Potential structural role of non-coding and coding RNAs in the organization of the cytoskeleton at the vegetal cortex of Xenopusoocytes

RNAs function in a wide variety of processes in the nucleus and cytoplasm,including regulation of translation, RNA editing and gene regulation. Additionally, a class of localized RNAs found in specific regions of cells or embryos perform many different functions(Bashirullah et al., 1998; Jansen, 2001; Kloc et al., 2002a; Palacios and St. Johnston,2001).

In Xenopus laevis, Vg1 was the first mRNA identified that was localized at the oocyte vegetal cortex(Rebagliati et al., 1985). Now, many different RNAs are known to be localized at the vegetal cortex(Bashirullah et al., 1998; King et al., 1999; Kloc et al., 2002b; Palacios and St. Johnston,2001). Localized RNAs such as Vg1(Joseph and Melton, 1998; Kessler and Melton, 1995); the Xenopus homolog of Bicaudal-C (Wessely and de Robertis, 2000); and the T-box family member VegT (Xanthos et al., 2001; Zhang and King,1996) are involved in mesoderm and/or endoderm formation. Recently, Hermes was shown to regulate the cleavages of the vegetal blastomeres (Zearfoss et al.,2004). Xcat2 and Xpat mRNAs are thought to have functions specific to germ cells because of their association with germ plasm, which contains the germ cell determinants within structures called germinal granules(Yisraeli et al., 1990; Kloc and Etkin, 1995; Forristall et al., 1995; Kloc et al., 1998; Kloc et al., 2002a).

Localized RNAs co-fractionate and co-immunoprecipitate with cytoskeletal elements that serve in both RNA transport and anchoring(Alarcon and Elinson, 2001; Boccaccio et al., 1999; Elinson et al., 1993; Forristall et al., 1995; Jankovics et al., 2002; Klymkowsky et al., 1991; Marikawa et al., 1997; Pondel and King, 1988; Shan et al., 2003; Stebbings, 2001; Yisraeli et al., 1990; Zhao et al., 1991). Interestingly, a noncoding RNA, Xlsirts(Kloc and Etkin, 1994) and VegT mRNA (Heasman et al.,2001; Zhang and King,1996), when destroyed with antisense oligonucleotides (ODNs),caused release of specific RNAs from the vegetal cortex. The present study demonstrates that VegT and Xlsirts RNAs are integrated within the cytokeratin cytoskeleton and that destruction of these RNAs disrupts the cytoskeleton in specific ways resulting in the release of several different RNAs. This suggests the intriguing possibility that both coding and noncoding RNAs may have several roles, including structural organization of the vegetal cortex in Xenopus oocytes.

Xenopus laevis wild-type females were anesthetized and pieces of ovary were surgically removed and placed in oocyte medium (OCM)(Heasman et al., 1991). Stage IV and VI oocytes were defolliculated manually, and injected and incubated in the same medium.

Total RNA was prepared from control and antisense Xpat ODN-injected stage VI oocytes (100 oocytes in each group) using Trizol (Invitrogen) according to manufacturer protocol. Poly(A) RNA was isolated from total RNA using Poly(A)purist kit from Ambion. Northern blot was prepared according to standard methods and hybridized to digoxigenin-labeled antisense Xpat RNA probe synthesized from Xpat 3′UTR (Xpat3′UTR in Bluescript was a gift from Hugh Woodland, University of Warwick, UK), and antisense Xcat2 probe synthesized from Xcat2-Sport (gift from M. L. King, University of Miami, FL)in ULTRAhyb buffer (Ambion) according to manufacturer protocol.

We injected 10 ng per oocyte of phosphorothioated oligonucleotides(synthesized by Sigma Genosys, Woodlands, TX; modified bases marked by the *)of the following sequences: antisense VegT, 5′C*A*G*C*AGCATGTACT*T*G*G*C 3′ (Zhang et al., 1998; Heasman et al., 2001);antisense Xlsirts 1, 5′ C*A*G*GTATAGTAGGG*A*G*A 3′(Kloc and Etkin, 1994);antisense Xlsirts 2, 5′ T*C*T*CTGGGAAGGGA*G*T*G 3′(Kloc and Etkin, 1994);antisense XlCaax, 5′ C*T*G*CGCTTAGAGAA*C*C*C 3′(Kloc and Etkin, 1994);antisense Xpat, 5′ T*T*C*T*GCCTTCAAAGCCAT*A*G*A 3′; sense VegT,GC*C*A*A*GTACATGCT*G*C*T*G; sense XlCaax, CG*G*T*TCTCTAAGCG*C*A*G; sense Xlsirts 1:TC*T*C*CCTACTATA*C*C*TG; sense Xlsirts 2, CA*C*T*CCCTTCCCAG*A*G*A(Kloc and Etkin, 1994); VegT morpholino, 5′CCCGACAGCAGTTTCTCATTCCAGC 3′ injected at 30 ng (in 10 nl) per oocyte (see Heasman et al.,2001).

Molecular beacons, with 2'-O-methylribonucleotide backbone and Texas Red at the 5′ end and BHQ2 (Black hole quencher 2) at the 3′ end were synthesized as in Bratu et al. (Bratu et al., 2003). The molecular beacon oligonucleotides of the following sequences were injected into the oocytes (10 ng in 10 nl): VegT, 5′ cacacCAGCAGCATGTACTTGGCgtgtg 3′; a 1:1 mixture of Xlsirts number 1 (5′ cacactCAGGTATAGTAGGGAGAgtgtg 3′) and Xlsirts number 2 (5′ caagtTCTCTGGGAAGGGAGTGacttg3′); and Xpat, 5′ caagt TTCTGCCTTCAAAGCCATAGAacttg3′ (the bold nucleotides are the arm sequences). Injected oocytes were incubated in OCM for 4-6 hours at 18°C. Then, 1 nl of an anti-pancytokeratin-FITC antibody (C11 antibody, Sigma) was injected into oocytes, and the vegetal tip of the oocyte was immediately cut off, mounted in PBS on a microscope slide under a coverslip and observed under a Nikon fluorescence microscope within 15-30 minutes of C11-FITC injection.

Host transfer and rescue was carried out according to Heasman et al.(Heaseman et al., 1991) and Zearfoss et al.(Zearfoss et al., 2004).

Actin was stained using rhodamine-phalloidin (Molecular Probes)(Kloc et al., 2004). Cytokeratin immunostaining was carried out according to the method of Alarcon and Elinson (Alarcon and Elinson,2001).

Light and electron microscopy of whole-mount in situ hybridization of oocytes and embryos were performed as described by Kloc and Etkin(Kloc and Etkin, 1994) and Kloc et al. (Kloc et al.,2002a).

To determine how Xlsirts and VegT may function in anchoring other RNAs, we depleted Xlsirts RNA and VegT mRNA using antisense oligodeoxynucleotides(ODNs). Destruction of either VegT or Xlsirts resulted in release of Vg1 mRNA from the vegetal cortex of stage VI oocytes(Fig. 1A,B) consistent with previous studies showing Vg1 mRNA being released and not degraded after destruction of specific RNAs (Kloc et al., 1994; Heasman et al., 2001). Injection of antisense ODNs against another localized mRNA, Xpat did not cause the release of Vg1 RNA (Fig. 1A). Fig. 1C shows that antisense Xpat ODN destroys Xpat mRNA.

We hypothesized that the release of Vg1 RNA from the vegetal cortex in VegT or Xlsirts depleted oocytes was caused by disruption of the oocyte cortical cytoskeleton. Therefore, we investigated the effect of destruction of Xlsirts and VegT RNAs on the organization of the cytokeratin and the actin cytoskeleton within the vegetal cortex of full-grown stage VI oocytes, oocytes that were matured in vitro by addition of progesterone, and mature oocytes that were activated by pricking with a glass needle.

In control stage VI oocytes the cytokeratin cytoskeleton is visible as an intricate network within the vegetal cortex(Fig. 2A, part 1). During maturation of control stage VI oocytes this network is disrupted, resulting in a punctate appearance consisting of foci of cytokeratin(Fig. 2A, part 2). Following activation by pricking with a glass needle the cytokeratin reforms into a more complex network-like arrangement (Fig. 2A,part 3) (Clarke and Allan,2003; Klymkowsky et al.,1991; Klymkowsky and Maynell,1989).

The depletion of Xlsirts or VegT RNA with specific ODNs in stage VI oocytes caused the disruption of the cytokeratin network at the oocyte vegetal cortex(Fig. 2A). Interestingly, the effects of VegT depletion and Xlsirts depletion were different. Destruction of Xlsirts RNA caused a very pronounced `thinning' or `flattening' of the cytokeratin network and its collapse inward towards the yolk mass(Fig. 2A, part 4). The cytokeratin network looked well organized but much thinner and less `three dimensional' than the network in control oocytes, and, in contrast to the control oocytes, the yolk platelets were clearly visible below the cytokeratin network (Fig. 2A, part 4). Destruction of VegT mRNA resulted in fragmentation and disruption of the cytokeratin network (Fig. 2A, part 7).

To analyze the effect of destruction of Xlsirts and VegT RNAs during the processes of maturation and activation, the antisense ODNs were injected into stage VI oocytes, and injected oocytes were incubated for 24 hours, matured by addition of progesterone and activated by pricking. After maturation,cytokeratin foci were sporadically visible in antisense Xlsirts eggs and very rarely visible or not visible at all in antisense VegT egg(Fig. 2A, parts 5 and 8). Upon activation, the cytokeratin network was partially reconstituted in antisense Xlsirts-injected eggs but was less regular than the network in control eggs(Fig. 2A, part 6). However, in antisense VegT activated eggs the cytokeratin network was not reconstituted,and only a few short cytokeratin filaments were visible(Fig. 2A, part 9).

Destruction of vegetally localized Xpat mRNA, a non-localized RNA XlCaax,and sense oligonucleotides against all of the RNAs had no effect on the appearance of the cytokeratin network (Fig. 2B, parts 1 and 2). Injection of VegT morpholino that inhibits translation of VegT protein (Heasman et al., 2001) had no effect on the cytokeratin network in stage VI oocytes (Fig. 2B3) or during maturation and activation (data not shown). These findings support the conclusion that disruption of the cytokeratin network in oocytes injected with antisense VegT oligonucleotides is due to elimination of RNA and not protein. In addition, no noticeable effect on the cortical actin microfilament cytoskeleton was observed in oocytes, maturing oocytes or activated eggs when XlCaax, Xlsirts or VegT was depleted (data not shown).

We also compared the effects of injection of the anti-pancytokeratin C11 antibody that blocks cytokeratin function(Alarcon and Elinson, 2001) on the cytokeratin network in stage VI oocytes, with those observed after the injection of antisense Xlsirts and antisense VegT ODNs. Thirty minutes after the injection of C11-FITC antibody, the cytokeratin network remained comparable with the control (Fig. 3A). However, ∼40 minutes. after the injection of antibody,the network started to disintegrate and, after 3 hours, its remnants were visible as the brightly fluorescing foci(Fig. 3B). When antibody-injected oocytes were matured overnight, the cytokeratin foci,similar to the punctate foci observed in control matured oocytes were visible(Fig. 3C). Following activation, the cytokeratin partially reformed into short filaments, but in contrast to control activated eggs, it did not reform the network-like arrangement (Fig. 3D). These results showed that the disruption of cytokeratin network by anti-pancytokeratin antibody, had similar, although not identical, effects to the disruption of cytokeratin caused by the injection of antisense VegT.

We also determined the relationship between Xlsirts RNA, VegT RNA and the cytokeratin network. Upon injection of Texas Red-labeled RNAs into vegetal region of stage VI oocytes, the Xlsirts, VegT and Xpat RNAs formed particles measuring 0.5-1.0 μm; however, only Xlsirts and VegT particles had an affinity to the cytokeratin and aggregated on its filaments(Fig. 4A). The difference between the number of Xlsirt, VegT and control Xpat particles aggregated on cytokeratin filaments was statistically significant(Fig. 4B).

Although these results were informative, they did not provide any indication of the distribution of endogenous RNAs in relation to cytokeratin filaments. To investigate the spatial distribution of endogenous Xlsirts and VegT RNAs on the cytokeratin network, we used the technique of in vivo in situ hybridization with molecular beacons (Fig. 5). Molecular beacons are short synthetic hairpin 2′-O-methylribonucleotides complementary to their cellular RNA targets that possess a fluorophore on one end of the molecule and a quencher on the other end. The hairpin configuration holds the fluorophore and quencher in close proximity, thus blocking the fluorescence signal. Upon injection into an oocyte and binding to the cognate RNA, the hairpin configuration becomes disrupted such that the fluorophore is removed from the vicinity of the quencher and the fluorescence of the probe is restored(Bratu et al., 2003; Marras et al., 2004; Tyagi and Kramer, 1996).

We injected Texas Red-conjugated molecular beacons against Xlsirts, VegT and Xpat RNAs into stage VI oocytes. After a 6-hour incubation we injected an anti-pancytokeratin C11-FITC antibody, and within 30 minutes, we examined the distribution of molecular beacons bound to their target RNAs in relation to the cytokeratin network. We found that the pattern of distribution differed for each of the three RNAs. Xlsirts RNA formed two different types of aggregates: smaller ones intimately associated with cytokeratin(Fig. 5A, parts 1 and 2) and much larger ones that we believe represent the germ plasm islands (data not shown). VegT mRNA formed an intricate network of bright foci or aggregates associated with cytokeratin filaments (Fig. 5A, parts 3 and 4). Very often the regular spacing of the VegT RNA on cytokeratin filaments was clearly visible(Fig. 5A, part 4). By contrast,Xpat mRNA, was absent from the cytokeratin network and was present only in large aggregates that probably represented the germ plasm islands(Fig. 5A, part 5; not shown). Statistical analysis of frequency of association of RNA particles with cytokeratin filaments confirms these results(Fig. 5B). In addition, VegT and Xlsirts RNAs were detected in particles that measured ∼0.5-1.0 μm in diameter. This is very similar to the size of the particles formed from injected Xlsirts and Vg1 RNAs while they were translocating to the vegetal cortex (Kloc et al.,1996).

The results of our analysis indicate that endogenous VegT mRNA and Xlsirts RNA are components of the cytokeratin network in the vegetal cortex of stage VI Xenopus oocytes, and that the two RNAs have unique patterns of association with cytokeratin filaments. Although our results do not demonstrate if the interaction with the cytoskeleton is direct or indirect,they do demonstrate the presence of these RNAs within the cytokeratin network.

We next wanted to see if the cytokeratin network disintegrated by the depletion of VegT mRNA could be reconstituted and rescued upon the injection of VegT RNA. Stage IV and VI oocytes were injected with antisense VegT ODN,followed by injection with VegT mRNA known to rescue antisense VegT depletion(Zhang and King, 1996; Heasman et al., 2001; Horb and Thomsen, 1997). The rationale behind injection into stage IV oocytes was that stage IV oocytes (in contrast to stage VI oocytes) are able to localize injected RNA to the vegetal cortex (Yisraeli and Melton,1988). The cytokeratin network in vegetal cortex of control stage IV oocytes differs significantly in appearance from the network of stage VI oocytes (Clarke and Allan,2003); it is irregular and much less intricate(Fig. 6A,B), thus the disruption of the network upon antisense VegT ODN injection into stage IV oocytes causes seemingly much less dramatic effect than in stage VI oocytes(Fig. 6C,D).

The exogenous VegT mRNA, but not Xpat or Vg1 mRNA, was able to reconstitute and rescue the cytokeratin network in stage IV and stage VI oocytes(Fig. 6E,F; data not shown). Interestingly, the reconstituted network in stage IV oocytes was strikingly more complex and multidimensional than in controls, and similar to the network in stage VI oocytes (Fig. 6E,F). The rescuing effect of VegT mRNA occurred not only in stage IV but also in stage VI oocytes, when injected RNA is unable to localize. Thus, the ability of RNA to localize was irrelevant for rescue. We also found that the exogenous VegT RNA was able to reconstitute, to considerable extent,the cytokeratin network in the oocytes injected with anti-pancytokeratin C11 antibody (see Fig. 3E). In addition, we found that the injection of VegT RNA did not change the cytokeratin network appearance when injected to control oocytes(Fig. 6G,H). This result, and the fact that VegT RNA also reconstitutes the network in antibody-injected oocytes, suggests that the rescuing effect of VegT RNA operates at the level of cytokeratin network-subunit assembly.

We next investigated the biological consequences of disruption of the cytokeratin cytoskeleton. Residing at the vegetal cortex are the germinal granules, which are components of the germ plasm that specifies the germ cell lineage. Germinal granules are electron-dense structures that contain several different RNAs, including Xcat2 and Xpat. The Xlsirts RNA is located within the germ plasm in the germ plasm matrix between the germinal granules(Hudson and Woodland, 1998; Kloc et al., 2001; Kloc et al., 2002b).

As the depletion of Xlsirts and VegT RNAs affects the organization of the germ plasm-containing region of the vegetal cortex, we analyzed the effect of depletion of these RNAs on the morphogenesis of the germinal granules in matured oocytes, activated eggs and embryos of different stages acquired by host transfer (Heasman et al.,1991). Stage VI oocytes injected with either Xlsirts, VegT, XlCaax antisense ODNs, VegT morpholino or anti-pancytokeratin C11 antibody, were incubated for 24 hours, matured and placed within the host female. In matured and activated eggs originating from control, antisense XlCaax, antisense Xlsirts and VegT morpholino-injected oocytes, the germ plasm and germinal granules appeared normal at the electron microscopic level(Fig. 7A-D). However, in antisense VegT injected oocytes, the germinal granules were strikingly abnormal, forming long intricate chains of electron dense material(Fig. 7E) that looked similar to the germinal granule aggregates normally present much later in cleaving embryos (Kloc et al., 2002b). This finding suggested that the depletion of VegT mRNA (but not VegT protein)from the oocyte cortex causes untimely and premature aggregation of the germinal granules.

Next, we analyzed the effect of Xlsirts and VegT depletion, and the effect of anti-cytokeratin antibody on germinal granule structure and on the association of Xcat2 mRNA with the granules in early embryos. Although the appearance of the germ plasm in whole mounts was similar in control and antisense ODN-injected embryos, the intensity of Xcat2 labeling was clearly lower in antisense Xlsirts embryos (Fig. 8A, parts 1 and 2). Higher magnification revealed that in antisense Xlsirts-injected embryos, the Xcat2 mRNA was present in the germ plasm region but was located outside of the germinal granules, which were mostly unlabeled thus barely visible. Labeled granules were observed only sporadically, and the labeling intensity was lower than in control embryos(Fig. 8B, parts 1 and 2; see Fig. 10A, parts 3 and 3′; Fig. 10B). By contrast, the germ plasm of antisense VegT or anti-pancytokeratin antibody injected embryos, instead of containing a high number of small aggregates, contained a lower number of large stringy aggregates of labeled germinal granule-like structures(Fig. 8A, parts 3 and 4; Fig. 8B, parts 3 and 4). Interestingly, the cytokeratin network was present not only in control embryos but also in the embryos originating from antisense VegT, antisense Xlsirts and anti-pancytokeratin antibody injected eggs(Fig. 9). This indicates that disruption of the cytokeratin network in oocytes and eggs does not prevent the reconstitution of a seemingly normal network during cleavage, and shows that the untimely aggregation of germinal granules in antisense VegT-injected embryos is a consequence of the disruption of the cytokeratin network in the oocytes and not in the embryos.

The ultrastructure of germinal granules in antisense and anti-cytokeratin antibody injected embryos was analyzed at the electron microscopic level(Fig. 10A). In embryos injected with antisense XlCaax or antisense Xpat ODNs, the Xcat2 mRNA was located throughout the germinal granule and no abnormalities were detected(Fig. 10A, parts 1, 2, 1′,2′; not shown). In antisense Xlsirts embryos, the germinal granules were largely devoid of Xcat2 mRNA(Fig. 10A, parts 3 and 3′; Fig. 10B). However, in antisense VegT ODN-injected embryos, Xcat2 mRNA remained partially on the granules, which now resembled abnormally large aggregates (Fig. 10A, part 4). These were probably formed by the further aggregation of untimely coalesced stringy aggregates present in VegT-depleted oocytes (see Fig. 7E). The germinal granules in anti-pancytokeratin antibody-injected embryos formed large aggregates that were similar, but not identical, to the germinal granules aggregates in antisense VegT embryos (Fig. 10A, part 5). However, in contrast to antisense VegT embryos, the amount of Xcat2 mRNA in anti-pancytokeratin antibody embryos was similar to control embryos (Fig. 10B). Statistical analysis of Xcat2 mRNA distribution within the germinal granules of control and experimental embryos showed that the number of Xcat2 mRNA grains (silver grains) per 0.25 μm² area of germinal granule was similar in control, anti-cytokeratin C11 antibody and antisense Xpat ODN-injected embryos, and significantly lower in antisense VegT ODN- and antisense Xlsirts ODNs-injected embryos(Fig. 10B). Thus, disruption of the cytokeratin network within the vegetal cortex of stage VI oocytes by either depletion of Xlsirts and VegT RNAs or by the injection of anti-pancytokeratin antibody has a major effect on the formation, structure and molecular composition of the germinal granules. This indicates an important link between the structural integrity of the vegetal cortex and subsequent events that occur during development, supporting our conclusion that Xlsirts and VegT RNAs do function in the maintenance of the cytokeratin network in the oocyte cortex.

To further assess developmental defects, we analyzed primordial germ cells(PGCs) in embryos produced by the host transfer method following depletion of Xlsirts or VegT RNA in stage VI oocytes. Xpat mRNA was used as a marker to identify PGCs (Hudson and Woodland,1998). In agreement with Heasman et al.(Heasman et al., 2001) and Zhang and King (Zhang and King,1996), antisense VegT injected embryos are unable to undergo gastrulation or neurulation properly; therefore, we analyzed PGCs in blastula embryos. This, at least partially, eliminated the possibility that changes in PGCs were a consequence of an abnormal development. It is known that in mid to late blastula embryos (stage 7-9), the dividing PGCs occupy the vegetal tip of blastula. At this stage the germ plasm is visible as small aggregates, on the vegetal tip of blastula (Fig. 11A) (Hudson and Woodland,1998). In VegT-depleted embryos, the germ plasm was visible as a single large aggregate (reminiscent of large aggregates present in early embryos) and as smaller dispersed aggregates(Fig. 11B,C). In Xlsirts-depleted embryos, the germ plasm was either barely visible or visible as tiny germ plasm islands scattered throughout the vegetal blastomeres instead of being found in compact group(Fig. 11D,E). Therefore,disruption of the cytokeratin network within stage VI oocytes not only affected the formation of the germinal granules but also had a direct consequence on the appearance and behavior of the germ plasm and possibly on the formation of the primordial germ cells in later-stage embryos, as judged by the distribution of Xpat mRNA within the germ plasm of blastula PGCs.

Our data support a model in which the non-coding Xlsirts RNA and the coding VegT mRNA play a structural role in the organization of the cytokeratin network but not the actin cytoskeleton within the vegetal cortex of Xenopus oocytes. We cannot exclude possibility that actin cytoskeleton is also affected but in more subtle way that was undetectable by our methods. We also found that destruction of Xlsirts or VegT RNA affects the cytokeratin network in different ways. Loss of Xlsirts RNA does not completely destroy the cytokeratin network in stage VI oocytes; rather, it causes a thinning of cytokeratin network and its collapse towards the yolk mass, with partial reformation of the network in activated eggs. By contrast, loss of VegT mRNA disrupts the network in stage VI oocytes, after which the network never reforms in activated eggs, but, interestingly, does reform in cleaving embryos, which suggests that, at least partially different mechanism(s) govern the cytokeratin network assembly in oocytes and embryos.

An important issue is whether Xlsirts and VegT RNAs interact directly with cytokeratin fibers. Our data show an intimate association of the localized Xlsirts and VegT RNA complexes with the cytokeratin network within the vegetal cortex. Our findings agree with previous findings by others that several different localized RNAs, including VegT, co-immunoprecipitate with the cytokeratin network (Alarcon and Elinson,2001). Furthermore, we found that depletion of the Xlsirts RNA and especially VegT mRNA resulted in loss of integrity of the cytokeration network, and that exogenous VegT RNA was able to reconstitute cytokeratin network. We interpret this finding as an indication that these RNAs are integrated within the network holding the filaments together and/or are needed for the assembly of the network subunits(Fig. 12). Further analyses will be required to determine if the RNAs interact directly with the cytokeratin or if other proteins within the RNA complexes are required. The large size (0.5-1 μm) of these complexes suggests that they probably contain a multitude of RNA-associated proteins.

We propose that an intact cytokeratin network in stage VI oocytes and/or abundant cytokeratin foci in matured eggs prevent the untimely aggregation of germinal granules that normally takes place in embryos during early cleavage. This is supported by our data showing that the appearance of germinal granules in antisense VegT embryos is very similar to that in anti-cytokeratin antibody injected embryos. At present, there is no information on how cytokeratin filaments interact with the germ plasm or germinal granules. One possibility is that cytokeratin anchors the germinal granules, either directly or indirectly via other proteins. Thus, the severance of the cytokeratin network in oocytes injected with antisense VegT oligonucleotides or anti-cytokeratin antibody causes the release and untimely aggregation of germinal granules(Fig. 12). Another possibility is that cytokeratin itself does not have any structural or functional connection to the germinal granules but, as a structural component of the vegetal cortex, is necessary to preserve its integrity. If the latter possibility is true, then the untimely and premature aggregation of germinal granules is the result of intrinsic changes in the organization of the oocyte cortex in oocytes injected with antisense VegT oligonucleotides. Further studies are needed to explain how and why the depletion of Xlsirts RNA in stage VI oocytes caused the release of Xcat2 mRNA from the germinal granules in cleaving embryos and caused the changes in the morphology and the distribution of germ plasm islands in PGCs in blastula-stage embryos.

The importance of an intact cytokeratin network within the cortex has now been demonstrated in two studies. The first was a previous study showing that disruption of the cytokeratin in the oocyte cortex using anti-cytokeratin antibodies results in the release of several different RNAs(Alarcon and Elinson, 2001). The second is our current study, which demonstrated that the injection of anti-cytokeratin antibody and the depletion of Xlsirts and VegT RNAs (which specifically disrupted the cytokeratin cytoskeleton) had a dramatic effect on the morphogenesis and molecular composition of the germinal granules. This resulted in a phenotype affecting PGC formation in blastula embryos. The evidence that the VegT mRNA and not protein is serving a structural role is that antisense morpholino against VegT that inhibit translation of VegT protein (Heasman et al., 2001)did not produce this effect. Xlsirts does not encode protein, thus it must be functioning as an RNA.

Although the findings of this study apply to only a very specialized situation such as the anchoring of localized RNAs within the vegetal cortex of oocytes, it is quite likely that other RNAs play similar roles in both germ cells and somatic cells of other organisms. In the case of Xenopusoocytes, it is clear that integration of the RNAs into the cytoskeleton is of extreme importance for the proper formation and migration of the germ cells. In other systems, the role of the RNA-cytoskeletal network may serve a different function; however, our results open up the exciting possibility of a new role for RNAs in maintaining the structural integrity of the cellular cytoskeleton.

This work was supported by grants from NSF (to L.D.E.), the Polish State Committee for Scientific Research-KBN grant BW/4/IZ/2002 (to K.W.), and Core Grant CA16672 and grant GM070357-01 to Sanjay Tyagi. We also thank Dr Sanjay Tyagi and Dr Salvatore Marras from the Department of Molecular Genetics(Public Health Research Institute, Newark, NJ) for their help in designing the molecular beacons and their helpful discussion of methods. We also thank Mr Kenneth Dunner, Jr, for his superb electron microscopy work.