Insulin Receptor Substrate and p55 Orthologous Adaptor Proteins Function in the Caenorhabditis elegans daf-2/Insulin-like Signaling Pathway*

An insulin-like signaling pathway regulates development and lifespan in Caenorhabditis elegans. Genetic screens that identified many components of the C. elegansinsulin pathway did not identify homologs of insulin receptor substrates or the phosphoinositide 3-kinase (PI3K) adaptor/regulatory subunit, which are both required for signaling by mammalian insulin/insulin-like growth factor I pathways. The C. elegans genome contains one homolog of each protein. The C. elegans versions of insulin receptor substrate (IST-1) and PI3K p50/p55 (AAP-1) share moderate sequence similarity with their vertebrate and Drosophila counterparts. Genetic experiments show that ist-1 and aap-1 potentiate C. elegans insulin-like signaling, although they are not required for signaling in the pathway under most conditions. Worms lacking AAP-1 activity because of the mutation aap-1(m889) constitutively arrest development at the dauer larval stage when raised at high temperatures. aap-1 mutants also live longer than wild-type animals, a phenotype observed in other C. elegansmutants with defects in DAF-2 signaling. Interestingly, IST-1 appears to be required for signaling through a pathway that may act in parallel to AGE-1/PI3K.

In vertebrates and Drosophila, insulin-like signaling utilizes adaptor proteins that link activated growth factor receptors to intracellular signaling pathways (18 -20). Ligand binding activates the tyrosine kinase activity of the insulin and IGF-I receptors, resulting in the phosphorylation of the insulin receptor substrate (IRS). Phosphotyrosines on IRS act as binding sites for downstream targets of growth factor receptors such as PI3K and Ras/MAPK pathways. Biochemical and genetic studies in mammalian cells have shown that PI3K is a major component of signaling downstream from insulin and IGF-I receptors (21). The PI3K adaptor subunits p85, p55, p50␣, and p85␤ bind to phosphotyrosines within a YXXM motif on IRS and recruit the p110 catalytic subunit to the membrane for access to the lipid substrates. In addition, the interaction between the adaptor and catalytic subunits activates the p110 lipid kinase activity (22,23).
We have investigated the functions of two C. elegans genes, one homologous to mammalian PI3K adaptor subunits named aap-1 and the other homologous to IRS named ist-1. Genetic analysis demonstrates that aap-1 and ist-1 both function in the daf-2/insulin-like pathway. The physical interactions between the PI3K catalytic and adaptor subunits are conserved in the worm homologs despite low sequence conservation in the intersubunit interaction domains. Previous studies have shown that dauer arrest in age-1 mutants is suppressed by a gain-offunction mutation in Akt/protein kinase B, akt-1(mg144gf) (5). Both ist-1 and aap-1 are required for akt-1(mg144gf) to suppress age-1 mutations. These findings indicate that the DAF-2/insulin receptor-like protein may activate both AGE-1 PI3K signaling and a parallel pathway that remains to be identified.
Identification and cDNA Cloning of aap-1 and ist-1-The aap-1 (Y110a7a.10) and ist-1 (C54D1.3) gene structures were predicted using the GeneFinder and Blast gene prediction and analysis programs and were confirmed by sequencing partial cDNA clones (aap-1:yk442b3, ist-1:yk26d7, and yk110e3), which were obtained from the Genome Biology Laboratory at the National Institute of Genetics (Mishima Japan). The 5Ј ends of these genes were cloned using 5Ј-RACE (Strat- FIG. 1. A, AAP-1 encodes a p55-like PI3K adaptor subunit. Relative positions of SH2 domains I and II are shown with percent identity to human p85 and Drosophila p60 PI3K subunits shown below. For comparison, percent identity to the SH2 domain of human vav, the next closely related sequence, is shown. B, alignment of the SH2 domains of AAP-1, Drosophila p60 and human p85␤. Shaded residues are identical. C, the aap-1(m889) mutation truncates AAP-1 after the amino-terminal SH2 domain. D, AGE-1 amino-terminal domain (1-286) binds to AAP-1containing beads. Bacterially expressed AAP-1 with a COOH-terminal His 6 tag was bound to nickel-charged agarose beads. The AGE-1-(1-268) fragment in bacterial crude lysates binds to beads containing AAP-1 (lane 3) and not to beads lacking AAP-1 (lane 2). Lane 1 shows the AGE-1-(1-268)-containing lysate prior to incubation with AAP-1 beads. agene, La Jolla, CA) with total RNA obtained from a mixed population of wild-type (N2 Bristol) animals. The nucleotide coding sequences for aap-1 and ist-1 have been deposited in GenBank under accession numbers AF209707 and AY064245, respectively. Sequence alignments were performed using GCG Pileup or SMART (Simple Modular Architecture Research Tool, smart.embl-heidelberg.de/), and the output was displayed using GCG Prettybox (GCG, Madison, WI).
Identification and Characterization of aap-1(m889)-The aap-1(m889) mutation was identified in a screen for long-lived mutants (28) and was back-crossed three times with the fer-15(b26) parent prior to analysis. To test whether the wild-type Y110A7A.10 sequence could rescue the 27°C dauer arrest phenotype of the aap-1(m889) mutation, a 3.7-kb fragment including the Y110A7A.10 coding region and 963 bp of 5Ј sequence were amplified from genomic DNA by nested PCR using outer primers GATATTCTCCACGAGTACCG and TGCTCATTGGACT-TGTTGGC and inner primers GTCGTCAGCCGAGGAGCGCC and GGCCTGGTGGCTGGCAACGG. The PCR product was gel-purified and co-injected into the germ line of aap-1(m889) with the rol-6(s1006) marker plasmid. To test the Daf-c phenotype, Rol animals were placed at 27°C, and both Rol and non-Rol progeny were scored. 42 Rol and 2 non-Rol progeny matured to the adult, whereas 52 non-Rol segregants that did not carry the transgenic array arrested at the dauer stage, demonstrating that the transgene complements the m889 mutation.
Reduction of aap-1 and ist-1 Function by RNA Interference-For aap-1 and ist-1 RNA interference, a mixture of sense and antisense RNA (MEGAscript kit, Ambion, Austin, TX) of the aap-1 cDNA clone (yk442b3) or the ist-1 cDNA clone (yk26d7 or yk100e3) was injected at 1-5 g/l into L4 larvae and young adult hermaphrodites (25). The injected animals were allowed to recover overnight at 15°C and then shifted to 25°C unless otherwise indicated to lay eggs. The injected parents were removed from the plate, and dauer arrest was assayed in the progeny 72 h later.

AAP-1 Is the C. elegans Homolog of PI3K Adaptor Subunits-
Searches of the C. elegans genome sequence identified an open reading frame (Y110A7A.10) that contains the hallmarks of PI3K p50/p55 adaptor subunits, a pair of SH2 domains flanking a short inter-SH2 domain. We named this gene aap-1 (AGE-1 adaptor protein). The mammalian PI3K adaptor subunits p85, p55 ,p50␣, and p85␤ bind to phosphotyrosine within a YXXM motif in growth factor receptors and IRS (26). This binding is mediated by SH2 domains within the PI3K adaptor subunit. The SH2 domains in AAP-1 are most closely related in sequence to those in mammalian PI3K p85 and p55, suggesting that AAP-1 may have the same binding specificity as the mammalian proteins (Fig. 1, A and B). Although AAP-1 does not contain the extended amino terminus of p85, the worm genome does not contain any other p85-like genes, suggesting that AAP-1 is sufficient to provide adaptor subunit activity for AGE-1/p110. In mice, p50/p55␣ can compensate for the deletion of p85␣ in insulin-stimulated PI3K activation, although the absence of p85 causes altered glucose homeostasis (27).
Binding by the adaptor subunit serves to stabilize and activate the PI3K p110 catalytic subunit. The intersubunit interactions are mediated by the p110 amino-terminal domain and the adaptor subunit inter-SH2 domain (22,23). The predicted intersubunit interaction domains of AAP-1 and AGE-1 are not well conserved with the Drosophila and vertebrate homologs (data not shown). To determine whether AAP-1 binds to the AGE-1 amino terminus, this interaction was tested in vitro. When bound to beads, bacterially expressed AAP-1 co-precipitated an amino-terminal fragment of AGE-1-(1-268) containing the predicted interaction domain (Fig. 1D). The AGE-1-(1-268) fragment did not bind to beads lacking AAP-1. This evidence indicates that the intersubunit interaction domains of PI3K are conserved in AGE-1 and AAP-1 despite low sequence conservation.
An AAP-1::GFP translational fusion with 1.8 kb of upstream sequence (containing the presumptive promoter) was used to identify cells in which AAP-1 might function. Although AAP-1::GFP expression was weak, fluorescence was consistently observed in the intestine and in neurons and was occasionally observed in body wall muscles and the hypodermis (data not shown). Expression was observed at all developmental stages beginning with early embryos and continued through adulthood. AAP-1::GFP did not appear to perturb aap-1 or age-1 function in vivo, because AAP-1::GFP-containing animals appeared grossly normal and did not arrest at the dauer stage (data not shown). The expression pattern of aap-1 is similar to that of the age-1 effectors akt-1, pdk-1, and daf-16, which are also widely expressed throughout development (5-7). However, the GFP expression pattern may not represent the full expression pattern of the endogenous aap-1 gene.
AAP-1 Is Required for Full Signaling through the DAF-2 Pathway-PI3K signaling by AGE-1 in C. elegans is required for wild-type adult lifespan and to bypass dauer larval arrest (4,(12)(13)(14). Genetic analysis shows that a loss of aap-1 gene activity also lengthens adult lifespan and causes dauer arrest, consistent with a function in the same pathway as age-1. The aap-1(m889) mutation was identified in a selection for thermotolerant mutants (28) and was mapped to the left arm of chromosome I between unc-38 and dpy-5. An additional three-factor cross-mapped aap-1(m889) to a 0.34-map unit (242 kb) region between unc-38 and unc-63. The 41-kb YAC sequence between cosmid C30F8 and unc-63 contains 14 predicted genes including Y110A7A. 10. The sequencing of Y110A7A.10 revealed an amber nonsense mutation in the m889 mutant at Trp-325 (TGG to TAG), which would result in a truncated protein that lacks the COOH-terminal 198 amino acids including the second SH2 domain (Fig. 1C).
At 25.5°C, aap-1(m889) mutant adults have a mean lifespan of 33 days, nearly twice as long as the controls (Table I). When grown at 27°C, aap-1(m889) animals arrest development as dauer larvae (Table I). Dauer arrest at 27°C is rescued by wild-type maternal aap-1 activity (data not shown), similar to the maternal rescue of age-1 mutations (13,14). The DAF-16 forkhead-like transcription factor is necessary for dauer arrest and long lifespan in age-1 pathway mutants as well as the phenotypes of aap-1(m889) mutants (28). In addition, the presence of aap-1(m889) in a double mutant does not affect daf-2(m41) adult longevity. This is consistent with placing aap-1 into the same pathway as age-1.
To further support the hypothesis that aap-1 functions in the same pathway as age-1, we tested whether the loss of aap-1 function by RNA interference (RNAi) could enhance weak mutations in age-1 or in daf-2, encoding an insulin receptorlike protein that activates AGE-1 PI3K (1). aap-1(RNAi) did not cause any detectable phenotype in wild-type animals grown at 25.5 or 27°C, the restrictive temperatures for many mutations in the daf-2 pathway. However, we found that aap-1(RNAi) could enhance the phenotype of weak mutations in the daf-2 pathway. age-1(hx546) is a weak allele of age-1 that causes dauer arrest at 27°C but not at 25.5°C (17,29). aap-1(RNAi) dramatically enhanced dauer arrest at 25.5°C in the age-1(hx546) background (Fig. 3A). aap-1(RNAi) also enhanced dauer arrest in daf-2(e1370) mutants at a semipermissive temperature (22°C) (Fig. 3B). Together, these results show that aap-1 potentiates signaling by the daf-2/ age-1 pathway.
A TGF-␤ pathway also controls C. elegans dauer arrest in parallel to the daf-2/age-1 pathway (30, 31). The gene daf-1 encodes the type I TGF-␤ receptor essential for signaling through this pathway (31). aap-1(RNAi) in the daf-1(m40) mutant at semi-permissive temperature did not enhance dauer arrest, indicating that aap-1 activity is not required for signaling via the TGF-␤ branch of the dauer pathway (Fig. 3B).
IST-1 Is a Homolog of the Mammalian IRS Proteins That Acts in the DAF-2 Pathway-A predicted protein distantly related to vertebrate IRS proteins was identified in the C. elegans genome (32), which we named ist-1 (insulin receptor substrate). IRS proteins contain amino-terminal PH and phosphotyrosine binding (PTB) domains for binding to activated insulin and IGF-I receptors (19). The IRS carboxyl terminus contains tyrosine substrates for insulin/IGF-I receptor kinase that upon phosphorylation act as docking sites for downstream effectors including PI3K and Ras/MAPK. The C. elegans IST-1 protein contains recognizable PH and PTB domains, although the overall sequence similarity is low. The PH domain of IST-1 is only 13-21% identical to the PH domains of other IRS proteins. The IST-1 PTB domain is better conserved and is 29% identical to the PTB domain of human IRS-1. IST-1 contains 22 tyrosines, some of which may recruit downstream proteins containing SH2 domain (Fig. 2C). There is a single YXXM motif for AAP-1/AGE-1 PI3K binding (Tyr-530). IRS proteins from other species also contain at least one YXXM binding site for PI3K. In addition, Tyr-321 in the PTB domain matches consensus binding sites for Grb-2 (YY/I/VXF/L/I/V) and SH-PTP2 (YV/IXV/I/ L/P) (26,33). To date, neither protein has been implicated as a component of the DAF-2/insulin-like signaling pathway, although C. elegans homologs have been characterized (34,35).
To determine whether IST-1 functions in the DAF-2 pathway, RNA interference was used to inactivate ist-1 function. As with aap-1, reduced ist-1 function is predicted to decrease daf-2 pathway activity, resulting in constitutive arrest at the dauer larval stage. Arrest at the dauer larval stage was not observed in wild-type animals with ist-1(RNAi) when grown at either 27 or 25.5°C (data not shown). However, ist-1(RNAi) enhanced dauer arrest in the age-1(hx546) background at 25.5°C as observed with aap-1(RNAi) (Fig. 3A). This result is consistent with placing IST-1 in the DAF-2 pathway.
The DAF-2 Receptor May Activate Parallel Downstream Pathways-Phosphorylation of IRS by the insulin receptor provides docking sites for downstream pathways including PI3K and Ras/MAP kinase pathways. To date, DAF-2/insulin receptor signaling has only been shown to act through the AGE-1/ PI3K pathway. However, evidence suggests that DAF-2 also acts through a parallel AGE-1-independent pathway. A gainof-function mutation in akt-1 suppresses dauer arrest in age-1 null mutants but not in daf-2 mutants (5). The akt-1(mg144gf) mutation contains an A183T substitution in the linker region between the PH and kinase domains and may activate the AKT-1 kinase activity in the absence of phospholipids. The inability of akt-1(mg144gf) to suppress daf-2 mutants has been postulated to suggest that signaling bifurcates downstream of DAF-2 to activate both AGE-1/PI3K and a parallel pathway. In addition, Gems et al. (3) proposed that some daf-2 mutants (e.g. e1370) have reduced signaling in both parallel pathways, whereas others (e.g. m41) are affected in only one. Class 1 mutants like m41 are dauer-constitutive, thermotolerant, and long-lived. Class 2 mutants exhibit these traits but are more pleiotropic with high levels of embryonic and L1 arrest, de- FIG. 3. aap-1 and ist-1 are required for full activity of the daf-2/insulinlike pathway. Dauer arrest and arrest as sterile adults are shown at the percent of a population of animals with aap-1 or ist-1 loci inactivated by RNA interference. The progeny of animals injected with double-stranded RNA of the corresponding gene were grown at 25.5°C (A) or at indicated temperature (B) for 72 h before dauer arrest was scored.

AAP-1 and IST-1 Potentiate Signaling Downstream from the DAF-2
Receptor-Using a combination of forward and reverse genetic approaches, we have demonstrated that the C. elegans DAF-2/insulin receptor utilizes an IRS-like protein, IST-1, and a PI3K p55-like adaptor subunit, AAP-1, to activate downstream signaling pathways. The description and characterization of these genes is necessary for understanding how signals are transduced from the DAF-2 receptor to AGE-1, the PI3K p110 catalytic subunit. Our results show that both AAP-1 and IST-1 are required for full DAF-2 pathway signaling. However, the relatively subtle phenotypes revealed by RNAi and by the aap-1(m889) mutation suggest that these genes may not be essential for DAF-2 pathway signaling under normal growth conditions.
Mutations in aap-1 and ist-1 have not been identified in forward genetic screens for dauer arrest mutants. The aap-1(m889) allele causes both developmental and lifespan phenotypes similar to those caused by mutations in other daf-2 path-way genes. In addition, the RNAi phenotypes provide support for placing aap-1 and ist-1 into the daf-2 pathway. RNAi can phenocopy the loss-of-function phenotypes of most genes. However, genes expressed in the C. elegans nervous system are usually resistant to RNAi. Both daf-2 and age-1 act in the nervous system to regulate dauer arrest and lifespan, although dauer arrest can be regulated by daf-2 pathway activity in some non-neuronal cell types such as intestine and muscle (36,37). The weak phenotypes resulting from aap-1 and ist-1 RNAi may reflect the inability to sufficiently remove the activities of the genes from the nervous system.  4. ist-1 is required for reproductive development when insulinlike signaling is mediated by gain-offunction AKT. Dauer arrest was scored in the progeny of age-1(mg44);akt-1(mg144gf) animals with ist-1(RNAi) or aap-1(RNAi). Animals were grown at 25.5°C for 72 h before development was scored.
An alternative explanation for the weak RNAi phenotypes we observed is that AAP-1 and IST-1 are dispensable for DAF-2 pathway signaling in otherwise wild-type backgrounds. This would suggest that the AGE-1/p110 catalytic subunit can be activated directly by the DAF-2 receptor or redundantly by another mechanism. The finding that null mutations in the Drosophila PI3K adaptor subunit Dp60 are not as severe as null mutations in the p110 catalytic subunit (Dp110) supports this view (38). One explanation for these results may be that p110 has a low basal level of lipid kinase activity in the absence of receptor stimulation. In the fly and the worm, this low basal activity may be sufficient for normal or near-normal development under most conditions. Only when PI3K signaling is compromised in other ways, i.e. by mutations in the p110 catalytic subunit, is the requirement for the adaptor subunit revealed.
In the mouse, signals from the insulin receptor are transduced via IRS to several redundant PI3K adaptor subunits p85␣, p55␣, p50␣ (all transcribed from the same gene), p85␤, or p55 PIK (39). Recent genetic analysis of these adaptor proteins suggests that they act both positively and negatively with respect to insulin signaling. Knock-out mutants in either the gene encoding p85, p55, p50␣, or p85␤ displayed increased insulin sensitivity as compared with wild type, suggesting that reductions in PI3K adaptor/regulatory protein activity is beneficial in this instance (27,40,41). We find that the lack of aap-1 activity weakens the daf-2 pathway, suggesting that negative inputs from PI3K adaptor subunits are not a general feature of insulin-like signaling pathways. The increased complexity and redundancy of the mammalian insulin pathways relative the C. elegans DAF-2 pathway may account for these phenotypic differences.
We found that aap-1(RNAi) slightly enhanced dauer arrest in the age-1(mg44); akt-1(mg144gf) background. We had expected that aap-1(RNAi) would have no effect, because age-1(mg44) is a nonsense mutant that deletes the entire lipid kinase domain (4). In this background, aap-1 activity should be dispensable. One explanation for this result is that AAP-1 weakly potentiates all signaling by the DAF-2 receptor, perhaps by competing with negative regulators for binding to the DAF-2.
IST-1 May Be Required to Activate Parallel Outputs of the DAF-2 Receptor-Our genetic experiments in the akt-1(mg144gf) background suggest that the C. elegans IRS homolog, IST-1, is necessary to activate an AGE-1-independent output of the DAF-2 receptor (Fig. 5). In mammalian cells, IRS can couple to multiple outputs including the Ras/MAPK pathway and PI3K to mediate insulin receptor signaling (42). These proteins contain SH2 domains that recognize phosphotyrosines within specific amino acid motifs (33). For example, the p85 PI3K subunit binds to phosphotyrosine within a YXXM motif. The sequence of the Drosophila IRS protein, Chico, contains binding sites for PI3K, a known output of Drosophila insulin signaling as well as potential GRB2/DRK and SH-PTP2 binding sites (20). IST-1 contains a YXXM motif for PI3K binding and a potential SH-PTP2 consensus binding site (YV/IXV/I/ L/P) (26). SH-PTP2 binding to IRS appears to attenuate responses of mammalian cells to insulin, suggesting that this protein could antagonize DAF-2 signaling in the worm (43).
In summary, we have characterized the genetic activities of the C. elegans homologs of mammalian PI3K adaptor/regulatory subunits, AAP-1, and IST-1. Reducing the functions of these genes in wild-type backgrounds did not disrupt development, suggesting that they may be dispensable for growth under normal conditions. However, an aap-1 nonsense mutation extends adult longevity. Both AAP-1 and IST-1 are required for development in the background of weak mutations in the insulin-like signaling pathway of the worm. By using a combination of reverse genetics and sensitized backgrounds, the roles of other proteins that potentiate signaling by the DAF-2 insulin pathway may be addressed.