Developmentally programmed epigenome regulates cellular plasticity at the parental-to-zygote transition

During metazoan development, the dramatic potency change from germline to embryos raises an important question regarding how the new life cycle is reset. Here, we report a tightly regulated epigenome landscape change from the parental germline to embryos in C. elegans. The epigenome is enriched with histone H3 in early-stage germ cells but switches to a histone variant H3.3-enriched epigenome in the mature egg. This H3.3-dominant epigenome persists in early-stage embryos until gastrulation, when the epigenome becomes H3 abundant again. We further demonstrate that this developmentally programmed H3 → H3.3 → H3 epigenome landscape change is regulated through differential expression of distinct histone gene clusters and is required for both germline integrity and early embryonic cellular plasticity. Together, this study reveals that a bimodal expression of H3 versus H3.3 is important for epigenetic reprogramming during gametogenesis and embryonic plasticity. One Sentence Summary Developmentally programmed epigenome resets cellular plasticity at the parental-to-zygote transition in C. elegans.


Main text：
A fundamental and important developmental biology question is how a fertilized egg, upon division and differentiation, becomes a highly organized embryo. On the one hand, gametogenesis represents one of the most dramatic terminal cellular differentiation pathways. On the other hand, upon fertilization, the totipotency is regained in the zygote (1). During these processes, epigenetic information is robustly and extensively established, erased, and reset (2)(3)(4).
Both canonical histones and histone variants play important roles in epigenetic regulation. The canonical, or replication-coupled (RC) histones (i.e., H3, H4, H2A, H2B) are expressed during S-phase and mainly incorporated into the genome during DNA replication. Histone variants are typically replication-independent (RI), expressed throughout the cell cycle, and regulate a variety of biological processes (5)(6)(7). Intriguingly, unicellular organisms such as yeast only have H3.3like histones, whereas metazoans have both H3.3-like and H3-like histones, suggesting that H3 could be responsible for more specialized roles in metazoan development (8,9), such as regulating distinct cell fates.
Metazoan RC histones are typically encoded by gene clusters found at multiple chromosomal locations (10). For example, the human genome has 14 histone H3 genes on two chromosomes, while the C. elegans genome has 15 histone H3 genes on four chromosomes (11).
Due to the high sequence similarity among the multiple H3 genomic loci, defining contributions of individual histone H3 genes has remained a challenge due to the lack of precise genetic tools.
Furthermore, H3 and the histone variant H3.3 share a 97% similarity among their amino acid sequences, and are two of the most conserved proteins among all eukaryotic organisms (8,12).
Functionally, H3.3 is often associated with active transcription and enriched with posttranslational modifications (PTMs) such as H3K36me2 and H3K4me3 (16)(17)(18)(19)(20). In contrast, PTMs associated with more repressive chromatin, such as H3K27me2/3 and H3K9me2/3, occur preferentially on H3 (17)(18)(19). Furthermore, recent studies in several organisms have suggested a conserved role for H3.3 during gametogenesis and early embryonic development in mice (21)(22)(23)(24), Drosophila (25)(26)(27), and X. Laevis (28). In C. elegans, removal of H3.3 is not lethal, but reduces fertility and viability in response to stress (29). However, it is unknown whether and how individual H3 genes act in C. elegans and how they interact with H3.3. Here, we characterized expression patterns and roles of individual histone H3 genes, and compared them to the histone variant H3.3, during C. elegans gametogenesis and embryogenesis. We identified unique developmentally programmed expression and function of the 15 histone H3 genes, which uniquely influence the epigenome in the germline and restrict embryonic plasticity by providing a barrier against cellular reprogramming.
To investigate the spatiotemporal dynamics of each H3 gene cluster in the C. elegans genome, we used CRISPR/Cas9 genome engineering to systematically tag endogenous histone genes (30), leading to a series of histone H3 knock-in lines that produce fusion proteins with different fluorescent tags such as Dendra2, mCherry, and eGFP, at their C-termini (31). Using quantitative, live cell imaging to characterize histone H3 incorporation patterns, we uncovered two major classes of H3 genes based on their distinct expression patterns during gametogenesis and early embryogenesis. Surprisingly, only a small subset (4 of 15 H3 genes, Class I) are expressed in both the germline and somatic lineages, which include his-45, his-55, his-59, and his-63 ( Fig. 1A and B, fig. S1A and B). The majority of H3 genes (10 of 15, Class II) are not detectable in the germline but only in the somatic cell lineages, including members of the HIS1 (his-2), HIS2 (his-6), HIS3 (his-9, his-13, his-25, and his-42), HIS4 (his-17, his-27 and his- 49), and HIS5 (his-32) histone gene clusters ( Fig. 1B and fig. S1C). Notably, expression patterns among different H3 genes within the same gene cluster are consistent with each other. Of the 15 histone H3 genes, his-40 is unique considering it can only be detected in a subset of somatic lineages ( Fig. S1D and table S1). Consistent with our results, recent RNA profiling assays have revealed two major RC histone gene clusters based on their distinct expression patterns. The Class I histone H3 genes (his-45, his-55, his-59, and his- 63) show significant expression at the earliest embryonic timepoints as well as from the gonad, suggesting their potential roles in intergenerational epigenetic inheritance (32). In contrast, the Class II histone H3 genes show a largely zygotic expression pattern. The consistent patterns between histone H3 proteins and transcripts indicate that their expression is not subject to significant post-transcriptional regulation.
In the C. elegans gonad, germ cells are organized in a temporally and spatially organized manner along the longitude axis (Fig. 1A). Examination of the four Class I H3 genes revealed that his-45 is the highest expressed in the germline ( Fig. 1C and fig. S1B). In order to directly compare H3 versus H3.3 quantitatively in a spatiotemporally specific manner, we generated two CRISPR-tagged H3.3 strains, H3.3::Dendra2 and H3.3::mCherry, respectively ( Fig. 1B and fig.   S2A). We then used the H3.3::Dendra2 for quantitative temporal imaging analyses to compare with H3::Dendra2 strains. On the other hand, the H3.3::mCherry was used with different H3::eGFP strains for their potential distinct localization and chromosomal association. Based on these studies, we found that H3 and H3.3 display distinctly dynamic patterns: As germline nuclei move proximally, an approximate 4-fold increase of H3.3 was detected in late pachytene germ cells (Fig. 1D). In contrast, H3 is enriched in the distal mitotic region. In addition, the condensed state of chromosomes during meiotic prophase allows identification of individual chromosomes ( Fig. S1A). Using antibodies against specific histone PTMs, H3K36me2 is enriched on autosomes while the X chromosome is deficient of H3K36me2 but is primarily associated with H3K27me2/3, consistent with the repressive chromatin status for the X chromosome (33,34). In accordance with previous reports, H3.3 was found to accumulate on the autosomes but is depleted from the X chromosome, based on its anticorrelation with H3K27me2/3 and colocalization with H3K36me2 (35-37) (Fig. S2B). Contrastingly, germline-expressed H3 was enriched in H3.3-depleted regions, including the X chromosome, marked by its co-localization with H3K27me2/3 but anticorrelation with H3K36me2 (Fig. S2B). These results suggest that H3 carries unique modifications that establish and/or maintain repressed regions of the epigenome. Furthermore, we found both H3.3 and H3 are retained in mature sperm and oocyte chromatin ( Fig. S2A), indicating that they may regulate the transition from germline to zygote.
In C. elegans, embryos initiate zygotic transcription in the blastomeres beginning at the four-cell stage, while the onset of gastrulation begins at the 26-cell stage (38)(39)(40). Interestingly, we observed a low level of H3 but a high level of H3.3 throughout the rapid cell cycles of early embryogenesis ( Fig. 2A). However, when gastrulation initiates, a drastic increase of all 15 histone H3 genes was detected ( Fig. 2A-B). Consistent with the above results ( Figure 1), the expression timing of the H3 genes coincide with each other within the same class (i.e., Class I versus Class II), suggesting they are developmentally coordinated for co-expression. Using his-45 (Class I) and his-6 (Class II) to represent the two classes of H3 genes, the his-45 was detectable throughout embryogenesis, including the earliest cell divisions, and increased gradually upon gastrulation ( Fig. 2A-B). Remarkably, expression of the soma-specific his-6 gene was not detectable until the onset of gastrulation, and increased rapidly over the subsequent cell divisions ( Fig. 2A-B). Quantifying expression of these two H3 classes highlights the global change in the epigenome, from an H3.3-enriched epigenetic landscape in late germline and early embryos to an H3-enriched epigenome in late embryos ( Fig. 2A-C). Furthermore, this change is specific to the somatic lineage cells, as live cell imaging revealed that the P lineage for primordial germ cells does not display such an increased H3 incorporation throughout embryogenesis but remained enriched with H3.3 ( Fig. 2C and movie S1).
We next asked whether the germline expressed Class I H3 genes are required for normal gametogenesis and fertility. To address this question, we generated deletion alleles in two of the lowly expressed Class I H3 genes, his-59 and his-55, using CRISPR/Cas9-mediated genome editing. Interestingly, removal of his-59 and his-55 is already sufficient to result in detectable phenotypes, including reduced brood size and increased germline apoptosis. We observed a variable penetrance of shrunken and collapsed germlines that exhibit a reduction in germ cell number. To understand whether this germ cell loss is due to a reduction of meiotic or/and mitotic cells, we used a reporter that specifically labels the mitotic cells by GFP (41), while all germ cells are labeled by mCherry. Comparing wild-type and his-59; his-55 gonads, the progenitor mitotic zones were comparable, whereas the meiotic regions were greatly reduced in the double H3 mutant ( Fig. 3A and B). Consistently, his-59; his-55 double mutants displayed significantly reduced brood sizes (Fig. 3C). To assess whether the germ cell loss and brood size reduction were due to increased cell death, we used a CED-1::GFP reporter to visualize apoptotic germ cells in adults (42). Increased germ cell apoptosis was detected in the his-59 single H3 mutant, and such a phenotype is enhanced in the his-55; his-59 double mutant ( Fig. 3D and E). Taken together, these data indicate that compromising H3 levels in the germline by removing two Class I H3 genes causes increased germ cell apoptosis and germline atrophy. In contrast, previous studies demonstrate that H3.3 is dispensable, as knockout of multiple H3.3 genes did not lead to any detectable germline defects, using a strain that lacks all H3.3 homologues, including his-69, his-70, his-71, his-72, and his-74 (29). Taken together, these results indicate that H3 has an essential role in establishing or maintaining a unique chromatin state in the differentiating germline and this role cannot be compensated by H3.3.
Prior to gastrulation, embryonic blastomeres are characterized by decondensed chromatin and wide cellular differentiation capacity, termed as pluripotent (43)(44)(45). As embryos transit through gastrulation, they acquire distinct cell fates, accumulate heterochromatin, and restrict the developmental plasticity (44,46,47). To determine whether the increased incorporation of H3 during gastrulation is necessary for restricting embryonic plasticity, we generated a mutated form of H3 containing a single substitution of histidine (His, H) to aspartic acid (Asp, D) at the 113th position (H113D). The C-terminal H113 of H3 is a key residue that functions at the interface between the two H3-H4 dimers (H3 and H3' in Fig. 4A). By changing the positively charged H to the negatively charged D, the H113D mutant is thought to drastically destabilize the H3:H3' interface, preventing the complex formation between an H3-H4 dimer and the histone chaperone CAF-1. This mutant has been reported to have a dominant negative inhibition on CAF-1 mediated nucleosome deposition in vitro and in vivo (48,49). In order to avoid detrimental developmental defects, we generated this mutation at the his-6 locus, a somatic-specific Class II histone H3 gene, which limits the expression of the H113D restrictively in the somatic lineage and specifically at the onset of gastrulation. The his-6(H113D) mutant worms were viable, with no apparent embryonic abnormalities or change in the timing of embryogenesis. We then examined whether H113D acts dominant negatively to reduce the deposition of wild-type H3.
The molecular processes underlying declined cellular plasticity during embryogenesis have been previously investigated (44,45,50). For example, all somatic lineages can be converted into other somatic cell types, depending on the ectopic expression of master cell fate defining transcription factors (51)(52)(53). However, this reprogramming flexibility is gradually lost in cells from embryos after initiating gastrulation (44,47,51,(54)(55)(56). It has been shown that ectopic expression of the transcription factor che-1/Zn-finger, which is normally expressed solely in the bilaterally symmetric ASE neuron pair (ASEL and ASER), is sufficient to induce the expression of a neuronal cell fate reporter (gcy-5::GFP) and aberrant adoption of neuronal cell fates in early embryonic blastomeres (47,(57)(58)(59)(60). However, the ability of blastomeres to respond to ectopic expression of che-1 is progressively lost during gastrulation (47). To determine whether the developmentally programmed increase of H3 deposition during gastrulation restricts the embryonic plasticity, we challenged embryos to adopt alternative cell fates at progressively later developmental timepoints. Without ectopic expression of che-1, the neuronal cell fate reporter, gcy-5::GFP, is expressed solely in the ASER neuron in late-stage embryos ( Fig. 4E and   F). In this assay, non-ASE cells that actively express gcy-5::GFP in response to che-1 ectopic expression are considered plastic. When che-1 is induced during early embryonic stages (480 and 550 min.) in a wild-type background, gcy-5::GFP is broadly activated (Fig. 4F). Consistent with the progressive loss of plasticity, wild-type embryos that are challenged later in embryogenesis (670 and 760 min.) demonstrate a reduced ability to respond to ectopically expressed che-1 by turning on gcy-5::GFP (Fig. 4F). In contrast, in his-6(H113D) embryos, the number of cells that respond to che-1 overexpression exhibited a significant increase in neuronal cell fate induction and gcy-5::GFP expression, when compared to wild-type controls throughout embryogenesis (Fig. 4F). Together, these results demonstrate that histone H3 incorporation during gastrulation progressively restricts embryonic plasticity and reduced H3 incorporation is sufficient to extend the window of embryonic plasticity.
Despite their widely divergent morphologies, invertebrates and vertebrates go through largely similar early embryonic stages. For example, it has been shown that the de novo establishment of heterochromatin domains during early embryogenesis is conserved from Drosophila, C. elegans, Zebrafish, to mammals (46,(61)(62)(63). Heterochromatin is known to be a potent epigenetic barrier against cellular reprogramming in C. elegans, as well as in mammals (47,(64)(65)(66). In C. elegans, the timing of heterochromatin formation during embryogenesis has been tracked using transmission electron microscopy and immunostaining with antibodies against heterochromatin-enriched PTMs (46). These studies have provided significant evidence that heterochromatin domains are established during gastrulation, coinciding with the drastic increase of H3 gene expression and H3 incorporation identified in this work. Consistent with our results, extension of embryonic plasticity has also been identified in animals deficient in histone PTMs associated with heterochromatin, such as H3K9 and H3K27 methylation (44,47).
Furthermore, recent studies in C. elegans have identified that MET-2, an H3K9 methyltransferase, translocates from the cytosol to the nucleus during gastrulation. This changed localization and activity of the H3K9 methyltransferase may work together with the dynamic upregulation of chromatin-associated H3 to co-regulate heterochromatin formation (46). Intriguingly, mouse early embryos transit from H3.3-enriched epigenetic landscape that is distributed evenly across the genome to a more canonical pattern (67). Therefore, our finding addresses the mechanism governing the developmental timing of de novo heterochromatin formation with its potential roles in antagonizing the cellular plasticity.
Histone PTMs have also been found to transit from both maternal and paternal chromatin to the zygote in C. elegans, as well as in mammals (68)(69)(70). For example, H3K27me3, which is generated by the Polycomb Repressive Complex 2, can be inherited both maternally and paternally (69). Therefore, histone PTMs established during oogenesis and spermatogenesis transmit an epigenetic memory from gametes to the zygote (69). Future investigations will be needed to determine whether unique histone PTMs are inherited preferentially on histone H3 or H3.3, as well as where H3 and H3.3 are distributed throughout the parental and early embryonic genomes in C. elegans.
In metazoans, histone genes are clustered together. For example, in the human genome, a large cluster on chromosome 6 and two small clusters on chromosome 1 contain all the RC histone genes. While in the mouse, a large cluster on chromosome 13 and two small clusters on chromosome 3 contain all RC histone genes (11,71). In C. elegans, several clusters of histone genes are dispersed on four distinct chromosomes (Fig. S1). Here, our results demonstrate that the expression of each histone cluster is developmentally regulated for tightly coordinated expression, which is important for both germline fitness and embryonic development. Results from other metazoan embryos, such as sea urchin, suggest a potential conserved regulation of distinct histone gene clusters during embryogenesis (72,73). Therefore, our results start to shed light on the functional roles of selective histone cluster expression during development. Given the conserved features of histone genes in different metazoan species, it will be interesting to investigate whether such a programmed histone gene expression is a common feature and what cue(s) may trigger such a co-expression.