All of the EMAP-like proteins identified to date, including vertebrate EMLs and invertebrate ELPs, share a common domain organization with a short, 60–70 amino acid, hydrophobic EMAP-like protein (HELP) motif preceded by a series of WD repeat domains (Figure 1). Although the function of the HELP motif is unknown, it is unique to this gene superfamily. There is evidence in the human genome for 6–8 EML genes, however at this time only EMLs 1–5 have a confirmed gene product. The apparent or predicted molecular mass of most EMAP-like proteins ranges from ~70 to ~120 kDa (Figure 1). Human EML5 is the exception with a predicted Mr of 220 kDa and three repeats of the HELP and WD domains.

The C. elegans genome contains a single EMAP-like gene, elp-1, that maps to the right of unc-51 on the extreme right arm of linkage group (LG) V (Figure 2). Genefinder™ analysis predicts that the elp-1 gene consists of 16 exons that encode an 891-amino acid polypeptide of 98 kDa. We used reverse transcription coupled to the polymerase chain reaction (RT-PCR) to show that there are at least two alternatively spliced variants: a full-length cDNA transcript composed of exons 1 through 16 (elp-1a); and an alternatively spliced transcript composed of 15 exons (elp-1b). The elp-1b transcript lacks exon 5, an 81 bp segment that encodes for a 27-amino acid sequence that is more than 46% serine and threonine residues (Figure 2C). The elp-1b transcript is identical to the cDNA EST yk209e10 that is predicted to be translated into an 864-amino acid polypeptide of ~95 kDa 17.

Exon 5 is conserved in nematodes but not detected in other EMAP-like proteins. Twenty-four of the twenty-seven predicted amino acids (88%) are identical in Caenorhabditis briggsae and C. elegans. All three substitutions are conservative (V15I, I16V, and S19N; numbers refer to the 27 amino acids coded by this exon). This domain shows a slightly higher level of sequence conservation when compared to the HELP domain protein sequence (81% identical) and the full-length protein sequence (80% identical) of C. elegans and C. briggsae. Although the function of this 27 amino acid region, rich in potential phosphorylation sites, is unknown, these observations indicate that exon 5 may be important for a nematode-specific function.

To learn more about the function of elp-1, we obtained a deletion allele, ok347, from the C. elegans Knockout Consortium. The ok347 allele deletes 1301 nucleotides, spanning intron 1 through intron 4 (Figure 2A, 2D). Starting from the ATG of the ELP-1a open reading frame, ok347 is missing base pairs 421 to 1721. This genome deletion does not remove ELP-1 function however, since elp-1(ok347) worms retain two nearly full-length transcripts (Figure 2E). These transcripts are predicted to encode two proteins that include both the HELP and WD repeat domains, a finding which implies that these proteins could retain partial function.

To learn how ELP-1 contributes to development, physiology and behaviour, we examined the expression pattern of the elp-1 gene. Three GFP reporter constructs were generated: an ELP-1 promoter construct [pKA99-3], a truncated ELP-1 with an NLS [pKA99-2]; and a full-length ELP-1 [pKA99-1] (Figure 2B). All three constructs were expressed in the same cells and tissues. For clarity, the protein expressed from the pKA99-1 construct will be noted as ELP- 1Δ::NLS::GFP and the full-length protein expressed from the pKA99-2 construct will be written as ELP-1::GFP.

In embryos, ELP-1::GFP expression first appeared during the comma stage. As the embryo matured into the 1-1/2 fold stage, the strongest expression was seen in the hypodermal cells, with only diffuse staining throughout the rest of the embryo (Figure 3). During larval development, expression of the fusion protein was progressively refined to muscle, neurons and epithelial cells. In the adult, ELP-1::GFP and ELP- 1Δ::NLS::GFP were expressed in body wall muscle, spermatheca, vulval muscle, seam cells, the intestine, touch receptor neurons (TRNs) and inner labial 1 (IL1) neurons of the head. Expression of ELP- 1Δ::NLS::GFP is shown in Figure 4A–D. The truncated ELP-1 construct with the NLS was used in order to identify the neuronal cell bodies.

As indicated above ELP-1::GFP and ELP-1Δ::NLS::GFP were expressed in cells that were identified as mechanoreceptor neurons (Figures 4E &4F). Although the NLS did not restrict expression to the nucleus, it was generally easier to identify neurons with this construct than with the full-length ELP-1::GFP construct. Specifically, ELP-1Δ::NLS::GFP was found in the six touch receptor neurons (ALML/R, AVM, PVM, and PLML/R) responsible for detecting light touch applied to the body surface (Figure 4E, F) 18. These cells were identified by the position of their cell bodies and the long nerve cell processes that extend anteriorly or posteriorly over half of the body length. Touch receptor neurons (TRNs) are not ciliated or organized into sensilla, but extend neurites along the midline and lateral line of the worm with their dendritic receptors lying within 150 nm of the inner border of the cuticle 19.

ELP-1::GFP was also expressed in the dendrites and the ciliated endings of six neurons in the head (Figure 4G, J–L). Candidates for these six neurons included the six inner labial 1 neurons (IL1s), six inner labial 2 neurons (IL2s), and six outer labial neurons (2OLLs and 4 OLQs) 20. To determine the identity of these neurons, the worms were labeled with the lipophilic tracer molecule DiD which gains access to the sensory cilia that project through an opening in the cuticle. In the presence of calcium acetate, DiD is exclusively taken up into the six IL2 neurons, although occasional dye-filling of the amphid neurons can be observed (personal communication, Elizabeth Ryder, Worcester Polytechnic Institute). The ciliated nerve endings of IL2 and IL1 both extend through the inner labial sensillum. However, only the IL2 nerve endings extend through a hole in the cuticle and are able to take up the dye. Dye-filling in the presence of calcium acetate allowed visualization of the IL2 neurons and thus allowed for a comparison to be made between the position of the IL2 neurons relative to the ELP-1::GFP expressing neurons (Figure 4J–L). DiD staining in the presence of calcium acetate showed that the IL2 neuronal cell bodies were located anterior to the ELP-1::GFP expressing neurons. This finding suggests that these head neurons are the mechanosensory IL1 neurons. We confirmed this identification by crossing ELP-1::GFP worms with deg-1(u38) mutations. The toxic DEG-1 protein caused the degeneration of the neurons which normally express the deg-1 gene: IL1, PVC, AVG and AVD 21. In the deg-1(u38) background, we found fewer than three GFP- labelled neurons in the head (Table 1). These results confirm that the IL1 mechanosensory neurons express ELP-1::GFP.

In addition to the mechanoreceptor neurons described above, ELP-1::GFP was associated with the nine pairs of rays of the male tale (Figure 4H). The male tale is a sensory apparatus used to find the hermaphrodite vulva during mating. The relatively diffuse and broad band of fluorescence appears to originate from the hypodermal cells (hyp7) and one or more of the neuronal cells of the ray (RnA, RnB).

At the subcellular level, ELP-1::GFP was associated with adhesion sites in both males and hermaphrodites. In the hermaphrodite body wall muscle, ELP-1::GFP was associated with the muscle arms that terminate in neuromuscular junctions (Figure 5). The muscle arms are long cellular processes that extend to make contact with motor axons. The muscle cell shown in Figure 5A shows four muscle arms, which is consistent with the three to five muscle arms classically observed per body wall muscle cell 2223. The ELP-1::GFP fluorescence is found throughout the muscle arm.

Figure 5B shows a male with ELP-1::GFP associated with repeating focal attachment points in the sex-specific muscles of the worm. These muscles are located in the posterior end of the worm and function in various phases of male mating behaviour 24. ELP-1 expression was prominent at the adhesion sites located at the sarcolemma. Unlike the dense body adhesion sites described below, these sites occur at the muscle ends in a plane perpendicular to the long axis of the sarcomere.

Hermaphrodites were examined by fluorescent microscopy with the optical axis perpendicular to the longitudinal axis of a body wall muscle cell. Affinity-purified antibodies against ELP-1 and the full-length ELP-1::GFP fusion protein localized to thin linear rows near the cell surface that were not quite parallel to the long axis of the muscle cell (Figure 6A, B). At higher resolution there was a periodicity to these lines that approximated the distance between the dense body adhesion sites (Figure 6C, D).

Body wall muscle cells are anchored along their length to the basement membrane by the dense bodies, finger-like projections analogous to the vertebrate Z-lines. Anchorage of actin in the myofilament lattice to the dense bodies is necessary for force transduction in body wall muscle. Dense body puncta are distinctive because they are aligned in a row that runs at a 6-degree pitch from the longitudinal axis of the muscle cell 25.

In Figure 6E, individual dense bodies are shown as phase-dark puncta. The fluorescent image of ELP-1::GFP was examined at the same focal plane (Figure 6F) and these micrographs were overlaid and shown in Figure 6G. The ELP-1::GFP fluorescence slightly overlaps the edges of the dense bodies but is not superimposable with the dense bodies at this angle. These results indicate that ELP-1 is unlikely to be a component of the dense bodies.

To further examine the association of ELP-1 with dense bodies, we double-stained muscle cells with anti-ELP-1 antibodies and a monoclonal antibody MH25 against β integrin (PAT-3), an integral membrane component that anchors the dense bodies to the sarcolemma. Figure 6 H-J shows that integrin and ELP-1 have overlapping staining patterns at this level of resolution. Because of the relatively impenetrable cuticle that surrounds the nematode, these worms were frozen and cracked open prior to antibody staining. Occasionally a portion of the muscle cell membrane was removed carrying along its complement of dense bodies (see white outline in Figure 6J). In areas lacking dense bodies, ELP-1::GFP was sometimes lost. However most of the ELP-1::GFP was retained in a linear pattern. These results indicate that the membrane and dense bodies can be separated from the ELP-1::GFP.

Although the truncated construct and the full-length construct were expressed in the same cells and tissues, only the full length ELP-1::GFP construct localized to an elaborate array of fluorescent filaments. These filaments, which resembled microtubules, were obvious in the larger cells of the worm including the body wall muscle (Figure 7) and intestine (Figure 8). An obliquely striated fluorescent pattern emerges approximately 0.8 μm apically from the base of the dense body. Deeper within the muscle body the filaments appear to separate from the linear track and branch out into the cytoplasm.

To test the idea that the filaments decorated by ELP-1::GFP were microtubules, we treated worms with the microtubule inhibitor, nocodazole. Because the worm cuticle is an effective barrier to nocodazole, we gently pressed worms between a slide and coverslip to extrude their intestines. This process was done in the presence of levamisole, an acetylcholine agonist, which was used to paralyze the worms for microscopy. In Figure 8, the worm was flattened and the intestines were gently extruded. The filaments in the extruded layer were stable for 30 minutes or more in the absence of nocodazole. The decorated filaments began to degrade within seconds and were entirely de-polymerized after two minutes in nocodazole. When the fluorescence disappeared the intestinal fragments noticeably flattened and the remnants of the worm contracted. This result indicates that ELP-1::GFP associates with microtubules in situ.

Finally, we show that ELP-1 was enriched in preparations of paclitaxel-stabilized microtubules in vitro. Paclitaxel-stabilized microtubules were prepared from a mixed-stage worm preparation and examined by means of Western blotting with an affinity-purified antisera against a bacterially expressed ELP-1 fusion protein (Figure 9). An ELP-1-reacting band co-purified with microtubules and migrated at ~100 kDa, a mass that corresponded to the Genefinder™ predictions for the larger ELP-1a polypeptide. Two smaller and less abundant proteins that cross-react with the ELP-1 antibodies also co-purified with microtubules. These may be isoforms generated by alternative splicing (i.e. ELP-1b) or proteolytic fragments of ELP-1a. This experiment and the one described above showed that ELP-1 is microtubule-associated both in situ and in vivo. Whether ELP-1 binds directly to tubulin remains to be determined.

Based on its expression in the TRNs and its intimate association with microtubules, we hypothesized that ELP-1 plays a role in gentle-touch sensation. We tested this idea by measuring touch-sensitivity in mutants carrying defects in the elp-1 gene and in animals treated with RNAi directed against ELP-1. We found that the ok347 allele significantly increases the proportion of touch-insensitive animals (Figure 10). This is unlikely to be the null phenotype, however, since ok347 mutants retain two transcripts predicted to encode nearly full-length ELP-1 proteins containing both the HELP and WD repeat domains (Figure 2E). Thus, ok347 is predicted to be a partial loss-of-function allele. Consistent with this idea, the proportion of touch-insensitive animals was dramatically increased when the elp-1(ok347) allele was placed in trans to a deficiency that covers the elp-1 gene (ozDf1) and when ELP-1 expression was decreased by feeding RNAi-expressing bacteria (Figure 10). These results demonstrate that wild-type ELP-1 is needed for normal touch sensitivity in C. elegans and indicate that EMAP-like proteins could play a critical role in sensory mechanotransmission.

Stock solutions of levamisole, DiD (1,1'-dioctadecyl-3, 3,3', 3'-tetramethylindodicarbocyanine, 4-chlorobenzenesulfonate salt, Invitrogen-Molecular Probes, Carlsbad, CA), paclitaxel (Taxol™, Merck-Calbiochem, San Diego, CA), and nocodazole were stored at -20°C and prepared as follows (solvent, concentration): levamisole (M9 buffer, 100 mM); DiD (1 mg/ml, dimethyl formamide); paclitaxel (10 mM, dimethylsulfoxide); and, nocodazole (10 mM, dimethylsulfoxide).

Wild-type worms (N2, Bristol) were grown on NGM plates seeded with the E. coli strain OP50 at 20°C, unless otherwise indicated 4849. MT8189 lin-15(n765ts) X, KP3948 eri-1(mg366) IV; lin-15B(n744) X, TU38 deg-1(u38) X, RB639 elp-1(ok347) V, and BS518 ozDf1/sdc-3(y52y180) unc-76(e911)V animals were supplied by the Caenorhabditis Genetics Center, Minneapolis. Genotypes of new strains generated to examine tissue specific expression include lkEx4[P_elp-1::elp-1(exon Δ11–16)::nls::gfp (pKA99-1); rol-6(su1006)] (KA15-17), (lin-15(n765); lkEx1[P_elp-1::elp-1::gfp (pKA99-2); lin-15(+)]) (KA6-8), (lin-15(n765); lkEx3[P_elp-1::gfp (pKA99-3); lin-15(+)]), (KA14) and deg-1(u38)X;lkEx4[P_elp-1::elp-1::gfp (pKA99-2); rol-6(su1006)] (KA37).

The elp-1(ok347) strain, RB639, was obtained from the C. elegans Gene Knockout Consortium at the Oklahoma Medical Research Foundation (Oklahoma City, OK), out-crossed four times and the exact location of the deletion was confirmed by sequencing a PCR product that spanned the deletion (Figure 2D).

To test whether elp-1(ok347) was a null or hypomorphic allele of elp-1, the elp-1(ok347) allele was placed in trans to ozDf1. The deficiency ozDf1 maps to linkage group V (left end: 10.318; right end 25.1193) and deletes dpy-21, pro-3, srf-4, and unc-51 50 and covers elp-1. The absence of elp-1 in strain BS518 was confirmed via PCR.

A 2595-bp cDNA clone for elp-1b (yk209e10) was kindly provided by Y. Kohara at the DNA Data Bank of Japan http://www.ddbj.nig.ac.jp. The insert was sub-cloned into a pET14b expression vector (Novagen, Madison, WI). The 6-His-tagged ELP-1b (aa1-864) was expressed in E. coli (BL21(DE3)) and purified under denaturing conditions via chromatography on an iminodiacetic acid Sepharose column 51. Antibodies were generated in rabbits against the purified 6-His-tagged ELP-1b immunogen and affinity-purified as specified 52.

Proteins were separated on 8% acrylamide mini-gels, transferred to nitrocellulose, probed with affinity-purified anti-ELP-1 antibodies (1:1000), and visualized with alkaline phosphatase-conjugated secondary antibodies and chemiluminescence.

Vectors that express ELP-1 fused to GFP were created from the Fire Lab Vectors http://www.addgene.org by PCR amplification or restriction digestion of the genomic cosmid, F38A6.2. Standard molecular techniques were used and all PCR products were cloned and sequenced to be certain that no errors were introduced during amplification. The P_elp-1::elp-1(Δ11–16)::nls::gfp construct (pKA99-1) was generated by ligating a SphI-BamHI genomic fragment that contains 4 kb of the elp-1 5'UTR region (P_elp-1) and 3 kb of the elp-1 gene (exons 1–10) into the pPD95.67 vector upstream of a nuclear localization sequence (NLS) and the gfp gene. The construct expresses a truncated ELP-1 protein fused to GFP with an NLS under the endogenous elp-1 promoter. The P_elp-1::elp-1::gfp construct (pKA99-2) contains 9 kb of the elp-1 gene with the endogenous elp-1 promoter (P_elp-1) inserted into the SphI-XbaI site in the pPD95.75 vector and expresses the full-length ELP-1 protein (exons 1–16) fused to GFP. elp-1 was amplified from the F38A6 cosmid with the Expand Long Template PCR system (Roche-Boehringer Mannheim; Alameda, CA) with the primers, EMAPUSGFP1-5', corresponding to sequence upstream of the SphI genomic restriction site (5'-AACACCGAACTTGATGAAATATTCGGTGCAAC-3') and EMAPSTP2GFP2-3' (5'CCTC

TCTAGA

Transgenic strains carrying each of these extrachromosomal arrays (200 ng/μl) were established with rol-6(su1006) (100 ng/μl) as a transformation marker in N2 worms or with the rescue reporter lin-15(+) (100 ng/μl) injected into a lin-15(n765ts) mutant strain. Three independent array-bearing lines were isolated for each array with each line showing the same tissue expression pattern.

The inner labial 2 (IL2) sensory neurons were labelled with 10 μg/ml DiD in M9 buffer containing 50 mM calcium acetate (personal communication from Elizabeth Ryder, Worcester Polytechnic Institute). After incubation for one hour, the animals were washed twice with dH₂O and put onto an OP50 feeding plate for 1 hr to allow the excess DiD to pass through the intestines. Worms were washed off the plate with M9 and placed on 2% agar pads in a drop of M9 buffer containing 10 mM sodium azide, covered with a coverslip, and examined via fluorescence light microscopy.

Microtubules were disrupted in adult worms via a protocol adapted from a previous study of the microtubule binding protein ZYG-9 53. ELP-1::GFP expressing worms were gently flattened between a coverslip and slide to extrude a portion of the intestine. Squashed worms were perfused three times with a small volume (10 μl) of M9 buffer containing levamisole (25 mM) and observed for ~15 minutes. We disrupted microtubules in this preparation using nocodazole, an antagonist of microtubule polymerization. Nocodazole (10 μM) was applied by superfusing squashed worms with 10 μl of solution.

Immunofluorescent staining of worms was carried out following the methods of Miller and Shakes (1995) 54. A mixed population of worms was freeze-cracked by immersion in liquid nitrogen, fixed in ice-cold methanol (15 min) and in ice-cold acetone (10 min). Animals were washed twice in PBT [PBS containing 0.1% Triton-X (v/v) and 0.1% BSA (w/v)] and stained with primary antibodies overnight at 4°C. The primary antibodies used were the affinity-purified anti-ELP-1 antibodies (diluted 1:50 in PBST with BSA) and a mouse monoclonal anti-PAT-3 antibody MH25 (a gift from Michelle Hresko, Washington University School of Medicine, St. Louis, 38) (diluted 1:250 in PBST). The worms were washed three times in PBST and then incubated with the secondary antibodies: Cy2-conjugated donkey anti-rabbit [1:250] or Cy3-conjugated donkey anti-mouse [1:250] in PBT buffer overnight at 4°C. Lastly, the slides were washed three times in 1× PBT and mounted in PBT buffer.

C. elegans were examined by means of bright field, differential interference contrast (DIC), and fluorescence microscopy 49. Worms were anesthetized with a drop of 1% 1-phenoxy-2-propanol in M9 buffer and placed on 2% agar pads on glass slides. The highest resolution images were taken with a 60× 1.4 NA Plan-Apochromat Nikon objective and captured with a Hamamatsu Orca-ER camera (Hamamatsu City, Japan) driven by the OpenLab software V3 (Improvision, Lexington, MA). Individual files were compiled and contrast-balanced when necessary using Adobe PhotoShop. All the images for each dataset were treated the same.

Recently hatched larvae of the RNAi-hypersensitive mutant, lin-15b;eri-1, were fed RNAi-expressing bacteria from the Ahringer RNAi feeding library 55.

Young adult animals were alternately touched on the anterior and posterior region of the body a total of ten times with an eyebrow hair glued to a toothpick 56. Animals that failed to respond to all ten touches were considered touch-insensitive. Three independent assays of 30–50 animals were tested. Tests were conducted blind to RNAi treatment or to genotype.