Alternative polyadenylation results in a truncated daf-4 BMP receptor that antagonizes DAF-7-mediated development in Caenorhabditis elegans.

The DAF-4 receptor kinase, which promotes larval development, is encoded by a 2.9 kb mRNA transcribed from the only type II TGF-beta/BMP receptor gene in Caenorhabditis elegans. Here we report that alternative polyadenylation in intron 5 of daf-4 results in a 2.0 kb mRNA that encodes an open reading frame including only the N-terminal secretion signal and ligand-binding domains, and not the transmembrane or kinase domains, of DAF-4. Northern blots and real-time RT-PCR amplifications using RNA samples from developmentally staged animals show that expression levels of both the 2.9 kb and 2.0 kb transcripts are relatively constant, and their abundances similar, except for the transition between non-dauer and dauer stages. In dauer larvae, the steady-state level of the 2.0 kb mRNA increases more than 10-fold and exceeds the 2.9 kb transcript, coincident with an absence of signaling from DAF-4. Transgenic expression of a recombinant daf-4 transgene that encodes only the 2.0 kb mRNA enhances the Daf-c phenotype of a daf-4 hypomorph, whereas the same transgene with a nonsense mutation does not. These data suggest that a polypeptide encoded by the 2.0 kb transcript can function as an antagonist of full-length DAF-4 signaling. Alternative processing of type II receptor transcripts to generate an antagonist is a novel mechanism for negative regulation of a TGF-beta signaling pathway.

The Caenorhabditis elegans dauer diapause is regulated by the nervous system in response to temperature, food, and pheromone (1). Molecular genetic analyses have shown that the nervous system responds to environmental stimuli to regulate larval development via TGF-␤ 1 (2), cyclic GMP (3), and insulinlike (4,5) signaling pathways, which in turn act on a nuclear receptor (6) to control dauer versus non-dauer morphogenesis (7). Dauer larvae are physically adapted for survival in stressful environments and are long-lived (8,9). In the soil, development to adulthood resumes after dauer larvae disperse to a fresh environment.
The concentration of a secreted dauer-inducing pheromone is an indicator of C. elegans population density (10). The daf-7 gene, which encodes a TGF-␤/BMP-like ligand expressed in amphid sensory neurons in the head, is actively transcribed in low pheromone environments, whereas high concentrations of pheromone inhibit its transcription, leading to dauer larva development (2). Secreted DAF-7 activates signaling via DAF-4 (11) and DAF-1 (12,13) transmembrane receptor serine/threonine kinases. These receptors in turn control the activities of SMAD transcription factors encoded by daf-8 (14), daf-14 (15), and daf-3 (16), which regulate genes controlling dauer and non-dauer morphogenesis. DAF-7 signaling also stimulates resumption of development following diapause, with reinitiation of daf-7 transcription an early event in dauer recovery (2).
A dual mechanism of activation has been proposed for mammalian TGF-␤ (17), activin (18), and BMP (19) receptor complexes, as well as DPP receptors thick-veins and punt (20) in Drosophila. In these systems, the type II receptor binds ligand and then associates with the type I receptor in a heteromeric ligand-receptor complex. Phosphorylation of type I by a type II kinase releases conformational inhibition of type I receptor kinase activity, which can then phosphorylate transcriptionregulating SMADS. Based on analysis of the completed C. elegans genome sequence (21), DAF-4, which can bind mammalian BMPs in cultured cells (11), is the only type II receptor in C. elegans. In addition to regulating dauer larva development with DAF-1, DAF-4 also functions with the SMA-6 type I receptor (22) in response to the DBL-1 ligand (23) to promote body size and male tail morphogenesis. DAF-1, unlike most type I receptors (which absolutely require type II transphosphorylation for activation), maintains some ligand-dependent signaling activity in the absence of the DAF-4 kinase (13).
Negative regulators of TGF-␤/BMP receptor signaling have been identified, some of which are secreted proteins that prevent interaction between TGF-␤/BMP ligands and their receptors. Follistatin antagonizes signaling by interaction with the BMP-receptor complex (24). Chordin and Noggin both bind extracellular BMPs and prevent their interaction with receptors on the cell surface (25), as do the secreted antagonists Gremlin (26), Cerberus (27), and Dan (28). Here we report an alternatively polyadenylated transcript of daf-4 that encodes only the ligand-binding domain of DAF-4 and is expressed at all stages of development. Overexpression of genomic DNA that encodes only this alternative form of mRNA enhances dauer larva development in a daf-4 temperature-sensitive mutant, suggesting that the truncated protein can function as an antagonist of DAF-4/DAF-1 receptor signaling.

EXPERIMENTAL PROCEDURES
Growth and Maintenance of Strains-Strains used in this work include the wild-type N2, and the daf-4 III mutants described in Fig. 5. For analysis of Daf-c phenotype, all mutants were maintained at 15°C on NG agar plates seeded with the Escherichia coli uracil auxotroph OP50 (29) prior to analysis. For each strain, 10 or 20 gravid adults were transferred to a fresh plate and incubated at 15, 20, 22.5, or 25°C. After 6 -12 h, the adults were removed, eggs were returned to the experimental temperature, and hatched progeny were scored for the presence of dauer larvae at the following times: 2 days at 25°C, 3 days at 22.5°C, 4 days at 20°C, and 5 and 6 days at 15°C. Dauer larvae and adults were counted and removed at the time of scoring. Animals that were neither dauer nor adult were incubated overnight at the same temperature and scored on the following day. For isolation of RNA from specific developmental stages, mixed stage N2 cultures were grown at 20°C in S medium containing 5% w/v E. coli strain 1666. Eggs were purified from these animals and hatched to L1 larvae in M9 buffer (30). Synchronized populations of fed L1, L2, L3, L4, and adult N2 worms were harvested as aliquots of a starter culture of purified and hatched L1 larvae grown in 5% 1666. Dauer larvae were cultivated in S medium containing 0.5% 1666, and were then purified by their survival in 1% SDS. Following the transfer of dauer larvae to fresh S medium with 5% 1666, post-dauer populations were harvested at 6 h (PD1, feeding dauer larvae) and 24 h (PD2, L4 larvae). Since the daf-4 mutant larvae analyzed in Fig. 5 primarily arrest as dauer larvae when grown in liquid at 15°C, RNA was isolated exclusively from dauer larvae preparations of all strains in Fig. 5C.
Reverse Transcription and PCR Amplification-RT-PCR reactions with reagents from Display Biotech were conducted according to the manufacturer's protocols using a reverse transcription primer within intron 5 (5Ј-TTCGCCTGAACGCTTCTACATTAG-3Ј) and a second primer corresponding to the start of the daf-4 open reading frame (5Ј-CGCGAGCTCCCCGGGATGAATCAGAAAGGGACA-3Ј) for amplification. Amplified DNA was resolved by agarose electrophoresis and transferred to nylon, whereupon a 1.1 kb product tested positively for hybridization with both exon 1-6 and intron 5 32 P-labeled probes. The gel-purified fragment was cloned into pGEM-T (Invitrogen) and sequenced. For rapid amplification of cDNA ends by RT-PCR, primers specific for the conserved splice leader 1 (SL1) sequence (5Ј-TCTAGAAT-TCCGCGGTTTAATTACCCAAGTTTG-3Ј) and intron 5 (same as above) were used for 5Ј-RACE, whereas oligo dT (5Ј-AACTGCAGGATCCTCG-AGTTTTTTTTTTTTTTTTT-3Ј) and intron 5 (5Ј-ACAAATTTCCGGTG-GAAGTCAATC-3Ј) primers were used for 3Ј-RACE. The 3Ј-RACE product was cloned into pGEM-T and sequenced to determine the site of cleavage and poly(A) addition.
Real-time RT-PCR-Total RNA prepared as described for Northern analysis was treated with Amplification Grade DNase I (Invitrogen) to remove genomic or transfected DNA, and purified using the Qiagen RNeasy Mini kit. Alternatively, small scale quantities of total RNA were prepared from worms using Ambion's RNA-4RT-PCR miniprep kit. Reverse transcription and real-time PCR were conducted using the TaqMan EZ RT-PCR kit and an ABI Prism 7700 Sequence Detection System according to the manufacturer's specifications (Applied Biosystems Incorporated). Three sets of oligomers (Integrated DNA Technologies) were selected to quantitate PCR-amplified reverse transcription products from the two daf-4 mRNAs. Exon 4/5 primers amplify sequence common to 2.0 and 2.9 kb transcripts (sense, antisense, and probe sequences shown 5Ј to 3Ј: CTCCTGTCGCCAAGGACG, TTGGT-CGAACAGCAGCACA, 6-FAM-TCATTTCGCGGCGGAATTGGA-BHQ-1). INTR 5/UTR primers amplify sequence in intron 5 that is spliced from the 2.9 kb transcript, but maintained in the 3Ј-untranslated region of the 2.0 kb transcript (ATGTAGAAGCGTTCAGGCGAAA, CCCTAT-TACTGTGGCCGAGC, 6-FAM-CGAGAGAGTTTGGCGATGTTTGGTT-TG-BHQ-1). Exon 8 primers amplify sequence that is unique to the 2.9 kb spliced transcript (AAAAACTGACGCACTGGAAGC, CCACCATTT-CAATCATTTCCTCT, 6-FAM-AATGTGCCTCTCGTGGAGCCGG-BH-Q-1). Each RNA sample was analyzed in triplicate with each probe. Real-time PCR data in Fig. 6B are derived from RNA samples isolated from developmentally staged worms grown on two different dates, 1 year apart.
To estimate the copy number of RNA template in the reactions by measurement of probe fluorescence following each cycle of PCR, a dilution series of a plasmid encoding daf-4 genomic DNA (from 10 3 to 3 ϫ 10 5 plasmid copies) was analyzed with the 3 probes during each PCR run. A standard curve for each fluorescent probe was generated with the data from these dilutions, and the copy number of template in the unknown samples was estimated by linear regression. Because each standard curve is based on amplification values from a DNA template, no adjustment for differences in the efficiency of reverse transcription between primers can be made, and absolute differences in quantities of the 3 PCR products within a particular stage cannot be accurately calculated. Changes in transcript abundance during development are depicted by calculating the ratio of PCR product produced from each primer set in the L1 stage relative to every other stage.
Transformation of C. elegans-A 6.6-kb SalI/XbaI restriction fragment was isolated from a plasmid containing a 10-kb SalI fragment of daf-4 genomic DNA (kindly provided by Garth Patterson) and ligated into pBluescript (Stratagene) to generate DR 416. This plasmid was cut with Age I, and the complementary ends were filled in using E. coli DNA polymerase I Klenow fragment and re-ligated to generate DR 438. daf-4(m592) mutant hermaphrodites were microinjected either with DR 416 or DR 438 and with pRF4 (31), containing the dominant rol-6(su1006) gene as a transformation marker, at a concentration of 100 g/ml for each plasmid. A control line was constructed by injection of pRF4 alone. Roller (Rol) F1 progeny were isolated, and transgenic lines were maintained by continued propagation of Rol offspring. The percent of Rol and non-Rol progeny that developed as dauer larvae in each line was calculated at 20 and 22.5°C to determine the effect of each transgene on Daf-c phenotype.

RESULTS
A 2.0 kb daf-4 Transcript Encodes the Extracellular Domain-A daf-4 cDNA probe hybridized with 2.9 and 2.0 kb transcripts in Northern blots of poly(A)-enriched RNA prepared from mixed stage wild-type (N2) populations grown in abundant food (Fig. 1A). The 2.9 kb mRNA contains an 11 exon, 2232-bp open reading frame (11) encoding a 744-amino-acid polypeptide with extracellular, membrane-spanning, and cytoplasmic kinase domains (Fig. 1C). To determine which of these domains are encoded in the 2.0 kb transcript that is most abundant in dauer larvae, Northern blots of mixed stage N2 RNA were hybridized with five cDNA probes corresponding to exons in the extracellular domain, the transmembrane, and juxtamembrane regions, the kinase domain, and the C terminus (Fig. 1B). Whereas all five probes hybridized with the 2.9 kb transcript, only two probes (corresponding to the extracellular domain) hybridized with the 2.0 kb transcript, indicating that the smaller transcript is an alternatively processed form of daf-4 mRNA that terminates prior to exon 7.
Since the coding sequence encompassing exons 1-6 is less than 700 bp, we questioned whether any large intron sequences, which would normally be spliced out of the full-length transcript, are maintained in the 2.0 kb RNA. Consequently, a 1.6 kb probe corresponding to the sequence of intron 5 was found to hybridize exclusively to the 2.0 kb transcript, and not to the 2.9 kb transcript (Fig. 1B). We were unable to identify any cDNA clones containing intron 5 from the C. elegans data base, or to isolate one from existing cDNA libraries. We successfully synthesized a first-strand cDNA with a primer complementary to intron 5 using a thermostable reverse transcriptase that functions at 65°C, suggesting that there may be considerable secondary structure within intron 5 at lower temperatures, and that this is likely to be the cause for absence of 2.0 kb cDNAs from existing libraries. PCR amplification with a second primer at the start of the daf-4 open reading frame yielded an 1135-bp fragment, the sequence of which included exons 1-5 contiguous with intron 5, consistent with Northern blot results.
The 2.0 kb Transcript Is Trans-spliced and Polyadenylated-At the 5Ј-end of the daf-4 open reading frame is a 6-bp consensus for trans-splicing adjacent to the start codon (Fig. 2).
Having previously demonstrated that the 2.9 kb daf-4 cDNA is trans-spliced (11), we determined whether the 2.0 kb daf-4 transcript is similarly processed using RT-PCR amplification with primers specific for the SL1 splice leader and intron 5. The resulting product matched the length (1170 bp) of the predicted 2.0 kb RNA and hybridized with an intron 5 probe in a Southern blot (data not shown). Chemical sequencing of an exon-1primed extension product from N2 poly(A) RNA showed a 5Јterminus that matched both the length and sequence of the SL1 spliced leader. To identify the 3Ј-end of the 2.0 kb transcript, we employed RT-PCR with primers that complement the poly(A) tail and the intron 5 sequence unique to this mRNA. The polyadenylation site mapped 225 nt upstream of the intron 5/exon 6 junction. The putative 6-nt poly(A) recognition sequence (AGUAAA), which matches the sequence found in some other C. elegans mRNAs (32), is 10 nt upstream from the site of cleavage and poly(A) addition (Fig. 2).
We conclude that the smaller daf-4 transcript is 2032 nt in length, and includes a 22-nt SL1 spliced leader sequence, 621 nt of exons 1-5, and 1389 nt of intron 5 upstream of the poly(A) addition site. The putative translation product of the single open reading frame is a 206-amino acid polypeptide that includes 204 amino acids of the DAF-4 extracellular domain (the transmembrane motif of DAF-4 is proposed to begin at Phe-228) and a novel Gly-Glu dipeptide from the junction of exon 5 and intron 5 (Fig. 1C). Although missing 24 amino acids of the DAF-4 extracellular domain encoded in exon 6, this truncated receptor maintains polypeptide sequence that is highly conserved among type II receptors (see Supplementary figure).
Sequence Differences in Caenorhabditis Species-To determine whether the two daf-4 transcripts in our N2 laboratory strain reflect the natural wild-type state of C. elegans, we used Northern blotting to examine daf-4 transcripts in three diverged, wild-type isolates. The strains were from the UK (CB4852), California (CB4854), and Hawaii (CB4856). These strains, which had divergent patterns of Tc1 insertions in their genomes (33), exhibited the 2.9-and 2.0-kb daf-4 transcripts found in N2 (data not shown).
The recent completion of the genome sequence of Caenorhabditis briggsae (34), one of C. elegans closest known relatives (35), afforded us a unique opportunity to compare the amino acid sequences and splicing organization of two daf-4 genes from related, but evolutionarily distinct species. We call attention to three features of the alignments depicted in Fig. 3: First, C. elegans daf-4 protein coding sequence is highly homologous with C. briggsae from exons 2 through 5, and again from exons 7 through 11, but there is an additional exon (C. elegans exon 6) encoding 20 amino acids that juxtapose the C terminus of the type II receptor Cys-box motif. Second, exon boundaries are conserved between exons 2 and 11, except for C. elegans exon 6. Third, the alternative polyadenylation site in intron V that is responsible for multiple C. elegans daf-4 transcripts is absent from C. briggsae.
Extrachromosomal Expression of a 2.0 kb RNA Enhances Dauer Larva Formation-If the 206-amino acid DAF-4 polypeptide encoded by the 2.0 kb RNA open reading frame can bind ligand, we hypothesize that it would compete with fulllength DAF-4 and DAF-1 for DAF-7, but lacking any intracellular kinase activity, would not contribute to intracellular signaling. In this case, overexpression of the open reading frame should enhance the formation of dauer larvae by reducing the concentration of free ligand. To test this hypothesis, a 6.6-kb fragment of daf-4, including 2.8 kb of upstream regulatory sequence and complete exon and intron sequences through intron 5, was subcloned and microinjected with a plasmid expressing rol-6(su1006) (31) into wild-type N2 hermaphrodites. We also microinjected the truncated daf-4 gene and rol-6(su1006) into daf-4(m592), as well as into daf-1(m40) Daf-c temperature-sensitive mutant hermaphrodites, to look for enhancement of the Daf-c phenotype at temperatures permissive for growth. A mutant version of the transgene was constructed as a control for enhancement of the Daf-c phenotype in the absence of translation of the 2.0 kb ORF. Klenow-mediated fill-in of an Age I restriction site in exon 1 shifts the daf-4 open reading frame by ϩ1, resulting in premature termination at an AUG stop codon following amino acid 50 (see Fig. 2).
By cloning roller (Rol) F2 progeny from cloned F1 Rol mothers, we cultivated various transformed lines harboring extrachromosomal concatameric arrays that express the 2.0 kb wildtype or mutant transcripts in a percentage of their progeny. Synchronous populations containing Rol and non-Rol progeny from each transgenic line were scored for dauer formation to measure the effect of each daf-4 transgene on the Daf-c phenotype of each line (Table I). No dauer larvae developed from N2 transgenic lines grown at temperatures between 15 and 25.5°C (data not shown). In three daf-4(m592) transgenic lines (DR2161, DR2162, DR2163) carrying the wild-type daf-4 genomic sequence (2.0 kb WT), constitutive dauer formation in abundant food was significantly higher in Rol progeny relative to non-Rol siblings. Between 2.4 and 8.9% of Rol animals formed dauer larvae, whereas all non-Rollers grew to adulthood at 20°C, similar to the untransformed mutant strain DR1359. At 22.5°C, less than 2% of non-Rol progeny developed as dauer larvae, whereas 37-70% Rol offspring in the transgenic lines were dauers. Enhancement of Daf-c by the transgene was also observed in three daf-1(m40) stable lines (data not shown). It is important to note that the daf-4(m592) line (DR2333) carrying the frame-shifted daf-4 ORF (2.0 kb MT) showed no significant difference in the percent dauer formation of Rol and non-Rol animals at 20 or 22.5°C. Furthermore, no difference in dauer larva formation was observed between Rol and non-Rol animals in a control line (DR2270) transformed with rol-6(su1006) alone. These controls permit the conclusion that an intact 2.0 kb open reading frame is essential for enhancing the formation of dauer larvae. These data support the hypothesis that, when over-expressed, the putative 206-amino acid product of the open reading frame can function as an antagonist in the DAF-4/DAF-1 signaling pathway that promotes continuous growth.
To verify that each transgene overexpresses daf-4 mRNA, and show that enhancement of Daf-c is not due to titration of an endogenous dauer repressor by the transgenic DNA, we employed real-time PCR to assess daf-4 expression levels in daf-4(m592) and three of the m592 transgenic strains (expressing 2.0kb WT, 2.0kb MT or rol-6 only) (Fig. 4). Three different daf-4 sequence probes were used to measure expression: exon 4/5 primers amplify sequence common to 2.0 and 2.9 kb mRNAs, intron 5 primers are unique to sequence in only the 2.0 kb transcript, and exon 8 primers amplify coding sequence exclusive to the 2.9 kb sequence. Transgenic line DR2163, which expresses 2.0 kb WT, demonstrates 12-and 15-fold increases in sequences encoded in the 2.0 kb RNA (exons 4/5 and intron 5) relative to daf-4(m592), but no increase the copy number of sequence specific to the 2.9 kb RNA (exon 8). 3-5-fold overexpression of exon 4/5 and intron 5 sequences were also observed in the 2.0 kb MT line DR2333. Since the penetrance of the Rol phenotype in DR2270 and DR2163 was similar (ϳ50% Rol offspring), we speculate that nonsense-mediated mRNA degradation may have reduced the amount of PCR product in 2.0 kb MT relative to WT strains. No increase in RT-PCR products from any daf-4 sequences was observed in the rol-6(su1006) control strain DR2270 relative to DR1359.

Effects of Mutations in daf-4 on Gene Expression and Dauer
Phenotype-Our model predicts two antagonistic activities encoded by daf-4: a full-length DAF-4 receptor kinase that signals continuous development, and a truncated DAF-4 protein that binds DAF-7 and competitively inhibits the signaling activated by ligand-bound DAF-4 and DAF-1. Since mutations in the sequence of the extracellular domain would affect the function of both full-length receptor and antagonist DAF-4, the phenotype of extracellular domain mutants could be significantly different from intracellular domain mutants, which would express wild-type 2.0-but mutant 2.9 kb transcripts. We analyzed the genotype and phenotype of numerous daf-4 mutants for evidence supporting or disputing the antagonist hypothesis. The sites of lesions in nine daf-4 alleles were located using chemical mismatch detection (13), and then sequenced to identify each mutation. Notably, seven of the eight daf-4 mutations (one was the same in 2 alleles) are located within the kinase domain sequence (Fig. 5A), and hence are unique to the 2.9 kb mRNA. These include four missense mutations, one splice site mutation and one deletion. Among the missense mutants, we predict that substitutions in m44 (L393F) and b11/m326 (E487K) result in a complete loss of kinase activity for two reasons: (1) These mutations substitute amino acids conserved in other kinase domains with residues of very different hydrophobicity or charge (2). The severity of each mutant's Daf-c phenotype (Fig. 5B) is similar to the previously identified nonsense mutant, m72 (10,11), also depicted in Fig. 5. Although the amino acid substitution in m76 (G509R) is at a residue conserved in other kinases, the mutant displays a less severe phenotype than m72, and therefore is likely to maintain some kinase activity. The m580/m592 mutation is a conservative amino acid substitution (R560K) in a region of the kinase motif that is not highly conserved. Since m580/m592 animals are the only daf-4 mutants that are temperature-sensitive not only for dauer larva formation, but also for the small body size phenotype (37), this mutant kinase is likely to be thermosensitive.
Although the m46 intron 9 3Ј-splice site has a single base change, the mutant still expresses nearly normal levels of 2.9 kb mRNA (Fig. 5C) and a moderate Daf-c phenotype, suggesting continued splicing at or near the mutant site. A 7-bp deletion in m78 shifts the reading frame, resulting in a nonsense codon within the kinase domain that also confers a phenotype similar to m72. In daf-4(e1364), the only allele with a mutation in the extracellular domain (13) Fig. 2). The resulting gene expresses three RNAs (Fig. 5C): a 1.8 kb transcript that is missing 212 nt of daf-4 and is polyadenylated in intron 5, a 2.9 kb product that lacks 31 nt of exon 5, but is spliced to exon 6 at a cryptic 5Ј-site 48 nt downstream of the deletion (shaded in Fig. 2), and a 4.2 kb transcript that contains all 11 exons and intron 5, minus the 212 nt deletion. The reading frame of all three transcripts terminates at a UGA codon 42 nt downstream from the deletion, which results in replacement of the C-terminal 12 amino acids of the 2.0 kb putative translation product with 14 amino acids encoded by intron 5 (italicized amino acids in Fig. 2). Because none of the daf-4(e1364) translation products encode a kinase, any activity of the DAF-1 receptor is independent of DAF-4 transphosphorylation in this strain.
Interpretation of the phenotypic affect of each mutation is confounded by the maintenance of full-length 2.9 kb transcripts in all strains. Northern analysis of RNA isolated from kinasedomain mutants show that none of these mutations signal rapid decay of their respective 2.9 kb mutant transcripts (Fig.  5C). Translation of the mutant mRNAs would generate DAF-4 transmembrane proteins that still can bind extracellular DAF-7, but have defects in phosphorylation because of amino acids changes or premature termination in the kinase domain. Therefore, the Daf-c phenotype of these strains may be enhanced not only by the antagonist encoded by 2.0 kb daf-4 RNA, but also by full-length DAF-4 mutant proteins (i.e. no kinase activity). The relatively moderate phenotype of e1364 relative to kinase-domain mutants may be due to an absence of any full-length DAF-4 mutant proteins, and/or to antagonism of DAF-1 signaling by DAF-4 translated from the mutant 2.0 kb mRNA. Without definitive and complete deletion of the 2.0 and 2.9 kb daf-4 coding sequences, the individual contributions of the full-length and truncated DAF-4 polypeptides on dauer formation cannot be measured.
Developmental Regulation of daf-4 mRNA Abundance-Signals from the full-length DAF-4/DAF-1 receptor contribute to activation of non-dauer pathways of gene regulation when food is plentiful and animals are not overcrowded. One key factor modulating the signaling activity of DAF-4 and DAF-1 during  4. Overexpression of daf-4 2.0 kb RNA in transgenic animals. In the genome of the parent strain DR1359 daf-4(m592), DR2163 expresses wild-type 2.0 kb daf-4 and rol-6 (gray shading), DR2333 expresses mutant 2.0 kb daf-4 and rol-6 (black-filled) and DR2270 expresses rol-6 alone (no shading). 100 ng of total RNA from these transgenic lines were analyzed by real-time RT-PCR with three different daf-4 sequence probes (exons 4/5, intron 5, exon 8) to estimate the copy number of daf-4 2.0 kb and 2.9 kb mRNAs. The relative abundance of the 2.0 kb to the 2.9 kb daf-4 mRNAs in the transgenic lines, relative to the parent strain, was calculated by division of the transgenic strain copy number by the parent strain copy number. All values are the mean of three reactions, with S.D. depicted by the error bars. development is the steady-state level of daf-7 mRNA, which is highest in L1 larvae, and negligible in dauer larvae (2). Changes in the steady-state level of either 2.0 or 2.9 kb daf-4 messages, which could alter the ratio of full-length receptor to antagonist, may also affect the intracellular signaling activity of DAF-4 and DAF-1. We compared the steady-state levels of 2.9 to 2.0 kb daf-4 mRNAs at individual stages of larval development, including dauer and post-dauer, by Northern analysis of poly(A) RNA isolated from synchronized populations of N2 worms. In a separate experiment, total RNA from synchronized populations was analyzed for changes in the expression levels of the two daf-4 transcripts using real-time RT-PCR.
Northern hybridization signals from daf-4 2.0 kb and 2.9 kb RNAs (Fig. 6A) were quantified using a phosphorimager, and then evaluated by the ratio of 2.0 to 2.9 kb signal for each sample. The 2.0 kb transcript was slightly more abundant than the 2.9 kb during all stages grown in abundant food, with a ratio between 1.2 and 1.9 in L2-L4 larvae and adults (AD). In cultures where food was scarce, and animals were crowded, developing larvae entered and remained in dauer diapause (DA), and the ratio of 2.0 to 2.9 kb transcript signals increased to 5.0. Dauer larvae transferred to fresh medium with plentiful food resumed growth through post-dauer stages PD1 and PD2, and the 2.0 to 2.9 kb ratio decreased correspondingly to levels in well fed populations.
To more accurately measure the change in steady-state RNA levels, daf-4 total RNA was reverse transcribed and the resulting cDNAs were analyzed using real-time PCR probes specific for sequences in the 2.0 kb (intron 5) or the 2.9 kb (exon 8) RNA. A standard curve derived from real-time PCR data on known concentrations of a plasmid encoding the entire daf-4 gene was used to estimate the copy number of each daf-4 reverse transcription product. The quantity of PCR product in the L1 stage was set to a baseline value of 1.0, and we calculated changes in the copy number at all other stages relative to L1 (Fig. 6B). In favorable growth conditions, copy number of the 2.0 kb product decreased slightly as animals developed through larval stages, but returned to L1 levels in the adult. Copy number of the 2.9 kb product also demonstrated minor deviations from L1 levels. In dauer larvae, there is a 2-fold increase in full-length daf-4, and a 13-fold increase in the 2.0 kb transcript. Post-dauer animals (PD1, PD2) that recovered from diapause in plentiful food show a reduction in the amount of both transcripts to near-L1 levels.
Although the real-time PCR and Northern blot data for daf-4 expression cannot be compared (one assay measures the relative abundance of 2.0 to 2.9 kb RNAs within a single developmental stage, and the other measures expression of either 2.0 kb or 2.9 kb RNAs between different stages), they are similar in that they share the same pattern in response to environmental change: In dauer larvae, steady-state levels of the smaller transcript increase, and ultimately exceed levels of the larger transcript. A decline in the 2.0 kb steady-state RNA level to near equal the 2.9 kb corresponds with recovery from the dauer state. Although the mechanism that drives these changes in transcript abundance is not addressed with these data, they clearly demonstrate a significant correlation between the relative abundance of the two daf-4 transcripts and growth state, suggesting that environmental conditions that lead to dauer formation also favor increased steady-state levels of the 2.0 kb daf-4 mRNA. DISCUSSION We have identified a 2032-nt daf-4 transcript that encodes only the extracellular domain of DAF-4 and terminates at an alternative polyadenylation site in intron 5. This mRNA is expressed at levels similar to the full-length 2.9 kb daf-4 transcript in well fed worms, but is not represented in numerous cDNA libraries, perhaps due to RNA secondary structure in intron 5 that promotes termination of reverse transcription. The open reading frame encodes a putative polypeptide that includes the first 204 amino acids of the 227-amino acid DAF-4 extracellular domain, and is terminated by two novel amino acids and a translation stop codon encoded in intron 5. Overexpression of a transgene encoding only the 2.0 kb transcript enhances dauer formation in daf-4 and daf-1 hypomorphic mutants, whereas the same transgene with a nonsense mutation overexpresses RNA that has no effect of dauer formation.
Alternative Polyadenylation of daf-4 in Caenorhabditis Species Evolution-A comparison of the C. elegans and C. briggsae daf-4 gene sequences demonstrates that C. briggsae does not share the same juxtamembrane sequence encoded by C. elegans exon 6 or the intron V alternative polyadenylation site, leading us to predict that a truncated DAF-4 receptor would not be present in C. briggsae. The origin of this difference between C. elegans and C. briggsae daf-4 genes could be either a sequence insertion in ancestral C. elegans, or a loss of intron/exon sequences in C. briggsae after the divergence between these species, estimated at 80 -110 million years ago (36). Based on their own analyses, as well as findings from previous comparisons between genome sequences of Caenorhabditis species, Kiontke et al. (38) suggest that "species-specific introns in C. elegans and C. briggsae genomes were more likely to have resulted from intron loss in one or the other ancestral lineage than from species-specific intron gain." Despite the reported high frequency of intron loss in the Caenorhabditis group, we hypothesize that C. elegans exon 6 and the flanking introns are the result of a species-specific insertion. No other type II receptor in the superfamily of TGF-␤, activin, BMP, and DPP receptors in mammals, frogs, or fruit flies maintains a juxtamembrane sequence (between the Cys-box and the transmembrane domain) with length comparable to C. elegans DAF-4 (see Supplemental figure). In future phylogenetic analyses, the presence or absence of exon 6 and/or alternative polyadenylation in intron V may serve as a useful evolutionary benchmark in refining ancestral relationships between Caenorhabditis species.
The 2.0 kb Open Reading Frame Encodes a DAF-4 Antagonist-We hypothesize that the open reading frame encoded by the 2.0 kb RNA produces a weak antagonist of the DAF-7 signaling pathway, and propose that the primary mechanism underlying its action is competition with full-length DAF-4 and DAF-1 for DAF-7 binding. The effectiveness of the putative protein as a competitor depends on its concentration relative to DAF-7 and to full-length receptors. Developmental changes in the steady-state RNA levels for two (daf-7 and 2.0 kb daf-4) of these three components suggest that the competitive effect of truncated DAF-4 would be strongest in environments that promote dauer larva formation.
In a favorable growth environment (Fig. 7A), the daf-7 gene is actively transcribed and the ligand is secreted from amphid neurons in quantities sufficient to stimulate continuous growth (2) via cells that express DAF-4 and DAF-1, including many neurons in the head, ventral-cord and tail (13). As shown in Northern blot hybridizations, 2.9 kb daf-4 RNA is present in amounts similar to the 2.0 kb during continuous growth. In this scenario, continuous development occurs because there is a sufficient quantity of DAF-7 available to occupy full-length DAF-4 receptors, which in turn associate with and activate the DAF-1 receptor kinase that phosphorylates the SMAD transcription factors (14 -16). Also depicted is the weak kinase activity of ligand-bound DAF-1 in the absence of DAF-4 transphosphorylation (13).
As environmental conditions change to favor the formation of dauer larvae (i.e. crowding), there is a decrease in the steadystate level of daf-7 RNA (2) that is likely to restrict levels of DAF-7 translation product in dauer larvae (Fig. 7B). Concomitantly, steady-state levels of the 2.0 kb daf-4 transcript rise more than 10-fold, which may increase the concentration of the putative truncated DAF-4 protein. In this scenario, we predict that DAF-7 ligand has a much greater likelihood of binding to the truncated DAF-4 protein, and therefore not participating in the formation of DAF-4/DAF-1 receptor complexes. Higher levels of truncated DAF-4 protein may contribute to silencing of the signaling pathway required for continuous growth, and possibly prevent exit from the dauer stage in response to a transient DAF-7 signal. Regulation of the expression of this antagonist provides an additional level of fine-tuning to the FIG. 6. Changes in daf-4 transcript abundances during development. A, 1 g of poly(A) RNA samples from developmentally synchronized populations of N2 worms were blotted and then hybridized with an exon 1-6 cDNA probe (L, larval; AD, adult; DA, dauer; PD, post-dauer). In the panel below, the signal strength (quantified using a Fuji phosphorimager) of the 2.9 kb band for each developmental stage was adjusted to equal 1 unit on the y-axis, and the intensity of the 2.0 kb band is shown as a multiple of this unit value (filled bar). B, 100-ng isolates of total RNA from developmentally synchronized N2 populations were used for real-time RT-PCR with primer sets specific for daf-4 2.0 kb mRNA (shaded bars, intron V probe) and 2.9 kb mRNA (filled bars, exon 8 probe). Template copy number for each sample was calculated by linear regression from a DNA template real-time PCR standard curve created for each probe. Changes in transcript abundance during development are depicted by calculating the ratio of PCR product produced from each primer set in the L1 stage relative to every other stage (i.e. a y-axis value of 5 would indicate that the template copy number of the tested sample is 5-fold higher than the L1 copy number). All values are the mean and S.D. of triplicate RT-PCR reactions from 2 sets (8 stages/set) of template RNAs prepared from unique populations of synchronized worms. dauer/non-dauer developmental switch, a decision that is critical to the reproductive success of C. elegans. We speculate that this additional level of regulation in C. elegans may contribute to species differences in the propensity for arrest and dispersal in a given set of environmental conditions. The Daf-c phenotype of these daf-4 alleles is attributable to the sum of three different effects on DAF-1 signaling: (1) loss of DAF-1 activation due to mutation or deletion of the DAF-4 kinase, (2) antagonism of DAF-4-independent DAF-1 signaling by DAF-7 binding to the truncated DAF-4 protein encoded by the wild-type 2.0 kb mRNA, and, in the case of kinase domain mutants (3) competitive inhibition of DAF-1 by interaction with DAF-4 mutant receptors that lack kinase activity but maintain ligand-binding activity (encoded by the 2.9 kb transcript).
Negative effects on signaling have been observed with other TGF-␤ receptors that lack kinase activity, either through liganddependent or ligand-independent mechanisms (23, 39 -41). Ligand binding and signaling activity of the full-length DAF-4 receptor, as well as the antagonistic activity of the putative truncated protein, should be absent from a daf-4-null allele. We predict that the dauer-constitutive phenotype of a null allele would be less severe than alleles in which the kinase activity is eliminated but the 2.0 kb-encoded antagonist remains. Experimental tests of this hypothesis await development of daf-4 deletion mutants that completely eliminate expression of both daf-4 mRNA isoforms.
Factors Establishing the Relative Abundance of 2.0 kb daf-4 mRNA-The steady-state level of the daf-4 2.0 kb mRNA is determined by a combination of factors that minimally include: an absence of intron 5 splicing, cleavage, and polyadenylation within intron 5, and stability of the resulting RNA. The biochemical processes of splicing and 3Ј-end formation are tightly linked to transcription, as well as each influencing the specificity and efficiency of the other (43). Many of the components involved in splicing and polyadenylation of RNA polymerase II (Pol II) transcripts have been identified and cloned, and are conserved between yeast and mammalian cells. Recent experiments suggest that functional coupling of splicing and cleavage/polyadenylation to transcription is achieved by proteinprotein interactions between RNA processing factors and components of the Pol II transcriptional apparatus (44). One of the key elements in the coupling of these events is the Cterminal domain of the Pol II large subunit, which when deleted, abrogates both splicing and 3Ј-end processing (45). Although less well studied, the cis-and trans-acting components of 3Ј-end processing identified in nematodes appear to follow the eukaryotic paradigm (32).
Exon 5 of daf-4 is a "composite" exon; one that is followed by an intron with a poly(A) site and that can behave either as a 3Ј-terminus or an internal exon depending on the occurrence or absence of splicing (46). Studies of eukaryotic transcription units with an alternative poly(A) site located within a gene indicate that polyadenylation at the promoter-proximal site competes with splicing to influence the expression of multiple RNAs, and that poly(A) site recognition can be inhibited by the proximity of an upstream 5Ј-splice site. In mammals, processing of composite exons in the RNA from each of the five classes of immunoglobulin heavy chain genes leads to differential expression of membrane-bound versus secreted Ig proteins during B-lymphocyte differentiation (47). In the IgM heavy chain gene, the internal poly(A) site is repressed by the upstream 5Ј-splice site in early developing B-cells, resulting in production of the membrane form of this protein. In plasma cells, processing of the upstream poly(A) site is more efficient, and outcompetes the splicing reaction, resulting in synthesis of the secreted form. Other examples of composite exon processing include the epidermal growth factor (48) and the fibroblast growth factor (49) receptor-tyrosine kinases, which also produce secreted proteins that interfere with the activity of the membrane-bound receptors. In C. elegans, a small isoform of the basement membrane protein UNC-52/perlecan is encoded FIG. 7. A model for regulation of signaling by a truncated DAF-4 protein. A, in favorable growth conditions, DAF-7 is available for binding to full-length and truncated forms of DAF-4, as well as to the DAF-1 type I receptor alone. DAF-1 in the absence of DAF-4 maintains some kinase activity (dashed arrow), but phosphorylation by DAF-4 enhances signaling activity (bold arrow). The putative targets for this receptor kinase are SMAD proteins that promote non-dauer development. B, in conditions leading to dauer larva development, high concentrations of pheromone negatively regulate daf-7 transcription, resulting in a decline in the concentration of secreted DAF-7 ligand. Available ligand is bound by the truncated secreted DAF-4 that is now more abundant, whereas full-length DAF-4 and DAF-1 remain unbound. In the absence of ligand, the DAF-1 kinase is not activated by DAF-4 phosphorylation. The phosphorylation-independent activation of SMADs by DAF-1 may also be hampered by non-productive interactions with the truncated DAF-4 protein.
by a 4.0 kb mRNA that is polyadenylated in intron 10, omitting 26 exons from the full-length gene, and encoding 32 C-terminal amino acids that are not found in the other gene products (42).
Since polyadenylation and splicing of composite exons are mutually exclusive processes competing for the same nascent RNA, changes in the activities of polyadenylation and/or splicing factors that occur during development, or in different cellular environments, are likely to play an important role in the differential expression of each gene product. In the case of daf-4, it is possible that the environmental factors regulating dauer larva formation (nutrient availability, pheromone concentration, temperature) may also affect the activities of RNA processing components, or the efficiency of these processes, which could lead to higher efficiency of the daf-4 intron V cleavage/polyadenylation site in stressful environments. In this context, alternative processing of daf-4 will function as a particularly useful genetic system for investigating the interactions between Pol II transcription, splicing, and polyadenylation processing events throughout C. elegans development.