PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Dev Biol. Author manuscript; available in PMC Jun 1, 2009.
Published in final edited form as:
PMCID: PMC2442716
NIHMSID: NIHMS53354
Control of Cell Cycle Timing during C. elegans Embryogenesis
Zhirong Bao,^*+ Zhongying Zhao,* Thomas J Boyle, John I Murray, and Robert H Waterston
Department of Genome Sciences, University of Washington, Seattle, WA 98195, USA.
^Corresponding author. Email: baoz/at/mskcc.org
*These authors contributed equally to the paper.
+Current address: Developmental Biology Program, Sloan-Kettering Institute, 1275 York Avenue, New York, NY 10065, USA.
As a fundamental process of development, cell proliferation must be coordinated with other processes such as fate differentiation. Through statistical analysis of individual cell cycle lengths of the first eight out of ten rounds of embryonic cell division in C. elegans, we identified synchronous and invariantly ordered divisions that are tightly associated with fate differentiation. Our results suggest a three-tier model for fate control of cell cycle pace: the primary control of cell cycle pace is established by lineage and the founder cell fate, then fine-tuned by tissue and organ differentiation within each lineage, then further modified by individualization of cells as they acquire unique morphological and physiological roles in the variant body plan. We then set out to identify the pace-setting mechanisms in different fates. Our results suggest that ubiquitin-mediated degradation of CDC-25.1 is a rate-determining step for the E (gut) and P3 (muscle and germline) lineages but not others, even though CDC-25.1 and its apparent decay have been detected in all lineages. Our results demonstrate the power of C. elegans embryogenesis as a model to dissect the interaction between differentiation and proliferation, and an effective approach combining genetic and statistical analysis at single-cell resolution.
Keywords: statistics, single cell, fate differentiation, cdc25, Skp1-related
Cell cycle control is an essential aspect of metazoan development. Proliferation must be coupled to differentiation and controlled by fate to coordinate the size and shape of different tissues and organs. In many organisms, cells exit the cell cycle upon terminal differentiation. Disruption of this exit, or inappropriate reentry into the cell cycle can lead to cancer. More subtly, the pace of each cell cycle needs to be controlled to allow appropriate transient cell-cell interactions important for fate induction and morphogenesis. However, the mechanisms of developmental control of cell cycle pace are still poorly understood.
A known mechanism for fate to control the pace of cell cycle is through cell type-specific transcription of rate-limiting components of the cell cycle machinery, such as cdc25/String or cyclin E in Drosophila (Jones et al., 2000; Lehman et al., 1999) and cyclin D1 in mammals (Shtutman et al., 1999; Tetsu and McCormick, 1999). Timing of such transcription in turn sets the length of the gap phase (G1 and G2) and consequently the pace of cell cycle.
In many organisms that deploy rapid embryogenesis such as insects (Lee and Orr-Weaver, 2003), fish (Zamir et al., 1997) and amphibians (Tikhmyanova and Coleman, 2003), the early embryonic cell cycles present a different problem: these cycles oscillate between DNA replication (S phase) and mitosis (M phases) and are supported by a maternal supply of cell cycle regulators. These cycles have been thought of as rapid, synchronous division for the purpose of rapid proliferation. However, in Drosophila, more careful measurements show differences in cell cycle length (Ji et al., 2004). Developmental regulation of non-gap phases is also seen in other organisms. For example, in mouse embryo, M phase takes about 109 minutes in the first division and 70 minutes in the second (Ciemerych et al., 1999).
C. elegans provides a promising model to study developmental control of cell cycle pace. C. elegans has an invariant cell lineage and a fixed fate map (Sulston et al., 1983). That is, each cell has a known, unique fate, greatly simplifying the developmental context in which cell cycle occurs. Leveraging on this clarity and the reproducible cell cycle length of the early founder cells, Brauchle et al studied the asynchrony of cell division occurring at the two-cell stage (Brauchle et al., 2003). At this stage, the anterior cell AB, which gives rise to the majority of the soma, divides about 2 minutes earlier than its posterior sister P1, which gives rise to the germline and the rest of the somatic cells. These cells as well as the other founder cells oscillate between the S and M phases. Brauchle et al’s results show that the asynchrony between AB and P1 requires differential activation of a DNA replication checkpoint. A recent whole-genome RNAi screen revealed 13 more genes involved in this process (Sonnichsen et al., 2005).
An intriguing open question is whether later development of C. elegans provides similar opportunities to dissect how the many conserved differentiation pathways interact with cell cycle control. Homeotic fate transformation in C. elegans is frequently accompanied by the transformation of division pattern and cell cycle lengths (Draper et al., 1996; Kaletta et al., 1997; Maduro et al., 2005). While such observations suggest that the pace of cell cycle is tightly regulated by fate, others caution that individual cell cycle lengths might be too variable to be a meaningful consequence of fate (Schnabel et al., 1997).
We sought to address the uncertainty regarding fate differentiation and cell cycle pace during C. elegans embryogenesis. Exploiting our automated lineaging system (Bao et al., 2006), we examined cell division timing and cell cycle length through the first eight (out of ten) rounds of embryonic cell division. Our results show that through the first eight rounds, division timing and cell cycle lengths are reproducible, providing the statistical basis for assaying cell cycle pace. Furthermore, the quantitative measurements of variability allowed us to discover various patterns of synchronous as well as ordered cell divisions that are tightly associated with fate differentiation within all founder lineages, revealing the biological traits that can be used to study how different fates control cell cycle. More importantly, the patterns suggest that fate controls cell cycle pace at three levels, first by lineage, then by tissue- and organ-type, then by the unique identity of individual cells, providing a simplifying framework to understand fate control of cell cycle. Finally, we present genetic evidence that through the first eight rounds of division, ubiquitin-mediated degradation of cdc-25.1 is used to set the wild type pace of the E (gut) and P3 (muscle and germline) lineages, but not others. Our results suggest that indeed C. elegans provides ample opportunities to study how the diverse differentiation pathways control cell cycle, many of which are conserved in other organisms, and lay some of the necessary groundwork to fully exploit C. elegans as a model for developmental cell cycle control.
Worm strains and culture
Worms were cultured with standard procedure (Brenner, 1974) and were well fed for at least two generations before embryos were collected from young adults and imaged at room temperature (21 to 23 °C).
To facilitate lineage analysis, all worms analyzed carry an integrated transgene, a protein fusion of his-72 and GFP (Ooi et al., 2006). For the analysis of cdc-25.1 and skr-1/2 function, the worms also carries a second integrated transgene, which is a his-24::mCherry driven by a pha-4 promoter (3.5 Kb upstream of the first exon). The mCherry is highly expressed in gut cells. Occasionally, it is also weakly expressed in pharyngeal precursor cells. Because the gut mCherry signal is about two orders of magnitude higher than that in the pharynx, the high expression of this reporter specifically marks the gut. The ij48 allele of cdc-25.1 was obtained from the Caenorhabditis Genetics Center (CGC) as strain IA123. A deletion allele of skr-1/2, ok1938, was provided by the C. elegans Gene Knockout Consortium via CGC.
For RNAi of skr-1/2, we used clones in the Ahringer library, following their feeding protocol (Kamath et al., 2003). We transferred L4 worms to RNAi plates, and waited for 36 to 48 hours before collecting embryos. This time window allowed us to avoid sterility and get consistent RNAi phenotype.
Lineage analysis
Embryos were imaged and lineaged as described in (Murray et al., 2006). Without convenient markers of the cell membrane, cell division was called based on the mitotic figures. Hence the cell cycles measured are from anaphase or telophase to anaphase or telophase. The time of division of ABa in each embryo is arbitrarily defined as 27 minutes after first cleavage. This time is estimated from 3 embryos where the cell cycle of AB is ~14 minutes and ABa ~13 minutes. We analyzed 20 embryos for the wild type, 3 for cdc-25.1 (ij48), 4 for skr-1/2 RNAi and 5 for skr-1/2 RNAi and cdc-25.1 (ij48) combined.
Statistical analysis
Mean and standard deviation were calculated according to their mathematical definition. Linear correlations were fitted with either R or gnuplot. To estimate the probability of a given cell cycle length being wild type, we assumed that the wild type distribution estimated from the 20 embryos is Gaussian. We then calculated the probability based on the z score. To test the significance of observed asynchrony of division in the AB lineage, we used random simulation to estimate the distribution of pairwise difference in division timing if all observed differences are due to stochastic variation. For a given round of AB division, we randomly sample two divisions from each of the 20 embryos and calculated the mean difference. We did this for 105 times in order to estimated the distribution of pairwise difference in division timing. Our estimated distribution for the 6th round is −0.0012 ± 0.4639 min, and for the 7th, −0.0024 ± 0.9761. The significance of real-life differences was then estimated based on these distributions using z scores.
Statistical variability of division timing and cell cycle lengths
To test the statistical validity of quantifying the pace of cell cycle in C. elegans embryogenesis, we first examined wild type variability of cell division timing and cell cycle lengths. To this end, we lineaged 20 embryos at one-minute interval through the first eight rounds of embryonic cell division (Bao et al., 2006) under controlled conditions see (Methods).
As shown in Figure 1, division timing correlates linearly between embryos. This suggests that each embryo has a general clock, or speed of development, and each cell adjusts proportionally according to the general clock. It also suggests that there are two layers of variability: the variation of the general clock from embryo to embryo and the deviation of individual cells of an embryo from its general clock.
Figure 1
Figure 1
Correlation of division timing of individual cells between embryos. Plotted are the median vs the slowest or the fastest developing embryo among the 20 embryos examined.
To quantify the two layers of variation, we estimated the idealized embryogenesis under our experimental conditions by calculating the average time of each division. We then examined the correlation between individual embryos and the ideal embryo. The slope of such correlations, or more specifically, the inverse of the slope 1/k, reflects the clock of an embryo, and the correlation coefficient r2 reflects the intrinsic variability of the cells of the embryo. Based on 20 embryos, we estimated that the developmental clock has a relative standard deviation of 4.5%. Presumably, this variation is due to the small variation of experimental conditions such as temperature. The r2 ranges from 0.997 to 0.999. On average, 95% of the cell divisions of an embryo deviate less than 2% from its general clock. We observed no correlation between an embryo’s clock and its intrinsic variability, suggesting that this intrinsic variability is less sensitive or not affected by the experimental conditions (within the range of our settings).
We further analyzed each cell cycle (Figure 2, Supplemental Table 1). Among the first eight rounds of division, cell cycle length ranges from ~15 to ~60 minutes, and the standard deviation from ~1 to ~6 minutes. The standard deviation is roughly proportional to cell cycle length (Supplemental Figure 1), with the average relative standard deviation being 7%. Among the cells analyzed, P4, the germline precursor, has an obviously higher variability (Figure 2). However, this may only reflect the fact that P4 has a longer cycle length compared to its contemporaries rather than a looser control of cell cycle (Supplemental Figure 1).
Figure 2
Figure 2
The lineage with vertical lines proportional to the average cell cycle length and the magenta bars to the standard deviation.
Thus, our results suggest that the natural variation of C. elegans embryogenesis is small under room temperature and other normal experimental conditions, both in terms of overall development and individual cell cycles. This small variability suggests that precise measurements of cell cycle length and division timing may provide useful quantitative phenotypes to study the regulation of cell cycle pace during development
Lineal patterns of cell division and founder cell fates
The initial stage of C. elegans embryogenesis involves the generation of the founder cells, each of which gives rise to a lineage with different number and types of cells (Figure 2). The lineage tree shows that cell cycle lengths also show a lineage-based pattern: different founder lineages have different rhythms of proliferation, punctuated by rounds of cell divisions. It was therefore proposed that each lineage has its own endogenous and autonomous clock (Deppe et al., 1978).
We found that these lineal patterns of cell division have a simple property: over the successive rounds of division within a founder lineage, cell cycle lengths follow a geometric sequence. That is, the ratio between the cell cycle lengths of a given round of division and the previous round is a constant (Figure 3).
Figure 3
Figure 3
Cell cycle length follows a geometric sequence. Plotted are correlations of average cell cycle length vs average time of division of each round of division in the C and MS lineages. By definition, time of division is the sum of the cell cycle lengths (more ...)
Mathematically, the “shape” of a geometric sequence is determined by two factors: the value of the first element (cell cycle length of the founder cell) and the common ratio between consecutive elements (ratio between the lengths of consecutive cell cycles). It is known that the founder cells have different and reproducible cell cycle lengths, resulting in an invariant order of division (Deppe et al., 1978; Sulston et al., 1983). We further found that the common ratio, or the rate at which cell cycles decelerate, is also different from lineage to lineage (Figure 3). Hence, the two combined explains the different lineal patterns, or Deppe et al’s lineal clock.
Asynchrony of cell division and further differentiation of cell fate
The rounds of cell division within each founder lineage are often described as synchronous, with no consensus in the field whether the subtle differences in division timing within each round are stochastic variation or biologically meaningful. While stochastic variation is a major factor, we found statistically significant patterns of synchrony and asynchrony (invariantly ordered divisions) in all lineages. Furthermore, these synchrony and order are tightly coupled to fate differentiation of cells, suggesting that the lineal control of division timing is modified, first by tissue and organ specification and then by individualization of cells of the same tissue or organ for their unique morphological and physiological roles in the invariant body plan.
The C lineage provides examples of both layers of fate control (Figure 4). This lineage produces bilaterally symmetric groups of hypodermal and muscle cells. Comparison of division timing and cell cycle lengths within each round of C division shows three features regarding fate and cell cycle:
Figure 4
Figure 4
Synchrony and asynchrony of cell division in the C lineage and their association with fate differentiation. Horizontal lines in the tree mark the time of division. Numbers are time of division / cell cycle length. l1/r1 etc. are bilateral pairs, and * (more ...)
First, within each round of division, the hypodermal precursors divide faster than the muscle precursors (see also Supplemental Figure 2). Table 1 further compares the division of the C granddaughters (i.e., C4, which means that C has divided into four cells (Sulston et al., 1983)). The Table shows that the small differences are statistically meaningful. Specifically, for all possible pairs of cells we calculated the difference between their time of division (mean and standard deviation over 20 embryos). Across tissue types, the differences are statistically non-zero (p < 3×10−3). In all 20 embryos examined, the hypodermal precursors divide before the muscle precursors. In contrast, between cells of the same tissue, the difference is not significantly deviated from zero (p < 0.31). Among the 20 embryos, there is no consistent order of division within the same tissue.
Table 1
Table 1
Differences of Division Time Between C4 Cells
Second, bilaterally symmetric pairs, which have essentially the same fate, have the most similar time of division and cycle length compared to others cells of the same round (l1/r1 etc). Caap and Cpap (stars in Figure 4) break the fate symmetry within the C lineage: Cpap produces four hypodermal cells as the other hypodermal precursors (Caaa and Cpaa); Caap however produces one hypodermal cell, two neurons and a cell death. Caap and Cpapa also have different cell cycle lengths.
Third, the four muscle precursors of the C8 round fall in two groups, Capa-Cppa and Capp-Cppp, in terms of timing. Because each group is a bilaterally symmetric pair, we hypothesize that the difference between the two groups is due to cells acquiring individual identity within a tissue type.
Like C, the other lineages also show coupling of fate differentiation and cell cycle changes. In the MS lineage, pharyngeal precursors divide faster than muscle precursors (Supplemntal Figure 2). In the D and E lineages, each of which gives rise to a single tissue (muscle for D and gut for E), cells exhibit individuality. These lineages also show strong symmetry of timing between bilaterally symmetric cells (Table 2).
Table 2
Table 2
Comparison of Division Time of Bilateral Pairs
Across founder lineages, cells of the same tissue or organ do not converge to the same cell cycle length (Supplemental Figure 2). Thus, differentiation only modifies the cell cycle lengths set up by the founder cell fate, rather than creating tissue- or organ-specific cell cycles.
The AB lineage has a more complex fate map than the lineages discussed above. Cells of different tissues and organs are intermingled. Sublineages do not become single-fate until after the 5th round of division, while some remain multi-fate till the terminal division. Intriguingly, asynchrony also becomes statistically significant after the 5th round. During the 6th round, the fastest dividing cell (ABarpaa) on average divides 3.8±1.1 minutes earlier than the slowest (ABalaap), which has a different fate (p < 10−12, see Methods). Asynchrony becomes more significant the next round, with the largest difference of division timing being 11.1±2.3 minutes (p < 10−18). Thus we suggest that fate also changes cell cycle in the AB lineage, even though fate does not simply equate with tissue or organ type as in the other lineages.
Lineage-specific regulation of cell cycle by cdc-25.1
The statistical analysis above revealed a three-tier regulation of cell cycle by fate differentiation during C. elegans embryogenesis. This model is consistent with previous observations that homeotic fate transformation can lead to the transformation of division pattern and cell cycle lengths (Draper et al., 1996; Kaletta et al., 1997; Maduro et al., 2005). With this model, we started to investigate the molecular mechanisms by which the diverse developmental context of the cells regulates their cell cycle machinery. Specifically, we studied the cell-specific functions of a known cell cycle regulator, cdc-25.1.
Cdc-25.1 is one of the four homologs of cdc25/string in C. elegans (Ashcroft et al., 1999). CDC25 is a phosphatase that positively regulates cell cycle progression by promoting the G1-to-S and G2-to-M transition. CDC-25.1 is provided maternally in all early cells, and its abundance decreases over time based on antibody staining. Despite its ubiquitous presence, two gain-of-function (gf) mutations have been isolated for cdc-25.1 that cause the shortening of the E lineage (gut lineage) cell cycles as well as hyperplasia of the gut with other parts of the worm seemingly normal based on gross morphology, the number of pharyngeal cells and the wild type-looking D lineage (Clucas et al., 2002; Kostic and Roy, 2002). Both mutations affect a conserved protein sequence motif near a substrate recognition motif for the E3 ubiquitin ligase (Hebeisen and Roy, personal communication. See also below), suggesting that the gain of function of these mutations could be due to reduced degradation of CDC-25.1. Indeed, CDC-25.1 persists longer than the wild type in at least one of the alleles (Kostic and Roy, 2002).
With the systematic statistics on cell cycles of the wild type, we reexamined the phenotype of cdc-25.1 using allele ij48, a gain of function allele (Clucas et al., 2002) (Figure 5). We observed hyperplasia of the E lineage in ij48 homozygous animals as previously described (Supplemental Figure 3). Further analysis of individual cells revealed an important detail: the cycles of E and its daughters (E2) are not affected while the cycles of E4, E8 and E16 are shortened (Figure 6). Hence, gain of function of cdc-25.1 causes accelerated cell cycle in E4 and beyond and delayed exit of cell cycle in E16.
Figure 5
Figure 5
Correlation of division timing in a cdc-25.1 (ij48) embryo and the wild type average (over 20 embryos).
Figure 6
Figure 6
Comparison of wild type cell cycle length in the E and P3 lineages with those in cdc-25.1 (ij48), skr-1/2 RNAi and cdc-25.1 (ij48)+skr-1/2 RNAi. The branch lengths of the underlying lineage tree are proportional to the wild type average cycle lengths. (more ...)
Surprisingly, systematic analysis of all lineages showed that the phenotype of ij48 is not E-specific (Figure 5). Contrary to the previous report, ij48 affects the D lineage. Furthermore, it also affects P4, the germline precursor and sister of D. The cycles of D and P4 are shortened (Figure 6), consistent with the function of cdc-25.1 as a positive regulator of cell cycle. The cycles of the D2 and D4 are, however, lengthened. Interestingly, the lengthening of D2 and D4 cycles compensates the shortening of the D cycle so that their division times are not significantly different from the wild type (Figure 5). Among these changes, the shortening of the D cycle is the least significant (p < 10−5, n = 3). Finally, the phenotypes in the D lineage and P4 are unlikely the result of fate transformation as the D cells and the daughters of P4 (Z2 and Z3) occupy normal positions after gastrulation and cdc-25.1 (ij48) animals are fertile.
Potential role of cdc-25.1 degradation in its lineage-specific function
To identify genes interacting with cdc-25.1 in the E, D and P4 lineages, we analyzed the function of skr-1 and skr-2. Skr-1/2 are two of the 21 homologs of Skp1 in C. elegans (Nayak et al., 2002). Skp1 is a conserved component of the SCF (Skp1-Cullin-F-box) complex, which is an E3 ubiquitin ligase. As a general machinery for protein degradation, SCF has broad functions and is known to target over a dozen cell cycle regulators including CDC25 (Nakayama and Nakayama, 2006).
Among the 21 Skp1 homologs, skr-1 and skr-2 were shown to be required for embryogenesis (Nayak et al., 2002). RNAi of skr-1 and skr-2 leads to embryonic lethality. (Because skr-1 and skr-2 are recently duplicated genes, RNAi against either could affect the other. Therefore, skr-1/2 is typically used to refer to both.) Specifically, embryos die with nearly twice as many cells as the wild type, suggesting that skr-1/2 are broadly required for embryonic cell cycle exit.
Skr-1/2 RNAi leads to variable phenotypes in initial embryogenesis. Some show prolonged mitosis, with the M phase increases from ~3 minutes to ~7 minutes. Some show defects in the separation of sister nuclei (the crossed-eye phenotype, (Sonnichsen et al., 2005)). A deletion allele, ok1938, leads to more severe phenotypes, with clusters of non-separated nuclei. To avoid potential complications by the early defects, we used RNAi and analyzed embryos with essentially normal development up to the 4-cell stage. Presumably these represent embryos with the weaker loss of function of skr-1/2. We observed no difference between RNAi of skr-1 and RNAi of skr-2(data not shown).
We found that skr-1/2 RNAi phenocopies cdc-25.1 (ij48) in the E lineage (Figure 6). Like cdc-25.1 (ij48), skr-1/2 RNAi shortens the cycle of E4, E8 and E16, but does not affect that of E and E2. It also causes extra cell divisions in the E lineage (Supplemental Figure 3), although excessive division is not restricted to the E lineage. Furthermore, skr-1/2 RNAi enhances the phenotype of cdc-25.1 (ij48) by further shortening the cycle of E4, E8 and E16. The average E2 cycle lengths are within the wild type range, but exhibit greater variability. E is still not affected.
These changes in the gut lineage are not due to fate transformation. The E cells express a pha-4 reporter (Supplemental Figure 3, see also Methods) and accumulate gut granules as in wild type. They also gastrulate to the interior of the embryo and occupy normal gut position, a process that depends on proper differentiation of the gut fate (Kaletta et al., 1997). Because the E lineage follows its normal fate, the phenotypes suggest that skr-1/2 is used by the wild type E lineage to regulate its pace of cell cycle, possibly by degrading CDC-25.1.
While the E lineage maintains its normal gut fate, not all apparent gut cells (based on marker expression and cell position, Supplemental Figure 3) in skr-1/2 RNAi-treated embryos are derived from E. Lineage analysis showed that in some embryos, the MSp and Cp sublineages are transformed to gut fate, generating >60 apparent gut cells (data not shown).
As in cdc-25.1 (ij48), skr-1/2 RNAi also shortens the cycle length of the D cell and P4 (Figure 6). Skr-1/2 RNAi and cdc-25.1 (ij48) combined further shortens the cycle of these cells. However, skr-1/2 RNAi does not show the compensatory lengthening of the D2 and D4 cycles. The D2 cycles are shortened. The D4 cycles are within the range of the wild type. In the wild type, Dxa divide about 5 minutes after their sisters, Dxp. In skr-1/2 RNAi, this asynchrony is lost. Skr-1/2 RNAi in the cdc-25.1 (ij48) background shows similar results. Hence, skr-1/2 is required for both the invariant order of division among D4 and the compensatory lengthening of the D2 and D4 cycles caused by cdc-25.1 (ij48).
Interestingly, skr-1/2 RNAi also shortens the cycle of P3, mother of D and P4. While cdc-25.1 (ij48) alone does not cause statistically significant changes of the P3 cycle, it slightly enhances the shortening by skr-1/2 RNAi, suggesting that cdc-25.1 also regulates P3. Finally, skr-1/2 RNAi causes the daughters of P4 (Z2 and Z3) to divide, which normally exit embryonic cell cycle.
Reproducibility of cell cycle
C. elegans is renowned for its invariant cell lineage and fixed fate map. We further showed that the temporal control of cell division during C. elegans embryogenesis is statistically reproducible. Our data suggest two sources of variability: variation of the global developmental clock of each embryo, which is presumably sensitive to the variation of experimental conditions, and the variation of individual cell cycles from the embryo’s global clock, which appears to be intrinsic and insensitive to conditions. The two combined result in ~7% relative standard deviation in the raw measurements of individual cell cycles. A tighter control of the temperature would likely to further reduce the variance (by reducing the variation of the developmental clock).
Our systematic estimates are consistent with the ad hoc observations of timing in the literature. For example, while Sulston claimed no statistical significance of his measurement of timing, we found that many of the invariantly ordered cell divisions we encountered are reflected in his lineage drawing (Sulston et al., 1983). Deppe et al also noticed the asynchrony among the great granddaughters of C (C8) (Deppe et al., 1978). Their analysis was prior to the knowledge of the complete lineage and the fate map, hence they were not able to associate the asynchrony with fate differences. Based on their lineage drawing, it appears that their naming of the C8 cells is reversed compared to the later standard that Sulston established (eg, the canonical Caaa is named as Caap by Deppe).
We also compared with measurements by Schnabel et al (Schnabel et al., 1997). While we focused on the first eight rounds of embryonic division, Schnabel et al focused on the last two rounds (the 9th and 10th) in certain sublineages. Despite the difference in stages, their results show a similar level of variability as ours when translated into comparable terms (see Supplemental Materials for detailed comparisons). Interestingly, with a comparable level of observed variability, they raised the concern that individual cell cycle lengths might be too variable to be meaningful, while we came to the conclusion that there are statistically significant patterns of synchrony and asynchrony. It is possible that these patterns break down in the last two rounds of division, where they focused. It is also possible that they focused on comparing individual embryos while we focused on statistical patterns and trends. For example, while a bilaterally symmetric pair on average divides at the same time, they do not divide precisely in sync in all embryos. Such statistical patterns may not be apparent from comparing individual embryos, especially not from comparing the extreme cases.
Fate control of cell cycle
Our analysis suggests that the primary control of cell cycle timing during C. elegans embryogenesis is laid by lineage and founder cell fate, which is then modified by tissue and organ differentiation and then by individualization of cells. The three-tier control is in accordance with our knowledge of fate differentiation during C. elegans embryogenesis. The initial differentiation is the specification of the founder cells, with master fate switches function in lineage-based patterns (Baugh et al., 2005; Bowerman et al., 1992; Draper et al., 1996; Mello et al., 1992). The next is the specification of tissues and organs, with a cohort of transcription factors being expressed in cells of the same tissue and organ regardless of their lineage identity (Andachi, 2004; Fukushige et al., 2006; Gilleard et al., 1999; Hallam et al., 2000; Labouesse et al., 1994; Maduro et al., 2005; Page et al., 1997). The final stage is the individualization of cells of the same tissue or organ for their unique morphological and physiological roles in the invariant body plan (Costa et al., 1988; Fukushige et al., 2005). Thus, it appears that each step of fate differentiation leaves its mark on cell cycle.
However, there is a subtle difference between fate differentiation and cell cycle control: while tissue and organ differentiation and cell individualization take over fate differentiation from lineage-based master switches, they only modify the geometric sequences of cell cycle lengths set up by the founder cell fate. Thus, it suggests that the three tiers of cell cycle control are additive.
The geometric sequences describing lineal patterns of C. elegans embryonic cell cycles are intriguingly simple, and suggest that some simple factors may play a critical role in determining the cell cycle lengths. One candidate is cell volume, which follows a geometric sequence over division. Previous reports on the effect of cell volume on cell cycle length are however contradictory and hence inconclusive (Brauchle et al., 2003; Schierenberg, 1984). Another hypothesis that invokes simple mechanisms is the consumption of energy or materials. If the rate of consumption and the rate of cell cycle progression are both proportional of the concentration of the rate-limiting substance, the pace of cell cycle would decelerate at a constant rate, i.e., cell cycle lengths would follow a geometric sequence. However, when the E lineage cell cycle is disturbed by the cdc-25.1 (ij48) of skr-1/2 RNAi, it no longer follows a geometric sequence. Thus, it suggests that the simple mathematic feature may not be a direct derivative of a simple factor but may instead involve complex mechanisms. It is not clear to us at the moment if the geometric sequence is a pure coincidence or has indeed implications about cell cycle regulation and physiological constraints on cell cycle.
Another general question concerning cell cycle regulation is autonomy. Because of the invariant lineage, C. elegans development was initially thought to be deterministic and autonomous. Work in the last two decades however revealed complex cell-cell signaling during fate differentiation. Regulation of cell cycle on the other hand, is still assumed as autonomous (Schnabel et al., 1997). With fate regulating cell cycle, the simple model becomes questionable. One may argue that if signaling regulates fate and fate regulates cell cycle, then by definition cell cycle regulation is not autonomous. However, perhaps a more meaningful question is whether there are signals that directly regulate cell cycle independent of the fate signals. In cdc-25.1 (ij48), the cycles of D2 and D4 cells are lengthened suffieciently to compensate the shortening of the D cell cycle, implying regulation. Besides intercellular signaling, vertical signals from mother to daughters could also lead to compensation, which, in Drosophila, involves negative feedback via cyclin-dependent kinases to E2F, ultimately affecting cyclin E and cdc25/string (Reis and Edgar, 2004).
Fate-specific regulation by general cell cycle regulators
Measuring cell cycle length with single cell resolution allowed a deeper understanding of cdc-25.1 function. In particular, we found that the gain of function mutation ij48 affects not only the E lineage, but also D and P4. Within the E lineage, we found that it does not affect E and E2 but shortens later cell cycles.
Our results suggest that both cdc-25.1 and skr-1/2 are used in the E lineage to set the proper cell cycle length. Our analysis further shows additivity between cdc-25.1 (ij48) and skr-1/2 RNAi. Because we analyzed the weaker partial loss of function by skr-1/2 RNAi, the additivity suggests that cdc-25.1 and skr-1/2 could act in either the same pathway or parallel pathways. However, other evidence suggests that skr-1/2 directly regulates cdc-25.1 through protein degradation. As mentioned above, the gain of function mutations of cdc-25.1 affects a conserved protein motif near a substrate recognition motif for the F-box protein β-TrCP. The ortholog of β-TrCP in C. elegans is lin-23. Interestingly, Hebeisen and Roy (personal communication) showed that loss of function of lin-23 also phenocopies cdc-25.1 gain of function, and that lin-23 loss of function disrupts the degradation of CDC-25.1. Both skr-1/2 and lin-23 are maternal and present in all embryonic cells. The two shows similar phenotypes in not only the E lineage but also in the general exit of embryonic cell cycles. Thus, skr-1/2 and lin-23 are likely functional partners during embryogenesis. Combined, our results and Hebeisen and Roy’s suggest that the E lineage employs SCF for ubiquitin-mediated degradation of CDC-25.1 to lengthen its cell cycle during normal development.
The altered pace of cell division in the P3 lineage suggests that SCF-mediated degradation of CDC-25.1 is also required for proper regulation of cell division timing in this lineage. However the function of cdc-25.1 and skr-1/2 appears to be more complex in the P3 lineage, as seen in the compensatory lengthening of the D2 and D4 cycles.
Cdc-25.1 is a general cell cycle regulator. It is maternal and present in all cells. Antibody staining shows that CDC-25.1 is degraded over development in all lineages (Kostic and Roy, 2002). Thus, it remains an open question why this degradation process exerts lineage-specific effects. Similarly, it is also unclear why within the E lineage E and E2 behave differently than E4 and later cells.
The other side of the lineage-specificity of SCF-mediated degradation of CDC-25.1 is how the other lineages pace their cell cycle. Or more broadly, what are the possible ways through which the cell cycle machinery interfaces with differentiation pathways. In this regard, it is interesting to consider the fact that Drosophila embryogenesis also exploits cdc25/string for tissue specific regulation of cell cycle. However, unlike in C. elegans where degradation of maternal CDC25 is used, in Drosophila cdc25/string bears transcription enhancers that are recognized by tissue-specific transcription factors. This difference, as well as the difference between the lineages within C. elegans, suggests the diversity and evolutionary flexibility of the interfacing between cell cycle and fate differentiation.
New opportunities for studying developmental cell cycle control
In order to study developmental control of cell cycle, one needs the ability to track cells from division to division during development and the knowledge about the differentiation of the cells being tracked. With the invariant lineage and the recently developed automated lineaging system that tracks every cell at every minute during embryogenesis, C. elegans offers great advantage on both aspects.
The systematic analysis of cell cycle lengths strengthens C. elegans as an emerging model to study developmental cell cycle control (Fay, 2005; Kipreos, 2005; van den Heuvel, 2005). The pace of cell cycle is intrinsically a quantitative question. The evaluation of the statistical reproducibility of cell cycle length and division timing establishes the meaningfulness to measure them as quantitative traits. The analysis of the functional specificity of cdc-25.1 further demonstrates the effectiveness of such systematic quantitative analysis at single-cell resolution, both in terms of assaying individual cell cycles and studying potential complications by fate changes.
The patterns of synchronous and invariantly ordered cell divisions provide the biological traits that can be used to experimentally dissect the molecular mechanism of fate control of proliferation. The tight coupling of fate and cell cycle throughout the lineage, as suggested by the patterns, provides opportunities to study the role of numerous differentiation pathways in cell cycle control, many of which are conserved in other organisms. A question worth noting is whether and how the differentiation pathways control the introduction of gap phase during embryogenesis, a phenomenon that is common to many animals. In this regard, further improvement of the imaging techniques would be needed to acquire more information on cell cycle. Using histone-GFP fusions, we already capture the progression of the M phase by the mitotic figures. With S phase-specific markers, one would be able to delineate the four phases of cell cycle (G1, S, G2 and M) and further determine at which stage differentiation pathways act.
Our data suggest that fate differentiation creates small but reproducible differences of division timing in cells that would otherwise be synchronous. Thus, besides the molecular connection between differentiation and cell cycle, C. elegans also allows us to address broader questions, such as how cells achieve such accurate control, and how temporal control of development can be so robust yet so sensitive to the input of different events.
Supplementary Material
03
04
05
Acknowledgment
We thank Drs. Laura Buttitta and James Thomas for their discussions and critical reading of the manuscript. We are also grateful to Michael Hebeisen and Dr. Richard Roy for communicating their results before publication. The study was partly funded by NIH and a Damon Runyon Cancer Research Foundation Fellowship to ZB (DR 1813-04). Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). A deletion allele was provided by the C. elegans Gene Knockout Consortium.
Footnotes
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
  • Andachi Y. Caenorhabditis elegans T-box genes tbx-9 and tbx-8 are required for formation of hypodermis and body-wall muscle in embryogenesis. Genes Cells. 2004;9:331–344. [PubMed]
  • Ashcroft NR, et al. RNA-Mediated interference of a cdc25 homolog in Caenorhabditis elegans results in defects in the embryonic cortical membrane, meiosis, and mitosis. Dev Biol. 1999;206:15–32. [PubMed]
  • Bao Z, et al. Automated cell lineage tracing in Caenorhabditis elegans. Proc Natl Acad Sci U S A. 2006;103:2707–2712. [PubMed]
  • Baugh LR, et al. The homeodomain protein PAL-1 specifies a lineage-specific regulatory network in the C. elegans embryo. Development. 2005;132:1843–1854. [PubMed]
  • Bowerman B, et al. skn-1, a maternally expressed gene required to specify the fate of ventral blastomeres in the early C. elegans embryo. Cell. 1992;68:1061–1075. [PubMed]
  • Brauchle M, et al. Differential activation of the DNA replication checkpoint contributes to asynchrony of cell division in C. elegans embryos. Curr Biol. 2003;13:819–827. [PubMed]
  • Brenner S. The genetics of Caenorhabditis elegans. Genetics. 1974;77:71–94. [PubMed]
  • Ciemerych MA, et al. Control of duration of the first two mitoses in a mouse embryo. Zygote. 1999;7:293–300. [PubMed]
  • Clucas C, et al. Oncogenic potential of a C.elegans cdc25 gene is demonstrated by a gain-of-function allele. Embo J. 2002;21:665–674. [PubMed]
  • Costa M, et al. Posterior pattern formation in C. elegans involves position-specific expression of a gene containing a homeobox. Cell. 1988;55:747–756. [PubMed]
  • Deppe U, et al. Cell lineages of the embryo of the nematode Caenorhabditis elegans. Proc Natl Acad Sci U S A. 1978;75:376–380. [PubMed]
  • Draper BW, et al. MEX-3 is a KH domain protein that regulates blastomere identity in early C. elegans embryos. Cell. 1996;87:205–216. [PubMed]
  • Fay DS. The cell cycle and development: lessons from C. elegans. Semin Cell Dev Biol. 2005;16:397–406. [PubMed]
  • Fukushige T, et al. Defining the transcriptional redundancy of early bodywall muscle development in C. elegans: evidence for a unified theory of animal muscle development. Genes Dev. 2006;20:3395–3406. [PubMed]
  • Fukushige T, et al. Transcriptional control and patterning of the pho-1 gene, an essential acid phosphatase expressed in the C. elegans intestine. Dev Biol. 2005;279:446–461. [PubMed]
  • Gilleard JS, et al. ELT-3: A Caenorhabditis elegans GATA factor expressed in the embryonic epidermis during morphogenesis. Dev Biol. 1999;208:265–280. [PubMed]
  • Hallam S, et al. The C. elegans NeuroD homolog cnd-1 functions in multiple aspects of motor neuron fate specification. Development. 2000;127:4239–4252. [PubMed]
  • Ji JY, et al. Both cyclin B levels and DNA-replication checkpoint control the early embryonic mitoses in Drosophila. Development. 2004;131:401–411. [PMC free article] [PubMed]
  • Jones L, et al. Tissue-specific regulation of cyclin E transcription during Drosophila melanogaster embryogenesis. Development. 2000;127:4619–4630. [PubMed]
  • Kaletta T, et al. Binary specification of the embryonic lineage in Caenorhabditis elegans. Nature. 1997;390:294–298. [PubMed]
  • Kamath RS, et al. Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature. 2003;421:231–237. [PubMed]
  • Kipreos ET. C. elegans cell cycles: invariance and stem cell divisions. Nat Rev Mol Cell Biol. 2005;6:766–776. [PubMed]
  • Kostic I, Roy R. Organ-specific cell division abnormalities caused by mutation in a general cell cycle regulator in C. elegans. Development. 2002;129:2155–2165. [PubMed]
  • Labouesse M, et al. The Caenorhabditis elegans gene lin-26 is required to specify the fates of hypodermal cells and encodes a presumptive zinc-finger transcription factor. Development. 1994;120:2359–2368. [PubMed]
  • Lee LA, Orr-Weaver TL. Regulation of cell cycles in Drosophila development: intrinsic and extrinsic cues. Annu Rev Genet. 2003;37:545–578. [PubMed]
  • Lehman DA, et al. Cis-regulatory elements of the mitotic regulator, string/Cdc25. Development. 1999;126:1793–1803. [PubMed]
  • Maduro MF, et al. Genetic redundancy in endoderm specification within the genus Caenorhabditis. Dev Biol. 2005;284:509–522. [PubMed]
  • Mello CC, et al. The pie-1 and mex-1 genes and maternal control of blastomere identity in early C. elegans embryos. Cell. 1992;70:163–176. [PubMed]
  • Murray JI, et al. The lineaging of fluorescently-labeled Caenorhabditis elegans embryos with StarryNite and AceTree. Nat Protoc. 2006;1:1468–1476. [PubMed]
  • Nakayama KI, Nakayama K. Ubiquitin ligases: cell-cycle control and cancer. Nat Rev Cancer. 2006;6:369–381. [PubMed]
  • Nayak S, et al. The Caenorhabditis elegans Skp1-related gene family: diverse functions in cell proliferation, morphogenesis, and meiosis. Curr Biol. 2002;12:277–287. [PubMed]
  • Ooi SL, et al. Histone H3.3 variant dynamics in the germline of Caenorhabditis elegans. PLoS Genet. 2006;2:e97. [PMC free article] [PubMed]
  • Page BD, et al. ELT-1, a GATA-like transcription factor, is required for epidermal cell fates in Caenorhabditis elegans embryos. Genes Dev. 1997;11:1651–1661. [PubMed]
  • Reis T, Edgar BA. Negative regulation of dE2F1 by cyclin-dependent kinases controls cell cycle timing. Cell. 2004;117:253–264. [PubMed]
  • Schierenberg E. Altered cell-division rates after laser-induced cell fusion in nematode embryos. Dev Biol. 1984;101:240–245. [PubMed]
  • Schnabel R, et al. Assessing normal embryogenesis in Caenorhabditis elegans using a 4D microscope: variability of development and regional specification. Dev Biol. 1997;184:234–265. [PubMed]
  • Shtutman M, et al. The cyclin D1 gene is a target of the beta-catenin/LEF-1 pathway. Proc Natl Acad Sci U S A. 1999;96:5522–5527. [PubMed]
  • Sonnichsen B, et al. Full-genome RNAi profiling of early embryogenesis in Caenorhabditis elegans. Nature. 2005;434:462–469. [PubMed]
  • Sulston JE, et al. The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev Biol. 1983;100:64–119. [PubMed]
  • Tetsu O, McCormick F. Beta-catenin regulates expression of cyclin D1 in colon carcinoma cells. Nature. 1999;398:422–426. [PubMed]
  • Tikhmyanova N, Coleman TR. Isoform switching of Cdc6 contributes to developmental cell cycle remodeling. Dev Biol. 2003;260:362–375. [PubMed]
  • van den Heuvel S. The C. elegans Research Community, (Ed.) WormBook; 2005. Cell-cycle regulation. http://www.wormbook.org/doi/10.1895/wormbook.1.28.1". [PubMed]
  • Zamir E, et al. Transcription-dependent induction of G1 phase during the zebra fish midblastula transition. Mol Cell Biol. 1997;17:529–536. [PMC free article] [PubMed]