Embryonic Cleavage Cycles: How Is aMouse Like a Fly?

Patrick H. O’Farrell1, Jason Stumpff2 and Tin Tin Su2

The evolutionary advent of uterine support of embry-onic growth in mammals is relatively recent.Nonetheless, striking differences in the earliest stepsof embryogenesis make it difficult to draw parallelseven with other chordates. We suggest that use offertilization as a reference point misaligns the earlieststages and masks parallels that are evident whendevelopment is aligned at conserved stages sur-rounding gastrulation. In externally deposited eggsfrom representatives of all the major phyla, gastrula-tion is preceded by specialized extremely rapid cleav-age cell cycles. Mammals also exhibit remarkably fastcell cycles in close association with gastrulation, butinstead of beginning development with these rapidcycles, the mammalian egg first devotes itself to theproduction of extraembryonic structures. Previousattempts to identify common features of cleavagecycles focused on post-fertilization divisions of themammalian egg. We propose that comparison to therapid peri-gastrulation cycles is more appropriate andsuggest that these cycles are related by evolutionarydescent to the early cleavage stages of embryos suchas those of frog and fly. The deferral of events inmammalian embryogenesis might be due to an evolu-tionary shift in the timing of fertilization.

The demands on frog or Drosophila eggs, which aredeposited in the environment to fend for themselves,are very different from the demands on a mouse egg,which is held in a protective and nutritive environment.Frog and fly eggs need to produce a feeding animalwith the reserves within the egg. This produces acascade of problems and solutions that appears tohave become an integral part in the early develop-mental programs of freely developing organisms [1].The first problem is to produce a whole feeding organ-ism from an egg. The solution is to make eggs espe-cially large cells to provide adequate reserves. Thesecond problem is that the single allotment of DNA inan egg does not have the capacity to rapidly changethe composition of RNAs in the huge cytoplasm of theegg. The solution is to use maternally encoded geneproducts and to quickly amplify the number of nucleito provide a transcriptional output that is adequate forthe developmental events to come.

The mammalian egg is faced with the very differenttask of developing a machinery to take advantage of

the nutritive environment in which it is located. Thus, itdevelops extraembryonic tissues for interaction withthe uterus and, in doing so, defers the events of earlydevelopment. Mammalian eggs lack the massivematernal contributions of freely developing eggs, anddevelopment begins at a more leisurely pace basedlargely on zygotic synthesis of components. There isno obvious reason that the mouse egg would need tohave especially rapid cleavages, except that a mammalmay well rely on developmental programs that evolvedduring the more than 250,000,000 years of metazoanevolution that preceded the appearance of mammals.

Rapid cleavage cycles are found in all major meta-zoan phyla, including chordates. Nonetheless, themammalian embryo begins development with slowdivisions and shows rapid cell cycles only at a laterstage. Because they do not immediately follow fertil-ization, these later rapid cycles in mammals are notordinarily considered homologous to early cleavagecycles of other embryos. Here, we suggest that fertil-ization should not be used to align the developmentalprogram of mammals with that of other organisms.Instead, when the highly conserved events surround-ing gastrulation are aligned, the rapid division cycles ofthe mammalian embryo come into correspondencewith the cleavage cycles of other metazoan embryos.We summarize evidence suggesting that the mam-malian rapid cycles are homologous to the rapid cleav-age cycles of other metazoans.

The alignment of embryonic events that we advocateemphasizes that the post-fertilization events of mam-malian development begin with the generation of a trophoblast, the key contributor to the mammalian pla-centa. This process appears to have no analog in thepost-fertilization events of non-placental vertebrates.We suggest that these steps may have had an evolu-tionary precursor in events that contribute to oogenesisin other species. If one considers the maternal events inthe oocyte lineage and the zygotic events that followfertilization as a continuum, a shift in the timing of fertil-ization with respect to other events occurring in thislineage could shift processes from maternal to zygoticcontrol, or vice versa. We propose that, during the evo-lution of mammals, fertilization was advanced to anearlier stage such that events that occurred late in thecell lineage of the oocyte in the progenitors of mammalswere displaced and modified to become the earliestevents of post-fertilization development in mammals.

Unique Features of Embryonic Cleavage CellCyclesBecause Xenopus and Drosophila are important modelsystems for cell cycle control, we possess detailed infor-mation that allows identification of common features ofthe early mitotic cycles in chordates and arthropods.

First, pre-gastrulation cycles are unusually fast. Afrog egg undergoes six divisions in 3 hours, averagingabout 30 minutes per division cycle [2]. A Drosophila

Review

Current Biology, Vol. 14, R35–R45, January 6, 2004, ©2004 Elsevier Science Ltd. All rights reserved. DOI 10.1016/j.cub.2003.12.022

1Department Biochemistry and Biophysics, GH-S372CGenentech Hall, UCSF, San Francisco, CA 94143-2200, USA.E-mail: ofarrell@cgl.uscf.edu2Department Molecular, Cellular and Developmental Biology,347 UCB, University of Colorado, Boulder, CO 80309-0347.

egg undergoes 13 embryonic cycles in 2 hours with aprogressive lengthening of the cycles from 8.3 to 23minutes [3]. In contrast, proliferating larval tissueshave about an 8 hour cell cycle time [4,5].

Second, embryonic cleavage cycles occur withoutgrowth, so that cells become progressively smaller.Indeed, this is inevitable, as the embryos have nooutside source of nutrition. The progressive reductionin size of cells stands in contrast to most cell cycles,wherein cells grow prior to division to roughly maintainsize constancy during proliferation [1,6,7].

Third, the cleavage cycles appear to lack the gapphases that usually intervene between mitosis and Sphase, and between S phase and mitosis. Whereas thevery first mitotic cycle following fertilization has a shortG2 phase in frogs and perhaps also in Drosophila [8],the subsequent phases have no detectable gap phases[9,10]. In other words, the very short interphases consistexclusively of S phase. Replication of the entire 1.8·108

base pair (bp) genome of Drosophila and the 1.7·109 bpgenome of Xenopus [11] is completed in as little as 3.4and 15 minutes, respectively. This remarkable feat isachieved by the use of many more origins of DNA repli-cation than are active in longer cell cycles [12–15].

Fourth, the embryonic cell cycles of frogs and fliesrely on maternally deposited products and can run inthe absence of transcription of the zygotic genome.Thus, cell cycle transitions and the regulation of thesetransitions are independent of transcription.

Fifth, the early cell cycles of Xenopus andDrosophila appear to lack certain checkpoint controlsthat ordinarily coordinate progression through thevarious cell cycle events. Thus, whereas cells fromdiverse sources (different species, tissues, or cell lines)arrest cell cycle progress when DNA synthesis isblocked, cells progress to mitosis with catastrophicconsequences, when DNA replication is blocked byaphidicolin in Xenopus and Drosophila embryos[16–19]. As a result of these observations, it was ini-tially concluded that the early cycles lacked the check-point controls required to arrest the cells. Newerobservations suggest that some checkpoint mecha-nisms are in place, but in some species are too weakto enforce an arrest [19,20] (see below). The above-described cycles are followed by gastrulation in fly andfrog and slower cell cycle times [3,2]. It has been rea-soned that early cycles rapidly generate the cells thatbecome fodder for gastrulation and creation of thebody layers. Additionally, the exponential increase inthe transcriptional capacity has been suggested to beimportant for the switch to control by zygotic tran-scription, which occurs in parallel with the completionof the early rapid cycles.

Fast Embryonic Cycles Exist in All Major AnimalPhylaThe frog and the fly are model organisms for earlydevelopment in part because they develop quickly.One important question is whether the organization ofthe early cycles is more general. While analyses inother systems are less detailed, key features of theearly cleavage cycles — their speed, near synchrony,and the progressive decline in cell size — are obvious

in descriptive analyses that were pursued widely at theturn of the last century. In his classical book ‘The Cellin Development and Heredity’ [21], E.B. Wilson attrib-utes a generalization that embryogenesis begins with‘a series of rapidly succeeding mitotic divisions, thussplitting up [the egg] into blastomeres or embryoniccells’ to studies of Kölliker and Remak in the mid 19th

century. This early evidence of generality is bolsteredby more recent and detailed investigations of specificrepresentatives of the major phyla:

Annelids: During the first seven stages of embryo-genesis in the leech, Helobdella triserialis, early blas-tomeres contain short cell cycles that lack G1 phases.Following these divisions, primary blast cells cycle,still without a G1 phase, but with a much longer G2phase [22].

Echinoderms: In the sea urchin, Paracentrotuslividus, the first four division cycles are synchronous inall blastomeres and last approximately 30 minuteseach. The mitotic index in these embryos is high duringthe blastula stage (60% of cells are in mitosis in 6 hourold blastulae) and drops dramatically before hatching(11% in hatching blastulas), suggesting a lengtheningof interphase [23]. These events precede gastrulationand the overall pattern is consistent with sea urchinsexhibiting fast cycles before gastrulation.

Nematodes: In C. elegans, cell division patterns arehard-wired, with cells of each lineage dividing withstereotypical timing that is invariant from embryo toembryo [24]. With the exception of the first embryoniccell division, which occurs at about 40 minutes afterfertilization, cell cycles that precede gastrulation last10–30 minutes. Cell cycles lengthen after the onset ofgastrulation, although the extent of the lengtheningvaries between lineages. For example, the first cellcycle after gastrulation ranges from 30–60 minutes inthe C-lineage and from 70–90 minutes in the E-lineage.

Molluscs: In the surf clam Spisula solidissima, thefour mitotic cycles following fertilization last 25–35minutes each [25,26].

When added to the Drosophila and frog models,these examples represent the major phyla in the evo-lutionary tree of the metazoa: Mollusca (clam), Arthro-poda (fruit fly), Annelida (leeches), Echinodermata (Starfish and sea urchin), Chordata (frog) and Nematoda (C.elegans) [27]. It is notable that frogs are not unusualamong the Chordata in having rapid cleavage cycles,which are found broadly among birds, amphibians, fishand ascidians.

The early cycles in chick are notable because of therelatively tight evolutionary connection between birdsand mammals: like mammals, birds develop an amni-otic sac in early development and clear parallels can bedrawn between steps of gastrulation in birds andmammals. The first 22.5 hours of post-fertilization chickdevelopment occur in the oviduct, as the albumin andshell are deposited. When the egg is laid, the blastodischas about 60,000 cells [28] (R. Ivarie, personal commu-nication). This requires at least 15 or 16 cell doublingsin 22.5 hours and, thus, an average doubling time ofless than 1.5 hours. In chick, as in all the characterizedexamples, gastrulation is tightly coupled with the laststages of the rapid early cleavage cycles.

ReviewR36

Integration of Rapid Cell Cycles with EmbryonicPatterningIn Xenopus and Drosophila, the cleavage cycles endwith an abrupt transition that is followed by the onset ofgastrulation. The transition at the end of the cleavagestages, referred to as the midblastula transition (MBT)in frog and the maternal to zygotic transition (MZT) inflies, is dramatic in that numerous fundamental featuresof cell behavior change at this time including onset ofhigh level transcription, initiation of cell movements andintroduction of a gap phase into the cell cycle. It is notclear why so many changes occur in concert at the endof the cleavage divisions, but there are suggestions thatthe consuming investment into cellular replication isincompatible with some aspects of morphogenesis andgene expression. Experimental induction of mitosisduring gastrulation disrupts the morphogenetic move-ments [29–31]. Furthermore, rapid cycles interfere withgene expression and the arrest of rapid early cell cyclescan advance zygotic transcription [32,33]. Presumably,the longer interphase associated with the introductionof gap phases provides time for cytoskeletal changesthat underlie cell movement during gastrulation andopportunity for the relatively time consuming polymer-ization of lengthy transcripts [32].

The dramatic nature of the MBT/MZT has promoteda simple view in which the cell cycle exists in eitherembryonic or adult forms, separated by one majorembryonic transition. However, the cell cycle has manyfaces, and even after the MBT/MZT embryos canexhibit very fast cycles. For example, in Drosophila themesoderm, which invaginates and is internalized in thefirst cell cycle after the MZT, goes through two subse-quent cycles that are about 45 minutes long. The neu-roblasts have a similarly rapid cycle [34,35]. In bothcases, there are no evident gap phases. Thus, duringgastrulation and patterning of the embryo, the shortestcycles are five times longer than the cleavage cycles,but are still more than ten times faster than later mitoticcycles in the larval tissues.

It should also be emphasized that the model organ-isms that we know best are not fully representative.Drosophila exemplifies a very successful late branch ofarthropod evolution, the long-germ-band insects. OtherArthropods exhibit a more basal developmental modethat is more easily related to events in other phyla. Inshort-germ-band insects and crustaceans, only theanterior part of the body plan is patterned at the onsetof gastrulation and a proliferative group of posteriorblast cells supplies cells that build successively moreposterior body regions [36]. Thus, whereas a dramatictransition in cell cycle marks the end of the cleavagecycles, local rapid proliferation remains a feature ofembryos at gastrulation and later.

Are Mammals Exceptions?The early cycles following fertilization of the mammalianegg are not unusually fast and appear to resemblemore canonical cell cycles. In the mouse, the first cellcycle is long – the fertilized egg reaches the 2-cell stageat 1.5 days post-coitum (dpc). The next four cell cyclesaverage about 12 hours each leading to the 32 cell early

blastocyst at 3.5 dpc [37]. Cell cycles from the earlyblastocyst stage to implantation of the late blastocyst(~120 cells) take on average about 24 hours. The dura-tion of these cycles is not only comparable to that oftypical proliferating cell populations, the cycles alsoinclude features lacking in early cleavage cycles. Theearly mouse cycles have a G1 and a G2, they arrest inresponse to aphidicolin inhibition of DNA replicationand they exhibit a radiation-induced arrest in G2 ([38],see below). As the post-fertilization cycles have little incommon with the early cleavage cycles of Drosophilaor frogs, it is a widely held view that the fast cycles ofmodel organisms are not relevant to mammalianembryonic cell cycle regulation. We agree that the post-fertilization cycles differ, but nonetheless suggest thatthere are mammalian cell cycles that are homologousto the rapid cleavage cycles of the model organisms,only that these cell cycles are at a different stage.

Aligning DevelopmentThere is a discontinuity in the manner in which devel-opment of mammals is aligned with that of other organ-isms. While the earliest stages are aligned based on theuse of fertilization as a reference point for the beginningof development, other common features of embryonicpatterning have led to an independent alignment ofembryogenesis at later stages. The latter is based onremarkably conserved features of morphology, pat-terning events and expression of conserved genes.

As noted by von Baer and emphasized by Haeckelmore than a century ago [21], all vertebrate embryoslook remarkably similar after the establishment of bodyaxes, neurulation and the beginning of somite formation(Figure 1A). Similarly, the analysis of embryos of diversearthropods has identified remarkable similarities afterestablishment of the body axis, production of a ventralnerve cord and initiation of segmentation [39,40]. Thisstage has been referred to as the phylotypic stage,because all of the organisms within a phylum appear toresemble each other at this stage. However, at thisstage the similarities are not only apparent within aphylum, but also between phyla. Thus, the body plansof arthropods and vertebrates share features such as acentral nerve cord, relative position of yolk, gut andnerve cord, and subdivisions along the anterior poste-rior axis (somites and segments; Figure 1). These mor-phological parallels are reinforced by parallels inpatterns of expression of important determinants ofdevelopmental fate (see below). Perhaps we shouldexpect such similarities across phyla at this stage. Justas conserved domains can be recognized in distantlyrelated protein sequences, it is the conserved steps ofdevelopment that can be most easily recognized in theembryos of distantly related organisms. We argue that,just as the alignment of distantly related proteins isbased on conserved domains rather than simply align-ing the sequences starting at the amino terminus, sothe alignment of distantly related developmental pro-grams ought to be based on alignment of the most con-served stages.

Whereas embryo morphology and size are remark-ably conserved at the phylotypic stage, it is commonly

Current BiologyR37

recognized that morphology diverges at later stages, asspecies specific anatomy develops [41]. Furthermore,as the phylotypic stage is the most conserved stage, itshould not be a surprise that earlier embryos also showa more highly diverged morphology (Figure 1). Indeed,as one examines progressively earlier stages, homolo-gies between species become gradually less evident.

For example, the similarity of neural tube formation viaa neural fold is clearly evident in diverse chordates, butcentral nerve cord formation in other phyla occurs indifferent ways. Even within chordates, however, theparallels in development become less evident at evenearlier stages. Challenging mental gymnastics arerequired to draw parallels between the gastrulationmovements of fish, frog, chick and mouse and the par-allels at pregastrulation stages are not clear. Nonethe-less, molecular studies and patterns of gene expressionhave shown that extensive parallels do exist.

Molecular analyses have detected parallels in thegastrulation processes of organisms belonging to dif-ferent phyla. In organisms from Drosophila to human,a conserved set of genes encodes an extracellular sig-naling pathway that governs dorsal/ventral patterningof the embryo. In this pathway, a BMP type of signal-ing molecule acts as a ventralizing signal in verte-brates, and, due to a switch in spatial reference-pointsand hence names, as a dorsalizing signal in arthro-pods [42,43]. Additionally, other interacting compo-nents of this signaling system (e.g., Chordin/Sog, andTwisted gastrulation) are also involved in diversespecies [44,45]. Although the pathway remains to befully elucidated, studies in Xenopus have shown thatBMP signaling regulates localized expression of thehomeodomain protein Goosecoid and the conservedDNA binding protein Brachyury [46,47]. Brachyury andits homologs appear to specify posterior embryonicstructures in species extending from C. elegans tomammals, whereas Goosecoid and its homologsspecify anterior structures [48–52].

While the details of the conservation and mechanismof action of the pattern forming genes are of tremen-dous importance, we wish to emphasize here theirutility in aligning analogous stages of the developmentof different groups. The BMP signaling cascade inXenopus acts very early, as the egg is undergoing therapid cleavage cell cycles. The onset of localizedexpression of Goosecoid and Brachyury precedes gas-trulation slightly and persists as a distinctive markduring gastrulation [53]. In species from echinoderms tomice, the expression of these genes shows a similarassociation with gastrulation [53–57]. Notably, thesegenes are not expressed in mammalian blastocysts,which are often presented as the analog of the frogblastula (Figure 1,2). Instead, they are expressed in theegg cylinder stage just prior to the onset of gastrulationin mouse (Figure 2; [58]). It has been argued by severalauthors that the common roles and expression patternsof these genes are the result of evolutionary conserva-tion, or evolution by descent [59–61]. We infer from thisthat the stage of mammalian embryogenesis that isanalogous to the frog blastula and the Drosophila blas-toderm is the pregastrulation stage, at which all ofthese conserved features of patterning occur. Mousegastrulation occurs at the egg cylinder stage (6.5 dpc)[37]. Thus, it is the egg cylinder stage that is analogousto the frog blastula and fly blastoderm.

In reviewing the mechanisms involved in embryonicaxis specification in Chordates, Eyal-Giladi similarly con-cluded that the mammalian blastocyst is not the analogof the blastula [62]. She argues that ‘the homology of the

ReviewR38

Figure 1. Phylotypic stages of chordates and arthropods.(A) The morphology of chordate embryos, represented by sala-mander, chick, and human embryos, initially converges to lookmore similar at the so called pharyngula stage or phylotypicstage [91,92] and subsequently diverges. The high degree of sim-ilarity between organisms allows an unambiguous alignment ofstages near the phylotypic stage, but ambiguities in alignmentcan occur at other stages. We argue that the illustrated and gen-erally accepted alignment of the human blastocyst with the blas-tula stages of amphibian and chick is incorrect (see Figure 2 fora more correct alignment). (B) The phylotypic stage of arthro-pods, represented by tick, spider, fruit fly and amphipod (freshwater shrimp) embryos, is proposed to occur just after gastrula-tion [39,93]. Parallels in body segmentation and the position ofthe yolk are apparent. Anterior is up. The existence of the phylo-typic stage has been questioned by Richardson who pointed outexaggerated homologies in Haeckel’s drawings [94]. We do not,however, feel that this is significant criticism of the concept ofthe phylotypic stage, during which the similarities betweenembryos across a phyla remain apparent relative to stagesbefore and after. Figure adapted from [95–99] and Flybase athttp://flybase.bio. indiana.edu/).

Egg(to scale)

Blastula Humanblastocyst(incorrectlyaligned)

Gastrula

Phylotypicstage

Salamander Chick Human

Tick Spider Fruit fly Amphipod

Current Biology

A

B

germ layers and of the cavities of a mammalian embryoto all other types of amniotic embryos … is quite clear’,and that it is ‘the narrow slit separating the epiblast fromthe hypoblast that should be identified as a blastocoele’.This ‘narrow slit’, which develops within the inner cellmass at about the time that the mammalian blastocystimplants, is distinct from the large cavity of the blasto-cyst, which can be considered an empty yolk vesicle(see below; Figure 2 and 3).

It should be noted that gastrulation in the mouse isoften considered to begin with the separation of prim-itive endoderm from the epiblast around 4.5 dpc.However, the primitive endoderm is homologous to thehypoblast, an extraembryonic tissue. For comparisonwith other systems we use the more widely accepteddefinition of gastrulation based on the formation of theembryonic germ layers. In mouse, this occurs in con-junction with primitive streak formation at day 6.5.Given this alignment of embryogenesis, should we lookfor the mammalian analog of the cleavage stages justbefore gastrulation or should we look almost a weekearlier, immediately after fertilization? We have usedgastrulation as our reference.

Peri-Gastrulation Cycles Show Features of FastEmbryonic CyclesAs first noted by Snow, the leisurely pace of cell cycleprogression that characterizes early mouse embryoge-nesis increases dramatically during the egg cylinderstage at 6.5 days [37,63]. The extraembryonic tissuesdo not undergo especially rapid cycles, but areas withinthe embryonic ectoderm have cell cycle times as short

as 2.2 hours. Snow [63] suggested that the divisions arelimited to a ‘proliferative zone’, whereas MacAuley et al.[64] suggested that all cells passing through the primi-tive streak undergo rapid cycles. Because cells moveduring gastrulation, the region of high proliferation is nota constant group of cells; rather, many of the embryoniccells pass through a period of rapid division in closecoordination with their gastrulation movements. As theprimitive streak lengthens across the embryo, a subsetof the embryonic ectodermal cells change shape andmove out of the ectodermal layer to form the futuremesoderm and endoderm. The primitive streak cellpopulation is, therefore, dynamic with ectodermal cellsmoving in to replace those moving out to form themesoderm. During this process, cells of the ectodermproliferate rapidly to populate the primitive streak.

Remarkably, the rapid pre-gastrulation (at the cellu-lar level) or peri-gastrulation (at the embryonic level)divisions in mammalian embryos share many featuresof pre-gastrulation cell cycles in flies and frogs. First,the feature that identified these cycles, their speed,sets them apart from the cycles of other mammaliancells and is an important parallel to the embryoniccycles in frogs and flies. As mentioned above, cellcycles of the proliferative zone in a 6.5 day old mouseembryo take on average 2.2 hours [37,63]. Ratembryos of a similar stage exhibit cell cycle times ofless than 3–3.5 hours [64]. Furthermore, like the cleav-age cycles of flies and frogs, the peri-gastrulationcycles of rat embryos show non-existent or short(0–30 min) G1 and G2 phases [64]. Second, cellulargrowth and division appears uncoupled to a certain

Current BiologyR39

Figure 2. Alignment of early embryonic stages among three chordates: frog, chick and human.Pregastrulation stages (bottom row) of mammal, bird and amphibian as represented by human, chick and Xenopus, show parallels.Cells of the human epiblast (blue) will begin gastrulation into the space between the epiblast and the hypoblast (yellow); cells of thechick epiblast (blue) will likewise move into the space between the epiblast and the hypoblast (yellow). In frogs, cells of the animalcap, which are proposed to be analogous to the epiblast [62], will gastrulate into a space between the animal cap and a subpopula-tion of vegetal cells, which are proposed to be analogous to the hypoblast cells [62]. The cavity into which human epiblast will gas-trulate is, therefore, equivalent to the blastocoelic cavity of chick and frog, but is not named as such. Rather, the yolk vesicle that liesbeneath the hypoblast is referred to, incorrectly we propose, as the cavity of the blastocoel. This (incorrect) nomenclature correlateswith the (incorrect) alignment of the human blastocyst (top row) with chick and frog blastulae (bottom row). We draw an additionalparallel in the migration of hypoblast cells (pale yellow) to form the yolk sac that surrounds the yolk vesicle (in humans) and the yolk(in chick). In an analogous process, cells of the primitive endoderm (yellow), which are equivalent to the hypoblast in mouse (Figure3) differentiate into visceral endoderm that forms the yolk sac. Figure adapted from [62,95].

Empty yolk vesicle(cavity of theblastocyst)

ICM

Trophoblast

Trophoblast

Epiblast

Presumed location ofblastocoel

HypoblastYolk sac(derived fromhypoblast)Yolk vesicle

Blastodisc

Yolk

Yolk

Epiblast

Blastocoel

HypoblastYolk sac(derived fromhypoblast)

Animal polecells

Yolkyvegetal polecells

Epiblastanalog

Blastocoel

Hypoblastanalog

Pre-gastrulation

stage

TIM

E

Mammal Bird FrogCurrent Biology

extent in the peri-gastrulation cycles of mouse and rat,such that cells of the primitive streak are smaller thanthose of their predecessors in the ectoderm [64]. Thismay be an inevitable consequence of a short G1phase during which cellular growth typically occurs inmany somatic cell types. Again, this is a feature that isshared by rapid embryonic divisions of frogs and flies

that subdivide the large egg mass. Third, studies in ratshow that pre-gastrulation cycles lack a checkpointthat inhibits mitosis in the presence of DNA damage.Rather than execute a cell cycle delay, these cells die[65]. Similarities of these cycles to cleavage cycles offrog and fly with regard to checkpoint control arefurther discussed below.

ReviewR40

Figure 3. Aligning fly and mouse development at gastrulation rather than at fertilization.During Drosophila oogenesis (left), the germ cell lineage (green and blue stripes) produces both the extraembryonic nurse cells (green)and the oocyte (blue). The somatic lineage (red) produces follicle cells that surround the developing oocyte. Fertilization in Drosophilais followed by 13 rapid nuclear divisions within a common cytoplasm. Cellularization follows to produce a cellular blastoderm in whichcortical cells enclose a central yolk mass. Gastrulation ensues next.In the mouse (right), the fertilized egg (green and blue striped) splits into extra-embryonic lineages (green) and the inner cell mass(ICM; dark blue), which will subsequently shed additional rounds of extra-embryonic lineages (yellow) as well as produce the embry-onic epiblast (light blue). Early specified extra-embryonic lineages (green) include the trophoblast cells, which endoreplicate like thenurse cells of Drosophila. Extra-embryonic primitive endoderm (yellow in late blastocyst stage), which delaminates from ICM is equiv-alent to the hypoblast of chick and human (yellow in Figure 2). Primitive endoderm will further differentiate into parietal endoderm andvisceral endoderm. At the egg cylinder stage, epiblast cells will migrate during gastrulation into the space between the epiblast andvisceral endoderm (hypoblast equivalent). This is therefore equivalent to the space between epiblast and hypoblast into which epi-blast cells move during gastrulation in chicken and humans (Figure 2). Embryonic ectodermal cells of the epiblast exhibit rapid divi-sion cycles prior to gastrulation, just as nuclei of Drosophila syncytium exhibit rapid divisions prior to cellularization and gastrulation.We do not wish to propose a one-to-one alignment of the fly developmental stages (left) to mouse developmental stages (right).Rather, we propose an alignment at gastrulation in accord with other authors, and suggest the absence of correspondence at fertil-ization. While we depict an approximate alignment of early stages, we expect that divergence at stages distant from the phylotypicstage has altered the coordination of different events to the point that precise alignment will change depending on the criterion usedto assess developmental stage. Nonetheless, this alignment differs significantly from prior alignments at the stage of fertilization inflies and mice. (Figures adapted from [88,95,100].)

Matureegg

Fertilization

Fertilization

Early blastocyst

Late blastocyst

Egg cylinder stage

Blastoderm

Germcells

Extaembryonicsomatic cells

Nurse cells(endoreplicate)

Oocyte

Yolk

Visceral endoderm

Epiblast

Yolk sac cavityTrophoblast

EpiblastProamniotic cavityParietal endodermVisceral endoderm

Extraembryonic ectodermEctoplacental cone

Primitive endoderm

Blastocysticcavity

Blastocysticcavity

Trophoblast(endoreplicate)

ICM

Current Biology

Fly

Gastrulation begins

Mouse

New Evidence of Parallels Between the CleavageCycles of Frog and Fly and Peri-Gastrulation Cyclesof Vertebrates

The repair of DNA damage is tied to progressionthrough the cell cycle. In species from yeast to human,genes have been characterized that arrest progress ofthe cell cycle to mitosis when DNA is damaged [66].This coupling gives cells the opportunity to repair thedamage prior to irrevocable genetic damage. In allspecies examined, the genes coupling cell cycle pro-gression to DNA damage are dispensable for undis-turbed cell cycle progression — at least in mostcycles. Indeed, the genes in this regulatory pathwayare the quintessential checkpoint genes — geneswhose ability to modulate cell cycle progression isthought to be engaged only when events go awry, aswould occur upon irradiation. A highly related check-point pathway senses the completion of DNA replica-tion and prevents inappropriate, premature progressto mitosis. Again these genes are generally dispens-able in undisturbed cell cycles.

In the context of the prevailing idea that theembryonic cleavage cycles lack checkpoint controls[16,17,19,67], it came as a surprise that analysis ofmutations in Drosophila ATM/ATR (mei-41) and Chk1(grapes) showed that these genes are uniquelyrequired in the early cycles [20,68,69]. Thus, genesthat are dispensable when they function in check-point control are required in the cycles in which wehad inferred they were not active. Clearly they mustbe active in the cleavage cycles, but what are theydoing if the cells do not exhibit a checkpoint? Apartial answer comes from more detailed studies ofcheckpoint action in the early cycles.

In vitro studies in a cycling Xenopus extract paral-leled findings in the intact embryo in that blocking DNAreplication did not block cycling as assessed either byoscillations of cyclin/Cdk kinase or by entry of nucleiinto M phase [19]. However, when the density of nucleiin the extract was increased, the cycling of the extractbecame dependent on replication [70]. It was inferredthat the extract, and presumably the embryo, is capableof coupling mitosis to S-phase but that the signal gen-erated at the very low nuclear to cytoplasmic ratio inthe early embryo was not sufficient to inhibit progressto mitosis. Other findings are also consistent with aquantitative interpretation.

Xenopus embryos and the cycling extracts alsoappear to lack a spindle checkpoint, as the use ofdrugs blocking spindle formation does not prevent exitfrom mitosis. However, when the nuclear density wasincreased in the extract, a drug-induced arrest ofmitosis became evident, as was seen in the case of thereplication checkpoint. This arrest depends on geneproducts homologous to those acting in checkpointregulation of mitotic progression in yeast [67].

In Drosophila, it was found that aphidicolin inhibitionof S-phase during embryonic cycle 11 or 12 brieflydelays the subsequent mitosis [20]. This suggests thatthe early embryos do have a mechanism that can delaymitosis when replication is incomplete. The transientnature of the block suggests that the mechanism mightnot have the quantitative ability to fully block the

activators of mitosis present during these stages. Inaddition to these indications of weak checkpoint activ-ity in the models that had originally suggested anabsence of checkpoints, there is full-fledged check-point activity in some systems. Disruption of the spindlein Drosophila embryos arrests cleavage nuclei inmetaphase, clam embryos exhibit checkpoint arrests inresponse to both microtubule depolymerizing drugsand inhibitors of DNA replication, and sea urchinembryos arrest in response to inhibitors of DNA repli-cation [25,71–73]. Thus, it now appears that checkpointpathways are present during cleavage cycles.

Drosophila mutants in the checkpoint genes mei-41(ATM/ATR) and grapes (Chk1) are viable, but they arematernal effect lethals, i.e. mutant adult females giveno or few progeny [20,68,74]. Embryos from mutantmothers show severely defective cell cycles by thetime of mitosis 12. Because the requirement for thesefunctions is seen in the absence of any perturbation,this finding indicates that control of the early cycles isdistinct from that of other cell cycles. It is unclear whyundisturbed rapid divisions should require checkpointactivities. In cycles without the leeway provided by aG2-phase, these functions are perhaps needed toensure that mitosis does not initiate until after DNAsynthesis is completed [20,74]. Alternatively, theunusually fast mitoses perhaps rely on this pathway toserve a different role that ensures the proper orderand timing of mitotic events [69].

Regardless of the mechanism that underlies theunique requirement for these checkpoint genes duringthe early cleavage stages of Drosophila, it is striking thatthe mouse embryo exhibits a similar requirement at thetime of the rapid peri-gastrulation divisions. Analysis ofmouse mutations in ATR and Chk1 shows that thesegenes are dispensable for cell divisions shortly after fer-tilization, but they become essential in peri-gastrulationmouse embryos. ATR−/− embryos develop to 3.5 dpc,but die between implantation (after 4.5 dpc) and 7.5dpc, which encompasses the peri-implantation stagesunder discussion. Chk1−/− embryos die between 3.5and 7.5 dpc, which again encompasses the peri-gas-trulation stages [75,76]. In culture, cells from ATRmutant embryos display defects that are consistent withcell division problems, such as small size, decreasedproliferative index and chromosome fragmentation [77].As in the case of fly embryos, the reason for the require-ment for ATR and Chk1 in mouse embryogenesisremains unclear. Nonetheless, it would be interesting toknow if ATR and Chk1 homologs play essential rolesduring rapid cell cycles in other phyla.

The rapid early embryonic cycles of the fly shareanother unusual feature with the perigastrulation cellcycles of the mouse egg cylinder. During the earlycycles in Drosophila, embryos are remarkably radiationsensitive. The dose of ionizing radiation that kills 50% ofembryos undergoing cleavage divisions is about 250Rads, compared to ~4000 Rads in larvae [78,79].Detailed analysis of the timing of the change in sensitiv-ity showed that embryos develop a much increasedradiation tolerance at the close of the rapid mitoticcycles. Real time analysis following irradiation revealedthat early embryos fail to arrest progress into mitosis

Current BiologyR41

within the cycle in which they are irradiated, and thatthey show severe defects in subsequent mitosis [68,80].Interestingly, the cell cycles of gastrulating mouseembryos also lack a checkpoint response that inhibitsmitosis in the presence of DNA damage [65]. Ratherthan execute a cell cycle delay, these cells readilycommit cell death, reminiscent of the radiation sensitiv-ity of the cleavage stage fly embryo. Consequently, cellsof mouse embryos show higher sensitivity to killing byionizing radiation during gastrulation than at an earlierstage. Likewise, cells of the frog embryo show highersensitivity to killing by ionizing radiation during cleavagecycles than at later stages [81]. Although more com-plete time courses are needed, the results suggest thatembryonic cells are unusually sensitive to radiationduring the rapid cell cycles. It will also be of interest todetermine whether the inability to regulate the entry intomitosis in response to damaged DNA is a featureshared by rapidly cycling embryonic cells of differentphyla. While many aspects of the roles of the check-point genes in the embryonic cycles need to be defined,it is striking that the early cleavage cycles of modelorganisms share with the peri-gastrulation cycles ofmammals a unique requirement for the checkpointpathway and an especially high sensitivity to irradiation.

Post-Fertilization Cycles in Mammalian EmbryosMay Have Been Derived from Pre-FertilizationProcessesIf peri-gastrulation cell cycles of mammals are equiva-lent to the cleavage of early embryos in other metazoa,why are they not seen immediately following fertiliza-tion? In mice and humans, divisions immediately afterfertilization do not produce the embryo proper butextra-embryonic tissues that will supply nutrients to theembryo. Two generations of primary extraembryonictissues are produced before the mammalian embryoinitiates events of embryogenesis: the post-fertilizationdivisions produce a blastocyst composed of a sphereof trophectoderm (extraembryonic) and inner cell mass(ICM), and later (4 days into mouse embryogenesis) theICM produces the primitive endoderm (extraembryonic)and the primitive ectoderm (embryonic) [37]. The firstgeneration of extraembryonic tissue, the trophoblast,will differentiate into the placenta and the chorion, whilethe later formed primitive endoderm/hypoblast will firstdifferentiate into parietal and visceral endoderm andlater into the yolk sac (Figure 3). Thus, strictly speaking,the divisions that immediately follow fertilization are notembryonic divisions yet, but rather produce tissuesinvolved in nourishing the embryo.

In most vertebrates, it is the yolk sac that performsthe nutritive role. In meroblastic embryos, which arederived from yolk laden eggs and have incomplete earlycytokinesis (e.g. chick), the yolk sac, true to its name,sits at the interface with the abundant yolk and harvestsyolk material to provide for the embryo. These organ-isms lack the first wave of extraembryonic tissue gen-eration and show no evidence of a trophoblast [62,82].The primitive placenta of sharks and viviparous reptilesis formed by a secondary specialization of the yolk sacwhich serves a dual role of providing nutrients first fromthe yolk and then from the mother. While the yolk sac

retains some nutritive functions in mammals in which itfunctions to take up material from uterine fluid, the tro-phoblast-derived placenta performs the major nutritiverole. It appears as if evolution has added a new stageto early embryogenesis in order to generate this nutri-tive organ. It is of interest to consider whether the tro-phoblast had an evolutionary precursor and if so, whatit might be.

Given arguments that the cavity of the blastocyst isanalogous to an empty yolk vesicle (see above; Figure2,3), perhaps the evolutionary precursors of the tro-phoblast cells will surround the yolk mass in the prede-cessors of mammals. In mammals, the cavity of theblastocyst comes to be surrounded by two layers ofcells; the shell of trophoblastic cells defines this cavityand cells derived from the primitive endoderm/hypoblast migrate over the inner surface of the tro-phoblast to form the second layer, the yolk sac. In theembryos of birds, hypoblast cells migrate over thesurface of the yolk to form the yolk sac, but there is nooverlying tissue analogous to the trophoblast. However,earlier, during oogenesis, there are maternally derivedgranulosa cells surrounding the yolk. These cells com-prise a monolayered epithelium that contributes to theextraordinary accumulation of yolk during oogenesis. Insummary, it appears that the trophoblast is a novelfeature of embryogenesis that was added during theevolutionary history of mammals. Birds and presumablyreptiles, though lacking any obvious zygotic analog ofthe trophoblast, possess maternal tissue that is spa-tially and physically analogous.

How was the developmental program modified tointroduce a new stage and accommodate the tro-phoblast? The above comparison to chick as well asbroader evolutionary considerations suggest how thischange may have occurred. In contrast to mammalianeggs, the eggs of non-uterine animals grow to verylarge sizes during oogenesis. The huge expansion ofthe oocyte is promoted by specialized nutritive tissues.Detailed analysis, largely outside chordates, shows thatthere are two categories of nutritive cells, nurse cellsand follicle cells. In diverse organisms, nurse cells aresister cells of oocytes (e.g., the beetle, Dytiscus or theleech, Pisciola; [21]). In chordates, the differentiatingoocytes appear to recruit cells that become the granu-losa cells or follicle cells that perform this nutritive role[83]. However, the lineages that give rise to granulosacells in different chordates and their relationship to thenurse cells and follicle cells of non-chordate speciesare ill defined. We speculate that in non-mammalianchordates at least some of the granulosa cells are sistercells of the oocyte, much like the nurse cells in otherphyla. If, during the evolution of mammals, a change inthe timing of events occurred, so that meiosis and fer-tilization shifted to precede the cell divisions that sepa-rated the later derived granulosa cells form the oocyte,some of the granulosa cells, namely the one’s pro-duced late in the lineage, would then be produced afterfertilization. According to this scenario, these zygoti-cally produced ‘granulosa cells’ would be the tro-phoblast cells. They would continue to perform anutritive function but now would be providing nutritionto the developing embryo rather than the oocyte.

ReviewR42

Though speculative, this hypothesis explains featuresof early mammalian development beyond the origins ofthe trophoblast. For example, the transition to the smallsize of the mammalian egg could be explained by its‘precocious’ maturation. Furthermore, the switch frommaternally regulated to zygotically regulated earlydevelopment would follow from the change in relativetime of fertilization, which would switch many eventsfrom pre-fertilization to post-fertilization. An adaptationleading to viviparous fish reveals another case in whichthe relative timing of fertilization is altered. In someviviparous fish (some Poecilid teleosts), the egg is fer-tilized within the follicle and is supported throughembryonic development within the follicle prior to ‘ovu-lation’/birth [84].

Our proposal would predict that early zygotic devel-opment of mammals might resemble steps in folliculardevelopment in monotremes (e.g. Platypus), birds, rep-tiles and other non-therian species. In the formation ofchordate follicles, the developing oocyte and a few‘granulosa cells’ are isolated within a basement mem-brane, the granulosa cells begin to form an epitheliumon the outside and the oocyte grows on the inside (e.g.[85]). One can make analogies between this stage andthe morula of mammalian embryogenesis, a stage atwhich the central cells commit to an embryonic fateand the outer cells begin to take on epithelial featuresof the trophoblast. Other similarities can be drawnbetween the ‘granulosa cell’ layer and the developmentof the trophoblast; however, if they are related bydescent, modification of subsequent events gives thesetissues distinct roles.

Despite the enormous evolutionary distance, fea-tures of the developmental programs in the nurse cellsof non-chordate species might be taken as support forour proposal. As mentioned, the nurse cells are pro-duced from germ cell precursors and hence are sistercells of the oocytes [21]. In some cases the fates ofnurse cells and the oocytes are separated extremelylate. Indeed, in the leech Pisciola the nurse cells enterthe meiotic divisions. If the sibling relationship of nurseand germ cells were to persist only slightly longer, thelineages might separate after fertilization, as wesuggest for mammals. Furthermore, throughout mostof metazoan phylogeny, nurse cells contribute to thegrowth of the egg and accumulation of maternally pro-vided yolk stores. The granulosa cells of non-mam-malian chordates make a similar contribution, whiletrophoblast cells transfer nutrients to the embryo, arole not unlike that of nurse/granulosa cells. Addition-ally, nurse cells in diverse species grow to very largesizes and develop huge polyploid nuclei. The enor-mous, branched nucleus of the nurse cells of an earwigprovides a dramatic example of this [21]. The nursecells of Drosophila and other species are polyploid andproduce prodigious levels of RNA (e.g., the polychaeteOphryotrocha labronica and the dipteran, Calliphora[86,87]). As suggested by their name, the trophoblasticgiant cells are very large, and they also endocycle tobecome polyploid. Perhaps this unusual behavior ofthe trophoblast cells has a primordial origin in nursecell developmental programs that evolved in non-chor-date predecessors.

According to the proposal made here, the early cellcycles of mammalian eggs would not be analogous tothe cleavage cycles in other embryos, but to cell divi-sions producing the oocyte and its sister cells. While wehave been unable to locate, among published works,descriptions of these events in the non-mammalianchordates, these events are well known in Drosophila.Four cell divisions of a precursor cell produce theoocyte and the 15 nurse cells that populate a Drosophilaegg chamber. These four egg chamber divisions take onaverage 6 hours per cell cycle, much longer than earlycleavage cycles [88]. Based on our proposal that thetiming of fertilization is displaced with respect to otherevents in mammals, we propose the speculative align-ment shown in Figure 2.

A Precedent for Our ProposalBiology provides an independent example of deferreddevelopment and of production of extraembryonictissues. Like the embryos of uterine animals, theembryos of parasitic wasps develop in a highly nutritiveenvironment. After fertilization, the eggs of Copidosomafloridanum undergo dramatic growth and proliferation toproduce multiple twin embryos within the hemolymph ofparasitised caterpillars [89]. To accomplish this, embry-onic patterning and gastrulation are deferred, in a similarmanner as in mammalian development. Additionally,these embryos possess extraembryonic membranes[90]. These membranes develop from polar nuclei, thereduction nuclei of meiosis that are usually discarded. Inthis organism, it is unambiguous that the events repre-sent a deferral of embryonic development because thelater patterning of the multiple twinned embryos can beclearly aligned with the developmental programs of otherinsects. Thus, it is undeniable that deferral of earlyembryonic programs evolved at least once. We proposethat mammals represent a second example.

The establishment of analogies between modelsystems and mammalian development extends theimpact of studies in model organisms. In the view pre-sented in this article, the studies of rapid cell cycles infrogs and flies will directly benefit the understandingof mammalian embryonic cell cycles.

AcknowledgmentsWe thank Gail Martin, Gilles R. Hickson, Pascale Dijck-ers, Joan Ruderman, and Michael Stowell for helpfulcomments.

References1. O’Farrell, P.H. (2003). How metazoans reach their final size: the

natural history of bigness. In Cell Growth, Hall, M., Raff, M, Thomas,G., eds. (Cold Spring Harbor Laboratory Press).

2. Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., and Walter,P. (2002). Molecular Biology of the Cell (4th edn), Garland Science.

3. Foe, V.E., Odell, G.M., and Edgar, B.A. (1993). Mitosis and morpho-genesis in the Drosophila embryo. In The Development ofDrosophila melanogaster, Bate, M., Martinez Arias, A., eds. (ColdSpring Harbor Laboratory Press).

4. Milan, M., Campuzano, S., and Garcia-Bellido, A. (1996). Cell cyclingand patterned cell proliferation in the Drosophila wing during meta-morphosis. Proc. Natl. Acad. Sci. USA 93, 11687–11692.

5. Milan, M., Campuzano, S., and Garcia-Bellido, A. (1996). Cell cyclingand patterned cell proliferation in the wing primordium ofDrosophila. Proc. Natl. Acad. Sci. USA 93, 640–645.

6. Mitchison, J.M. (1971). The Biology of the Cell Cycle (CambridgeUniversity Press).

Current BiologyR43

7. Conlon, I., and Raff, M. (2003). Differences in the way a mammaliancell and yeast cells coordinate cell growth and cell-cycle progres-sion. J. Biol. 2, 7.

8. Ferrell, J.E., Jr., Wu, M., Gerhart, J.C., and Martin, G.S. (1991). Cellcycle tyrosine phosphorylation of p34cdc2 and a microtubule-asso-ciated protein kinase homolog in Xenopus oocytes and eggs. Mol.Cell. Biol. 11, 1965–1971.

9. Newport, J., and Kirschner, M. (1982). A major developmental tran-sition in early Xenopus embryos: I. characterization and timing ofcellular changes at the midblastula stage. Cell 30, 675–686.

10. Newport, J.W., and Kirschner, M.W. (1984). Regulation of the cellcycle during early Xenopus development. Cell 37, 731–742.

11. Baldari CT, Amaldi F. (1976). DNA reassociation kinetics in relationto genome size in four amphibian species. Chromosoma 59, 13–22.

12. Blumenthal, A.B., Kriegstein, H.J., and Hogness, D.S. (1974). Theunits of DNA replication in Drosophila melanogaster chromosomes.Cold Spring Harb. Symp. Quant. Biol. 38, 205–223.

13. McKnight, S.L., and Miller, O.L., Jr. (1977). Electron microscopicanalysis of chromatin replication in the cellular blastodermDrosophila melanogaster embryo. Cell 12, 795–804.

14. Spradling, A., and Orr-Weaver, T. (1987). Regulation of DNA repli-cation during Drosophila development. Annu. Rev. Genet. 21,373–403.

15. Walter, J., and Newport, J.W. (1997). Regulation of replicon size inXenopus egg extracts. Science 275, 993–995.

16. Raff, J.W., and Glover, D.M. (1988). Nuclear and cytoplasmic mitoticcycles continue in Drosophila embryos in which DNA synthesis isinhibited with aphidicolin. J. Cell Biol. 107, 2009–2019.

17. Raff, J.W., and Glover, D.M. (1989). Centrosomes, and not nuclei,initiate pole cell formation in Drosophila embryos. Cell 57, 611–619.

18. Glover, D.M., Alphey, L., Axton, J.M., Cheshire, A., Dalby, B.,Freeman, M., Girdham, C., Gonzalez, C., Karess, R.E., Leibowitz,M.H., et al. (1989). Mitosis in Drosophila development. J. Cell Sci.Suppl. 12, 277–291.

19. Newport, J., and Dasso, M. (1989). On the coupling between DNAreplication and mitosis. J. Cell Sci. Suppl. 12, 149–160.

20. Sibon, O.C., Stevenson, V.A., and Theurkauf, W.E. (1997). DNA-replication checkpoint control at the Drosophila midblastula transi-tion. Nature 388, 93–97.

21. Wilson, E. B. (1925). The cell in development and heredity, 3rdEdition (New York: The MacMillan Company).

22. Bissen, S.T., and Weisblat, D.A. (1989). The durations and compo-sitions of cell cycles in embryo of the leech Helobdella triserialis.Development 106, 105–118.

23. Parisi, E., Filosa, S., De Petrocellis, B., and Monroy, A. (1978). Thepattern of cell division in the early development of the sea urchin.Paracentrotus lividus. Dev. Biol. 65, 38–49.

24. Wood, W.B. (1988). Embryology. In The Nematode Caenorhabditiselegans, Wood, W.B., ed. (Cold Spring Harbor Laborarory Press).

25. Hunt, T., Luca, F.C., and Ruderman, J.V. (1992). The requirementsfor protein synthesis and degradation, and the control of destruc-tion of cyclins A and B in the meiotic and mitotic cell cycles of theclam embryo. J. Cell Biol. 116, 707–724.

26. Wright, S.J., and Schatten, G. (1990). Teniposide, a topoisomeraseII inhibitor, prevents chromosome condensation and separation butnot decondensation in fertilized surf clam (Spisula solidissima)oocytes. Dev. Biol. 142, 224–232.

27. Tudge, C. (2000). The Variety of Life (Oxford University Press).28. Eyal-Giladi, H., and Kochav, S. (1976). From cleavage to primitive

streak formation: a complementary normal table and a new look atthe first stages of the development of the chick. I. General mor-phology. Dev. Biol. 49, 321–337.

29. Grosshans, J., and Wieschaus, E. (2000). A genetic link betweenmorphogenesis and cell division during formation of the ventralfurrow in Drosophila. Cell 101, 523–531.

30. Mata, J., Curado, S., Ephrussi, A., and Rorth, P. (2000). Tribblescoordinates mitosis and morphogenesis in Drosophila by regulat-ing string/CDC25 proteolysis. Cell 101, 511–522.

31. Seher, T.C., and Leptin, M. (2000). Tribbles, a cell-cycle brake thatcoordinates proliferation and morphogenesis during Drosophilagastrulation. Curr. Biol. 10, 623–629.

32. Shermoen, A.W., and O’Farrell, P.H. (1991). Progression of the cellcycle through mitosis leads to abortion of nascent transcripts. Cell67, 303–310.

33. Edgar, B.A., and Schubiger, G. (1986). Parameters controlling tran-scriptional activation during early Drosophila development. Cell 44,871–877.

34. Hartenstein, V., and Campos-Ortega, J. (1985). Fate-mapping inwild-type Drosophila melanogaster. I. The spatio-temporal patternof embryonic cell divisions. Wilhelm Rouzs Arch. Dev. Biol. 194,181–195.

35. Hartenstein, V., Rudloff, E., and Campos-Ortega, J. (1987). Thepattern of proliferaion of the neuroblasts in the wild-type embyo ofDrosophila melanogaster. Wilgelm Rouxs Arch. Dev. Biol. 196,473–485.

36. Patel, N.H., Ball, E.E., and Goodman, C.S. (1992). Changing role ofeven-skipped during the evolution of insect pattern formation.Nature 357, 339–342.

37. Hogan, B., Beddington, R., Costantini, F., and Lacy, E. (1994).Manipulating the Mouse Embryo (2nd edn), Cold Springer Labora-tory Press.

38. Fulka, J., Jr., First, N.L., Fulka, J., and Moor, R.M. (1999). Check-point control of the G2/M phase transition during the first mitoticcycle in mammalian eggs. Hum. Reprod. 14, 1582–1587.

39. Thomas, J.B., Bastiani, M.J., Bate, M., and Goodman, C.S. (1984).From grasshopper to Drosophila: a common plan for neuronaldevelopment. Nature 310, 203–207.

40. Patel, N.H., Kornberg, T.B., and Goodman, C.S. (1989). Expressionof engrailed during segmentation in grasshopper and crayfish.Development 107, 201–212.

41. Gould, S.J. (1977). Ontegeny and Phylogeny (Belknap Press).42. Holley, S.A., Neul, J.L., Attisano, L., Wrana, J.L., Sasai, Y., O’Con-

nor, M.B., De Robertis, E.M., and Ferguson, E.L. (1996). TheXenopus dorsalizing factor noggin ventralizes Drosophila embryosby preventing DPP from activating its receptor. Cell 86, 607–617.

43. Bachiller, D., Klingensmith, J., Shneyder, N., Tran, U., Anderson, R.,Rossant, J., and De Robertis, E.M. (2003). The role of chordin/Bmpsignals in mammalian pharyngeal development and DiGeorge syn-drome. Development 130, 3567–3578.

44. Garcia Abreu, J., Coffinier, C., Larrain, J., Oelgeschlager, M., and DeRobertis, E.M. (2002). Chordin-like CR domains and the regulationof evolutionarily conserved extracellular signaling systems. Gene287, 39–47.

45. Oelgeschlager, M., Reversade, B., Larrain, J., Little, S., Mullins,M.C., and De Robertis, E.M. (2003). The pro-BMP activity of Twistedgastrulation is independent of BMP binding. Development 130,4047–4056.

46. Re’em-Kalma, Y., Lamb, T., and Frank, D. (1995). Competitionbetween noggin and bone morphogenetic protein 4 activities mayregulate dorsalization during Xenopus development. Proc. Natl.Acad. Sci. USA 92, 12141–12145.

47. Isaacs, H.V., Andreazzoli, M., and Slack, J.M. (1999). Anteroposte-rior patterning by mutual repression of orthodenticle and caudal-type transcription factors. Evol. Dev. 1, 143–152.

48. Stott, D., Kispert, A., and Herrmann, B.G. (1993). Rescue of the taildefect of Brachyury mice. Genes Dev. 7, 197–203.

49. Kispert, A., Herrmann, B.G., Leptin, M., and Reuter, R. (1994).Homologs of the mouse Brachyury gene are involved in the speci-fication of posterior terminal structures in Drosophila, Tribolium,and Locusta. Genes Dev. 8, 2137–2150.

50. Woollard, A., and Hodgkin, J. (2000). The Caenorhabditis elegansfate-determining gene mab-9 encodes a T-box protein required topattern the posterior hindgut. Genes Dev. 14, 596–603.

51. Schulte-Merker, S., van Eeden, F.J., Halpern, M.E., Kimmel, C.B.,and Nusslein-Volhard, C. (1994). no tail (ntl) is the zebrafish homo-logue of the mouse T (Brachyury) gene. Development 120,1009–1015.

52. Giovannini, N., and Rungger, D. (2002). Antisense inhibition ofXbrachyury impairs mesoderm formation in Xenopus embryos. Dev.Growth Differ. 44, 147–159.

53. De Roberts, E.M., Blum, M., Niehrs, C., and Steinbeisser, H. (1992).Goosecoid and the organizer. Dev. Suppl. 1, 167–171.

54. Blum, M., Gaunt, S.J., Cho, K.W., Steinbeisser, H., Blumberg, B.,Bittner, D., and De Robertis, E.M. (1992). Gastrulation in the mouse:the role of the homeobox gene goosecoid. Cell 69, 1097–1106.

55. Kispert, A., Ortner, H., Cooke, J., and Herrmann, B.G. (1995). Thechick Brachyury gene: developmental expression pattern andresponse to axial induction by localized activin. Dev. Biol. 168,406–415.

56. Shoguchi, E., Satoh, N., and Maruyama, Y.K. (2000). A starfishhomolog of mouse T-brain-1 is expressed in the archenteron ofAsterina pectinifera embryos: possible involvement of two T-boxgenes in starfish gastrulation. Dev. Growth Differ. 42, 61–68.

57. Gross, J.M., and McClay, D.R. (2001). The role of Brachyury (T)during gastrulation movements in the sea urchin Lytechinus varie-gatus. Dev. Biol. 239, 132–147.

58. Wilkinson, D.G., Bhatt, S., and Herrmann, B.G. (1990). Expressionpattern of the mouse T gene and its role in mesoderm formation.Nature 343, 657–659.

59. McGhee, J.D. (2000). Homologous tails? Or tales of homology?Bioessays 22, 781–785.

60. Adell, T., Grebenjuk, V.A., Wiens, M., and Muller, W.E. (2003). Isola-tion and characterization of two T-box genes from sponges, thephylogenetically oldest metazoan taxon. Dev. Genes Evol., in press.

ReviewR44

61. De Robertis, E.M., Fainsod, A., Gont, L.K., and Steinbeisser, H.(1994). The evolution of vertebrate gastrulation. Dev. Suppl. 1,117–124.

62. Eyal-Giladi, H. (1997). Establishment of the axis in chordates: factsand speculations. Development 124, 2285–2296.

63. Snow, M.H.L. (1977). Gastrulation in the mouse: growth and region-alization of the epiblast. J. Exp. Emryol. Exp. Morphol. 42, 293–303.

64. Mac Auley, A., Werb, Z., and Mirkes, P.E. (1993). Characterizationof the unusually rapid cell cycles during rat gastrulation. Develop-ment 117, 873–883.

65. Heyer, B.S., MacAuley, A., Behrendtsen, O., and Werb, Z. (2000).Hypersensitivity to DNA damage leads to increased apoptosisduring early mouse development. Genes Dev. 14, 2072–2084.

66. Elledge, S.J. (1996). Cell cycle checkpoints: preventing an identitycrisis. Science 274, 1664–1672.

67. Minshull, J., Sun, H., Tonks, N.K., and Murray, A.W. (1994). A MAPkinase-dependent spindle assembly checkpoint in Xenopus eggextracts. Cell 79, 475–486.

68. Fogarty, P., Campbell, S.D., Abu-Shumays, R., Phalle, B.S., Yu,K.R., Uy, G.L., Goldberg, M.L., and Sullivan, W. (1997). TheDrosophila grapes gene is related to checkpoint gene chk1/rad27and is required for late syncytial division fidelity. Curr. Biol. 7,418–426.

69. Su, T.T., Campbell, S.D., and O’Farrell, P.H. (1999). Drosophilagrapes/CHK1 mutants are defective in cyclin proteolysis and coor-dination of mitotic events. Curr. Biol. 9, 919–922.

70. Dasso, M., and Newport, J.W. (1990). Completion of DNA replica-tion is monitored by a feedback system that controls the initiationof mitosis in vitro: studies in Xenopus. Cell 61, 811–823.

71. Edgar, B.A., Sprenger, F., Duronio, R.J., Leopold, P., and O’Farrell,P.H. (1994). Distinct molecular mechanism regulate cell cycle timingat successive stages of Drosophila embryogenesis. Genes Dev. 8,440–452.

72. Ikegami, S., Taguchi, T., Ohashi, M., Oguro, M., Nagano, H., andMano, Y. (1978). Aphidicolin prevents mitotic cell division by inter-fering with the activity of DNA polymerase-alpha. Nature 275,458–460.

73. Su, T.T., Sprenger, F., DiGregorio, P.J., Campbell, S.D., and O’Far-rell, P.H. (1998). Exit from mitosis in Drosophila syncytial embryosrequires proteolysis and cyclin degradation, and is associated withlocalized dephosphorylation. Genes Dev. 12, 1495–1503.

74. Sibon, O.C., Laurencon, A., Hawley, R., and Theurkauf, W.E. (1999).The Drosophila ATM homologue Mei-41 has an essential check-point function at the midblastula transition. Curr. Biol. 9, 302–312.

75. Liu, Q., Guntuku, S., Cui, X.S., Matsuoka, S., Cortez, D., Tamai, K.,Luo, G., Carattini-Rivera, S., DeMayo, F., Bradley, A., et al. (2000).Chk1 is an essential kinase that is regulated by Atr and required forthe G(2)/M DNA damage checkpoint. Genes Dev. 14, 1448–1459.

76. Takai, H., Tominaga, K., Motoyama, N., Minamishima, Y.A., Naga-hama, H., Tsukiyama, T., Ikeda, K., Nakayama, K., and Nakanishi,M. (2000). Aberrant cell cycle checkpoint function and early embry-onic death in Chk1(−/−) mice. Genes Dev. 14, 1439–1447.

77. Brown, E.J., and Baltimore, D. (2000). ATR disruption leads to chro-mosomal fragmentation and early embryonic lethality. Genes Dev.14, 397–402.

78. Wurgler, F.E., and Ulruch, H. (1976). Radiosensitivity of embryonicstages. In The Genetics and Biology of Drosophila, Ashburner, M.,Novitski, E., eds. (Academic Press).

79. Ashburner, M. (1989). Drosophila: A Laboratory Handbook (ColdSpring Harbor Laboratory Press).

80. Sibon, O.C., Kelkar, A., Lemstra, W., and Theurkauf, W.E. (2000).DNA-replication/DNA-damage-dependent centrosome inactivationin Drosophila embryos. Nat. Cell Biol. 2, 90–95.

81. Anderson, J.A., Lewellyn, A.L., and Maller, J.L. (1997). Ionizing radi-ation induces apoptosis and elevates cyclin A1-Cdk2 activity beforebut not after the midblastula transition in Xenopus. Mol. Biol. Cell 8,1195–1206.

82. Solnica-Krezel, L. (1999). Pattern formation in zebrafish–fruitfulliaisons between embryology and genetics. Curr. Top. Dev. Biol. 41,1–35.

83. Rothchild, I. (2003). The yolkless egg and the evolution of eutherianviviparity. Biol. Reprod. 68, 337–357.

84. Thibault, R., and Shultz, R. (1978). Reproductive adaptations amongviviparous fishes (Cyprinodontiformes and Poecelidae). Evolution32, 320–333.

85. Prisco, M., Loredana, R., and Piero, A. (2002). Ultrastructural studieson developing follicles of the spotted ray Torpedo marmorata. Mol.Reprod. Dev. 61, 78–86.

86. Emanuelsson, H. (1985). Autoradiographic analysis of RNA synthe-sis in the oocyte-nurse cell complex of the polychaete Ophry-otrocha labronica. J. Embryol. Exp. Morphol. 88, 249–263.

87. Ribbert, D. (1979). Chromomeres and puffing in experimentallyinduced polytene chromosomes of Calliphora erythrocephala.Chromosoma 74, 269–298.

88. Spradling, A. (1993). Dev. Genet. of Oogenesis. In The Developmentof Drosophila melanogaster, Bate, M., Martinez Arias, A., eds. (Vol.1), Cold Spring Harbor Laboratory Press.

89. Grbic, M., Nagy, L.M., Carroll, S.B., and Strand, M. (1996). Polyem-bryonic development: insect pattern formation in a cellularized envi-ronment. Development 122, 795–804.

90. Corley, L.S., and Strand, M.R. (2003). Evasion of encapsulation bythe polyembryonic parasitoid Copidosoma floridanum is mediatedby a polar body-derived extraembryonic membrane. J. Invertebr.Pathol. 83, 86–89.

91. Kuratani, S., Nobusada, Y., Horigome, N., and Shigetani, Y. (2001).Embryology of the lamprey and evolution of the vertebrate jaw:insights from molecular and developmental perspectives. Philos.Trans. R. Soc. Lond. B Biol. Sci. 356, 1615–1632.

92. Slack, J.M., Holland, P.W., and Graham, C.F. (1993). The zootypeand the phylotypic stage. Nature 361, 490–492.

93. Patel, N.H., Martin-Blanco, E., Coleman, K.G., Poole, S.J., Ellis,M.C., Kornberg, T.B., and Goodman, C.S. (1989). Expression ofengrailed proteins in arthropods, annelids, and chordates. Cell 58,955–968.

94. Richardson, M.K. (1999). Vertebrate evolution: the developmentalorigins of adult variation. Bioessays 21, 604–613.

95. Gilbert, S. (1991). Developmental Biololgy (3rd edn), Sunderland.96. Elinson, R. (1987). Changes in developmental patterns: embryos of

amphibians with large eggs. In Development as an evolutionaryprocess, Raff, R., Raff, E., eds. (Alan R. Liss).

97. Abele, L.G. (ed.) (1982). Embryology, Morphology and Genetics(Academic Press).

98. Foelix, R.F. (1996). Biology of Spiders (Oxford University Press).99. Sonenshine, D.E. (1991). Biology of Ticks (Oxford University Press).100. Behringer, R.R., Wakamiya, M., Tsang, T.E., and Tam, P.P. (2000). A

flattened mouse embryo: leveling the playing field. Genesis 28,23–30.

Current BiologyR45