Regulation of Caenorhabditis elegans body size and male tail development by the novel gene lon-8

Background In C. elegans and other nematode species, body size is determined by the composition of the extracellular cuticle as well as by the nuclear DNA content of the underlying hypodermis. Mutants that are defective in these processes can exhibit either a short or a long body size phenotype. Several mutations that give a long body size (Lon) phenotype have been characterized and found to be regulated by the DBL-1/TGF-β pathway, that controls post-embryonic growth and male tail development. Results Here we characterize a novel gene affecting body size. lon-8 encodes a secreted product of the hypodermis that is highly conserved in Rhabditid nematodes. lon-8 regulates larval elongation as well as male tail development. In both processes, lon-8 appears to function independently of the Sma/Mab pathway. Rather, lon-8 genetically interacts with dpy-11 and dpy-18, which encode cuticle collagen modifying enzymes. Conclusion The novel gene lon-8 encodes a secreted product of the hypodermis that controls body size and male ray morphology in C. elegans. lon-8 genetically interacts with enzymes that affect the composition of the cuticle.


Background
The regulation of body size is an important yet poorly understood aspect of animal development [1]. Unlike vertebrate size, nematode body size is largely dependent on cell size rather than cell number [2][3][4]. Nevertheless, Caenorhabditis elegans is an attractive model to study the genetics of body size determination because many mutants that specifically affect size have been isolated and characterized. In C. elegans, the hypodermis, which is the epithelial structure that encapsulates the worm, is the primary site where growth is regulated. The hypodermis consists of several syncytia of fused epidermal cells [5]. The largest of these syncytia that covers the entire mid-body region of the animal is called hyp7. During larval development, hyp7 increases in nuclear content and cell size as progressively more daughters of the lateral seam cells fuse with it [4,5]. The syncytial nuclei undergo several rounds of endoreduplication during the fourth larval stage and adulthood [2,6]. In C. elegans and other nematode species, a substantial part of growth takes place after larval development is completed [2]. The amount of post-larval growth is correlated with the amount of endoreduplication that occurs in the hypodermis during adulthood [2,7]. Therefore, growth is at least partially dependent on the DNA content of the hypodermis [2,4,7].
Another factor that affects body size in C. elegans is the composition of the exoskeletal cuticle, which is synthesized by the hypodermis before hatching and at the end of each larval stage (reviewed in [8,9]). Mutations in cuticle components can result in dumpy (Dpy) or squat (Sqt) phenotypes, meaning animals are shorter and thicker than wild-type [10]. Many of these dpy and sqt genes encode collagens or the enzymes required to process them [8,9]. There are also mutations that affect body size without affecting morphology. In these cases the animals' size is altered but their proportions are more or less intact, resulting in small (Sma) or long (Lon) phenotypes [10]. So far, only two lon genes have been cloned and characterized. Both genes encode molecules that are synthesized and secreted by the hypodermis. lon-3 encodes a cuticle collagen that is required for body size regulation primarily during late larval development [11,12]. lon-1 encodes a member of the CRISP (Cysteine-Rich Secretory Protein) family of unknown function [13,14]. lon-1 animals are of wild-type length at hatching, but grow substantially faster than wild-type during both larval and adult stages. Part of the increase in body size of lon-1 animals is proposed to be due to an increase in the ploidy of the hypodermal nuclei [14].
In C. elegans, growth is controlled by a highly conserved TGF-β pathway, the Sma/Mab (for Small/Male abnormal) pathway (reviewed in [15]). Mutants of components of this pathway are smaller than wild-type, but not obviously dumpy. The ligand, DBL-1, is a dose dependent regulator of body size. dbl-1 loss-of-function causes a Sma phenotype, whereas overexpression causes a Lon phenotype [16,17]. Mutants of dbl-1 or positive downstream regulators of the Sma/Mab pathway are of wild-type length at hatching, but by the time they reach adulthood they are about a third shorter [16][17][18][19][20]. In addition, adult growth and endoreduplication of the hypodermal nuclei is reduced in dbl-1 animals [7]. Therefore, the Sma/Mab pathway is required for larval elongation as well as adult growth.
The effects of dbl-1 signaling are mediated in part through the regulation of lon genes. dbl-1 is thought to regulate translation, modification or degradation of the LON-3 protein [12], whereas lon-1 is a negatively regulated transcriptional target of DBL-1 [13,14]. In addition, mutations in lon-3 and lon-1 can suppress the small phenotype of Sma/Mab pathway mutants [11][12][13][14]. Here, we identify a novel regulator of body size, lon-8, which is also required for male tail morphology. lon-8 encodes a novel secreted protein that is highly conserved within the phylum Nematoda. Our results suggest that LON-8 regulates body size independently of the Sma/Mab pathway.

lon-8 encodes a small protein that is likely to be secreted
We found that Y59A8B.20 RNAi by feeding causes an increase in body length (1.3 ± 0.08 mm in animals treated with control vector versus 1.4 ± 0.08 mm in animals treated with lon-8 RNAi, n = 75), as previously observed by others [21]. Upon amplification and sequencing of the Y59A8B.20 transcript from a C. elegans cDNA library, it became apparent that the sequence of this transcript differs from the predicted Y59A8B.20 gene structure in that it contains an additional final fourth exon ( Figure 1). The complete Y59A8B.20 sequence (Genbank accession number EF495354) encodes a 162 amino acid protein containing an N-terminal signal sequence. There are no other conserved motifs such as transmembrane domains or signals for lipid mediated anchoring in the sequence. Y59A8B.20 is strongly expressed in the hypodermal syncytium (see below). Clear localization to this cell is only visible when the signal sequence is deleted, implying that in the presence of this sequence Y59A8B.20 is secreted and diffuses. Y59A8B.20 is therefore likely to be a secreted protein.
To investigate the loss-of-function phenotype further, we identified and isolated two deletions of Y59A8B.20 (see Methods). Both alleles gave a clear Lon phenotype, therefore Y59A8B.20 was named lon-8. The first allele generated, hu187, is a deletion of the entire lon-8 first exon and surrounding sequences. A shorter transcript can still be detected in hu187 animals by Northern blotting ( Figure  1). We amplified this fragment by reverse transcription PCR (RT-PCR, see Methods) and sequenced it. The alternative transcript found in hu187 mutants is predicted to encode an N-terminal truncation of the full-length LON-8 protein starting from an in-frame methionine present in the second exon. This transcript cannot be detected in wild-type animals by Northern blotting or RT-PCR. Importantly, if the hu187 transcript were to be translated, the resulting protein would lack the signal peptide. Furthermore, treatment of hu187 mutants with lon-8 RNAi did not enhance the Lon phenotype (1.5 ± 0.08 mm in hu187 animals on control vector versus 1.4 ± 0.08 mm in hu187 animals treated with lon-8 RNAi, n = 75). Therefore, it seems unlikely that this shorter version of LON-8 is functional. The second deletion allele generated, hu188, lacks the entire second exon and surrounding sequences ( Figure 1). Alternative splicing of lon-8 from the first to the third exon would result in a frame-shift leading to a nonsense transcript, moreover, no transcript can be detected in these animals by Northern blotting. Both alleles show similar phenotypes in the experiments described below. Taken together, these results strongly suggest that both alleles represent a strong loss-of-function or null phenotype.

lon-8 loss-of-function affects body size during larval development and early adulthood
To determine at which developmental stages body length is affected in lon-8 mutants, we synchronized animals by collecting them an hour after hatching and letting them grow for different periods of time before measuring their length ( Figure 2). We compared the lon-8(hu187) phenotype with that of lon-1(e185) animals, which grow more than wild-type animals during both larval development and adult growth [13,14]. At hatching, lon-8(hu187) mutants were no longer than wild-type animals, but throughout larval development, they outgrew their wildtype counterparts at a rate comparable to that of lon-1(e185) animals. At around 96 hours post-hatching, the growth rate of lon-8(hu187) slowed down compared to lon-1 animals, which kept growing at the same rate. This Lon phenotype resulting from mutations in lon-8 Figure 1 Lon phenotype resulting from mutations in lon-8. (A) N2 wild-type hermaphrodite grown at 20°C for 120 hours. Bar is 200 µm. (B) lon-8(hu187) hermaphrodite grown at 20°C for 120 hours. (C) Structure of lon-8 gene. SL1 is the SL1 trans-splice leader sequence. The region which encodes the signal peptide is shaded. The lon-8(hu187) and (hu188) deletions are indicated. (D) Northern blot analysis; RNA was isolated from mixed-stage populations of animals. Blots were probed with the amplified full-length sequence of the lon-8 cDNA. eft-2 was used as a control for loading [13]. Animals containing hu187 express a transcript which is recognized by the lon-8 probe, but is shorter than the wild-type transcript. suggests that lon-8 is more important for body size determination during larval elongation and early adult growth than later in life.
As lon-8 acts at stages in development when there is cell proliferation in the hypodermis, it could be influencing the number of seam cells or the number of syncytial nuclei, which could lead to an increased body size [2,4,7]. We stained N2, hu187 and hu188 animals with DAPI and counted the number of hypodermal and seam cell nuclei located between the postdeirid sensillum and the anus to determine whether the number of hypodermal nuclei is elevated in lon-8 mutants compared to wild-type. We did not observe a difference in the number of hypodermal nuclei (in each case, 21 ± 1 nuclei, n = 20 for each strain). We also investigated whether the total number of seam cells is affected by crossing the seam cell specific GFP marker wIs51 [scm::gfp] [22] into the hu187 background and counted the total number of seam cells lining the flank of the worm. Again, we observed no difference in seam cell number between wild-type and lon-8(hu187) animals (16-17 seam cell nuclei, n = 32 animals for each strain). This suggests that the increase in body length of lon-8 animals correlates with an increase in cell size rather than cell number.
Adult growth in C. elegans is thought to be regulated by the number of endoreduplication rounds completed by the nuclei of the hypodermal syncytium [2,7]. In lon-1 mutants, some of the hypodermal nuclei undergo additional rounds of endoreduplication, leading to an increased ploidy compared to wild-type, and this is thought to contribute to their excessive growth [11,14]. In contrast, lon-3 mutants have no detectable increase in hypodermal ploidy [11]. To see if adult growth by hypodermal endoreduplication is affected in lon-8 mutants, we measured the DNA content of individual hypodermal nuclei in the posterior lateral hypodermis of synchronized populations of 120 hour old animals using DAPI staining and microdensitometry [2,23] as well as PI staining and confocal analysis [23]. We were unable to detect a difference in hypodermal ploidy between wild-type and lon-8 animals using either of these techniques (a ploidy of 7.9 ± 1.6 for N2 and 8.2 ± 2.4 for lon-8(hu188) when analyzing 10 animals using PI staining, see Methods for details). Taken together, these results lead us to conclude that lon-8 does not have an obvious effect on the nuclear content of the hypodermis.

lon-8 is expressed in the hypodermis throughout development
We also investigated where lon-8 function might be required for its role in body size determination. To this end, a transcriptional reporter was constructed, consisting of the lon-8 upstream sequence and part of the genomic sequence ( Figure 3A; see Methods), which was inserted into a vector containing the coding sequence for nuclear localized GFP. Because the LON-8 N-terminus contains a signal peptide, which could cause the secretion and diffusion of GFP, another variant was made excluding the signal sequence. Multiple lines carrying extra-chromosomal arrays of these constructs were generated by microinjection. The lines carrying the variant of the reporter construct including the signal peptide mostly showed weak GFP expression in the posterior gut (data not shown) but nowhere else, presumably because the GFP is secreted and diffuses. However, the lines carrying the variant lacking the signal peptide showed clear expression in two of the hypodermal syncytia, hyp4, which surrounds the anterior part of the pharynx, and hyp7, which surrounds the midbody region of the animal (Figure 3). The earliest detectable GFP expression can be observed in comma stage embryos, where there is abundant expression in the newly generated hyp4 nuclei and, albeit less frequently, in hyp7 nuclei. This expression pattern remains consistent throughout embryonic and larval development up until adulthood. There is faint expression in the hindgut, which Growth curve of lon-8(hu187) mutants Figure 2 Growth curve of lon-8(hu187) mutants. Wild type, lon-1(e185) and lon-8(hu187) animals were synchronized and measured at seven time points. The animals are of indistinguishable length at hatching, but lon-1 and lon-8 animals grow faster than wild type animals throughout larval development and during early adulthood. Later in adult life, lon-1(e185) animals continue to grow at a faster rate (steeper slope angle) than wild type animals, whereas the growth rate of lon-8(hu187) animals stabilizes. At least 100 animals were measured for each strain. a, b, c, for wild type/lon-1 and wild type/ lon-8 p < 0.001. d, e for wild type/lon-1, wild type/lon-8 and lon-1/lon-8 p < 0.001. P values were calculated using Student's t-test.

lon-8 cannot suppress the Sma phenotype of dbl-1 mutants
Both lon-1 and lon-3 can suppress the small phenotype of Sma/Mab mutants [11,13,14]. Several lon-1 alleles even suppress the Sma phenotype so strongly that double mutant combinations can become longer than wild-type [13,14]. To determine whether lon-8 mutations can likewise suppress the phenotype of Sma mutants, we generated double mutant combinations of hu187 and hu188 with dbl-1, and of hu187 with sma-6, which encodes the Type I receptor for DBL-1 [18] (Figure 4). If lon-8 were an important downstream effector of dbl-1, one would expect that double mutants of lon-8 with dbl-1(nk3) or sma-6(e1482) would reach a body length longer than or equal to wild-type. Although lon-8 has an effect on the maximum length that dbl-1 animals can reach, there is no clear suppression of the Sma phenotype in dbl-1(nk3) lon-8 animals. In double mutants between sma-6(e1482) and lon-8(hu187), there is a slightly stronger suppression of the Sma phenotype than in dbl-1(nk3) lon-8(hu187) animals, which could be due to the fact that e1482 is a hypomorphic allele of sma-6. In any case, both dbl-1(nk3) lon-8(hu187) and sma-6(e1482); lon-8(hu187) double mutants still show a strong and penetrant Sma phenotype. Taken together, these data indicate that lon-8 is unlikely to be an important effector of DBL-1 signaling. We also analyzed the body size phenotype of double mutant combinations of lon-8 and three other lon genes; lon-2, which functions upstream of dbl-1 in the Sma/Mab pathway [24], lon-1 and lon-3. In all cases, lon-8(hu187) enhanced the average single mutant body size to a similar degree, which is around 10%.

lon-8 is not transcriptionally regulated by the Sma/Mab pathway
Two sets of experiments were carried out to further study the relationship between lon-8 and the Sma/Mab pathway. Firstly, we examined the expression of the lon-8 transcript in several Sma/Mab pathway mutants by Northern blotting ( Figure 5). Others have shown that the lon-1 transcript is clearly upregulated in the Sma/Mab pathway mutants dbl-1 and sma-6, whereas it is downregulated in animals overexpressing dbl-1, thus establishing that lon-1 is a negatively regulated transcriptional target of dbl-1 signaling [13,14]. Whereas the lon-1 transcript is clearly upregulated in dbl-1 and sma-6, the expression level of the lon-8 transcript is similar to the wild-type level in these mutants. In addition, we crossed the extrachromosomal arrays of our transcriptional reporter construct without the signal peptide into dbl-1(nk3). As some of our lines were generated using a red fluorescent injection marker expressed in the pharynx (see Methods), we could use Expression pattern of a lon-8 transcriptional reporter these extrachromosomal arrays to compare the proportion of transgenic animals (with a red pharynx) that also expressed GFP in hyp4 and/or hyp7 in a wild type versus a dbl-1 background. We found no difference in the number of transgenic animals that expressed GFP under control of the lon-8 promoter between wild-type and dbl-1 animals (n > 100), nor could we detect any difference in GFP intensity by eye (data not shown). Taken together, these results imply that lon-8 is not transcriptionally regulated by the Sma/Mab pathway.

lon-8 is required for male ray development
C. elegans males possess a copulatory tail consisting of nine bilaterally symmetric pairs of sensory structures called rays and two sharp, hardened structures called spicules [25]. The ray pairs have distinct identities with respect to morphology and chemosensitivity. As the name of the pathway suggests, Sma/Mab pathway mutants show extensive defects in ray specification. [16][17][18][19]26]. In addition, all Sma/Mab pathway mutants have crumpled spicules. Male mutant dbl-1, daf-4, sma-2, sma-3, sma-4 and sma-6 animals exhibit dorsal to ventral transformations of rays 5, 7 and 9, which take on the fate of their anterior neighbors 4, 6, and 8 and often fuse with them [16][17][18][19]26]. dbl-1 overexpression has the opposite effect, dorsalizing ray 4 which can lead to fusions of rays 3 and 4 [17]. If lon-8 is acting in parallel to the Sma/Mab pathway in body size regulation, does it also affect male tail development? To address this question, we analyzed the morphology of the male rays using a him-5(e1490) mutant background. We found that lon-8(hu187) and lon-8(hu188) mutants have extensive morphological ray defects, but we did not observe any defects of the spicules. All animals examined have similar ray defects but the severity of the phenotype is variable: in some animals the rays appear grossly thickened but can still be identified. In these cases, the correct positional identity of the rays does not seem to be affected. In most males, the rays have such an abnormal appearance that they can no longer readily be distinguished ( Figure 6). All rays appear morphologically abnormal, but rays 5 and 6 are most severely affected, whereas rays 8 and 9 are less deformed than their more anterior neighbors. Surprisingly, lon-8(hu187) and lon-8(hu188) males can mate, suggesting that ray sensory function is intact even though ray morphology is grossly abnormal. This phenotype is most reminiscent of the phenotype of mutants in a class of 8 genes that give a Ray Abnormal Morphology (Ram) phenotype [27]. These genes act after the cell lineage is complete to regulate the apposition of the cuticle around the rays. Although mutations in the ram genes result in an abnormal, lumpy ray morphology, mutant animals appear otherwise normal and can mate. Finally, we examined the expression of the lon-8::gfp transcriptional reporter described above in him-5 animals ( Figure 6). In males, the reporter is expressed in hyp4 and hyp7 like in hermaphrodites. In addition, GFP is expressed in the nuclei V6.pppaa, T.aa, R6.p, T.apapa, R7.p, R8.p and R9.p. These cells fuse and make up the ventral part of the male tail hypodermis [25]. In conclusion, LON-8 is expressed in hypodermal tissues that surround the developing rays.

Functional interaction between lon-8 and the Ram genes dpy-11 and dpy-18
In addition to their role in male tail development, some of the mutations that cause a Ram phenotype also have a significant effect on body size, causing a dumpy (Dpy) phenotype [27]. Two of the genes affected by these mutations, dpy-11 and dpy-18, encode enzymes required for the posttranslational modification of cuticle collagens [28][29][30]. As mutations in dpy-11 and dpy-18 cause a male tail phenotype similar to lon-8, we investigated whether they might interact with lon-8 in body size regulation. We found that both dpy-11(RNAi) and dpy-18(RNAi) completely suppress the lon-8 body size phenotype (Figure 7), strongly suggesting that the function of these three genes are intimately linked.

LON-8 has been highly conserved throughout nematode evolution
Sequence similarity searches of the NCBI protein database and a recently established database of parasitic nematode tail ventral hypodermis expressed sequence tags (ESTs) [31], indicated that LON-8 is a nematode specific protein. Clear LON-8 homologs were found in Caenorhabditis briggsae, Haemonchus contortus, Pristionchus pacificus, Parastrongyloides trichosuri, Meloidogyne incognita and Brugia malayi, representing each of the three clades of the order Rhabditida [32,33]. When the sequences from the different nematode species are aligned, the only major divergence is in the N-terminal part of the protein (Figure 8). Despite this difference in amino acid sequence, all the putative N-termini are predicted signal peptides. Therefore, the function of this part of the LON-8 protein is conserved even though the sequence has diverged. This suggests that lon-8 has an important function in all Rhabditid nematodes.

Discussion lon-8 regulates body size independently of the Sma/Mab pathway
In this study, we describe lon-8, a novel regulator of body size. So far, all the lon genes that have been cloned show clear interactions with the Sma/Mab pathway [11][12][13][14]. Our epistatic analysis of lon-8 and other body size mutants imply that lon-8 is an exception in that it is functioning in parallel to the Sma/Mab pathway. We observed that dbl-1 suppresses the lon-8 body size phenotype, indicating that lon-8 does not function downstream of dbl-1. Furthermore, the finding that lon-8 expression is not affected in dbl-1 or sma-6 mutants suggests that it is not a transcriptional target of the DBL-1 pathway. The suppression of the lon-8 phenotype by dbl-1 could also be explained by a function of lon-8 upstream of dbl-1. Although this cannot be excluded, the body size phenotypes of double mutant combinations of lon-8 and the three different lon genes that are part of the DBL-1 pathway argue against this possibility. We found that lon-8 enhances the body length of these lon mutants by a similar proportion, an additive effect which would not be expected if they were in a linear pathway. Taken together, the evidence suggests a parallel function of LON-8 and the Sma/Mab pathway in body size regulation.

lon-8 regulates body size primarily during larval and early adult stages
The lon-8 body size phenotype is more subtle than those of the other lon genes; whereas lon-1(e185) animals are approximately 30% longer than wild-type, and lon-3 animals about 25%, lon-8 mutants have a more modest increase of 10% of the wild-type body length on average. Another distinction is that lon-8 appears to be affecting growth largely during larval development, whereas lon-1 acts throughout larval and post-larval growth (this study, [11,14]) and lon-3 acts primarily during late larval stages [11,12]. In addition, lon-1 partially exerts its effects by regulating hypodermal ploidy, which accounts for most, if not all of adult growth [2,11,14]. The increase in body size Loss of dpy-11 or dpy-18 suppress the lon-8 phenotype Figure 7 Loss of dpy-11 or dpy-18 suppress the lon-8 phenotype. Animals were grown on the indicated dsRNA expressing bacteria and were measured as described in Methods. At least 75 animals were measured for each strain. Each dot represents one animal. Average length and standard deviation are indicated. dpy-11 and dpy-18 RNAi in wild type and lon-8(hu187) backgrounds result in equal body lengths (p > 0.05). of lon-8 mutants shows no clear correlation with the nuclear content of the hypodermis. Taken together with the finding that the growth of lon-8 mutants is affected during larval development and early adulthood rather than throughout adult growth, we hypothesize that lon-8 is involved in regulating growth processes that are not dependent on hypodermal nuclear content.

The lon-8 male tail phenotype is indicative of a function for LON-8 in cuticle retraction
lon-8 is the only lon gene studied so far to affect the development of the male tail. Although the rays are severely deformed in lon-8 animals, they do not seem to exhibit cell fate alterations as Sma/Mab pathway mutants do [17][18][19]26]. In Sma/Mab mutants, the identities of specific rays are altered to resemble that of their anterior or posterior neighbor, and this leads to fusions between rays with similar fates. In lon-8 mutants, the rays appear to be deformed rather than misspecified and that is why they can no longer readily be distinguished. The spicules do not look affected and the males can mate. This phenotype is quite different from that of Sma/Mab mutants, again suggesting that lon-8 is not acting in this pathway. Rather, the lon-8 phenotype resembles the ram mutant phenotype [27]. Animals with a Ram phenotype have a defect in the extension of the rays and this leads to abnormal male tail morphology. Although the mutant rays are lumpy, their sensory function seems to be unaffected as ram males can mate. We hypothesize that, like the ram genes, lon-8 is required for the retraction of the hypodermis that accompanies the terminal differentiation of the rays.

lon-8 genetically interacts with enzymes required for the modification of cuticle components
A number of the ram genes have been found to encode components of the extracellular cuticle (reviewed in [34]). Two of the genes that give a Ram phenotype are the previously identified genes dpy-11 and dpy-18, which encode enzymes synthesized by the hypodermis required for the posttranslational modification of cuticle components [28][29][30]. Epistatic analysis shows that loss-of-function of these genes can fully suppress the lon-8 body size phenotype, indicating that dpy-11 and dpy-18 function genetically downstream of lon-8. There are many possible hypotheses as to the actual mechanism of the interaction. Taking into account that LON-8 is most likely a secreted component of the cuticle, it might for instance function to modulate the conformation or localization of targets of DPY-11 and DPY-18.

The conservation of LON-8 implies that it performs an important function in nematodes
So far, the C. elegans and C. briggsae genomes have been fully sequenced and made publicly available. In addition, over half a million ESTs have been generated from around 40 species of nematodes representing all but one clade [35], enabling extensive cross-species comparisons. Although the Rhabditid nematode species share morpho-  logical similarity, there is a great deal of genetic diversity between them. For example, C. elegans and C. briggsae, which are the most closely related of the set and are morphologically indistinguishable, are separated by an estimated 100 million years of evolution, and 10% of their genes differ [36]. Analysis of nematode ESTs showed that LON-8 is highly conserved within species of the order Rhabditida, which is one of the major nematode orders and contains free living as well as animal and plant parasitic nematodes [32]. LON-8 mediated regulation of body size may thus be an essential feature of nematode development.

Conclusion
We have identified a novel gene that controls body size and male ray morphology in C. elegans and is conserved in nematodes. Although dbl-1 signaling regulates both body size determination and male tail development, lon-8 does not appear to be an effector of the Sma/Mab pathway. The presence of a secretion signal and the fact that a lon-8 promoter fusion is expressed in the hypodermis suggest that LON-8 is a secreted component of the extracellular cuticle. The interaction between lon-8 and genes encoding cuticle collagen modifying enzymes further suggests that LON-8 may be important for cuticle structure, thereby influencing body size as well as male ray extension.

Sequencing of lon-8
To amplify the lon-8 cDNA from a yeast-two-hybrid cDNA library (a gift of M. Walhout and M. Vidal) we used a primer directed to the 5' side of the pPC86 poly-linker with the lon-8 primer CTGCAGAATTACTGACTGTT-GGGATC and a primer directed to the 3' side of pPC86 with the lon-8 primer CAGTCAGTAATTGCG-GATTTG. All amplified products were purified and sequenced from both directions to obtain the full-length lon-8 cDNA sequence.
To sequence the lon-8 transcript present in wild type and hu187 animals, we used 1 µg of total RNA from N2 and hu187 (see below) to generate cDNA by RT-PCR using an oligo-dT primer. We then amplified the lon-8 cDNA using primers for the SL1 sequence and CTGCAGATTC-CTCGCTTTCGCTG corresponding to the 3' end of the fourth exon of lon-8 and sequenced the purified products from both directions to obtain the wild-type and hu187 cDNA sequences.

Isolation and characterization of lon-8 alleles
The deletion alleles lon-8(hu187) and lon-8(hu188) were generated by constructing and screening a deletion library as described by [38]. The primers used to identify and sequence the hu187 deletion allele were TGCAATAGGTG-TACGAAAAG and CCACAACACATACATTCCAT as the external pair and GAGTGGAATGCTAATAGGGA and GTGAAGCAGCGTATCAACTA as the internal pair. The primers used to identify and sequence the hu188 deletion allele were TTTTCTTTGAGCCGCAAAAT and CAGG-TAACGCCATGAGGAAT as the external pair and CAAAG-GAGGTCGAAAGTGGA and CCATTCTGGCAGTGATCTCC as the internal pair.

Body size measurements
Strains were synchronized by gently flushing 9 cm plates of a mixed stage population to remove all adults and larvae, leaving the unhatched eggs. One hour later the plates were gently flushed to collect the larvae hatched within this interval. For the growth curve, synchronized animals were kept at 20°C and one plate of each strain was used for measurements at each time point. A minimum of 75 random animals was sampled every time. For the epistatic analysis an N2 control was taken along with each synchronization to ensure consistent results as body size may vary due to external conditions such as humidity [11]. To measure, the animals were mounted on 2% agarose slides containing 10 mM Sodium azide and immediately photographed at a magnification of 33.5 times using a Zeiss Axioplan2 microscope and attached Axiovision digital camera. Body length was determined by using Object Image 2.13 software. The data was analyzed using Graph-Pad Prism.

lon-8p::gfp fusion constructs
To generate the Plon-8::gfp fusion construct pGS25, sequence corresponding to the 2 kb upstream region up to the second exon of the predicted lon-8 transcript was amplified from N2 genomic DNA using primers (PstI)-CTTGACCGTAGAGCTTCC and (PstI)-CAAATCCGCAAT-TACTGACTG. This fragment was cloned into the pPD95.69 GFP insertion vector (A. Fire, personal communication) using the PstI cloning site. To generate the construct excluding the predicted signal peptide, pGS26, two fragments were amplified from pGS25: the cloned lon-8