Genetic Control of Cell Division Patterns in the Drosophila Embryo

The stg Phenotype Suggests that stg Activity Triggers Mitosis

The eight alleles of stg isolated by Jürgens et al. (1984) are recessive embryonic lethals, and were originally identified because they cause a rather nonspecific reduction in the number of differentiated cuticular structures. The probable cause of this reduction was discovered by Y. N. Jan (personal communication), who noticed that stg embryos have a markedly reduced number of cells relative to wild-type embryos during the later stages of embryogenesis (Figure 1). This suggested that stg mutations caused a defect in cell proliferation.

To investigate the character of this cell-proliferation defect, we observed both live stg embryos under Nomarski optics, and fixed stg embryos that we stained to reveal the morphology of DNA, microtubules, and the nuclear envelope. We found that embryos homozygous for one the strongest alleles of stg (7B69) undergo the first 13 nuclear divisions with normal timing and morphology, and generate a cellular blastoderm with a wild-type number of nuclei, stg^7B69 embryos first deviate from normal development during the early stages of gastrulation, when they fail to initiate the 14th mitosis, which normally begins at about 200 min AED (after egg deposition, 25°C). We fixed and stained embryos homozygous for the three stongest alleles of stg (7B69, 8A83, 7L105), and found no evidence of chromosome condensation, nuclear envelope break-down, or spindle formation between 3 and 12 hr AED (Figure 1). In contrast, most cells in wild-type embryos divide at least twice during this period (Hartenstein and Campos-Ortega, 1985). We infer that the cells formed during interphase 14 in stg homozygotes remain in interphase for the duration of embryogenesis.

To determine the stage of cell-cycle arrest in stg mutants, we measured the DNA in nuclei of stg^7B69 embryos by scanning microdensitometry of squashed, feulgen-stained samples at about 6 hr AED. We found that these nuclei were 4C, and were thus arrested in G2 (data not shown). We also labeled DNA in stg^7B69 and wild-type embryos by injecting BrdU during G2 of interphase 14 (240 min AED, 22°C; see Edgar and Schubiger, 1986) and allowing the embryos to develop for 3 hr (Figure 2). In wild-type embryos, which progressed through S phases 15 and 16 during the labeling period, all nuclei became heavily labeled. In stg embryos only the giant, polyploid nuclei of the amnioserosa were labeled. These amnioserosa nuclei normally undergo multiple rounds of DNA synthesis after mitosis 13, but never divide (Hartenstein and Campos-Ortega, 1985). Their labeling in stg embryos indicates that the stg mutation does not block the normal program of polyploidization, and implies that the DNA synthetic machinery is functional in stg embryos. Thus, although we know that failure to complete DNA replication can cause cell-cycle arrest in interphase 14 (Edgar and Foe, unpublished data), it seems unlikely that this is the cause of cell-cycle arrest in stg embryos.

stg embryos undergo a remarkably normal program of postblastoderm development despite their cell-cycle arrest. They have a normally timed series of morphogenetic movements (see Figure 1), and the 5000 cells present in stg homozygotes eventually express markers found in most of the differentiated tissues of wild-type embryos, which hatch with perhaps 50,000 cells (Jürgens et al., 1984; Edgar, unpublished data; V. Hartenstein, personal communication). This lack of pleiotropy suggests that stg activity is required rather specifically for mitotic initiation. Our preliminary analysis of some of the weaker (hypomorphic) stg alleles supports this conclusion. Embryos homozygous for these alleles (9A19, 4B51, 9K21) support some mitotic activity after interphase 14. The mitoses that occur seem to be delayed; they occur over a longer period of development than in wild-type embryos, but nevertheless generate fewer cells. We saw no signs of polyploidy, aneuploidy, or spindle defects in these embryos. Our observations suggest that lowering the level of stg activity delays the timing of mitotic initiation without perturbing the coordination of cell-cycle events such as S phases, chromosome condensation, and spindle formation.

Cloning stg

stg was originally mapped to the region between bands 98A and 99A at the tip of chromosome arm 3R (Jürgens et al., 1984). We confirmed the more detailed mapping of Weigel and Jürgens (personal communication) using a series of terminal deficiencies, stg was uncovered by deficiencies distal to 98F, but not those distal to 99A. Thus, the region 98F–99A is required for stg function. We tested a P element insertion mutant at 99A, I(3)neo62 (Cooley et al., 1988), for its ability to complement the eight EMS-induced stg alleles of Jurgens et al. (1984). I(3)neo62 is a recessive embryonic lethal, but does not show a stg cuticle phenotype when homozygous. Nevertheless, it failed to complement the lethality of the strongest stg alleles (8A83, 7L105, 7B69), and only partially complemented the weaker stg alleles (13D22, 9A19, 7M53, 4B51, 9K21). Moreover, lethal transheterozygotes (stg^7L105/I(3)neo62) exhibited a weak stg cuticle phenotype.

We reverted the lethality of I(3)neo62 by transposase-mediated excision of the P element insertion (Robertson et al., 1988), and found that the reverted chromosome 3 was restored in its ability to complement stg^7L105 and stg^7B69. After cloning the genomic region containing this P insertion (see below), we confirmed that the P transposon had indeed “hopped out” of the 99A insertion site in the reverted chromosome (Figure 3). In addition to I(3)neo62, three more recessive-lethal P insertions in the same 2.2 kb genomic EcoRI fragment have been isolated (I(3)neo61: Cooley et al., 1988; I(3)3A1,1(3)1D3: Y. N. Jan, personal communication; see Figure 3). These also fail to complement stg, and one (I(3)3A1) exhibits a strong stg phenotype when homozygous.

Based on these observations, we concluded that the P element insertions at 99A caused stg mutations, and were thus in the proximity of the stg gene. We used plasmid rescue of I(3)neo62 (Cooley et al., 1988), followed by chromosome walking, to isolate 33 kb of the genomic DNA surrounding the P element insertion site (Figure 3). We confirmed that we had cloned sequences from 99A by hybridizing these sequences to Drosophila polytene chromosomes (not shown).

DNA fragments from this 33 kb genomic region were used to probe developmental Northern blots, and we found at least two transcription units in the cloned region (Figure 3). The transcribed region nearest the P element insertion site (less than 1 kb downstream) produces mRNAs of approximately 2.8 and 3.0 kb. The expression of these transcripts is as expected of a product involved in cell division: they are abundant during periods of much cell division (mitoses 1–13 and mitoses 14–16) and scarce during periods of little cell division (G2 of cycle 14 and postmitosis 16). The other RNAs transcribed from regions not adjacent to the P insertion site do not follow this profile (Figure 3). This suggests that the transcription unit adjacent to the P insertion site encodes the stg product, a conclusion that was strongly supported by further analysis in which we determined the sequence of mRNAs from this region and detailed their spatial patterns of expression (see below). Moreover, we found that embryos homozygous for the P element insertion I(3)3A1, a strong allele of stg, failed to express detectable amounts of the putative stg transcripts during postblastoderm development. This was determined by in situ hybridization of a stg cDNA probe (see below) to embryos derived from I(3)3A1/TM3 parents (data not shown).

stg Is Homologous to cdc25 of S. pombe

Using genomic DNA probes, we isolated cDNAs corresponding to the putative stg mRNAs. We sequenced a 2308 base stg cDNA and found a single long open reading frame (Figure 4). Searches of the Dayhoff protein sequence data base revealed that the predicted stg protein of 479 amino acids includes a C-terminal region of 187 amino acids that is 34% identical to the C-terminal portion of the cdc25 protein of the yeast S. pombe (Figure 5). The molecular function of cdc25 is not known, but genetic experiments indicate that it plays a role in cell-cycle regulation similar to stg: in a wild-type background, cdc25 is required for progression from G2 to mitosis, and when overexpressed it causes premature initiation of mitosis (Fantes, 1981; Russell and Nurse, 1986). In addition to the sequence homology, stg cDNA clones can complement the cdc25–22 temperature-sensitive mutation in S. pombe, showing that stg is a functional cdc25 homolog (P. Russell, personal communication). Another functional homolog of cdc25, MIH1, has recently been found in S. cerevisiae (Russell et al., 1989). MIH1 shares sequence homology with cdc25 in the same regions that stg does (Figure 5).

stg mRNA Expression Predicts the Sequence of Embryonic Cell Divisions

The phenotype of stg mutants shows that stg gene expression is required for the initiation of mitosis 14. Thus, in theory, differential expression of stg prior to mitosis 14 could cause the complex spatio–temporal pattern of mitotic initiation seen at this stage of development. To test this possibility, we analyzed the expression of stg mRNA during embryonic development by in situ hybridization of a stg cDNA probe to whole and sectioned embryos (Figures 6 and 7).

The Drosophila egg contains a large amount of maternal stg mRNA that is distributed uniformly throughout the yolk and cytoplasm (data not shown, but see Figures 3e and 3f). We know this RNA is maternal because it is present in unfertilized eggs, and because our studies with transcription and protein synthesis inhibitors indicate that stg transcription is negligible through cycle 13 (data not shown, but see Edgar et al., 1986b, and Weir et al., 1988, for examples of this kind of study). Maternal stg mRNA persists at a constant level from fertilization through cycle 13, and is degraded to a level undetectable by in situ hybridization during the first 20–30 min of interphase 14. Although stg has not yet been tested for a maternal effect, it seems likely that stg protein produced from maternal mRNA is required for the early, rapid nuclear cycles. If this is the case, the abrupt degradation of maternal stg mRNA after mitosis 13 could cause the termination of these cycles.

Zygotically synthesized stg mRNA begins to accumulate in the latter half of interphase 14, 25–35 min before the first cells enter mitosis 14. Accumulation is limited to cells that are destined to divide, and occurs with a temporal sequence nearly identical to, but preceding, the sequence of mitoses (Figures 6 and 7). The spatio–temporal pattern of mitosis 14 has been studied in detail by Foe (1989), who constructed an atlas of mitotic domains. She defines a mitotic domain as a group of cells that divide together, with timing distinct from their neighbors, and designates these domains with numbers indicating their relative order of entry into mitosis. The earliest zygotic stg expression is in a wide ribbon of cells along the ventral surface of the blastoderm that will invaginate to form the mesoderm (Figure 6A). This region corresponds to mitotic domain 10. Several minutes later, as the cephalic furrow begins to form, expression begins in a set of bilaterally symmetric stripes and spots in the head region (mitotic domains 1–3, 5–8), and the tail (mitotic domain 4; Figure 6B). Following this, as the first mitoses occur in the head, stg expression ensues in the dorsolateral regions, which correspond to mitotic domain 11. Expression in domain 11 initiates in five distinct spots (Figure 6C) and then spreads to fill the entire domain (Figure 6D). Twenty-five to thirty-five minutes after expression is first observed, mitoses begin to traverse the domain in a similar manner, initiating at five centers and spreading to the borders of the domain (Foe, 1989). At about the same time as expression in domain 11 becomes detectable, expression begins along the margin of the ventral furrow (mitotic domain 14; Figure 6E), and in expanded regions of the head (mitotic domains 9 and 15; Figure 7A) and tail (mitotic domains 12 and 13; Figure 6D). Just after this, expression begins in a series of segmentally repeated spots in the abdominal ventral ectoderm (mitotic domains 16 and 17, Figure 6E). In all of these cases we are able to correlate the regions of stg expression with mitotic domains using both the atlas constructed by Foe (1989) and the distribution of mitotic figures visible in our in situ preparations. Moreover, minor variations in the shapes and symmetry of the mitotic domains seen between different embryos are also reflected in the stg RNA patterns.

Our in situ preparations allowed us to assess, albeit crudely, relative levels of transcript in different regions of the embryo. We found that the sequence with which cells achieve high levels of stg RNA and the sequence of mitotic initiation are highly correlated (Figure 7). This can be seen within single domains, such as domain 11, where high levels of stg RNA are reached earliest in the five origins of mitosis mentioned above (Figures 6C and 6D). It can also be seen between domains; for example, although zygotic stg RNA is detected first in mitotic domain 10 (Figure 6A), accumulation proceeds more rapidly in domains 1–9, which initiate expression slightly later but nevertheless divide earlier (Figure 6B). These observations suggest that the level of accumulated stg activity determines the timing of mitosis. This notion is supported by our observation that the reduction of stg activity by hypomorphic stg mutations delays the timing of mitosis.

After mitosis, stg RNA disappears from most of the early dividing domains, and begins to accumulate in the ventral ectoderm, where many of the later dividing domains (21, 22,24,25, M, and N) are located (Figures 6F and 6G). During mitotic cycles 15 and 16, stg RNA expression recurs in a series of rapidly changing and increasingly complex patterns (data not shown). These patterns correlate well with maps of the mitotic pattern during these stages (Hartenstein and Campos-Ortega, 1985; Foe, personal communication), and with the distribution of mitotic figures in our in situ preparations, but we have not yet attempted a detailed analysis. Most cell lineages stop dividing after the 16th mitosis, during germ-band shortening. At this stage, stg RNA expression also ceases in most of the embryo. Cell proliferation continues primarily along the ventral surface of the nerve cord, in the brain cortex, and in the peripheral nervous system, where stg mRNA expression also continues (data not shown).

What Is stg?

Embryos lacking zygotic stg activity undergo cell-cycle arrest in G2 of interphase 14, just prior to the first zygotically controlled mitosis. This effect is highly specific; other aspects of development, such as cell movements and differentiation, proceed for many hours after the cell-cycle block. When stg activity is reduced by hypomorphic mutations, morphologically normal mitoses occur but their number is reduced and their timing delayed. These facts suggest that the stg product specifically catalyzes the initiation of mitosis, possibly in a level-dependent fashion. Since mitotic initiation seems to be the primary point of regulation in the early embryonic cell cycles (see Introduction), stg probably regulates not only mitosis, but cell-cycle progression in general.

The finding that stg is a structural and functional homolog of cdc25 from S. pombe allows a more specific speculation about its activity. Based on genetic studies, Russell and Nurse (1986) proposed that cdc25 is a level-dependent activator of another factor required to initiate mitosis, the cdc2 protein kinase. cdc2 encodes a component of MPF (mitosis promoting factor), an activity that is ubiquitous in proliferating eukaryotic cells (Dunphy and Newport, 1988b; Gautier et al., 1988). In both cultured human cells and Xenopus embryos, cdc2 protein is present during all phases of the cell cycle, but its kinase activity is activated prior to each mitosis (Draetta and Beach, 1988; Arion et al., 1988; see also Cyert and Kirschner, 1988; Dunphy and Newport, 1988a). This activation is presumed to trigger a cascade of phosphorylation events leading to nuclear envelope breakdown, chromosome condensation, spindle formation, and other mitotic functions (Newport and Kirschner, 1984; Karsenti et al., 1987; see Fantes, 1988, Lee and Nurse, 1988, and Dunphy and Newport, 1988c, for reviews). This set of relationships suggests that stg may trigger mitosis by activating a Drosophila cdc2 homolog.

Does stg Control the Developmental Pattern of Mitoses?

A simple way to regulate cell divisions differentially during development would be to express all but one of the required factors constitutively, and to then modulate expression of the single remaining, rate-limiting factor. Our findings are consistent with this type of regulation, and suggest that stg mRNA is the rate-limiting factor whose expression governs mitotic patterns after interphase 14. The most suggestive of these findings are first, that stg mRNA is expressed in a spatio–temporal pattern that anticipates the mitotic pattern, and second, that the stg gene is required zygotically for initiation of the first patterned mitosis. Since cell-cycle arrest occurs earlier in stg mutants than in any other characterized zygotic mutant, it seems likely that maternal supplies of other cell-cycle factors persist past the time when stg becomes required. The work of Lehner and O’Farrell (1989) on Drosophila cyclin A provides a clear example of a product that is required for mitosis yet seems not to be rate-limiting during the first zygotically controlled divisions. In contrast to stg, cyclin A derived from maternal mRNA is sufficient to support divisions through mitosis 15, and the level of cyclin A protein has little influence on the timing of mitoses.

We have demonstrated directly only that stg is required for mitosis 14, but our observations of mitotic patterns in hypomorphic and temperature-sensitive stg mutants suggest that stg is required for subsequent embryonic mitoses (Edgar, unpublished data). In addition, clonal analysis has indicated that stg is required for cell divisions in the imaginal discs during metamorphosis (Terle and Saint, personal communication). Although such observations do not address whether stg continues to be rate-limiting during later mitoses, the continued correlation of stg expression with mitotic patterns suggests that stg expression could determine the timing of all the postblastoderm divisions.

How Is stg Expression Controlled?

Perhaps our most significant finding is that the pattern of embryonic cell divisions may be predicted from the expression pattern of stg messenger RNA (Figure 7). This implies that the mitotic pattern may be controlled through differential rates of stg transcription and/or RNA degradation. While we know little about RNA degradation in Drosophila, recent studies have defined a plethora of factors involved in generating complex spatio–temporal patterns of transcription during embryogenesis (see Ingham, 1988, for review). These factors are encoded by the “selector” genes that set up patterns and determine cell identities in the embryo. To date, most of the selector genes studied at the molecular level encode DNA binding proteins of the zinc finger or homeodomain types that reside in the nucleus, and are widely believed to be transcription factors (see Gaul and Jäckle, 1987; Scott et al., 1989). This is fitting, since it has long been thought that selector genes determined cell fates by modulating the expression of “cytodifferentiation” genes encoding products directly responsible for cell structure, movement, and division (Garcia-Bellido, 1975). Our understanding of stg’s function clearly classifies it as a cytodifferentiation gene, and thus as a likely target of selector-gene regulation.

There are more compelling reasons than theory, however, to believe that stg is a selector-gene target. One is that the selector-gene expression patterns exhibit uncanny similarities to the stg expression patterns. Preceding mitosis 14, stg expression along the dorsoventral axis of the embryo is broken up into at least six distinct patterns that fall into the six dorsoventral domains defined by selector genes such as zerknüllt, twist, and those of the spitz group (Ingham, 1988; Mayer and Nüsslein-Volhard, 1988). Along the anteroposterior axis, stg expression is divided into different patterns in the head, thorax, abdomen, and tail, regions that are determined by differential expression of selector genes of the gap and homeotic classes (Gaul and Jäckle, 1987; Gehring, 1987). Within these regions, stg expression occurs in reiterated patterns with double- and single-segment periodicity that resembles pair-rule and segment-polarity gene expression patterns (Ingham, 1988).

In theory, the various selector-gene expression patterns subdivide the embryo into enough uniquely specified domains to account for virtually all aspects of stg expression during cycle 14. Moreover, the patterns of mitosis 14 are altered in predictable ways in many selector-gene mutants (Foe, personal communication). We expect that the altered patterns of mitosis in these mutants will be correlated with altered patterns of stg expression. Accordingly, we would like to propose that stg activity, and thus the mitotic pattern, is regulated at the transcriptional level by a variety of selector-gene products. Much of this regulation may be indirect, but the relative timing of selector-gene and stg expression suggests that direct interactions, perhaps using combinations of selector-gene encoded transcription factors, are involved.

Is Mitosis Important?

One higher order question that the stg mutations and clones will help us address is the relationship of cell-division patterns to other developmental processes. Cell division is clearly essential to generate the correct number of cells in different tissues and organs during development, and also plays a role, in conjunction with control of the orientation of division, in forming morphological structure (see Foe, 1989). It has also often been proposed that cell-cycle functions might somehow feed back on states of gene expression, thereby influencing determination and differentiation during development (Holtzer et al., 1975; Edgar and McGhee, 1988). Despite these expectations, the surprising degree of differentiation that occurs in stg mutants, which lack all postblastoderm cell divisions, indicates that the cell-division program in Drosophila may have little influence on morphogenetic movements, cell determination, or differentiation. Instead, different aspects of cell fate such as mitotic timing, movement, and structural differentiation seem to be regulated independently. Since all of these functions are directed by selector-gene expression, it would seem that the selector genes regulate many different types of target genes independently.

Mapping and Reversion of stg

The cytological position of stg was confirmed by crossing stg ^7B69/TM3 or stg ⁴⁸¹/TM3 to T(Y:3)R78/TM6, T(Y:3)J151/TM6, T(Y:3)B152/TM6 (breakpoints in 98F), and T(Y:3)In(3R)B172/TM6 (relevant breakpoint in 99A). The lethal progeny of these crosses were scored for the stg cuticle phenotype. Only T(Y:3)In(3R)B172/TM6 did not give stg embryos, indicating stg is proximal to 99A.

I(3)neo62 was reverted by crossing I(3)neo62 r e mwh/TM3 Sb e flies to Δ2–3 Sb/TM6 Ubx e flies that carry a stable transposase source on chromosome 3 (Δ2–3; Robertson et al., 1988). Dysgenic male progeny of the genotype I(3)neo62 r e mwh/Δ2–3 Sb were then mated to I(3)neo62 r e mwh/TM3 Sb e females. Homozygous Sb⁺ r e mwh progeny occurred at a frequency of about 2%–4%. These presumably heterozygous revertants of the genotype I(3)neo62 r e mwh/neo62^R r e mwh were crossed to stg^7B69 e/TM3 Sb e flies to test for stg complementation. Sb⁺ e flies, presumably of the genotype stg^7B69 e/neo62^R r e mwh occured at a frequency of 43%. These were then mated to each other, and homozygous r e mwh progeny were picked and mated to each other to establish the homozygous revertant lines whose DNA was used in Southern analysis (Figure 3).

Cloning and Blot Hybridizations

Plasmid rescue was carried out according to Cooley et al. (1988) using genomic DNA isolated from I(3)neo61/TM3 or I(3)neo62/TM3 flies cut with SaII, EcoRI, or BamHI. Similar flanking fragments were recovered from both strains. Genomic DNA clones were isolated from a λ-dash library (a gift from L. and Y. N. Jan), and cDNA clones were isolated from the L. Kauvar λgt10 library according to standard protocols, with conditions as described in Lehner and O’Farrell (1989). DNA preparations for Southern blots were carried out as in Lehner and O’Farrell (1989). Northern blots were done as described in Edgar et al. (1986b), using total RNA from 100 individually staged embryos per lane, formaldehyde-agarose gels, and Hybond-N nylon filters (Amersham). Probes were made from gel-isolated restriction fragments by the random primer method (Hodgson and Fisk, 1987).

In Situ Hybridizations

Whole-mount hybridizations were done as follows. Embryos were dechorionated in 50% bleach for 2 min, fixed for 20 min in a two-phase mixture of heptane over PBS, 50 mM EGTA, and 10% formaldehyde, and devitellinized by replacing the aqueous fixative with absolute methanol and shaking vigorously. After two rinses in methanol, they were rinsed twice in 95% ethanol and stored in 95% ethanol at −20°C for 1–20 days. The embryos were rehydrated by two rinses in PBS containing 0.05% Triton X-100 (PTx), and acetylated by rinsing for 5 min in 1 × PBS, 0.1 M triethanolamine (pH 7.8), 0.5% acetic anhydride (added just before the embryos). They were then rinsed twice in PTx, and pre-hybridized for 1 hr at 50°C in 50 μl of a solution containing 50% de-ionized formamide, 0.6 M NaCl, 10 mM Tris 7.5, 1 mM EDTA, 1% SDS, 1× Denhardt’s, 10% PEG 8000, 0.03% Triton X-100, 0.25 mg\ml E. coli tRNA. Hybridizations were done for 12 hr at 50°C in 50 μl of the above solution containing 10⁷ cpm of antisense RNA probe. We snythesized antisense RNA probes with T7 polymerase from a cDNA template of 1.3 kb from the 5′ region of the stg message (pC8). Probes were labeled with [³⁵S]UTP or [³H]ATP, GTP, and UTP of the highest specific activity available (Amersham). After probe synthesis, the DNA template was removed with DNAase, and the probes were base-hydrolyzed to a mean length of about 100 bases by incubation in 40 mM NaHCO₃, 60 mM Na₂CO₃ for 90 min at 60°C. After hybridization, the embryos were rinsed three times in PBS, and incubated for 60 min at 37°C in 100 μg/ml RNAase A, 0.5 M NaCl, 10 mM Tris 7.5. They were then washed three times in 1× SSPE, 50% formamide, 0.05% SDS, 0.05% Triton X-100, and three more times in 0.1 × SSPE, 0.05% SDS, 0.05% Triton X-100. Washes were done in 800 μl, for 30 min, at 50°C. Finally, the embryos were rinsed in PTx. Ten to twenty embryos were mounted with even spacing onto a formaldehyde-activated TESPA-coated slide (Berger, 1986) in about 30 μl of PTx. A siliconized 22 mm² cover slip was applied, and as much liquid as possible was wicked away by applying gentle pressure and touching a filter paper to the edges of the cover slip. This flattened the embryos, allowing an even application of the photographic emulsion. The cover slip was then removed by rinsing the slide in a bath of 95% ethanol, and the slide was air dried. We applied 66% Kodak NTB-2 nuclear track emulsion by standard methods, and exposed the slides for 6 days (³⁵S probes) or 5 months (³H probes). After developing the slides, the embryos were stained in 10 μg/ml Hoechst 33258 for 3 min, rinsed in water for 15 min, air dried, and mounted in a mixture of Canada balsam and methyl salycilate that was the consistency of maple syrup.