The process of localizing a maternal messenger RNA in Xenopus oocytes

Xenopus oocytes have a clearly defined animal-vegetal (a−v) polarity. In previtellogenic oocytes, the mitochondrial cloud and the chromosomal attachment site in the germinal vesicle (gv) define a polarity that is proposed to be the future a−v axis (Al-Mukhtar and Webb, 1971; Heasman et al. 1984). Pigment granules migrate to the animal hemisphere in middle-stage, vitellogenic oocytes (Dumont, 1972), followed by the asymmetric deposition of dense yolk platelets in the vegetal hemisphere (Danilchik and Gerhart, 1987). Around this same time, the germinal vesicle migrates to near the animal pole from its location in the center of the oocyte. After fertilization, the dorsal-ventral axis is generated orthogonally to the a−v axis (Vincent and Gerhart, 1987), and it is along these primary axes that the germ layers become defined.

In addition to these morphological markers, numerous molecular asymmetries exist (see Gerhart, 1979). In particular, a small class of RNAs have been identified which are unevenly distributed in oocytes and eggs (Rebagliati et al. 1985; King and Barklis, 1985). The best studied of these RNAs is Vgl, a vegetally localized mRNA whose protein product is a member of the TGF-μ family (Weeks and Melton, 1987). Vgl is initially synthesized in previtellogenic oocytes and is uniformly distributed until the oocyte reaches the end of stage III (Melton, 1987). By the middle of stage IV, Vgl mRNA is found almost exclusively in a tight shell at the vegetal cortex (Melton, 1987; Yisraeli and Melton, 1988). The RNA is released from its tight localization after maturation but remains in the vegetal hemisphere, presumably as a result of cellularization, throughout embryogenesis (Weeks and Melton, 1987).

We are interested in understanding how polarity is generated and interpreted in oocytes and eggs. By culturing oocytes in vitro, we have been able to analyze and interfere with the localization of Vgl mRNA. Injected, in vitro synthesized Vgl mRNA can be localized in cultured oocytes in a manner identical to the localization of the endogenous message, suggesting that all the information necessary for proper localization is encoded by the RNA itself. Insoluble pellets of detergent extracts of oocytes, which maintain most of the cytoskeletal elements of the cell, preferentially retain Vgl mRNA. Using cytoskeletal inhibitors in middle- and late-stage oocytes, we have been able to distinguish two separate steps in the translocation process. The movement of Vgl RNA to the vegetal hemisphere is inhibited by drugs which depolymerize microtubules, but not by those that affect microfilaments. Anchoring of the Vgl mRNA, however, can be disrupted by microfilament inhibitors, but not microtubule inhibitors. Thus, at least in the case of Vgl RNA, the oocyte utilizes common cytoskeletal elements to help generate specific asymmetries in the cell.

Large oocytes cultured in the presence of vitellogenin-containing serum continue to grow and incorporate vitellogenin (Wallace et al. 1980). In order to make use of this in vitro system for our studies, it was important to show that smaller oocytes retained their ability to localize endogenous Vgl message in culture. As shown in figure 1, oocytes grown in vitro in the presence of frog serum containing vitellogenin localize endogenous Vgl mRNA in the same pattern as oocytes grown in vivo. Oocytes incubated in either saline or medium w’ithout serum demonstrate no localization of the endogenous mRNA (data not shown). Only those oocytes grown in the presence of vitellogenin increase in diameter, a result of micropinocytosis of vitellogenin, the precursor for the yolk proteins (Wallace et al. 1970). In situ hybridization to oocytes of different stages (Melton, 1987; Yisraeli and Melton, 1988; unpublished observations) suggests that localization of Vgl RNA occurs within a small window of oogenesis between the end of stage III and the middle of stage IV. In vivo, this period of growth probably requires several weeks to a month (Keem et al. 1979). Oocytes cultured under the conditions described here grow at least three to four times faster. Nevertheless, the progressive accumulation of Vgl mRNA along the vegetal cortex coupled with the graded loss of the mRNA from the cytoplasm in an animal-to-vegetal direction is identical in vivo and in vitro.

The endogenous, steady-state level of Vgl mRNA remains constant from previtellogenesis until after maturation (Melton, 1987). In addition, as mentioned above, localization of Vgl mRNA occurs in a particular pattern, regardless of whether that movement takes five to six days or three to four weeks. Finally, the accumulation of Vgl message at the vegetal cortex is accompanied by a graded disappearance of message, first in the animal hemisphere and then throughout the oocyte, as opposed to a general loss of message everywhere. These observations suggested that the localization of Vgl mRNA is not the result of specific degradation of the message away from the vegetal cortex, but rather the accumulation of Vgl message at its proper location. In order to explore this possibility further, and to begin to determine how the specificity of localization is achieved, we synthesized Vgl and globin RNAs in the presence of both radioactively labelled UTP and GpppG (a cap analog whose incorporation at the 5’ end of synthetic RNA transcripts prevents 5’ exonucleolytic degradation). Late stage III oocytes injected with synthetic Vgl message were capable of localizing the exogenous message when cultured in vitellogenin-containing serum for five days (figure 2). The kinetics and pattern of this localization are virtually identical to those of the endogenous message and the fact that the injected RNA is relatively stable throughout the entire incubation period further argues that localization is not due to degradation or differential stability (Yisraeli and Melton, 1988). Injected globin RNA, however, becomes distributed throughout both the animal and vegetal hemispheres over the course of the culture period. The information necessary for the specific localization of Vgl mRNA is thus present in the naked Vgl RNA molecule itself. Studies are now under way to define further the cry-acting sequences which are involved in the process. Initial studies using a deleted Vgl RNA lacking 62 nucleotides from the 5’ end, including the start codon and putative signal sequence, have shown that this region is not necessary for the proper localization of the RNA (Yisraeli and Melton, 1988).

The striking distribution of Vgl mRNA at the vegetal cortex suggested that cytoskeletal elements might be involved in its anchoring. A large number of RNAs have been shown to be associated with a detergentinsoluble fraction of extracts from different cells, a fraction that has been shown to contain cytoskeletal elements such as microtubules, microfilaments, and various intermediate filaments (e.g. Lenk et al. 1977; Jeffery, 1984). If Vgl message is specifically associated with the cytoskeleton in these late-stage oocytes one would expect that Vgl but not non-localized RNAs would be associated with the detergent-insoluble pellet. As shown in figure 3, Vgl mRNA is highly associated with the cytoskeletal pellet, with less than 10% of the total Vgl mRNA in the soluble fraction. This association is not a result of non-specific trapping, because fibronectin mRNA, which is approximately twice the size of the Vgl message, is found almost exclusively in the soluble fraction. Less than 2% of the total oocyte poly(A)+ RNA is found in the detergent-insoluble pellet, and equal oocyte equivalents of pellet and soluble RNA are compared on the gel. Translatability of a message appears not to be important for its association with the cytoskeletal pellet because Vgl, fibronectin, and several other RNAs are all translated in oocytes but only Vgl mRNA is found in the pellet (data not shown). These results are similar to those recently reported by Pondel and King (1988).

The above results suggested that Vgl mRNA is associated with cytoskeletal elements at the cortex in latestage oocytes. In order to determine what these elements are and how they are interacting with Vgl mRNA, we treated late-stage oocytes with various cytoskeletal inhibitors overnight and then looked at the localization of the Vgl message by in situ hybridization. Cytochalasin B treatment, which depolymerizes microfilaments, had a dramatic effect on the distribution of Vgl mRNA, releasing it from its tight shell and allowing it to diffuse (fig. 4); this distribution is very similar to that observed in unfertilized eggs (Weeks and Melton, 1987). Nocodazole and colchicine, drugs that bind tubulin and cause depolymerization of microtubules, have no effect on the localization of Vgl message (fig. 4 and data not shown). Interestingly, analysis of the partitioning of RNA in detergent extracts of the treated oocytes only partially reflects the effects of the drugs. As expected from the in situ results, nocodazole treatment does not affect the association of Vgl mRNA with the cytoskeletal pellet (fig. 4). Cytochalasin B treatment, however, despite releasing Vgl mRNA from its cortical localization, has only a slight effect on Vgl mRNA distribution in extracts. Presumably, although the depolymerization of microfilaments in late-stage oocytes allows Vgl mRNA to diffuse away from the cortex, the mRNA remains bound with other factors that may cause it to be pelleted during the detergent extraction.

The experiments with late-stage oocytes address the question of how Vgl mRNA is tethered at the vegetal cortex. To look at how the message is translocated to this site, we studied the effects of the cytoskeletal inhibitors on middle-stage oocytes undergoing localization. Cytochalasin B had little if any effect on the translocation of Vgl mRNA, although the message seemed to accumulate further from the cortex than in untreated oocytes (fig. 5). Nocodazole, however, completely inhibits localization (fig. 5), and colchicine has an identical effect (data not shown). None of the drug treatments had any effect on the stability of Vgl mRNA or on the synthesis of protein over the five day culture period (data now shown). All of the drugs severely inhibited growth of the oocytes in culture, perhaps by inhibiting the uptake of vitellogenin from the medium. Nonetheless, Vgl mRNA was translocated in cytochalasin B-treated oocytes, indicating that growth and translocation are independent and separable phenomena, and that the inhibition of translocation by microtubule inhibitors is not a result of the lack of growth of the oocyte.

The data presented here demonstrate that microtubule and microfilament inhibitors have different and stagespecific effects on the Vgl localization machinery. The simplest model to explain these results is the two-step process diagrammed in figure 6, with microtubules involved in the translocation process and microfilaments involved in anchoring the message at the cortex. Although the data described above appear to argue against a specific degradation or instability of nonlocalized Vgl mRNA. it is hard to distinguish between active, as opposed to passive, movement of Vgl mRNA. The graded disappearance of Vgl message during its localization both in vivo and in vitro in the

animal to vegetal direction may be the result of an active localization of the message. Alternatively, reversible binding of Vgl mRNA to a membrane-bound receptor would result in the same ‘window shade’ effect, even if movement of the mRNA were passive. The apparent involvement of microtubules in the translocation of the message may indicate an active movement of the message. A number of different types of cell make use of microtubule-mediated motors to actively direct the transport of macromolecules and organelles (see Vale, 1987). Indeed, calculated rates for slow axonal transport, which is mediated by microtubules, are similar to the estimated rate of localization of the Vgl mRNA in vitro (OTmm/day; Lasek, 1982). Alternatively, however, microtubules may act simply as ‘highways’ or tracks, expediting the otherwise random movement of Vgl mRNA without actually propelling it.

The model in figure 6 implies that cytoplasmic microtubules are disrupted in middle-stage oocytes by nocodazole or colchicine, and that cortical microfilaments are depolymerized by cytochalasin B in late-stage oocytes. Microtubule arrays begin to form early in oogenesis; tubulin staining appears first perinuclearly and then in radial arrays extending completely around the oocyte from the germinal vesicle (gv) to the cortex (Palecek et al. 1985; Dent and Klymkowsky, 1988; manuscript in preparation). As oogenesis proceeds and the gv migrates to the animal hemisphere, the microtubule arrays become less easily detectable in the vegetal hemisphere but remain radially aligned, emanating from the gv to the animal cortex. In both middle- and late-stage oocytes, these arrays are depolymerized by nocodazole and colchicine. Microfilaments, by comparison, appear mainly along the cortex early in oogenesis and remain there until maturation, when there is a reorganization of cytoskeletal elements throughout the oocyte (Franke et al. 1976; Dent and Klymkowski, 1988). Thus, the temporal and spatial organization of microtubules and microfilaments is precisely what would be expected from the two-step localization scheme outlined in figure 6.

The association of Vgl mRNA with the detergentinsoluble fraction of extracts from late-stage oocytes is consistent with the release of Vgl mRNA after cytochalasin B treatment. A similar detergent extract of oocytes has been shown to contain significant amounts of several intermediate filament proteins (Pondel and King, 1988), although there is no evidence at present for any Vgl mRNA association with these proteins. The data presented here clearly indicate that Vgl mRNA is somehow associated with microfilaments in the cortex but do not rule out interactions with other proteins as well. The association of Vgl mRNA with the cytoskeletal pellet even in oocytes where the message has been released from its tight cortical shell by cytochalasin B may suggest a continued association of other detergentinsoluble proteins with the message. Presumably, the specificity for the localization process is a combination of the presence of cw-acting signals in the RNA itself, specific factors that recognize these sequences, and the cytoskeletal framework within which the RNA moves and is anchored properly. By more precisely defining the RNA signals and identifying the cytoskeletal elements involved in the localization, it should be possible more fully to understand how oocytes generate and interpret intracellular polarity.