Developmental Cell, Vol.
6, 183–192, February, 2004, Copyright 2004 by Cell Press
The Lateral Signal for LIN-12/Notch in C. elegans
Vulval Development Comprises Redundant
Secreted and Transmembrane DSL Proteins
Ning Chen1 and Iva Greenwald2,*
1
Integrated Program in Cellular, Molecular
and Biophysical Studies
2
Howard Hughes Medical Institute and
Department of Biochemistry and Molecular
Biophysics
Columbia University
College of Physicians and Surgeons
701 West 168th Street, Room 720
New York, New York 10032
Summary
The vulval precursor cells (VPCs) are spatially pat-
terned by a LET-23/EGF receptor-mediated inductive
signal and a LIN-12/Notch-mediated lateral signal. The
lateral signal has eluded identification, so the mecha-
nism by which lateral signaling is activated has not
been known. Here, we computationally identify ten
genes that encode potential ligands for LIN-12, and
show that three of these genes, apx-1, dsl-1, and lag-2,
are functionally redundant components of the lateral
signal. We also show that transcription of all three
genes is initiated or upregulated in VPCs in response
to inductive signaling, suggesting that direct tran-
scriptional control of the lateral signal by the inductive
signal is part of the mechanism by which these cell
signaling events are coordinated. In addition, we show
that DSL-1, which lacks a predicted transmembrane
domain, is a natural secreted ligand and can substitute
for the transmembrane ligand LAG-2 in different func-
tional assays.
Introduction
Vulval precursor cell (VPC) specification in C. elegans
has been a valuable paradigm for elucidating cell signal-
ing pathways. There are six vulval precursor cells (VPCs),
consecutively numbered P3.p–P8.p (see Figure 1). Each
has the potential to adopt one of three fates, called 1⬚,
2⬚, or 3⬚ (Figure 1; reviewed in Greenwald [1997] and
Kornfeld [1997]). In wild-type hermaphrodites, P6.p
adopts the 1⬚ fate, and P5.p and P7.p adopt the 2⬚ fate;
their descendants form the vulva. The three other VPCs
adopt the 3⬚ fate and generate daughters that join the
hypodermal syncytium.
The VPCs express LET-23, a receptor tyrosine kinase,
and LIN-12, a Notch protein. LIN-3, an EGF-like ligand
produced by the anchor cell (AC) of the somatic gonad
(Hill and Sternberg, 1992), activates LET-23 and a canon-
ical Ras-MAP kinase (MAPK) cascade to promote the 1⬚
fate (reviewed in Greenwald [1997] and Kornfeld [1997]).
The LET-23-Ras pathway modulates the activity of sev-
eral different transcription factors or cofactors (Miller et
al., 1993; Singh and Han, 1995; Tuck and Greenwald,
1995; Beitel et al., 1995; Howard and Sundaram, 2002),
*Correspondence: greenwald@cancercenter.columbia.edu
at least some via direct phosphorylation by MAPK (Ja-
cobs et al., 1998; Tan et al., 1998). An important role
of inductive signaling may be to antagonize inhibitory
effects from the hypodermis that promote the 3⬚ fate
(Herman and Hedgecock, 1990).
Laser ablation experiments led to the suggestion that
there is a lateral signal that prevents adjacent VPCs from
adopting the 1⬚ fate (Sternberg, 1988). Genetic evidence
suggested that LIN-12 is the receptor for the lateral
signal, as alleles of lin-12 that cause constitutive activity
cause all six VPCs to adopt the 2⬚ fate, whereas no VPCs
adopt the 2⬚ fate in lin-12 null mutants (Greenwald et al.,
1983). However, the lateral signal itself has been elusive.
In order to understand the nature of the lateral signal-
ing event, it is important to address three outstanding
issues. One issue is whether this event truly involves
a lateral signal, i.e., whether the signal originates as
hypothesized within a VPC (Sternberg, 1988). A second
issue is whether the production of the lateral signal is
controlled by the inductive signal, as is implicit in the
lateral signal hypothesis. A third issue is whether the
lateral signal is regulated at the transcriptional or post-
transcriptional level. The Ras pathway can regulate the
activity of downstream genes by many different mecha-
nisms (e.g., Engelberg et al., 1994). In the case of the
lateral signal, transcriptional control seems plausible,
given that known transcription factors are modulated
by the inductive signaling pathway. However, a possible
mechanism for posttranslational control is suggested by
the finding that LIN-12 protein must be downregulated in
P6.p for lateral signaling to occur (Shaye and Green-
wald, 2002).
Genes encoding the lateral signal have not been iden-
tified despite extensive genetic screens for mutations
that might in principle have identified it (e.g., Ferguson
and Horvitz, 1985; Eisenmann and Kim, 2000). The failure
to identify the lateral signal genetically suggests that it
may be encoded by more than one member of a gene
family or that mutations in the gene for the lateral signal
may cause pleiotropic effects that have precluded their
recovery in screens for vulval mutants.
In this study, we have used a computational approach
to identify genes encoding the lateral signal, predicated
on the assumption that the lateral signal would resemble
characterized ligands for LIN-12/Notch proteins. Most
known ligands for LIN-12/Notch proteins are transmem-
brane proteins of the Delta/Serrate/Lag-2 (DSL) family,
with an amino-terminal DSL motif followed by one or
more EGF motifs. Three ligands with this predicted
structure were previously known in C. elegans: lag-2,
apx-1, and arg-1. lag-2 and apx-1 were identified geneti-
cally and shown to function in several distinct cell fate
decisions (Lambie and Kimble, 1991; Tax et al., 1994;
Mango et al., 1994; Mello et al., 1994), and arg-1 had
been identified by cross-hybridization to a probe from
apx-1 (Mello et al., 1994). Our computational analysis
also identified seven additional potential ligands of
LIN-12/Notch based on the presence of the hallmark
Developmental Cell
184
porter egl-17::gfp or egl-17::cfp-lacZ at the “Pn.px” stage as a marker for a VPC that adopted the 1⬚ fate (Burdine et al., 1998; Yoo et al., in
press). We have also used the presence of the GFP-tagged protein AJM-1::GFP (expressed from jcIs1), a cell junction marker (Koppen et al.,
2001), at the Pn.pxx stage as a marker for VPCs that have adopted vulval fates.
motifs. Surprisingly, while all contain the hallmark mo-
tifs, at least five of them lack predicted transmem-
brane domains.
The available genetic and expression information ar-
gued against roles for lag-2 and apx-1 in VPC specifica-
tion, but here we provide evidence that these two genes
are components of the lateral signal. We also provide
evidence that one of the computationally identified
genes, dsl-1, encodes a secreted ligand and is a compo-
nent of the lateral signal. In addition to defining the
lateral signal, we demonstrate that the transcription of
all three genes in P6.p is controlled by activity of the
LET-23-Ras pathway.
Results
Ten Potential Ligands for LIN-12/Notch
in C. elegans
To identify genes encoding candidate ligands for
LIN-12/Notch, we used the DSL motif of the known li-
gands LAG-2 and APX-1 to search the complete C. ele-
gans genome sequence database, and analyzed the
highest-scoring proteins by SMART (Letunic et al.,
2002). This analysis revealed ten predicted proteins of
the DSL family (Figure 2).
Sequence analysis (see Figure 3A) indicates that three
DSL proteins have highly probable predicted transmem-
brane domains; these were encoded by the three genes
that had been identified previously, lag-2, apx-1, and
arg-1. Two of the proteins identified by our computa-
tional analysis, DSL-2 and DSL-6, may also have trans-
membrane domains, although the potential transmem-
brane domains were assessed as lower probability in
prediction programs. Surprisingly, the five remaining
genes (dsl-1, dsl-3, dsl-4, dsl-5, and dsl-7) encode DSL
proteins that are predicted to lack transmembrane do-
mains and hence are likely to be secreted. We verified
that the predicted ORFs encoding these proteins are
correct by analyzing 5⬘ and 3⬘ RACE products (see Ex-
perimental Procedures). These five proteins also lack
Figure 1. Schematic View of VPC Specifica-
tion and Relevant Markers for Cell Fate
In the L3 stage, six initially equivalent VPCs,
P3.p–P8.p, adopt a 3⬚-3⬚-2⬚-1⬚-2⬚-3⬚ pattern
of cell fates, the net outcome of inductive,
lateral, and inhibitory signaling. The inductive
signal LIN-3, represented by arrows, appears
to form a spatial gradient (Sternberg and Hor-
vitz, 1986; Yoo et al., in press). The lateral
signal is the subject of this study. It has been
hypothesized to originate in the presumptive
1⬚ cell (Sternberg, 1988) and is represented
by open arrows. An inhibitory signal, believed
to originate in the hyp7 hypodermal syncy-
tium (Herman and Hedgecock, 1990), is not
shown. The descendants of cells that adopt
the 1⬚ and 2⬚ fates form the vulva; the progeny
of cells that adopt the 3⬚ fate fuse with the
hypodermal syncytium. The three cell fates
can be distinguished by cell lineage and the
pattern of marker gene expression. We have
used expression of the transcriptional re-
predicted GPI (glycosyl-phosphatidyl-inositol)-linkages
(Eisenhaber et al., 2003). Analysis of available C. brigg-
sae orthologs supports our inferences about the C. ele-
gans proteins (see supplemental data [http://www.
developmentalcell.com/cgi/content/full/6/2/183/DC1]).
The SMART database (http://smart.embl-heidelberg.de)
(Letunic et al., 2002) includes potential secreted ligands
in several different vertebrate and insect species; these
predictions must be considered provisional until the
transcripts are confirmed by RACE analysis. However,
the sequence of the two spider DSL proteins included in
this database had been confirmed by RACE (Stollewerk,
2002). When we analyzed their sequences, we found
that they, too, appear to lack transmembrane domains
(Figure 3B), suggesting that naturally secreted DSL pro-
teins exist in other phyla.
Functional Identification of Components
of the Lateral Signal
We assessed lateral signal function by reducing the ac-
tivity of individual genes using available mutations or
using “feeding RNAi” in order to bypass embryonic re-
quirements (e.g., lag-2 and apx-1 are required for viabil-
ity). We assessed VPC fates by looking at the expression
of egl-17::gfp, a marker of the 1⬚ fate (Burdine et al.,
1998). Reduced lateral signaling may be manifested as
expression of egl-17::gfp in adjacent VPCs. We saw no
effect of reducing dsl gene activity in a wild-type or
rrf-3 (Simmer et al., 2002) background; however, lin-
12(RNAi) also has little effect on lateral signaling in a
wild-type background, suggesting that it may be difficult
to inhibit lateral signaling sufficiently by RNAi to see a
strong phenotype (data not shown).
As a potential way to sensitize the system, we at-
tempted to reduce lateral signaling in genetic back-
grounds in which inductive signaling is constitutively
activated in all six VPCs. In such backgrounds, it ap-
pears that the activation of lateral signaling superim-
posed upon the constitutive Ras pathway activity results
in an alternating pattern of 1⬚ and 2⬚ fates (Sternberg,
Lateral Signal and Vulval Development
185
1988). We first used a lin-15(n309) background, the same
background that was used to infer the existence of a
lateral signal (Sternberg, 1988). In a lin-15(n309) back-
ground (unlike in wild-type), depletion of lin-12 activity
results in a highly penetrant lateral signaling defect (Fig-
ure 4A). When we performed RNAi for each candidate
gene in the lin-15(n309) background, we observed a
significant increase in the proportion of pairs of adjacent
VPCs expressing egl-17::gfp when the activity of apx-1,
dsl-1, or lag-2 was reduced (Figure 4A). To test whether
this observation is specific to lin-15, we performed RNAi
in a kuIs14 background (Sundaram et al., 1996), in which
constitutively activated LET-60 Ras is expressed from
the transgene, and obtained similar results (Figure 4B).
We tried various combinations of existing mutations
and RNAi to examine the effect of depleting the activity
of all three putative components of the lateral signal
(see Experimental Procedures). The most informative
experiments involved the dsl-1(ok810) null allele (see
Experimental Procedures). dsl-1(ok810) hermaphrodites
are viable and fertile (data not shown), and display a
4% penetrant defect in lateral signaling (Figure 4D). In
a lin-15(⫹) background, dsl-1(ok810) combined with
feeding RNAi for apx-1 ⫹ lag-2 caused about 13% of
hermaphrodites to display a defect in lateral signaling,
which is comparable to the effect of lin-12 RNAi in the
Figure 2. Predicted C. elegans DSL Proteins
(A) Schematic domain structure of predicted
C. elegans DSL proteins. The diagnostic DSL
and EGF motifs are shown, as are predicted
signal sequence and transmembrane do-
mains.
(B) Phylogenetic relationship among the ten
C. elegans DSL proteins, based on CLUS-
TALW analysis of DSL domains. The first EGF
domain from Drosophila Notch and other EGF
domains from other LIN-12/Notch proteins
(not shown) were included along with the DSL
domains of the proteins shown to create the
alignment. It is curious that all of the putative
secreted DSL proteins, except for DSL-4, ap-
pear to be more closely related to each other
than to the canonical DSL proteins. This re-
semblance extends to the overall domain
structure, as these six proteins all have only
a single EGF domain. C. briggsae orthologs
of LAG-2, APX-1, DSL-1, DSL-3, and DSL-7
have been identified (see supplemental
figures).
same background (Figure 4D). When we combined
apx-1 ⫹ lag-2 RNAi with dsl-1(ok810) in a lin-15 mutant
background, over 90% of hermaphrodites contained at
least one pair of adjacent VPCs that adopted the 1⬚ fate,
and almost 70% of all pairs of adjacent VPCs adopted
the 1⬚ fate (Figure 4C).
These genetic data indicate that apx-1, dsl-1, and
lag-2 are functionally redundant components of the lat-
eral signal. Given the limitations of feeding RNAi, the
data suggest that apx-1, dsl-1, and lag-2 are the main
components of the lateral signal, although we cannot
rule out the possibility that there is another component
that we have not yet identified. The identification of
these three genes has enabled us to explore the cellular
origin of the lateral signal and the level at which its
production is regulated.
Transcriptional Reporters and Analysis
of VPC Expression
We made transcriptional reporters to assess the expres-
sion of apx-1, lag-2, and dsl-1 (see Experimental Proce-
dures and Figure 5A). In the early L3 stage, as assessed
by the extent of gonad extension, apx-1 and dsl-1 are
not expressed. Later, when the gonad is more extended
and beginning to reflex, apx-1 and dsl-1 begin to be
expressed in P6.p (Figure 5C), the cell in which LET-23
Developmental Cell
186
activity is maximally activated by the inductive signal.
Expression of the transcriptional reporter for lag-2 is
evident in all six VPCs in the early L3 stage (Figure 5B),
but then appears to be much more highly expressed in
P6.p while the expression in the other VPCs diminishes
or becomes undetectable (Figure 5C). Expression of all
three reporters continues in P6.p descendants (Figure
5C). T. Vellai et al. have seen a similar expression pattern
using an independent transcriptional reporter for lag-2
(personal communication). The restricted expression or
upregulation of apx-1, dsl-1, and lag-2 in P6.p strongly
supports the hypothesis that the lateral signal originates
in the presumptive 1⬚ VPC and directly activates LIN-12
in the neighboring VPCs. We note that transcriptional
reporters for other dsl genes are not expressed in the
VPCs, further supporting the inference from RNAi exper-
iments that they are not part of the lateral signal (Experi-
mental Procedures and data not shown).
apx-1, dsl-1, and lag-2 Expression Is Responsive
to Inductive Signaling and Mediator
Complex Activity
The expression patterns observed for apx-1, dsl-1, and
lag-2 suggest that their transcription is initiated or
upregulated in response to the inductive signal. Indeed,
loss of lin-3 activity prevents expression of these re-
porter genes (Figure 6A), and in a lin-15 mutant, the
reporter genes display a striking pattern concordant
with the alternating 1⬚ (reporter on) and 2⬚ (reporter off)
fates (Figure 6A). However, in lin-3 and lin-15 mutants,
the VPCs also undergo complete cell fate transforma-
tions, making interpretation of these observations
somewhat problematic.
Figure 3. Some DSL Proteins Lack Predicted
Transmembrane Domains
(A) Assessment of potential transmembrane
domains for C. elegans DSL proteins. We
used several different programs to analyze
the DSL protein sequences. The plots shown
here were generated using TMAP (Persson
and Argos, 1997), which gives an indication
of the hydrophobicity throughout the protein
sequence. Peaks represent the predicted sig-
nal sequence (S) and potential transmem-
brane domains, both indicated in black. The
probability values calculated by TMHMM
(Krogh et al., 2001) are indicated for potential
transmembrane domains; 1.0 is the highest
value, and where probability values are not
shown, no potential membrane spanning do-
main was found by this program. We note
that TMHMM has been evaluated as the best-
performing program for this purpose of many
available (Moller et al., 2001; Melen et al.,
2003). The presence or absence of predicted
transmembrane domains is conserved in C.
briggsae (see supplemental figures).
(B) Two spider DSL proteins also lack pre-
dicted transmembrane domains.
Another mutant background, sur-2, is more informa-
tive. sur-2 encodes a component of the Mediator tran-
scription complex, and acts downstream of Ras (Singh
and Han, 1995; Boyer et al., 1999). In sur-2 mutants,
P6.p usually adopts a recognizable 1⬚ fate, but P5.p and
P7.p usually adopt the 3⬚ fate, suggesting that lateral
signaling is lost (Singh and Han, 1995; see also Figure
6B). Expression of transcriptional reporters for all three
genes is dramatically reduced in a sur-2 mutant back-
ground (Figure 6B), suggesting that they are regulated
by sur-2 as targets of the inductive signaling pathway.
We note that there is substantial but variable expres-
sion of all three ligand genes in ventral cord motor neu-
rons in the vicinity of the VPCs (see Figure 5). The func-
tional relevance of this neuronal expression of DSL
ligands is unknown. However, the nerve cord expression
of the lag-2 reporter is similar in wild-type (27/41 her-
maphrodites display detectable nerve cord staining),
lin-3 (31/54), or sur-2 (31/53) backgrounds, suggesting
that it is not likely to be relevant to VPC specification.
DSL-1 Is Secreted and Can Function as an
Activating Ligand in Other Cell Fate Decisions
The predicted structure of DSL-1 as a secreted protein
is unusual for a LIN-12/Notch ligand. In the VPCs, DSL-1
functions in concert with the transmembrane ligands
LAG-2 and APX-1. We therefore asked whether DSL-1
can functionally replace LAG-2. When lag-2(q420ts)
eggs are grown at 25⬚C, larvae arrest development with
a characteristic Lag phenotype; when lag-2(q420) L1
larvae are shifted to 25⬚C, hermaphrodites have 2 anchor
cells, as in lin-12 null mutants, instead of 1 AC, as in wild-
type (Lambie and Kimble, 1991). These highly penetrant
Lateral Signal and Vulval Development
187
defects indicate that lag-2 is the only functionally rele-
vant ligand for LIN-12 during these cell fate decisions.
When we expressed DSL-1 using lag-2 regulatory se-
quences in the background of the temperature-sensitive
lag-2(q420) allele, both larval lethality and the 2 AC de-
fect could be rescued (Figure 7A). In addition, expres-
sion of DSL-1 under the control of myo-3 regulatory
sequences, which drive expression in body wall muscles
and the gonadal sheath cells (Okkema et al., 1993),
causes a low-penetrance tumorous germline phenotype
(data not shown), suggesting that GLP-1 has been ec-
topically activated (Berry et al., 1997; Pepper et al.,
2003). These observations suggest that DSL-1 can sub-
stitute for transmembrane ligands and can activate both
LIN-12 and GLP-1.
We used the tumorous germline phenotype to develop
a functional assay for DSL ligand secretion from coe-
lomocytes. Each coelomocyte is surrounded by a base-
ment membrane, and the germline is ensheathed in so-
matic gonadal cells, which are enclosed by a basement
Figure 4. Reduction of apx-1, dsl-1, or lag-2
Activity Can Cause Adjacent VPCs to Adopt
the 1⬚ Fate
The ayIs4[egl-17::gfp] marker for the 1⬚ fate
was also present. For each genotype, the
number of hermaphrodites scored at the
Pn.px stage is indicated below the corre-
sponding bar. % of worms affected ⫽ % of
worms in which there is at least one pair of
adjacent VPCs that both adopted the 1⬚ fate.
% adjacent 1⬚ fate ⫽ % of adjacent pairs in
which both VPCs adopted the 1⬚ fate.
(A and B) lin-15(n309) and let-60(kuIs14)
backgrounds. lin-12(RNAi) results in a highly
penetrant 2 AC phenotype; we do not know
to what extent this 2 AC phenotype may con-
tribute to the high frequency of lateral signal-
ing defects observed. The penetrance of the
2 AC phenotype in lag-2(RNAi) hermaphro-
dites is not complete, and the frequency of
lateral signaling defects observed in her-
maphrodites having only 1 AC is shown.
(C) dsl-1(ok810); lin-15(n309) background.
We performed a Z-test to ascertain whether
the apparent differences are statistically sig-
nificant. As the “% worms affected” is high
even for the control, we focused on differ-
ences for the “% adjacent 1⬚ fate.” All are
significantly different from the control. lag-2
is different from apx-1 ⫹ lag-2 (p ⬍ 0.0001),
and apx-1 is different from apx-1 ⫹ lag-2
(p ⫽ 0.0004).
(D) dsl-1(ok810) background. The jcIs1[ajm-
1::gfp] marker for the vulval fate was also
present. Absence of ajm-1::gfp in P5.px and
P7.px indicates loss of 2⬚ fate. The low pene-
trance loss of 2⬚ fate observed in the control
is due to the dsl-1 allele and not the ajm-
1::gfp marker (data not shown). The Z-test
indicates that both apx-1 ⫹ lag-2 (p ⫽ 0.006)
and lin-12 (p ⫽ 0.002) are significantly differ-
ent from the control and are not different from
each other (p ⫽ 0.865).
membrane (Schedl, 1997). The cell membranes of the
coelomocytes and germline are therefore not in direct
contact, but the germline can take up yolk and perhaps
other proteins from the pseudocoelom. Expression of
DSL-1 or a truncated form of APX-1 that lacks the trans-
membrane domain under the control of a coelomo-
cyte-specific promoter (Loria et al., 2004; see Experi-
mental Procedures) causes a tumorous germline pheno-
type, whereas expression of full-length, transmembrane
APX-1 does not (Figures 7B and 7C). These observations
support the inference based on sequence analysis that
DSL-1 is indeed secreted.
Discussion
Sternberg (1988) proposed that a LIN-12-mediated lat-
eral signal contributes to VPC patterning. Since then,
however, the lateral signal has eluded identification, so
its cellular source and how it is regulated has not been
Developmental Cell
188

Figure 5. Expression of Transcriptional Reporters for apx-1, dsl-1, or lag-2

(A) Schematic view of the apx-1, dsl-1, and lag-2 genomic regions. Dark gray boxes represent exons of apx-1, dsl-1, or lag-2. Open boxes
represent the next gene upstream. Dashed lines represent the extent of the 5⬘ flanking regions that were used to drive the transcriptional reporter.
(B) Expression of lag-2 in VPCs at a time inferred to be prior to inductive signaling (“lag-2early”), based on the size of the worm and development
of the gonad. Here and in (C), arrows indicate P6.p or its descendants, lines indicate other VPCs or their descendants, and n indicates ventral
cord neurons. MH27 was stained with Cy5-conjugated secondary antibody (blue), and lacZ was stained with Cy3-conjugated secondary
antibody (red) (see Experimental Procedures).
(C) Expression of apx-1, dsl-1, or lag-2 is shown at the Pn.p (VPC) stage at a time inferred to be after inductive signaling. Expression in the
descendants of P6.p at the Pn.px and Pn.pxx stages is also shown.

known. We have provided evidence that three DSL pro-
teins, encoded by apx-1, lag-2, and dsl-1, function re-
dundantly as components of the lateral signal. All three
genes are transcribed in P6.p, the presumptive 1⬚ VPC,
consistent with function as a hypothesized lateral signal
between P6.p and the neighboring VPCs. Furthermore,
transcriptional initiation or upregulation requires the ac-
tivity of the transcriptional Mediator complex, which acts
downstream of the LET-23-Ras pathway to promote the
2⬚ fate. These observations suggest that lateral signaling
appears to involve the transcription of apx-1, lag-2, and
dsl-1 as a direct response to activation of LET-23.
Understanding how inductive and lateral signaling are
coordinated is the key to understanding how the precise
spatial pattern of VPC fates is specified. Our finding that
the inductive signaling pathway appears to regulate the
transcription of genes encoding the lateral signal is con-
sistent with the “sequential induction” model for the
patterning events that operate during VPC specification
(see Koga and Ohshima, 1995; Simske and Kim, 1995).
However, a simple, linear transcriptional cascade does
not adequately describe the complex molecular events
underlying VPC patterning. The inductive signal appears
to be spatially graded (Sternberg and Horvitz, 1986; Yoo
et al., in press), and its effects become limited to P6.p
in part through the transcription of multiple negative
regulators of the LET-23-Ras pathway in P5.p and P7.p
(Berset et al., 2001; Yoo et al., in press). In addition,
cell-cycle-dependent gating of the inductive and lateral
signaling pathways appears to contribute to a temporal
sequence of commitment, to the 1⬚ fate in G1 and to
the 2⬚ fate after S phase (Ambros, 1999). Finally, down-
regulation of LIN-12 protein in response to inductive
signaling in P6.p is necessary for lateral signaling to
occur (Shaye and Greenwald, 2002).
The Mediator complex appears to be dispensable for
some aspects of the 1⬚ fate, but is required for lateral
signaling (Singh and Han, 1995; Tuck and Greenwald,
1995). The requirement for the Mediator complex for
lateral signal gene transcription suggests that the im-
pairment of lateral signaling in Mediator complex mu-
tants may be a direct consequence of failure to express
components of the lateral signal. However, while tran-
scription of the lateral signal must be a prerequisite for
lateral signaling to occur, it is not sufficient, as removal
of LIN-12 protein from the presumptive 1⬚ VPC is also
Lateral Signal and Vulval Development
189
Figure 6. Transcriptional Regulation of apx-1 by Inductive Signaling
(A) Expression of apx-1::lacZ depends on inductive signaling. arIs89
is an integrated transgene carrying the apx-1::lacZ transcriptional
reporter (see Experimental Procedures). Relevant Pn.px cells are
indicated by arrows. MH27 was stained with Cy5-conjugated sec-
ondary antibody (blue), and lacZ was stained with Cy3-conjugated
secondary antibody (red). apx-1 reporter expression is not seen in
lin-3(e1417) hermaphrodites; ectopic expression is seen in alternat-
ing VPCs in lin-15(n309) hermaphrodites. Similar results were also
obtained for transcriptional reporters for dsl-1 and lag-2 (data not
shown).
(B) Expression of transcriptional reporters for apx-1, dsl-1, and lag-
2 is specifically reduced in a sur-2 null mutant background. Note
that most sur-2(0) hermaphrodites express the marker egl-17::cfp-
lacZ, indicating that by this criterion, they have adopted the 1⬚ fate.
For each genotype, the number of hermaphrodites scored at the
Pn.px stage is indicated above the corresponding bar.
necessary for lateral signal activity (Shaye and Green-
wald, 2002). The Mediator complex also controls down-
regulation of LIN-12 in P6.p, presumably through the
transcription of a factor that promotes LIN-12 endocyto-
sis and degradation (Shaye and Greenwald, 2002). Po-
tential interplay between the regulation of lateral signal
gene expression and LIN-12 downregulation will be an
interesting possibility for future exploration.
In contrast to canonical transmembrane LIN-12/Notch
ligands, DSL-1 and four other C. elegans DSL proteins
lack predicted transmembrane domains and instead ap-
pear likely to be secreted or peripheral membrane pro-
teins. We have also demonstrated that DSL-1 can acti-
vate LIN-12/Notch at a distance from its cellular source,
indicating that it is secreted. When we analyzed avail-
able DSL proteins of confirmed sequence by the same
criteria, we found that two DSL proteins recently identi-
fied in spiders (Stollewerk, 2002) also lack predicted
transmembrane domains. These spider DSL proteins ap-
pear to be required for Notch-mediated signaling (Stolle-
werk et al., 2003), supporting the view that secreted DSL
Figure 7. DSL-1 Can Functionally Substitute for LAG-2 and Is Se-
creted
(A) Rescue of lag-2(ts) by DSL-1. Experiments were conducted as
described in Experimental Procedures. For each genotype, the num-
ber of animals scored for each trait is indicated above the corre-
sponding bar. No transgene ⫽ lag-2(q420). Control ⫽ lag-2(q420);
arEx486[cdh-3::gfp]. lag-2p::DSL-1 ⫽ lag-2(q420); arEx489[cdh-
3::gfp ⫹ lag-2p::DSL-1].
(B and C) Ectopic activation of GLP-1 in the germline by DSL-1
expressed in coelomocytes provides evidence that DSL-1 is a se-
creted protein. (B) Photomicrographs of DAPI-stained gonads. The
germline of the unc-122p::apx-1(⫹) hermaphrodite appears wild-
type, and that of unc-122p::dsl-1(⫹) displays a tumorous phenotype
indicative of GLP-1 hyperactivation. (C) Quantitation of the tumorous
phenotype associated with expression of secreted ligands in coe-
lomocytes. Each bar represents an independent transgenic line.
Because sterility caused by expression of the extracellular domain
of APX-1 [APX-1(extra)] is so highly penetrant, lines could not be
maintained, so the F2 progeny of individual F1 transgenic hermaph-
rodites were scored.
proteins can function as ligands for LIN-12/Notch in
other phyla. Perhaps additional sequence analysis will
identify secreted ligands in vertebrates.
In addition to functioning as a component of the lateral
signal, DSL-1 can substitute for the transmembrane li-
gand LAG-2 in different cell fate decisions. Whether the
putative secreted ligands must undergo endocytosis for
full activity, as transmembrane ligands apparently do
Developmental Cell
190
(Parks et al., 2000; see Le Borgne and Schweisguth,
2003), is not known. Perhaps secreted ligands have spe-
cial modifications that promote oligomerization or asso-
ciation with membranes, bypassing the need for a trans-
membrane domain or endocytosis. In this context,
however, we note that engineered secreted forms of
APX-1 and LAG-2 are functional (Fitzgerald and Green-
wald, 1995; this study), suggesting that natural secreted
ligands may not need any special carboxy-terminal
modifications to function.
In terms of VPC specification, the cell biology of lateral
signaling offers a rationale for why a secreted or periph-
eral membrane protein ligand might be a component of
the lateral signal. The VPCs are polarized epithelial cells:
they have an apical region and a basolateral domain,
separated by adherens junctions (Kaech et al., 1998).
As the apical regions of adjacent VPCs appear to be in
contact only in the vicinity of the cell junctions (Kaech
et al., 1998), and LIN-12 is distributed over the whole
apical surface (e.g., Shaye and Greenwald, 2002), a
transmembrane ligand on the surface of one VPC might
have access to LIN-12 on the apical surface of its neigh-
bor only in a relatively limited area. A ligand that can
diffuse may be available to activate LIN-12 over a greater
region of the apical domain, affording one solution to
such a topographical problem.
Experimental Procedures
Confirmation of the Predicted dsl Coding Regions
The 5⬘ and 3⬘ ends of the predicted dsl messages were determined
by comparing the predicted ends (from Wormbase [http://www.
wormbase.org] and Intronerator [http://www.cse.ucsc.edu/ⵑkent/
intronerator/splidi.html]) and the sequences obtained by PCR ampli-
fication from three cDNA libraries and RACE (rapid amplification of
cDNA ends) products. Details of our analysis are available upon
request. The information in Wormbase now reflects our analysis.
We verified 5⬘ ends that are consistent with the Wormbase predic-
tions for all dsl genes except for dsl-5 and dsl-6. For dsl-5, we were
unable to verify the 5⬘ end; we have used the Wormbase prediction
as the basis for our analysis, as none of the 5⬘ RACE products and
the PCR fragments we obtained from the cDNA libraries for dsl-5
contained a spliced leader sequence. The 5⬘ end of the longest
clone was 77 bp downstream of the predicted start codon and there
is no EST available for dsl-5. For dsl-6, the 5⬘ RACE products showed
a predicted start codon that was consistent with Intronerator but
not Wormbase. The start codon is 94 bp downstream of the pre-
dicted ATG (Wormbase), resulting in protein that is 31 amino acids
shorter but contains a predicted signal sequence.
The 3⬘ ends are consistent with the Wormbase predictions for
all dsl genes except for dsl-7, which matches to the prediction
from Intronerator.
Genetics and RNAi
All RNAi data shown are the sum of at least two independent trials;
in all cases, results were consistent from trial to trial.
The lin-3(e1417) and lin-15(n309) alleles (Ferguson and Horvitz,
1985) and sur-2(ku9) (Singh and Han, 1995) were used to manipulate
the activity of the inductive signaling pathway. Transgenes used
as cell fate markers were: ayIs4[egl-17::gfp] (Burdine et al., 1998),
jcIs1[ajm-1::gfp] (Koppen et al., 2001), and arIs92[egl-17::cfp-lacZ]
(Yoo et al., in press). Routine manipulations were generally per-
formed at 20⬚C, but strains containing lacZ reporters were grown
at 25⬚C prior to fixation for antibody staining.
Null alleles of lag-2 are lethal (Lambie and Kimble, 1991).
lag-2(q420ts) behaves as a strong loss-of-function mutation at 25ⴗC,
causing highly penetrant Lag and 2 AC phenotypes (Lambie and
Kimble, 1991; Fitzgerald and Greenwald, 1995). Rescue of these
lag-2 (q420ts) defects is therefore an effective assay for ligand ac-
tivity.
All available alleles of apx-1 are not null; they cause maternal-
effect lethality due to loss of a glp-1 maternal function (Mango et
al., 1994; Mello et al., 1994). We used apx-1(zu183) (Mello et al.,
1994) for genetic studies, but as this allele, and presumably others
with the same phenotype, affects translation of maternal mRNA, it
may not offer a sensitized background for a study of the function
of apx-1 in vulval development.
The Caenorhabditis elegans Knockout Consortium (http://
eleganskoconsortium.omrf.org/) generated dsl-1(ok810). We se-
quenced dsl-1(ok810) and found that it contains a deletion starting
at 1038 bp 5⬘ of the start codon and removing the entire coding
region, ending at 364 bp 3⬘ of the stop codon.
RNAi was performed by “feeding” (Timmons and Fire, 1998), using
a cDNA fragment for each dsl gene subcloned into the feeding
vector pPD129.36 and transformed into bacterial strain HT115. Syn-
chronized L1 larvae were placed on a lawn of such bacteria and
allowed to grow for 19 hr at 25⬚C, and egl-17::gfp or ajm-1::gfp
expression was analyzed with a Zeiss Axioplan 2 microscope.
In addition to the experiments described in the Results, we per-
formed additional RNAi experiments in the following genetic back-
grounds (data not shown): (1) rrf-3(pk1420); ayIs4 and (2) lin-15
(n309); rrf-3(pk1420); ayIs4. The rrf-3 mutant is hypersensitive to
RNAi (Simmer et al., 2002), but RNAi performed for individual dsl
genes gave similar results as in the rrf-3(⫹) background. (3) lag-2
(q420ts); jcIs1; arIs92 and (4) lag-2(q420ts); lin-15(n309); ayIs4. Her-
maphrodites grown in the presence of bacteria expressing dsRNA
at 15⬚C until the L2 molt so that they would have one AC were
transferred to 25⬚C, and grown to the late L3/early L4 stage. We
performed apx-1 ⫹ dsl-1 RNAi in this background. No difference
was observed as compared to a lag-2(⫹) background. (5) apx-1(zu183)/
nT1; ayIs4. We performed dsl-1 ⫹ lag-2 RNAi in this background.
No obvious loss of 2⬚ fate was observed.
Transcriptional Reporter Genes
Transcriptional reporters were based on gene structures dia-
grammed in Figure 5A. The 5⬘ flanking regions of apx-1 (9 kb up-
stream), lag-2 (6.2 kb upstream), and dsl-1 (8 kb upstream) were
inserted into vector pPD16.51. The 5⬘ flanking regions were de-
signed to extend to the next upstream gene, based on the Wormbase
predictions and RACE (see above). The data shown in the Figures are
for transgenes formed from these constructs, listed below. Previous
reporter genes for lag-2 (Henderson et al., 1994; Fitzgerald and
Greenwald, 1995) that do not show upregulation in P6.p used only
3.4 kb of 5⬘ flanking region.
We made other transcriptional reporters for each of the dsl genes
except for lag-2 and apx-1 using a design originally developed for
expression analysis of lin-12 (see Wilkinson et al., 1994). lacZ was
inserted at the position of the predicted ATG into a large genomic
region encompassing the entire dsl coding region as well as the 5⬘
and 3⬘ flanking regions, extending to the neighboring predicted
genes. Such constructs only express lacZ in a smg-1 background.
None of the reporters for the other dsl genes were expressed in the
VPCs (data not shown). For apx-1, the very large genomic region
(26 kb) precluded the use of such a construct.
Transgenic Lines for Expression Studies
The transcriptional reporters described above were injected at 100
␮g/ml, together with plasmids pMH86 [dpy-20(⫹)] (50 ␮g/ml; Han
and Sternberg, 1991) and pCW2.1 [ceh-22::gfp] (20 ␮g/ml; Okkema
et al., 1997). Mixes containing the apx-1, dsl-1, or lag-2 reporters
were injected into dpy-20(e1282) animals. Mixes containing the
smg-1-dependent reporters were injected into unc-54 smg-1; dpy-
20(e1282) animals. F1 progeny that expressed GFP in the pharynx,
indicating the presence of ceh-22::gfp, were picked individually to
establish independent extrachromosomal lines.
Extrachromosomal arrays carrying apx-1::lacZ or dsl-1::lacZ con-
structs were integrated using standard methods (Mello and Fire,
1995). All integrated lines were backcrossed five times to N2 prior
to further analysis. We analyzed two independent integrants for
apx-1::lacZ (arIs88, derived from arEx479, and arIs89, derived from
arEx480), two independent integrants for dsl-1::lacZ (arIs94, derived
Lateral Signal and Vulval Development
191
from arEx481, and arIs95, derived from arEx482), and eight extra-
chromosomal arrays for lag-2::lacZ (arEx471-arEx478). Different
transgenes show the same expression pattern for each individual
gene.
Constructs and Transgenic Lines for DSL-1 Rescue and
Secretion Studies
pNC21.1 [lag-2p::DSL-1] contains a DSL-1 cDNA extending from
the start to the stop codon in the lag-2 expression vector p226S
(described in Fitzgerald and Greenwald, 1995) at the SpeI site. The
expression of DSL-1 is under the control of the lag-2 regulatory
region previously shown to be sufficient to rescue the viability and
2 AC defects of lag-2 mutants (Tax et al., 1994). This regulatory
region includes 3.4 kb of 5⬘ flanking region. lag-2(q420ts); smg-1
animals were coinjected with cdh-3::gfp (25 ␮g/ml), pBluescript (75
␮g/ml), and without or with pNC21.1 (2 ␮g/ml). arEx486-arEx487
and arEx488-arEx491, respectively, were obtained at 15⬚C. cdh-
3::gfp expression was used to identify and follow transgenic worms,
and also served as an AC marker (Pettitt et al., 1996).
To assess rescue, synchronized eggs or L1 larvae were shifted
to 25⬚C. Worms with one or more AC expressing GFP were analyzed
with a Zeiss Axioplan 2 microscope. arEx489 rescued both the viabil-
ity and 2 AC phenotypes; arEx490 rescued the viability but not 2
AC phenotype (data not shown).
pNC31 [unc-122p::dsl-1(⫹)] contains the full-length DSL-1 cDNA
and a fragment of the unc-122 promoter that drives coelomocyte-
specific expression (Loria et al., 2004). pNC32 [unc-122p::apx-1(ex-
tra)] and pNC33 [unc-122p::apx-1(⫹)] contain either the “APX-1(ex-
tra)” cDNA, a truncated APX-1 cDNA that lacks the TM domain
(Fitzgerald and Greenwald 1995), or the full-length APX-1(⫹) cDNA
and the unc-122 promoter. pha-1(e2123) animals were coinjected
with ceh-22::gfp (20 ␮g/ml), pBX [pha-1(⫹)] (50 ␮g/ml), and pNC31,
pNC32, or pNC33 (each at 100 ␮g/ml), and grown at 15⬚C for 4
days. Three extrachromosomal arrays were obtained for each DNA
mixture by switching the plates to 25⬚C to select for pha-1 rescue
(Granato et al., 1994). ceh-22::gfp expression was also used here
to confirm the presence of the arrays.
Worms with GFP expression in the pharynx were picked and
stained with DAPI, and analyzed for tumorous germline phenotype
with a Zeiss Axioplan 2 microscope.
Immunofluorescence
Synchronized L1 larvae were grown for 19 hr at 25⬚C and fixed
according to the modified Finney-Ruvkun procedure (Bettinger et
al., 1996). Fixed animals were first incubated with monoclonal mouse
anti-LacZ (Promega) (1:500). After being washed with buffer, these
animals were incubated with Cy3-conjugated goat anti-mouse sec-
ondary antibody (1:300). In the same fashion, these animals were
further incubated with polyclonal rabbit anti-GFP (Molecular Probes)
(1:300), and monoclonal antibody MH27 against AJM-1 (Priess and
Hirsh, 1986) (1:1000), and then with Cy2-conjugated goat anti-rabbit
(1:300) and Cy5-conjugated goat anti-mouse (1:300) secondary anti-
bodies. All conjugated secondary antibodies are from Jackson Im-
munoresearch. DAPI (Sigma) was added to the last wash at a final
concentration of 0.5 ␮g/ml. Stained worms were kept in SlowFade
reagent (Molecular Probes), and viewed with a Nikon Eclipse E800
microscope. Photomicrographs were acquired with a Bio-Rad MRC
1024ES laser scanning confocal attachment.
Acknowledgments
We are indebted to Daniel Shaye for generous help, advice, and
insight throughout this project. We are grateful to Xinlan Zhou and
Richard Ruiz for excellent technical assistance, Oliver Hobert for
advice about sequence analysis, Hannes Buelow for advice about
statistical analysis, Paula Loria and Oliver Hobert for the coelomo-
cyte-specific promoter, Tibor Vellai and Fritz Mueller for sharing
results prior to publication, and Bob Barstead, Gary Moulder, and
the C. elegans Knockout Consortium for providing the dsl-1(ok810)
allele. We also thank Oliver Hobert, Daniel Shaye, Gary Struhl, and
Natalie de Souza for helpful comments on the manuscript. This
work was supported by NIH CA095389. I.G. is an Investigator of the
Howard Hughes Medical Institute.
Received: December 14, 2003
Revised: December 22, 2003
Accepted: December 30, 2003
Published: February 9, 2004
References
Ambros, V. (1999). Cell cycle-dependent sequencing of cell fate
decisions in Caenorhabditis elegans vulva precursor cells. Develop-
ment 126, 1947–1956.
Beitel, G.J., Tuck, S., Greenwald, I., and Horvitz, H.R. (1995). The
Caenorhabditis elegans gene lin-1 encodes an ETS-domain protein
and defines a branch of the vulval induction pathway. Genes Dev.
9, 3149–3162.
Berry, L.W., Westlund, B., and Schedl, T. (1997). Germ-line tumor
formation caused by activation of glp-1, a Caenorhabditis elegans
member of the Notch family of receptors. Development 124,
925–936.
Berset, T., Hoier, E.F., Battu, G., Canevascini, S., and Hajnal, A.
(2001). Notch inhibition of RAS signaling through MAP kinase phos-
phatase LIP-1 during C. elegans vulval development. Science
291, 1055–1058.
Bettinger, J.C., Lee, K., and Rougvie, A.E. (1996). Stage-specific
accumulation of the terminal differentiation factor LIN-29 during
Caenorhabditis elegans development. Development 122, 2517–
2527.
Boyer, T.G., Martin, M.E., Lees, E., Ricciardi, R.P., and Berk, A.J.
(1999). Mammalian Srb/Mediator complex is targeted by adenovirus
E1A protein. Nature 399, 276–279.
Burdine, R.D., Branda, C.S., and Stern, M.J. (1998). EGL-17(FGF)
expression coordinates the attraction of the migrating sex myo-
blasts with vulval induction in C. elegans. Development 125, 1083–
1093.
Eisenhaber, F., Eisenhaber, B., Kubina, W., Maurer-Stroh, S., Neu-
berger, G., Schneider, G., and Wildpaner, M. (2003). Prediction of
lipid posttranslational modifications and localization signals from
protein sequences: big-Pi, NMT and PTS1. Nucleic Acids Res. 31,
3631–3634.
Eisenmann, D.M., and Kim, S.K. (2000). Protruding vulva mutants
identify novel loci and Wnt signaling factors that function during
Caenorhabditis elegans vulva development. Genetics 156, 1097–
1116.
Engelberg, D., Klein, C., Martinetto, H., Struhl, K., and Karin, M.
(1994). The UV response involving the Ras signaling pathway and
AP-1 transcription factors is conserved between yeast and mam-
mals. Cell 77, 381–390.
Ferguson, E.L., and Horvitz, H.R. (1985). Identification and charac-
terization of 22 genes that affect the vulval cell lineages of the
nematode Caenorhabditis elegans. Genetics 110, 17–72.
Fitzgerald, K., and Greenwald, I. (1995). Interchangeability of Caeno-
rhabditis elegans DSL proteins and intrinsic signalling activity of
their extracellular domains in vivo. Development 121, 4275–4282.
Granato, M., Schnabel, H., and Schnabel, R. (1994). Genesis of an
organ: molecular analysis of the pha-1 gene. Development 120,
3005–3017.
Greenwald, I. (1997). Development of the vulva. In C. elegans II, D.L.
Riddle, T. Blumenthal, B.J. Meyer, and J.R. Priess, eds. (Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory Press), pp. 519–542.
Greenwald, I.S., Sternberg, P.W., and Horvitz, H.R. (1983). The lin-
12 locus specifies cell fates in Caenorhabditis elegans. Cell 34,
435–444.
Han, M., and Sternberg, P.W. (1991). Analysis of dominant-negative
mutations of the Caenorhabditis elegans let-60 ras gene. Genes
Dev. 5, 2188–2198.
Henderson, S.T., Gao, D., Lambie, E.J., and Kimble, J. (1994). lag-2
may encode a signaling ligand for the GLP-1 and LIN-12 receptors
of C. elegans. Development 120, 2913–2924.
Herman, R.K., and Hedgecock, E.M. (1990). Limitation of the size
Developmental Cell
192
of the vulval primordium of Caenorhabditis elegans by lin-15 expres-
sion in surrounding hypodermis. Nature 348, 169–171.
Hill, R.J., and Sternberg, P.W. (1992). The gene lin-3 encodes an
inductive signal for vulval development in C. elegans. Nature 358,
470–476.
Howard, R.M., and Sundaram, M.V. (2002). C. elegans EOR-1/PLZF
and EOR-2 positively regulate Ras and Wnt signaling and function
redundantly with LIN-25 and the SUR-2 Mediator component. Genes
Dev. 16, 1815–1827.
Jacobs, D., Beitel, G.J., Clark, S.G., Horvitz, H.R., and Kornfeld, K.
(1998). Gain-of-function mutations in the Caenorhabditis elegans
lin-1 ETS gene identify a C-terminal regulatory domain phosphory-
lated by ERK MAP kinase. Genetics 149, 1809–1822.
Kaech, S.M., Whitfield, C.W., and Kim, S.K. (1998). The LIN-2/LIN-7/
LIN-10 complex mediates basolateral membrane localization of the
C. elegans EGF receptor LET-23 in vulval epithelial cells. Cell 94,
761–771.
Koga, M., and Ohshima, Y. (1995). Mosaic analysis of the let-23 gene
function in vulval induction of Caenorhabditis elegans. Development
121, 2655–2666.
Koppen, M., Simske, J.S., Sims, P.A., Firestein, B.L., Hall, D.H.,
Radice, A.D., Rongo, C., and Hardin, J.D. (2001). Cooperative regula-
tion of AJM-1 controls junctional integrity in Caenorhabditis elegans
epithelia. Nat. Cell Biol. 3, 983–991.
Kornfeld, K. (1997). Vulval development in Caenorhabditis elegans.
Trends Genet. 13, 55–61.
Krogh, A., Larsson, B., von Heijne, G., and Sonnhammer, E.L. (2001).
Predicting transmembrane protein topology with a hidden Markov
model: application to complete genomes. J. Mol. Biol. 305, 567–580.
Lambie, E.J., and Kimble, J. (1991). Two homologous regulatory
genes, lin-12 and glp-1, have overlapping functions. Development
112, 231–240.
Le Borgne, R., and Schweisguth, F. (2003). Notch signaling: endocy-
tosis makes delta signal better. Curr. Biol. 13, R273–R275.
Letunic, I., Goodstadt, L., Dickens, N.J., Doerks, T., Schultz, J., Mott,
R., Ciccarelli, F., Copley, R.R., Ponting, C.P., and Bork, P. (2002).
Recent improvements to the SMART domain-based sequence anno-
tation resource. Nucleic Acids Res. 30, 242–244.
Loria, P.M., Hodgkin, J., and Hobert, O. (2004). A conserved postsyn-
aptic transmembrane protein affecting neuromuscular signaling of
C. elegans. J. Neurosci., in press.
Mango, S.E., Thorpe, C.J., Martin, P.R., Chamberlain, S.H., and Bow-
erman, B. (1994). Two maternal genes, apx-1 and pie-1, are required
to distinguish the fates of equivalent blastomeres in the early Caeno-
rhabditis elegans embryo. Development 120, 2305–2315.
Melen, K., Krogh, A., and von Heijne, G. (2003). Reliability measures
for membrane protein topology prediction algorithms. J. Mol. Biol.
327, 735–744.
Mello, C., and Fire, A. (1995). DNA Transformation. In Caenorhabditis
elegans: Modern Biological Analysis of an Organism, H.F. Epstein,
and D.C. Shakes, eds. (San Diego: Academic Press), pp. 451–482.
Mello, C.C., Draper, B.W., and Priess, J.R. (1994). The maternal
genes apx-1 and glp-1 and establishment of dorsal-ventral polarity
in the early C. elegans embryo. Cell 77, 95–106.
Miller, L.M., Gallegos, M.E., Morisseau, B.A., and Kim, S.K. (1993).
lin-31, a Caenorhabditis elegans HNF-3/fork head transcription fac-
tor homolog, specifies three alternative cell fates in vulval develop-
ment. Genes Dev. 7, 933–947.
Moller, S., Croning, M.D., and Apweiler, R. (2001). Evaluation of
methods for the prediction of membrane spanning regions. Bioinfor-
matics 17, 646–653.
Okkema, P.G., Harrison, S.W., Plunger, V., Aryana, A., and Fire, A.
(1993). Sequence requirements for myosin gene expression and
regulation in Caenorhabditis elegans. Genetics 135, 385–404.
Okkema, P.G., Ha, E., Haun, C., Chen, W., and Fire, A. (1997). The
Caenorhabditis elegans NK-2 homeobox gene ceh-22 activates pha-
ryngeal muscle gene expression in combination with pha-1 and is
required for normal pharyngeal development. Development 124,
3965–3973.
Parks, A.L., Klueg, K.M., Stout, J.R., and Muskavitch, M.A. (2000).
Ligand endocytosis drives receptor dissociation and activation in
the Notch pathway. Development 127, 1373–1385.
Pepper, A.S., Lo, T.W., Killian, D.J., Hall, D.H., and Hubbard, E.J.
(2003). The establishment of Caenorhabditis elegans germline pat-
tern is controlled by overlapping proximal and distal somatic gonad
signals. Dev. Biol. 259, 336–350.
Persson, B., and Argos, P. (1997). Prediction of membrane protein
topology utilizing multiple sequence alignments. J. Protein Chem.
16, 453–457.
Pettitt, J., Wood, W.B., and Plasterk, R.H.A. (1996). cdh-3, a gene
encoding a member of the adherin superfamily, functions in epithe-
lial cell morphogenesis in Caenorhabditis elegans. Development
122, 4149–4157.
Priess, J.R., and Hirsh, D.I. (1986). Caenorhabditis elegans morpho-
genesis: the role of the cytoskeleton in elongation of the embryo.
Dev. Biol. 117, 156–173.
Schedl, T. (1997). Developmental Genetics of the Germ Line. In C.
elegans II, D.L. Riddle, T. Blumenthal, B.J. Meyer, and J.R. Priess,
eds. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press),
pp. 519–542.
Shaye, D.D., and Greenwald, I. (2002). Endocytosis-mediated down-
regulation of LIN-12/Notch upon Ras activation in Caenorhabditis
elegans. Nature 420, 686–690.
Simmer, F., Tijsterman, M., Parrish, S., Koushika, S.P., Nonet, M.L.,
Fire, A., Ahringer, J., and Plasterk, R.H. (2002). Loss of the putative
RNA-directed RNA polymerase RRF-3 makes C. elegans hypersen-
sitive to RNAi. Curr. Biol. 12, 1317–1319.
Simske, J.S., and Kim, S.K. (1995). Sequential signalling during
Caenorhabditis elegans vulval induction. Nature 375, 142–146.
Singh, N., and Han, M. (1995). sur-2, a novel gene, functions late in
the let-60 ras-mediated signaling pathway during Caenorhabditis
elegans vulval induction. Genes Dev. 9, 2251–2265.
Sternberg, P.W. (1988). Lateral inhibition during vulval induction in
Caenorhabditis elegans. Nature 335, 551–554.
Sternberg, P.W., and Horvitz, H.R. (1986). Pattern formation during
vulval development in C. elegans. Cell 44, 761–772.
Stollewerk, A. (2002). Recruitment of cell groups through Delta/
Notch signalling during spider neurogenesis. Development 129,
5339–5348.
Stollewerk, A., Schoppmeier, M., and Damen, W.G. (2003). Involve-
ment of Notch and Delta genes in spider segmentation. Nature
423, 863–865.
Sundaram, M., Yochem, J., and Han, M. (1996). A Ras-mediated
signal transduction pathway is involved in the control of sex myo-
blast migration in Caenorhabditis elegans. Development 122, 2823–
2833.
Tan, P.B., Lackner, M.R., and Kim, S.K. (1998). MAP kinase signaling
specificity mediated by the LIN-1 Ets/LIN-31 WH transcription factor
complex during C. elegans vulval induction. Cell 93, 569–580.
Tax, F.E., Yeargers, J.J., and Thomas, J.H. (1994). Sequence of
C. elegans lag-2 reveals a cell-signalling domain shared with Delta
and Serrate of Drosophila. Nature 368, 150–154.
Timmons, L., and Fire, A. (1998). Specific interference by ingested
dsRNA. Nature 395, 854.
Tuck, S., and Greenwald, I. (1995). lin-25, a gene required for vulval
induction in Caenorhabditis elegans. Genes Dev. 9, 341–357.
Wilkinson, H.A., Fitzgerald, K., and Greenwald, I. (1994). Reciprocal
changes in expression of the receptor lin-12 and its ligand lag-2 prior
to commitment in a C. elegans cell fate decision. Cell 79, 1187–1198.
Yoo, A.S., Bais, C., and Greenwald, I. (2004). Cross-talk between the
EGFR and LIN-12/Notch pathways in C. elegans vulval development.
Science, in press.