A Genetic Screen for Neurite Outgrowth Mutants in Caenorhabditis elegans Reveals a New Function for the F-box Ubiquitin Ligase Component LIN-23

Corresponding author: Oliver Hobert, College of Physicians and Surgeons, 701 W. 168th St., HHSC 724, New York, NY 10032., or38{at}columbia.edu (E-mail)

Communicating editor: P. ANDERSON

	ABSTRACT

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION LITERATURE CITED

Axon pathfinding and target recognition are highly dynamic and tightly regulated cellular processes. One of the mechanisms involved in regulating protein activity levels during axonal and synaptic development is protein ubiquitination. We describe here the isolation of several Caenorhabditis elegans mutants, termed eno (ectopic/erratic neurite outgrowth) mutants, that display defects in axon outgrowth of specific neuron classes. One retrieved mutant is characterized by abnormal termination of axon outgrowth in a subset of several distinct neuron classes, including ventral nerve cord motor neurons, head motor neurons, and mechanosensory neurons. This mutant is allelic to lin-23, which codes for an F-box-containing component of an SCF E3 ubiquitin ligase complex that was previously shown to negatively regulate postembryonic cell divisions. We demonstrate that LIN-23 is a broadly expressed cytoplasmically localized protein that is required autonomously in neurons to affect axon outgrowth. Our newly isolated allele of lin-23, a point mutation in the C-terminal tail of the protein, displays axonal outgrowth defects similar to those observed in null alleles of this gene, but does not display defects in cell cycle regulation. We have thus defined separable activities of LIN-23 in two distinct processes, cell cycle control and axon patterning. We propose that LIN-23 targets distinct substrates for ubiquitination within each process.

AXONS are extraordinarily dynamic cellular structures. During development, the growth cone of an axon is steered toward its target area through the concerted action of intrinsic and extrinsic molecular cues. The responsiveness of a growth cone to specific extracellular cues can significantly change along the path that an axon takes to reach its target. Upon reaching its target area and achieving target recognition, a growth cone is transformed into a synapse. A mature synapse undergoes further dynamic remodeling, a process that is dependent on electrical activity at the synapse. Post-translational protein modification through ubiquitination has been found to affect each of these dynamic cellular processes (HEGDE and DIANTONIO 2002 ; MURPHEY and GODENSCHWEGE 2002 ).

The ubiquitination machinery is composed of several proteins (HEGDE and DIANTONIO 2002 ). Ubiquitin, a 76-amino-acid protein, is initially activated by an E1 enzyme. It is then transferred to the E2-conjugating enzyme, which associates with an E3 ligase complex to transfer ubiquitin to the eventual target protein. Several classes of E3 ubiquitin ligase complexes exist. One of these, the SCF E3 ubiquitin ligase complex, consists of four proteins, an Skp1-like protein, a Cullin, an F-box protein, and a RING domain containing Rbx-like protein (CRAIG and TYERS 1999 ; JOAZEIRO and WEISSMAN 2000 ). The SCF ligase complex is constitutively active but recognizes only phosphorylated substrates. Its substrate specificity is determined by its F-box protein subunit. The large number of F-box proteins in databases prompted the "F-box hypothesis," which posits that specific substrates are recruited through defined F-box proteins (BAI et al. 1996 ). Consistent with this notion, F-box proteins are bipartite in composition, containing an F-box domain required for interaction with the core ligase complex and other protein-protein interaction motifs that may bind and recruit protein substrates (CRAIG and TYERS 1999 ; KIPREOS and PAGANO 2000 ).

The consequences of protein ubiquitination on a molecular level are diverse. Originally identified as a tag that targets proteins to the degradation apparatus (HOCHSTRASSER 1996 ), more recent studies have emphasized the involvement of protein ubiquitination in a variety of other cellular processes such as intracellular protein trafficking, transcriptional control, and DNA repair (BACH and OSTENDORFF 2003 ; HICKE and DUNN 2003 ).

The cellular and organismal consequences of protein ubiquitination have begun to be elucidated. Several connections between ubiquitination and neuronal development were made with the use of genetically tractable model organisms. The identification of Drosophila bendless as an E2-conjugating enzyme provided the first insights into the importance of ubiquitination as a regulatory mechanism for axon pathfinding (THOMAS and WYMAN 1982 ; MURALIDHAR and THOMAS 1993 ). A surprising feature of the bendless mutant phenotype is its high selectivity, exemplified through bendless affecting only one of several connections in the giant fiber system (THOMAS and WYMAN 1982 ). More recently, specific axonal cues were found to be controlled through protein ubiquitination. Axon pathfinding at the midline in Drosophila requires the Robo axon guidance receptor whose cell surface expression is downregulated by commissureless (Comm), a lysosomal endocytotic sorting protein (KELEMAN et al. 2002 ; MYAT et al. 2002 ). Ubiquitination of Comm by the ubiquitin ligase Nedd4 is thought to be required for Comm to traffic the Robo receptor from the plasma membrane to endosomes (MYAT et al. 2002 ). Ubiquitinated Comm may also have similar roles in downregulating cell surface proteins during neuromuscular synapse formation (WOLF et al. 1998 ; HEGDE and DIANTONIO 2002 ). In vertebrates, the axon guidance cue netrin elicits changes in the ubiquitination state of proteins in isolated retinal growth cones (CAMPBELL and HOLT 2001 ) and, furthermore, the Netrin receptor DCC is subjected to ubiquitin-mediated protein degradation (HU et al. 1997 ), thus illustrating the widespread use of ubiquitination in distinct axon pathfinding events.

The importance of ubiquitination extends from axon outgrowth and target recognition to the processes of synaptic growth and synaptic plasticity. In Drosophila, overexpression of the deubiquitinating enzyme, fat facets, causes defects in synaptic size and strength (DIANTONIO et al. 2001 ). An enhancer of fat facets, highwire, displays similar defects. Highwire encodes a RING finger protein and has been proposed to be a member of a family of E3 ubiquitin ligases (WAN et al. 2000 ). Its worm homolog rpm-1 has also been implicated in synaptogenesis (SCHAEFER et al. 2000 ; ZHEN et al. 2000 ). Another link between ubiquitination and synaptic transmission is provided by the observation that postsynaptic clustering of the Caenorhabditis elegans glutamate receptor glr-1 is regulated by ubiquitination in vivo; mutations that decrease the ubiquitination of GLR-1 lead to altered locomotion behavior in a manner that is consistent with increased synaptic strength (BURBEA et al. 2002 ). Studies in Aplysia have further underscored the importance of ubiquitination in synaptic plasticity, an effect that is exerted through the regulated degradation of defined intracellular signaling components such as the regulatory subunit of protein kinase A and the transcription factor C/EBP (HEGDE et al. 1993 ; HEGDE et al. 1997 ; YAMAMOTO et al. 1999 ).

In this article, we describe a novel aspect of ubiquitin function in the nervous system, namely its role in terminating axon outgrowth. We show that LIN-23, an F-box-containing component of an SCF E3 ubiquitin ligase complex, is required cell autonomously for normal axonal outgrowth. Our newly isolated lin-23 allele furthermore uncouples roles of lin-23 in axonal outgrowth and cell cycle regulation. Our studies therefore provide an example of the diversity of cellular processes in which a single ubiquitin ligase is involved.

	MATERIAL AND METHODS

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains:
The following strains were used: N2 C. elegans wild-type var. Bristol; RW7000 C. elegans wild type, RW subclone of Bergerac BO; CB4856, isolated from a pineapple field in Hawaii in 1972 by L. Hollen; ET003 lin-23(e1925)II/dpy-10(e128); ET009 lin-23(e1521)II/dpy-10(e128); ET012 lin-23(e1883)II/mnC1; ET032 lin-23(rh194)II/mnC1; ET041 lin-23(rh294)II/bli-2(e768); ET068 lin-23(rh293)II/bli-2(e768); ET073 lin-23(m731), unc-4(e120)II/mC6; NJ582 cul-1(e1756)/unc-69(e587)III; CB845 unc-30(e191)III; GS1063 rol-6(e187)II; MT301 lin-31(n301)II; NM1448 rpm-1(js410)V; CB61 dpy-5(e61)I; CB184 dpy-13(e184)IV; SP552 mnDf39 unc-4(e120)/mnC1; dpy-10(e128); unc-52(e444)II; SP543mnDf30unc-4(e120)/mnC1dpy-10(e128)unc-52(e444)II; CB2196daf-4(e1364ts)unc-32(e189)III; OH110lim-6(nr2073)X; and GS1081 sel-10(ar41)him-5(e1490).

The following reporter transgenes were used (all chromosomally integrated): oxIs12, expresses unc-47::gfp (MCINTIRE et al. 1997 ); uIs25, expresses mec-18::gfp (WU et al. 2001 ); jsIs42, expresses unc-4::snb-1::gfp (NONET 1999 ); otIs92, expresses flp-10::gfp (derived from an extrachromosomal array kindly provided by C. Li); mgIs18, expresses ttx-3::gfp (ALTUN-GULTEKIN et al. 2001 ); and nuIs63, expresses ceh-24::gfp (HARFE and FIRE 1998 ), kindly provided by J. Kaplan.

DNA constructs:
For the transcriptional reporter lin-23prom::gfp, 2.2 kb of sequence 5' to the start codon of lin-23 was amplified from N2 wild-type genomic DNA (upstream primer sequence, 5'-GAATAGAGCGGATGTTCG-3'; upstream nested primer sequence, 5'- CCAAATTTGCCTCTGATTCCG-3'; gfp fusion primer, 5'-ctagagtcgacctgcaggc TTAATGTCCTGAATTAAATGG-3'). The 3.0-kb amplicon was PCR fused to gfp using a previously described protocol (HOBERT 2002 ). For the translational reporter lin-23::gfp, the same upstream primers were used, yet the fusion to gfp was at the 3' end of the lin-23 locus so as to include all exons and introns of lin-23 (gfp fusion primer, 5'- ctagagtcgacctgcaggcTGGGCCACCATCTGGCAT-3'). lin-23::gfp constructs that contained the ot1 or rh294 mutations (P610S and Q12Stop, respectively) were constructed using the same primers for the PCR fusion reaction. The amplicons were derived from ot1 or rh294 homozygous mutant DNA. All constructs were injected at 20 ng/µl, together with 50 ng/µl rol-6(d) and 30 ng/µl Stratagene (La Jolla, CA) pBluescript II SK+, to bring the DNA concentration to a total of 100 ng for stable array formation.

The lin-23 full-length cDNA clone yk784a08 was obtained from the Kohara expressed sequence tag collection. The amplified 2.1-kb cDNA was subcloned into SmaI/KpnI restriction sites of unc-47::MCS. The cDNA was sequenced after subcloning and a single-amino-acid change was found and corrected back to wild type. unc-47::MCS was created by subcloning a 1.2-kb region of the unc-47 promoter into PstI/BamHI sites of pPD95.75. The gfp coding region was then replaced with a multiple cloning site (MCS).

Scoring neuroanatomy:
Neuroanatomy was scored under a Zeiss Axioplan 2 microscope equipped with a Hamamatsu Orca CCD digital camera using Openlab as image processing software. Neurons were visualized with gfp markers, listed above, or by DiI filling, using a previously described protocol (HEDGECOCK et al. 1985 ).

Genetic screen and mapping:
Three independent screens were performed. In screens 1 and 3, oxIs12 animals were mutagenized with EMS and 3620 haploid genomes were screened. In screen 2, an unc-30(e191); oxIs12 strain was mutagenized to prevent D-type motor neurons from obscuring the AVL and DVB axons. In this screen, 1700 haploid genomes were screened. In each screen, five F₁ progeny each were placed onto single plates and allowed to self-fertilize at 25° until their progeny had almost entirely starved out the plate. The entire population was washed off with M9 medium, mounted onto coverslips, and examined under a Zeiss Axioplan 2 microscope for defects in DVB and AVL axon morphology. To isolate homozygous animals from populations in which we noted animals with mutant phenotypes, we blindly singled out 12 animals per population and assessed whether any of those would throw a brood with at least a 20% penetrant mutant phenotype. Animals from these plates were again singled out to confirm that each one would throw >20% mutant animals. Animals were then backcrossed multiple times. Upon backcrossing we noted that one mutant strain, OH2224, which displayed a 96% penetrant AVL/DVB axon defect, contains two separable mutations, each of which alone gave only a lowly penetrant mutant phenotype. We did not pursue this mutant any further.

For genetic mapping, a combination of three-factor, deficiency, sequence-tagged site (STS; using RW7000 as a mapping strain; WILLIAMS et al. 1992 ), and single-nucleotide polymorphism (SNP; using CB4856 as a mapping strain) methods were used. ot1 was initially mapped onto LGII between stP101 and maPI using RW7000. A three-factor cross between lin-31(n301)rol-6(e187) and ot1; oxIs12 produced 6/21 lin-non-rol animals and 18/23 rol-non-lin animals that retained ot1. SNP mapping with the triple-mutant strain lin-31(n301)ot1rol-6(e187);oxIs12 and the Hawaiian strain CB4856 resulted in the identification of 1/114 recombinants that retained ot1 and recombined within the genomic region defined by cosmid C32D5. Deficiency analysis indicated that mnDf30 [–0.9 to –0.21 map units (MU)] failed to complement ot1, but mnDf39 (–0.4 to –0.21 MU) complemented ot1. Since mnDf39 is contained within the mnDf30 deficiency, further mapping of mnDf39 dead embryos indicated that the left-hand boundary of the deficiency is between SNPs on T25D10 and K03H9. Thus, the ot1 mutation was narrowed down to a 62-kb region covered by five cosmids from C32D5 to K03H9. ot2 was mapped between the Hawaiian strain CB4856 SNP markers on W03D8 (–5.6 MU) and VF39H2L (2.9 MU) on LGI. ot3 was mapped between the RW7000 STS markers stP156 (–5.33 MU) and stP103 (–1.35 MU) on LGX. ot4 was linked to LGII through crossing with lin-31(n301)rol-6(e187)II. Lin-rol animals were picked in the F₂ generation and the progeny were scored for axonal outgrowth defects similar to ot4. In total, 0/20 linrol animals showed axonal outgrowth defects. ot6 was initially mapped between the bP1 (3.15 MU) and stP6 (6.0 MU) STS markers. On LGV a three-factor cross between ot6 and unc-42lin-25 gave 7/12 unc-non-lin recombinants that retained ot6. ot8 was first mapped onto LGIII through SNP mapping and through three-factor crossing between daf-4 (–1.46 MU) and unc-32 (0.0 MU). ot7 was linked onto LGIV by crossing it with dpy-13(e184)IV. Dpy animals were picked in the F₂ generation and the progeny were scored for axonal outgrowth defects. In total, 0/10 F₂ dpy animals showed axonal outgrowth defects. ot9 was placed on LGX through its cosegregation with the X-linked oxIs12 marker. ot40 was initially mapped between Hawaiian SNPs on W03D8 (–5.6 MU) and T22A3 (5.0 MU) onto LGI and then to the left of dpy-5(e61) (0 MU).

	RESULTS

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION LITERATURE CITED

A screen for mutants that affect axon anatomy of the AVL and DVB motor neurons:
The complete set of GABAergic neurons in the nervous system of C. elegans can be labeled with oxIs12, a chromosomally integrated transgene expressing gfp under the control of the promoter for the GABA transporter unc-47 (MCINTIRE et al. 1997 ; Fig 1). Various forward and reverse genetic approaches have been used in the past to define genes that affect cell fate determination and axon outgrowth of these GABAergic neurons (MCINTIRE et al. 1992 , MCINTIRE et al. 1993 ; JIN et al. 1994 ; EASTMAN et al. 1999 ; HOBERT et al. 1999 ; HAMMARLUND et al. 2000 ; HUANG et al. 2002 ). We have previously shown that loss of the LIM homeobox gene lim-6 causes developmental and axonal defects of the GABAergic ventral cord motor neurons AVL and DVB (HOBERT et al. 1999 ; Fig 2A). Of the multiple types of neurons in which lim-6 is expressed, only the axon anatomy of the AVL and DVB motor neurons is affected by loss of lim-6 function. In contrast, genes such as unc-34, unc-44/ankyrin, unc-61, unc-71, and unc-76 affect axon anatomy of AVL and DVB (data not shown) as well as of many other if not all neurons of the nervous system (HEDGECOCK et al. 1987 ; MCINTIRE et al. 1992 ; ANTEBI et al. 1997 ). Animals carrying mutations in these pleiotropically acting genes hence display obvious morphological, locomotory, and reproductive defects.

View larger version (26K): In this window In a new window Download PPT slide   Figure 1. Schematic of a selection of the neurons tested for axonal and cell proliferation defects. The oxIs12 transgene (MCINTIRE et al. 1997 ) was used to visualize the GABAergic nervous system, uIs25 (WU et al. 2001 ) was used to observe the touch sensory neurons, and jsIs42 (NONET 1999 ) was used to assess the SAB motor neurons. oxIs12 and jsIs42 show a broader expression pattern; only neurons of interest are shown here.

View larger version (82K): In this window In a new window Download PPT slide   Figure 2. DVB/AVL anatomical defects in eno mutants. Anatomy of the GABAergic neurons in a selected number of eno mutants is shown. See Table 1 for quantification of defects. For comparison, the previously described lim-6 mutant phenotype is shown (HOBERT et al. 1999 ). Neurons are visualized with the oxIs12 transgene. Arrows point to the DVB cell body and arrowheads to defects. More anteriorly located cell bodies are those of the D-type motorneurons (see wild type, left). An extra arrowhead in eno(ot1) (right) points to the branched nature of the ectopic axon; as demonstrated in Fig 5, this tail axon is not an ectopic sprout from DVB but rather an overextended AVL axon. Note that the ectopic axons in eno mutants can derive from either the cell body (eno-2 and eno-3, left) or the axon shaft (eno-2 and eno-3, right). eno-6 mutants show both ectopic axon branching (left) and axon misrouting (a premature termination is shown, right). eno-7 also shows axon misrouting and axon branching (arrowheads). Both eno-6 and eno-7 and are shown in an unc-30(e191) mutant background to prevent D-type motor neurons from obscuring DVB anatomy. A dashed line indicates the location of the ventral nerve cord.

To identify genes that like lim-6 affect axon outgrowth of AVL and DVB in a cell-specific manner, we sought to undertake a genetic screen that would prevent the isolation of genes that affect broad aspects of neuronal development. To this end, we mutagenized animals in which the GABAergic nervous system is labeled with gfp, picked five F₁ progeny per plate from the mutagenized P₀ generation, and allowed these animals to produce self-progeny for several generations. We reasoned that growing the animals for several generations before anatomic analysis would "dilute out" animals compromised in viability and brood size. We then analyzed these progeny on a population basis with a high-power compound fluorescence microscope specifically focusing on the tail region, which contains the axon terminus of AVL and the cell body of DVB (Fig 1 and Fig 2).

Screening through 5300 mutagenized haploid genomes, we isolated 12 mutants with anatomical defects of the AVL and/or DVB motor neurons, at least 10 of which define different complementation groups located in distinct chromosomal intervals (Fig 2; Table 1). On the basis of their axonal phenotypes, mutants were termed eno (ectopic/erratic neurite outgrowth of the AVL or DVB motor neurons). All eno mutants are viable, show an approximately normal brood size, and display normal locomotory behavior, with the exception of ot2 animals, which are sluggish. All eno mutants are recessive, again with the exception of ot2, which is semidominant (data not shown). Two types of neuroanatomical defects of GABAergic neurons can be observed to a variable degree in eno mutants. One type of defect is characterized by what appears to be ectopic axons that emanate from the cell body or main axon shaft; the other type is characterized by the main axon of DVB and/or AVL straying off their normal path and/or terminating inappropriately (Fig 2; Table 1). We also isolated one mutant, ot32, which displayed cell shape defects reminiscent of those seen in sax-1 mutant animals (ZALLEN et al. 2000 ). ot32 maps to LGX but complements sax-1. We have not pursued a further characterization of this mutant.

The penetrance of the axonal defects in several of the eno mutants is significantly reduced by lowering the cultivation temperature to 15° (Fig 3A). This temperature sensitivity may reflect a temperature-sensitive nature of the mutant protein product or an underlying temperature sensitivity of the eno phenotype. The latter case would be supported by the observation that null alleles of genes affecting neuronal activity show temperature-sensitive axon sprouting defects (PECKOL et al. 1999 ).

View larger version (36K): In this window In a new window Download PPT slide   Figure 3. Characterization of GABAergic neuronal defects in eno mutants. (A) The eno phenotype of several eno mutants is temperature dependent. AVL/DVB axon defects were scored with oxIs12. (B) The eno mutants show normal to moderately defective defecation behavior. Five to 10 defecation cycles were scored for each genotype at 25°. Each asterisk represents a single animal. EMC, enteric muscle contractions.

Since the GABAergic DVB and AVL motor neurons mediate the expulsion step of the defecation motor program, we next examined whether abnormal axon outgrowth of DVB or AVL in eno mutants would affect the functional output of these neurons. We found that eno-2(ot2), eno-4(ot4), eno-7(ot7), eno-8(ot8), eno-9(ot9), and eno-11(ot40) show expulsion defects, which are weaker than the defects observed in GABA-deficient unc-25 mutants (Fig 3B). The moderate nature of the defects and the lack of significant defects in eno(ot1), eno-3(ot3), and eno-6(ot6) may be explained by eno mutants differentially affecting DVB and AVL, both of which have a redundant function in regulating enteric muscle contractions (MCINTIRE et al. 1993 ; e.g., as we show below, eno(ot1) affects only AVL but not DVB axon outgrowth). Alternatively, ectopic axon outgrowth may have no impact on synapse formation and neuron function.

Other nervous system defects in eno mutants:
To survey the extent of neuroanatomical defects in the eno mutants, we visualized several classes of neurons with a panel of gfp reporter genes as well as by dye filling. We observed defects in the morphology of D-type motor neurons in eno-2(ot2) and eno-8(ot8), which display axon sprouting defects in the commissural axon tracts (Fig 4; Table 1; data not shown). Amphid sensory neurons, which were visualized with the lipophilic dye DiI, displayed defects in eno-2(ot2), eno-6(ot6), eno-7(ot7), and eno-11(ot40), but not in any other eno mutant (Fig 4; Table 1). The amphid sensory neuron defects fall into two different categories. eno-2(ot2) and eno-11(ot40) display amphid axon sprouting defects, whereas eno-6(ot6) and eno-7(ot7) display ASJ axon guidance defects, such that the main ASJ axon fails to extend appropriately along the normal route taken by amphid sensory axons (Fig 4; Table 1).

View larger version (119K): In this window In a new window Download PPT slide   Figure 4. Additional neuroanatomical defects in eno mutants. Anatomy of head amphid sensory neurons is visualized by DiI staining. eno-2(ot2) and eno-11(ot40) animals show ectopic axon sprouting defects (arrowhead). eno-6(ot6) shows an amphid axon misrouting defect, characterized by amphid axon(s) not being part of the amphid commissure (arrow). Similar defects are seen in eno-7(ot7) (not shown). Anatomy of the D-type motorneurons is visualized with the oxIs12 transgene and AIY interneuron anatomy is visualized with the mgIs18 transgene. Aberrant sprouting is indicated with an arrowhead. eno-2(ot2) shows ectopic outgrowth of axons from either the cell body of AIY (arrowhead) or the main axon shaft (not shown).

Amphid axon sprouting defects have been observed in some of the sax mutants (sensory axon defective), namely sax-1, sax-2, and sax-6 (ZALLEN et al. 1999 ). We found that these three sax mutants also yielded an eno phenotype in the AVL and DVB motor neurons [penetrance: sax-1(ky211), 58%, n = 41; sax-2(ky216), 95%, n = 93; sax-6(ky214), 35%, n = 37]. Mapping and complementation data indicate that eno and sax genes define distinct loci (Table 1).

After analyzing motor neurons (AVL, DVB, and D-type) and sensory neurons (amphid neurons), we next examined interneuron anatomy using a cell-type-specific marker for the AIY interneuron class (ttx-3::gfp). We found that eno-2(ot2) but no other examined eno mutant displayed an interneuron axon sprouting phenotype (Fig 4; Table 1). Further extending our analysis of eno-2(ot2), we found that all other examined neuron classes (ASE chemosensory, mechanosensory, and SAB motor neurons) also displayed axon sprouting defects (data not shown).

Taken together, the eno mutants fall into different categories. One category affects axonal outgrowth of a subset of neurons [eno(ot1), eno-3(ot3), eno-4(ot4), eno-8(ot8), eno-9(ot9), and eno-11(ot40)] and another category affects all neuron classes tested [eno-2(ot2)]. Mutants in both categories have axons and dendrites that take their correct path, but either fail to terminate appropriately or extend additional, aberrant axon sprouts. In contrast, mutants in the last category [eno-6(ot6) and eno-7(ot7)] display axonal sprouts as well as axon pathfinding defects of their main axon.

We chose to pursue a more detailed cellular and molecular characterization of the ot1 allele due to its highly penetrant eno phenotype in the GABAergic neurons and due to its apparent cell-type specificity.

eno(ot1) affects axon outgrowth and targeting of specific classes of sensory and motor neurons:
eno(ot1) mutant animals display a GABAergic axon outgrowth defect, characterized by an overextended axon that can be detected in the posterior end of the animal; this aberrantly extended axon also often displays ectopic branches (Fig 2 and Fig 5A). The overextended axon can be an inappropriately terminated AVL motor axon or an inappropriately terminated D-type motor axon or may be an ectopic sprout derived from the DVB motor neuron. To distinguish among these possibilities, we genetically eliminated the D-type motor neurons using an unc-30(e191) mutant background (JIN et al. 1994 ; EASTMAN et al. 1999 ) and found that the appearance and penetrance of eno(ot1) outgrowth defects remains the same in eno(ot1) and eno(ot1);unc-30(e191) animals (Fig 5B, data not shown). Thus, the overextended axon in eno(ot1) animals cannot derive from D-type ventral cord neurons. The eno(ot1); unc-30(e191) strain also allowed us to score termination of the DVB axon at the vulva, which we found to be unaffected. In addition, we observed no DVB axonal defects upon visualization of DVB with a gfp transgene that is expressed in DVB but not in AVL (otIs92). We conclude that the axon termination defect in eno(ot1) can be assigned to the AVL motor neuron. Consistent with the notion of defective termination of axon outgrowth, we furthermore note that the outgrowth of the AVL axon continues through larval stages, resulting in a more penetrant mutant phenotype in adult animals compared to young larvae (data not shown).

View larger version (76K): In this window In a new window Download PPT slide   Figure 5. Axon outgrowth defects in eno(ot1) mutant animals. (A, left) The AVL axon (red) travels along the ventral cord and terminates just before the DVB cell body (blue; arrow). AVL overgrowth defects are denoted by arrowheads. (Middle) Touch sensory neurons ALM (blue) and AVM (red) normally extend their axons toward the tip of the nose (arrows). All lin-23 alleles show outgrowth defects including a failure of AVM to terminate appropriately (middle arrowhead) or a failure to extend past the nerve ring (bottom arrowhead). The defects of the ot1 mutant allele in the touch neurons can be observed only at 25°. (Right) SAB motor neurons, which normally terminate in the tip of the nose (arrows), show outgrowth defects (arrowheads) in lin-23 mutants. (B) Overextended GABAergic tail axons in ot1 mutants derive from the AVL motor neuron (arrowhead). Animals carry the unc-30(e191) mutation and the oxIs12 transgene. The unc-30(e191) allele eliminates oxIs12-conferred gfp expression in the D-type motor neurons of the ventral cord (JIN et al. 1994 ; EASTMAN et al. 1999 ).

As described above, eno(ot1) mutants do not display defects in DVB motor neuron, amphid sensory neuron, or AIY interneuron morphology. We further extended our analysis of neuroanatomical defects in eno(ot1) mutants by crossing eno(ot1) mutant animals with transgenic gfp reporter strains that label additional sets of sensory, inter-, and motor neurons (Fig 1). We found no defects in the axon anatomy of ASE chemosensory neurons or PVQ ventral cord interneurons (data not shown). However, we found that SAB head motor neurons show multiple defects in eno(ot1) mutants including ectopic sprouting from the axon, misrouting of the neuron from its original path, and an overgrowth of the axon, that is, a failure to terminate outgrowth appropriately (Fig 5A). Mechanosensory axons also fail to terminate outgrowth, an effect that is strictly temperature dependent (Fig 5A).

The appropriate expression of several gfp-marked genes (unc-47, mec-18, and unc-4) suggests that execution of cell fate is normal in eno(ot1) mutants. Hence, the axon outgrowth and pathfinding defects observed using these markers may represent a direct consequence of defective axon outgrowth machinery rather than a consequence of inappropriate execution of cell fate.

ot1 is an allele of lin-23:
Through a combination of multifactor, single-nucleotide polymorphism-based mapping and deficiency analysis, we localized the eno(ot1) locus to a 62-kb region on chromosome II. This region contains the previously described lin-23 locus. Complementation tests (not shown), transformation rescue (Fig 6), and allele sequencing (Fig 7A) demonstrated that ot1 is allelic to the lin-23 locus. From here on, we refer to eno(ot1) as lin-23(ot1). lin-23 was previously shown to encode an F-box protein with WD40 repeats that is a component of SCF-type E3 ubiquitin ligase complexes (Fig 7C; KIPREOS et al. 2000 ). In comparison to its vertebrate, fly, and yeast orthologs, the predicted LIN-23 protein has a significantly extended C terminus (>100 amino acids; Fig 7A). ot1 is a missense mutation leading to a single-amino-acid change (proline 610 to serine) within the nonconserved C-terminal tail (Fig 7A and Fig B). Proline 610 lies in a motif of the protein (PAPP) that constitutes a potential binding site (PXXP) for SH3 domain or WW domain containing protein ligands (MACIAS et al. 2002 ), suggesting that in ot1 mutants a specific protein interaction of LIN-23 may be disrupted (Fig 7C). To corroborate this notion, we examined whether the extended C-terminal tail of C. elegans LIN-23 and its PAPP motif is conserved in the C. briggsae ortholog of LIN-23. The recently released genome sequence of C. briggsae contains a single LIN-23 ortholog, which is 89% identical to C. elegans LIN-23. The C termini of the two proteins, however, are highly diverged, yet the PAPP motif is completely conserved (Fig 7B). The conservation of this motif within a region of little overall homology underscores the functional relevance of this motif.

View larger version (30K): In this window In a new window Download PPT slide   Figure 6. ot1 is an allele of lin-23. Transformation rescue experiments are shown. K10B2 is a cosmid that contains the lin-23 gene as well as four other predicted genes. T25D10 is an unrelated control cosmid that neighbors K10B2. lin-23::gfp is a translational fusion of the lin-23 locus to gfp. unc-47p is a GABAergic neuron-specific promoter that was injected either as a fusion with lin-23 or with a control sequence (gfp). All DNA was injected at 20 ng/µl with rol-6(d) as an injection marker.

View larger version (62K): In this window In a new window Download PPT slide   Figure 7. LIN-23 structure and alleles. (A) LIN-23, its mutant alleles, and orthologous proteins are shown. The molecular nature of m731 has not been described before. The ot1 allele affects a proline in a predicted SH3/WW domain binding site (PAPP). All other mutant alleles have previously been described by KIPREOS et al. 2000 . (B) The PAPP motif is conserved in the C. briggsae ortholog of LIN-23. The F-box at the N terminus is boxed in black; the seven consecutive WD40 repeats are shaded in gray; and the PAPP motif, located in the otherwise highly divergent C terminus, is boxed in black. The P610S mutation in ot1 is indicated by an arrow. (C) SCF-type E3 ubiquitin ligase complexes. Structural studies revealed that cullin serves as a bridge to bring together different components of the complex (ZHENG et al. 2002 ). Substrates are recruited by protein-protein interaction domains (in this case, WD40 domains) of the F-box protein, while the F-box recruits other components of the ligase complex. We speculate that the PAPP motif is involved in recruiting a set of substrate(s) required for axon outgrowth while it has no role in recruiting substrates involved in cell cycle control; those may be recruited by the WD40 repeats.

We next tested the effect of previously isolated lin-23 alleles, most of which are likely null alleles (KIPREOS et al. 2000 ), on axon outgrowth. Using gfp markers for the GABAergic motor neurons, touch neurons, and SAB-type motor neurons, we found that all other known lin-23 alleles also affected axon anatomy of these neuron classes (Fig 5A and Fig 9). The cell-type specificity also appears to be the same; that is, cell types that were not affected in ot1 mutants (amphid sensory neurons) were also not affected in lin-23 null mutants. Furthermore, the GABAergic and SAB motor neuron defects are qualitatively and quantitatively indistinguishable between ot1 and previously described null alleles (Fig 5A and Fig 9A and Fig C). In contrast, however, the defects of the mechanosensory neurons appear different. The putative lin-23 null alleles cause the mechanosensory axons to leave their appropriate track, while the ot1 mutation affects only the axon termination site rather than the track taken by the neuron (Fig 5A and Fig 9B).

View larger version (69K): In this window In a new window Download PPT slide   Figure 8. Cellular proliferation defects in lin-23 mutants. (A) Neuronal proliferation defects in lin-23 mutants. (Left) Effect of lin-23 alleles on ventral cord D-type motor neurons, visualized with the oxIs12 transgene (cell bodies are labeled with gray lines). Six D-type neurons, the DD neurons, are embryonically generated, while 13 D-type neurons, the VD neurons, are postembryonically generated. (Right) Effect of lin-23 alleles on touch sensory neurons, visualized with the uIs25 transgene. Wild-type, postembryonically generated sensory cell bodies are shown with arrows and ectopic cell bodies in a representative example are indicated with arrowheads. lin-23(e1925) animals show proliferation defects in AVM, PVM, and also PLM. A quantification of the defects is shown in Fig 9. (B) Nonneuronal proliferation in lin-23 mutants. (Left) The overall body morphology and appearance of ot1 mutant animals are indistinguishable from those of wild-type animals. In contrast, animals harboring canonical lin-23 null alleles, such as lin-23(e1925), are elongated and pale. (Middle) ot1 animals display no distal tip cell (arrows) proliferation defects or ectopic gonad arms (arrowheads) as observed in e1925 animals (bottom and KIPREOS et al. 2000 ). (Right) The proliferation and migration of postembryonically generated sex muscles are affected in canonical lin-23 alleles (bottom, inappropriately proliferating sex myoblasts are scattered along the body, indicated with arrowheads), but not in ot1 mutant animals. Sex muscles were visualized with a ceh-24::gfp reporter transgene (HARFE and FIRE 1998 ), nuIs63, which also labels enteric muscle in the tail of the worm (arrows).

View larger version (28K): In this window In a new window Download PPT slide   Figure 9. Quantification of axonal defects and neuronal proliferation defects. (A) Quantification of effects of lin-23 on D-type motor neuron proliferation and axon outgrowth. The wild-type number of D-type cell bodies was scored as 19 ± 1. Percentages of animals within a population that show defects are shown. An animal was scored as defective if it contained >20 cell bodies; up to 25 cell bodies were observed. Axon outgrowth defects refer to DVB and AVL axons. (B) Quantification of touch neuron defects (scored with uIs25). Axon outgrowth defects subsume wandering and premature termination (hatched bars). Axon overgrowth, i.e., failure to terminate can be observed only in the ALM touch neuron (shaded bars). The touch neuron cell proliferation defects are a sum of those observed in AVM, PVM, and PLM. (C) Quantification of SAB motor neuron defects (scored with jsIs42). Axon outgrowth defects include wandering, sprouting, and overgrowth.

The neuroanatomical defects of lin-23 null mutants in the AVL neuron and the touch sensory neurons are not accompanied by behavioral defects in enteric muscle contractions (regulated by AVL and DVB), anterior body wall muscle contraction (regulated by AVL only), or touch sensation (regulated by the six touch neurons; data not shown). An inability of an axon to appropriately terminate axon outgrowth may therefore not affect its normal patterns of neuronal connectivity and function. In this context, it is important to keep in mind that in C. elegans synapses are made en passant, rather than at axon termini (WHITE et al. 1986 ).

ot1 separates axonal function from a cell cycle function of lin-23:
Previously described lin-23 null mutant animals are characterized by an overproliferation of various postembryonically generated cell lineages leading to an elongated appearance and sterility of the animals (KIPREOS et al. 2000 ). These previous studies on lin-23 cell proliferation defects have largely relied on observing cell lineages with Nomarski optics. However, it was not clear whether ectopic cells produced by cell cycle defects would be able to differentiate appropriately. We addressed this question by analyzing the fate of postembryonically generated lineages with neuronal cell-type-specific gfp reporter strains (Fig 1). The GABAergic neuron marker unc-47::gfp (oxIs12) labels the postembryonically generated VD and DVB motor neurons while the touch neuron marker mec-18::gfp (uIs25) labels the postembryonically generated AVM and PVM neurons (Fig 1). We found excessive fluorescent cell bodies in the postembryonically generated neuronal lineages examined in all except one of the previously described putative null alleles of lin-23 (Fig 8A and Fig 9; rh294, the exception, is described in more detail in the next section). We also found the embryonic PLM lineage to have cell proliferation defects in lin-23 alleles. The gfp markers that we have used, a GABA transporter and a touch neuron-specific protein, can be classified as terminal differentiation markers. In spite of uncontrolled cell proliferation, the expression of terminal cell differentiation markers in ectopic neurons indicates that individual neuronal cell fates are still appropriately executed.

Intriguingly, the ot1 allele, unlike all previously characterized lin-23 alleles, displays no readily observable cell proliferation defects. ot1 animals do not display the sterility or elongated appearance indicative of overproliferation of various tissues that is characteristic of previously described lin-23 alleles (Fig 8B). No ectopic neuronal D-type motor neuron or touch neuron cell bodies were observed in ot1 mutants (Fig 8A and Fig 9). Moreover, no obvious proliferation defects are apparent in two types of nonneuronal tissue that we specifically examined, the distal tip cells and the postembryonically generated sex muscles; both cell types are affected by the lin-23 null alleles (Fig 8B; KIPREOS et al. 2000 ). Lowering the dosage of lin-23(ot1) by placing this allele over a chromosomal deficiency that eliminates lin-23, mnDf30, also causes no proliferation defects, but recapitulates the axonal outgrowth defects (data not shown). Furthermore, lin-23(ot1) rescues the sterility of lin-23 deficiency.

The lack of cell proliferation defects in lin-23(ot1) animals is in striking contrast to the observed axon outgrowth defects, which show a similar severity and cell-type specificity in ot1 and lin-23 null mutant animals. The absence of cell proliferation defects in ot1 mutants also eliminates the possibility that the aberrant axons that we observe in all lin-23 mutant alleles derive from aberrantly proliferated and hence possibly "confused" neurons. Taken together, these results suggest that LIN-23 has separable functions in cell cycle control and axon outgrowth (Fig 7C).

LIN-23 is a broadly expressed cytoplasmic protein:
As a first step toward delineating the site of lin-23 action, we examined its expression pattern. RNA in situ analysis demonstrated broad embryonic expression of lin-23, but lacked cellular resolution (KIPREOS et al. 2000 ). To analyze the expression pattern in more detail, we generated two gfp reporter gene constructs, one in which 2.2 kb of the 5' upstream regulatory region of lin-23 is fused to gfp (lin-23prom::gfp) and one in which the exons and introns of the gene are included (lin-23::gfp; Fig 10A). The lin-23::gfp construct completely rescues the mutant phenotype (Fig 6). Both constructs show similar expression patterns. Broad and possibly ubiquitous expression is first visible during gastrulation. Postembryonically, LIN-23 reporter fusions are expressed in many but not all neurons (e.g., no expression is observed in commissural motor neurons), in body wall and enteric muscles, and in hypodermal cells (Fig 10B). The fusion protein is excluded from the nucleus and is uniformly distributed throughout the cytoplasm (Fig 10B). The protein also shows uniform distribution along axons (Fig 10B). LIN-23 expression is maintained in all tissues throughout adulthood. Given the absence of somatic cell divisions in adulthood, this observation is consistent with a role of LIN-23 that goes beyond cell cycle control in the developing organism and may hint at further roles of LIN-23 in fully differentiated cells.

View larger version (45K): In this window In a new window Download PPT slide   Figure 10. LIN-23 is a broadly expressed cytoplasmic protein. (A) Schematic of expression constructs. Q12Stop corresponds to the mutation in the rh294 allele and P610S to the ot1 allele. A question mark indicates potential alternative start codons. (B) Expression of the lin-23 transcriptional reporter (lin-23prom::gfp) and a lin-23 translational reporter (lin-23::gfp). LIN-23::GFP first reveals expression during gastrulation and is strictly restricted to the cytoplasm in embryonic cells and throughout the cell body and axon in the ventral cord (arrows point to left and right cord) in adult animals. (C) Comparison of expression of wild-type and mutated forms of LIN-23::GFP in an adult wild-type background [slightly twisted due to the rol-6(d) injection marker]. All constructs were injected at the same concentration (20 ng/µl). At least three lines were scored per construct. Speckles are gut autofluorescence. Arrows point to expression in head ganglia.

The LIN-23::GFP rescuing construct provided us with an assay to test the stability and localization of lin-23 mutant alleles. By introducing the ot1 mutation (P610S) into the LIN-23::GFP construct (Fig 10A), we asked whether this mutation may potentially destabilize or mislocalize the protein. We found that LIN-23^P610S::GFP animals looked indistinguishable from LIN-23::GFP animals (Fig 10B and Fig C), consistent with our hypothesis that ot1 has no gross effects on LIN-23 function but may affect only its interaction with a specific subset of target proteins. It is possible, however, that a destabilizing effect of the P610S mutation may be compensated through the addition of the GFP moiety.

The rh294 mutation, which was originally proposed to be a null allele, is an early stop codon 12 amino acids past the start site (KIPREOS et al. 2000 ). In contrast to the other putative lin-23 null alleles, we found that rh294 mutant animals affected the proliferation of touch neurons, but not of the VD ventral nerve cord motor neurons; axon outgrowth is also less severely affected in rh294 animals (Fig 9). The hypomorphic nature of defects observed in rh294 indicated that a more downstream start site may be used in these animals (e.g., a methionine at position 56 or 78, both of which are still upstream of the F-box; Fig 10A). To test this notion we introduced the rh294 mutation into the LIN-23::GFP fusion construct to create LIN-23^Q12Stop::GFP (Fig 10A). LIN-23^Q12Stop::GFP shows appreciable, though largely reduced expression levels (Fig 10C). The cellular and subcellular sites of expression appear unaffected.

lin-23 acts cell autonomously in GABAergic neurons to affect axon outgrowth:
Defects in axon termination of the AVL motor neuron in lin-23 mutants may be caused through defects in the neuron itself or in its synaptic target cells (for example, the enteric muscles), which may fail to provide an appropriate stop signal for the innervating neuron. To determine the focus of lin-23 action, we expressed a lin-23 cDNA under the control of the 5' regulatory region of the unc-47 gene, which is exclusively expressed in the GABAergic neurons including AVL (MCINTIRE et al. 1997 ). We find that lin-23 is required cell specifically within the GABAergic neurons to rescue the AVL outgrowth defects and conclude that the gene acts cell autonomously (Fig 6).

Examination of other ubiquitin ligases in the nervous system:
F-box proteins recruit protein substrates to SCF-type ubiquitin ligase complexes (Fig 7C). There are >300 F-box proteins in the C. elegans genome (KIPREOS and PAGANO 2000 ), yet only 2 of these F-box proteins, SEL-10 and T01E8.4, share a LIN-23-like domain composition of one F-box and several WD40 repeats. We asked whether SEL-10, the only one of the two LIN-23-like proteins for which mutant alleles are available (HUBBARD et al. 1997 ), may play a role in axonal development. We first examined those neurons whose morphology is not affected by LIN-23 (D-type motor neurons, DVB, and amphid sensory neurons) but found no defects in sel-10(ar41) null animals (data not shown). sel-10 mutants also do not display defects in mechanosensory neurons, SAB motor neurons, or synaptic VAMP::GFP clustering within the GABAergic neurons. We conclude that neurons that are unaffected by lin-23 contain either another type of ubiquitin ligase or a mechanism independent of ubiquitination to ensure their correct axogenesis.

RING finger proteins constitute a different class of E3 ubiquitin ligases (JOAZEIRO and WEISSMAN 2000 ), one example being Drosophila highwire, which is involved in synaptic signaling (WAN et al. 2000 ). Its worm ortholog rpm-1 is, like lin-23, also a broadly expressed gene and shows a spectrum of mutant phenotypes that is remarkably similar to those of lin-23, specifically, touch neuron axon overgrowth and misrouting of SAB motor neurons (SCHAEFER et al. 2000 ; ZHEN et al. 2000 ). We expanded the comparison of mutant phenotypes of rpm-1 and lin-23 and found that, unlike lin-23 mutants, rpm-1 mutants show no AVL overgrowth defects, but display DVB axon sprouting (data not shown). Furthermore, we found that lin-23 mutants do not show the presynaptic VAMP::GFP clustering defects in GABAergic ventral cord motor neurons that are observed in rpm-1 mutants (SCHAEFER et al. 2000 ; ZHEN et al. 2000 ; data not shown). We conclude that different types of E3 ubiquitin ligase have some overlapping but also distinct functions in neuronal development.

	DISCUSSION

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION LITERATURE CITED

eno mutants may define proteins involved in cell-specific aspects of axon guidance:
We have described here mutants retrieved from a forward genetic screen that display axon outgrowth defects in the AVL and DVB motor neuron classes. Mutants defective in DVB morphology display ectopic outgrowth defects from the axon or cell body. Several of the eno mutants also show axon sprouting defects in sensory and interneurons.

Ectopic axon outgrowth can be indicative of a failure to establish a functional connection of a neuron with its target cell (BROWN et al. 1981 ; ZHAO and NONET 2000 ; LORIA et al. 2003 ). Hence, eno mutants could affect steps relating to target recognition, synapse formation, or synaptic or retrograde signaling. eno mutants do not show defects normally associated with broadly acting synaptic transmission mutants, such as aldicarb resistance or uncoordinated locomotion (data not shown), and at least some of them show only a very restricted set of neuroanatomical defects. This may simply be an indication of a hypomorphic nature of our available eno alleles or may indicate that eno mutants affect cell-specific aspects of axogenesis rather than broad aspects of neurotransmission.

Some of the eno mutants show not only axon branching defects but also axon extension, misrouting, and defasciculation defects. These defects also display cell-type specificity; for example, ot6 and ot7 affect DVB and ASJ axon routing and fasciculation within the amphid commissure and ventral cord, but not AIY axon guidance and lin-23(ot1) displays AVL, but not DVB axon termination defects. Our genetic screen has thus defined a multitude of genes each affecting specific aspects of axogenesis in different cell types.

LIN-23, ubiquitin, and axogenesis:
The C. elegans genome contains at least 326 predicted F-box proteins (KIPREOS and PAGANO 2000 ). Surprisingly, besides SEL-10, which regulates the degradation of LIN-12/Notch and SEL-12/presenilin (HUBBARD et al. 1997 ; WU et al. 1998 ), and FOG-2, an F-box protein involved in translational repression (CLIFFORD et al. 2000 ), LIN-23 represents the only other C. elegans F-box protein for which a function has been reported. In addition to the previously documented role of lin-23 in the cell cycle, we have described here a requirement for the lin-23 gene in axon outgrowth of specific types of motor (AVL, but not DVB) and sensory neuron classes (touch neurons but not amphid neurons).

The AVL motor neuron axon normally makes several en passant synapses before it terminates in the preanal ganglion, where it makes synaptic contacts with preganglionic axons and the enteric muscles (WHITE et al. 1986 ; AVERY and THOMAS 1997 ). In lin-23 animals, the AVL axon takes a normal trajectory, but fails to terminate near the enteric muscles and instead overextends into the tail. These defects may be indicative of a failure of the axon to interpret a stop signal derived from one of its postsynaptic targets. When we genetically removed the enteric muscle target of AVL in hlh-8 mutant animals (CORSI et al. 2000 ), we observed no AVL outgrowth defects (data not shown). Neurotransmission may also contribute in a negative feedback loop to the termination of axon outgrowth (BROWN et al. 1981 ; ZHAO and NONET 2000 ), yet we have found that genetic removal of AVL's neurotransmitter, GABA, does not cause AVL overgrowth defects (data not shown). However, these data do not rule out the possibility that a complete and combined removal of all synaptic targets of AVL and/or a complete elimination of all synaptic transmission (GABA plus some unknown peptidergic transmission event; LORIA et al. 2003 ) may impart a lin-23-like axon termination defect. It is also conceivable that axon stop signals are unrelated to the synaptic targets of a neuron and are rather presented by some distinct cellular source. No matter from where the stop signal is provided, the cell-autonomous function of lin-23 in AVL prompts us to hypothesize that the reception and/or interpretation of this axon termination signal is disrupted in lin-23 mutant animals.

A similar failure of axon termination is observed in the mechanosensory neurons of lin-23 mutant animals. Interestingly, within the mechanosensory neuron context, an allelic series reveals differential defects based on severity of the lin-23 mutation. These axons fail to follow the correct trajectory in strong loss-of-function lin-23 alleles. However, in the ot1 allele only the appropriate termination of axon outgrowth, but not the axonal trajectory, is affected. Yet a different scenario is observed in the SAB motor neurons. Here, all mutant alleles show similar defects in choosing incorrect axon trajectories. The distinctiveness of axonal outgrowth defects observed in lin-23 mutants in diverse cell types and with various mutant alleles may be caused by a distinct substrate spectrum of LIN-23 in different cell types. Alternatively, LIN-23 may affect a common set of substrates that function in a cellular context-dependent manner.

Our findings provide another example of the importance of ubiquitination in axon outgrowth, first recognized upon cloning of the bendless E2 ubiquitin ligase, which affects axon targeting in Drosophila (MURALIDHAR and THOMAS 1993 ) and later corroborated by the implication of ubiquitination in Robo-mediated axon guidance at the fly midline (MYAT et al. 2002 ). A new and intriguing aspect of our studies is the implication of lin-23 in two distinct and genetically separable processes, cell cycle control vs. axonal development. The genetic separability of these two functions is consistent with several distinct possibilities. In one scenario, LIN-23 may act through a single target protein that controls both cellular processes and whose ubiquitination is differentially affected in lin-23 null vs. lin-23(ot1) mutations. To our knowledge, there is only a single potential molecular link between cell cycle and axon outgrowth. The p21^Cip1/WAF1 protein, initially described as a nuclear inhibitor of the cyclin-Cdk kinase complex (HARPER et al. 1993 ), has recently also been implicated in regulating neurite remodeling by inhibiting Rho-kinase activity in the cytoplasm (TANAKA et al. 2002 ). In an alternative scenario, LIN-23 may affect the ubiquitination of distinct targets that control different cellular processes (Fig 7C). One (set of) target(s) may be recognized through the extended C terminus, in which the ot1 mutation is localized, affecting the ubiquitination of proteins required for axon outgrowth. Distinct binding sites for substrates involved in cell cycle control may be located within the WD40 repeats (Fig 7C). The multisubstrate scenario is consistent with the notion that the fly and vertebrate orthologs of LIN-23, termed Slimb and ßTrCP, regulate the degradation of distinct components of Wingless, Hedgehog, and NFB signaling pathways (JIANG and STRUHL 1998 ; YARON et al. 1998 ). Slimb has furthermore been implicated in a variety of cellular processes such as centrosome duplication (WOJCIK et al. 2000 ), circadian timing (GRIMA et al. 2002 ; KO et al. 2002 ), and transcriptional regulation (HERICHE et al. 2003 ), further underscoring the distinct roles of a single ubiquitin ligase component.

In summary, we have provided here further evidence for the importance of ubiquitination in axon outgrowth. LIN-23 may cause ubiquitination of cell surface protein(s) required for axon guidance and target recognition. The failure to degrade or alter the subcellular targeting of such cell surface substrate(s) may lead to the axonal outgrowth defects observed in lin-23 mutants. Alternatively, LIN-23 may have a more indirect impact on axon pathfinding through an involvement in ubiquitin-mediated, yet proteolysis-independent transcriptional regulation (BACH and OSTENDORFF 2003 ).

	ACKNOWLEDGMENTS

We thank Thomas Boulin, Stephane Conte, Ruby Hsu, Jane Rosen, Wendy Yip, and Jason Tien for their involvement in the genetic screens and mutant analysis; Mike Nonet, Josh Kaplan, and Erik Jorgensen for providing nematode strains; and Iva Greenwald and Edward Kipreos for comments on the manuscript. This work was funded by a grant from the National Institutes of Health (NIH; R01 NS 39996), the Muscular Dystrophy Association, and the Christopher Reeve Paralysis Association. O.H. is a Klingenstein, Rita Allen, Irma T. Hirschl, and Sloan Fellow. P.M.L was supported by an NIH postdoctoral fellowship (NS11144). N.M. was supported by NIH Molecular Aspects of Aging (T32 AG00289-12) and NIH Vision Sciences training grants (T32 EY 13933-01).

Manuscript received July 22, 2003; Accepted for publication October 14, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALTUN-GULTEKIN, Z., Y. ANDACHI, E. L. TSALIK, D. PILGRIM, and Y. KOHARA et al., 2001  A regulatory cascade of three homeobox genes, ceh-10, ttx-3 and ceh-23, controls cell fate specification of a defined interneuron class in C. elegans. Development 128:1951-1969.[Abstract/Free Full Text]

ANTEBI, A., C. NORRIS, E. M. HEDGECOCK and G. GARRIGA, 1997 Cell and growth cone migrations, pp. 583–610 in C. elegans II, edited by D. L. RIDDLE, T. BLUMENTHAL, B. J. MEYER and J. R. PRIESS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

AVERY, L., and J. H. THOMAS, 1997 Feeding and defecation, pp. 679–716 in C. elegans II, edited by D. L. RIDDLE, T. BLUMENTHAL, B. J. MEYER and J. R. PRIESS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

BACH, I. and H. P. OSTENDORFF, 2003  Orchestrating nuclear functions: ubiquitin sets the rhythm. Trends Biochem. Sci. 28:189-195.[CrossRef][Medline]

BAI, C., P. SEN, K. HOFMANN, L. MA, and M. GOEBL et al., 1996  SKP1 connects cell cycle regulators to the ubiquitin proteolysis machinery through a novel motif, the F-box. Cell 86:263-274.[CrossRef][Medline]

BROWN, M. C., R. L. HOLLAND, and W. G. HOPKINS, 1981  Motor nerve sprouting. Annu. Rev. Neurosci. 4:17-42.[CrossRef][Medline]

BURBEA, M., L. DREIER, J. S. DITTMAN, M. E. GRUNWALD, and J. M. KAPLAN, 2002  Ubiquitin and AP180 regulate the abundance of GLR-1 glutamate receptors at postsynaptic elements in C. elegans. Neuron 35:107-120.[CrossRef][Medline]

CAMPBELL, D. S. and C. E. HOLT, 2001  Chemotropic responses of retinal growth cones mediated by rapid local protein synthesis and degradation. Neuron 32:1013-1026.[CrossRef][Medline]

CLIFFORD, R., M. H. LEE, S. NAYAK, M. OHMACHI, and F. GIORGINI et al., 2000  FOG-2, a novel F-box containing protein, associates with the GLD-1 RNA binding protein and directs male sex determination in the C. elegans hermaphrodite germline. Development 127:5265-5276.[Abstract]

CORSI, A. K., S. A. KOSTAS, A. FIRE, and M. KRAUSE, 2000  Caenorhabditis elegans twist plays an essential role in non-striated muscle development. Development 127:2041-2051.[Abstract]

CRAIG, K. L. and M. TYERS, 1999  The F-box: a new motif for ubiquitin dependent proteolysis in cell cycle regulation and signal transduction. Prog. Biophys. Mol. Biol. 72:299-328.[CrossRef][Medline]

DIANTONIO, A., A. P. HAGHIGHI, S. L. PORTMAN, J. D. LEE, and A. M. AMARANTO et al., 2001  Ubiquitination-dependent mechanisms regulate synaptic growth and function. Nature 412:449-452.[CrossRef][Medline]

EASTMAN, C., H. R. HORVITZ, and Y. JIN, 1999  Coordinated transcriptional regulation of the unc-25 glutamic acid decarboxylase and the unc-47 GABA vesicular transporter by the Caenorhabditis elegans UNC-30 homeodomain protein. J. Neurosci. 19:6225-6234.[Abstract/Free Full Text]

GRIMA, B., A. LAMOUROUX, E. CHELOT, C. PAPIN, and B. LIMBOURG-BOUCHON et al., 2002  The F-box protein Slimb controls the levels of clock proteins Period and Timeless. Nature 420:178-182.[CrossRef][Medline]

HAMMARLUND, M., W. S. DAVIS, and E. M. JORGENSEN, 2000  Mutations in beta-spectrin disrupt axon outgrowth and sarcomere structure. J. Cell Biol. 149:931-942.[Abstract/Free Full Text]

HARFE, B. D. and A. FIRE, 1998  Muscle and nerve-specific regulation of a novel NK-2 class homeodomain factor in Caenorhabditis elegans. Development 125:421-429.[Abstract]

HARPER, J. W., G. R. ADAMI, N. WEI, K. KEYOMARSI, and S. J. ELLEDGE, 1993  The p21 Cdk-interacting protein Cip1 is a potent inhibitor of G1 cyclin-dependent kinases. Cell 75:805-816.[CrossRef][Medline]

HEDGECOCK, E. M., J. G. CULOTTI, J. N. THOMSON, and L. A. PERKINS, 1985  Axonal guidance mutants of Caenorhabditis elegans identified by filling sensory neurons with fluorescein dyes. Dev. Biol. 111:158-170.[CrossRef][Medline]

HEDGECOCK, E. M., J. G. CULOTTI, D. H. HALL, and B. D. STERN, 1987  Genetics of cell and axon migrations in Caenorhabditis elegans. Development 100:365-382.