Sperm granules mediate epigenetic inheritance

Abstract

Epigenetic inheritance is the transfer of non-DNA information across generations. A new study identifies sperm-specific PEI granules as essential for paternal epigenetic inheritance. PEI granule partitioning to sperm requires palmitoylation and myosin VI activity, suggesting lipidation-dependent granule transport on vesicles.

Gametes — oocytes and sperm — transmit DNA and other molecules, including proteins, RNA and metabolites, to offspring. The transmission of non-DNA information between generations is termed epigenetic inheritance and is important for adaptation of progeny to environmental conditions experienced by parents¹. Epigenetic inheritance often involves the transmission of small RNAs (sRNAs) that act as guides for Argonaute proteins to facilitate both transcriptional and post-transcriptional gene regulation. sRNAs loaded in oocytes and sperm target complementary transcripts in embryos. In the oocytes of Drosophila, Caenorhabditis elegans and zebrafish, Argonaute proteins localize to discrete cytoplasmic condensates (termed germ granules) that are thought to function as vehicles for the transmission of sRNA from parent to progeny². Similar condensates have been observed during spermatogenesis³, but not in mature sperm, leaving open the question of how sperm transmit sRNAs. In this issue of Nature Cell Biology, Schreier et al.⁴ provide an answer: they identify a previously undescribed class of condensate in sperm, PEI granules, required for paternal epigenetic inheritance in C. elegans.

During the final steps of sperm development, sperm cells discard most of their cytoplasm into a structure called the residual body. Only nuclei, mitochondria and specialized organelles required for sperm motility and fertilization are maintained in mature sperm. In C. elegans, the so-called fibrous body-membranous organelles (FB-MOs) are actively transported into spermatids away from the residual body⁵. Notably, Scherier et al.⁴ find that PEI granules hitchhike on FB-MOs to gain access to spermatids and avoid being left behind in the residual body. PEI-related proteins function during spermatogenesis in mammals⁶, raising the possibility that a similar mechanism may facilitate paternal epigenetic inheritance in other organisms. This study also adds to a growing body of evidence that implicates membrane association as a key regulator of condensate assembly and dynamics⁷.

With 27 Argonaute-encoding genes and thousands of sRNAs, C. elegans has emerged as an outstanding model system to study transgenerational inheritance⁸. Both hermaphrodites and males transmit epigenetic information, and both oocytes and sperm contain sRNAs that are carried over in fertilized eggs⁹. Until now, however, Argonaute-containing condensates (P granules) had been observed only in oocytes. P granules disassemble during spermatogenesis and their components are discarded in the residual body¹⁰.

In the course of studying the Argonaute WAGO-3, Schreier et al.⁴ noticed that, unlike other Argonautes, WAGO-3 localizes to puncta in spermatids. Proteomic analyses on WAGO-3 immunoprecipitates identified a previously undescribed protein PEI-1 (paternal epigenetic inheritance defective-1). PEI-1 is expressed exclusively during spermatogenesis, localizes to WAGO-3 puncta, and is required for the partitioning of WAGO-3 to spermatids. Importantly, wago-3 and pei-1 mutant males are defective in transmitting epigenetic information to progeny. These findings led Schreier et al.⁴ to propose that PEI-1 and WAGO-3 define a previously unappreciated condensate, the PEI granule, that transmits epigenetic information via sperm (Fig. 1a).

Fig. 1 |. Membrane-tethered Pei granules partition to spermatids to mediate paternal epigenetic inheritance.

a, PEI-1 recruits the Argonaute WAGO-3 to sperm-specific PEI granules via a putative interaction with the PEI-1 intrinsically disordered region (IDR). b, PEI granules partition to budding spermatids by association with fibrous body-membranous organelles (Fb-MOs). Partitioning is dependent on the myosin VI motor SPE-15. c, PEI-2 tethers PEI granules to membranes via predicted palmitoylation by the palmitoyltransferase SPE-10. Note that the putative PEI-2 residue(s) modified by palmitoylation remain to be defined.

Biomolecular condensates are micrometre-scale assemblies of proteins, and often nucleic acids, that are not enclosed by limiting membranes. These have been proposed to form via phase separation, a process driven by low-affinity, multivalent interactions that cause polymers to de-mix into dense (granule) and dilute (cytoplasm) phases in a concentration-dependent manner¹¹. Molecules that self-associate using multivalent binding sites function as ‘scaffolds’ and recruit other molecules with lower valency as ‘clients’. Consistent with a potential scaffolding role, PEI-1 contains a BTB fold, a BACK domain, and an intrinsically disordered region (IDR), all of which could potentially engage in self interactions¹². Deletion of the BTB and BACK domains caused a partial reduction in the number of PEI granules, consistent with the role of these domains in promoting granule assembly. Notably, deletion of the PEI-1 IDR caused WAGO-3 to no longer co-localize with PEI-1 and remain in the residual body. PEI-1 missing the IDR, however, was expressed at reduced levels and did not assemble into robust granules, raising the possibility that the effect on WAGO-3 distribution is indirect. Overall, these data suggest that PEI-1 functions as a granule scaffold that recruits WAGO-3 as a client. Reconstitution experiments with purified proteins will be required to explore this model further.

An interesting question is what causes PEI granules to partition into spermatids. Correlative light and electron microscopy showed that PEI granules are directly adjacent to FB-MOs⁴. FB-MOs are partitioned into spermatids by SPE-15⁵, a myosin VI motor that is also required to partition PEI granules (Fig. 1b). Proper FB-MO biogenesis requires the transmembrane palmitoyltransferase SPE-10: in spe-10 mutants, FB-MOs disassemble and FBs are left behind in the residual body¹³. Schreier et al.⁴ found that in spe-10 mutants, PEI granules are misshapen and are unable to partition to spermatids. Furthermore, immunoprecipitation experiments with PEI-1 identified a second BTB and BACK domain protein, PEI-2, that seems to be modified in a SPE-10-dependent manner. PEI-2 localizes to PEI granules, and although not essential for PEI granule assembly, is required for partitioning to spermatids⁴. Together, these data suggest that palmitoylation of components of PEI granules mediates membrane anchoring and partitioning with FB-MOs during spermatid budding (Fig. 1c). Similar to wago-3 and pei-1 mutants, pei-2 mutants are defective in epigenetic inheritance, which confirms the idea that segregation of PEI granules into spermatids is essential for the paternal transmission of epigenetic information.

The discovery of PEI granules represents an exciting advance in our understanding of paternal epigenetic inheritance, and motivates questions about the composition, dynamics and functions of these structures. At present, only three proteins — WAGO-3, PEI-1 and PEI-2 — have been characterized as PEI granule components. Therefore, an important future goal will be to fully characterize the proteome of PEI granules. It also remains unclear whether PEI granules represent biomolecular condensates or a different type of cellular assembly. Schreier et al.⁴ used fluorescence recovery after photobleaching (FRAP) and 1,6-hexanediol treatment to show that PEI granules are relatively non-dynamic⁴. These types of assay are commonly used to probe material properties of potential condensates, but further experiments will be required to determine whether PEI granules form by phase separation.

The functional importance of PEI granule interaction with membranes is unknown. Previous studies have suggested that membrane association lowers the threshold for phase separation, as membranes restrict protein diffusion to a two-dimensional surface⁷. Membrane anchoring by PEI-2, therefore, could facilitate granule nucleation. PEI granules, however, still form in pei-2 mutants, which suggests that membrane anchoring in this case may be required only for segregation into spermatids. Asymmetric partitioning has also been observed for the oocyte-derived P granules, which segregate with the germline lineage during embryonic cleavages¹⁰. In zygotes, the segregation of P granules involves polarized disassembly and reassembly, but in later stages, the asymmetry of P granules seems to depend on association with nuclear membranes¹⁴. RNA granules in neurons have also been observed to ‘hitchhike’ on lysosomal membranes during axonal transport¹⁵. Membrane association may therefore be a common mechanism for localizing condensates in cells⁷.

Chromatoid bodies are granule-like structures that assemble in developing spermatids in mammals. Similar to PEI granules, chromatoid bodies contain Argonaute proteins and sRNAs, associate with vesicles, and are dynamic³. Unlike PEI granules, however, chromatoid bodies are not thought to persist in mature sperm. During spermatid elongation, chromatoid bodies split in half. One half is discarded in the residual body; the other associates transiently with the elongating sperm tail before disappearing from fully elongated spermatids³. It will be interesting to test whether mammalian homologues of PEI-1 and PEI-2 form granules that persist into mature sperm. The discovery of PEI granules offers a tractable system to explore the function and assembly of condensates in sperm. Many interesting questions remain, including the fate of PEI granules and their Argonaute or sRNA cargo after fertilization.