We describe here the results of genetic screens for Caenorhabditis elegans mutants in which a single neuronal fate decision is inappropriately executed. In wild-type animals, the two morphologically bilaterally symmetric gustatory neurons ASE left (ASEL) and ASE right (ASER) undergo a left/right asymmetric diversification in cell fate, manifested by the differential expression of a class of putative chemoreceptors and neuropeptides. Using single cell-specific gfp reporters and screening through a total of almost 120,000 haploid genomes, we isolated 161 mutants that define at least six different classes of mutant phenotypes in which ASEL/R fate is disrupted. Each mutant phenotypic class encompasses one to nine different complementation groups. Besides many alleles of 10 previously described genes, we have identified at least 16 novel “lsy” genes (“laterally symmetric”). Among mutations in known genes, we retrieved four alleles of the miRNA lsy-6 and a gain-of-function mutation in the 3′-UTR of a target of lsy-6, the cog-1 homeobox gene. Using newly found temperature-sensitive alleles of cog-1, we determined that a bistable feedback loop controlling ASEL vs. ASER fate, of which cog-1 is a component, is only transiently required to initiate but not to maintain ASEL and ASER fate. Taken together, our mutant screens identified a broad catalog of genes whose molecular characterization is expected to provide more insight into the complex genetic architecture of a left/right asymmetric neuronal cell fate decision.
APART from expressing a core set of features that distinguish neuronal from nonneuronal cell types, individual cell types in the nervous system express distinct batteries of genes that generate the structural and functional diversity of neuronal cell types. The diversification of cell fate in the developing nervous system presumably relies on the interplay of a host of regulatory factors. Screens for mutant animals defective in neuronal fate specification provide a powerful and unbiased approach to identify these regulatory factors. Such screens have been successfully conducted in various model systems, most prominently flies and worms, yielding valuable insights into the molecular mechanisms that control neuronal fate specification. We describe here genetic screens for neuronal fate mutants in the nematode Caenorhabditis elegans that focus on a cell fate decision executed by a single neuron class, the ASE gustatory neurons. This neuron class is composed of two morphologically bilaterally symmetric neurons, ASE left (ASEL) and ASE right (ASER). ASEL and ASER are the main taste receptor neurons of C. elegans and sense multiple chemosensory cues in a left/right asymmetric manner (Bargmann and Horvitz 1991; Pierce-Shimomura et al. 2001) (our unpublished data). Left/right asymmetric chemosensory functions correlate with the left/right asymmetric expression of a class of putative chemoreceptors encoded by the gcy (guanylyl cyclase) gene family (Yu et al. 1997; Pierce-Shimomura et al. 2001; Ortiz et al. 2006) (Figure 1). These left/right asymmetric features of ASEL and ASER provide a model to understand how functional laterality is superimposed on a morphologically symmetric structure, a hallmark of many nervous systems across phylogeny (Hobert et al. 2002; Sun and Walsh 2006).
We have previously reported the identification of mutants that affect the development of the left/right asymmetry of the ASE neurons (Chang et al. 2003; Johnston and Hobert 2003, 2005; Johnston et al. 2006). These mutants, which we termed lsy mutants (for laterally symmetric) to indicate their role in controlling the fate of two left/right asymmetric neurons, fell into four distinct classes (Figure 1A). In class I mutants, both ASE neurons adopt the fate of the ASEL neuron, with a concomitant loss of ASER cell fate (“2 ASEL” mutants). In class II mutants, the opposite phenotype is observed, with both neurons expressing ASER fate and losing ASEL fate (“2 ASER” mutants). In class III mutants, both ASEL and ASER fates are lost and in class IV mutants, either the ASEL or the ASER fate is mixed, expressing both ASEL and ASER features (Figure 1A). The molecular characterization of these mutants revealed a complex gene regulatory network composed of several transcription factors and miRNAs that control ASEL/R fate specification (Figure 1B) (Hobert 2006). These regulatory factors interact with one another in specific network configurations, including feedback and feedforward loops, to control ASEL/R fate specification (Hobert 2006). Given that both cells have the ability to adopt either the ASER or the ASEL fate, the system essentially classifies as a bistable system (Johnston et al. 2005). Conceptually similar bistable systems that also utilize double-negative feedback loops composed of gene regulatory factors include the decision of phage lambda to adopt either the lytic or the lysogenic state (Ptashne 1992).
ASEL and ASER fates are adopted after the two neurons have passed through a transient hybrid precursor state. During the hybrid precursor state, both neurons coexpress genes that during ensuing development become restricted to either ASEL or ASER (Johnston et al. 2005). However, the decision regarding which of the two ASE neurons adopts the stable ASER or ASEL fate is not controlled at the stage when the ASE neurons progress from the hybrid precursor state to their terminal state. It is rather determined very early in embryogenesis by a signal sent along the anterior/posterior axis through the glp-1/Notch signaling system (Figure 1B) (Poole and Hobert 2006). This signal represses expression of two T-box genes at the four-blastomere stage (Good et al. 2004), thereby instructing a descendant of the ABa blastomere to eventually adopt the ASEL fate and a descendant of the ABp blastomere to adopt the ASER fate (Figure 1B) (Poole and Hobert 2006). Since the Notch signal and the eventual adoption of the ASEL and the ASER fate are separated by nine cell divisions, one central unanswered question is how this information is transmitted throughout development (indicated by a “?” in Figure 1B).
Genetic screens for mutants that affect ASEL/R fate specification may not only provide the link between early embryonic signaling and terminal neuron specification but also reveal additional components of the bistable loop shown in Figure 1B. Since the bistable feedback loop involves several miRNAs, a class of regulatory molecules whose mechanism of action is poorly understood (Nilsen 2007), another expected outcome of genetic screens for ASE specification mutants is the discovery of molecules that may act together with miRNAs to control the expression of their target genes.
In this article, we provide a first step toward these goals by reporting the results of significantly expanded genetic screening efforts aimed at uncovering the complete genetic regulatory architecture required for ASEL/R fate specification. Previously published mutants (Chang et al. 2003; Johnston and Hobert 2003, 2005; Johnston et al. 2006) were derived from relatively small-scale screens totaling ∼15,000 haploid genomes (Table 1). We describe here mutants from those previous screens that were not yet reported and report the results of an almost 10-fold expansion of these screening efforts, resulting in the screening of ∼120,000 haploid genomes (Table 1). We describe informative alleles of previously known genes, such as temperature-sensitive alleles that allowed us to determine the timing of action of the bistable feedback loop described above. We describe new genes that fall into previously described phenotypic categories as well as genes that define novel phenotypic categories. Our mutant collection illustrates the genetic complexity of neuronal cell fate decisions.
MATERIALS AND METHODS
Strains and reporter transgenes:
N2 Bristol wild-type (Brenner 1974) and CB4856 Hawaiian wild-type isolates (Hodgkin and Doniach 1997) were used. Transgenes that label ASEL and ASER fates include ASEL markers otIs3V = Is[gcy-7prom∷gfp; lin-15 (+)], otIs131 = Is[gcy-7prom∷dsRed2; rol-6(d)], otIs6, and otIs114I = Is[lim-6prom∷gfp; rol-6(d)]; ASER markers ntIs1V = Is[gcy-5prom∷gfp; lin-15 (+)] and otEx2333 = Ex[gcy-5prom∷cherry, rol-6(d)]; and ASEL/R marker otIs151V = Is[ceh-36prom∷rfp; rol-6(d)]. Reporter transgenes that assay the 3′-UTR regulation in ASEL vs. ASER are otEx2646 = Ex[ceh-36prom∷gfp∷unc-543′-UTR] and otIs185 = Is[ceh-36prom∷gfp∷cog-13′-UTR].
Screening for lsy mutants:
In all screens, animals were mutagenized with EMS according to standard protocols (Brenner 1974), one to two F1 progeny of mutated P0 animals were singled on individual plates, and their ensuing progeny (F2 and F3 generation) were screened under a stereomicroscope equipped with a fluorescent light source. Animals with mutant phenotypes were picked, and their Lsy phenotype was confirmed in ensuing generations and subsequently backcrossed [alleles of previously characterized genes were generally backcrossed zero to two times, and new lsy genes and the cog-1(ot221ts) allele were backcrossed two to five times] and then mapped. Temperature sensitivity was tested by maintaining mutant animals for several generations at 15° and 25° before scoring.
As indicated in Table 1, four independent screens were conducted using three different L/R asymmetric fate reporters, expressed from four different chromosomally integrated transgenic arrays, lim-6prom∷gfp (otIs114 and otIs6), gcy-7prom∷gfp (otIs3) (all ASEL markers), and gcy-5prom∷gfp (ntIs1; ASER marker). Each reporter has its own individual advantages and disadvantages. While all markers are visible under a stereomicroscope equipped with a fluorescent light source, the ASEL reporters lim-6prom∷gfp and gcy-7prom∷gfp produce substantially more gfp fluorescence than the ASER reporter gcy-5prom∷gfp. To facilitate the screening through a large number of genomes, we therefore conducted most of our screens (100,000/119,440 haploid genomes) with ASEL markers. Among those markers, we preferred lim-6prom∷gfp over gcy-7prom∷gfp since the strong expression of lim-6prom∷gfp in the excretory gland cells provides a convenient internal control to exclude that loss of lim-6prom∷gfp expression in ASEL is not simply caused by global array loss. lim-6prom∷gfp integrants also appear to be more healthy than several independently derived gcy-7prom∷gfp integrants. One shortcoming of the usage of the ASEL markers gcy-7 or lim-6 for screening is that in embryos, these markers are bilaterally expressed in ASEL and ASER and become restricted to ASEL only postembryonically (Johnston et al. 2005). Mutant animals that arrest at embryonic stages can therefore only be assessed for the absence of gfp expression, but not for aberrant expression of ASEL fate in ASER, observed in class I (2 ASEL) mutants in larval and adult stages. This problem cannot be easily overcome through the use of gcy-5prom∷gfp, which is restricted to ASER even in embryos, since 2 ASEL mutants would produce mutants that fail to express gcy-5prom∷gfp. In practice, such non-gcy-5prom∷gfp-expressing, arrested embryos are difficult to identify in a large population of mutagenized animals. Non-gcy-5prom∷gfp-expressing, arrested embryos may have also simply died before the ASE neurons are born. Since we attempted to be as unbiased as possible in our phenotypic categories, we nevertheless also screened a substantial number of genomes with gcy-5prom∷gfp (Table 1). The use of several independent gfp arrays also controls for possible genetic background effects in the transgenic strains.
The screens with the otIs6, otIs3, and ntIs1 transgenes (34,840 haploid genomes screened) were conducted at 20° (i.e., animals were constantly maintained at this temperature) while the screen with the otIs114 transgene (84,600 haploid genomes screened) was conducted at 25° to allow for the isolation of temperature-sensitive alleles. For the screen with ntIs1, a gcy-7prom∷rfp transgene (otIs131) was contained in the background. Rfp expression from this array is too low to allow for primary screening under a stereo microscope, but this transgene allows us to quickly distinguish under a compound microscope whether aberrant gcy-5prom∷gfp expression was paralleled by aberrant ASEL marker expression. As the otIs131 array contained a dominant roller mutation, this array also facilitated the assessment of left/right asymmetric gcy-5prom∷gfp expression since animals can be observed from multiple distinct perspectives.
Mapping and allele identification:
We first tested mutants for linkage to the X chromosome by crossing them with wild-type males and examining a potential mutant phenotype in hemizygous, male F1 cross-progeny. If a phenotype was observed in the F1 male progeny, it was tested whether this is due to dominance rather than to X-linkage. If a mutant was indeed X-linked, it was, depending on the mutant phenotypic class, either complementation tested or sequenced to determine whether it was allelic to known X-linked genes with a Lsy phenotype (ceh-36, lsy-2, lim-6). If the mutant was not X-linked, a mapping strategy was employed that was contingent on the known set of previously identified, non-X-linked genes involved in ASE fate specification (class I, cog-1, unc-37; class II, lin-49, die-1, lsy-6; class III, che-1; class IV, fozi-1) (Chang et al. 2003, 2004; Johnston and Hobert 2003, 2005; Johnston et al. 2006). Each mutant that failed to express one of the ASEL or ASER fate markers could, depending on the gfp marker used, constitute either a class III (“no ASE differentiation”) or a class I or II phenotype (loss of either ASEL or ASER fate, with inappropriate execution of the opposite cell fate). To avoid remapping already known genes, we did not cross further gfp markers into the individual mutants to precisely define that phenotype, but rather immediately conducted complementation tests. All mutants with a lack of expression of either a left- or a right-specific gfp fate marker were first complemented with the sole gene that defines the class III mutant class, che-1. A failure to complement was followed by sequencing of the che-1 locus. If the mutant complemented che-1, further complementation tests were contingent on the transgenic array and the mutant phenotype. Mutants that constituted a potential class II phenotype (2 ASER phenotype; i.e., bilateral expression of the ASER marker gcy-5 or loss of expression of the ASEL marker lim-6 or gcy-7) were complemented against known, non-X-linked class II genes (die-1, lsy-6, lin-49); failure to complement was followed by sequencing of the respective loci. Mutants that constituted a potential class I phenotype (2 ASEL phenotype; i.e., bilateral expression of ASEL marker gcy-7 or lim-6 or lack of expression of ASER marker gcy-5) were initially not complementation tested against known class I genes since we had found that the first two described class I mutants, cog-1 and unc-37, displayed nonallelic noncomplementation (Chang et al. 2003; in the course of characterizing new class I mutants, we subsequently found that nonallelic noncomplementation is not a general feature of all class I genes; supplemental Table 2 at http://www.genetics.org/supplemental/). New class I mutants were therefore rather linked to individual chromosomes using single-nucleotide polymorphisms (SNPs) in the CB4856 Hawaiian wild-type isolate (Wicks et al. 2001; Swan et al. 2002; Davis et al. 2005). If located on the same chromosome as cog-1 (LGII), unc-37 (LGI), or fozi-1 (LGIII; this is a different mutant class, but like class I mutants, it also displays ectopic expression of ASEL markers in ASER), the respective loci were sequenced, leading to the identification of 28 alleles of the previously known cog-1, unc-37, and fozi-1 genes.
Mutants that were not allelic to previously known genes were further mapped using SNP markers (supplemental Table 2 at http://www.genetics.org/supplemental/). Mutants locating in similar intervals were complementation-tested against each other (supplemental Table 2). With the exception of X-linked and dominant genes, these complementation tests, as well as complementation tests mentioned above, were usually done by crossing two mutants and determining whether F1 male cross-progeny displayed the mutant phenotype.
Reporter gene constructs:
The functional consequence of the lsy-6(ot150) promoter mutation was tested by introducing this mutation into a reporter construct that monitors expression in ASEL. The wild-type reporter construct was generated by subcloning a 930-bp PCR product, extending from −930 to −1 bp relative to the lsy-6 hairpin into the pPD95.75 expression vector, using an XbaI site and a HindIII site introduced at either end of the PCR product. This construct, termed lsy-6promC∷gfp, contains the same sequences as the previously described lsy-6 promoter construct lsy-6prom∷gfp (incorrectly annotated in Johnston and Hobert 2003 as containing 2 kb of promoter sequence). In contrast to the previously described lsy-6prom∷gfp, which was generated by PCR fusion and injected with a rol-6(d) injection marker, lsy-6promC∷gfp was subcloned and was injected at 50 ng/μl with 45 ng/μl elt-2∷gfp injection marker. We note that lsy-6prom∷gfp transgenes [both extrachromosomal arrays and the integrated derivative otIs160 (Johnston et al. 2005)] are expressed with high penetrance in ASEL and also in a few other head and tail neurons (Johnston and Hobert 2003), while four different lsy-6promC∷gfp extrachromosomal arrays (otEx3071, otEx3072, otEx3079, and otEx3080) produce lowly penetrant (12–33%) expression in ASEL, generally lower gfp levels in ASEL and no readily detectable gfp expression in other neurons. These differences in gfp expression may be due to differences in copy numbers of the DNA on the arrays (linear DNA is more efficiently incorporated on DNA than circular DNA) or to the usage of different injection markers.
The lsy-6promC∷gfp construct was mutated using the Stratagene (La Jolla, CA) QuickChange II XL site-directed mutagenesis kit to introduce the ot150 allele into the reporter construct. Like the wild-type construct, this construct was also injected at 50 ng/μl with 45 ng/ul elt-2∷gfp injection marker.
Primer sequences (5′–3′) for these constructs are as follows:
For the 3′-UTR sensor construct, two previously described control sensor constructs (cog-1 3′-UTR and unc-54 3′-UTR) were rescored (Didiano and Hobert 2006). A cog-1 3′-UTR sensor construct replicating the ot123 deletion was generated by PCR amplification (see primer sequences below) of the wild-type cog-1 3′-UTR; the amplicon was digested with EagI and EcoRI and was used to replace the unc-54 3′-UTR in the previously described ceh-36prom sensor construct (Didiano and Hobert 2006). The sensor construct was injected at 5 ng/μl with rol-6(d) at 100 ng/μl as the injection marker.
Primer sequences (5′–3′) for these constructs are as follows:
All sensor construct-expressing strains were maintained at 20° prior to scoring. All lines were scored under a Zeiss Axioplan 2 microscope. To minimize the inclusion of mosaic animals only those animals were scored in which gfp expression in the AWCL and AWCR neurons was observed. The fluorescence intensity of the 3′-UTR sensor constructs was compared between ASEL and ASER in each individual animal and scored as ASEL > ASER, ASEL = ASER, or ASER > ASEL (see Didiano and Hobert 2006 for a more detailed explanation of scoring criteria).
Microscopy, laser ablation, and phenotypic observations:
Laser ablations were done as previously described (Poole and Hobert 2006). Embryos were dissected from gravid hermaphrodites, mounted at the one- to four-cell stage in a drop of water on a 5% agar or agarose pad, and sealed between a coverslip and slide using melted Vaseline. A Photonics dye laser attached to the microscope was used to ablate embryonic blastomeres. Larger early blastomeres were irradiated for 3–5 min with the laser beam, while smaller later blastomeres were irradiated for 1–2 min in and around the nucleus. Increased cytoplasmic movements were often observed and in many cases the nucleus was observed to break down. Irradiated blastomeres usually did not divide although on some occasions late aberrant divisions were observed. Laser ablations were performed at 25°. A copper ring attached to the microscope objective through which temperature-controlled water was passed maintained the constant temperature of the slide. Following ablation the slides were left overnight for ∼12–15 hr at 25° and then scored for the number of cells exhibiting transgene expression at late threefold/L1 stages.
RESULTS AND DISCUSSION
Overview of the screen
We used three different L/R asymmetrically expressed gfp reporter genes to monitor ASEL or ASER fate. lim-6prom∷gfp is expressed in ASEL and in the excretory gland cells, gcy-7prom∷gfp is expressed in ASEL and weakly in the excretory cell of adult animals, and gcy-5prom∷gfp is expressed exclusively in ASER. We conducted separate F1 semiclonal screens for mutants in which the expression of the respective gfp marker is different from expression in wild-type animals (see materials and methods for a detailed description of the screen and for an explanation on why different markers were used). A total of 119,440 haploid genomes were screened (100,000 with ASEL markers and 19,440 with an ASER marker) and 161 mutants with gfp expression defects were retrieved (Table 1). Mutant phenotypes are 2–100% penetrant. We focused our analysis on 123 mutants. The other mutants were not pursued since they display defects that are <10% penetrant (34 mutants), have multiple loci mutated (2 mutants), or have the gfp-expressing array affected (2 mutants) (supplemental Table 1, A–C, at http://www.genetics.org/supplemental/). Most of the isolated mutant strains were viable and recessive (Table 3). By complementation testing, mapping, and allele sequencing (see materials and methods), these 123 alleles were found to define 30 complementation groups (Tables 2 and 3; supplemental Table 2 at http://www.genetics.org/supplemental/). The mutation rate of 1 mutant allele per 740 haploid genomes is within the range of the average mutation rate observed in other mutant screens (Anderson 1995).
We observed six mutant classes that show aberrant expression of ASEL/R fate markers (classes I–VI), two of which had not been previously described (class V and class VI) (Figure 2). All mutant classes can be summarized as follows. As mentioned in the Introduction (Figure 1A), in class I mutants, ectopic expression of the ASEL fate is observed in ASER with concomitant loss of the ASER fate markers (i.e., lim-6prom∷gfp or gcy-7prom∷gfp is expressed in two cells rather than one cell and gcy-5prom∷gfp expression is lost); in class II mutants, the opposite phenotype is observed; in class III mutants ASEL and ASER cell fate is not executed (lim-6prom∷gfp, gcy-7prom∷gfp, and gcy-5 not expressed); and in class IV mutants (“mixed fate”), ASEL fate is ectopically expressed in ASER, but the ASER fate marker continues to be expressed (or, vice versa, ASER fate is ectopically expressed in ASEL with ASEL fate markers being unaffected) (Figure 1A). In two mutant classes not previously described, we observe a “heterogeneous phenotype” with a combination of loss and gain of gfp expression of individual markers in ASEL or ASER within a mutant population (class V mutants) and an ectopic gfp expression phenotype in which cells other than the ASE neurons express the cell fate marker (class VI mutants). In the following sections, we describe these mutants and phenotypic classes in more detail.
Class I mutants (2 ASEL neurons)
Alleles of known class I genes:
We have isolated a total of 19 alleles of the previously known class I gene cog-1, a homeobox gene (Palmer et al. 2002; Chang et al. 2003), and 2 alleles of unc-37, the previously known C. elegans ortholog of the Groucho corepressor family (Table 3; Figure 3) (Pflugrad et al. 1997; Chang et al. 2003). Each allele displays a characteristic 2 ASEL phenotype, manifested by the gain of ASEL markers in ASER and the concomitant loss of ASER markers in ASER (Figures 1A and 2A). Representative class I mutant animals display the phenotypic defects expected by the loss of ASER fate; i.e., these mutants have defects in their ability to sense ASER-specific chemosensory cues (data not shown). This observation complements previous observations made with class II mutants (defective in ASEL-sensed cues; Chang et al. 2004), class III mutants (defective in sensing both ASEL and ASER cues; Uchida et al. 2003), and class IV mutants (discrimination defects of ASEL and ASER cues; Pierce-Shimomura et al. 2001).
Most cog-1 alleles lie within the homeobox (Figure 3). We were unable to identify mutations in the coding region or UTRs of cog-1 in two mutants, ot119 and ot201, which we conclude are cog-1 alleles because they (a) are tightly linked to the cog-1 locus, (b) fail to complement a canonical cog-1 mutation (supplemental Table 2 at http://www.genetics.org/supplemental/), and (c) are rescued by a fosmid, WRM067cF11, that contains the cog-1 locus [two of seven lines rescue the 62%-penetrant 2 ASEL phenotype of ot119 each to 15% (n = 55) and 7% (n = 52), respectively; two of eight lines rescue the 80%-penetrant 2 ASEL phenotype of ot201 to 9% (n = 51) and 30% (n = 65), respectively]. These mutations may reside in regulatory elements that control cog-1 expression.
By conducting most of the genetic screens at 25° (84,600 of the total of 119,440 haploid genomes screened; Table 1), we aimed to identify temperature-sensitive alleles of known genes that would help us address the question of at what stage the previously described regulatory factors (Figure 1B) act to control ASE fate determination. We indeed found that five alleles of cog-1 show a strong temperature-sensitive phenotype. All five alleles harbor the same Ala > Val substitution that is highly conserved in most homeodomain proteins (Figure 4A) and is located in the third α-helix of known homeodomain structures (Figure 4B) (Piper et al. 1999). A slightly more bulky Val residue in this position may affect the packing of the three α-helixes, thereby possibly destabilizing the protein at higher temperatures. We note that a mutation in an adjacent position of the homeodomain of UNC-4 (Leu > Phe) also causes a temperature-sensitive phenotype (Miller et al. 1993). Other less specific impacts of Ala > Val mutations, e.g., on protein synthesis, cannot be ruled out.
We characterized one of these alleles, ot221, in more detail. ot221 animals raised at 15° show an almost 100% penetrant wild-type phenotype, while animals raised at 25° show a ∼50% penetrant 2 ASEL phenotype (Figure 4C). By performing both up- and downshift experiments at different developmental stages we found that cog-1 gene activity is required embryonically only right around or shortly after the birth of the ASE neurons to control terminal ASEL and ASER fate, as assessed by the expression of three different differentiation markers (Figure 4D). Raising or lowering cog-1 gene activity after the threefold stage has no effect on the expression of the three terminal differentiation markers that we tested (Figure 4D).
These temperature-shift experiments demonstrate that the bistable feedback loop, of which cog-1 is a critical component, is required only embryonically to initiate the induction of the L/R asymmetric, terminal differentiation features of ASEL and ASER. After this initial choice has been made, terminal differentiation features must rely on some other mechanisms to maintain their activity. This finding is perhaps unexpected since feedback loops are often involved in stabilizing and maintaining cellular fates (Edlund and Jessell 1999) and since the expression of bistable loop components, including cog-1, persists in ASE throughout adulthood, as measured by various reporter gene constructs (data not shown). Given our temperature-shift experiments, we speculate that the ASE feedback loop may rather act to amplify an initial, perhaps transient or low-level input into the system, rather than stabilizing it. Fate stabilization and maintenance may be ensured by positive autoregulation of transcription factors that provide the output from the loop (die-1) or act downstream of the loop, such as lim-6, which we have previously shown to autoregulate (Johnston et al. 2005). Moreover, we note that in a newly isolated allele of the class IV gene fozi-1, a transcription factor, the mutant phenotype appears more penetrant in adult vs. larval stage animals (ot191 allele), which argues for a continuous requirement of the gene throughout larval and adult stages. The lack of requirement for the bistable feedback loop postembryonically may also explain a puzzling observation that we previously reported, namely that in lim-6 mutants, lsy-6 expression is partially lost without having an effect on expression of terminal fate markers such as gcy-7 (Johnston et al. 2005). We infer that the loss of lsy-6 in lim-6 mutants occurs too late to have an impact on the temporally restricted, ASEL-inducing activity of lsy-6.
In addition to class I cog-1 alleles, we isolated two recessive alleles of the unc-37 Groucho-type transcriptional cofactor (Table 3; Figure 3). Both alleles cause an embryonic lethal phenotype that is maternally rescued. The Lsy phenotype is not maternally rescued and can be readily observed in the homozygous offspring of heterozygous mothers (Table 3).
On the basis of their physical interactions in other systems as well as genetic interaction tests, we proposed that the COG-1 and UNC-37 proteins directly interact to control the postmitotic adoption of terminal ASEL and ASER fates (Chang et al. 2003). However, a recent analysis of maternal unc-37 function in the early embryo reveals a role for this gene in early, Notch-mediated lineage decisions (Neves and Priess 2005). unc-37 acts as a corepressor with the ref-1 family of bHLH transcription factors downstream of Notch signaling in the early embryo. In ref-1 mutants, a second ectopic ASEL neuron is generated from the ABara blastomere (Poole and Hobert 2006). To test whether the cell that ectopically expresses ASEL fate in hypomorphic unc-37(e262) mutants (Chang et al. 2003) is caused by an early lineage transformation in the ABara lineage or whether it is caused by the transformation of the ABp-derived ASER neuron into ASEL, we laser ablated the ABp blastomere in unc-37(e262) mutants. This manipulation abolished one of the two cells that expresses ASEL fate, which indicates that this cell is not derived from the ABara blastomere and which is consistent with this cell being ASER (data not shown). Taken together, unc-37 functions independently of its early embryonic patterning role to cause a 2 ASEL phenotype.
New class I mutants:
In addition to cog-1 and unc-37 alleles, we found six recessive class I mutations that define four novel class I complementation groups, termed lsy-5, lsy-16, lsy-17, and lsy-22 (Table 3, supplemental Table 2 at http://www.genetics.org/supplemental/). Up to two alleles per locus were identified. The penetrance of the mutant phenotypes ranges from 35 to 100% penetrant (Table 3). lsy-16 and lsy-17 are fertile and viable, while the lsy-5 and lsy-22 genes are required for viability. The embryonic lethality of lsy-22 mutants is maternally rescued but the Lsy phenotype is not. The most severe allele of lsy-5 mutants, ot37, displays zygotic embryonic and larval lethality. Escapers are uncoordinated (Unc) and display a protruding vulva (Pvl) phenotype but produce no progeny. A weaker allele of lsy-5, ot240, is homozygous viable but also Unc and Pvl.
We previously noted that mutations in the first two identified class I genes, cog-1 and unc-37, failed to complement each other (Chang et al. 2003). The nonallelic noncomplementation is likely a reflection of a direct physical interaction of COG-1 and UNC-37 proteins, which we infer from the ability of COG-1 orthologs (Nkx6 family) to physically interact via the EH1 domain with vertebrate UNC-37 orthologs (Muhr et al. 2001). Nonallelic noncomplementation is, however, not a general feature shared by class I genes. For example, lsy-5 and lsy-16 each complement unc-37 and the new class I lsy genes show complementation among each other (supplemental Table 2 at http://www.genetics.org/supplemental/).
Future molecular characterization and genetic epistasis analysis will reveal how lsy-5, lsy-16, lsy-17, and lsy-22 fit into the known regulatory architecture of ASE fate specification shown in Figure 1B. Given that they fall into the same phenotypic category as cog-1, it is possible that some of these genes may mediate the control of the die-1 3′-UTR by the cog-1 gene. Our previous work indicated that cog-1 may exert its effect on the die-1 3′-UTR through mir-273, a miRNA that is expressed in a cog-1-dependent manner in ASER and that is sufficient to downregulate die-1 expression (Chang et al. 2004; Johnston et al. 2005). However, mir-273 alleles were not recovered from our screens, perhaps because of a potential redundancy of mir-273 with several sequence-related miRNAs (mir-51 through mir-56) or because another factor (miRNA or protein) may have a more prominent role in controlling the die-1 3′-UTR. In that latter case, it is conceivable that the molecular identity of lsy-5, lsy-16, lsy-17, and lsy-22 will provide a better understanding of how cog-1 controls the die-1 3′-UTR.
Class II mutants (2 ASER neurons)
Alleles of known class II genes:
We have isolated a total of 22 recessive alleles of known class II genes (die-1, lsy-2, ceh-36, lin-49, and lsy-6) and one gain-of-function allele of a previously described class I gene, cog-1.
die-1 encodes a C2H2 Zn-finger transcription factor (Heid et al. 2001). In contrast to the early embryonic lethality of a die-1 null mutant allele (Heid et al. 2001), all die-1 alleles that we isolated are viable and all except one of the die-1 alleles show a 100% penetrant Lsy phenotype (Table 3). With the exception of ot100, a point mutation in the first Zn finger, all other alleles cluster at the C terminus of the protein and constitute either nonsense or missense mutations (Figure 3). This C-terminal region contains no recognizable sequence motif, but missense mutations alter residues that are conserved in three nematode species (data not shown). Considering the viability of these alleles and their completely penetrant Lsy phenotype, it is possible that these alleles have revealed genetically separable functions of distinct parts of the DIE-1 protein. For example, the C terminus of DIE-1 may interact with a cell-type-specific cofactor in ASEL. Alternatively, all of these alleles may simply produce less overall gene activity and the essential function of die-1 in the hypodermis (Heid et al. 2001) may require less gene activity than die-1 function in the ASE neurons.
lsy-2 encodes a C2H2 Zn-finger transcription factor (Johnston and Hobert 2005). All previously described alleles of lsy-2 cause a completely penetrant sterility (Ste) phenotype, whose basis is currently unknown (Johnston and Hobert 2005). We have now also recovered an allele, ot90, a splice donor mutation (Figure 3), which displays the Lsy but not the Ste phenotype.
We mapped the only dominant gain-of-function mutation retrieved in our screen, ot123 [48% penetrant 2 ASER phenotype in heterozygous state (n = 114); 100% penetrant in homozygous state (n = 76)] to the cog-1 locus. Sequence analysis revealed a 329-bp deletion that eliminates the last 105 bp of the cog-1 3′-UTR (Figure 5A). The cog-1 3′-UTR is downregulated in ASEL by the lsy-6 miRNA (Johnston and Hobert 2003) and ectopic expression of cog-1 in ASEL transforms ASEL into ASER (Chang et al. 2003). A failure to downregulate endogenous cog-1 expression therefore causes the same 2 ASER phenotype that we observe in ot123 animals (Table 3). To experimentally test whether the 3′-UTR deletion in ot123 mutants indeed causes a failure to downregulate the cog-1 3′-UTR, we employed a previously described sensor gene strategy (Johnston and Hobert 2003; Didiano and Hobert 2006). In this approach the ASEL/ASER-expressed promoter of the ceh-36 gene is fused to gfp, followed by the cog-1 3′-UTR. This sensor construct is downregulated in ASEL and this downregulation can be observed by assessing relative gfp levels in ASEL vs. those in ASER (Figure 5B). Introduction of a deletion into this sensor that corresponds to the ot123 mutation causes a failure to downregulate gfp expression, and therefore the cog-1 3′-UTR, in ASEL (Figure 5B). We conclude that in cog-1(ot123) animals, ectopic expression of cog-1 in ASEL causes the cell to adopt an ASER fate.
We note that the ot123 deletion does not delete the lsy-6 complementary site in the cog-1 3′-UTR (Figure 5A), which is required for cog-1 downregulation (Johnston and Hobert 2003). As the lsy-6 complementary site is not sufficient to confer downregulation of a 3′-UTR (Didiano and Hobert 2006), we conclude that the ot123 mutation affects another cis-regulatory or structural element of the cog-1 3′-UTR that is required for lsy-6-mediated downregulation of cog-1.
We had previously described the isolation of a single allele of the miRNA lsy-6, a deletion allele that eliminates the complete miRNA (ot71; Figure 6A) (Johnston and Hobert 2003). Our expanded screening efforts resulted in the isolation of three more lsy-6 alleles, two point mutations (ot149 and ot150) and a deletion allele (ot182) that eliminates the entire lsy-6 locus (Figure 6A). All alleles are completely viable, have normal brood sizes, and display no obvious pleiotropies. With a total of four alleles, the allele frequency of the lsy-6 locus is one in 24,000 haploid genomes. The only two other cloned genes besides lsy-6 whose null phenotype does not impinge on brood size, overall health, or viability are more frequently isolated (che-1, 1/5400; fozi-1, 1/10,000; Table 2) but as these protein-coding loci are also much larger than lsy-6 they are more likely to be targeted by a random mutagen.
The ot149 point mutation in the lsy-6 locus is a G > U substitution. The mutated G nucleotide normally participates in base pairing within the lsy-6 hairpin precursor (Figure 6B, top) and in base pairing between the mature lsy-6 miRNA with its complementary sequence in the cog-1 3′-UTR (Figure 6B, bottom).
In contrast to the other lsy-6 alleles, the lsy-6(ot150) allele is only weakly penetrant (Table 3) and, like the ot149 and ot182 alleles, was identified as a lsy-6 allele by noncomplementation with the lsy-6 reference allele ot71 (data not shown). lsy-6(ot150) animals harbor a C → T mutation ∼100 nucleotides upstream of the predicted lsy-6 hairpin structure (Figure 6A). We considered the possibility that the mutation might affect a cis-regulatory element required for lsy-6 expression. Consistent with such a possibility, the alignment of the lsy-6 upstream region in C. elegans, C. briggsae, and C. remanei reveals that the ot150 allele lies in a conserved patch of nucleotides, consistent with this region being functionally relevant (Figure 6A). To test the functional relevance of this region experimentally, we introduced this mutation into a reporter gene construct that monitors expression of lsy-6 in the ASEL neuron. The introduction of this mutation completely abolishes expression of the reporter in ASEL (Figure 6C). The ot150 mutation therefore indeed disrupts a cis-regulatory element required for lsy-6 expression.
A recent analysis of cis-regulatory elements of genes expressed in ASE identified a cis-regulatory element, termed the ASE motif, that is required for expression of genes in ASE and constitutes a binding site for the Zn-finger transcription factor CHE-1 (Etchberger et al. 2007). The ot150 mutation affects one of the invariant positions of the ASE motif (Figure 6D).
ceh-36 and lin-49:
Notably, we did not recover additional alleles of the previously described lin-49 and ceh-36 genes, two transcription factors required for expression of the ASEL fate (Chang et al. 2003), even though we have screened ∼10 times as many haploid genomes as in the previously described smaller-scale screens. Since our original publication of the only ceh-36 allele that resulted from our initial screen, ot79 [a premature stop codon very late in the coding sequence (Chang et al. 2003)], two additional, also completely viable alleles of ceh-36, ky640, and ky646, have become available (Lanjuin et al. 2003). Both ky640 animals and ot79 mutant animals harbor a premature stop codon late in the gene, and each allele displays a partially penetrant class II (2 ASER) phenotype (Chang et al. 2003; Lanjuin et al. 2003). In contrast, the ky646 allele, which harbors a premature stop codon at the beginning of the homeobox, was reported to display a much weaker Lsy phenotype (Lanjuin et al. 2003). To exclude that this surprising result is not caused by different scoring criteria or conditions in different laboratories, we compared adult mutant animals side-by-side at carefully controlled temperatures. We observed that all alleles display a previously unnoted strong temperature sensitivity with a much stronger phenotype at 15°, compared to 25° (Table 4). This is particularly evident for ky646 animals, which display no phenotype at 25°, but a >50% penetrant phenotype at 15°. This observation suggests that ky646, in spite of its premature stop codon, may not be a null allele. Indeed, two ceh-36 deletion alleles, independently isolated by two different knockout consortia, display embryonic lethality (http://www.wormbase.org), which likely represents the true null phenotype of the gene. Such a phenotype is consistent with an early embryonic expression of ceh-36 at epidermal closure (data not shown). Moreover, RNAi of ceh-37, a closely related paralog of ceh-36 (Lanjuin et al. 2003), shows no effect on viability or ASE laterality in a wild-type background, but causes embryonic lethality if done in a viable ceh-36(ky646) mutant background (data not shown). Taken together, the cold sensitivity and early embryonic pleiotropy of ceh-36 gene function may decrease the frequency of allele recovery in our screens, which were mainly conducted at 25°. Pleiotropies that affect the viability of strong lin-49 loss-of-function alleles (Chamberlin and Thomas 2000; Chang et al. 2003) may also explain the low recovery rate of lin-49 mutants.
New class II mutants:
We recovered nine class II mutant alleles that define four novel class II genes, lsy-12, lsy-14, lsy-15, and lsy-19 (Table 3, supplemental Table 2 at http://www.genetics.org/supplemental/). All five loci are recessive, viable, and defined by one to five alleles, ranging in penetrance from 50 to 100% (Table 3). One viable mutant, lsy-12(ot86), has the unique property of being maternally rescued as revealed by scoring the homozygous progeny of heterozygous parents. An egg-laying-defective (Egl) phenotype associated with lsy-12 is not maternally rescued.
Future molecular characterization and genetic epistasis analysis will reveal how lsy-12, lsy-14, lsy-15, and lsy-19 fit into the known regulatory architecture of ASE fate specification (Figure 1B). As these genes show similar phenotypes to the lsy-6 miRNA, it is conceivable that one of these loci may be a miRNA-specific cofactor of lsy-6 activity that may provide a better understanding of miRNA function. Such cofactors are beginning to emerge from other systems (Nolde et al. 2007). As expected, we have not retrieved generic miRNA processing enzymes, such as Dicer, due to early development pleiotropies associated with loss of these genes (Grishok et al. 2001). As is the case for novel class I genes, the novel class II genes described here may also act far outside the bistable feedback loop to provide a link between the early embryonic Notch induction and the bistable feedback loop (? in Figure 1B). An early embryonic function is a particularly attractive possibility for lsy-12, which appears to be contributed maternally.
Class III mutants (no ASEL/R differentiation)
All class III mutants that we identified define a single complementation group on chromosome I, che-1 (Figure 3; Table 3), which codes for a Zn-finger transcription factor (Chang et al. 2003; Uchida et al. 2003). We have identified a total of 22 che-1 alleles, which include splice site, missense, and nonsense mutations in the coding region of the gene (Figure 3). All of the nonsense mutations are located before the C-terminal Zn-finger domains. All of the amino acid-changing missense mutations reside in three of the four Zn-finger domains (Zn-finger 2, 3, and 4). We have recently shown that Zn-finger 3 and 4, but not Zn-finger 1 and 2, are involved in binding of CHE-1 to the ASE motif of the gcy-5 gene (Etchberger et al. 2007). The ASE motif is a cis-regulatory element that is present in many ASE-expressed promoters (Etchberger et al. 2007). The apparent requirement of Zn-finger 2 for ASE development, revealed by our genetic analysis, suggests that CHE-1 may show a differential domain requirement for binding to different ASE motifs.
The loss of genes that control the tightly regulated expression of che-1 in ASE (Chang et al. 2003; Uchida et al. 2003) would be expected to yield a class III mutant phenotype, yet no class III mutants other than che-1 were found. We presume that either those upstream regulators have pleiotropies that prevented their retrieval (e.g., essential functions in early embryonic development) or a redundant set of regulatory factors may control che-1 expression.
Class IV mutants (mixed fate of ASEL or ASER neurons)
Alleles of known class IV genes:
We have previously identified two genes with a class IV “mixed” phenotype, the LIM homeobox gene lim-6 and the C2H2 Zn-finger factor fozi-1 (Hobert et al. 1999; Johnston et al. 2006). In lim-6 mutants, ASER fate is ectopically expressed in ASEL without the concomitant complete loss of ASEL fate, while in fozi-1 mutants, ASEL fate markers are ectopically expressed in ASER without the concomitant loss of ASER fate markers (schematically shown in Figure 1A). Our extended screening efforts have retrieved 11 additional alleles of fozi-1, many of them premature stop codons (Table 3; Figure 3). Two missense mutations map into the first Zn-finger domain, thereby corroborating its previously reported importance in fozi-1 function (Johnston et al. 2006). No missense mutations in the FH2 domain were identified, corroborating our previous conclusion that the FH2 domain is not essential for fozi-1 function (Johnston et al. 2006). The ot191 and the ot61 allele, both nonsense mutations in the FH2 domain, may destabilize the mRNA and/or protein. Animals that carry the ot191 allele show a more penetrant phenotype in adult compared to larval stages (data not shown), suggesting that fozi-1 may be continuously required to maintain ASER fate.
We have not isolated any alleles of the LIM homeobox gene lim-6, a previously known class IV gene. This gene was the first regulatory gene to be implicated in ASE laterality, on the basis of the analysis of a single, reverse-genetically engineered deletion allele, nr2073, which eliminates the complete DNA-binding domain of lim-6 (Hobert et al. 1999). lim-6(nr2073) mutants show only partially penetrant and expressive effects on the expression of the gfp markers that we used for screening (Johnston et al. 2005, 2006), thereby providing a possible explanation for our failure to retrieve lim-6 alleles.
New class IV mutants:
We have identified three novel class IV mutant loci, lsy-18, lsy-20, and lsy-26 (Table 3). The mutant phenotypes of both lsy-20 and lsy-26 are defined by the ectopic expression of the ASER marker gcy-5 in ASEL. In contrast to class II mutants, the ectopic expression of the ASER marker is not accompanied by a loss of expression of the ASEL fate marker (Table 3). The ASEL neuron therefore displays a mixed fate, characterized by coexpression of ASEL and ASER fate markers. This phenotype is similar to that of lim-6(nr2073) mutants (Hobert et al. 1999). lim-6 has distinct functions in different neuron types (Hobert et al. 1999). It is conceivable that lsy-20 and lsy-26 are transcription factors that may act together with lim-6 to determine its ASE-specific function in controlling gcy and flp gene expression patterns in ASEL (Figure 1B). Neither lsy-20 nor lsy-26 share the Egl phenotype of lim-6 that is caused by loss of lim-6 function in the uterus (Hobert et al. 1999).
In contrast to lsy-20 and lsy-26, lsy-18 displays a fozi-1-like mixed phenotype in which ASEL fate is derepressed in ASER without a concomitant loss of the ASER fate marker gcy-5. Like in fozi-1 mutants (Johnston et al. 2006), the derepression of ASEL fate in ASER is fully penetrant (100% of animals show derepression of lim-6prom∷gfp in ASER) but only partially expressive (levels of ectopic gfp expression in ASER vary and often do not reach the level of expression normally seen in ASEL). lsy-18 may act together with fozi-1 to distinguish between the roles of fozi-1 in neuronal and mesodermal cell specification (Johnston et al. 2006; Amin et al. 2007).
Class V mutants (heterogeneous phenotype)
We recovered five alleles with novel phenotypes. Four of these alleles define a new complementation group on chromosome IV, termed lsy-9, which displays a heterogeneous phenotype. Within a population of mutant animals, a spectrum of phenotypes is observed, including the loss of the ASEL marker lim-6, ectopic expression of lim-6 in ASER, and, surprisingly, exclusive expression of lim-6 in ASER (Figure 2B; Table 3). The ASER fate marker gcy-5 is frequently lost, but never ectopically expressed in ASEL (Table 3).
One allele, ot147, defines a new locus on chromosome II, lsy-21, which displays another type of a heterogeneous phenotype that varies within a population. Some animals lose expression of the ASEL marker lim-6 in ASEL while a subset of these animals also lose expression of the bilateral ASEL/R marker ceh-36 in ASEL but not in ASER (data not shown), indicating a partial loss of ASE cell fate.
At this point it is difficult to speculate on the potential role of class V genes. It is conceivable that these mixed phenotypes reflect a reduced but not completely eliminated gene dosage. The characterization of null phenotypes of these loci will eventually resolve this issue.
Class VI mutants (ectopic ASEL/R fate)
We identified a substantial number of mutants that display normal L/R asymmetric expression of the ASE fate markers in the ASE neurons, but ectopic expression of ASEL or ASER markers in cells other than ASEL or ASER (Figure 2C). In spite of the normal ASEL/R laterality, we still refer to this phenotype as a Lsy phenotype as the mutants were isolated with gfp markers that monitor ASEL/R laterality.
Cell death mutants:
The sisters of ASEL and ASER normally undergo programmed cell death (Sulston et al. 1983). We retrieved a substantial number of alleles of previously known regulators of cell death, namely 12 alleles of ced-3 and 5 alleles of ced-4 (Yuan et al. 1993; Shaham and Horvitz 1996; Figure 3). All alleles display ectopic expression of ASEL fate markers in the undead sister of ASEL (Figure 2; Table 5). Similarly, undead ASER sisters execute the ASER fate (Table 5). Although we found that a reference loss-of-function allele of the cell death gene egl-1 also causes a class VI phenotype (data not shown), we did not retrieve egl-1 alleles, possibly because the locus is very small (Conradt and Horvitz 1998). Undead cells have previously been shown to execute the fate of their lineal sisters (e.g., Avery and Horvitz 1987; Guenther and Garriga 1996). In the case of the ASE neurons, the observation that the undead ASEL or ASER sister cell expresses the appropriate fate [rather than being “stuck” in the bilateral ASE precursor state (Johnston et al. 2005)] provides a further argument for left/right asymmetric fate being programmed into the ASEL and ASER lineage before the ASE neurons are generated (Poole and Hobert 2006). Also, it argues that the cleavage plane of the last cell division generating the ASE neurons has no impact on the adoption of ASE fate (i.e., both the anterior and the posterior daughter of the ASE mother cell can adopt the ASE fate).
Ion channel mutants:
Two mutant alleles displayed normal expression of lim-6 in ASEL, but also ectopic expression in a pair of bilaterally symmetric sensory neurons located anteriorly to ASE, the AFD thermosensory neurons (Figure 2C; Table 6). Mapping and complementation tests revealed that these mutations affect the two subunits of a previously identified, AFD-expressed cyclic nucleotide gated ion channel, tax-2 and tax-4 (Figure 3, Table 6) (Coburn and Bargmann 1996; Komatsu et al. 1996). Unlike previously described tax-2 alleles, tax-2(ot25) appears to be a clear molecular null allele (early stop codon; Figure 3). Similar defects in the expression of the ASEL fate marker lim-6 can be observed using previously described reference alleles of tax-2 and tax-4 (Table 6). Ectopic expression of ASE fate in AFD is restricted to the lim-6 gene. Neither the ASEL marker gcy-7, nor the ASER marker gcy-5, nor the bilateral ASEL/R marker ceh-36 display any expression defects in tax-2/tax-4 mutants (Table 6 and data not shown).
Loss of tax-2/tax-4 was previously shown to affect gene expression of a putative olfactory receptor in the ASI and AWC neurons (Troemel et al. 1999; Peckol et al. 2001); an effect of these channels on the regulation of gene expression is therefore not unprecedented. However, in contrast to the previously described case, the expression of lim-6 in AFD is not controlled by the calcium-dependent UNC-43 kinase since neither gain- nor loss-of-function mutations in the unc-43 gene affect lim-6 expression (Chang et al. 2003).
We identified mutations in two genes on LGII and LGX, termed lsy-23 and lsy-24, that show normal ASEL/R laterally but display ectopic lim-6prom∷gfp expression in the AVL motor neuron (Figure 2C). As is the case for tax-2/tax-4, the ectopic expression is observed only with the lim-6 fate marker and not with the alternative ASEL marker gcy-7prom∷gfp (data not shown).
We isolated a single allele of a gene on LGIV, termed lsy-25(ot97), which displays a completely penetrant embryonic lethal phenotype that is maternally rescued. Homozygous lsy-25 embryos derived from a heterozygous mother show no aberrant ASEL/R specification, but the unviable offspring of these animals show up to two gcy-5prom∷gfp-expressing cells (wild-type animals at similar embryonic stages of development always show only one gcy-5prom∷gfp-expressing cell) and up to two gcy-7prom∷gfp-expressing cells [two is normal for wild-type animals as mature ASEL markers label both ASE neurons in the embryo (Poole and Hobert 2006)] (Table 7). The apparent 2 ASER phenotype of this mutant is, however, not caused by a transformation of ASEL into ASER (as in class II mutants), but caused by a lineage transformation within the ABp lineage. In wild-type animals, the ASER neuron is generated from ABp and the ASEL neuron from ABa. ABa ablation in wild-type animals therefore eliminates ASEL and ABp ablation eliminates ASER (Table 7) (Poole and Hobert 2006). In lsy-25(ot97) animals, the ablation of ABp eliminates both gcy-5prom∷gfp-expressing cells (Table 7). As ASEL is derived from ABa, the ectopic gcy-5prom∷gfp cell can therefore not be an ASEL > ASER transformed neuron, but is rather a descendant of the ABp lineage that normally produces ASER.
lsy-25(ot97) animals not only generate an additional ASER cell from the ABp lineage, but also lack the normally ABa-derived ASEL neuron. We derived this notion from laser ablation of ABp in lsy-25(ot97) animals expressing gcy-7prom∷gfp, which normally marks both ASEL and ASER in embryos. ABp ablation completely eliminates gcy-7prom∷gfp expression. We conclude that in lsy-25(ot97), the ABa lineage fails to produce an ASEL neuron, while a lineage transformation in ABp generates an ectopic ASE lineage that expresses the ASER fate. We therefore classify lsy-25 as a class VI mutant.
Saturation of the screens
Having determined the number of complementation groups recovered by our screens, we considered the degree of saturation of our genetic screens. Hints about saturation are contrasting: On the one hand, (a) 13 nucleotide positions in six different loci have been hit multiple times (one even five times; summarized in supplemental Table 3 at http://www.genetics.org/supplemental/), (b) multiple alleles of the small miRNA lsy-6 have been retrieved, (c) several loci are represented by >10 alleles (Figure 7), and (d) viable hypomorphic alleles of essential genes (e.g., die-1, lin-49, lsy-5) have been isolated. On the other hand, eight loci are represented only by single alleles (Figure 7). These eight single-allele loci do not reside within a particular class (Table 2, Figure 7), indicating that alleles of any one class are not harder to uncover than others. Genes defined by single alleles may have pleiotropic functions and these single alleles may be rare alleles that do not provide enough activity for their function in ASE development but do provide enough activity for other functions.
To analyze the degree of saturation in a more quantitative manner, we utilized a method introduced by Pollock and Larkin (2004) that calculates the maximum-likelihood estimates of the number of alleles that remain to be found. The algorithm attempts to fit the given data (number of alleles vs. number of loci) to various models that describe a different probability of discovering mutants across the genome. The model that fits our allele distribution the best predicts a 74% saturation of our screen (supplemental Table 3 at http://www.genetics.org/supplemental/), indicating that up to 12 class I–class V loci in addition to the currently known 23 class I–class V loci (Table 2) remain to be identified. We caution that this is only a rough estimate that does not take into account 34 mutants with a <10% penetrant mutant phenotype. Future analysis of these mutants may increase allele coverage for several loci, thereby causing a different saturation estimate.
One approach to understanding how cellular diversity in the nervous system is genetically encoded is to employ genetically amenable model systems to screen for mutants in which developmental programs in the nervous system are disrupted. This endeavor has been substantially aided by the introduction of reporter gene technology that allows the visualization of the differentiation program of neurons on a single-cell level in live animals (Chalfie et al. 1994). We have made use of gfp reporter gene technology to dissect the genetic program of a single-neuron class-specific fate decision at unprecedented genetic depth. In light of an average allele frequency of 1/2000 haploid genomes (Anderson 1995), our screening through ∼120,000 mutagenized haploid genomes provides a broad catalog of genes that govern ASE cell fate specification. Our screens also recovered informative alleles of previously known genes, such as a conditional cog-1 allele that allowed us to demonstrate that a previously described bistable feedback loop is required to initiate but not to maintain terminal fates. This is an unexpected finding in light of the persistent expression of bistable loop components throughout the life of the animal. The retrieval of hypomorphic alleles of essential genes further illustrates the advantage of genetic screens over genomewide knockout approaches that would not have revealed functions of several genes due to early pleiotropies. Apart from the previously known 10 loci that control ASE specification, we have identified 16 novel loci and estimate that another 12 loci remain to be identified. We expect that the molecular characterization of all these loci will continue to provide insights into the architecture of gene regulatory networks that induce and maintain neuronal cell fate decisions.
We thank Q. Chen for expert injection assistance; various high school, undergraduate, and rotation students for help with mapping mutant alleles; B. Brady for extensive mapping and technical assistance; E. Bashllari for laser ablations of unc-37 mutants; L. Shapiro for providing information on the HoxB1 homeodomain structure; B. Tursun for scoring ot94; and Hobert lab members, I. Greenwald and S. Clark, for comments on the manuscript. This work was funded by grants R01 NS050266-01 and 2R01NS039996-05 from the National Institutes of Health to O.H. and F31 predoctoral grants to D.D. (NS052089-03) and S.S. (NS054540-01). O.H. is an Investigator of the Howard Hughes Medical Institute.
↵1 These authors contributed equally to this work.
Communicating editor: D. I. Greenstein
- Received May 7, 2007.
- Accepted June 4, 2007.
- Copyright © 2007 by the Genetics Society of America