The localization of mRNAs in developing animal cells is essential for establishing cellular polarity and setting up the body plan for subsequent development. Cellular and molecular mechanisms by which maternal mRNAs are localized during oogenesis have been extensively studied in Drosophila and Xenopus. In contrast, evidence for mechanisms used in the localization of mRNAs encoded by developmentally important genes has also been accumulating in several other organisms. This offers the opportunity to unravel the fundamental mechanisms of mRNA localization shared among many species, as well as unique mechanisms specifically acquired or retained by animals based on their developmental needs. In addition to maternal mRNAs, the localization of zygotically expressed mRNAs in the cells of cleaving embryos is also important for early development. In this review, mRNA localization dynamics in the oocytes/eggs of Drosophila and Xenopus are first summarized, and evidence for localized mRNAs in the oocytes/eggs and cleaving embryos of other organisms is then presented.
Strategies for localizing specific mRNAs to particular sub-cellular domains are used on a number of biological occasions, including the mating decision in mother and daughter cells of budding yeast, early development, particularly oogenesis and embryogenesis, fibroblast cell migration, and synapse formation during brain development, to effectively target proteins to the sites where they function (Du et al. 2007; Becalska & Gavis 2009; Holt & Bullock 2009; Martin & Ephrussi 2009; Meignin & Davis 2010). The regulation of intracellular protein distribution by mRNA localization is considered to be advantageous for several reasons: (i) transport costs are reduced as a single mRNA molecule can be translated multiple times to yield numerous protein molecules; (ii) the translation and function of unwanted proteins in any other location than the target site is limited as translation is often blocked during the mRNA transportation process (Besse & Ephrussi 2008; Besse et al. 2009); and (iii) cells are capable of the temporal regulation of translation with high resolution, such as rapid on-site translation in response to an inductive signal (Martin & Zukin 2006; Wang et al. 2010).
mRNA localization relies on the formation of functional ribonucleoprotein (RNP) complexes, which often begin to form in the nucleus where specific RNA-binding proteins recognize mRNAs targeted for localization. After being transported into the cytoplasm, the resulting RNP complexes are remodeled by recruiting additional proteins (Kress et al. 2004). These proteins, called localization factors, precisely control and coordinate the multiple steps involved in mRNA localization, including recognition of mRNAs by their cis-acting sequences for selective transportation, organelle- and/or motor-based transport and anchorage along the cytoskeleton, and translation of mRNAs in the cortex (Du et al. 2007; Becalska & Gavis 2009; Holt & Bullock 2009; Martin & Ephrussi 2009; Meignin & Davis 2010).
Since the first visualization of localized mRNAs in the eggs of the ascidian, Styela plicata (Jeffery et al. 1983), intensive analyses of mRNA localization have been conducted using the oocytes and eggs of a wide range of animals, including Drosophila (St. Johnston 2005; Becalska & Gavis 2009), Xenopus (King et al. 2005; Kloc & Etkin 2005), several species of ascidians (Prodon et al. 2007; Sardet et al. 2007), and the cindarian Clytia (Momose & Houliston 2007; Momose et al. 2008). mRNA localization in oocytes and eggs is particularly important for early development because local proteins synthesized from maternally supplied mRNAs control development until the initiation of zygotic transcription. Local proteins play various roles in the execution of early developmental processes, such as fate determination and axis formation, at their targeted locations, leading to a diversity of cell types from the very early stages of development. Motor-based transport may be a fundamental mechanism to localize mRNAs in relatively large cells, such as oocytes and eggs, as it provides rapid and long-distance movement, although other conserved mechanisms are also evident. In contrast, relatively few examples of mRNA localization have been identified in cleavage- or later-stage embryos. One such example involves asymmetric mRNA segregation during cell division, in which specific mRNAs are localized within the mother cell and are inherited by only one of the daughter cells upon mother cell division (Broadus et al. 1998; Lambert & Nagy 2002; Hughes et al. 2004; Jedrusik et al. 2008; Takatori et al. 2010, described below for more detail). In the mechanism of localizing mRNAs at these later developmental stages, similar processes to those described in oocytes and eggs appear to be used.
In the following sections, mRNA localization dynamics in the oocytes/eggs of Drosophila and Xenopus are first summarized, and evidence for localized mRNAs in the oocytes/eggs and cleaving embryos of other organisms is then presented.
The four maternal mRNAs are transcribed in the ovarian nurse cells and, therefore, must be transported into the oocyte. bcd and grk mRNAs are thought to accumulate in the same RNP complex and to be transported on a dynein-mediated transport pathway toward the ring canals that connect the nurse cells to the oocyte (Fig. 1B, dark blue circles and green stars on left side) (Clark et al. 2007; Mische et al. 2007). In this pathway, concentrated microtubule arrays at the ring canals, which appear to project into the cytoplasm of nurse cells (Fig. 1B, two green lines on left side), serve to transport mRNA (Theurkauf et al. 1992). However, how the RNP complexes encounter the microtubule arrays within nurse cells is not understood, although it may occur through random movement and an entrapment process (Theurkauf & Hazelrigg 1998; Clark et al. 2007; Mische et al. 2007). Once the RNP complexes with specific mRNAs reach the ring canals, they undergo remodeling, which involves an exchange of localization factors, and sort mRNAs according to their target destination for further transportation within the oocyte (St. Johnston 2005; Mische et al. 2007; Becalska & Gavis 2009). Several osk RNP localization factors, including Tropomyosin II, Barentsz and Mago nashi, are thought to replace the dynein motors that mediate the nurse cell-to-oocyte transportation with a kinesin motor at the ring canals (Newmark & Boswell 1994; Erdélyi et al. 1995; van Eeden et al. 2001). Events that occur in nurse cells are important for correct mRNA localization within oocytes. For example, exogenously injected bcd mRNA is only properly localized to the anterior region of oocytes when injected into nurse cells, as mRNA directly injected into oocytes fails to localize (Cha et al. 2001).
In contrast to the longstanding model that microtubules are polarized along the anteroposterior axis of oocytes, with minus ends nucleated at the anterior pole and plus ends projecting towards the posterior pole, such that kinesin-associated osk mRNA is transported to the posterior pole, recent live imaging studies suggest that microtubules are largely randomly distributed (Fig. 1B, numerous green lines on the right side). It is proposed that osk mRNA is transported in a kinesin-dependent manner on randomly oriented microtubules with a slight bias towards the posterior during the mid-stage of oogenesis (Fig. 1B, red and light red circles on right side) (Zimyanin et al. 2008). The slight polarization bias is considered sufficient for osk mRNA to be enriched at the posterior region, because the transportation of mRNA requires a relatively long period (6–10 h). Localization of osk mRNA to the posterior cortex requires two steps: long-range, kinesin-dependent transportation throughout the oocyte into the posterior cytoplasm, as described above, and short-range, myosin-dependent translocation or entrapment at the posterior cortex (Krauss et al. 2009). The notion that microtubules are largely randomly distributed may be consistent with the above experimental result for bcd mRNA, in which exogenously injected bcd mRNA into oocytes accumulated at the nearest cortex in a dynein- and microtubule-dependent manner (Cha et al. 2001; Mische et al. 2007). Notably, mislocalization of bcd mRNA occurs even when it is injected into nurse cells that lack functional nurse cell-derived Exuperentia (Exu) proteins (Cha et al. 2001). Therefore, the presence of Exu allows bcd RNPs to locate and use a microtubule subpopulation with minus ends oriented toward the anterior cortex in the background of largely randomized microtubules. The discriminatory use of a microtubule subpopulation has recently been suggested for Vg1 mRNA localization in Xenopus oocytes, where Vg1 RNPs may use a microtubule subpopulation with its plus ends directed to the vegetal cortex (Fig. 1F, light blue stars and brown lines) (Messitt et al. 2008).
bcd mRNA continues to accumulate at the anterior cortex during the late-stage of Drosophila oogenesis in a dynein- and microtubule-dependent manner (Fig. 1C, dark blue circles and thin dark blue arrows). This late accumulation accounts for the majority of bcd mRNA present in the anterior region of the embryo (Weil et al. 2006) and uses distinct transport mechanisms from those involved during the mid-stage of oogenesis. Swallow and Staufen are only required for the late transportation pathway (St. Johnston et al. 1989; Weil et al. 2006, 2008). Swallow plays an indirect role in bcd mRNA localization by organizing the cytoskeleton, whereas Staufen contributes directly to transportation (Weil et al. 2010). nos, cyclin B, germ cell-less and polar granule component mRNAs, are also accumulated during the late-stage of oogenesis, but unlike bcd, these mRNAs are localized to the posterior region (Dalby & Glover 1992; Jongens et al. 1992; Wang et al. 1994; Nakamura et al. 1996).
Interestingly, nos RNPs are passively transported by diffusion and become trapped at the posterior cortex via posteriorly-localized germ plasm (Fig. 1C, pink stars and yellow circles) (Forrest & Gavis 2003). However, the localization of nos mRNA to the posterior cortex is not entirely microtubule-independent, as parallel arrays of microtubules run beneath the oocyte cortex (Fig. 1C, curved green lines) to create a cytoplasmic flow, termed ooplasmic streaming (Fig. 1C, light blue arrow), and facilitate the probability of RNPs encountering the germ plasm (Forrest & Gavis 2003; Weil et al. 2008). Recently, osk mRNA was shown to be transported to the posterior cortex not only during mid-oogenesis, as described above, but also at the late-stage of oogenesis (Fig. 1C, red circles) (Sinsimer et al. 2011). The transportation of osk and nos mRNAs during late-oogenesis requires the same localization factors Lost and Rumpelstiltskin (Sinsimer et al. 2011). Although only 4% of nos mRNA is localized to the posterior pole (Bergsten & Gavis 1999), after fertilization, nos in other locations is subject to deadenylation and degradation by the multifunctional RNA binding protein Smaug, resulting in greater than a hundred-fold enrichment of nos mRNA at the posterior pole (Zaessinger et al. 2006). The protection/degradation mechanism of localization mediated by Smaug is also used for hsp83 mRNA in the posterior pole plasm (Tadros et al. 2007).
mRNA anchoring in Drosophila
Recent developments in live-imaging techniques, particularly fluorescence recovery after photobleaching (FRAP), have made it possible to separate the anchoring and transport phases in the mRNA localization pathway. In FRAP, mRNA is considered to be stably anchored if the photobleaching of fluorescent-tagged mRNA at the localization site does not result in fluorescence recovery from adjacent regions. In contrast to the major role microtubule cytoskeleton plays in mRNA transportation, actin-dependent anchoring of mRNA to the oocyte cortex has been shown in several studies (Becalska & Gavis 2009; Kugler & Lasko 2009). Anchoring of mRNA is associated with two major functions: stable mRNA localization and translation. Although translation is often suppressed during mRNA transportation, upon arrival and anchoring of mRNA at the target destination, translation is initiated locally. Indeed, anchoring is required for the translation of nos and osk mRNA (Gavis & Lehmann 1994; Markussen et al. 1995; Rongo et al. 1995).
bcd mRNA is continuously transported to the anterior cortex in a dynein- and microtubule-dependent manner, as described above, until the completion of oogenesis, at which point the transport of bcd mRNA is shifted to a stable cortical actin-dependent anchoring process (Fig. 1C, dark blue circles and purple bars) (Weil et al. 2008). The stable anchoring of bcd mRNA is likely needed as mature oocytes can be dormant for weeks prior to fertilization. Upon egg activation, however, bcd mRNA is released from its anterior cortical tether as a result of re-organization of the actin cytoskeleton (Weil et al. 2008). The release of mRNA from a cortical tether upon egg activation is also observed in Xenopus, where Vg1 mRNA is released from the vegetal cortex (Yisraeli et al. 1990). The dispersal of Vg1 mRNA before initiating cell divisions may ensure that this developmentally important factor is inherited by the proper number of vegetal cells for establishing later embryonic patterning. Similarly, the dispersal of bcd mRNA during egg activation could be important for antero-posterior patterning of Drosophila embryos.
grk mRNA exhibits a dynamic distribution during the mid-stage of oogenesis: it is localized to the posterior at an early stage (Fig. 1A), subsequently moves to the anterior, and eventually accumulates around the oocyte nucleus in the dorso-anterior cortex (Neuman-Silberberg & Schüpbach 1993). The two steps involved in the transportation of grk mRNA towards the anterior and dorso-anterior regions are both dynein and microtubule dependent (Fig. 1B, green stars and arrows) (MacDougall et al. 2003). Of the three target sites, grk mRNA is stably anchored only in the dorso-anterior cortex within large electron-dense cytoplasmic structures known as sponge bodies (Delanoue et al. 2007; Jaramillo et al. 2008). Interestingly, the anchoring of grk mRNA is also dynein-dependent (Fig. 1B, orange rectangles) (Delanoue et al. 2007), indicating that dynein functions in two distinct roles: a motor in mRNA transportation and as an mRNA anchor. One grk mRNA localization factor, Squid (Sqd), is required for grk mRNA to enter the sponge bodies (Fig. 1B, light purple rectangles), as grk mRNA remains associated with transport particles and does not exhibit dorso-anterior localization in a sqd mutant (Delanoue et al. 2007).
mRNA transport and anchoring in Xenopus
Xenopus oocytes, like those of Drosophila, have multiple transport pathways with distinct mechanisms for mRNA localization that operate in different stages of development (Fig. 1D–F). A subset of maternal mRNAs is distributed differentially either at the animal or vegetal poles of oocytes. Whereas the mechanism(s) of mRNA enrichment in the animal hemisphere are unclear, two major distinct pathways are known to mediate mRNA transport to the vegetal cortex (Forristall et al. 1995; Kloc & Etkin 1995): the early MEssage TRansport Organizer (METRO) pathway that is activated in stage I-II oocytes (Fig. 1D,E) and the late pathway (Fig. 1F) that begins as the METRO pathway ceases to function.
The METRO pathway localizes several mRNAs, including nanos-related Xcat2 (Mosquera et al. 1993; Forristall et al. 1995; Kloc & Etkin 1995), Xdazl (Houston et al. 1998), and DEAD-South (MacArthur et al. 2000), many of which encode different families of RNA-binding proteins that are components of the germ plasm, and thus specify the germline cells in later development. The mRNAs within the germ plasm may not be translated for many months (MacArthur et al. 1999; Houston & King 2000). These early localized germline mRNAs are targeted to the germ plasm-forming METRO region within a subcellular structure called the mitochondrial cloud (MC) or Balbiani body (Heasman et al. 1984; Kloc et al. 1996; Zhou & King 1996a), which is enriched with endoplasmic reticulum (ER), Golgi, and mitochondrial membranes (Fig. 1D,E, green circles). The MC/Balbiani body is located asymmetrically on the future vegetal side of the nucleus (Fig. 1D). Initially, the asymmetrical distribution of the mitochondrial and organelle precursors of the MC/Balbiani body is established through the targeted growth of the organelle aggregate that contains the centrosome by dynein-dependent transport of these organelle precursors toward the centrosome (Heasman et al. 1984; Pepling et al. 1999; Kloc et al. 2002, 2004), while other organelle aggregates surrounding the nucleus remain the same size. The germline mRNAs localize at the MC/Balbiani body during stage I of oogenesis (Fig. 1D, red stars) by both diffusion and an entrapment mechanism (Kloc et al. 1996; Zhou & King 1996a,b; Chang et al. 2004), which is conserved between Drosophila and Xenopus for the localization of nos (Forrest & Gavis 2003) and Xcat2 mRNAs, respectively. The smooth ER-like membranes within the METRO region of the MC/Balbiani body are associated with these mRNAs and therefore may serve as the site for the entrapment (Chang et al. 2004). In stage II oocytes, the MC/Balbiani body associated with the localized mRNAs moves to the vegetal cortex (Fig. 1E).
Although the late pathway initiates in the nucleus, several factors are known to be recruited to Vg1 RNPs in the cytoplasm after their export from the nucleus and play roles in transportation and anchoring. For example, XStau, a Xenopus orthologue of the Drosophila RNA binding protein Staufen, which plays important roles in osk RNP localization (St. Johnston et al. 1991), interacts with Vg1 mRNA and kinesin I, and is required for proper Vg1 mRNA localization (Yoon & Mowry 2004). Another factor, Prrp, is proposed to act as a mediator for actin-based anchoring of Vg1 mRNA to the cortical cytoskeleton (Schlüter et al. 1997; Zhao et al. 2001). In addition to the two functions of anchoring described above, namely stable mRNA localization and translation, the reported anchoring of Vg1 mRNA to the vegetal cortex may indicate a third function: the bias for mRNA transportation to a particular location in the background of randomly oriented microtubules (Fig. 1F, green and brown lines), where Vg1 mRNA is thought to be transported bidirectionally between the vegetal cortex and a halfway point to the cortex in the vegetal hemisphere (Fig. 1F, light blue arrows), with mRNA anchoring at the vegetal cortex influencing the targeting localization in that direction (Messitt et al. 2008).
Despite the spatial and temporal differences and distinct mechanisms used between the early and late transportation pathways, a number of parallels have been also observed, suggesting that common cellular machinery mediates mRNA localization throughout oogenesis. For example, a few localized mRNAs classified as intermediate have been identified (e.g., fatvg [Chan et al. 2007]), which appear to have characteristics indicative of use of both the early and late pathways. In addition, a number of germline mRNAs, including Xcat2, Xpat, and Xlsirts, are capable of using the late pathway when injected into stage III/IV oocytes (Zhou & King 1996b; Hudson & Woodland 1998; Allen et al. 2003), suggesting that a fail-safe mechanism may operate to ensure that germline mRNAs, even when transcribed late, are segregated to the vegetal cortex for proper germline formation. The use of multiple localization mechanisms for a germline mRNA is observed in Drosophila, where osk mRNA is localized to the posterior cortex by means of microtubule-dependent transportation during mid-oogenesis and ooplasmic streaming at the late stage of oogenesis (Glotzer et al. 1997; Sinsimer et al. 2011), as described above.
mRNA localization in other organisms
The use of mRNA localization in oocytes/eggs as a strategy to achieve asymmetric protein distributions is not restricted to Drosophila and Xenopus, as it has been observed in a wide range of organisms, including the wasp Nasonia vitripennis (Fig. 2A–C), the cnidarian Clytia hemisphaerica (Fig. 2D–G), zebrafish (Fig. 2H–L), and several ascidian species (Fig. 2M–Q). In addition, similar mechanisms by which mRNAs are localized in the oocytes/eggs of Drosophila and Xenopus are used in other organisms as well. It is noteworthy that mRNAs for Vg1 and VegT orthologues in the directly developing frog Eleutherodactylus coqui, in contrast to Xenopus, are localized to the animal pole of oocytes (Beckham et al. 2003), indicating that widening mRNA localization studies to include other organisms is also important.
The study of mRNA localization in the wasp Nasonia provides an interesting opportunity to investigate the mechanism by which localized mRNAs pattern the embryo along the anterior-posterior axis, as well as specify the germline cells, in an arthropod model with a long-germ type of embryogenesis other than Drosophila (Fonseca et al. 2009), where the bcd gene is absent. In Nasonia oocytes, caudal (cad) and nanos (Nv-nos) mRNAs are localized to the posterior pole and direct posterior patterning (Olesnicky et al. 2006; Lynch & Desplan 2010), while giant mRNA is targeted to the anterior pole and mimics a part of bcd function with respect to anterior patterning (Brent et al. 2007). Further, orthodenticle1 (otd1) mRNA is localized to both the anterior and posterior poles and is required for anterior patterning in a manner reminiscent of the Bcd gradient, and for posterior development (Lynch et al. 2006). otd1 mRNA is first targeted to the posterior at the mid-stage of oogenesis (Fig. 2A, dark blue circles) and then localized also to the anterior cortex at the late-stage (Fig. 2B, dark blue circles) (Lynch et al. 2006). The localization of these four maternal mRNAs in Nasonia is achieved using at least two distinct mechanisms (Olesnicky & Desplan 2007). Anterior otd1 and giant, as well as posterior cad mRNA localization, requires microtubules (Lynch & Desplan 2010). The localization of giant and cad mRNAs is transient and they diffuse at later stages and form a gradient: a gradient distribution of cad is observed in the late stage of oogenesis (Fig. 2B, green stars) and that of giant in the freshly laid embryos (Fig. 2C, red circles) (Olesnicky et al. 2006; Brent et al. 2007). In addition, anterior localization of otd1 and giant mRNAs differs in that giant is localized in a perinuclear manner (Fig. 2A) but that otd1 is not (Olesnicky & Desplan 2007). In contrast to the anteriorly localized mRNAs, posterior localization of Nv-nos and otd1 mRNAs is actin-dependent (Lynch & Desplan 2010). The localization of Nv-nos mRNA to the posterior pole occurs at an early stage of oogenesis (mid-oogenesis) (Fig. 2A, pink stars) independently of ooplsamic streaming and this early localization does not appear to depend on the assembly of the germ plasm (Lynch & Desplan 2010). This mode of localization distinctly differs from that used for nos in Drosophila oocytes (described above). However, in the freshly laid eggs/embryos, Nv-nos as well as otd1 mRNAs accumulate at the oosome, a structure that is analogous to the germ plasm in Drosophila, in the posterior region (Fig. 2C, dark blue circles, pink stars, and yellow circle) (Lynch et al. 2006; Lynch & Desplan 2010).
Due to its phylogenetic position, the study of mRNA localization in the cnidarian Clytia also provides a unique opportunity to evaluate the extent to which the mechanisms of egg and embryo polarization via mRNA localization are conserved across metazoans. In Clytia, the two antagonizing Wnt receptor mRNAs, Frizzled 1 (CheFz1) and 3 (CheFz3) (Momose & Houliston 2007), and the Wnt ligand mRNA, CheWnt3 (Momose et al. 2008), are localized to distinct regions along the animal-vegetal (oral-aboral) axis in growing and maturing oocytes. All these mRNAs are located throughout the cytoplasm and most are concentrated around the GV at an early stage of oogenesis (Fig. 2D, dark blue circles, and pink and yellow stars). CheFz1 mRNA, then, adopts a graded distribution in the cytoplasm of the animal hemisphere by a mechanism that does not involve anchoring, but might involve preferential stabilization around the animally localized germinal vesicle (GV) (Fig. 2E, dark blue circles), while CheFz3 and CheWnt3 mRNAs are distributed at the periphery of the oocyte (Fig. 2E, pink and yellow stars) (Amiel & Houliston 2009). The polarized cytoplasmic distribution of CheFz1 directs the development of oral fate (Momose & Houliston 2007). During meiotic maturation, CheFz3 mRNA is localized and anchored underneath the vegetal cortex in a microtubule-dependent manner (Fig. 2F, pink stars), although not likely involving microtubule-directed transport but maybe involving mRNA degradation (Amiel & Houliston 2009), and opposes the function of CheFz1 to define an aboral region (Momose & Houliston 2007; Amiel & Houliston 2009). In contrast, CheWnt3 mRNA is concentrated in the animal cortex in a microtubule-independent manner (Fig. 2G, yellow stars), possibly via cortical/cytoplasmic flow during maturation, and induces oral fate (Momose et al. 2008; Amiel & Houliston 2009).
During early oogenesis in zebrafish, mRNAs of the germ plasm components vasa, nanos 1, and dazl localize to the MC/Balbiani body and are transported to the vegetal cortex (Fig. 2H, pink and yellow stars, dark blue and green circles) (Kosaka et al. 2007). The distribution pattern of these mRNAs in stage I oocytes is reminiscent of the Xenopus early METRO pathway. In fact, the zebrafish vasa 3′UTR directs GFP mRNA to the germ plasm in Xenopus oocytes, indicating that these animals share conserved localization machinery (Knaut et al. 2002). The localization of these three mRNAs to the MC/Balbiani body is abrogated in a bucky ball (buc) zebrafish mutant (Marlow & Mullins 2008; Bontems et al. 2009). buc mRNA itself is localized to the MC/Balbiani body and is considered to be a germ-cell determinant, as it is required and sufficient for germ cell formation (Bontems et al. 2009), similar to the factor osk in Drosophila. In the buc mutant, animal localization of several mRNAs, such as pou2 and Vg1, is also disrupted during the later stages of oogenesis (stages III and IV) (Marlow & Mullins 2008). An additional localization pathway may be operated during oogenesis as bruno-like (brul) mRNA localizes at the vegetal cortex in a timing later than vasa/nanos1/dazl mRNA localization (Fig. 2I, light blue stars and arrows), similar to the late pathway in Xenopus (Suzuki et al. 2000).
Vegetally-localized mRNAs in zebrafish, such as buc, vasa, and nanos 1, change their distribution in stage II oocytes (Fig. 2I, dark blue circles and arrows, and yellow stars and arrows) and eventually localize at the animal pole in late-stage oocytes or in freshly laid eggs (Theusch et al. 2006; Kosaka et al. 2007; Bontems et al. 2009). After fertilization, vasa and nanos1 mRNAs, which are distributed in a broad ring that encircles the animal pole of the egg (Fig. 2K, dark blue circles and yellow stars) and appear to be bound to a cortical actin network, move towards the periphery, leaving a large mRNA-free region at the animal pole, in a microtubule-dependent manner (Fig. 2L, dark blue and yellow vertical arrows), and further migrate laterally to accumulate at the forming cleavage furrows (Fig. 2L, dark blue and yellow horizontal arrows) (Theusch et al. 2006), where the germ plasm forms. In contrast, other vegetally-localized mRNAs such as dazl and brul, remain at the vegetal cortex until the late-stage of oogenesis (Fig. 2J, light blue and pink stars); however, following fertilization, they translocate along the plane of the cortex towards the animal pole and are incorporated in the germ plasm (Fig. 2L, pink stars and arrows) (Theusch et al. 2006). Due to the use of the different localization pathways after vegetal localization, these two classes of germline mRNAs (i.e., vasa and nanos 1 vs. dazl) are localized in distinct regions within the germ plasm (Theusch et al. 2006).
Differential distribution of different classes of germline mRNAs within the germ plasm has also been observed in other organisms. In Xenopus, Xcat2, Xpat, and DEAD-South mRNAs are associated with germinal granules, while mRNAs such as Xlsirts and Xdazl reside in the matrix between germinal granules (Kloc et al. 2002). These mRNAs all use the same METRO pathway early to localize at the vegetal cortex (described above). Likewise, non-coding RNAs of Drosophila, including mitochondria large ribosomal RNAs, which are required for germ cell formation (Kobayashi et al. 1993), are often detected in association with polar granules (Kashikawa et al. 1999), while other RNAs are found in the matrix. In addition to the active cytoplasmic transportation of mRNAs, a protection/degradation mechanism operates for zebrafish vasa mRNA to ensure its germline-specific expression during the cleavage stages (Wolke et al. 2002).
Two classes of localized maternal mRNAs have been identified in ascidian species, an immediate sister group to vertebrates. Type I postplasmic/PEM RNAs, such as PEM1, macho1, PEM3, and POPK1, are concentrated at the vegetal cortex of fertilized eggs by microfilament-driven cortical contractions (Fig. 2P, light blue stars and arrows, and green stars and arrows), and are then relocated towards the future posterior pole by sperm-aster microtubule-driven translocations and microfilament-dependent relaxation (Fig. 2Q, light blue stars and arrows, and green stars and arrows) (Roegiers et al. 1999; Sasakura et al. 2000; Yamada et al. 2005; Kumano & Nishida 2007; Prodon et al. 2007; Sardet et al. 2007; Paix et al. 2009). The second class, type II postplasmic/PEM RNAs, which include vasa and PET1, are distributed evenly throughout the cytoplasm (Fig. 2P, red circles), and progressively localize to the posterior pole during embryogenesis (Fig. 2Q, red broken arrows) (Prodon et al. 2007; Sardet et al. 2007). This class, however, may be better categorized as type I in that in some ascidian species, numerous postplasmic/PEM RNAs including those previously regarded as type II exhibit significant accumulation at the vegetal cortex of zygotes (Fig. 2P, red broken arrows), as well as uniform distribution throughout the cytoplasm (Paix et al. 2009). A cortical structure called the centrosome attracting body (CAB) (Hibino et al. 1998), which comprises three distinct parts: the sub-membranous polarity plot layer (Patalano et al. 2006), the cER-mRNA domain (Sardet et al. 2003), and vasa-positive granules (Paix et al. 2009), forms in association with type I and type II postplasmic/PEM RNAs in the posterior pole as development proceeds, and is inherited by the posterior-most blastomere during the cleavage stages. The cER-mRNA domain and vasa-positive granules are already present but dispersed in oocytes (Paix et al. 2009), and after fertilization they get compacted together to form an ascidian equivalent to germ plasm within the CAB (Iseto & Nishida 1999; Fujimura & Takamura 2000; Takamura et al. 2002; Shirae-Kurabayashi et al. 2006; Paix et al. 2009).
In stage I oocytes, Ciona intestinalis PEM3 (Ci-PEM3) mRNA is found located around the GV (Fig. 2M, light blue stars) (Prodon et al. 2006, 2007). However, in full-grown stage III oocytes, Ci-PEM3 and Ci-PEM1 mRNAs are both distributed as cortical patches, which are found uniformly at the entire cell cortex (Fig. 2N, light blue and green stars) (Prodon et al. 2006), suggesting that Ci-PEM3 mRNA is transported from the GV periphery to the cell cortex (Fig. 2N, light blue arrows) as oocytes grow. Although the mechanisms underlying this transportation are not known, the MC/Balbiani body is not observed in ascidian oocytes (Prodon et al. 2006). Prior to germinal vesicle break down (GVBD), Ci-PEM1, Halocynthia roretzi PEM1 (Hr-PEM1), Phallusia mammillata PEM1 (Pm-PEM1) and Pm-macho1 mRNAs are distributed as patches, which are located uniformly at the cell periphery in full-grown oocytes (Prodon et al. 2008; Paix et al. 2011). Ci- and Hr-PEM1 mRNAs are anchored to a network of rough cER in a microfilament-dependent manner (Prodon et al. 2005, 2008). After GVBD, these mRNAs are excluded from the animal pole by vegetally directed and microfilament-driven surface, cortical, and cytoplasmic flows and exhibit a graded distribution along the animal-vegetal axis (Fig. 2O, light blue and green stars) (Prodon et al. 2008). Following fertilization, the mRNAs are further concentrated vegetally (Fig. 2P) and subsequently translocated to the future posterior pole (Fig. 2Q) as described above, while maintaining their association with cER. To date, many localized mRNAs that are associated with cER have been identified in addition to Ci- and Hr-PEM1, and include Hr-macho1, Hr-POPK1, Hr-ZF1, Hr-Wnt5, Pm-PEM1, and Pm-macho1 (Sardet et al. 2003; Nakamura et al. 2005; Paix et al. 2009). A recent study suggests that localized mRNAs are translated on or near the cER (Paix et al. 2011). In contrast to type I postplasmic/PEM RNAs, Hr-vasa mRNA (type II) is located around the GV in full-grown stage-III oocytes (Fig. 2N, red circles) and distributed uniformly throughout the cytoplasm after GVBD (Fig. 2O, red arrows) (Prodon et al. 2009). Hr-vasa mRNA is then thought to accumulate at the future posterior pole after fertilization. Pm-vasa mRNA, in contrast to Hr-vasa mRNA, accumulates at the vegetal cortex (Fig. 2O, red broken arrows) and also displays uniform distribution throughout the cytoplasm of non-fertilized eggs (Paix et al. 2009). Interestingly, Pm-vasa, Pm-POPK1, and Pm-PEM3 mRNAs are localized in granules in non-fertilized eggs (Fig. 2O, two adjacent red circles), which represents a different distribution from that of Pm-PEM1 and Pm-macho1 mRNAs on the cER network (Paix et al. 2009), as described above. Therefore, ascidian germline mRNAs also appear to reside in distinct subdomains within the germ plasm, a conclusion that is supported by the recent findings that PEM1 plays a role in germline transcriptional repression (Kumano et al. 2011; Shirae-Kurabayashi et al. 2011), and POPK1 regulates CAB/germ plasm formation (Nakamura et al. 2005).
mRNA localization dynamics in cleavage stage embryos
Localized maternal mRNAs in oocytes/eggs are inherited by particular subsets of blastomeres within the embryo through compartmentalization resulting from mitotic cell divisions after fertilization, and specify cell fates and embryonic body axes when translated in the inherited regions or cells. Occasionally, these asymmetrically distributed maternal mRNAs remain anchored to their target sites and segregate into specific cell types during cell cleavage. One such example is maternally supplied germline mRNAs, which are inherited only by germline, and maintain their association with the germ plasm, and play key roles in germ cell formation during embryogenesis (Fig. 3A–D, green stars). Even in animals other than those described above, such as Caenorhabditis elegans, specific mRNAs are maintained in the germline (Fig. 3E–I, green stars) and localize with germ plasm in cleaving embryos (Seydoux & Fire 1994; Schisa et al. 2001). In contrast to these maternal mRNAs, other populations of mRNAs are not localized in oocytes/eggs, but are found localized for the first time in the cleavage stages, and thus do not depend on the localization systems that operate in the oocytes/eggs for asymmetric distribution in the later stages of development. These mRNAs are localized in a polarized manner to one side of the mother cell and are inherited by only one of the daughter cells through cell divisions. Here, three examples of such mRNAs are presented: centrosomally localized mRNAs in the molluscs, Ilyanassa obsolete (Fig. 3J–L); Not mRNA in the ascidian, Halocynthia roretzi (Fig. 3M–P); and cdx2 mRNA in the mouse (Fig. 3Q,R).
In early development of the snail Ilyanassa, a number of mRNAs display specific localization to the centrosomes during interphase in cleaving embryos (Fig. 3J, pink stars) (Lambert & Nagy 2002; Kingsley et al. 2007), where the cells at the four-cell stage, called macromeres, divide synchronously and successively towards the animal pole to produce sets of four smaller cells termed micromeres. These localized mRNAs include several developmental patterning genes known to control cell fate specification in the embryos of Ilyanass, as well as those of other organisms (Lambert & Nagy 2002; Kingsley et al. 2007; Rabinowitz et al. 2008; Swartz et al. 2008; Rabinowitz & Lambert 2010; Chan & Lambert 2011). The centrosomally localized mRNAs subsequently move to the cell cortex during prophase (Fig. 3K) and are inherited by one of the daughters through cytokinesis (Fig. 3L). The pattern of mRNA segregation depends on the species of mRNA, the particular cells, and the stage of development, which, in combination with all of the patterns for localized mRNAs, results in distinct distribution of mRNAs in the different quartets of micromeres. In Ilyanass, the migration of mRNAs, which are initially distributed diffusely throughout the cytoplasm, to the centrosomes is shown to be microtubule-dependent, whereas cortical localization in the second phase of intracellular movement is mediated by actin filaments and requires the previous centrosomal accumulation of mRNAs (Lambert & Nagy 2002). Localization to the centrosomes may represent an analogous situation to what occurs in vertebrate oocytes where the MC/Balbiani body, which is associated with centrosomes, is responsible for mRNA localization (described above). Similarly, in oocytes of the bivalve mollusc Spisula solidisima, several mRNAs are also localized to centrioles (Alliegro et al. 2006; Alliegro & Alliegro 2008). In these regards, the mechanism of mRNA localization identified in Ilyanassa could be more widespread across metazoans (Lambert 2009). A similar mRNA localization pattern was recently observed during the cleavage stages of the closely related mollusk species, Crepidula fornicate (Henry et al. 2010).
In cleaving embryos of the ascidian Halocynthia, asymmetrically localized Not mRNA within mesendoderm mother cells at the 16-cell stage is responsible for the segregation of germ layer fates into daughter cells, mesoderm and endoderm (Takatori et al. 2010). Not mRNA is partitioned into the mesoderm cell where it promotes mesoderm and suppresses endoderm fates. The migration of the nucleus is essential for this partitioning process. While transcribing Not mRNA, the nuclei of mesendoderm mother cells migrate towards the future mesoderm-forming regions (Fig. 3M, light blue circle and pink stars), where Not mRNA will be released from the nucleus into the cytoplasm (Fig. 3N, pink stars). The nuclei subsequently return to the central parts of the cells (Fig. 3O, light blue circle), leaving Not mRNA at the mesoderm poles (Fig. 3O, pink stars), and form the cell division planes that separate Not mRNA-containing mesoderm from the endoderm (Fig. 3P). Nuclear migration also serves an essential role in mRNA localization in Drosophila, where grk mRNA associates with the oocyte nucleus (Neuman-Silberberg & Schüpbach 1993; Saunders & Cohen 1999) and is mis-localized when the nuclear migration to the dorso-anterior cortex is disrupted (Lei & Warrior 2000). Although grk mRNA is transcribed in the oocyte nucleus (Saunders & Cohen 1999), as well as in nurse cells (Thio et al. 2000), the nurse cell-derived grk mRNA is sufficient for proper dorso-ventral patterning in Drosophila oocytes (Cáceres & Nilson 2005). In both the cases of Halocynthia and Drosophila, the anchoring of mRNA following transportation is critical for its correct localization. Wnt5 signaling might be involved in anchoring Not mRNA to the cortex on the mesoderm side (Fig. 3N) (Takatori et al. 2010). Therefore, the localization of mRNA by means of nuclear migration and subsequent anchoring to the cortex may be a conserved mechanism during many developmental stages and among various embryos. A recent study of Drosophila embryos also reported a role for the nucleus in the deposition of mRNA-containing germ plasm in the pole cells of germline precursors (Lerit & Gavis 2011). Of equal interest, Not mRNA appears to colocalize with mitochondria (Takatori et al. 2010), which is a known target site for mRNA localization (King et al. 2005; Kloc & Etkin 2005).
The localization of mRNA that is important for early development is observed even in the mouse embryo, although its involvement seems to be limited because of its developmental liability. In the first cell fate decision in early embryogenesis, namely the segregation of pluripotent inner cell mass (ICM) and extra-embryonic trophectoderm, transcripts for the transcription factor Cdx2 are localized to the apical domains of 8- and outer 16-cell blastomeres (Fig. 3Q, pink stars), where they are partitioned by the outside daughter cells in cell divisions (Fig. 3R) and appear to play a role in trophectoderm specification (Jedrusik et al. 2008).
As presented in this review, several intriguing similarities and differences in mRNA localization dynamics and anchoring exist between different stages of development in the same species and also between species. However, more detailed analyses of these dynamics will be necessary, particularly in organisms other than Drosophila and Xenopus, in order to clarify which mechanisms of mRNA localization are fundamental across metazoans and which have evolved to meet species specific-developmental needs. The application of recent visualization techniques to monitor mRNA localization events in real time at high spatial and temporal resolutions are expected to facilitate our understanding of the underlying mechanisms, as many localization patterns in lesser-studied organisms have only been observed as the end-stage accumulation of mRNAs by in situ hybridization. Models of mRNA localization and anchoring have been greatly revised and improved by the use of live imaging techniques in Drosophila (Becalska & Gavis 2009). Even in well-studied organisms such as Drosophila and Xenopus, the development of novel fluorophores and microscopes may be key to monitor the entire path of mRNA localization at or near physiological levels. Such developments may also help to answer several basic questions, such as the timing of recruitment of specific mRNAs or localization factors to RNPs and whether the same localization factors are used for different RNPs, by allowing the labeling of different mRNAs and proteins simultaneously.
The identification of localization factors would also aid in better understanding the mechanisms by which mRNAs are localized. Although several localization factors have been identified in Drosophila and Xenopus, almost no such factors have been found in other organisms. Expanding the list of known localization factors is important for understanding not only the mechanisms of mRNA localization regulation, but also how distinct pathways are separated mechanistically and/or mRNA localization and translational repression are coordinated. To date, Staufen is the only well-studied localization factor that is conserved between different species. In parallel with efforts to identify more localization factors, extensive analyses on cis-acting elements within localized mRNA molecules would also help clarify how localization factors regulate localization processes. However, such analyses are difficult because the cis-acting elements involved in mRNA localization do not share significant primary sequence similarity and often differ substantially in length. Recently progress has been made, as a model for a complete link between cis-acting elements of minus-end-directed mRNAs and microtubules has been proposed, in which divergent cis-acting signals are recognized by the same factor, called Egalitarian (Dienstbier et al. 2009).
I would like to thank Christian Sardet for critically reading the manuscript and the members of Hiroki Nishida’s laboratory for many stimulating discussions. The author acknowledges the support from the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT) and the Japan Society for the Promotion of Science (JSPS) to G.K.