Maternally synthesized mRNAs and proteins control early animal development until the maternal to zygotic transition (MZT), at which point the zygotic genome assumes control. Synthesis of maternal transcripts begins early in oogenesis. Because transcription is either absent or unnecessary in late-stage oocytes as well as during the maternally controlled period at the beginning of embryogenesis, posttranscriptional control of maternal mRNA is a key means of regulating gene expression at these developmental stages.
Two key control points during these stages of early development are oocyte maturation, which leads to the production of fully developed eggs that are primed for embryogenesis, and egg activation, which triggers the already mature egg to initiate the cellular and molecular processes required for embryogenesis. Maturation and activation have been studied extensively in amphibians and echinoderms where large numbers of oocytes can be obtained and induced to undergo these processes in vitro. Little analysis has been done, however, in a system that is both genetically tractable and biochemically accessible, such as Drosophila. The advantage of such a system is that it allows for examination of both the mechanisms and the functions of posttranscriptional regulation. By mutating regulatory components, for example, their roles in the organism's development can be assessed. Additionally, the function of specific target transcripts can be studied by introducing transgenes that express mRNAs with mutated cis-acting elements, which cause these transcripts to be misregulated.
This review provides a comparative discussion, with a focus on Drosophila, of two different levels of posttranscriptional control: translation and transcript stability. We examine how this regulation is coordinated with and by the developmental events that occur from late oogenesis to the mid-blastula transition (MBT) which, in Drosophila, is coincident with the MZT. Simultaneously, we highlight the developmental functions of translational control and transcript destabilization in the early embryo. Our focus will be largely on what might be termed “instructive” maternal mRNAs: ones that encode proteins that instruct the embryo as to spatial pattern, cell fate, and cell division. Many of these are transcription factors, RNA-binding proteins, cell surface receptors, signal transduction molecules, and cell cycle regulators. Little consideration will be given to mRNAs that encode “permissive” proteins, such as general components of the cytoskeleton, the ribosome, and chromatin.
POSTTRANSCRIPTIONAL CONTROL DURING OOGENESIS AND OOCYTE MATURATION
Oogenesis can take a very long time: from several days in Drosophila, to several months or even years in amphibians such as Xenopus and Rana, to several decades in humans. Maternal transcripts are posttranscriptionally regulated from the moment of their synthesis, during oogenesis, often because they are not used for an extensive period after their production. Indeed, as will be seen, posttranscriptional control of these transcripts can play a key role in the coordination of developmental events.
Translational Repression During Oogenesis (Prematuration)
Maternal mRNAs are associated with discrete messenger ribonucleoprotein (mRNP) complexes in many animal species (reviewed in Johnstone and Lasko, 2001). One major component of these mRNPs are Y-box proteins—whose hallmark is a cold shock domain. In Xenopus, these proteins have been shown to function as translational repressors (reviewed in Matsumoto and Wolffe, 1998). Best characterized is FRGY2, which has been shown to translationally mask maternal mRNAs in oocytes (Bouvet and Wolffe, 1994; Table 1). Translational repression by FRGY2 requires transcription in vivo because injected in vitro synthesized transcripts are not repressed; this finding suggests that association with nuclear factors, possibly FRGY2 itself, occurs upon transcription to repress target transcripts.
Table 1. Factors Involved in Posttranscriptional Regulation in Drosophila and Xenopusa
Data relevant to post-transcriptional regulation
References available in text unless otherwise indicated. —, no known homolog; ?, function unknown; N/A, not applicable.
In addition, DEAD-box helicases such as Xp54, the Xenopus homolog of Saccharomyces cerevisiae Dhh1, have been shown to oligomerize on masked mRNAs and to repress their translation (Minshall et al., 2001; Minshall and Standart, 2004). Xp54 interacts in a complex with CPE binding protein (CPEB, Minshall and Standart, 2004). CPEB binds to a particular cis-acting element—the cytoplasmic polyadenylation element (CPE)—in the 3′-untranslated region (UTR) of target mRNAs (reviewed in Mendez and Richter, 2001). Translational repression is accomplished by Maskin, which interacts both with CPEB and with eIF4E (the m7G cap-binding protein), thus preventing eIF4E from interacting with eIF4G to recruit the 43S translation initiation complex (Fig. 1).
mRNPs are also found throughout Drosophila oogenesis, which proceeds in an assembly line manner through 14 morphologically distinct stages (reviewed in Spradling, 1993). Stage 1 begins with the production of a cyst of 16 cells that are interconnected by means of cytoplasmic bridges called ring canals. One of these cells differentiates into the oocyte, while the remaining 15 become polyploid nurse cells. The vast majority of the late-stage oocyte's cytoplasmic contents—including its mRNAs—are supplied by the nurse cells. Nurse cells actively transcribe and shuttle mRNAs into the oocyte during the early stages of oogenesis (stages 1 through 8), which last almost 2 days. Stages 9 through 10A—which last a total of 12 hr—see a dramatic increase in nurse cell transcription. Microfilament-dependent contraction of the nurse cells begins at stage 10B, taking approximately 6 hr to dump their cytoplasmic constituents into the oocyte before the nurse cells degenerate during stage 12.
In Drosophila, maternal mRNPs are found in electron-dense sponge-bodies, which can be visualized in both nurse cell and oocyte cytoplasm. The most specific marker for these sponge bodies is a protein called Exuperantia (Exu, Wilsch-Brauninger et al., 1997; see Table 1). Exu moves with the sponge bodies from the nurse cells, through the ring canals into the oocyte and is required for the proper localization of two developmentally important maternal transcripts, bicoid (bcd) and oskar (osk) (Berleth et al., 1988; St. Johnston et al., 1989; Wilhelm et al., 2000). Also present in these particles and directly interacting with Exu is the RNA-binding protein Ypsilon Schachtel (Yps), another member of the Y-box family of proteins (Thieringer et al., 1997; Wilhelm et al., 2000). Although a clear role for Yps in translational regulation has not been demonstrated, it has been shown to interact genetically with the fly CPEB homolog Oo18 RNA-binding protein (Orb) (Mansfield et al., 2002; see below). Another member of this complex is Me31B (maternal expression at 31B) (Nakamura et al., 2001), the Drosophila homolog of Xenopus Xp54 (see above). In flies, Me31B is required for the translational repression of osk (oskar) and Bicaudal-D (BicD) transcripts.
In early Xenopus oocytes, Xp54 localizes to the Balbiani body (also known as the mitochondrial cloud, which shares some structural homology to Drosophila sponge bodies) (Smillie and Sommerville, 2002). Many transcripts involved in axis determination and germ plasm assembly comigrate with the mitochondrial cloud to localize in the vegetal cortex of growing oocytes. Interestingly, some of these transcripts, like Xcat2 (which encodes a vertebrate homolog of Drosophila Nanos) and Xdazl, are translationally repressed until early embryogenesis (MacArthur et al., 1999; Houston and King, 2000). This finding suggests remarkable conservation of both the mechanism of translational regulation and function of maternal mRNPs across species.
Translational Regulation During Oocyte Maturation
A universal characteristic of immature oocytes is their arrest in prophase I of meiosis. This developmental arrest can last for a few days in Drosophila, a few years in amphibians or up to 50 years in humans. This arrest is thought to be required to coordinate the production of mature gametes only when the ideal environmental conditions, such as mate availability and seasonal cues, are met. Maturation, the final stage of oogenesis, is marked by release from the prophase I arrest and ends with an arrest at a second point in the cell cycle. Progression from meiotic prophase to metaphase requires the up-regulation of maturation promoting factor (MPF), a complex of Cdc2 kinase and cyclin B. MPF promotes entry into metaphase by phosphorylating factors involved in nuclear envelope breakdown, chromosome condensation, and spindle formation (reviewed in Kishimoto, 2003). An oocyte arrested at the second arrest point awaits release by activation and/or fertilization (discussed below).
In many species, maturation is marked by changes in the polyadenylation status (i.e., poly(A) tail length) and, consequently, the translation of several transcripts. Theories on how the length of a poly(A) tail influences the efficiency of translation are based on the “closed-loop model” (Munroe and Jacobson, 1990). This model states that an actively translated transcript is held in a circularized configuration where the poly(A) tail is bound by multiple molecules of poly(A) binding protein (PABP) that interact with eukaryotic initiation factor (eIF) 4G, which in turn binds eIF4E, the cap-binding protein. This eIF4E-4G complex is able to recruit the 43S preinitiation complex and begin translation. It is also thought that bringing the two ends of the transcript together allows the ribosome to easily reinitiate translation upon completion of a round of translation. An alternative model is that polyadenylation may induce ribose methylation of the cap (Kuge and Richter, 1995; Kuge et al., 1998). Most forms of translational regulation are thought to target initiation, particularly the eIF4E-4G interaction, because, in eukaryotes, this stage is often the rate limiting step (reviewed in Pestova et al., 2001).
Maturation-induced changes in polyadenylation status have been best studied in Xenopus (reviewed in Kishimoto, 2003). Before maturation, there is translational masking of transcripts that contain a CPE in their 3′UTR. As discussed above, masking is thought to be achieved by the formation of a stable bridge between CPEB, Maskin, and eIF4E. Upon progesterone stimulation, Aurora-A phosphorylates CPEB, allowing it to bind cytoplasmic polyadenylation specificity factor (CPSF; Fig. 1). CPSF then recruits poly(A) polymerase (PAP), which results in the lengthening of the transcript's poly(A) tail. This product is then bound by multiple copies of PABP. PABP interacts with eIF4G, allowing it to bind eIF4E in place of Maskin and, thus, recruits the 40S ribosomal subunit, leading to initiation of translation (Fig. 1).
The CPE-mediated pathway leads to the translational activation of several mRNAs that encode proteins such as c-Mos, Xkid, Ringo/Speedy, Cdc6, and Cyclins B1 and B4. These proteins have distinct functions during maturation. Ringo/Speedy—a novel protein—is thought to promote meiotic entry by binding and activating Cdc2 (Ferby et al., 1999; Lenormand et al., 1999). The translation of Xkid as well as of cyclins B1 and B4, on the other hand, is required for the subsequent meiosis I-to-II transition (Hochegger et al., 2001; Perez et al., 2002). The synthesis of Mos, an activator of the MAPK pathway, is crucial in preventing DNA replication during meiosis and for maintaining the metaphase II arrest (Dupre et al., 2002). The translation of Cdc6 is important for attaining competence to initiate replication later after fertilization (Lemaitre et al., 2002; Whitmire et al., 2002).
In mammals, maternal mRNA translation upon maturation also appears to involve regulation of masking factors and poly(A) tail length. For example, tissue-type plasminogen activator (tPA) is translationally dormant in mouse primary oocytes but is activated upon maturation (Huarte et al., 1992). Masking during early oogenesis is accomplished by the binding of an 80-kDa factor to a UA-rich element in the 3′UTR (Stutz et al., 1998). Unmasking can be achieved by displacement of this “silencing” factor using, for example, injected competitor transcripts. It has been suggested that the determining event in translational activation of tPA mRNA is removal of the 80-kDa factor, whereas extension of its poly(A) tail is merely a correlated process.
Virtually nothing is known about the mechanisms of oocyte maturation in Drosophila—in fact, this stage in oogenesis has not yet been clearly defined. The oocyte is in prophase I for most of oogenesis. However, at late stage 13, this arrest is relieved, the nuclear envelope breaks down, and meiosis progresses to metaphase I (King, 1970). During the final stage of oogenesis—stage 14—eggs remain arrested in metaphase I before activation. Using the above criteria, oocyte maturation in Drosophila, therefore, may be considered to occur during a brief, probably 1-hr period, spanning late stage 13 and early stage 14, thus taking approximately one tenth as long as maturation in Xenopus.
The trigger for Drosophila oocyte maturation, along with any downstream effects on posttranscriptional regulation, remains unknown. Drosophila has homologs of many of the proteins known to function in translational activation during Xenopus maturation (Table 1). In most cases, however, no analyses relevant to the issue of translational regulation have been conducted on these proteins during maturation, activation, and early embryogenesis. For instance, although a role for the CPEB homolog Orb in late oogenesis (stages 13–14) has not been investigated, Orb has been shown to function in translational regulation of osk mRNA during earlier stages. Osk protein, which is translated only in the oocyte posterior, is crucial for the assembly of the germ plasm, the cytoplasmic material that contains the abdominal and germ line determinants of the fly. Misexpression of Osk leads to ectopic abdominal structures; therefore, restriction of Osk synthesis to the posterior pole is crucial for normal development (Ephrussi and Lehmann, 1992). Although osk transcripts are actively transported to the oocyte posterior, mechanisms are in place both to prevent their translation while in transit and to promote their translation once they reach their destination. Orb has been shown to function in the latter mechanism (Chang et al., 1999; Castagnetti and Ephrussi, 2003): first, Orb complexes with osk mRNA in vivo; second, orb mutant egg chambers contain osk transcripts with shorter poly(A) tails than in wild-type; third, orb mutant egg chambers have low Osk protein levels. These results suggest that Orb may function in a manner similar to Xenopus CPEB, promoting cytoplasmic polyadenylation and translation of osk mRNA. Interestingly, Orb, Yps, and Exu are in an RNA-dependent complex and a yps mutation genetically rescues many of orb's oogenic phenotypes, including low Osk levels (Mansfield et al., 2002). Although this finding is consistent with the well-documented role of Y-box proteins as translational repressors, it will be interesting to see whether a similar relationship between Y-box proteins and CPEB exists in other systems.
Transcript Stability During Maturation
The mechanisms of mRNA degradation have been studied in most detail in budding yeast (reviewed in Parker and Song, 2004). Stable, polyadenylated mRNAs are kept in a closed loop configuration by an interaction between Pab1p, which is bound to the poly(A) tail, and eIF4G, a component of the cap-binding complex. Selective deadenylation of transcripts disrupts this interaction, leading to the loss of the decapping complex from the transcript. This loss exposes the 5′ cap to the decapping enzyme Dcp1p/Dcp2p. Uncapped mRNAs are then readily degraded by the 5′ to 3′ exonuclease Xrn1p. A secondary, deadenylation-dependent but decapping-independent pathway exists where transcripts are degraded in a 3′ to 5′ manner by a large multimeric complex called the exosome. In yeast, deadenylation is often the rate-limiting step for decay and, therefore, is a key target for regulation.
The majority of maternal mRNAs are stable during oogenesis. In Xenopus and mouse, “default deadenylation” of transcripts occurs upon oocyte maturation, typically of “housekeeping” mRNAs—such as those encoding actin and ribosomal proteins—that do not carry a CPE and, thus, are not cytoplasmically polyadenylated (Fox and Wickens, 1990; Varnum and Wormington, 1990; Paynton and Bachvarova, 1994). Although default deadenylation does not stimulate degradation of transcripts during oocyte maturation, it is a prerequisite for degradation later, at the blastula stage of embryogenesis (Audic et al., 1997; Voeltz and Steitz, 1998, see below). Default deadenylation is mediated by xPARN (Dehlin et al., 2000; Copeland and Wormington, 2001), which is functionally homologous to human PARN (Korner et al., 1998; Dehlin et al., 2000). Although not studied in detail, several “instructive” and “housekeeping” mRNAs are destabilized during maturation. These mRNAs include spindlin, β-actin, tPA, and c-myc mRNAs (Andeol et al., 1998; Oh et al., 2000).
Maturation-induced deadenylation/destabilization of transcripts has not been studied in Drosophila, but if it exists, it is unlikely to be exactly the same as in the vertebrate systems as there is no clear PARN homolog. However, during earlier stages of oogenesis, it is likely that this process occurs within the same mRNPs that function in translational regulation. For example, a protein involved in the stabilization of bcd transcripts, BSF (bcd stability factor, see below), has been localized in particles (Mancebo et al., 2001) that may in fact be the sponge bodies, which have been implicated in translational regulation (see above). Of interest, the budding yeast homolog of another sponge body component—Me31B—is Dhh1, an activator of decapping that is required for deadenylation-mediated mRNA decay (Coller et al., 2001; Fischer and Weis, 2002). That mRNA decay occurs in mRNPs is supported by recent data showing that, in yeast, mRNA decapping and degradation may be carried out in discrete cytoplasmic foci, called P-bodies, which contain components such as Dhh1 as well as RNA degradation intermediates (Sheth and Parker, 2003). As discussed above, the Dhh1/Me31B homolog in Xenopus, Xp54, is localized to Balbiani bodies, and there is also recent evidence that homologous factors may be found in discrete cytoplasmic foci in mammalian cells (reviewed in Wickens and Goldstrohm, 2003).
In summary, it is reasonable to speculate that the cellular and molecular functions of many of the proteins that have been implicated in posttranscriptional control during Xenopus oogenesis are likely to implement conserved roles in flies. However, to date, there is no direct evidence linking any of the Drosophila proteins to progression from prophase I to metaphase I at stage 13 (i.e., in oocyte maturation). Neither is it known whether mRNPs implicated in posttranscriptional control during oogenesis persist and function upon egg activation. Investigating these possibilities is likely to be a fruitful area for future research.
POSTTRANSCRIPTIONAL CONTROL DURING EGG ACTIVATION AND EARLY EMBRYOGENESIS
In all animals, the trigger for a mature egg to initiate the cellular and molecular events that permit embryogenesis is known as “activation,” which usually occurs simultaneously with fertilization. Most of our knowledge of egg activation comes from studies in organisms such as echinoderms and amphibians (reviewed in Runft et al., 2002). In these organisms, the canonical egg activation signaling pathway is thought to be initiated either by the induction of a cell surface receptor mediated by sperm–egg contact or by the diffusion of an activating particle brought about by sperm–egg fusion. A Src family kinase (SFK) then directs activation of phospholipase Cγ (PLCγ). (The exact involvement of SFK and PLCγ, however, is not confirmed in vertebrates.) PLCγ produces inositol-trisphosphate (IP3), which leads to the release of Ca2+ from the endoplasmic reticulum. This rise in the intracellular Ca2+ concentration is necessary and sufficient to reinitiate entry into the cell cycle and, in many species, causes cortical granule exocytosis, which establishes a block to polyspermy.
Drosophila egg activation occurs rapidly and internally within the female reproductive tract; hence, very little is known about its details. Interestingly, fertilization is not required for egg activation (Doane, 1960). Detailed in vivo analysis of Drosophila egg activation has shown that it begins upon oocyte entry into the lateral oviduct (Fig. 2, and see Heifetz et al., 2001). At this point, the vitelline membrane, the innermost layer of the eggshell, begins to become impermeable as key components such as sV23 protein are cross-linked. Meiosis resumes once the oocyte reaches the common oviduct. Fertilization occurs subsequently, when the egg reaches the uterus. Whether or not it is fertilized, within several minutes after egg laying, meiosis is completed and the eggshell is fully impermeable to aqueous solutions. The meiotic products condense to form one to three polar bodies in the dorsal anterior of the unfertilized egg. Egg activation also results in the depolymerization of the cortical microtubule network, presumably for use in the ensuing mitotic divisions (Theurkauf and Hawley, 1992; Page and Orr-Weaver, 1996).
In an effort to make Drosophila egg activation more amenable to experimental study, ways to in vitro activate oocytes have been developed (Mahowald et al., 1983; Page and Orr-Weaver, 1997). These methods were prompted by the observation that stage 14 oocytes were found to be wrinkled and flaccid in comparison to turgid newly laid eggs, suggesting that hydration might trigger activation. Furthermore, the oviduct matrix is hydrated in nonlaying virgin females compared with those that have begun egg laying, suggesting that its contents may be transferred to the oocyte (Mahowald et al., 1983). Indeed, dissected mature oocytes placed in a hypotonic buffer do complete several of the processes of egg activation, including the resumption of meiosis, the acquisition of impermeability and polysome assembly (Mahowald et al., 1983; Page and Orr-Weaver, 1997).
As mentioned previously, the period between egg activation and the MZT is a time of transcriptional quiescence; therefore, the instructive component of egg activation comes in the form of posttranscriptional control. In particular, egg activation has been known to stimulate changes in maternal transcript translation and stability in many different species (reviewed in Davidson, 1986). Unfortunately, the current protocol for Drosophila in vitro egg activation is not entirely useful for the study of posttranscriptional regulation. First, although bulk mRNA translation has been assayed in in vitro activated eggs (Page and Orr-Weaver, 1997), no comparative analyses have been conducted on in vivo activated eggs. Second, in vitro activated eggs do not undergo one of the key posttranscriptional processes—transcript destabilization—that is triggered in vivo (Tadros et al., 2003). It should, however, be possible to elucidate the pathway linking Drosophila egg activation to posttranscriptional regulation through a combination of two approaches: first, by refining in vitro activation such that it more closely mimics what occurs in vivo and, second, by applying what is known about these processes in other systems to flies.
Translational Activation During Egg Activation and Early Embryogenesis
Egg activation's effect on translational regulation is dual in nature: the activation of one subset of transcripts and the repression of another. Both processes are crucial to the control of early embryonic development. Translational activation has been studied in a host of species, including echinoderms, mouse, Xenopus, and Drosophila (reviewed in Davidson, 1986). In sea urchins, for example, there is a twofold increase in the poly(A) content of maternal mRNA over the first few cleavages. The rate of protein synthesis increases within minutes of fertilization, ultimately reaching a level 100-fold that in the unfertilized egg. This finding appears to be due in part to a 40- to 60-fold increase in the number of translationally active ribosomes, which are marked by their recruitment to polysomes. Likewise, after fertilization, clams demonstrate a three- to fourfold increase in protein synthesis and polysome content, whereas mice show a more modest, 40% increase in the rate of protein synthesis.
In Xenopus, some maternal mRNAs, such as those belonging to C11/eRF1, C12, activin receptor, and PP2Ac, are cytoplasmically polyadenylated and translated only after egg activation (reviewed in Paillard and Osborne, 2003). Regulation occurs by means of cis-acting sequences and trans-acting factors that are distinct from those operating during maturation. For example, UA-rich CPEs function during maturation, whereas U-rich or C-rich CPEs function after fertilization. In at least one case, however, embryonic translational control appears to be very similar to that induced by maturation: the same mechanism that regulates cyclin B1 translation during maturation operates during mitosis in the early embryo (Groisman et al., 2002). Specifically, cyclin B1 transcripts are polyadenylated and translated in a cell cycle-dependent manner that requires phosphorylated CPEB. Furthermore, blocking Maskin function keeps cyclin levels high. (That CPEB, Maskin, and cyc B mRNA are associated with the mitotic apparatus in Xenopus (Groisman et al., 2000) coupled with the observation that in Drosophila Cyclin B protein is also spindle associated in early embryos (Huang and Raff, 1999) raises the intriguing possibility that local cyc B translation may be conserved between these two systems.)
In Drosophila, only a handful of transcripts has been demonstrated to be translationally activated upon egg activation. These transcripts include bcd (Driever and Nusslein-Volhard, 1988b), nanos (nos; Wang et al., 1994), Toll (Tl; Toll protein is present, however, in stage 14 oocytes (Page, 1998), which suggests that it may be rapidly degraded upon egg activation and re-translated; Gay and Keith, 1992), hunchback (hb; Tautz and Pfeifle, 1989), caudal (cad; Dubnau and Struhl, 1997), smaug (smg; Dahanukar et al., 1999; Smibert et al., 1999), torso (tor; Casanova and Struhl, 1989), and string (stg; B. Edgar cited in Spradling, 1993).
Translational activation requires cytoplasmic polyadenylation; for example, bcd, torso, Toll, and hb mRNAs are polyadenylated upon egg activation (Fig. 3, and see Sallés et al., 1994; Wreden et al., 1997; Schisa and Strickland, 1998). Further evidence that polyadenylation leads to translational activation derives from the fact that overexpression of PAP in embryos causes both lengthening of the bcd mRNA poly(A) tail and an increase in Bcd protein accumulation (Juge et al., 2002).
The translation of bcd upon egg activation is crucial, because it sits at the top of a regulatory cascade that determines anterior patterning (Fig. 4): embryos from bcd null mutant females fail to develop head and thoracic segments (Frohnhöfer and Nüsslein-Volhard, 1986). The bcd mRNA is transcribed in nurse cells starting at stage 5 and is localized to the anterior of oocytes from stage 7 on (Berleth et al., 1988). During oogenesis, bcd mRNA is translationally silent and stable (Sallés et al., 1994; Surdej and Jacobs-Lorena, 1998). Egg activation results in the polyadenylation and translational activation of bcd mRNA (Sallés et al., 1994). The resulting graded Bcd protein distribution allows for concentration-dependent specification of developmental fates (Driever and Nusslein-Volhard, 1988a, b, 1989; Struhl et al., 1989).
A requirement for polyadenylation in translational activation of bcd mRNA has been demonstrated using transgenic transcripts (Sallés et al., 1994). The anterior defect of bcdE1 mutants is rescued by the injection of wild-type bcd transcripts but not of bcd mRNA lacking 537 nucleotides (nt) of the 3′UTR. The in vitro addition of 150–200 adenosine residues to the 3′ end of the injected mutant transcripts allows for partial rescue of the head defects. The specific cis-acting sequences required for polyadenylation and translation, however, have yet to be identified as replacement of the bcd 3′UTR with that from α-tubulin84B does not prevent its translational activation (Wharton, 1992).
Although trans-acting factors involved in transcript specific polyadenylation have also not been identified in Drosophila, a screen of maternal-effect lethal mutants has recovered two genes, grauzone (grau) and cortex (cort), which play a role in this process. Mutations in both genes perturb polyadenylation of several of the above-mentioned transcripts, including bcd where data suggest that this leads to its reduced translation (Lieberfarb et al., 1996). Since then, these genes have been cloned, revealing that Grau is a novel C2H2-type zinc-finger transcription factor whose only required function is the transcriptional activation of cort, which encodes a distant member of the Cdc20/Fizzy protein family (Chen et al., 2000; Harms et al., 2000; Chu et al., 2001). The molecular nature of these genes, however, does not reveal a direct role in cytoplasmic polyadenylation. This is also called into question owing to their pleiotropic mutant phenotypes in other aspects of egg activation such as meiosis, cytoskeletal rearrangement and transcript destabilization (see below) (Lieberfarb et al., 1996; Page and Orr-Weaver, 1996; Bashirullah et al., 1999).
Poly(A) elongation alone is not sufficient to activate translation of bcd mRNA. For instance, overexpression of PAP, which leads to the lengthening of the poly(A) tail of bcd mRNA in both oocytes and embryos, does not yield an increase in protein levels in oocytes (Juge et al., 2002). Furthermore, insertion of a short sequence within bcd's 5′UTR that is antisense to a portion of the open reading frame and is predicted to interfere with mRNA folding results in an transcript that is polyadenylated normally when injected into the anterior of wild-type embryos but is not translated (Verrotti et al., 1999).
Bcd's counterpart in posterior development is Nos which, when mutated, leads to absence of an abdomen (Lehmann and Nüsslein-Volhard, 1991). Restricting Nos expression to the posterior is crucial because ectopic Nos production in the anterior results in the formation of an abdomen at the expense of head and thoracic structures (Gavis and Lehmann, 1992). Although, nos mRNA is also translated upon egg activation, what distinguishes it from bcd is that the majority of nos transcripts are in the bulk cytoplasm (Bashirullah et al., 1999; Bergsten and Gavis, 1999); nevertheless, only those transcripts that are anchored in the pole plasm are translationally activated (Fig. 4, and see Gavis and Lehmann, 1994). The cis-acting sequences found within the 3′UTR have been shown to mediate translational repression in the bulk cytoplasm through factors such as Smaug and Cup (Dahanukar and Wharton, 1996; Smibert et al., 1996, 1999; Dahanukar et al., 1999; Nelson et al., 2004, see below). As can be seen by the above examples, although the mechanism of translational activation is currently not as clearly defined in Drosophila as in other systems, flies provide a crucial understanding of its function in development.
Translational Repression During Egg Activation and Early Embryogenesis
Several of the genes that are translated upon egg activation in flies themselves repress the translation of other maternal transcripts, restricting protein expression in a manner essential for proper development (Fig. 4). A prime example is provided by Bcd. In addition to its role as a transcriptional activator, Bcd acts as a regulator of translation: Caudal (Cad) is required for the establishment of the anteroposterior body axis (Mlodzik et al., 1985; Macdonald and Struhl, 1986). Maternal Cad protein is present in a posterior to anterior gradient soon after egg activation (Macdonald and Struhl, 1986). The exclusion of Cad from the anterior is achieved by Bcd, which represses the translation of cad mRNA by binding to a cis-element in its 3′UTR called the Bcd response element (Dubnau and Struhl, 1996; Rivera-Pomar et al., 1996; Chan and Struhl, 1997). Recently, it has been shown that Bcd likely mediates this repression by binding eIF4E and competing for binding to eIF4G at the 5′ end of the transcript (Niessing et al., 2002). The functional significance of this repression is highlighted by the fact that ectopic expression of Cad in the anterior causes the deletion of head and thoracic segments (Mlodzik et al., 1990).
Nos functions as a translational repressor of maternal hunchback (hb) mRNA in the posterior in an analogous manner to Bcd's role in repressing cad mRNA translation in the anterior (Fig. 4). In fact, genetic experiments have demonstrated that repression of maternal hb translation is Nos's only role in somatic patterning during embryogenesis (Hülskamp et al., 1989; Irish et al., 1989; Struhl, 1989). Very early in embryogenesis, when maternal hb mRNA is uniformly distributed throughout the cytoplasm, the Hb transcription factor begins to accumulate exclusively in the anterior, where it prevents the expression of genes involved in abdominal segmentation (Lehmann and Nüsslein-Volhard, 1987; Struhl et al., 1992). Ectopic expression of Hb in the posterior causes segmentation defects. The exclusion of Hb from the posterior is provided by the concerted action of Nos and Pumilio (Pum; Wharton and Struhl, 1991; Murata and Wharton, 1995; Sonoda and Wharton, 1999). Although Pum is present ubiquitously in the egg cytoplasm, repression of hb mRNA translation only occurs in the posterior because of the restricted localization of Nos. Pum binds a site within the Nanos response element (NRE) in the hb mRNA's 3′UTR. Nos associates with Pum and hb mRNA and recruits a third protein, Brain Tumor (Brat; Sonoda and Wharton, 1999). This complex causes the deadenylation and translational repression of hb mRNA in the posterior of the embryo (Wreden et al., 1997): The hb 3′UTR is polyadenylated when injected into the posterior of nos or pum embryos (i.e., derived from mutant females) but not when injected into wild-type embryos. Furthermore, an in vitro polyadenylated hb 3′UTR is completely deadenylated when injected into the posterior but not the anterior of wild-type embryos (Wreden et al., 1997). The exact mechanism of translational repression of hb mRNA by the Nos–Pum–Brat protein complex, however, is less clear than in the case of Bcd protein and cad mRNA. Experiments using transcripts with an internal ribosome entry site have shown that the repression does not act at the level of translational initiation (Wharton et al., 1998). Furthermore, Nos, Pum, and Brat may also repress translation by a poly(A)-independent mechanism (Chagnovich and Lehmann, 2001).
Nos and Pum act as translational repressors of at least one other transcript, cyclin B (cycB)—this time without the involvement of Brat (Asaoka-Taguchi et al., 1999; Sonoda and Wharton, 2001). cycB mRNA, which contains an NRE-like element in its 3′UTR (Dalby and Glover, 1993), is localized to pole cells during early embryogenesis. Both nos and pum mutants exhibit premature cycB translation in pole cells, beginning shortly after they form at 2 hr of embryogenesis (Asaoka-Taguchi et al., 1999); in wild-type, cycB mRNA is translationally silent for the first 11 hr of embryogenesis). Repression of cycB translation appears to be necessary to prevent premature mitosis in pole cells (Asaoka-Taguchi et al., 1999). Interestingly, Pum homologues in Xenopus, Caenorhabditis elegans, and budding yeast appear to be involved in translational repression and, in some cases, have also been shown to interact with CPEB (reviewed in Johnstone and Lasko, 2001). This finding suggests considerable conservation of this mode of translational repression.
The nos mRNA itself is also acted upon by mechanisms of translational repression (Fig. 4). Although what activates its translation in the posterior of the early embryo is unclear, more is known about how nos mRNA is translationally repressed throughout the bulk cytoplasm. The nos 3′UTR contains two stem–loop structures termed Smaug recognition elements (SREs), which are bound by the translational repressor Smg (Dahanukar and Wharton, 1996; Smibert et al., 1996, 1999; Dahanukar et al., 1999). Recent work suggests that Smg achieves this repression by binding Cup, which in turn interacts with eIF4E. This binding with eIF4E prevents its interaction with eIF4G, thereby disallowing assembly of the ribosome onto the nos transcript (Nelson et al., 2004). That repressed nos is also found associated with polysomes argues that, at least for a subset of nos transcripts, translation initiation is not blocked and repression occurs subsequently (Clark et al., 2000). The relief of Smg-mediated repression at the posterior is likely to involve Osk, as this protein interacts with Smg (Dahanukar et al., 1999). Osk may allow nos translation by preventing the Smg–Cup interaction, because Cup and Osk interact with the same region of Smg (Nelson et al., 2004).
Undoubtedly, the above examples illustrate the elaborate cascade of translational activation and repression that is set into motion by egg activation and how it affects the developmental fate of the embryo across its anterior–posterior axis (see Fig. 4).
Transcript Destabilization During Egg Activation and Early Embryogenesis
Egg activation also induces the destabilization of a subset of maternal transcripts. This phenomenon has been observed in a variety of model organisms such as mouse, rabbit, zebrafish, Xenopus, and Drosophila (Bachvarova and De Leon, 1980; Duval et al., 1990; Henrion et al., 1997, 2000; Bashirullah et al., 1999; Brunet-Simon et al., 2001; Kishida and Callard, 2001).
In Drosophila, several destabilized maternal transcripts have been studied in some detail, including Hsp70, Hsp83, nos, stg, cyclin B, twine, and Pgc (Fig. 5; Ding et al., 1993; Edgar and Datar, 1996; Bashirullah et al., 1999, 2001; Tadros et al., 2003; Semotok et al., in press). Unlike the case of translational regulation, the developmental function of transcript destabilization has not been demonstrated definitively. It is thought that, because maternal transcripts are often uniformly distributed throughout the egg's cytoplasm, it may be necessary to eliminate them to allow for spatially regulated zygotic transcription to have its proper effect. It is also possible, for example, that the half-life of maternal transcripts affects the timing of early development: increasing the maternal dosage of twine, which encodes a Cdc25 homolog, delays the onset of the MBT whereas reducing it has the opposite effect (Edgar and Datar, 1996). (Of interest, the functional significance of the degradation of cell cycle transcripts has been suggested in the adult mammalian cell cycle. The cell cycle-regulated expression of Cyclins A and B1 proteins correlates with the expression and half-lives of their mRNAs [Wang et al., 2000].) Destabilization also serves to localize maternal transcripts in the early embryo by a combination of their elimination in one subcellular compartment and their protection from degradation in another (reviewed in Lipshitz and Smibert, 2000). Hsp83, nanos, and Pgc are examples of transcripts that use this mechanism for their localization in the posterior of the embryo (Fig. 5; Bashirullah et al., 1999).
The trigger for maternal mRNA destabilization is egg activation. Dissected stage 14 oocytes incubated for several hours show no signs of destabilization, whereas transcripts are eliminated within 2.5 hr in activated, unfertilized eggs (Tadros et al., 2003). The kinetics of decay results from two activities: the first is maternally encoded, whereas the second requires zygotic transcription (Bashirullah et al., 1999, 2001; Cooperstock, 2002; Tadros et al., 2003). The maternal pathway is triggered by egg activation, whereas the zygotic pathway initiates in embryos 2 hr after fertilization. The combined activities of both pathways eliminates maternal transcripts by the MBT; if only the maternal or zygotic activities are functional, then transcript degradation occurs more slowly (Bashirullah et al., 1999). Hsp83 mRNA is targeted for degradation by both the maternal and zygotic degradation pathways; nanos mRNA is a target only of the maternal degradation pathway, because it is eliminated before 2 hr of embryogenesis; bcd transcripts appear to be targeted only by the zygotic pathway as they are stable in unfertilized eggs (Surdej and Jacobs-Lorena, 1998; Bashirullah et al., 1999). Interestingly, circumstantial evidence suggests that two degradation pathways—possibly maternal and zygotic as in flies—function in elimination of maternal transcripts in the mouse. It has been found that actin mRNA is partially degraded in the one-cell mouse embryo and falls to near the limit of detection in the late two-cell stage, whereas HPRT mRNA shows no change in early two-cell embryos but is deadenylated and declines greatly during the two-cell stage (Paynton et al., 1988). In aging unfertilized mouse eggs, however, most of these changes occur on a delayed schedule.
As for translational regulation, specific cis-acting elements, often in the 3′UTR, target certain maternal mRNAs for destabilization. For example, the HDE (Hsp83 degradation element, a 97-nt element) and the SRE (a 15-nt stem-loop) have been implicated in destabilization of Hsp83 and nos transcripts, respectively (Smibert et al., 1996; Bashirullah et al., 1999), whereas bcd transcripts contain a 43-nt element (the BIE, or bicoid instability element) that is sufficient to target an otherwise stable transgenic maternal mRNA for degradation (Surdej and Jacobs-Lorena, 1998). Just downstream of the BIE lies an NRE, which has been shown to be required for efficient deadenylation and destabilization (Gamberi et al., 2002). In addition, bcd transcripts appear to contain a separate “stability” element within stem loops IV/V, which acts to stabilize them during oogenesis (Mancebo et al., 2001, discussed above).
Redundancy seems to be a common theme among transcript destabilization elements: for example, deletion of the SRE or HDE from their respective full-length transcripts has only limited effects on transcript stability (Bashirullah et al., 1999; Cooperstock, 2002; Semotok et al., in press). Furthermore, no obvious similarities exist among the destabilization elements at the level of either primary sequence or secondary structure. The search for common cis-acting elements will be greatly aided by the use of gene expression profiling: recent genome-wide surveys have revealed that 15 to 30% of maternal transcript species are destabilized in early embryos (F. Menzies and H.D.L, unpublished observations; Arbeitman et al., 2002). Comparative analyses of these unstable transcripts may reveal conserved cis-acting elements.
Another major component of targeted transcript destabilization is the set of trans-acting factors that recognize these cis-elements. As with translational regulation, the approaches taken in Drosophila to elucidate these factors have been both genetic and biochemical. Genetic screens of maternal-effect lethal mutants have revealed several loci that, when mutated, result in degradation defects (Bashirullah et al., 1999; Tadros et al., 2003). As expected, general egg activation mutants such as cort and grau were recovered; however, the pleiotropic nature of these mutations suggests that these genes are not likely to be involved directly in degradation (as is the case with polyadenylation, mentioned above). Other mutants, such as pan gu (png), plutonium (plu), and giant nuclei (gnu), have only one other observed cellular defect, i.e., in the S-M transition at the end of meiosis (Tadros et al., 2003). Their gene products form a kinase complex that maintains cyclin protein levels, thereby ensuring the proper coordination between S and M phases during the early syncytial divisions (Elfring et al., 1997; Fenger et al., 2000; Lee et al., 2003; Renault et al., 2003). The role of Png, Plu, and Gnu in transcript degradation, however, has been demonstrated to be distinct from that in the cell cycle (Tadros et al., 2003).
To identify factors that interact directly with cis-elements, biochemical and/or candidate-gene approaches are necessary. For example, a trans-acting factor for the bcd mRNA IV/V region (described above) was discovered by means of ultraviolet cross-linking using ovary extracts. When the levels of this protein—BSF—are reduced by mutating the bsf gene, transgenic transcripts carrying the IV/V region are unstable during oogenesis (Mancebo et al., 2001).
Identifying factors involved in the destabilization process has been facilitated by an examination of mRNA decay pathways in other systems, particularly those in budding yeast (reviewed in Parker and Song, 2004). These have been summarized above in the discussions on oocyte maturation. As previously mentioned, in yeast, deadenylation is often the rate-limiting step for decay and, therefore, is a key target for regulation. This strategy is also true in Xenopus where control of a transcript's poly(A) tail length before and during the MBT affects that transcript's stability at the MBT (Audic et al., 1997; Voeltz and Steitz, 1998). Egg activation stimulates two different deadenylation activities in Xenopus (reviewed in Paillard and Osborne, 2003): the first requires A/U-rich elements and has been shown to operate on x-myc transcripts; the second uses the embryo deadenylation element (EDEN) and acts on Eg family and c-mos mRNAs. The trans-acting factor that recognizes EDEN and mediates deadenylation (and subsequent translational repression)—EDEN binding protein (EDEN-BP)—is a member of the Elav family of RNA binding proteins (Paillard et al., 1998). Of interest, it has been shown that EDEN can translationally repress reporter transcripts in Xenopus embryos as well as in Drosophila oocytes and embryos, suggesting that this mechanism is widely conserved (Ezzeddine et al., 2002). Recently, it been shown that a newly identified paralog of Bruno (Bru), Bru-3, shares 56% sequence similarity with EDEN-BP and specifically binds an EDEN-containing RNA probe (Delaunay et al., 2004). Bruno is a well-known repressor of osk mRNA translation in mid-oogenesis but, unlike Bru-3, is absent in embryos (Kim-Ha et al., 1995; Webster et al., 1997). It will be interesting, therefore, to assess whether Bru-3 functions in the deadenylation and subsequent destabilization and/or translational repression of maternal mRNAs upon egg activation in flies.
In Drosophila, deadenylation has been shown to play a role in the regulation of maternal hb and bcd mRNAs. As mentioned earlier, Nos and Pum promote the deadenylation of hb mRNA and, thereby, repress its translation in the posterior of the embryo. Deadenylation also plays a role in the destabilization of hb mRNA, which is partially stabilized in pum mutants (Fig. 3; Gamberi et al., 2002). The activity of Pum is not restricted to transcripts in the posterior of the embryo: bcd mRNA deadenylation and decay is also delayed in pum mutants. Deadenylation and degradation of bcd transcripts is mediated by the NRE in the bcd 3′UTR; the NRE was initially thought to have no biological significance, because bcd transcripts and Nos protein are localized to opposite ends of the embryo. Indeed, Pum's partner in targeting bcd transcripts for deadenylation remains unclear, because nos mutants only show a slight delay in this process.
How might trans-acting factors trigger the deadenylation-mediated destabilization of select transcripts? The major deadenylase activity in yeast is the Ccr4-Pop2-Not complex in which Ccr4 is the main catalytic subunit (reviewed in Parker and Song, 2004). Drosophila homologs of this complex exist and recently have been shown to have deadenylase activity (Temme et al., 2004). Recently, Smg, which had been characterized initially as a translational repressor (see above), has been found to be complexed with maternal Hsp83 mRNA in the early embryo and to be required for the deadenylation and destabilization of these transcripts (Semotok et al., in press). Smg physically interacts with the Drosophila Ccr4-Pop2-Not deadenylase complex in an RNA-independent manner, suggesting that Smg mediates transcript destabilization by recruiting the deadenylase complex to target mRNAs such as Hsp83 (Semotok et al., in press). Targeting of Smg to an otherwise stable mRNA, directs deadenylation and decay of that transcript. Furthermore, Smg is not synthesized until after egg activation and Smg protein disappears after the MBT (Dahanukar et al., 1999; Smibert et al., 1999). All of these results are consistent with the hypothesis that, initiating at egg activation, Smg is synthesized and acts to trigger deadenlyation and, thus, decay of transcripts such as Hsp83 by means of the “maternal” degradation pathway. Interestingly, Smg levels are reduced in png, plu, and gnu mutants, suggesting that the Png–Plu–Gnu kinase complex regulates translation of smg mRNA and/or stability of the Smg protein, thus linking cytoplasmic signaling upon egg activation to maternal transcript destabilization (W.T. and H.D.L., unpublished data). In light of Smg's previously studied role in translational repression of nos mRNA (see above), it appears that Smg is a multifunctional posttranscriptional regulator with particularly important roles during Drosophila egg activation. The presence of Smg homologs in other metazoa suggests that these factors may play a conserved role in posttranscriptional regulation of maternal transcripts.
Much remains to be answered with regard to both the mechanism and function of transcript destabilization during egg activation. Whereas the former can be addressed by continuing to apply what has been learned from organisms such as yeast and Xenopus to flies, doing the same with the latter will likely require the discovery of mutations that specifically perturb destabilization.
It is clear that posttranscriptional regulation is a key means of controlling gene expression during oocyte maturation and egg activation. During maturation, studies primarily in Xenopus have shown how translational regulation is important in controlling the cell cycle events of meiotic progression. Although very little is known about Drosophila maturation, there are several instances of conserved mechanisms of translational regulation earlier in oogenesis. Subsequently, upon egg activation, the specific activation and repression of translation is crucial for determining the developmental fate of various subregions of the early embryo. Transcript destabilization has also been shown to be regulated during these periods; however, its functional relevance to development has yet to be demonstrated comprehensively. This examination shows that, in many cases, the mechanisms of posttranscriptional regulation are re-used between maturation and activation. A primary example is the disruption of the “closed-loop” configuration as a key means of repressing translation. Furthermore, this review emphasizes the tight link between translation and transcript stability. Both processes heavily rely on the regulation of the poly(A) tail length; factors such as Nos, Pum, and Smg play roles in both; and, finally, both processes have been shown to occur in discrete cytoplasmic foci.
All of the data summarized in this review suggest that the fundamental cellular and molecular mechanisms used to regulate translation and transcript stability during oocyte maturation and egg activation are likely to be conserved throughout the animal kingdom. Studies using different model organisms will highlight conserved mechanisms as well as those that may be unique to a particular organism. In Drosophila, genetic approaches will facilitate elucidation of how specific processes are deployed and regulated in both a temporally and spatially specific manner. Thus, increasingly, analyses in different model organisms are likely to synergize to provide detailed insights into the mechanisms and, ultimately, the functions of posttranscriptional regulation in setting the stage for embryogenesis.
W.T. has been supported in part by an Ontario Graduate Scholarship and a studentship from the Ontario Student Opportunity Trust Fund - Hospital for Sick Children Foundation Student Scholarship Program. H.D.L. is Canada Research Chair (CRC; Tier 1) in Developmental Biology at the University of Toronto. Our research on transcript stability and localization is supported by funds from the CRC Program and an operating grant to H.D.L. from the Canadian Institutes of Health Research.
NOTE ADDED IN PROOF
Recent work has shown that Nanos is actually translated during stage 13 of oogenesis (Forrest KM, Clark IE, Jain RA, Gavis ER. 2004. Temporal complexity within a translational control element in the nanos mRNA. Development 131:5849–5857). This suggests that it may be the first example of a transcript regulated upon Drosophila oocyte maturation.