Complexity of Developmental Control: Analysis of Embryonic Cell Lineage Specification in Caenorhabditis elegans Using pes-1 as an Early Marker

Corresponding author: Heinke Schnabel, Institut für Genetik, Technische Universität Braunschweig, Spielmannstr. 7, D-38106 Braunschweig, Germany., schnabel{at}alpha.bio.nat.tu-bs.de (E-mail)

Communicating editor: R. K. HERMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In the early Caenorhabditis elegans embryo five somatic founder cells are born during the first cleavages. The first of these founder cells, named AB, gives rise to 389 of the 558 nuclei present in the hatching larva. Very few genes directly involved in the specification of the AB lineage have been identified so far. Here we describe a screen of a large collection of maternal-effect embryonic lethal mutations for their effect on the early expression of a pes-1::lacZ fusion gene. This fusion gene is expressed in a characteristic pattern in 14 of the 32 AB descendants present shortly after the initiation of gastrulation. Of the 37 mutations in 36 genes suspected to be required specifically during development, 12 alter the expression of the pes-1::lacZ marker construct. The gene expression pattern alterations are of four types: reduction of expression, variable expression, ectopic expression in addition to the normal pattern, and reduction of the normal pattern together with ectopic expression. We estimate that ~100 maternal functions are required to establish the pes-1 expression pattern in the early embryo.

DURING the early embryogenesis of the nematode Caenorhabditis elegans five somatic founder cells, AB, MS, E, C, and D, are formed in a series of asymmetric divisions. AB is the anterior daughter of the zygote P₀. AB contributes 389 of the 558 cells (nuclei) of the hatching larva. The AB-derived cells differentiate primarily into hypodermis and nervous system, but a great variety of specialized cells like pharyngeal muscles, excretory cells, or intestinal valve cells are also formed in a very complex pattern (SULSTON et al. 1983 ). A key problem in embryogenesis is to understand how such a complex pattern of specialized cells is determined. In contrast to the classical notion that nematode development proceeds autonomously, the AB lineage is actually specified by at least five inductions prior to the onset of gastrulation (for review see SCHNABEL 1996 ). These inductions do not specify tissues. Instead eight blastomere identities are specified, each of which is the origin of a complex lineage pattern resulting in the mosaic differentiation of tissues corresponding to the regional requirements of the body plan. Four of these inductions, which are all binary switches, depend on members of the notch/delta signal transduction pathway (glp-1, lin-12, apx-1; for reviews see SCHNABEL 1996 ; SCHNABEL and PRIESS 1997 ). Recent work on the gene lit-1 suggests that not just the early AB-derived blastomeres but most of the AB lineage is specified in consecutive binary switches. After each round of cell division anterior identity is the default and posterior identity is specified for one of the otherwise equivalent daughter cells (KALETTA et al. 1997 ). Thus the worm employs, at least formally, a series of binary switches to specify a very complex lineage pattern.

Genetic screens for genes involved in a specific process such as development of the AB lineage are usually designed to identify only mutants with very specific defects in this process. Pleiotropic genes repeatedly used during development or functioning in multiple processes may not be recognized in such screens. This may convey an unreal simplicity to a developmental process, and key aspects may be missed. The gene lit-1 is a particular example of this general problem. lit-1 is required in most lineages and would most probably be discarded in a screening protocol designed to identify genes required specifically for AB lineage development; all major tissues outside the AB lineage, which could serve as a control for AB specificity, are also affected. Nevertheless the gene has a central function in the patterning of the AB lineage (KALETTA et al. 1997 ).

The apparent simplicity in the specification of a complex lineage through a series of consecutive binary switches contrasts with the complexity of the C. elegans genome with ~16,500 genes, almost an order-of-magnitude more genes than anticipated previously from conventional genetic analysis (BRENNER 1974 ; S. JONES, http://www.sanger.ac.uk/Projects/C_elegans/sequencing.summary.html/). This high number of genes does not indicate a strong constraint on the number of genes available for processes like the specification of the AB lineage studied in this article. The genome projects reveal even larger numbers of genes for other eukaryotes (MIKLOS and RUBIN 1996 ). Central to the advancement of biology is an efficient means for assigning at least tentative functions to large numbers of genes. The complexity of genetic networks specifying individual characters is mostly unknown (LOOMIS and STERNBERG 1995 ). To assess the number of genes involved in a certain process a large number of genes must be tested with as little prejudice as possible for their involvement in that process.

Three advances now permit such an approach for identifying genes required for specification within the AB lineage: (i) the description of a very specific early expression pattern of the gene pes-1 in the AB lineage that permits assessment of the blastomere identities of six of the eight AB descendants present in the 12-cell stage embryo (HOPE 1991 , HOPE 1994 ); (ii) the development of a 4D-microscope system that allows a precise and rapid analysis of the expression of a pes-1::lacZ fusion gene in embryogenesis (SCHNABEL et al. 1997 ); and (iii) the isolation of a large collection of genetically balanced maternal-effect lethal mutants selected for their general requirement during embryogenesis (R. SCHNABEL, H. SCHNABEL, R. FEICHTINGER and T. KALETTA, unpublished results).

Here we describe that one-third, and thus a high proportion, of 36 tested genes that were selected to be involved in embryonic development are required for the proper expression of the marker in the complex AB lineage.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The basic methods of C. elegans culture and handling were as described previously (BRENNER 1974 ; WOOD 1988 ).

Strains and alleles:
The UL1 strain, derived from the wild-type N2 strain (var. Bristol), contains the genetic regulatory elements of pes-1 fused with the bacterial lacZ reporter gene integrated into chromosome V along with the rol-6(su1006) gene, which confers a dominant Rol phenotype (HOPE 1991 , HOPE 1994 ).

The strain UL56 also contains plasmids carrying the pes-1::lacZ fusion gene and the rol-6(su1006) gene, but the DNA is maintained as an extrachromosomal array.

The following genetic markers and balancers were used:

• linkage group II (LGII): dpy-2(e8), unc-4(e120), mnC1[dpy-10(e128) unc-52(e444)];

• LGIII: dpy-18(e499), unc-32(e189), sDf125, qC1[dpy-19 (e1259ts) glp-1(q339)];

• LGIV: him-3(e1147), unc-24(e138);

• LGV: dpy-11(e224); eT1(III; V); nT1(IV; V).

Some strains used in this study were obtained from the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health National Center for Research Resources. A strain carrying sDf125 was obtained from David Baillie.

The mutations used in this study were chosen from a collection of maternal-effect embryonic lethal mutations isolated on LGs II and III (R. SCHNABEL, H. SCHNABEL, R. FEICHTINGER and T. KALETTA, unpublished results). The 39 chosen genes were the following:

• LGII: arm-1(t1110), arm-2(t1214), arm-4(t1135), emb-40(t1108), arm-7(t1160), emb-62(t1180), emb-49(t1104), emb-50(t1181), emb-57(t1189), emb-67(t1218), emb-52 (t1265), emb-48(t1310), emb-46(t1212), emb-47(t1238), emb-53(t1270), emb-60(t1290), emb-61(t1315), mel-11 (t1282), mel-14(t1176), mel-18(t1144), mel-20(t1203), mod-1(t1157), sud-1(t1237);

• LGIII: arm-5(t1478), arm-6(t1677), arm-8(t1505), arm-9(t1635), emb-85(t1483), inx-1(t1460), lit-1(t1534) (KALETTA et al. 1997 ), mar-1(t1696), mel-32(t1597) (VATCHER et al. 1998 ), mod-2(t1447), mod-3(t1628), mod-4(t1701, t1663), mod-5(t1502), mod-6(t1653), mod-7(t1549), mod-8(t1530).

Names of new genes are as follows: emb, embryonic lethal; arm, embryos arrest during morphogenesis (i.e., some morphogenesis occurs); mod, morphogenesis defective (i.e., no morphogenesis occurs); sud, supernumerary cell divisions; inx, intestine excess; mar, embryos arrest with round cells and resemble a bag of marbles.

mel-11(t1282), mel-14(t1176), mel-18(t1144), and mel-20(t1203) are new alleles of maternal-effect lethal genes described by KEMPHUES et al. 1988A .

Mutants in previously characterized genes were used to determine the reliability of pes-1::lacZ expression as a cell identity marker. The following mutant strains were used:

• LGIII: glp-1(e2072 and e2144ts) (PRIESS et al. 1987 ); pie-1(t1682), par-3 (t1591);

• LGIV: skn-1(zu67);

• LGV: apx-1(t2063 and t2176).

The alleles pie-1(t1682), par-3 (t1591), and apx-1(t2063 and t2176) were isolated in the laboratory of R. Schnabel and H. Schnabel. A strain carrying skn-1(zu67) was obtained from B. Bowerman.

Genetics:
Strains carrying maternal-effect lethal mutations (mat) and the pes-1::lacZ fusion gene (called leIs1) from the UL1 strain were constructed in the following way: dpy-2 unc-4/mnC1; him-3; leIs1 for maternal lethal genes on chromosome II and dpy-18 unc-32/qC1; him-3; leIs1 for maternal lethal genes on chromosome III were first constructed by crossing UL1 males with L4 hermaphrodites of the genotypes dpy-2 unc-4/mnC1 or dpy-18 unc-32/qC1, respectively, and then by crossing F₁ progeny with males of the genotypes dpy-2 unc-4/mnC1;him-3 or dpy-18 unc-32/qC1; him-3, respectively. Rol progeny of these crosses were picked to individual plates and those segregating DpyUnc and mnC1 or qC1 progeny were kept. Self-progeny from these worms were picked to obtain a strain homozygous for leIs1 and him-3.

The strains dpy-2 mat/mnC1; him-3; leIs1 and unc-32 mat/qC1; him-3; leIs1 were then generated by crossing dpy-2 unc-4/mnC1; him-3; leIs1 or dpy-18 unc-32/qC1; him-3; leIs1 with males of the genotypes dpy-2 mat/mnC1; him-3 or unc-32 mat/qC1; him-3, respectively, followed by picking individual Rol cross progeny and selecting those that segregated Dpy and mnC1 or Unc and qC1 worms. Strains homozygous for leIs1 were established in the next generation by selecting worms that segregated only Rol progeny.

Construction of strains containing leIs1 and glp-1, pie-1, or par-3, which are all on LGIII, was carried out similarly.

To construct a strain carrying both apx-1 and leIs1, which both lie on LGV, it was necessary to isolate recombinants. This was done by first crossing UL1 hermaphrodites with unc-24/nT1; apx-1 dpy-11/nT1 males. Progeny were picked and F₂ Unc Rol hermaphrodites (genotype unc-24; leIs1 or unc-24; apx-1 leIs1/dpy-11 or unc-24; leIs1/apx-1 dpy-11) were then crossed with unc-24/nT1; dpy-11/nT1 males. Progeny were then picked to isolate hermaphrodites of the genotype unc-24/nT1; leIs1 apx-1/nT1. Construction of skn-1; leIs1 was also carried out in a nT1 background.

A second strain containing a chromosomally integrated pes-1::lacZ fusion gene was generated by UV irradiating L4 larvae of the strain UL56 for 2 min (2 W/m²/sec). Eight hundred F₂ larvae were cloned and one of these was found to stably produce Rol progeny, suggesting that the array had become integrated into a chromosome (strain GE3212).

Analysis of the pes-1 staining pattern:
Early eggs were collected by cutting open gravid hermaphrodites. The eggs were applied to eight-well multitest microscope slides coated with polylysine (1 mg/ml). Development was recorded to the appropriate stage using a 4D microscope (HIRD and WHITE 1993 ; SCHNABEL et al. 1997 ). Slides were frozen immediately and processed for ß-galactosidase staining as described previously (WOOD 1988 ; HOPE 1994 ). For each embryo the pattern of staining was recorded in a series of images along the z-axis and compared to the final images of the 4D-recording. The identity of the cells was then determined by lineaging the embryos with a newly developed computerized 4D-lineaging system (SCHNABEL et al. 1997 ). Laser ablations were carried out as described previously (HUTTER and SCHNABEL 1994 ).

Statistical analysis:
Because of the necessarily low numbers of mutant embryos analyzed, a deviation in the expression pattern of the pes-1 marker gene from that in wild type in only a few of the tested embryos may not be significant. Therefore we indicate in Figure 3 and Figure 4 the statistical significance of the pattern alterations. Chi-square analysis was used to assess the significance of reduction or increase in the number of staining cells as compared to wild type. The level of significance was set at P < 0.05.

View larger version (90K): In this window In a new window Download PPT slide   Figure 1. AB, MS, and D components of the pes-1::lacZ expression pattern. (a, b) ß-Galactosidase staining pattern of the AB and MS descendants at the 51-cell stage (32 AB cells) in the strains UL1 and GE3212 containing a chromosomally integrated pes-1::lacZ fusion gene. The two daughters of MSaa are indicated in the UL1 embryo (small arrows). Staining in these cells is weaker in comparison with other components of the pes-1::lacZ expression pattern and was not detected in GE3212. (c, d) Staining of the four D descendants in the 98-cell stage embryo. The intensity of this staining is generally in the range of that of the AB descendants at the 32-AB cell stage. Concerning the AB component of the pattern only the ABalp lineage stains more intensely. Anterior is to the left in all photos. The bottom of the figure shows the schematic diagram of the staining components used in Figure 3 and Figure 4 to show the pes-1::lacZ expression pattern in mutant embryos. The eight largest circles represent the first eight AB cells (ABala, ABalp, ABara, ...). The smaller circles inside represent the 16 daughters (ABalaa and ABalap, ABalpa and ABalpp, ...) and the 32 granddaughters (ABalaaa and ABalaap, ...) of these cells. lacZ expressing cells are shown as black circles; open circles denote cells that do not express the construct. The fractions near the cells correspond to the proportion of embryos where staining was observed. For instance, the fraction 9/10 above ABara means that ß-galactosidase staining was observed in the two daughters of the ABaraa cell in 9 out of 10 embryos.

View larger version (58K): In this window In a new window Download PPT slide   Figure 2. Localization of pes-1 mRNA in wild-type C. elegans embryos detected by whole-mount in situ hybridization. Embryos were hybridized to a pes-1 antisense probe visualized using alkaline-phosphatase-mediated detection. (a) Staining in two clusters of cells. In the left cluster we counted ~10 cells and in the other cluster 4 cells. The pattern is similar to that observed after ß-galactosidase staining of UL1 embryos at the 32-AB-cell stage. (b) An older embryo (~100 cells) shows two diffuse clusters of staining cells, which could correspond to the expression in the AB lineage, and on the right a cluster of a few cells brightly stained, which could correspond to the expression in the D lineage.

View larger version (53K): In this window In a new window Download PPT slide   Figure 3. The expression of the pes-1::lacZ marker in UL1 is coupled to the blastomere identity in the AB cell lineage. For explanation and references see RESULTS. The new identities of blastomeres in the mutants or after ablation of EMS are indicated. (A) Patterns in embryos in which the left-right induction in ABa does not occur. pes-1::lacZ expression in all ABa descendants is lost because fate transformations have eliminated the identities of blastomeres in which pes-1::lacZ is normally expressed. (B) Patterns in mutant embryos where the ABp identity is not induced. ABp now executes an ABa fate, which causes the blastomeres contacting MS to become left-right induced. (C) In glp-1 (e2144) embryos only those fates are present that normally do not express pes-1::lacZ. (D) In a par-3 mutation pes-1::lacZ expression becomes random. (E) pes-1::lacZ expression pattern in wild-type background (strain UL1). For nomenclature see Figure 1. To reduce the complexity of the diagram the embryos are shown at the 16-AB-cell stage. This is justified because no variability in the staining of the sister blastomeres was observed, i.e., either both sisters show staining or both sisters do not show staining. It does not mean, however, that staining has been observed before the 32-AB-cell stage.

View larger version (77K): In this window In a new window Download PPT slide   Figure 4. Patterns of pes-1::lacZ expression in mutant backgrounds. The first panel of each row shows ß-galactosidase staining of an embryo of the corresponding class. (A) Reduction in the number of staining cells as compared to UL1. (B) Absence of consistent pattern. (C) Ectopic staining in addition to the UL1 pattern. (D) Reduction in the number of staining cells as compared to UL1 together with ectopic staining. (E) UL1 pattern. For nomenclature see Figure 1. Where no numbers are indicated, the cells in all of the corresponding embryos showed no staining as in UL1.

In situ hybridization:
Digoxigenin-labeled RNA probes were synthesized by in vitro transcription using SP6 and T7 polymerases and digoxigenin-11-UTP. Probes were synthesized from a full-length pes-1 cDNA cloned in pCRscript plasmid. Whole-mount in situ hybridization on embryos was performed on wild-type N2 embryos as described (SEYDOUX and FIRE 1995 ) with the following modifications: after fixation in methanol the slides were immersed in acetone at -20° for 5 min; a proteinase K digestion step was included (15-min incubation in a 0.1 mg/ml solution); during prehybridization, embryos were preheated for 2 hr at 48° in a buffer containing 1x Denhardt's (0.02% Ficoll 400, 0.02% polyvinylpyrolidone, 0.02% bovine serumalbumin), 2 mM EDTA, 50% deionized formamide, 500 mg/ml ss-DNA, 4x SSC, 10% dextran sulfate; after overnight hybridization at 48° the embryos were washed as for cDNA-derived DNA probes (SEYDOUX and FIRE 1995 ). No signal was detected with the sense probe.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Background:
The gene pes-1 was identified by promoter trapping (HOPE 1991 ). In the transgenic strain UL1, DNA containing the genetic regulatory elements of pes-1 fused with the bacterial lacZ reporter gene is integrated into chromosome V (HOPE 1991 , HOPE 1994 ). The ß-galactosidase expression pattern in this strain has several components (HOPE 1994 ). The earliest component is a subset of the AB descendants. ß-Galactosidase is detected in 14 specific cells and their descendants from the 44-cell stage (32 cells in the AB lineage) up to the 220-cell stage (Figure 1). A second component of expression is in the lineage of the D founder cell where ß-galactosidase is detected in all descendants of D from the 4-D cell stage (98-cell stage) up to the 16-D cell stage (385-cell stage). The intensity of the expression in the D lineage corresponds to that in most of the AB descendants at the 32-AB-cell stage and can thus be used as a control for potentially unspecific alterations of expression in the AB lineage as will be discussed later. A third component of this pattern is observed in the two precursors of the somatic part of the gonad, Z1 and Z4. In UL1, in addition to the previously reported components of the pes-1::lacZ expression pattern, we detected ß-galactosidase staining in the lineage of the MS founder cell, specifically in MSaaa and MSaap (51-cell stage) (Figure 1). No staining was observed in the descendants of these two cells. The MS lineage staining is weaker in comparison with the other components of the pes-1::lacZ expression pattern and detected in only 7 out of 10 embryos.

Because this work aims to identify genes involved in the specification of AB-derived fates, it was critical to assess whether the expression of the pes-1::lacZ marker construct reflects the normal expression of the gene and whether the expression is indeed coupled to the identity (fate) of early blastomeres.

The pes-1::lacZ expression pattern in the AB and D lineages is independent of the site of integration in the genome:
The expression pattern for a chromosomally integrated pes-1::lacZ fusion gene was first determined in terms of the C. elegans cell lineage in the strain UL1 (HOPE 1994 ). To make sure that the expression pattern is independent of site of integration, we integrated the pes-1::lacZ fusion gene from an extrachromosomal array (strain UL56) into a chromosome (strain GE3212). The ß-galactosidase expression pattern in the AB and D lineages of GE3212 is identical to that of UL1 (Figure 1) except that expression is weaker. No staining was observed in the MS lineage; however, this is not relevant here. The weaker staining intensity of GE3212 as compared with that of UL1 may be caused by a different copy number of the pes-1::lacZ fusion genes integrated in this strain or by the general status of the chromatin in the region of integration and could be the explanation for the failure to detect expression in the MS lineage.

pes-1 mRNA distribution is consistent with the pattern observed using a reporter gene:
To confirm that the expression pattern observed in the UL1 strain was representative of the expression pattern of the endogenous pes-1 gene, we examined the distribution of pes-1 mRNA in N2 embryos by in situ hybridization. In wild-type N2 embryos at the 32-AB-cell stage a pes-1 antisense probe stains two clusters of cells, one with ~4 cells and the other with ~10 cells (Figure 2A), which corresponds to the earliest expression of ß-galactosidase in UL1. In later stages two diffuse clusters of weakly staining cells are detected, which correspond to the fading expression in the AB lineage, together with a cluster of a few brightly stained cells, corresponding to the expression in the D lineage as seen in UL1 (Figure 2B). Even though it is not possible to determine the identity of individual cells, the general pattern is similar enough to exclude a completely different expression pattern of the RNA. No pes-1 RNA could be seen in the two somatic germline precursors Z1 and Z4 or in the MS descendants that express the gene in UL1, perhaps because of a lack of sensitivity.

In summary these experiments show that the expression of pes-1::lacZ in the transgenic strains very probably corresponds to the normal expression of the pes-1 gene. We therefore expect that the promoter of the transgene is regulated normally and that alterations of the pes-1::lacZ gene activity in an altered genetic situation reflect the normal behavior of the pes-1 locus.

The pes-1::lacZ expression pattern indicates specific cell fates in the AB lineage:
As already described, the pes-1::lacZ fusion gene is expressed in a rather complicated pattern in 14 AB descendants at the 32-AB-cell stage. For the sake of simplicity we describe the pattern here in terms of 8 AB descendants because inductions specifying the identities of the AB-derived blastomeres can be explained comprehensively at this stage. Mutants or manipulations that interrupt or alter these inductions can be used to assess whether the pes-1::lacZ expression is coupled to the identities of specific blastomeres. An alteration of blastomere identity should cause a new expression pattern if the expression of pes-1::lacZ is coupled to the identity of a cell.

The four anterior AB blastomeres at the 8-AB-cell stage are derived from ABa. Among these only the anterior descendants of ABara, but both the anterior and posterior descendants of ABalp, express the gene in wild-type embryos. The pattern of expression in the four ABp-derived blastomeres is very different, with only the four anterior descendants of all four blastomeres expressing the marker (Figure 1).

The four different identities of the ABa granddaughters are induced during the specification of the primary left-right asymmetry by a glp-1-dependent signal from the MS blastomere. In the absence of this signal ABalp and ABara cell identities are not established. These cells instead adopt, respectively, the fates of ABala and ABarp, which do not express the pes-1::lacZ construct. Therefore a lack of left-right asymmetry should abolish the expression of pes-1::lacZ in the ABa-derived lineages but should not affect the pattern in ABp descendants.

The ABp fate is induced at the four-cell stage by P₂. This induction depends on the genes glp-1 and apx-1 whose products serve, respectively, as receptor and ligand in this induction. In their absence ABp executes an ABa fate. Two scenarios can be proposed in the absence of this induction. If the left-right induction is also prevented, there should be no more pes-1::lacZ expression because all AB descendants would adopt ABala or ABarp fates. If, however, the left-right induction still occurs, all blastomeres contacting MS would be left-right induced. The pes-1::lacZ expression pattern in the ABa descendants should remain normal. ABplp and ABprp that contact MS would be left-right induced. They would adopt the ABalp fate and express the pes-1::lacZ fusion gene. ABpla and ABpra would execute the ABala fate and should no longer express pes-1::lacZ (for review and original references see SCHNABEL 1996 ; SCHNABEL and PRIESS 1997 ). All these potential pattern alterations are quite radical and should be easy to score.

When the induction of left-right asymmetry is prevented either by interference with the glp-1 receptor in the weak e2072 mutant or by ablation of the EMS blastomere with a laser microbeam (HUTTER and SCHNABEL 1994 ) or genetically in the skn-1(zu67) mutant (BOWERMAN et al. 1993 ), pes-1::lacZ expression is in agreement with the predictions: it is eliminated in the ABa lineage but remains normal in ABp (Figure 3A). A failure to induce both the ABp fate and the left-right asymmetry in ABa in strong glp-1 (e2144) mutants (HUTTER and SCHNABEL 1994 ; MOSKOWITZ et al. 1994 ) eliminates pes-1::lacZ expression completely (Figure 3B).

We also observed the very characteristic new expression pattern expected in the ABp descendants when the induction of the ABp fate, but not the left-right induction, is prevented in apx-1 and pie-1 mutants (Figure 3B). The ligand for this induction is absent in the first case, and the identity of the P₂ blastomere is not established in the second case (MANGO et al. 1994 ; MELLO et al. 1994 ). In these mutants, ABp behaves like ABa. As explained above, ABplp and ABprp, which contact MS, now become left-right induced and adopt the ABalp fate, while those that do not contact MS, ABpla and ABpra, adopt an ABala or ABara fate.

As expected, no effect on the pes-1::lacZ expression in ABa descendants was observed in one new allele of apx-1, t2063. An unexpected reduction of pes-1::lacZ expression was observed in ABaraa in a different allele of apx-1(t2176) and in pie-1(t1682) (Figure 3B). No requirement for those genes in the ABa lineage has previously been reported (MANGO et al. 1994 ; MELLO et al. 1994 ). Because in the case of apx-1(t2063), as in all other experiments, pes-1::lacZ expression is in agreement with blastomere transformations previously described, we think that a reduced pes-1::lacZ expression in ABaraa in apx-1(t2176) and pie-1(t1682) indicates a requirement for the two genes to specify some characteristic of ABaraa rather than an unreliable expression of pes-1. The most parsimonious explanation is that in these mutants the left-right specification is also at least partially interrupted (HUTTER and SCHNABEL 1994 ).

The effect of a completely aberrant AB lineage was tested using the gene par-3. In mutants for this gene the first division is not unequal, as in wild type, but equal (KEMPHUES et al. 1988B ) and the AB blastomere executes diverse fates of either the AB or MS lineages in a mosaic fashion similar to the effect seen in mex-1 embryos (SCHNABEL et al. 1996 ). As seen in Figure 3D this leads to a random expression of the pes-1 marker. In this case the expression of the marker could reflect the expression of any of the components of the pes-1 pattern even if expression of AB or MS components is more likely.

In summary the presented results suggest that the expression of the pes-1::lacZ fusion gene is coupled with the identities of the AB blastomeres during early development.

The pes-1::lacZ expression pattern is often altered in strains carrying maternal-effect embryonic lethal mutations:
To identify genes involved in the specification of the AB lineage we used a subset of mutants from a large collection of strains carrying maternal-effect embryonic lethal mutations (R. SCHNABEL, H. SCHNABEL, R. FEICHTINGER and T. KALETTA, unpublished results). This collection includes 89 genes located on chromosome II, 21 with multiple alleles, and 125 genes located on chromosome III, 44 with multiple alleles. Of the 89 genes identified on LGII, 32 were considered to be involved in functions regulating development because the early cleavage pattern is normal, although embryos arrest with differentiated tissues arranged in an aberrant morphology. Of the 32 genes, 23 are analyzed here, including 8 genes defined by multiple alleles that were analyzed by investigating the strongest alleles. The genes defined by single mutants were selected randomly. On chromosome III we analyzed 13 genes of 53 possibly involved in embryonic development. For 1 gene on LGIII, mod-4, two alleles (t1663, t1701) were analyzed.

The collection contains many genes probably involved in general functions in the cellular machinery. For example, some mutations cause defects in the early divisions or result in cell proliferation arrest. These genes were generally excluded from this study. However, we tested three representative mutations from this group on LGIII to see whether pes-1::lacZ expression is also affected in this type of mutant. Embryos from mothers homozygous for emb-85(t1483) show a very variable phenotype. Sometimes early cell cleavages are aberrant and/or embryos arrest at different stages of development. The mutation mar-1(t1696) causes embryos to arrest with ~100 round cells that show some differentiation. The third mutation, t1597, is in the gene mel-32, which codes for the enzyme serinehydroxymethyltransferase involved in one-carbon metabolism (VATCHER et al. 1998 ). Embryos arrest with 100–200 cells and are completely undifferentiated.

Strains containing the chromosomally integrated pes-1::lacZ fusion gene from the UL1 strain and the 40 maternal-effect lethal mutations in 39 genes were constructed using genetic crosses. The pes-1::lacZ expression pattern was determined after first recording the development of several embryos up to the 64-AB-cell stage using a 4D-microscope system (SCHNABEL et al. 1997 ). The embryos were then stained for ß-galactosidase activity. The cell lineage was determined to identify the cells expressing the transgene. This procedure permits unambiguous identification of cells.

Mutations in 24 genes did not detectably affect the pes-1::lacZ expression pattern. These genes are:

• LGII: arm-4(t1135), arm-7(t1160), emb-40(t1108), emb-47(t1238), emb-50(t1181), emb-53(t1270), emb-57(t1189), emb-60(t1290), emb-61(t1315), emb-62(t1180), emb-67(t1218), mel-11(t1282), mel-14(t1176), mel-18(t1144), mel-20(t1203), mod-1(t1157), and sud-1(t1237);

• LGIII: arm-6(t1677), mod-2(t1447), mod-5(t1502), mod-7 (t1549), mod-6(t1653), mod-4(t1701, t1663), and mod-8(t1530).

Mutations in the other 15 genes, including 3 presumably involved in general cellular functions, cause an alteration in the AB component of the pes-1::lacZ expression pattern. These 15 genes can be classified into different groups according to the different types of pattern alterations seen.

Mutations in four genes [arm-5(t1478), emb-52(t1265), emb-85(t1483), and mod-3(t1628)] cause a reduction in the number of staining cells as compared to wild type (Figure 4A). Mutation in emb-85(t1483), which, as previously described, results in a very variable arrest of embryogenesis, is the only mutation that also affects the D lineage component of the pes-1::lacZ expression pattern. Although expression can be absent from any of the blastomeres, expression in the ABalp lineage, which in UL1 shows the strongest expression, is much more frequently retained. For one gene (arm-5) we also determined the pattern of pes-1 in trans to a deficiency to determine whether a potential further reduction of the gene activity would further reduce the expression pattern compared to the homozygous mutation. As seen in Figure 4A this was indeed the case and suggests that an amorphic mutation in this gene may completely suppress pes-1 expression in the AB lineage.

Mutants in another group of four genes [arm-1(t1110), arm-2(t1214), mar-1(t1696), and mel-32(t1597)] show a very variable, if not random, expression of the transgene. Expression is altered in many of the cells and different embryos display different pes-1::lacZ expression patterns. The alterations include a lack of expression in cells normally expressing pes-1 and ectopic expression in cells normally not expressing pes-1 (Figure 4B) as already observed in a mutant of par-3 (Figure 3D). Interestingly, two mutations are in genes with general cellular functions (mar-1 and mel-32) and result in arrest of development with 100 to 200 cells. In contrast to mutations in genes required specifically during development, mutations in these two genes also affect expression of pes-1::lacZ in the D lineage. In the case of mar-1 we cannot exclude that embryos arrest slightly too early to express pes-1::lacZ. It is, however, our experience that arrested embryos also express markers normally expressed only later in development (LAUFER et al. 1980 ; R. SCHNABEL, unpublished results).

Mutations in five genes, arm-8(t1505), emb-46(t1212), emb-48(t1310), emb-49(t1104), and lit-1(t1534), yielded ectopic expression in some AB-derived blastomeres, but never all of them, showing expression not seen in a wild-type genetic background. ß-Galactosidase staining is still observed in cells where expression is detected in UL1 (Figure 4C). Ectopic staining is observed more often in the daughters of ABarp (four out of five mutants) than in the daughters of ABpla, ABplp, ABpra, or ABprp. Only in one case did we observe aberrant expression in a daughter of ABala. These new patterns could, as will be discussed, very well reflect an alteration of cell identities in the AB lineage.

Mutations in two genes, arm-9(t1635) and inx-1(t1460), cause both a reduction in the number of staining cells as compared to wild type and an ectopic expression of the marker (Figure 4D). The new pattern in inx-1 is striking. The arm-9 (t1635) mutant pattern is similar to that of apx-1(t2063) embryos.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this analysis we used the expression pattern of the pes-1::lacZ fusion gene to screen maternal-effect embryonic lethal mutations for their effect on AB lineage specification in the early C. elegans embryo. Mutations in 12 out of 36 genes that were suspected to be involved in embryonic development cause an alteration of the expression of the transgene in the AB descendants but not in the D lineage.

Selection of mutants for the analysis:
During the extensive screens carried out previously to identify genes involved in the regulation of embryonic development, mutations were selected only for their embryonic lethality. After outcrossing, mapping, and genetic complementation, genes were grouped according to the primary defects observed in mutant embryos. Only well-differentiated mutants showing morphological defects with rather normal early cleavage patterns are considered to contain mutations in genes involved in regulatory functions. The operational criteria are that embryos can be lineaged using a 4D-microscope and that embryos develop to a stage where terminal fates like hypodermis, pharynx, intestine, muscle, etc., can be scored.

Mutations that do not alter the pes-1::lacZ expression pattern:
The finding that an embryonic lethal mutation does not affect a specific process also can be considered to be important information. The genes corresponding to the mutations that do not affect the pes-1 pattern are probably not directly or indirectly involved in those aspects of the specification of the early AB-derived blastomeres that can be assayed by the pes-1::lacZ expression pattern.

Variability of pattern alterations:
Many mutations cause variable alterations of the expression of the marker. This could be so for two reasons: either the mutations are only hypomorphic or the complete loss of gene activity causes a random distribution of cell fates. Unfortunately we are not able to discriminate between these possibilities in this study. The firm identification of the function of a gene by genetic means requires a careful definition of the amorphic phenotype. Hypomorphic mutations may only unravel the function incompletely or cause a variability of the expressed phenotypes (for an example see KALETTA et al. 1997 ). Some mutations studied here may indeed be hypomorphic, as shown for arm-5 (t1478) by lowering the gene dosage in trans to a deficiency.

Interpretation of altered expression patterns:
Alterations in pes-1::lacZ expression patterns could first be due to mutations specifically affecting pes-1 transcriptional regulation. This hypothesis of expression pattern alterations independent of cell fate could apply to the four mutants of group A, where the number of cells expressing the transgene is reduced and variable as compared to UL1. The observation that expression in the ABalp descendants is more frequently retained could be due simply to galactosidase expression being stronger for these cells. In UL1 these are the first cells to show expression within the AB lineage and this staining does appear stronger (HOPE 1994 ). With the exception of the t1473 mutant, in which the pes-1::lacZ expression pattern was determined because of the general defects of this mutant, no alteration was observed in the D lineage component of the expression pattern, which demonstrates general lineage specificity of the alterations.

Mutations in the four genes of group B cause a variable expression of the transgene throughout the whole AB lineage. The expression patterns for each mutant vary from embryo to embryo. The transgene is not always expressed in cells showing expression in UL1 and is often expressed ectopically. These inconsistent patterns could be explained by a random distribution of the cell fates in the AB lineage or a random expression of the fusion gene. The two mutations that cause an arrest of embryogenesis with 100–200 cells also fall into this class. The way mel-32, which codes for a metabolic enzyme involved in the one carbon metabolism (VATCHER et al. 1998 ), affects the pes-1::lacZ expression pattern remains to be determined.

In the mutants of group C the normal expression pattern is retained but additional blastomeres express the marker in consistent and specific patterns. In none of these mutants is the gene expressed in all AB-derived blastomeres. In four of the five mutants ABala never expresses pes-1. The significance of this finding will be discussed when the pattern alterations are considered in detail. In two mutants grouped in D (Figure 4) the expression pattern is specifically reorganized. A loss of expression in the blastomeres where the gene is normally expressed is accompanied by expression in other blastomeres. The pattern alterations seen in the last two groups suggest that the genes may be involved in the specification of fates in the AB lineage. These groups are thus worth further study.

No gene involved in the specification of the AB fate per se has yet been identified. Mutants that affect the AB fate, like those in the par or the mex-1 genes (KEMPHUES et al. 1988B ; MELLO et al. 1992 ; SCHNABEL et al. 1996 ), most probably affect AB indirectly by preventing the proper segregation of factors normally specifying other lineages, which then causes secondarily an aberration of the AB fate. Therefore it is not possible to predict the fates that would be executed in the AB lineage if the AB fate were not specified in this lineage. It is possible that blastomeres would not differentiate at all or that they would express any other differentiation potential or even several randomly, as observed for par-3 in this study. Mutants that interrupt the AB specification could thus, depending on the default fate acquired, show any pattern of pes-1 expression in the AB descendants. A loss or a random expression of the marker as observed in groups A and B could therefore still indicate an interference with the AB fate itself.

A random specification or loss of cell fate in the AB lineage in the mutants of group B is specifically indicated by the fact that all ABala blastomeres express the marker. The ABala identity is the ground state of the AB lineage expressed in the absence of all known inductions (HUTTER and SCHNABEL 1995 ). It should thus be the most stable fate if the AB identity is established at all. Therefore an expression of the marker in ABala indicates that the AB lineage per se may be affected in a mutant. In contrast to group B, in groups C and D, which show specific pattern alterations, the expression of pes-1 is with one exception still suppressed in ABala. The specificity of the mutants in groups C and D can be also deduced from the fact that most of the pattern alterations observed can be rationalized with the current knowledge of the pathway of specification of the AB lineage.

As discussed previously, the AB component of the pes-1::lacZ expression pattern in UL1 is directly linked with the identity of the AB blastomeres. From the 2-AB-cell stage on and up to the phase where the pes-1::lacZ fusion gene is expressed in the AB lineage, blastomere identities are specified by breaking the anterior-posterior or left-right equivalence of the AB descendants in binary switches (SCHNABEL 1996 ; KALETTA et al. 1997 ). In particular, at each round of cell division the anterior fate corresponds to the default state. Therefore all pattern alterations, where posterior sisters of cells normally expressing the transgene gain an expression of the marker, may identify genes involved in creating anterior-posterior asymmetry within the AB lineage. Candidates for this process are found in groups C and D [emb-46(t1212), lit-1(1543), arm-8(t1505), and arm-9(t1635)]. In five out of the seven mutants from these groups the descendants of ABarp express the marker. It is very difficult to rationalize with the current knowledge why this phenotype is apparently so common in the collection. A failure to specify the anterior-posterior polarity of the eight AB descendants causes ABarp to express the fate of the ABala blastomere (HUTTER and SCHNABEL 1994 ) that does not express the transgene. Therefore one would expect that any mutation altering the specification of the left-right asymmetry should cause a lack of expression in ABalp and not an ectopic one in ABarp.

Careful analysis of the phenotype of the lit-1 mutant phenotype (KALETTA et al. 1997 ) permits an explanation for the pes-1::lacZ expression pattern alteration observed in this mutant. The expression pattern of the marker in the lit-1 mutant t1534 is generally consistent with the finding that in this mutant the posterior fates of the 16-AB-cell stage are replaced by anterior ones. In lit-1 mutant embryos, pes-1::lacZ expression is observed in both anterior and posterior descendants of the blastomeres whose only anterior descendants, in UL1, express the transgene. pes-1::lacZ expression in ABarp is, however, not expected from the previous analysis (KALETTA et al. 1997 ). Whether expression in ABarp results from the alteration of an unidentified AB specification mechanism or suggests that pes-1 expression regulation is complex and does not depend only on cellular identity, remains to be clarified.

pes-1::lacZ expression patterns of mutants in two genes raise the possibility that these genes define new members in known pathways. emb-46(t1212) and lit-1(t1534) mutants show an identical expression pattern, which suggests that both genes may confer very similar functions. An epistasis analysis placed lit-1 upstream of pop-1, a gene of the Wnt pathway (LIN et al. 1995 ; KALETTA et al. 1997 ). The expression pattern of arm-9(t1635) mutant embryos resembles very much that of the apx-1 mutant. The gene may thus be involved in the notch/delta signaling pathway. Further work will have to be done by comparing in detail the phenotype of emb-46 and lit-1 or arm-9 and apx-1, respectively, to test the possibility that these genes may indeed be members of known pathways.

It appears that pes-1 is generally useful to prescreen a large number of mutants for possible defects in the AB and D lineage. The complex expression pattern of the pes-1::lacZ fusion gene is not a disadvantage for the study but a very useful means to assort genes into different groups that may correspond to different functions required for the proper expression of the gene. Expression of the gene in the D lineage was especially useful as an internal control in the analysis of the expression pattern alterations.

As discussed above, patterns resembling those of known genes may suggest processes in which new genes could be involved. The pes-1::lacZ expression patterns described here help to single out genes of interest, which then have to be characterized carefully to assess the gene function in detail. An analysis of the phenotype of a gene by immunochemistry and 4D-microscopy, necessary to assign defects to specific lineages because most tissues are of polyclonal origin, is a very laborious process that may take as much time as this analysis of 39 genes. A clear advantage of the use of an early expressed marker like pes-1, compared to immunochemical staining of terminal stage embryos, which requires approximately the same effort, is that it supplies very specific lineage information. The establishment of a system of markers for several, if not all, somatic founder cells, could greatly facilitate the analysis of the existing mutant libraries or of the large numbers of gene knockouts that could possibly be created in conjunction with the genome project.

Complexity of a developmental process, number of genes involved in AB lineage specification:
Our results indicate that many genes are involved in the specification of the complex AB lineage, which produces a great variety of different tissues. In contrast, we did not identify a gene required specifically for the specification of the D lineage, which produces only body wall muscle. This fact suggests that the D lineage is indeed genetically less complex, as one may expect for a lineage that gives rise to only 20 cells of the same tissue. Furthermore, the low number of mutations affecting expression of the transgene in the D lineage, as compared to the number of those affecting the AB lineage component, demonstrates that pes-1::lacZ expression is not generally sensitive to genetically modified situations. The pattern alterations observed in the AB lineage, therefore, more likely reflect the complexity of the normal regulation of the gene. A high complexity of the spatial regulation of genes during development has already been suggested by YUH and DAVIDSON 1996 , who identified more than 30 different binding sites in the promoter region of a sea urchin Endo16 gene. From our results, an estimation of the number of genes regulating, directly or not, pes-1 gene expression can be proposed. For the purposes of this estimation we exclude the three genes tested because of their presumed general cellular functions. Mutations in 12 out of 36 tested genes, from a pool of 85 genes suspected to be involved in developmental regulation and located on chromosomes II and III, affected the pes-1::lacZ expression pattern. By extrapolating to all genes in the pool one would expect that ~28 genes could affect pes-1. Considering the relative sizes of the six chromosomes of C. elegans, about a quarter of the genome was covered in the mutant screens by these two chromosomes. One could thus estimate that ~100 genes may regulate pes-1 expression in the AB lineage. This linear extrapolation may be even on the low side since FEICHTINGER 1995 showed that these extensive screens (e.g., 27,400 mutagenized lines for LGII) for maternal-effect embryonic lethal mutations (R. SCHNABEL, H. SCHNABEL, R. FEICHTINGER and T. KALETTA, unpublished results) did not lead to saturation.

Another and perhaps more realistic estimate of the number of genes involved in pes-1 expression can be made by using the numbers of parentally supplied genes essential for embryogenesis estimated by using similar screens on LGs IV and V. Because of statistical uncertainties, the number of genes is in the range of 300–1000, if only genes that have been uncovered by multiple alleles are considered (FEICHTINGER 1995 ). Considering that ~40% (for example, 85 of 214 genes on LGs II and III) of these may have regulatory functions in embryogenesis and again a third of these may influence the pes-1::lacZ expression pattern in the AB lineage, one can estimate that 40–120 genes are required for a proper expression of pes-1. These figures do not take into account genes that are only represented by a single allele in the mutant collection and that, when mutated, may also alter pes-1 expression as shown in this study. These genes may belong to a large group of genes with multiple functions during development (FEICHTINGER 1995 ). This estimate is, therefore, also probably on the low side. Whatever is the exact number of genes, even the lowest estimate reveals a high complexity of pes-1 expression regulation in the AB lineage.

	FOOTNOTES

¹ Present address: Department of Biology, University of Leeds, Leeds LS2 9JT, United Kingdom.
² Present address: Institut für Genetik, Technische Universität Braunschweig, 38106 Braunschweig, Germany.
³ Present address: deVGen, Technologiepark 9, B-9052 Gent-Zwijnaarde, Belgium.

	ACKNOWLEDGMENTS

We thank Elke Stenvers, Gaby Sowa, and Marie-Helene Faure for excellent technical assistance. This work was supported by the European Communities Human Capital and Mobility Programme (CHRX-CT, 94-0553).

Manuscript received July 20, 1998; Accepted for publication September 22, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BOWERMAN, B., B. W. DRAPER, C. C. MELLO, and J. R. PRIESS, 1993  The maternal gene skn-1 encodes a protein that is distributed unequally in early C. elegans embryos. Cell 74:443-452[Medline].

BRENNER, S., 1974  The genetics of Caenorhabditis elegans.. Genetics 77:71-94[Abstract/Free Full Text].

FEICHTINGER, R., 1995 Quantitative analysis of maternal gene functions of Caenorhabditis elegans. Ph.D. Thesis, University of Vienna, Austria.

HIRD, S. N. and J. G. WHITE, 1993  Cortical and cytoplasmic flow polarity in early embryonic cells of Caenorhabditis elegans.. J. Cell Biol. 121:1343-1355[Abstract/Free Full Text].

HOPE, I. A., 1991  Promoter trapping in Caenorhabditis elegans.. Development 113:399-408[Abstract].

HOPE, I. A., 1994  PES-1 is expressed during early embryogenesis in Caenorhabditis elegans and has homology to the fork head family of transcription factors. Development 120:505-514[Abstract].

HUTTER, H. and R. SCHNABEL, 1994  glp-1 and inductions establishing embryonic axes in C. elegans.. Development 120:2051-2064[Abstract].

HUTTER, H. and R. SCHNABEL, 1995  Specification of anterior-posterior differences within the AB lineage in the C. elegans embryo. Development 121:1559-1568[Abstract].

KALETTA, T., H. SCHNABEL, and R. SCHNABEL, 1997  Binary specification of the embryonic lineage in Caenorhabditis elegans.. Nature 390:294-298[Medline].

KEMPHUES, K. J., M. KUSCH, and N. WOLF, 1988a  Maternal-effect lethal mutations on linkage group II of Caenorhabditis elegans.. Genetics 120:977-986[Abstract/Free Full Text].

KEMPHUES, K. J., J. R. PRIESS, D. G. MORTON, and N. CHENG, 1988b  Identification of genes required for cytoplasmic localization in early C. elegans embryos. Cell 52:311-320[Medline].

LAUFER, J. S., P. BAZZICALUPO, and W. B. WOOD, 1980  Segregation of developmental potential in early embryos of Caenorhabditis elegans.. Cell 19:569-577[Medline].

LIN, R., S. THOMSON, and J. PRIESS, 1995  pop-1 encodes an HMG box protein required for the specification of a mesoderm precursor in early C. elegans embryos. Cell 83:599-609[Medline].

LOOMIS, W. F. and P. W. STERNBERG, 1995  Genetic networks. Science 269:649[Free Full Text].

MANGO, S. E., C. J. THORPE, P. R. MARTIN, S. H. CHAMBERLAIN, and B. BOWERMAN, 1994  Two maternal genes, apx-1 and pie-1, are required to distinguish the fates of equivalent blastomeres in the early Caenorhabditis elegans embryo. Development 120:2305-2315[Abstract].

MELLO, C. C., B. W. DRAPER, M. KRAUSE, H. WEINTRAUB, and J.R. PRIESS, 1992  The pie-1 and mex-1 genes and maternal control of blastomere identity in early C. elegans embryos. Cell 70:163-176[Medline].

MELLO, C. C., B. W. DRAPER, and J. R. PRIESS, 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[Medline].

MIKLOS, G. L. G. and G. M. RUBIN, 1996  The role of genome projects in determining gene function: insights from model organisms. Cell 86:521-529[Medline].

MOSKOWITZ, I. P., S. B. GENDREAU, and J. H. ROTHMAN, 1994  Combinatorial specification of blastomere identity by glp-1 dependent cellular interactions in the nematode Caenorhabditis elegans.. Development 120:3325-3338[Abstract].

PRIESS, J. R., R. SCHNABEL, and H. SCHNABEL, 1987  The glp-1 locus and cellular interactions in the C. elegans embryo. Cell 51:601-611[Medline].

SCHNABEL, R., 1996  Pattern formation: regional specification in the early C. elegans embryo. Bioessays 18:591-594[Medline].

SCHNABEL, R., and J. R. PRIESS, 1997 Specification of cell fates in the early embryo, pp. 361–382 in C. elegans II, Monograph 33, edited by D. L. RIDDLE, T. BLUMENTHAL, B. J. MEYER and J. R. PRIESS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SCHNABEL, R., C. WEIGNER, H. HUTTER, R. FEICHTINGER, and H. SCHNABEL, 1996  mex-1 and the general partitioning of cell fate in the early C. elegans embryo. Mech. Dev. 54:1-15.

SCHNABEL, R., H. HUTTER, D. MOERMAN, and H. SCHNABEL, 1997  Assessing normal embryogenesis in Caenorhabditis elegans using a 4D microscope: variability of development and regional specification. Dev. Biol. 184:234-265[Medline].

SEYDOUX, G., and A. FIRE, 1995 Whole-mount in situ hybridization for the detection of RNA in Caenorhabditis elegans embryos, pp. 323–337 in Methods in Cell Biology, Vol. 48, edited by H. F. EPSTEIN and D. C. SHAKES. Academic Press, San Diego.

SULSTON, J. E., E. SCHIERENBERG, J. G. WHITE, and J. N. THOMSON, 1983  The embryonic cell lineage of the nematode Caenorhabditis elegans.. Dev. Biol. 100:64-119[Medline].

VATCHER, G. P., C. M. THACKER, T. KALETTA, H. SCHNABEL, and R. SCHNABEL et al., 1998  Serine hydroxymethyltransferase is maternally essential in Caenorhabditis elegans.. J. Biol. Chem. 273:6066-6073[Abstract/Free Full Text].

WOOD, W., 1988 The Nematode Caenorhabditis elegans. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

YUH, C. H. and E. H. DAVIDSON, 1996  Modular cis-regulatory organization of ENDO16, a gut-specific gene of the sea urchin embryo. Development 122:1069-1082[Abstract].

This article has been cited by other articles:

				L Molin, A Mounsey, S Aslam, P Bauer, J Young, M James, A Sharma-Oates, and I. Hope Evolutionary conservation of redundancy between a diverged pair of forkhead transcription factor homologues Development, January 11, 2000; 127(22): 4825 - 4835. [Abstract] [PDF]