The multigenic family of mammalian Fe65s encodes three highly similar proteins with the same modular organisation: a WW domain and two phosphotyrosine-binding domains. The PTB2 domain of these proteins binds to the cytosolic domains of the Alzheimer's β-amyloid precursor protein APP and related proteins APLP1 and APLP2, generating a highly redundant system that is hard to dissect by reverse genetics. By searching potential Fe65-like genes in the nematode Caenorhabditis elegans, we identified a single gene, feh-1 (Fe65 homolog-1), encoding a protein with a high sequence similarity to mammalian Fe65s. FEH-1 is also functionally related to mammalian orthologues; in fact its PTB2 domain binds to APL-1, the product of the C. elegans orthologue of APP. Staining with specific antibodies show that the neuromuscular structures of the pharynx are the sites in which FEH-1 is present at highest levels. Expression studies with reporters indicate that the feh-1 gene is also expressed by a subset of the worm neurons.
We generated and isolated a deletion allele of feh-1, and the corresponding homozygous mutants arrest as late embryos or as L1 larvae, demonstrating for the first time an essential role for a Fe65-like gene in vivo. The pharynx of homozygous larvae does not contract and the worms cannot feed. Analysis of pharyngeal pumping in heterozygous worms and in feh-1 RNA-interfered worms indicates that dosage of feh-1 function affects the rate of pharyngeal contraction in C. elegans. Interference with apl-1 double-stranded RNA showed a similar effect on pharyngeal pumping, suggesting that FEH-1 and APL-1 are involved in the same pathway. The non-redundant system of the nematode will prove useful for studying the basic biology of the Fe65-APP interaction and the molecular events regulated by this evolutionarily conserved system of interacting proteins.
The main constituent of Alzheimer's amyloid plaques is a peptide, named Aβ, generated through the proteolytic processing of the precursor protein APP ( De Strooper and Annaert, 2000). The latter is in fact cleaved by alternative pathways mediated by α-secretase or by β- and γ-secretases; under pathological conditions, the β-amyloid peptide, generated by β- andγ -secretase activities, accumulates in the senile plaques typical of Alzheimer's disease. The identity of γ-secretase with presenilins, or with protein complexes involving presenilins, has been recently suggested ( Wolfe and Haass, 2001).
Binding of several proteins to the cytosolic domain of APP influences APP proteolytic processing ( Sabo et al., 1999; Borg et al., 1998; Sastre et al., 1998; Guènette et al., 1999; Ando et al., 2001). Some of these APP-interacting proteins belong to the family of the Fe65s, which includes Fe65, Fe65-L1 and Fe65-L2 ( Fiore et al., 1995; Guènette et al., 1996; Duilio et al., 1998; Russo et al., 1998). The common modular structure of the Fe65s, composed of a WW domain and two independent phosphotyrosine-binding domains (PTB), PTB1 and PTB2, suggests the function of molecular adaptors for these proteins. The PTB2 domain of the three proteins interacts with the cytosolic region of APP and of related proteins APLP1 and APLP2 at the level of a YENPTY sequence, which is common to all these APPs ( Guènette et al., 1996; Duilio et al., 1998; Russo et al., 1998, Zambrano et al., 1997). Other than the Fe65s, additional PTB-domain-containing adaptors, the X11 family members and m-Dab1, bind to the APP cytodomain ( Borg et al., 1996; Howell et al., 1999; Homayouni et al., 1999). In turn, these adaptors interact with other proteins; in the case of Fe65, the WW domain binds to several proteins, including Mena ( Ermekova et al., 1997) and the non-receptor tyrosine kinase Abl ( Zambrano et al., 2001). In the latter case, Fe65 recruits active Abl close to APP, which, upon phosphorylation of its tyrosine 682, binds the Abl SH2 domain ( Zambrano et al., 2001). The PTB1 domain may interact at the membrane level, with the low-density lipoprotein receptor-related protein LRP ( Trommsdorff et al., 1998), and in the nucleus, with the transcription factor CP2/LSF/LBP1 ( Zambrano et al., 1998) and the histone acetyltransferase Tip60 ( Cao and Sudhof, 2001).
Most of the genes encoding the proteins taking part in this complex molecular machinery have been isolated and, in some instances, characterised in C. elegans. In fact, apl-1 is the nematode orthologue of the APP gene family; the structural properties of the encoded protein suggest that APL-1 adopts the same topology as APP, and the similarity of the C-terminal cytosolic domain of APL-1 with that of APP is strikingly high ( Daigle and Li, 1993), suggesting an evolutionary conservation of the functions of this domain. Two different presenilin genes, sel-12 and hop-1, are present in C. elegans, where they control Notch signalling ( Levitan and Greenwald, 1995; Li and Greenwald, 1997). Nicastrin, and its nematode orthologue aph-2, have been proposed as molecular links between presenilin and APP machineries ( Yu et al., 2000). lin-10 is the C. elegans orthologue of X11, and it is needed for proper localisation of the let-23 gene product during vulval development ( Rongo et al., 1998). m-Dab and Mena orthologues have been first described in Drosophila, where they genetically interact with the tyrosine kinase abl gene ( Gertler et al., 1993; Gertler et al., 1995), and their orthologues are also present in the C. elegans genome.
The existence of three Fe65 proteins in mammals, all interacting with APP, renders the use of mouse genetic manipulation to study the functional role of Fe65s and of their complexes with APPs difficult. On the contrary, the lower complexity of the nematode genome enables the use of reverse genetic approaches in C. elegans to study the functions of the orthologues of mammalian gene families in a simpler, and in many cases, non-redundant genetic system. On the basis of this evidence, we attempted to identify, in the C. elegans genome, putative Fe65-like genes. In silico screening allowed us to find only one gene, which we call feh-1 (Fe65 homolog-1), encoding a protein with high degree of similarity to mammalian Fe65s. In fact, FEH-1 protein possesses the same modular organisation of the Fe65s, and its PTB2 domain interacts with the cytosolic domain of APL-1. Reverse genetics analyses demonstrated that the FEH-1 protein provides essential functions in C. elegans. Analysis of heterozygous worms, and of the phenotypes induced by RNA interference targeting either feh-1 or apl-1 transcripts, suggests that both genes are involved in a common molecular mechanism, whose alteration affects the rate of pharyngeal contractions.
Materials and Methods
Standard culture methods were used to handle the nematode strains used in this study ( Sulston and Hodjkin, 1988). The wild-type strain used was the Bristol N2; the following mutant alleles, obtained from the Caenorhabditis Genetic Center (St. Paul, MN) were also used: lin-15(n765) and dpy-1(s2170).
Bioinformatic tools, molecular methods and transgenic lines
Homology alignments were performed with the ClustalW protocol, available from EBI ( Thompson et al., 1994). Blast searches ( Altschul et al., 1990) were performed through the Sanger Centre Blast server. Analysis of the genomic sequences was performed with Gene Finder, which is available through the Baylor College of Medicine ( Favello et al., 1995). In silico isolation of the feh-1 gene was accomplished as described: the HVFRCEAPAKNIATSLHEICSKIMSERR sequence was used to test C. elegans databases with the TBLASTN protocol, and the expressed sequence tag, yk23f9.5 was identified. Its sequence allowed the isolation of the C. elegans genomic sequence Y54F10. Analysis of this sequence with GeneFinder allowed the identification of the coding regions of a putative feh-1 gene. The yk423e6 clone was obtained by Y. Kohara and fully sequenced by the Sequenase kit (Amersham Pharmacia Biotech). The resulting sequence was submitted to EMBL database (Accession #AJ345015).
Standard methods were used for generation and propagation of the recombinant constructs ( Sambrook et al., 1989). Restriction and modifying enzymes were from Roche; oligonucleotides were from CEINGE. Polymerase chain reactions (PCR) for cloning purposes were performed with Pfu DNA polymerase (Promega) or Gold Taq (Perkin Elmer) for mutant screening.
The feh-1 cDNA regions encoding the WW and the PTB2 domains were amplified by PCR, using the yk423e6 clone as a template; the APL-1 cytodomain was obtained by amplification of Bristol N2 genomic DNA. The primers were:
WW-F: 5′-AGTGGATCCCCGAAAGATTTACCACCAGG-3′; WW-R: 5′-AGTGAATTCGGTCTCACGGTTCACTGGT-3′; PTB2-F: 5′-ACGGGATCCGCTATCGAATCAGGAGAAAAGA-3′; PTB2-R: 5′-ACGGAATTCGGCGTCGAGCACTTTTTGGT-3′; APL-1F 5′-ATCGGATCCACCAACGCTCGTCGTCGC-3′; APL-1R 5′-ACT-GAATTCGGCCTTCGAGTCGAAGAATG-3′.
The purified and digested products were cloned into the pGEX2TK vector (Amersham Pharmacia Biotech) for expression in Escherichia coli.
The 5′ region of feh-1 (positions -2352 to +1311, +1 is at the A residue of the first methionine codon) was obtained by PCR amplification of Bristol N2 genomic DNA with the following primers: EXP-F1X: 5′-AAAATCTAGAGTATGTGTACGAGATTATCGCCT-3′; EXP-R1B: 5′-AATAAGGATCCCGATGTTCGGTCATAATTGT-TGTATC-3′.
The product was digested with XbaI and BamHI enzymes and cloned into the pPD21.28 and pPD95.75 vectors, kindly provided by A. Fire ( Fire et al., 1990), to generate, respectively, the 5′-feh-1::lac z, or the 5′-feh-1::GFP constructs. For the FEH-1 expression construct, which was used to rescue the mutant phenotype (fl-feh-1::HA), the complete gene, starting from the 5′ end used in the 5′-feh-1::GFP constructs, to codon 693, was cloned in pPD95.75. The primers were: EXPR-F1-P: 5′-AAAACTGCAGGTATGTGTACGAGATTATCGC-CT-3′; EXPR-R2-B: 5′-ACGGGATCCCGTTGTGCCGACCTAG-AAGGTACC-3′.
Green fluorescent protein (GFP) expression from the resulting construct was abolished by cloning, upstream of GFP, a 550 bp amplification product containing the hemagglutinin epitope (HA), a stop codon and the 3′ end of the gene. The primers used were: F-HA-10.8: 5′-ACAGGATCCCTACCCATATGATGTTCCAGATTACGC-TGCACAACGACTTTGAATTCTTCAT-3′; 11250R-B: 5′-ACAGG-ATCCCATACCAGCCTGGTTACCTACC-3′.
Recombinant constructs were co-injected into the ovary of young adult hermaphrodites to establish transformed lines; the transformation markers were the wild-type lin-15 for injections into temperature-sensitive lin-15(n765), multivulva (Muv) worms ( Clark et al., 1994). The expression of wild-type lin-15 allowed reversion of Muv phenotype at 20°C and the identification of recombinant lines bearing the arrayed constructs. Injections leading to phenotypic rescue were performed with the elt-2::GFP vector ( Fukushige et al., 1999) into feh-1 (+/gb561) worms.
Histochemical detection of β-galactosidase ( Edgar, 1995) was performed on acetone-fixed worms in a buffer containing 0.2 M sodium phosphate buffer, pH 7.5, 1 mM magnesium chloride, 0.004% (w/v) sodium dodecyl sulfate, 10 mM each of potassium ferricyanide and potassium ferrocyanide, 0.4% X-gal.
Antibodies, pull-down assays and immune detections
The recombinant proteins used were obtained by isopropyl-thio galactoside (IPTG) induction of bacterial cultures harbouring the corresponding constructs in the pGEX2TK vector. Affinity purification of the glutathione S-transferase (GST) fusion proteins was performed on glutathione-sepharose resin (Amersham Pharmacia Biotech). FEH-1 and APL-1 antisera were obtained by immunisation of rabbits with the GST-FEH-1-WW domain protein or with a peptide-ovalbumin conjugate of the juxtamembrane sequence of APL-1 (PRIMM). To obtain FEH-1 purified antibodies, immune sera were deprived of GST antibodies on a GST column, then purified on an antigen column. For APL-1 antibody purification, a specific peptide column was set. The GST and GST-FEH-1-WW columns, as well as the APL-1 peptide resin, were obtained by crosslinking the purified antigens to CNBr-activated sepharose (Amersham Pharmacia Biotech), following the instructions of the manufacturer. Fe65 antibodies have been described ( Zambrano et al., 1997). For preadsorption control, the FEH-1 antibody was challenged with 10 μg of purified antigen, then used in a western blot. Phosphatase treatment was performed on FEH-1 immunoprecipitates from 500 μg of C. elegans proteins as described ( Zambrano et al., 1998). For the pulldown experiments, equal amounts of fusion proteins (10 μg) were bound to glutathione-sepharose beads (10 μl) and challenged with nematode or mouse brain extracts obtained by lysis in a buffer containing 50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% glycerol, 50 mM NaF, 1mM Na vanadate, with a protease inhibitor cocktail (Complete, EDTA-free, Roche). The extracts were clarified at 16,000 g at 4°C, and the protein concentration determined by the Bio-Rad protein assay according to manufacturer's instructions. Unbound proteins were removed by washing the beads three times with lysis buffer, whereas retained proteins were resolved by SDS-PAGE, electroblotted to polyvinylidene fluoride Immobilon-P membrane (Millipore) and analysed by western blot with FEH-1 and APL-1 antibodies. Signals were detected with the ECL system (Amersham Pharmacia Biotech). Loading of the GST proteins was checked by staining the filters with Ponceau S (Sigma).
Whole-mount immunohistochemical detection ( Miller and Shakes, 1995) was performed in a buffer containing 1% (w/v) bovine serum albumin, 0.1% NP-40 (v/v), 1 mM EDTA, in phosphate buffered saline (PBS) and appropriate dilutions of the FEH-1 and MH27 antibodies. The latter was kindly provided by M. C. Hresko and R. H. Waterston ( Francis and Waterston, 1991). Fluorescent secondary antibodies (Jackson Immunoresearch Lab) were diluted 1:200. Texas-red conjugated phalloidin (Molecular probes), 1:20, was incubated for 15 minutes with worms previously stained with FEH-1 antibodies to detect pharyngeal muscles.
Mutant generation and RNA-mediated interference
Young adult worms were mutated using standard protocols ( Yandell et al., 1994). Briefly, Bristol N2 worms fed to DH5αE. coli cells were treated with 4, 5′, 8-trimethylpsoralen, 30 μg/ml, for 15 minutes, then irradiated for 1 minute with 366 nm UV source. 24 hours later, F1 eggs were allowed to hatch, then split at 1250 individuals per plate onto 96 NGM/agarose plates. Identification of mutant addresses was performed by PCR on DNA prepared from aliquots of nematode pools. Nested PCR was performed with two pairs of oligonucleotides. The outer set was: F454: 5′-CTCCAGATCATTCCCGTAGAGGA-3′; R4734: 5′-TGAAGCTTTCCGATGAGGTTTGC-3′. The inner set was: F756: 5′-ACGACTCTCGTGGTTACTCTTCG-3′; R4232: 5′-CAGCAGCTCTACATCATCCCTAC-3′. Optimal reaction conditions tested to detect preferentially the deleted allele were: 94°C, 30 seconds; 56°C, 30 seconds; 72°C, 90 seconds for 30 cycles. Nested PCR was performed on 2 μl of a 1:100 dilution of the first amplification following the same scheme. The wild-type allele was detected as a 3.5 kb band. One population, giving rise to a shorter, 2-kilobase product, was subjected to four cycles of sibling selection for isolation of the mutated clone. The isolated heterozygous line was balanced by crossing with dpy-1(s2170) males. This allele allows us to distinguish among +/+ and +/gb561 worms; +/+ worms have a dumpy phenotype, whereas +/gb561 appear normal. 250 +/+ and 250 +/gb561 individuals, as well as 500 growth-arrested L1s, were individually collected from the offspring of heterozygous worms and lysed in SDS buffer (100 mM Tris/HCl, pH 6.8, 4% SDS, 100 mM DTT) for detection of FEH-1 protein by western blot.
For double-stranded RNA-mediated interference (RNAi), feh-1 or apl-1 transcripts obtained by T7 RNA polymerase transcription of PCR-amplified products (RiboMax system, Promega) were injected into the ovary of young adult hermaphrodite worms. The primer pairs used for feh-1 were: T7-F1: 5′-taatacgactcactataggAAGCCGTGACGGAGCCATCTCT-3′; T7-R1: 5′-taatacgactcactataggCACCTTCATGTTTCTCCCATCC-3′, which amplify the third exon of the gene; T7-F2:5′-taatacgactcactataggTCGCCTACGTGTCCCGTGATCG-3′; T7-R2:5′-taatacgactcactataggGATATCTAACTCTGCACTCGGC-3′, which amplifies the seventh exon of feh-1. Lower case letters indicate the T7 promoter sequence. As a control for RNAi efficacy, the progeny of feh-1 dsRNA-injected worms, as well as a comparable number of mixed-stage individuals were lysed in SDS buffer for FEH-1 immunoblot.
For RNAi by feeding, the feh-1 cDNA from the yk423e6 construct was excised and cloned into pPD129.36 vector. The apl-1 cDNA region spanning positions 961-2045 was generated by RT-PCR on mixed-stage RNA samples with Superscript reverse transcriptase (Invitrogen) and Pfu polymerase (Promega). The reverse primer used for cDNA synthesis and PCR amplification was: apl-1R-X: 5′-ATATTCTAGACCTTCGAGTCGAAGAATGAGTACGT-3′; the forward primer was: apl-1F-H: 5′-ATCAAGCTTGAGATCGAGGCGGTTCATGAGGAG-3′.
The PCR product was digested with HindIII and XbaI enzymes and cloned into pPD129.36. This vector contains a polylinker flanked by two T7 RNA polymerase promoters. The corresponding apl-1 construct was also used to amplify, with a T7 promoter primer, the template for synthesis of the dsRNA used for apl-1 RNAi by injection. E. coli cultures harbouring the generated feh-1 and apl-1 constructs in the HT115(DE3) strain were used as source for RNAi by feeding ( Timmons and Fire, 1998). Bristol N2 embryos or L1 larvae were deposited onto NGM plates containing 10μ g/ml tetracycline, 100 μg/ml ampicillin, 1 mM IPTG, and IPTG-induced bacteria and allowed to lay eggs. F1 individuals were assayed for phenotype. Control worms were grown onto pPD129.36-transformed bacteria, without insert, producing vector-encoded dsRNA.
Pharngeal pumping was determined on 30 cloned individuals from each line or RNAi population per experiment, blindly. Such populations were scored four times during 48 hours of observation. Each experiment was repeated three times. Standard deviation was calculated for each group. For statistical comparisons, Student's t test was used.
feh-1 gene identification and characterisation of its product
Through the alignment of the sequences of mammalian Fe65, Fe65-L1 and Fe65-L2 proteins, we identified highly conserved sequence strings and used them to search potential Fe65-like gene(s) (see Materials and Methods). Translation of the longest cDNA identified from databases, the yk423e6 EST clone, gives rise to an open reading frame (ORF) of 554 amino acids; its 5′ end, although incomplete, allows us to identify the putative first exon as the one suggested by Genefinder analysis of the corresponding genomic sequence. This exon contains an ORF overlapping with that of yk423e6 cDNA and starting with a methionine 117 residues upstream of the first amino acid encoded by the cDNA. This ORF encodes a 695 amino-acid-long peptide. The Panel A of Fig. 1 shows, in bold characters, the translation of the yk423e6 EST; the first 117 residues, encoded by the first exon of the gene, and not present in the EST, are indicated by italic characters. The Fig. 1A also shows the alignment of the FEH-1 protein to the mammalian Fe65s. The highest similarity among the four proteins is present in the WW, PTB1 and PTB2 domains. Accordingly, the folding of the domains, deriving from secondary structure predictions with the Jpred2 program ( Cuff and Barton, 1999), indicate that their structure is maintained. It is worth noting that the cysteine residues, which are peculiar to the Fe65 PTB domains compared with other PTB domains, (residues C500 in the PTB1 and C655 in the PTB2 domain of Fe65) ( Borg et al., 1996; Zambrano et al., 1997; Russo et al., 1998), are present in corresponding positions of FEH-1 (C486 and C635, respectively). Single alignments of FEH-1 with each of the mammalian proteins suggest that, among the three Fe65s family members, the closest relatives to FEH-1 are Fe65 and Fe65L1. Fig. 1B shows the comparison of the three mammalian Fe65 proteins and the FEH-1 predicted structure. The size, but not the sequences of the two spacer regions separating the three domains, are similar among the four proteins, whereas the long N-terminal region is present only in FEH-1, Fe65 and Fe65-L1. Other than the short N-terminal region, Fe65-L2 also contains a longer C-terminal region. FEH-1 and Fe65 also share a high content of acidic residues in the N-terminal region.
A rabbit antibody, which was raised against the recombinant WW domain of FEH-1, fused to GST, was affinity purified and used in a western blot to detect FEH-1 in worm extracts. Its specificity was tested by probing C. elegans protein lysate with pre-immune serum and with antibody pre-adsorbed on the antigen ( Fig. 2A, lanes 1-2). Two groups of heterogeneously migrating bands were detected with the FEH-1 antibody, a fast one ranging from 60 to 70 kDa, and a slower one from 80 to 90 kDa ( Fig. 2A, lane 3). The presence of two discrete groups of polypeptides could be explained as deriving from translation starting from the two methionines, 1 and 142 ( Fig. 1A). Accordingly, expression of the cDNA contained in clone yk423e6, which is devoid of methionine 1, in COS7 cells generated a protein with a migration corresponding to the fastest group of bands (data not shown). The complex heterogeneity of migration found in both groups of bands may be interpreted as owing to post-translation modifications. Also, the migration of mammalian Fe65, shown in lane 4, produces a complex pattern on SDS-PAGE, which has previously been shown to be due, at least in part, to phosphorylation ( Zambrano et al., 1998). In order to demonstrate that, similarly to Fe65, FEH-1 is phosphorylated, we immunoprecipitated the protein from worm lysates and treated them with alkaline phosphatase. As shown in lane 6 of Fig. 2A, the mobility of FEH-1 is increased upon phosphatase treatment, compared with the untreated sample (lane 5), suggesting that FEH-1 is indeed phosphorylated in nematode cells.
The similarity of FEH-1 to mammalian Fe65s suggests that it may interact with APL-1, the C. elegans homolog of mammalian APP ( Daigle and Li, 1993). To address this point, we performed a pull-down assay by using the cytosolic domain of APL-1 fused to GST. The coprecipitation assay clearly shows that FEH-1 in worm lysates specifically interacts with the APL-1 cytodomain ( Fig. 2B). Conversely, the PTB2 domain of FEH-1, fused to GST, is able to coprecipitate with native APL-1 ( Fig. 2C).
Null feh-1 mutants are homozygous lethal
The feh-1 gene structure derived from our analysis differs slightly from the predicted coding sequence Y54F10AM.2, which is available through WormBase (genetic map position: III:-14.02). In fact, both by sequence analysis of yk423e6 cDNA clone and by RT-PCR analysis (data not shown), we have not been able to detect in the feh-1 transcript any sequence corresponding to that of the exon V, predicted in the Y54F10AM.2 sequence. In addition, the small predicted exon X is contained in a larger exon (IX in the actual gene), which is the last exon of the gene, as it contains the translation stop codon ( Fig. 3A).
In order to analyse the function of feh-1 in the nematode, we generated and characterized a deletion mutant of the gene, gb561. A genomic fragment containing this deleted allele, obtained by PCR amplification from the mutagenised line and fully sequenced, allowed us to characterise the deletion. The latter consisted of a double deletion in feh-1, removing regions from intron II to the 5′ end of intron IV, and from 3′ end of intron IV to most of intron V ( Fig. 3A, bottom). The resulting rearrangement of feh-1 will remove exons III to V, completely eliminating the WW domain, and the N-terminal region of PTB1 domain. The encoded, truncated protein will terminate with five extra amino acids after residue 147, as the fusion between exons II and VI will generate an out of frame transcript. Therefore, the gb561 allele is a null mutant as its locus can only encode a truncated FEH-1 devoid of the WW, PTB1 and PTB2 domains. This is confirmed by the western blot of Fig. 3B, in which we assayed protein lysates from wild-type, +/gb561 and gb561/561 worms for the levels of FEH-1. Reduced levels of protein are present in heterozygous worms, whereas no signal is detected in lysates from homozygous mutant worms.
The mutant worms present a recessive embryonic/larval lethal phenotype ( Table 1). About 75% of the offspring of heterozygous hermaphrodites (+/gb561) reaches the adult stage, and these worms always belong to the +/+ and +/gb561 genotypes, as assessed by single worm PCR analysis, indicating a lethal effect of the homozygous gb561/gb561 phenotype. These last worms (22.9%), confirmed to be gb561/gb561 by PCR, either do not hatch (n=25) or arrest as L1s (n=31) ( Table 1). Nomarski observation of the development of eggs laid by heterozygous hermaphrodites shows embryos that do not progress normally through development, and arrest during morphogenesis ( Fig. 4A).
Hatched homozygous mutant larvae arrest as L1. They have apparently normal morphology and movement and can remain alive on the plate for several days (up to 5-10 days). No pharyngeal pumping can be detected in these worms, suggesting that their growth and development may be blocked, at least in part, because they are not feeding efficiently.
Two approaches prove that the phenotypic defects of feh-1 gb561/gb561 embryos and larvae are indeed due to the disruption of the feh-1 gene and not to some other mutation present in the gb561 strain. First, we rescued the phenotype of the mutation with the fl-feh-1::HA construct (see experimental procedures) in which expression of the full-length FEH-1 protein is driven by 2.35 kb of sequences upstream of feh-1. The construct was injected in heterozygous feh-1 (+/gb561) worms together with a transformation marker, the elt-2::GFP construct, which drives a strong expression of GFP in gut cells beginning early in embryogenesis. We isolated three independent viable lines of feh-1 (gb561/gb561) worms carrying the transgenic array. The viability of worms from these lines was strictly dependent on the extra-chromosomal array, as all the adults express the GFP marker and the loss of GFP expression was always associated with embryonic lethality or larval arrest and with lack of detectable staining with an anti-FEH-1 antibody (data not shown). The second approach used to prove the phenotypes observed in gb561/gb561 worms are due to feh-1 ablation was based on the injection of feh-1 double-stranded RNA into wild-type young adult hermaphrodites, which resulted in F1 individuals presenting the same defects observed in feh-1 (gb561/gb561) mutants. 25 worms arrested as embryos, whereas 33 were arrested and pumping-impaired L1s ( Table 1). Also in this case, the presence of the phenotypes correlates well with a decreased amount of FEH-1 in the offspring of feh-1 dsRNA-injected worms ( Fig. 3C). The arrested embryos seen in the offspring of injected worms show a phenotype similar to that observed for the gb561/gb561, non viable embryos ( Fig. 4A,B).
feh-1 is expressed in the neuromuscular structures of the pharynx and in the nervous system
To study the expression of feh-1 in C. elegans and to try to correlate it with the phenotype observed in the mutant and in RNA-interfered worms, we used immune-detection of the protein and a reporter expression approach. FEH-1 antibodies in whole-mount immunofluorescence analysis showed a largely prominent localisation of the protein in the pharynx at all stages of development, from three-fold stage embryos to adults. Barely detectable staining could be observed also in some neurons in the nerve ring and in the nerve chord and in the tail (not shown). In the pharynx the fluorescent signals start from the most terminal end of the organ and extend posteriorly to the second pharyngeal bulb and the attachment to the intestine ( Fig. 5A). Part of the protein appears to uniformly fill the cytoplasm of the expressing cells but an important fraction appears to associate with some sub-cellular compartment/structure in the shape of filaments that form, at the level of the second bulb, a characteristic basket-like structure. Confocal microscopy analysis of worms double stained with anti-FEH-1 and anti-MH27 antibodies, which recognizes JAM-1, a component of adherens junctions of epithelia, tends to exclude FEH-1 staining from pertaining to the processes of epithelial cells, as the two signals never overlap and appear topologically complementary ( Fig. 5B). Confocal microscopy of worms stained with FEH-1 antibody and Texas-Red-conjugated phalloidin, which decorates actin-rich muscle cells, indicates, instead, that FEH-1, at least in part, colocalizes with actin in muscle cells in some areas of the two pharyngeal bulbs and in some areas just anterior to the second bulb ( Fig. 5C). However, costaining with Texas-Red phalloidin also shows association of FEH-1 with structures not stained by phalloidin. The majority of these structures have a filamentous shape and may resemble neuronal processes. The staining associated with the anterior end of the basket-like structure of the second bulb appears to be clearly different from that in the actin-containing muscle cells ( Fig. 5C).
The C. elegans pharynx is a complex organ in which muscle, neurons, epithelia and glands are tightly packed. In several cases a single cell type takes up functions and morphological features usually segregated to different ones, as in the case of the mioepithelial cells that serve a contractile role for pumping but also secrete the cuticle that lines the lumen of the pharinx. The shape of these cells is, as a consequence, often unconventional and it is difficult to identify, in whole mounts, the cells in which a protein is expressed. We thus turned to the study of worms that were transgenic for constructs in which expression of the reportersβ -galactosidase or GFP were driven by the sequences at the 5′ end of the feh-1 gene, which allowed the rescue of the gb561 mutant phenotype (see above). Furthermore, in theβ -galactosidase-expressing construct, a nuclear localisation signal allowed us to identify the nuclei of the cells expressing the transgene. The genomic fragment used for all of our expression constructs comprises 2352 nucleotides upstream of the starting ATG codon and extends 1311 nucleotides into the third exon (see Materials and Methods). The resulting constructs, 5′-feh-1::GFP and 5′-feh-1::NLS::lacZ, were injected along with a selectable marker in the gonad of young adult hermaphrodites and lines of transformed worms were established. Expression of reporters confirmed that the main site of synthesis is the pharynx, but it is also expressed in some neurons in the nerve ring, in the ventral chord and in the tail ( Fig. 6). During embryogenesis, the expression of the constructs is detectable unambiguously only after proliferation has ceased and the embryo has elongated to the two- to three-fold stage. As already indicated by immunodetection, expression of the reporters does not change substantially from embryogenesis and hatching to larval stages and adults. Expression is strong in several neurons and in pharyngeal cells. The GFP clearly outlines the shape of the expressing neurons with their processes (e.g. in the ventral nerve chord, Fig. 6A-C). On the contrary, the very strong GFP signal detected in the pharynx did not allow the identification of feh-1-expressing cells among those forming this organ. The analysis of nuclear-targeted β-galactosidase, expressed under the control of the feh-1 gene promoter, allowed us to clearly identify, in agreement with GFP reporter activity, the nuclei of some extrapharyngeal neurons ( Fig. 6D), as well as neurons of the ventral nerve chord (F) and of the tail (G). Furthermore, in the procorpus and in the first pharyngeal bulb, the nuclei of m3 and m4 muscle cells ( Albertson and Thomson, 1976) are also stained ( Fig. 6E). The staining of the nuclei in the second bulb, where the density is higher, does not allow us to unambiguously assign the feh-1-expressing cells; as shown in Fig. 6E, two main spots ofβ -galactosidase activity can be seen, which may correspond to the m5 nuclei.
The effect of FEH-1 dosage on pharyngeal pumping rate and the interaction with apl-1
In arrested feh-1 (gb561/gb561) L1s, which completely lack FEH-1, no visible contraction of the pharynx can be observed even with high-resolution optical microscopy. To elucidate the role of FEH-1 in pumping, we analysed pharyngeal contractions in heterozygous feh-1(+/gb561) worms. As expected, since gb561 is recessive, contractions can be observed; however, somewhat surprisingly, the rate of pharyngeal pumping of these heterozygous worms is always and significantly higher than that of wild-type N2 individuals ( Fig. 7A). We also analysed feh-1(gb561/gb561) worms in which the larval arrest phenotype of the mutation has been rescued by the fl-feh-1::HA construct (see above). In these worms the rate of pharyngeal contractions is also completely rescued to wild-type levels by the transgene ( Fig. 7A), indicating that the effect on pumping depends on feh-1 dosage. We next asked if we could reproduce the dosage effect on pumping rate observed in worms heterozygous for the feh-1-null allele gb561 by reducing, instead of completely abolishing, the expression of feh-1 and thus generating worms hypomorph for feh-1. For this, we used RNA interference through feeding. With this approach, it is possible to obtain a decrease in the levels of the corresponding mRNA and, in turn, of the cognate protein ( Timmons and Fire, 1998). In fact, by feeding adult N2 hermaphrodites with bacteria producing feh-1-specific dsRNA, we obtained RNA-interfered adult F1 individuals presenting an enhanced pumping rate ( Fig. 7B), similar to that of heterozygous worms. It is interesting to note that, with RNA interference obtained by feeding, we have not been able to produce the embryonic or larval arrest phenotypes. The latter was instead obtained by microinjection of synthetic feh-1 dsRNA, which reproduces the phenotype of the null mutants (see above; Table 1). These results suggest that reduced levels of FEH-1, although compatible with feeding and larval development, have a clear effect on the rate of pharyngeal contractions, whereas complete loss of this function, in feh-1 (gb561/gb561) homozygous larvae, results in the absence of pharyngeal contraction and larval development arrest, probably because feeding is impaired.
A crucial point that should be addressed is whether APL-1, which we have shown in this paper to physically interact with FEH-1, is involved in the same molecular pathways as FEH-1. To analyze this point, we injected into the worms a dsRNA of apl-1. This experiment resulted in a larval developmental arrest, similar to that observed in the case of feh-1 RNAi by injection (data not shown). Then, we generated a construct to perform RNA interference by feeding with bacteria producing apl-1-specific dsRNA. Worms grown on these bacteria show a normal development, with no obvious major abnormalities. However, their rate of pharyngeal pumping is significantly increased to the same extent observed in feh-1(+/gb561) heterozygous worms and in the feh-1 dsRNA-interfered worms, whereas the pumping rate of worms fed to control bacteria was not distinguishable from that of wild-type worms ( Fig. 7B).
In mammals, many of the proteins taking part in the complex network of interactions centred at the cytosolic domain of APP belong to gene families. This is in fact the case for APP-like genes, as well as for Fe65- and X11-like members. APP-like gene family members have been analysed in depth, in mammals, by reverse genetics. However, their redundancy prevented the generation of clear phenotypes from murine knock-outs of the single APP, APLP1 or APLP2 genes ( Zheng et al., 1995; von Koch et al., 1997; Heber et al., 2000), whereas mice with combined knock-outs displayed severe phenotypes, dying early and post-natally. Taken together, the functions of these genes have been proposed as essential and partially redundant in mammals ( Heber et al., 2000). A similar phenotype, although not strictly predictable, might occur if reverse genetic approaches were used with the mammalian Fe65 gene family members. This was one of the reasons that prompted us to investigate potential Fe65-like genes in the simpler genetic system of the nematode C. elegans. Genomic and EST database searches allowed us to isolate feh-1, which appears to be the unique nematode gene coding for a protein highly similar to mammalian Fe65s. We show, in fact, that the sequence of the encoded protein shows the same modular organisation of mammalian Fe65s, including the WW, PTB1 and PTB2 domains, confirming that the three protein-protein interaction domains constitute the main functional regions of these molecular adaptors. The PTB2 domain of FEH-1 is also functionally related to the corresponding domains of the mammalian homolog proteins, as it is able to interact with the cytosolic domain of APL-1. Furthermore, the high similarity between the WW and PTB1 domains of nematode and mammalian Fe65s suggests that these domains may be similarly involved in protein-protein interactions, in nematode cells. The possibility that these domains may be engaged in interactions with the nematode counterparts of mammalian ligands of WW and PTB1 domains is also strengthened by the presence, in C. elegans, of genes coding products similar to the mammalian WW ligands Mena and Ab1 and to PTB1 ligands CP2/LSF/LBP1, Tip60 and LRP ( Goddard et al., 1986; Yochem and Greenwald, 1993) (data not shown).
As discussed by Daigle and Li ( Daigle and Li, 1993), APP, APP-like proteins and APL-1 share high sequence similarity at the level of three different protein regions: two of them are located in the extracytosolic domain and contain the Cys-rich regions, whereas the third region of high similarity resides in the short, cytosolic domain. We have calculated, in the last 32 amino acids of the APL-1 cytodomain, corresponding to the APP region necessary for the interaction with Fe65s and other PTB-containing adaptors, that 60% of the residues are identical to human APP; this percentage gets to 75% if we consider the conserved substitutions occurring in this region between human and nematode proteins. This elevated degree of similarity suggests that the functional conservation of the APL-1 cytodomain in the nematode system, and the evidences provided in this study, confirms, both at the biochemical and at the functional level, this possibility.
A 2.35 kb region present in the 5′ region of feh-1 is able to direct a strong expression of the gene in various neurons throughout the body axis and in the pharynx. In the latter case, feh-1 expression is present both in muscular and nerve cells. In mammals, Fe65 and related genes show distinct expression patterns; in fact, Fe65 is highly expressed in the central and peripheral nervous systems ( Duilio et al., 1991; Simeone et al., 1994), whereas Fe65-L1 ( Guènette et al., 1996) and Fe65-L2 ( Duilio et al., 1998) show a broader expression in several organs. Fe65-L1 transcripts are particularly abundant in skeletal and cardiac muscles, as well as in the brain and kidney, whereas the Fe65-L2 gene is highly expressed in the brain and testis. In the nematode, the expression of feh-1 in the nervous system and in the pharynx, which is a neuromuscular organ, is in agreement with the complex expression of the three genes found in mammals, in which, however, the nervous system is a preferential site of expression.
On the basis of the functional redundancy of Fe65 gene products in mammals, we undertook reverse genetics approaches to reveal the functional roles of feh-1 in the nematode. We have shown, by analysis of both feh-1-null and RNA-interfered worms, that the absence or the marked reduction of FEH-1 is not compatible with proper development of C. elegans. Embryonic arrest phenotype results in severe developmental defects, leading to disorganised embryos, from which cells either detach or degenerate. One possible explanation for this evidence may be interpreted as owing to cell adhesion defects, although the complexity of this phenotype does not allow us to precisely interpret its molecular basis. The larval arrest phenotype has been characterised: the arrested worms were impaired in larval development because of the absence, or the severe reduction, of pharyngeal activity. The increased pharyngeal activity shown by the worms that were heterozygous for the deleted allele, or worms with milder reduction of feh-1 transcripts, obtained through RNAi by feeding, strongly associate feh-1 function with the control of the rate of pharyngeal contractions. Contraction of the pharynx results from the coordinated activity of its muscle cells, and the rate of this activity (pumping rate) is set by a cellular system functioning as pace-maker. An explanation of why reduction of FEH-1 function increases the rate of pumping, whereas complete loss results in no pumping, must await the elucidation of the precise role of FEH-1 in pumping. At present, a simple hypothesis to explain this apparent paradox is that FEH-1 is involved in negatively modulating the rate of pumping. In the complete absence of FEH-1, the rate would be set so high that, because of the intrinsic properties of its muscle cells, coordinated contraction of the pharynx is not possible anymore and therefore results in no visible or functional pumping.
The results obtained by RNA interference by feeding with either feh-1 or apl-1 dsRNAs combined with the documentation of the physical interaction between FEH-1 and APL-1, strongly suggest that the products of these genes are involved in a common molecular pathway, which controls pharyngeal pumping in C. elegans.
Many ion channels regulating the electrical events necessary for the activity of this neuromuscular organ have been identified ( Fleischhauer et al., 2000; Davis et al., 1999; Dent et al., 1997; Maryon et al., 1998). In addition, products of genes that regulate vesicular trafficking at the level of the neuromuscular junctions have also been shown to be important for pharyngeal function ( Nonet et al., 1998). The mechanisms through which FEH-1 and APL-1 influence the contraction rate of the pharynx remain to be understood. One potential mechanism could act by regulating the proper localisation of ion channels at the neuromuscular junction, but other possibilities can be hypothesised, such as a possible role of the FEH-1/APL-1 complex in determining the correct events of vesicle recycling at the pre-synaptic compartment. Interestingly, a phenotype similar to that obtained with feh-1 mutants has been shown in worms bearing a gain-of-function mutation in exp-2, a gene coding for a Kv-type ion channel. In this case, heterozygous worms presented an elevated rhythm of pharyngeal contractions and hyperactive head movements, whereas worms homozygous for the gain-of-function allele died as L1 larvae because of feeding abnormalities ( Davis et al., 1999). Recent results suggest two different and apparently unrelated roles for the FE65-APP complex. One concerns the possible involvement of mammalian Fe65 and APP in the regulation of gene expression. In fact, we have previously demonstrated that Fe65 is efficiently translocated into the nucleus and that APP functions as a cytosolic anchor for Fe65 ( Minopoli et al., 2001). These observations suggested a role for the APP processing machines in regulating the nuclear translocation of Fe65; it could affect gene transcription by interacting with transcriptional factors, such as CP2/LSF/LBP1 ( Zambrano et al., 1998) or with histone acetyl transferase Tip60 ( Cao and Sudhof, 2001). It is not excluded that a similar scenario may be also acting in C. elegans, as presenilin orthologues, which might be responsible for proteolytic processing of APL-1, are present in the nematode, where they control a well-defined pathway centred on the maturation of Notch membrane proteins, whose cytosolic domain is translocated to the nucleus and regulates gene transcription ( Kimble and Simpson, 1997; De Strooper et al., 1999; Struhl and Greenwald, 1999). The availability of mutant worms null for feh-1 will represent an ideal tool to identify the array of genes regulated by FEH-1. On the other hand, it has been recently discovered that the Fe65-APP complex is involved in the regulation of cell movement ( Sabo et al., 2001), and unpublished observations have been mentioned, concerning a possible role of APP and Fe65 in neuronal growth cone. This observation deserves further experiments to analyse the possible role of FEH-1 and APL-1 in the regulation of neural cell movements, which can be affected in mutant worms.
We thank Y. Kohara for yk423e6 EST clone, M.C. Hresko and R.H. Waterston for the MH27 antibody, A. Fire for pPD95.75 and pPD21.28 vectors, L. Timmons for the pPD129.36 vector and HT115(DE3) strain, J. McGhee for the elt-2::GFP construct, the members of Genetic Toolkit Project (NIH/NCRR) for dpy-1(s2170) strain, M. R. Sapio for help with mutagenesis, A. Croce for help with immunohistochemical analysis and A. Sollo for technical support.
This work was supported by grants from MURST-PRIN 2000, to N.Z., the V Framework program (Contract QLK6-1999-02238), the CNR-Italy `Programma Biotecnologie MURST L95/95' and `Progetto strategico: Basi biologiche delle malattie degenerative del sistema nervoso centrale' and Biogem, to T.R., the Associazione Italiana per la Ricerca sul Cancro, to P.B. and the Programma MURST, Cluster 02, to I.I.G.B.
- Accepted January 3, 2002.
- © The Company of Biologists Limited 2002