaling at peripheral cells. We generated transgenic animals expressing myo-3and unc-119-promoter-driven gon-1(sig_GON) in gon-1(tm3146); muIs71 worms and observed the localization of DAF-16a::GFP. DAF-16a::GFP was rarely observed in the nucleus of muscle cells (Fig 8k and 8o, arrowheads). However, DAF-16a::GFP was localized to the nucleus of cells excluding muscle cells (Fig 8k and 8o, arrows).
Because C. elegans adult lifespan is extended by reduced insulin signaling, we next investigated whether gon-1 affects longevity. gon-1(tm3146) mutants exhibited an extended lifespan relative to wild-type animals, whereas the lifespan of gon-1(tm3146);tmEx[gon-1p::gon-1(sig_-GON)] animals was similar to or slightly longer than that of wild-type animals, shorten than that of gon-1(tm3146) mutant animals. Expression of the GON domain shortened the lifespan of wild-type animals (Fig 9b and S2 Table). We next investigated whether overexpression of the GON domain alters lifespan in Daf-c and Daf-d mutants. The lifespan of daf-16 (mu86) and daf-2(e1368) mutants was shortened by expression of the GON domain under the control of the gon-1 promoter (Fig 9c and S2 Table). Moreover, we investigated the effect of gon-1 depletion in daf-16(mu86) mutants and found that gon-1(tm3146) has no effect on the lifespan of daf-16 mutants (Fig 9c and S2 Table).

Discussion
The GON domain is required at insulin secretory cells and peripheral tissues for normal insulin signaling GON-1 belongs to the ADAMTS protease family, which functions in ECM remodeling in C. elegans. Depletion of GON-1 results in gonadal proliferation defects and aberrant membrane morphology in hermaphrodite DTCs [13]. We found that GON-1 is required for the protein secretory pathway from the ER to the Golgi [12]. In the present study, we show that the insulin/IGF-like signaling pathway is compromised by the depletion of gon-1 in C. elegans. The insulin secretion defect in gon-1 mutants was restored by GON domain expression under the control of either the gon-1 promoter or the pan-neuronal unc-119 promoter. However, localization of DAF-16a::GFP to peripheral tissues was not restored by expression of either unc-119p::gon-1(sig_GON) nor myo-3p::gon-1(sig_GON) in the gon-1(tm3146) mutant background. Proper localization of DAF-16a::GFP in the whole body was restored by GON domain expression under the gon-1 promoter. The localization of DAF-16a::GFP was restored only in the body wall muscles by co-expressing unc-119p::gon-1(sig_GON) and myo-3p::gon-1(sig_GON) in the gon-1(tm3146) mutant background. This result indicates that the expression of myo-3p:: gon-1(sig_GON) allowed peripheral tissues to receive insulin, but the insulin secretion defect was not restored in gon-1(tm3146); tmEx[myo-3p::gon-1(sig_GON)] transgenic animals. Therefore, DAF-16a::GFP was localized to the nucleus in these animals. Insulin is normally secreted from neurons, and received at the body wall muscle cells in gon-1(tm3146); tmEx[unc-119p:: gon-1(sig_GON), myo-3p::gon-1(sig_GON)] transgenic animals. Consequently, DAF-16a::GFP was not localized to the nucleus at body wall muscle cells, leaving DFA-16a::GFP localized to nuclei in other cells. These data suggest that the GON domain alone, but not the protease domain, is required for insulin secretion and signaling. In addition, our data suggest that expression of GON-1 at insulin secretory cells and peripheral tissues is important for normal insulin signaling.
It has been reported that the DAF-2/DAF-16-mediated insulin signaling pathway is coordinated among different tissues by positive-feedback regulation of an insulin-like peptide [14,23]. We speculate that this positive feedback loop in insulin signaling by INS-7 is blocked in the gon-1(tm3146) mutant background. We have reported that knockdown of ADAMTS9/ GON-1 inhibited membrane protein transport [12]. The data presented in this paper raises the possibility that GON-1 depletion in peripheral tissues might impair the transport of membrane proteins that are required for the insulin/IGF-like signaling pathway, thereby resulting in insulin resistance. Future studies are necessary to confirm that the insulin receptor and glucose transporters, which are related to insulin resistance, are indeed functionally inhibited by GON-1 depletion. In mammals, insulin regulates glucose uptake in peripheral tissues by promoting the translocation of GLUT4 to the plasma membrane [24]. It has been reported that glucose levels change in response to insulin/IGF-like signaling in C. elegans [25]. However the C. elegans genome encodes members of the glucose transporter superfamily, there is no evidence which glucose transporter is regulated by insulin [26]. It would be interesting to determine which molecules account for this phenomenon.

Overexpression of GON domain alters dauer formation and longevity
In our current experiments, we showed that depletion of GON-1 induced dauer formation and longevity. Surprisingly, the dauer formation phenotype was stronger in gon-1 mutant animals than in ins-7 or daf-28 mutants. Consistent with this observation, DAF-7, another protein involved in dauer formation, was also inefficiently secreted (Fig 6). Thus, dauer formation was more strongly induced by the depletion of GON-1 than by the depletion of INS-7 and DAF-28. In addition, gon-1 mutants suffer from Unc phenotypes (data not shown) by secretion defects. Unc phenotypes cause a nutrition uptake defect, and may augment dauer phenotype. Thus, gon-1 mutants may tend to form dauers and show extended longevity.
The Daf-c phenotype of gon-1(tm3146) mutants was partially rescued by GON domain expression under the control of the gon-1 promoter. The lifespan of gon-1 mutants was shortened by GON domain expression. Moreover, the Daf-c phenotype of daf-2(e1368) was suppressed by expression of the GON domain. The lifespans of daf-2(e1368) and daf-16(mu86) worms were shortened by expression of the GON domain. These data suggest that the amount of secreted insulin may be increased by overexpression of the GON domain, and consequently dauer formation and lifespan may be affected. It was reported that overexpression of INS-7 in the intestine shortens lifespan [14]. Surprisingly, expression of the GON domain shortened the lifespan of daf-16(mu86) mutants. This result indicates that the GON domain may also alters DAF-16-independent molecules such as extracellular matrix proteins, to control lifespan [27].

A model of insulin resistance and insulin secretion defects
It has been reported that the gene encoding O-linked-N-acetyl glucosaminidase (OGA), MGEA5, is linked to insulin resistance, a hallmark of type 2 diabetes [28][29][30]. In C. elegans, knockout mutants of the oga-1, a homolog of human OGA, decrease fat stores and exhibit resistance to the insulin-like signaling pathway [31,32]. The knockout mutant of oga-1 was proposed as a genetic model of insulin resistance in peripheral tissues. Our data suggest that GON-1 is involved directly or indirectly both in insulin secretion by affecting insulin secretory cells and insulin signaling by affecting peripheral tissues. The gon-1 mutant may be a model of both insulin secretion from insulin secretory cells and insulin resistance in peripheral tissues. It has been reported that a variant of ADAMTS9 is associated with type 2 diabetes [1]. One study reported that the diabetes risk allele of ADAMTS9 was associated with reduced insulin sensitivity but not with insulin secretion [33]. Another study reported that the diabetes risk allele of ADAMTS9 was associated with impaired beta cell function [34]. However, the molecular mechanisms through which ADAMTS9 affects insulin sensitivity and/or secretion in the diabetes risk variant are unknown. Although future studies are needed to conclude whether ADAMTS9 is involved both in insulin secretion by beta cells and insulin signaling at peripheral tissues in humans, considering the similar functions of the GON domains in worms and humans, it is highly possible that ADAMTS9 is involved in both mechanisms.
In summary, we show that the GON domain of GON-1 is involved in protein secretion from insulin secretory cells and in insulin/IGF-like signaling at peripheral tissues. GON-1 may be involved in transport of membrane proteins that are required for the insulin/IGF-like signaling pathway at peripheral tissues. Moreover, at least in C. elegans, the GON domain may play an important role in the insulin feedback loop.
To construct ins-7p::ins-7::mCherry, the mCherry fragment was inserted into the pFX_VT vector plasmid using the NotI and BglII sites, giving rise to pFX_VT_Venus [37], yielding pFX_VT_mCherry. The promoter and genomic region of ins-7 was amplified from C. elegans genomic DNA using the primers 5'-GGTTCCGCGTGGATCCTTTTGCTTCGAAGGAT AACCCCG-3' and 5'-GCTCACCATGCGGCCGCAAGGACAGCACTGTTTTCGAATG-3'. The fragment was inserted into the plasmid pFX_VT_mCherry using the NotI and BamHI sites.
The micro-injection of the DNA constructs described above was performed as previously reported [38] with appropriate selection markers. The transgenic strains used in this work are listed in S1 Table. Images were taken with a BX-51 microscope (Olympus, Japan) or a LSM710 confocal laser-scanning microscope (Zeiss, Germany).

Measurement of fluorescence intensity in coelomocytes and neurons
Images were taken with a BX-51 microscope (Olympus, Japan). Exposure times were described at each figure legend. Data of fluorescence intensity in coelomocytes and neurons were collected and scored by a blinded observer.

Assays for the Daf phenotype
Dauer assays were performed similarly to those previously described [39]. Approximately 100-200 eggs of each genotype were placed on seeded 3.5 cm NGM agar plates and allowed to starve at 25°C. Four days after the food was exhausted, the plates were flooded with 1% SDS solution to select for surviving dauers. After 10-15 min, the plates were scored visually for dauers by using a microscope.

Lifespan analysis
Lifespan assays were performed as described previously [40][41][42]. Briefly, synchronous animal populations were generated by hypochlorite treatment of gravid adults. Late L4 larvae growing at 20°C were transferred to NGM plates with FUdR (40 μM) and incubated at 20°C. In lifespan assays, the first day of adulthood was defined as day 1. Nematodes that failed to display touchprovoked movement were scored as dead. Nematodes that died from causes other than aging, such as sticking to the plate walls or gonadal extrusion were censored as lost worms. Statistical analyses of lifespan were performed on Kaplan-Meier survival curves in Prism 6 by Logrank (Mantel-Cox) tests (S2 Table).  Table. Detailed parameters of adult lifespan of each genotype. † The number of animals scored in Fig 9b and 9c. † † The number of animals that were censored because they had crawled off the plate or were exploded. ÃÃÃ p < 0.005; ÃÃÃÃ p < 0.0001; ns, not significant. (XLSX)