INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Clathrin-coated pits and vesicles are ubiquitous organelles found in all the eukaryotic cells that mediate intracellular protein trafficking (Keen, 1990, Robinson, 1994, Hirst and Robinson, 1998). Clathrin-coated vesicles are composed of membrane fraction, selected membrane proteins, clathrin, and clathrin-associated proteins (APs). While clathrin is a structural unit common to all the clathrin-coated vesicles, APs can vary depending on the localization of the vesicles at the cellular and subcellular level (for example, Ahle et al., 1988, Dell'Angelica et al., 1997), and are thought to be important in selecting cargoes in the vesicles. There are four AP complexes identified so far in various species. All four complexes are similar in their composition and structure in that they are hetero-tetramers of two large chains, one small chain, and one medium chain. The medium chains of the clathrin AP complexes are known to interact with the tyrosine or dileucine residues of their cargo proteins (Ohno et al., 1995, Rodionov and Bakke, 1998, Hofmann et al., 1999). AP-1 complex contains 1 and  adaptin as large chains, µ1A or µ1B as a medium chain, and 1 (AP19) as a small chain. AP-2 has - and 2 adaptin as large chains, µ2(AP50) as a medium chain, and 2 (AP17) as a small chain. AP-3 complex consists of 3A or 3B and  adaptins as large chains, µ3A or µ3B as medium chains, and 3 as a small chain (Dell'Angelica et al., 1997, Simpson et al., 1997). AP-4 complex is a recently identified complex that consists of 4 and  adaptins as large chains, µ4 as a medium chain, and 4 as a small chain (Dell'Angelica et al., 1999a, Hirst et al., 1999). Some of the large chains share some similarity in their amino acid sequence (for example, Kirchhausen et al., 1989, Robinson, 1989), as do those of medium chains and small chains (Kirchhausen et al., 1991, Nakai et al., 1993, Nakayama et al., 1991, Phan et al., 1994, Thurieau et al., 1988). The localization of the AP-1 and AP-2 complexes is well known. AP-1 is at the trans-Golgi, AP-2 on the plasma membrane. The medium chains of the mammalian AP-1 complexes show tissue-specific expression (Ohno et al., 1999). AP-3 complexes were reported to be present in most cells, but some components of the AP-3 complexes were tissue-specific (Dell'Angelica et al., 1997, Pevsner et al., 1994). AP-4 was associated with the trans-Golgi network or with an adjacent structure in all cell types (Dell'Angelica et al., 1999a, Hirst et al., 1999).

Genetic analysis of medium chains has been reported in many systems. In yeast, mutations in µ1 is known to enhance the temperature-sensitive growth phenotype and the -factor processing defect caused by a temperature-sensitive allele of the clathrin heavy chain gene (Stepp et al., 1995). AP-3 is necessary for proper sorting of vacuolar alkaline phosphatase (Stepp et al., 1997). These phenotypes are mild in terms of viability of the organism. In the mammalian system, two µ1 chains have been characterized; one of them, µ1B, is expressed in epithelial cells and is required for basolateral targeting in these cells (Folsch et al., 1999, Ohno et al., 1999). -Adaptin was essential for embryonic development in mice by analyzing knockout mice (Zizioli et al., 1999). 3A was mutated in patients with human Hermansky-Pudlak syndrome (HPS; Dell'Angelica et al., 1999b) and in the mouse hypopigmentation mutant pearl (Feng et al., 1999), indicating that AP-3 functions in protein sorting to lysosomes. In the nematode C. elegans, one medium chain of AP-1, encoded by unc-101, was identified (Lee et al., 1994). Genetic analysis of unc-101 showed that mutations in this gene caused pleiotropic effects, including subviability, uncoordinated movement, a defect in neuronal dye uptake, male spicule defect, defecation defect, and suppression of a reduction-of-function mutation in the epidermal growth factor receptor (let-23EGFR) gene (Lee et al., 1994). Interestingly, putative null mutations of unc-101 do not cause 100% lethality, but only 50% lethality50% of progeny from a homozygous mother of unc-101/unc-101 genotype will survive(Lee et al., 1994). This suggests that another gene, encoding the medium chain of AP-1, can replace the essential function of unc-101 in individuals with defects in unc-101. Consistent with this, unc-101 mutations do not cause vulval defect, while unc-101 mutations have been isolated as extragenic suppressors of the let-23(sy1) vulvaless phenotype. In vulval tissues, there must therefore exist other genes that act redundantly with unc-101 as negative regulators. One such gene is sli-1. An unc-101; sli-1 double mutant displays greater-than-wild-type vulval induction (Sternberg et al., 1994). It is conceivable that there may be more redundant negative regulators of vulval development, because not all unc-101;sli-1 double mutants show multivulval phenotypes (Sternberg et al., 1994).

Here we report identification of a homolog of unc-101, which we named apm-1 (associated protein complex medium chain-1). We show that unc-101 and apm-1 are both ubiquitously expressed throughout development and that apm-1 plays redundant roles with unc-101 in embryogenesis and vulval development. We show that apm-1 also has distinct functions from those of unc-101. Finally, we report characterization of the functions of the two medium chains and compare them to those of the other AP-1 complex subunits, 1, 1, and .

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Strains and Culture

The Bristol strain N2 was used as the standard wild-type strain. The mutations used for apm-1 and unc-101 functional analysis are unc-101(sy108) (Lee et al., 1994) and let-23(sy1) (Aroian et al., 1990). rol-6(su1006) DNA and dpy-20(e1282) was used as selection markers for DNA microinjection as described below. The culture of C. elegans was previously described (Brenner, 1974).

cDNA Screening

We used the cDNA clone CEED20 (accession number T00259) from the GenBank database as a start point for cloning a full-length cDNA clones. We used the CEED20 DNA as probe in a cDNA screening for full-length clones. We used a standard hybridization procedure (Sambrook et al., 1989). We isolated three cDNA clones from a C. elegans cDNA library (Barstead and Waterston, 1989), all of which contained inserts of the same length. We determined the sequence of one of the clones, CEED20-3. Sequencing reactions were performed using Sequenase 2.0 and reagents from United States Biochemical (Cleveland, OH).

Sequence Analysis

Compiling of DNA and amino acid sequences were carried out using the Macvector program (IBI, Oxford Molecular Group, Hunt Valley, MD) and the GCG package v7.0, a software of the Genetics Computer Group (Madison, WI; Devereux et al., 1984). The BLAST program of the GCG software was used to search and compare homologies of the sequences. The Pileup and Gap programs were used to generate the comparisons of the amino acid sequences. The clustal w program was used to analyze and calculate genetic distances among the medium chain homologs. Tree view 1.5 (Win 32) version 1.5.2. (Page, 1998) was used to construct the dendrogram of the medium chain homologs. The sequences that were used in the sequence comparison and dendrogram were from C. elegans (CAPM-1, in this study; UNC-101, CAPM-2, Lee et al., 1994, CAPM-3, Cosmid F53H8.1, C. elegans genome project), humans(µ1B, NP  005489.1; µ2, sp P20172; µ3A, gb AAD43328.1; µ3B, sp P53677; µ4, gb AAD43328.1), mouse(µ1B, gb AAD28085.1; AP47, sp P35585), rat (p47A, sp P53676; p47B, sp P53678), fly(APM-1, emb CAA06918.1; APM-2, emb CAA06785.1; APM-3, emb CAA08768.1), and from yeast(YAP54, sp Q00776; APM-2, sp P38700; APM-3, emb CAA97989.1; APM-4, sp Q99186).

Expression Studies

To construct an unc-101 GFP reporter gene, we used the pJL1 plasmid and the vector pPD95.77 from Andy Fire (Carnegie Institute of Washington, Baltimore, MD). We amplified unc-101 genomic DNA from the K11D10 cosmid, using two PCR primers, and produced the pJL271 plasmid by replacing the unc-101 genomic region of pJL1 with PCR product. The 5' and 3' subcloning sites were HindIII and BamHI. The two PCR primers were K11-1, 5' CTCGTCGACCTGAT CGGTGTGC 3' and 101-C, 5' GGGATCCGTATTCTCCATTTTGAG 3'. Next, the 1.8 kb gfp fragment from pPD95.77 was subcloned into pJL271. The 5' and 3' subcloning sites were BamHI and SpeI. To construct the apm-1 GFP reporter, we subcloned amplified 6.0 kb genomic DNA using two PCR primers into the Fire vector pPD95.79. The subcloning sites were SalI and BamHI. The two PCR primers were F55-1, 5' GTGAAACTGCTGAAGGAAGC 3' and CE19, 5' GGGATCCTCATTTGATAATCTCCG 3'.

Construction of Hybrid Genes

Construction of the unc-101 hybrid genes was described (Lee et al., 1994). To construct an APM-1 hybrid gene, we amplified the APM-1 cDNA from nucleotide #325 through #1229 using two PCR primers. Both the 5' and 3' subcloning sites, NruI and EcoRV, are conserved in APM-1. The two PCR primers are: CE-6, 5'CGATAATTTCGTTATTATTTA TG3', and CE-7, 5'ATCCAGATTTCTCTATGAT TTT3'. The amplified DNA was ligated to the 7.2kb NruI/EcoRV fragment of pJL2. The resulting plasmid is the APM-1 hybrid gene. This construct contains the 5' promoter region of unc-101, the 5' coding region of unc-101 up to the unc-101 cDNA nucleotide #388, the apm-1 cDNA from NruI site to EcoRV site (corresponding nucleotides in unc-101 are #389 to #1281, the unc-101 3' region from #1282 to the end of cDNA, and the untranscribed 3' region of unc-101. The predicted protein from this construct contains 301 amino acid residues from APM-1, and 123 amino acid residues from UNC-101. A positive control construct, the unc-101 hybrid gene, contained all amino acids for UNC-101. A negative control construct contains only 123 amino acids from UNC-101, and the remaining amino acids from APM-2.

Microinjection Experiments and Double Strand RNA Interference

Microinjection of DNA into the gonad of C. elegans hermaphrodite adults was previously described (Mello et al., 1991). For expression studies, the GFP reporter constructs were coinjected into N2 wild-type animals with the pRF4 plasmid containing a dominant mutant gene for rol-6. The total concentration of injected DNA was 140 µg/ml (reporter 100 µg/ml and pRF4 40 µg/ml). The transgenic animals were selected by their rolling behaviors, and the animals were observed for GFP expression. For hybrid constructs, we used unc-101(sy108); let-23(sy1); dpy-20(e1282) animals as the host for microinjection. We coinjected the hybrid genes with a dpy-20(+) clone as a selection marker. The host animals have a dumpy body shape (Dpy phenotype), and the selection marker DNA can rescue the Dpy phenotype to wild-type body shape, enabling the selection of transgenic animals containing the microinjected genes. Following microinjection, we selected nonDpy transgenic animals, established stable lines that inherited the transgenes, and examined the phenotype of vulval differentiation. For RNAi, templates for RNA synthesis were produced by PCR amplification of full-length cDNA using T3 and T7 primers. RNAs were synthesized using a commercially available in vitro transcription kit (Promega, cat.#P2075, P2083) with T3 and T7 RNA polymerases. Unmodified RNA was resuspended for injection at 10 µg/ml to 500 µg/ml concentration in DEPC-treated water. Following microinjection of double strand RNA, the injected animals (P0) were transferred to new plates every 12 h and F1 progeny were counted and analyzed.

Microscopy

Differential interference contrast (DIC) microscopy (Nomarski optics) and fluorescence microscopy were used to observe the phenotypes and expression patterns. For DIC microscopy, we treated the worms with sodium azide at 1 mM concentration; for fluorescence microscopy, we treated the worms with levamisole at 100 ng/ml.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Molecular cloning of apm-1, a homolog of the unc-101 gene, encoding a medium chain of trans-Golgi clathrin-associated protein complex AP-1

A C. elegans cDNA sequencing project identified a cDNA clone (CEED20) containing a partial sequence similar to unc-101 (accession number T00259). We screened a cDNA library with probes made from the cDNA of CEED20. We isolated three full-length cDNA clones and determined the sequence of one of them. The sequence would encode a putative protein of 426 amino acids. We named this gene apm-1 (associated protein complex medium chain-1). Comparison of the sequence of APM-1 with other medium chain homologs indicated that this protein is more related to AP47 than to AP50 (Figure 1A). APM-1 has the same degree of similarity to mammalian AP47 as it has to UNC-101 in C. elegans (72% identity in both cases). Comparison of the amino acid sequences among the medium chain homologs from yeast to humans showed that APM-1 and UNC-101 are clearly grouped within the same subfamily of AP-1 medium chains (Figure 1B). DNA sequence comparison between apm-1 and unc-101 showed that the discrepancies are biased toward the third bases of codons (our unpublished results). There was minimal sequence identity in the 5' nontranslated region or in the 3' nontranslated region (our unpublished results), suggesting that these two genes might be subject to different types of regulation. The genome project later revealed that the genomic clone K11D2 (accession number Z83115) contained the full length apm-1 gene. A comparison of the genomic structures of the two genes showed that the boundaries of the first three exons and the last exon are conserved between the two genes, but the boundaries of the central exons are divergent (Figure 1C). The number of exons is also different: apm-1 has nine exons while unc-101 has only seven exons.

View larger version (105K): [in this window] [in a new window]   Figure 1.   (A) Comparison of amino acid sequences of APM-1, UNC-101, APM-2, and mouse AP47. The amino acids that are conserved in the three APM-1 homologs are in bold and shaded letters, and the asterisks underneath the sequences represent the amino acids conserved in all four proteins. Some of the amino acids are conserved in all four proteins, but others are conserved only in the AP-1 medium chains. (B) Diagram showing the evolutionary relationship among medium chains of clathrin AP complexes. Medium chains of all four types of complexes are compared. The circle on the center indicates a hypothetical ancestral medium chain gene. The prefix C indicates the sequence is from C. elegans, H from humans, R from rat, M from mouse, D from Drosophila, Y from yeast. For example, CAPM-1 is C. elegans APM-1 and DAPM-1 is the Drosophila APM-1 homolog. This dendrogram clearly shows that APM-1 is a member of the AP-1 medium chain family, together with UNC-101. (C) Genomic structures of the apm-1 and unc-101 genes. Exons are drawn as thick lines, and introns are drawn as thin lines outside the exon structures. Introns are not drawn in scale. The numbers indicate the numbers of nucleotides in each exon or intron.

apm-1 and unc-101 Are Expressed Ubiquitously throughout Development

To determine the expression patterns of the medium chains, we constructed GFP reporter constructs (see MATERIALS AND METHODS) and examined the GFP expression patterns in transgenic animals containing these reporter constructs. We found that unc-101 was expressed in most cells, if not all, at most embryonic and postembryonic stages. The highest level of expression was observed in muscles and pharyngeal regions (Figure 2B, G, K). While apm-1 was also expressed ubiquitously throughout development, its expression was stronger in neurons, as demonstrated by the bright fluorescence in the nerve ring (Figure 2, D, H, L). When examining the apm-1 expression by lacZ reporter assays, we also observed high expression of apm-1 in the intestine (our unpublished results). The expression of both genes was strong in the vulval cells when vulval tissues were undergoing morphogenesis (Figure 2, M-P). The overlapping expression pattern of apm-1 and unc-101 implied that these two medium chains were present in the same cells, possibly interacting with both shared and distinct cargo proteins during transport.

View larger version (61K): [in this window] [in a new window]   Figure 2.   Expression patterns of unc-101 and apm-1. The two genes are expressed ubiquitously throughout development. The left panels show expression patterns of unc-101 and the right panels those of apm-1. (A-D) show embryos consisting of ~ 100 cells, (E-H), animals at their L1 stage, (I-L), the head region of animals at their L4 stage, and (M-P) show the vulval regions of L4 stage animals. A, C, M, and N are Nomarski images of the embryos; E, F, I, and J are DAPI staining images showing locations of the nuclei. All others show GFP expression. Both genes are strongly expressed during embryogenesis and at the L1 stage. At the L4 stage, unc-101 is strongly expressed in the pharynx and in the vulval cells, and apm-1 is strongly expressed in the nerve ring, as well as in the vulval cells. The arrows in (M-P) are the invaginations of the vulvae. The scale bar is 10 µm.

RNAi of the Medium Chain Genes Specifically Phenocopies Reduction-of-Function Mutations of Corresponding Genes

To dissect the functions of apm-1 compared with those of unc-101, as well as other components of the AP-1 complex, we wanted to examine the phenotypes associated with disruption of apm-1 function. As there is no available mutation in the apm-1 gene, we used double-strand RNAi to phenocopy apm-1 reduction-of-function mutations. As APM-1 and UNC-101 share approximately 72% identity at the amino acid level, the possibility existed that the RNAi of one of these genes might interfere with the function of the other gene. We examined whether double-strand RNAs of apm-1 and unc-101 specifically and exclusively interfered with their respective targets. As shown in Figure 3, unc-101 RNAi caused reduction of GFP expression driven by the unc-101 gene while it did not cause any reduction in GFP expression driven by the apm-1 gene (Figure 3, C and D). Likewise, apm-1 RNAi interfered specifically with apm-1, not with unc-101 (Figure 3, E and F). In a control experiment, RNAi of apm-2, the medium chain of AP-2 complexes in the nematode, did not cause any reduction of either apm-1 or unc-101 expression (Figure 3, G and H). We therefore concluded that we could specifically phenocopy reduction-of-function mutations of either of these two medium chain genes by RNAi using the complementary RNAi or phenocopy double mutations of apm-1 and unc-101 by using double-strand RNAs of both genes simultaneously.

View larger version (117K): [in this window] [in a new window]   Figure 3.   Specific RNAi effect of unc-101and apm-1. The left panels show animals expressing the UNC-101::GFP reporter protein, and the right panels show animals expressing the APM-1::GFP reporter protein. (A and B) wild-type expression of unc-101 and apm-1, respectively. (C and D) GFP reporter expression after RNAi of unc-101. Only UNC-101::GFP expression is reduced, while APM-1::GFP is not affected. (E and F) GFP reporter expression after RNAi of apm-1. Only APM-1::GFP expression is reduced while UNC-101::GFP is not affected. (G and H) GFP reporter expression after RNAi of apm-2. APM-1::GFP and UNC-101::GFP expressions are not affected by apm-2 RNAi. The Dpy phenotype reportedly caused by disruption of apm-2 gene function is obvious as the animals are shorter and fatter (Shim and Lee, 2000).

Disruption of apm-1 Alone Causes Larval Lethality, and Simultaneous Disruption of apm-1 and unc-101 Causes Embryonic Lethality

We first examined the function of apm-1 during early development by RNAi. In a control experiment, we injected unc-101 double-strand RNA into N2 wild-type animals. At a high concentration (500 µg/ml), double-strand RNA caused, at most, 50% larval lethality compared with the putative null mutation of unc-101(sy108) animals, confirming that the null phenotype of unc-101 would not lead to 100% lethality. The animals showed arrest development at various stages of development. Most of the surviving animals showed the uncoordinated (Unc) phenotype (as is the case in unc-101 mutant animals) and did not show any embryonic lethality (Figure 4A). We then injected a high concentration of apm-1 double-strand RNA (200 µg/ml) into N2 animals and examined the phenotypes of the F1 progeny. We found that 100% of the F1 animals showed arrested development as L1 larvae (Table 1, Figure 4B); also, they did not show any embryonic lethality. Most animals displayed the typical phenotype consisting of movement at first after hatching, displacement over a short distance, then sudden arrest of movement in any direction. In addition, they had bloated anterior intestines, as if pumping was normal but that the ingested bacterial stream had stopped at the anterior intestine. The arrested L1 animals had very slow pumping motions with head and tail moving very little. The shape of intestine was abnormal in the animals with arrested development, with exaggerated curvature and uneven thickness of the intestine tubes. In addition, animals with arrested development had thinner posterior bodies than wild-type animals. Next, we injected the same concentration of apm-1 double-strand RNA into unc-101(sy108) animals to phenocopy double mutations of apm-1 and unc-101 and found that the F1 animals showed up to 100% embryonic lethality due to arrested development at the twofold stage within the eggshells (Table 2, Figure 4C). Thus, removal of both of the AP-1 medium chains leads to a synthetic embryonic lethal phenotype that cannot be achieved by the removal of either one of the two genes. This indicates that one of the two genes must be present for embryonic development, but that either one of the two genes is redundant in the presence of the other. In contrast to the redundant functions of apm-1 and unc-101 in embryogenesis, each medium chain may play a distinct role after hatching, because most animals in which only one of the two genes was disrupted could hatch but could not survive after the early larval stages.

View larger version (65K): [in this window] [in a new window]   Figure 4.   Redundancy of unc-101 and apm-1 during embryogenesis. Lethal phenotypes caused by RNAi of unc-101, apm-1, 1, 1, or . (A) F1 progeny from a wild-type animal injected with unc-101 double-strand RNA. About half of the animals arrested at various stages of development, and this specific figure shows an animal arrested at the L1 stage. (B) F1 progeny from a wild-type animal injected with apm-1 double-strand RNA. All the animals arrested as L1 larvae, and the anterior intestine was bloated, suggesting that, although pharyngeal pumping was not affected, the bacterial stream was blocked due to some malfunction or structural defect in the intestine. L1 animals arrested by apm-1 RNAi have a slightly different posture when they arrest. This is because they can move and feed on bacteria for a short period of time before suddenly stopping. On the other hand, arrested L1 animals of unc-101 mutants hardly move, if at all, and arrest in a straight posture. (C) F1 progeny from an unc-101(sy108) animal injected with apm-1 double-strand RNA. All the animals arrested as embryos, as shown in this figure. (D) F1 progeny from an N2 animal injected with 1 double-strand RNA. (E) F1 animal from a wild-type animal injected with 1 double-strand RNA. (F) F1 animal from a wild-type animal injected with  double-strand RNA. As can be easily noticed in (C-F), the phenotypes caused by double RNAi of both apm-1 and unc-101 are identical to those caused by single RNAi of 1, 1, or  chain gene. The scale bar is 10 µm.

apm-1 and unc-101 Are Redundant Negative Regulators of an EGF-Mediated Signaling Pathway

Unc-101 mutations suppress the vulvaless phenotype of let-23 EGFR reduction-of-function mutations, indicating that the normal function of unc-101 might be negative regulation of the EGF signaling pathway in vulval development (Lee et al., 1994). Because unc-101 single mutant animals do not show any abnormal vulval induction (Lee et al., 1994), the existence of redundant negative regulators of vulval development, acting in parallel with unc-101, has been suggested. One such gene is sli-1 (Yoon et al., 1995). Single mutants of either sli-1 or unc-101 do not exhibit any vulval phenotype, but double mutants for unc-101 and sli-1 have greater-than-wild-type vulval induction (Sternberg et al., 1994). We wished to examine if apm-1 was another redundant negative regulator of the signaling mediated by LET-23 EGFR.

We have previously shown that a hybrid construct, which contained the unc-101 promoter region and most of the mammalian AP47, could rescue phenotypes associated with an unc-101 mutation when introduced into the mutant animals by microinjection (Lee et al., 1994). To test whether APM-1 protein could complement functions of UNC-101 protein when overexpressed, we constructed and examined a hybrid gene (see MATERIALS AND METHODS). In the hybrid gene, we substituted two thirds of the UNC-101 protein from the C terminal end with APM-1. The rationale for this substitution was that the N-terminal region of the medium chains was required for interaction with the  chain, while the C-terminal region was important for interaction with target proteins(Aguilar et al., 1997). We comicroinjected this hybrid gene into the gonads of unc-101(sy108); let-23(sy1); dpy-20(e1284) animals with cloned DNA of dpy-20(+) as a selection marker. The resulting nonDpy transgenic animals were examined for their ability to complement the unc-101 mutations that result in suppression of the vulvaless phenotype of the let-23(sy1) mutation. The suppression of the vulvaless phenotype of the let-23(sy1) mutation by unc-101 mutations was rescued in the transgenic animals containing extra copies of the hybrid gene as 4 out of 8 transgenic animals at the L3 molt stage, observed under Nomarski optics, restored the vulvaless phenotype (Figure 5E). Our results indicate that overexpressed APM-1 may complement the functions of UNC-101 in the absence of functional, endogenous UNC-101 in the vulval cells.

View larger version (164K): [in this window] [in a new window]   Figure 5.   An APM-1 hybrid protein can complement UNC-101 proteins in vulval development: The apm-1 hybrid gene can rescue the suppression of vulvaless phenotype of unc-101(sy108). The animals in the photographs are at the L3 molt stage, which corresponds to the period of vulval induction. The arrowheads indicate the locations of the anchor cell that induces the vulval precursor cells (VPCs). The lines indicate the lineage of each VPC. The genotypes are: (A) wild-type; (B) let-23(sy1); (C) unc-101(sy108); let-23(sy1); (D) unc-101(sy108); let-23(sy1)::unc-101 minigene; (E) unc-101(sy108);let-23(sy1)::apm-1 hybrid gene; and (F) unc-101(sy108);let-23(sy1)::apm-2 hybrid gene. (A), wild-type vulval induction in which three vulval precursor cells (VPCs) have divided twice to make four progeny each. (B) There is no vulval induction because of the reduction-of-function mutation in the let-23EGFR gene. All VPCs have divided only once. (C) The unc-101 mutation suppresses the let-23(sy1) vulvaless mutant phenotype to greater-than-wild-type vulval induction. Four VPCs, instead of three VPCs as in normal induction, have divided twice. (D) The transgenic unc-101 minigene rescued the endogenous unc-101 (sy108) mutation, and thus no vulval induction occurred, as in the single mutant of let-23(sy1). (E) The apm-1 hybrid gene complemented the endogenous unc-101 (sy108) mutation, thus no vulval induction occurred, as in the single mutant of let-23(sy1). (F) Four VPCs were induced because the apm-2 hybrid gene could not complement the endogenous unc-101 (sy108) mutation. This phenotype is identical to that of unc-101(sy108); let-23(sy1) mutant animals. The scale bar is 10 µm.

We then asked whether apm-1 would act redundantly in the vulval induction pathway by examining the vulval phenotype of unc-101(sy108) animals injected with low concentrations of apm-1 double strand RNA. If the two genes are indeed redundant negative regulators of the pathway in vivo, we would expect to see the greater-than-wild-type vulval induction observed in unc-101; sli-1 double mutants. We used low concentrations of apm-1 double strand RNA for this experiment since a high concentration leads to 100% embryonic lethality. Out of 67 F1 postRNAi survivors, 15 animals showed greater-than-wild-type induction as observed using Nomarski optics (for example, Figure 6B), whereas unc-101(sy108) animals without the RNAi of apm-1 did not show increased induction compared with the wild-type (Figure 6A), indicating that apm-1 indeed acts redundantly with unc-101 in negatively regulating the vulval induction pathway. Next, we wished to determine whether the reduced function of apm-1 can directly suppress the vulvaless phenotype caused by the let-23(sy1) mutation even in the presence of the wild-type gene activity of unc-101. We injected double strand RNA of apm-1 into let-23(sy1) animals at low concentration. We found that at 20 µg/ml microinjected double strand RNA, 73% of F1 postRNAi survivors (n = 19) were suppressed for the vulvaless phenotype, thus exhibiting morphologically wild-type or greater-than-wild-type vulvae (for example, Figure 6C). 67% of F1 postRNAi survivors (n = 12) had functional vulvae, with which they could lay eggs. In contrast, in the let-23(sy1) animals without RNAi, only 10% showed morphologically wild-type vulvae, and 9% had functional vulvae (n = 231 and 222, respectively). These data indicated that the apm-1 gene acts as another negative regulator of the vulval induction pathway. Surprisingly, we also observed greater-than-wild-type induction of vulval precursor cells (VPCs) in transgenic N2 animals containing an apm-1::GFP reporter construct (Figure 6D), indicating that this reporter construct may have acted as a dominant negative mutation and may have interfered with the wild-type function of apm-1. This would have caused the VPCs to be induced excessively compared with wild-type. Thus, we infer that apm-1 and unc-101 are redundant negative regulators of the vulval induction pathway in the nematode.

View larger version (134K): [in this window] [in a new window]   Figure 6.   Redundancy of unc-101 and apm-1 in vulval development. The arrows indicate normal invagination of vulval tissues at the L4 stage in all figures. (A) An unc-101(sy108) animal showing wild-type vulval induction as indicated by a single invagination of the vulval tissue composed of the progeny of three VPCs. (B) An unc-101(sy108) animal injected with apm-1 double-strand RNA. An additional invagination by vulval cells, due to greater-than-wild-type induction, is indicated by an arrowhead. (C) Alet-23(sy1) animal injected with apm-1 double strand RNA. As in (B) an extra invagination is shown, as indicated by an arrowhead. (D) A transgenic animal of wild-type genetic background containing the apm-1::GFP reporter construct. An extra invagination is shown next to the normal vulval invagination. The scale bar is 10 µm.

The Two Medium Chains Are Shared by Other Components of AP-1 Complex in the Nematode

Thorough database searches failed to identity more than one homolog each for the large and small chains of AP-1 complexes in the nematode. We identified only single homologs of 1, 1, and  chain genes. To determine if the two medium chains, APM-1 and UNC-101, share other components when constituting the AP-1 complexes, we examined the phenotypes caused by RNAi of other components of the AP-1 complex. The F1 progeny from wild-type animals injected with double-strand RNA of the small chain 1 gene showed 100% embryonic lethality at the concentration of 100 µg/ml (table 3, Figure 4D). At lower concentration, development of most F1 progeny was arrested at the embryonic stage or at the larval stages (our unpublished results). RNAi of the 1 or  chain displayed identical phenotypes (our unpublished results and Figure 4, E and F). These results clearly imply that the removal of the single small chain of the AP-1 complex caused phenotypes as severe as those caused by the removal of both of the medium chains.

View this table: [in this window] [in a new window]   Table 3.  Embryonic lethality caused by RNAi of 1 in the wild type background

To further confirm the result, we examined the effect of RNAi of 1, the small chain of the AP-1 complex, on the expression patterns of apm-1 and unc-101. It has been reported that loss of one component of the clathrin-associated protein complex led to destabilization of the whole complex (Dell'Angelica et al., 1999b, Zizioli et al., 1999). We reasoned that the stability of both APM-1 and UNC-101 proteins would be decreased by loss of one of the components of the AP-1 complex if they indeed shared this component. We microinjected low concentration of 1 double-strand RNA into transgenic animals containing either an APM-1::GFP or an UNC-101::GFP reporter gene and obtained RNAi-affected embryos and animals at different larval stages. RNAi of 1 caused reduction in expression of both apm-1 and unc-101 genes in embryos and larvae (Figure 7, F, H, K, L), while 2 RNAi did not result in any difference of expression (Figure 7, M and N). This result indicates that the disruption of 1 destabilized both AP-1 complexes containing APM-1 or UNC-101 as their medium chain. It is therefore conceivable that a 1-containing AP-1 complex can have either APM-1 or UNC-101 as its medium chain.

View larger version (95K): [in this window] [in a new window]   Figure 7.   RNAi of 1 destabilizes both the UNC-101 and APM-1 proteins. The left panels show UNC-101::GFP expression and the right panels show APM-1::GFP expression. (A-H) show animals at their 2-3 fold embryonic stages, and (I-N) show animals at their L1 stage. A, C, E, G, I, and J are Nomarski images and all the others are GFP fluorescence images. (A-D) show the wild-type expression patterns of unc-101 and apm-1, and (E-L) show the results of RNAi of the 1 gene. Both UNC-101::GFP and APM-1::GFP expression are reduced at the embryonic stage and the L1 stage by the RNAi as shown in F, H, K, and L. On the contrary, RNAi of 2, the AP-2 small chain gene, neither reduced UNC-101::GFP nor APM-1::GFP expression as shown in M and N. Note that the nerve ring still contains APM-1::GFP expression even after RNAi of 1 in L.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In this article, we have identified a second homolog of AP47, the medium chain of the trans-Golgi clathrin-associated protein complex AP-1. Our functional analysis suggested that, on their own, apm-1 and unc-101 are dispensable for embryonic development as single disruption of either gene did not cause any embryonic lethality. However, when both of these genes are disrupted, embryogenesis is affected, indicating that the combined functions of APM-1 and UNC-101 proteins are essential for embryogenesis. Our results also indicated that the nematode AP-1 complexes can contain either UNC-101 or APM-1 as their medium chains along with other components, probably serving both shared and distinct functions. We propose that the AP-1 complex in the nematode can employ either one of the two medium chains as its medium chain for cargo transport.

Comparison of the genomic DNA sequences of the apm-1 and unc-101 genes revealed interesting structural features. The boundaries of the first three exons and the last one exon are conserved in both genes, whereas exon/intron boundaries as well as the number of central introns vary. It is possible that the genes were initially duplicated during evolution and that the introns in the central part of the genes were introduced later, at different loci, contributing to novel, distinct functions for each gene.

Mutations in the unc-101 locus cause pleiotropic effects, suggesting that unc-101 is not equivalent to apm-1 and that unc-101 has distinct functions from those of apm-1. However, several previous observations suggested that unc-101 may not be fully distinct from apm-1. For example, the phenotypes of unc-101 putative null mutations are not identical in all animals although they bear the same mutations in unc-101. The lethality associated with the putative null mutations is not complete as only 50% of the progeny of homozygous hermaphrodites actually die. The defecation defect shows more variety even in a single animal model (Thomas, 1990). Each defecation cycle in C. elegans is composed of an anterior body muscle contraction (aBoc), a posterior body muscle contraction (pBoc), and an expulsion (Exp) step. In an unc-101 null mutant animal, the aBoc step is missing in half of the defecation cycles, while in the other half of the defecation cycles, the aBoc is normal. Therefore, it is conceivable that a gene may exist that shares partial redundancy with unc-101. We infer that apm-1 may be one such gene. In this report, we have shown that apm-1 and unc-101 on their own are redundant for embryonic development and for regulating an EGF-mediated signaling pathway.

Concerning the extent of redundancy between apm-1 and unc-101, one possibility is that unc-101 and apm-1 have identical functions, and that full expression of both the genes is required for production of a sufficient amount of proteins. This is unlikely for the following reasons. Mutations in the unc-101 locus were not dosage-dependent but were fully recessive, indicating that the loss of one copy of unc-101 does not cause any defect. However, one still cannot exclude the possibility that there exists a threshold level of expression of these genes required for their proper functioning. Another prediction of the above hypothesis is that the apm-1 mutant animals will have the same phenotype as unc-101 mutant animals, but this is not the case either. Animals affected by RNAi of apm-1 displayed different phenotypes from those with unc-101 mutations. A second possibility is that APM-1 and UNC-101 are expressed in different types of cells, although they have the same functions. This possibility is unlikely either, since our expression studies did not reveal any difference in the pattern of unc-101 and apm-1 expression throughout development (Figure 2). A third hypothesis, which we prefer, is that APM-1 acts as a counterpart of UNC-101 (or vice versa) within the context of the AP-1 trans-Golgi clathrin-associated protein complex, and it may interact with distinct sets of target proteins from those of UNC-101 depending on the tissue type and the stage of development. It is possible that apm-1 can somehow compensate for the UNC-101 function when unc-101 is mutated. If this were the case, single mutants of apm-1 would probably have different phenotypes from those of unc-101 and double mutants for unc-101 and apm-1 would be 100% lethal. As a matter of fact, RNAi of apm-1 in the unc-101 background caused 100% embryonic lethality, while RNAi of apm-1 in the wild-type background resulted in different phenotypes from those of unc-101 mutants. As our results clearly showed, RNAi of the small or large chain of the AP-1 complex (1, 1, and ) caused 100% embryonic lethality, indicating that UNC-101 and APM-1 function as two alternative forms of the medium chain in the AP-1 complex.

One interesting feature of apm-1 function is its role in vulval development, which is mediated by the LIN-3 (EGF)-LET-23 (EGFR) signaling pathway. We propose that apm-1 is another negative regulator of vulval development on the bases of the following observations. First, a hybrid protein, composed of two thirds of the apm-1 gene and one third of the unc-101 gene, can complement a defective UNC-101 protein function in vulval development, when expressed under the control of the unc-101 promoter. Second, reduction in apm-1 function can suppress the vulvaless phenotype of let-23 (sy1) even in the unc-101 (+) background. This is also observed when unc-101 mutations are introduced into the apm-1(+) background. Third, disruption of both apm-1 and unc-101 caused greater-than-wild-type vulval induction. The phenotypes observed after the disruption of different sets of genes, let-23, unc-101, and/or apm-1 are summarized in table 4. These results suggest that apm-1 shares redundant functions with unc-101 for normal vulval development in wild-type animals.

Why were apm-1 mutations not identified in the genetic screen for suppressors of the vulvaless phenotype of the let-23(sy1) mutation? A possible answer is that the screen was not applied extensively enough to identify a reduction-of-function allele of a gene whose null phenotype is complete lethality. Nevertheless, a putative unc-101 null mutation was isolated in a screen (Lee et al., 1994). The fact that the apm-1:GFP reporter construct, which presumably overexpressed APM-1::GFP in the wild-type background, caused greater-than-wild-type induction of VPCs indicates that apm-1 might play a more important role in the vulval induction pathway than unc-101, although apm-1 was not isolated by a conventional genetic screen.

It is now clear that apm-1 and unc-101 play redundant roles in embryogenesis and in vulval development. However, as a single disruption of apm-1 or unc-101 clearly showed, a decreased function in either gene led to distinct phenotypes and therefore, the apm-1 and unc-101 genes have distinct functions. As described in the RESULTS section, apm-1 RNAi animals were arrested at the L1 larval stage with abnormal anterior intestine bloating although the larvae seemed normal after hatching. This indicates that apm-1 gene activity is required even in the presence of wild-type unc-101 activity during larval development. The intestinal phenotype that appears only in apm-1 RNAi animals may imply that APM-1 function is important in intestine cells. A new µ1 gene (µ1B) has been identified in mouse and humans and recent studies showed that the µ1B medium chain is epithelial cell-specific and important for polarized transport of proteins(Ohno et al., 1999, Folsch et al., 1999). It is possible that APM-1 may likewise have important roles in polarized cells, such as in the intestinal cells of the nematode. Another distinct function of apm-1 and unc-101 was drawn from the behavioral phenotype associated with a single disruption of each gene. While apm-1 RNAi animals do not seem to be Unc (uncoordinated movement), as they move for a short time after hatching, unc-101 mutant animals are severe Unc, indicating that UNC-101 protein may have major functions in the nervous system or muscles. Our RNAi results indicated that the Unc phenotype of unc-101 mutants may be due to muscle defects, since RNAi often does not interfere with neuronal gene activity (Travernarakis et al., 2000). Supporting this, GFP fluorescence in the neurons persisted after apm-1 RNAi in our RNAi experiment. Although apm-1 and unc-101 seem to be expressed in overlapping cells at all developmental stages, they exhibited different phenotypes, suggesting that both APM-1 and UNC-101 proteins interact with distinct target proteins.

In summary, we showed in this report that there exist two medium chains of the AP-1 complex in the nematode. Both are expressed ubiquitously throughout development and play redundant roles in embryogenesis and vulval development, but they appear to have distinct functions during early larval stages after hatching. The two medium chains are shared by other components of AP-1 complexes as in mammals. An important future direction of research should be to characterize specific functions for these two AP-1 complexes in more detail. The medium chains of AP complexes are known to recognize tyrosine signals and dileucine signals in specific target proteins. In order to elucidate the distinct functions of AP-1 complexes, it will be crucial to identify the cargo proteins that interact with UNC-101 and/or APM-1 protein.