Early transcription in Caenorhabditis elegans embryos

The stage at which embryonic transcription begins varies widely among embryos of different organisms, ranging from the 2-cell embryo in the mouse to several thousand cells in Xenopus (reviewed in Davidson, 1986). The relative importance of maternally supplied and embryonically produced gene products in the early events of establishing embryonic axes, patterning of cleavages and determining of cell fates likewise seems to vary between different embryos. In Drosophila, for example, initial asymmetries and regional specification in the embryo are established under maternal instruction, but genes important in determining cell fate and refining the pattern for subsequent morphogenesis are among the earliest to be embryonically expressed, transcribed in localized patterns one to two cell cycles before high levels of general transcription are attained following cellularization of the blastoderm (Edgar and Schubiger, 1986; Ingham et al., 1985; Karr et al., 1985; Knipple et al., 1985; Weir and Kornberg, 1985). In embryos of another invertebrate, the leech Helobdella triserialis, embryonic transcription is detected considerably earlier, at 5 cells for RNA polymerases I and III, and at about 25 cells for α-amanitin-sensitive transcription by RNA polymerase II. The early process of teloblast formation at stage 5 is affected by drug treatment, but subsequent blast cell divisions occur, suggesting that some early steps in differentiation require embryonically produced gene products even as other processes involve only maternally supplied components until a more general transition at stage 7, corresponding to a mid-blastula transition (Bissen and Weisblat, 1991). We report here an analysis of the onset of embryonic transcription and some aspects of the transition from sufficiency of maternally supplied gene products to dependency on embryonic transcription in early embryos of the nematode Caenorhabditis elegans.

Early embryogenesis in C. elegans follows an invariant pattern in which several unequal and asynchronous cleavages produce individually recognizable founder cells for the major somatic lineages (Deppe et al., 1978; Sulston et al., 1983). By the 30-cell stage, about 90 minutes after fertilization at 20 °C, all five somatic lineages and the germ-line lineage have been established, and gastrulation begins. An ensuing period of about 4 hours includes gastrulation, continuing cell division and the early stages of organogenesis and morphogenesis. Cell division is essentially completed during this period and, in the final 7.5 hours before hatching, the embryo elongates, secretes a cuticle and completes morphogenesis (for review see Wood, 1988).

The early determination of several lineages appears to depend on segregated maternal products (Laufer et al., 1980; Strome and Wood, 1983; Cowan and McIntosh, 1985; Kemphues et al., 1988; Bowerman et al., 1993). Genetic evidence, in the form of an abundance of maternal-effect embryonic lethal mutations, indicates that much of early embryogenesis is controlled by maternal information (Kemphues et al., 1988; Kemphues, 1989). A recent study of the defective phenotypes of embryos homozygous for chro-mosomal deficiencies totalling about half the genome also suggests that the patterning of early cleavages is dictated almost exclusively by maternal gene products (Storfer-Glazer and Wood, 1994).

However, this deficiency analysis also indicates that successful morphogenesis requires the activity of many genes transcribed in the embryo. At least some genes associated with differentiation functions are expressed very early during embryogenesis. For example, the MyoD homolog hlh-1 is expressed in four muscle precursor cells at the 28-cell stage (Krause et al., 1990), and the appearance of gut granules, an early lineage-specific marker related to tryptophan degradation, can be blocked by α-amanitin treatment of 16-cell embryos, only one cell cycle after birth of the E cell, which is the clonal progenitor of the gut (Edgar and McGhee, 1986; 1988).

The time of onset for embryonic transcription in C. elegans has previously been assessed indirectly by two different methods, with somewhat differing results. In situ hybridization with labelled oligo(dT) probes to detect poly(A)⁺ mRNAs showed the first detectable increase in nuclear labelling at approximately 80–100 cells (Hecht et al., 1981). Subsequently, a more sensitive assay using run-on transcription assays in vitro showed that extracts from staged early embryos of less than 30 cells produced a high level of incorporation into mRNAs, an estimated 20 to 30 of which are expressed at high levels only during this period (Schauer and Wood, 1990). However, these experiments provided only an approximation of the time of transcription onset.

We have carried out in vivo assays of early transcription to define more precisely when it begins. Utilizing methods for digesting the chitinous eggshell of C. elegans, mechanically stripping off the inner vitelline membrane and culturing the resulting permeabilized embryos, we have followed early transcription directly as incorporation of radioactively labeled nucleotides into RNA, detected by acid precipitation and scintillation counting or by autoradiography. The same culture system has been used to determine how far embryos will develop when α-amanitin is used to inhibit RNA polymerase II, and to ask whether pregastrulation cleavages are affected by inhibition of RNA polymerase II-mediated transcription. While these methods necessarily limit experiments to small numbers of embryos, statistical analysis of our results clearly shows that embryonic transcription begins before the 16-cell stage. The results of inhibiting RNA polymerase II transcription with α-amanitin indicate that, up to about 100-cell stage, maternally supplied gene products are sufficient to support cell division, with normal patterning and timing of cleavages. However, this result does not preclude the likely possibility that early embryonic transcripts are important for cell fate specification (see Discussion).

Embryos were made permeable to nucleotides and α-amanitin by chitinase digestion of the eggshell, followed by mechanical stripping of the vitelline membrane with a narrow pipette (Edgar and McGhee, 1988). Up to 200 early embryos could be staged and permeabilized in each preparation. Permeabilized embryos were cultured at 22 °C in embryonic growth medium (EGM), modified slightly from Edgar and Wood (1993) by the addition of trace minerals (ZnSO_4.7H₂O, 5× 10⁻⁶M; FeSO_4.7H₂O, 1×10⁻⁶M; MnSO_4.7H₂O, 5×10⁻⁶M; Na₂SeO₃, 1×10⁻⁷M) and 0.1 volume LPSR-1 [low protein serum-replacement, (Sigma L9263)]. Cell division in untreated permeabilized embryos proceeded to approximately the normal number of 550 cells, with a rate of about 75% that of nonpermeabilized embryos (data not shown). For α-amanitin treatment, 150–200 μg/ml of the drug was added to the medium. In this enriched medium, 100 μg/ml of α-amanitin produced the same physiological level of blocking on gut granules as 30 μg/ml using the earlier EGM recipe (Edgar and McGhee, 1988); presumably the higher level is required due to increased non-specific adsorption of the drug by media components (see Greenleaf et al., 1979). Embryos were incubated in drops of medium covered with silicon oil (Dow Corning) or in covered slide chambers for longer experiments.

[³²P]UTP medium was prepared by mixing 5 volumes 2× EGM (made without uridine), four volumes of aqueous α-[³²P]UTP (NEN, in 0.01 M tricine, pH 7.6, 3000 Ci/mmol, 10 mCi/ml, final concentration approximately 1.3 μM), and one volume of either H₂O or α-amanitin stock solution (1.5 mg/ml in H₂O, Sigma). Groups of embryos were staged at 2, 4 or 8 cells, incubated to the desired age and permeabilized. For labelling, embryos of a given age were incubated at 22 °C in 8 μl drops of [³²P]UTP EGM with or without α-amanitin at 150 μg/ml, under silicon oil for 20 minutes. Embryos were then washed twice in nonradioactive EGM, pipetted into 10 μl of GITC lysis buffer (4 M guanidium isothiocyanate; 0.5% Sarkosyl; 5 mM NaCl; 100 mM β-mercaptoethanol), containing 42.5 μg/ml yeast tRNA (Boehringer Mannheim) as carrier, and frozen. Equal volumes of the last wash medium were frozen separately as controls for background.

Pulse-labelling of a complete series of embryonic stages required 2–3 days; however, it was found that embryonic viability during incubation decreased as the [³²P]UTP aged. Consequently, freshly ordered [³²P]UTP was used prior to its calibration date. All embryos labelled from a given batch of [³²P]UTP were counted simultaneously. Each frozen sample was thawed with 100 μl 20 mM EDTA in diethylpy-rocarbonate-treated H₂O containing 50 μg salmon sperm DNA, and 5 ml cold 10% TCA was added. After 15 minutes on ice, precipitated nucleic acids were trapped on a Whatman GF/C filter and washed with 30 ml 10% TCA and 5 ml 100% ethanol. Filters were dried, suspended in fluor (Ready-Solv HP, Beckman), and counted in a Beckman Model LS8100 liquid scintillation counter.

For determining radioactivity incorporated into RNA, labelled embryos frozen in 10 μl of GITC buffer with carrier tRNA were thawed in 100 μl additional buffer, and 10 μl of each sample was filter-precipitated as above and assayed for radioactivity. The remainder was extracted twice with phenol:chloroform (1:1) and once with chloroform:octanol (10:1), back-extracting the organic phase each time with an additional 50 μl of buffer mix containing carrier RNA. Potassium acetate was added to 0.3 M, and nucleic acids were precipitated with 2.5 volumes of ethanol. The samples were dissolved in 40 μl TE with RNAsin (Promega, 25 μl/ml). To determine the fraction of radioactivity attributable to RNA, each of four 10 μl aliquots was treated according to one of the following procedures: (1) One aliquot was TCA-precipitated and counted directly to determine recovery. (2) One aliquot was mixed with an equal volume of 2× RQ1 buffer (80 mM Tris, pH 7.9; 20 mM NaCl; 12 mM MgCl₂;0.2 mM CaCl₂) and digested with RQ1 DNAse (Promega; 50 units/ml for 15 minutes at 37 °C). (3) One aliquot was mixed with an equal volume of 2× proteinase K buffer (200 mM Tris pH 7.8; 25 mM EDTA; 300 mM NaCl; 2% SDS) and digested with proteinase K (400 μg/ml, 15 minutes at 37 °C). (4) One aliquot was mixed with an equal volume of 4× SSPE (660 mM NaCl; 80 mM Na₂HPO₄; 1 mM EDTA) and digested with RNase A (10 μg/ml, 30 minutes at 37 °C).

For analysis by gel electrophoresis, nucleic acids were extracted and precipitated from staged and labelled embryos by the procedures above, redissolved in 12 μl TE, and digested with RQ1 DNAse after adding an equal volume of 2× buffer and RNAsin (25 μl/ml). Aliquots representing equal numbers of nuclei (calculated from the age and number of embryos in each sample) were electrophoresed in formaldehyde gels, which were dried and exposed to X-ray film (Kodak X-OMAT AR).

5,6-[³H]UTP or uridine (NEN, 35 Ci/mmol, 1 mCi/ml, in 50% ethanol) was evaporated in a microfuge to near dryness, resuspended in 1/10 volume H₂O, and added to 9 volumes of 1.1× EGM, ([³H]UTP molarity approximately 30 μM). Embryonic labelling was done in the same manner as with ³²P, using embryos of up to 20 cells in these experiments and incubating for 15 or 30 minutes. Each batch of early embryos was divided in two, and one aliquot was labelled in the presence of 200 μg/ml α-amanitin.

Embryos were washed twice in nonradioactive EGM (wash time approximately 3 minutes) and immediately placed in a drop of fix (2.5% paraformaldehyde; 0.1% glutaraldehyde; 125 mM phosphate, pH 7.2) on a gelatin-subbed slide and flattened under a coverslip. Cov-erslips were floated off with additional fix (total time 1 minute), and the slides were then fixed for 10 minutes in 3:1 methanol:acetic acid at 4°C, followed by 3 minutes in 100% methanol and air dried. Slides were then rinsed in distilled water, dipped in photographic emulsion (Kodak NTB2 diluted 1:1 with distilled H₂O), dried and exposed for autoradiography at 4 °C. Control slides were digested 30 minutes at 37 °C with 100 μg/ml RNAse A in PBS before dipping in emulsion. After 5–10 days exposure, slides were developed in Kodak D19 (2.5 minutes at 15 °C), rinsed in dH₂O, fixed 5 minutes in Kodak Rapidfix, washed and dried. Slides were stained for 5 minutes with DAPI (Boehringer Mannheim, 1 μg/ml in PBS) to visualize nuclei and mounted in Gelvatol (Monsanto). Microscopy was performed with a Zeiss Photomicroscope at 400×; individual embryos were pho-tographed with bright-field illumination for counting silver grains and with low-level bright-field plus fluorescent illumination for nuclear counting. Several background areas away from the embryos were also photographed on each slide for counting silver grains.

Lineage of permeabilized embryos was followed and timed using a Zeiss inverted microscope with differential interference contrast (DIC) optics at 400×. Embryos were incubated in sealed slide chambers in 30 μl drops of EGM with or without 200 μg/ml α-amanitin at 25 °C. Kodak TMax400 film was used for photography.

In preliminary experiments, small numbers of permeabilized embryos incubated in embryonic growth medium (EGM, see Materials and Methods) containing either [³H]uridine or [³²P]UTP readily incorporated radioactive counts detectable after TCA precipitation. Incorporation was linear for at least 60 minutes. When samples of 80 embryos of approximately 100 cells each were labelled for 30 minutes with [³²P]UTP, >90% of the TCA-precipitable radioactivity was recovered in the nucleic acids fraction following phenol extraction. Aliquots of this preparation were treated with DNAse, proteinase K and RNAse A to determine the distribution of the incorporated label. As shown in Table 1, most of the incorporation was into RNAse-sensitive material.

To characterize this product further, groups of embryos staged at <20 cells, 30 to 60 cells, and 250 to 350 cells were labelled for 15 minutes with [³²P]UTP, and their nucleic acids were extracted as above. Analysis on agarose gels after DNAse digestion showed labelled material in the molecular weight range characteristic of cellular RNA, including bands of the appropriate sizes for rRNAs and rRNA precursor (Fig. 1). Incorporation into mRNA was substantially reduced in the presence of α-amanitin at 150 μg/ml. These results indicated that incorporation of label into RNA could be measured by counting of TCA precipitates and that mRNA synthesis was effectively inhibited by α-amanitin.

To determine the rate of incorporation as a function of developmental stage, labelling was carried out either in the absence or presence of α-amanitin on embryos staged to within a single cell cycle throughout the first half of embryogenesis, from two cells to the 550 cells present when morphogenesis begins. Fig. 2 shows incorporated TCA-precipitable counts per nucleus for each group of staged embryos, as a function of embryonic age (time after first cleavage) and number of nuclei per embryo. Results are shown for three of several experiments which gave similar results. In each paired set of embryos, the uptake of label was without exception lower in the presence of α-amanitin. Between 30% and 50% of the incorporation was estimated to be α-amanitin-sensitive at all stages. The first incorporation above background was detected in the period between 40 and 60 minutes after first cleavage, when the embryos have between 8 and 12 cells at the end of the labelling period.

To assess transcription in individual embryos, preparations permeabilized as above were labelled for 15 minutes with either [³H]uridine or [³H]UTP, washed, fixed and autoradi-ographed (Fig. 3). Both the nucleoside and nucleotide were incorporated, although the rate of labelling was higher with [³H]UTP. RNase digestion prior to autoradiography eliminated almost all of the cytoplasmic and a large fraction of the nuclear grains (data not shown). Data for early embryos, grouped by number of nuclei and labelled with or without α-amanitin, are shown in Fig. 4. Each group of embryos corresponds to one early round of cell division. By the fourth cell cycle (9 to 16 cells), there is an increase in incorporation and the difference in means with and without α-amanitin is statistically significant. At the following cell cycle (17 to 24 cells), mean incorporation both with and without α-amanitin differs significantly from that in the preceding cycle, at the 95% confidence level. Since the labelling period for each embryo covered the preceding 15 minutes, or cell cycle, and the means include embryos spread across their particular age group, the earliest increase in incorporation in these experiments is estimated to begin somewhere between 8 and 16 cells. Therefore, these results agree with those presented for ³²P labelling in the previous section.

Early cell divisions in C. elegans are strikingly invariant from embryo to embryo with respect to timing, inequality of daughter-cell sizes in some divisions, asynchrony and orientation of cleavage planes (Sulston et al., 1983). To investigate whether any aspects of this pattern depend on early embryonic transcription, early cell divisions were examined in the presence of α-amanitin, beginning with the earliest possible time of permeabilization.

To assess effects on the extent of cell division, the final nuclear number was determined in embryos staged to within 5 minutes and cultured with α-amanitin from specific ages. Fig. 5A shows the average final nuclear number plotted against the age and cell number at which exposure to the drug commenced. In Fig. 5B, these data are replotted to show the number of cell cycles completed after drug addition at various ages. The final number of nuclei is consistent with the interpretation that maternally supplied gene products are sufficient to support the first 6 to 7 rounds of cleavage. Subsequent to this point (128 nuclei), cell division stops less than one cycle after addition, indicating dependence on embryonically transcribed message. The additional cleavage in embryos permeabilized very early (<4 cells) was seen in several experiments and is statistically significant by paired T-tests.

To determine whether embryonic transcription is necessary for the asymmetric pattern of the early divisions which differentially segregate maternal information, the pregastrulation cell lineages were followed with DIC microscopy in permeabilized embryos, both in the absence and in the presence of α-amanitin. Because the eggshell has been removed, permeabilized embryos take on a somewhat different form than untreated embryos (Fig. 6); some of the normal cell contacts are missing and division is about 25% slower than in intact embryos (data not shown; for intact embryos see Deppe et al., 1978). Nevertheless, the relative timing of divisions within different lineages (Fig. 7), their distinctive cleavage planes and the characteristic inequality of daughter cell sizes in several of the early cleavages were observed to be fairly normal through the first 5 to 6 cell cycles. No differences were found between embryos cultured with or without α-amanitin (200 μg/ml) and observed up to 87 cells (Figs 6,7). These observations were made to within one cell cycle before drug-treated embryos stop division. At about 32 cells, permeabilized embryos characteristically undergo cell movement in the P lineage in which the C cells at the end of the embryo contact the AB and MS cells, resulting in a compact round embryo. These movements appear to originate in the two E cells (the gut progenitors), and may reflect the cell movements of early gastrulation in a normal embryo within the egg shell. The movements occur even in the presence of α-amanitin (Fig 6M-O). Thus maternal information appears sufficient to control at least the patterns of the early determinative cleavages that lead to establishment of the founder cell lineages.

Earlier run-on transcription experiments showed that extracts of pregastrulation C. elegans embryos, averaging about 8 cells each, exhibited a high rate of α-amanitin-sensitive nucleotide incorporation (Schauer and Wood, 1990). However, since the embryos used to prepare these extracts ranged from the 2-cell to the 30-cell stages, the results indicated only that transcription must initiate before gastrulation onset at the 30-cell stage. In the experiments reported here, we have determined the stage of transcription initiation more precisely using staged, permeabilized early embryos to assay incorporation of [³²P]UTP by scintillation counting and incorporation of [³H]UTP by autoradiography.

Our results show that transcription is initiated before the 16-cell stage, but probably not before the 8-cell stage. The precise time of initiation is difficult to pinpoint, because of scatter in the data resulting from at least three possible sources of error. First, only small numbers of embryos were available for each age sample after staging, permeabilization and division into two portions for labelling in the presence and absence of α-amanitin. Therefore, in the earliest samples where incorporated counts are close to background, the sampling error is high. Second, the chitinase permeabilization is a difficult procedure to standardize and differences in viability and corresponding rates of labelled UTP uptake were observed in different preparations of embryos. Third, the assay of a complete set of developmental stages required several days of experiments, and both embryonic viability during labelling and rate of [³²P]UTP uptake decreased with age of the labelled substrate.

Despite these difficulties, the results of [³²P]UTP experi-ments were quite consistent, showing no incorporation over background during the first three cell cycles and a significant increase over background during the fourth cell cycle, between the 8- and 16-cell stages. Because embryos were synchronized only to the extent of having the same number of cells at the beginning of the labelling period, we could not determine when in this cell cycle transcription begins.

We obtained similar results from [³H]UTP incorporation assayed by autoradiography, which showed that grains over nuclei above the background level were first visible in 8-to 12-cell embryos. The considerable variation in grain count observed between embryos of a given stage in the same preparation could reflect either transient developmental delays due to damage during permeabilization or differences in transcription rate at different points in the cell cycle. The degree of synchrony in these experiments was not sufficient to test the latter possibility; however, we did not see labelling over nuclei in which the chromosomes were visibly condensed, indicating that transcription ceases during mitosis, as would be expected. We also did not observe any obvious differences in labelling between cells of different lineages in those embryos where the chromosomal morphology allowed us to assign lineage identities.

By the 16-to 20-cell stage, embryonic transcription appears to be proceeding at a high level, consistent with the earlier run-on transcription results (Schauer and Wood, 1990). Thus both these studies indicate clearly that embryonic transcription in C. elegans embryos begins well before gastrulation, in contrast to the conclusion of Hecht et al. (1981) that message transcription begins after gastrulation at 80 to 100 cells. The discrep-ancy may be due to the differences in measurement techniques, as the earlier studies measured in situ hybridization to nuclear poly(A)⁺ mRNA, detecting the first increase in embryos of >80 cells, whereas we have measured nucleotide incorporation directly. If early transcription includes a high proportion of poly(A)-mRNA, or if message is rapidly transported to the cytoplasm in early embryos, the hybridization methods might not be sensitive enough to detect nuclear-cytoplasmic differences in a high background of cytoplasmic maternal message. Embryonic transcription is initiated as early as the 4-to 8-cell stage in the large parasitic nematode Ascaris lumbricoides (Cleavinger et al., 1989), which has an early cleavage pattern very similar to that of C. elegans, but a much slower rate of cleavage. It has been argued that in Xenopus (Newport and Kirchner, 1982a,b; Takeichi et al., 1985; Kimelman et al., 1987) and Drosophila (Edgar et al., 1986) transcription does not occur during the rapid divisions in which S-phase occupies most of the cell cycle because transcription complexes cannot form until the lengthening cell cycles include a G₂ phase. However, this explanation is not supported by results from C. elegans or from the leech Helobdella, and is not likely to account for the difference in time of transcription onset between C. elegans and Ascaris. Presence of a G₂ phase is clearly not sufficient for embryonic transcription: in embryos of both Helobdella and Ascaris, which at all stages have extended G₂ phases, no transcription is detected before the 4-cell stage (Cleavinger et al., 1989; Bissen and Weisblat, 1991); neither does presence of a G₂ phase appear necessary for transcription. Blast cells in Helobdella embryos at stages 7-8 synthesize RNA during S phase as well as during G₁ and G₂ (Bissen and Weisblat, 1989; 1991). In C. elegans the first G₂ phase occurs in the E lineage at the 20-cell stage (Edgar and McGhee, 1988), but we detect transcription earlier than that in most if not all lineages by autoradiography. Thus we conclude that, as in the leech, the S phase block to transcription is not absolute.

Inhibition of RNA synthesis by α-amanitin provides an estimate of the relative contributions to transcription of mRNA synthesis by RNA polymerase II, which is sensitive to the drug, versus synthesis of other RNAs by polymerases I and III, which are relatively insensitive. In our incorporation experiments, significant inhibition by α-amanitin is already seen by the 16-to 20-cell stage, indicating that RNA polymerase II contributes significantly to overall transcription from the time when it is first detectable. At all stages tested, we find that 30% to 50% of total transcription is α-amanitin sensitive. This could be an underestimate of the fraction of transcription by RNA polymerase II if inhibition were incomplete. We cannot rule out this possibility. However, the method of permeabilization by complete removal of the vitelline membrane allows maximal access of the drug; no significant mRNA synthesis is seen by RNA blot analysis in α-amanitin-treated embryos (Fig. 1), and the doses used (150-200 μg/ml) are well above the 100 μg/ml that produces the maximal effect on limitation of final cell number as well as loss of expression of gut granules, gut esterase and several antigenic markers of differentiation.

The 30% to 50% α-amanitin inhibition that we observe differs from results obtained earlier in run-on transcription experiments. In early embryo extracts, Schauer and Wood (1990) reported 85-90% inhibition of incorporation by 1 μg/ml of α-amanitin. A similar study of embryonic run-on transcrip-tion in Ascaris found 25% inhibition at the 4-to 16-cell stages, 55% at the 30-cell stage, and >80% at the 60-cell stage (Cleav-inger et al., 1989). A possible explanation for these differences is that in run-on assays, initiated transcripts are completed but reinitiation does not occur (Schauer and Wood, 1990). If RNA polymerases I and III initiate at a higher rate than II, more α-amanitin-resistant transcripts would accumulate during the labelling period of our experiments than in the run-on assays. Thus our direct labelling experiments may reflect the in vivo ratios more accurately.

It is clear that at least a few maternal determinants are partitioned during early cleavage to the appropriate cell lineages, where they participate in the determination of gut, germ line and other cell identities (Laufer et al., 1980; Strome and Wood, 1983; Cowan and McIntosh, 1985; Bowerman et al., 1992; Mello et al., 1992; Bowerman et al., 1993). However, the demonstration of early embryonic transcription raises the possibility that some critical early events in embryogenesis require embryonically synthesized gene products. To investigate the extent of such requirements, we used α-amanitin to assess the effects of eliminating or greatly reducing RNA polymerase II-mediated transcription on early cell division, cell lineage patterns and morphogenesis.

When α-amanitin is added to the culture medium of permeabilized embryos, the final cell number achieved is relatively constant, approximately 100 cells, whether exposure to the drug begins at 4 cells or at close to 100 cells. Addition after the 100-cell stage blocks division within approximately one cell cycle. This result is consistent with the earliest arrest stage described for a nonmaternal embryonic lethal mutation; an emb-29 mutant arrests at approximately 140 cells (Hecht et al., 1987). A reasonable model for our results is that components of the cell-cycling machinery, maternally supplied as either mRNA or proteins, can support division to approximately 100 cells, at which point these components must be supplemented or replaced by embryonically produced gene products for cell division to continue.

Somewhat surprising, however, is the finding that exposure of embryos to α-amanitin before 4 cells results in a significantly higher final number of nuclei, close to the 140 seen in emb-29 mutant embryos. One possible explanation is that early products of embryonic transcription, not made in the presence of α-amanitin, destabilize or degrade maternal messages or proteins required for cell division. A similar phenomenon of an extra cell division has been observed in Drosophila embryos injected with α-amanitin before the onset of embryonic transcription (Edgar and Schubiger, 1986).

Not only does early cell cycling in C. elegans embryos proceed in the presence of α-amanitin, but the lineage-charac-teristic patterns of cell-cycle timing, cleavage orientations and unequal cleavages also appear to be unchanged. Earlier work has shown that normal early cell fate determination depends on these patterns; when they are changed by manipulation or by mutation, cell fates are also affected (Priess and Thomson, 1987; Kemphues et al., 1988; Schnabel and Schnabel, 1990; Wood, 1991; Bowerman et al., 1992; Mello et al., 1992). Therefore, it is possible that the apparently normal early cleavages in α-amanitin-treated embryos are also accomplish-ing normal partitioning of maternal components and normal early cell fate determination, although we have no evidence other than cell appearance to support this possibility. If we assume that α-amanitin is effectively blocking RNA poly-merase II transcription in these experiments, none of this early pattern, up to at least 100 cells, would require embryonic tran-scription, even though transcription demonstrably starts during this period. Our results differ in this regard from those obtained with the leech embryo, in which the divisions forming the teloblasts at about 25 cells are abnormal in the presence of α-amanitin, generating supernumerary yolk-filled blastomeres rather than yolk-free teloblasts (Bissen and Weisblat, 1991).

An analysis of early lineages in C. elegans embryos homozygous for any one of several deficiencies, representing embryonic losses covering nearly half the genome, also identified no chromosomal regions required for correct early pat-terning (Storfer-Glazer and Wood, 1994). All the deficiency embryos showed abnormalities in morphogenesis, but most arrested with the normal number of 550 cells. Only one (carrying a deficiency that included the emb-29 gene) arrested before 140 cells. In addition, all known mutations affecting the early lineages, including genes known to affect fate specification as well as early cleavage patterns, show strict maternal effects (Priess et al., 1987; Kemphues et al., 1988; Mains et al., 1990; Bowerman et al., 1992; Mello et al., 1992).

However, α-amanitin does affect morphogenesis as well as expression of early tissue-specific markers of differentiation. In α-amanitin-blocked embryos, no organ primordia are observed and the embryos arrest as masses of undifferentiated cells. The earliest tissue-specific markers of differentiation can be blocked by α-amanitin treatment well before the time at which cell division is arrested. Gut granules, normally first detectable at about 100 cells, are sensitive to α-amanitin treatment up to the 16-cell stage but appear if the drug is added subsequently (Edgar and McGhee, 1986; 1988). This implies that the latest transcription leading to gut granule expression occurs before 16 to 20 cells, or during the second E cell cycle after the gut lineage is established. Gut esterase, a second gut-specific marker, appears slightly after 100 cells, but is sensitive to α-amanitin only until the end of the second E cell cycle (Edgar and McGhee, 1986; 1988).

Our results and other recent studies referenced above support the following model for C. elegans embryogenesis. The pregastrulation cleavage patterns, including asymmetric partitioning of cytoplasmic components and determination of some but not necessarily all founder cell fates are controlled by maternally supplied gene products. Embryonic transcription, initiated during this period as the founder-cell lineages become established, is required for refinements of fate specification dependent on cell interactions, subsequent cellular differentiation and morphogenesis. An answer to the question of whether embryonically produced gene products are required for founder-cell fate determination must await further molecular and genetic characterization of the earliest tran-scripts, which we are now pursuing.

We are grateful to Andrea Wilson and Weiqing Li for technical assistance, to Barb Robertson and Tom Johnson for advice on statis-tical analysis, to Irene Schauer for helpful discussions and critical reading of the manuscript, and to Judith Taylor for help with its preparation. This research was supported by grants from the National Institutes of Health (HD14958) to W. B. W. and the American Cancer Society (NP696) to L. G. E.