While Bucky ball is needed to make the Balbiani body in this species, its role in the formation of the homologous structure found in man is enigmatic, since the human gene homologous to bucky ball/Xvelo1 is interrupted by stop codons. In this case it is possible that other redundant components may be able to substitute for it in forming a Balbiani body on an evolutionary timescale. Of course humans do not form germ plasm and it is not at all clear what the function of mammalian Balbiani bodies might be. However, the loss of Bucky ball protein coding ability in humans fits with a role in germ plasm RNP particle formation, since there is no evidence for RNA localisation in mammalian eggs. In zebrafish bucky ball RNA is present throughout oogenesis, but the expression characteristics of its protein are uncertain. Although, a Buc-GFP fusion enters the Balbiani body of early oocytes and the germ plasm of embryos, the endogenous protein has not been studied. However, the fact that the bucky ball mutation was only rescued by a translatable message suggests that Bucky ball protein is an essential player in germ plasm formation. Xenopus Xvelo1 exists as two splice variants, which we show interact with Hermes through their common N-terminal region. Both GFP-Xvelo fusions enter germ plasm in large oocytes and both enter the Balbiani bodies of previtellogenic oocytes. Staining of oocytes with Xvelo isoform-specific antisera demonstrates that both proteins are present in the RNP particles of germ plasm in large oocytes and fertilized eggs, where they co-localise with Benzethonium Chloride Cy5-nanos1 RNA in a distribution identical to their respective GFP fusions. However, only the larger variant, XveloFL, appeared to be naturally present in the earlier mitochondrial cloud. BiFC experiments supported the conclusion that Hermes interacts with Yunaconitine XveloFL and SV in germ plasm RNPs and that this is not simply because they are packed into these structures. Positive BiFC interactions between two candidate proteins may occur when they are less close than is required for FRET, i.e. up to 10 nm, so one might wonder if simply being packed into the same RNP particle is sufficient to give a positive signal. The negative results obtained using BiFC constructs for XveloSV and XveloFL reveal that this is not the case. Although, GFP fusion proteins for XveloFL and SV must localise into the same RNPs, combinations of their respective BiFC fusions gave no significant signal above background. The same argument applies to VC-XveloSV plus VN-XveloSV, which obviously must be in the same particles. As a consequence of their large size, these RNP particles must accommodate many XveloSV molecules; this must be so in order to yield the strong fluorescence seen when GFP constructs are expressed. Similarly, there was no signal when self-interaction of another particle protein, Poc1B was tested. We conclude that the positive signals detailed in Figure 3 and Table 2 identified genuine inter-molecular interactions within germ plasm RNPs. Examination of the protein structure of Xvelo1 gives little indication of its biological function. Both Xvelo isoforms contain a putative dynein light chain binding site, like the potential germ plasm protein 
Germes, which has been shown to interact with DLC-8, but we find that this motif is not essential for the localisation of GFP-tagged XveloFL to germ plasm islands. However, in these experiments endogenously expressed Xvelo proteins are also present, so we cannot exclude the possibility that these interact with dynein and enable exogenous mutant Xvelo to localise via direct or indirect protein-protein interactions.