We report the isolation of the first mutant histone allele recovered from a genetic screen of a multicellular organism. This mutation, a gain-of-function allele of the C. elegans histone H3 gene his-9, causes an unexpectedly specific defect: the establishment of the MI-e3D bilateral asymmetry is disrupted. The his-9 gain-of-function activity acts cell-autonomously in the MI mother cell and affects a process downstream of or parallel to the left-right asymmetric expression of the transcription factors NGN-1 and HLH-2. The CAF-1 protein complex is required for the MI-e3D asymmetry, and the mutant HIS-9 protein acts by inhibiting CAF-1-mediated nucleosome formation. These results suggest that CAF-1-mediated nucleosome assembly during S phase of the MI mother cell is required to specify the MI neuronal fate. The newly assembled nucleosomes might then be transmitted to the post-mitotic MI neuron and establish an MI-neuronal chromatin state.
It has been a matter of debate whether CAF-1, which interacts with histone H3.1 in mammalian cells (Tagami et al., 2004
), binds to an H3-H4 dimer consisting of one H3 and one H4 polypeptide or an H3-H4 tetramer consisting of two H3 and two H4 polypeptides; it has also been unclear whether CAF-1 deposits an H3-H4 dimer or an H3-H4 tetramer onto replicating DNA (Margueron and Reinberg, 2010
; Ransom et al., 2010
; Tagami et al., 2004
; Xu et al., 2010
). These issues are important for understanding the mechanistic basis of epigenetic inheritance of histone modifications: if CAF-1 binds to and deposits a single H3-H4 dimer, parental H3-H4 tetramers might be split and distributed onto daughter DNA strands as two H3-H4 dimers, each of which could then be paired with a newly synthesized H3-H4 dimer, thereby propagating parental histone modifications to the two daughter DNA strands (Tagami et al., 2004
); by contrast, a parental H3-H4 tetramer would be segregated to only one of the two daughter DNA strands if CAF-1 binds to and deposits an H3-H4 tetramer or if two CAF-1 proteins each bind to an H3-H4 dimer and interact and then together deposit an H3-H4 tetramer (Jackson, 1988
; Leffak et al., 1977
; Xu et al., 2010
). Our results demonstrate that mutant HIS-9 proteins inhibit CAF-1-mediated nucleosome formation even in the presence of wild-type H3 proteins. This observation suggests that mutant HIS-9 proteins have a dominant-negative effect and act to reduce the level of functional CAF-1 by directly interacting with CAF-1. Because mutant HIS-9 proteins are likely defective in H3-H4 tetramer formation and hence able to form a HIS-9-H4 dimer but not an H3-H4 tetramer, our findings strongly support the model that CAF-1 binds to an H3-H4 dimer. Given a previous report that CAF-1-CAF-1 dimer interaction is important for chromatin assembly (Quivy et al., 2001
), we suggest that the mutant HIS-9 H3.1-like protein forms a dimer with a histone H4 protein and then this mutant HIS-9-H4 dimer binds to CAF-1, forming a complex unable to interact with a second CAF-1-H3-H4 complex through the H3-H3 interaction surface and a CAF-1-CAF-1 dimer interaction surface; in this way, the mutant HIS-9 protein could reduce the level of functional CAF-1 (). Thus, our results suggest that during normal replication-coupled chromatin assembly CAF-1 binds to an H3-H4 dimer and that two CAF-1-H3-H4 complexes assemble to form an H3-H4 tetramer, thereby depositing an H3-H4 tetramer that consists entirely of newly synthesized H3 and H4 proteins. During this process of DNA replication, a parental H3-H4 tetramer remains intact and is transferred to newly replicated DNA.
Our observations also reveal that the two C. elegans homologs of ASF1 are not required to establish the MI-e3D asymmetry, indicating that an ASF1-independent pathway exists that leads to complex formation between CAF-1 and an H3-H4 dimer.
We previously showed that the specification of the MI-e3D asymmetry occurs at least three cell generations prior to the generation of MI and e3D within two sister cells, ABaraapa and ABaraapp, that first separate the left and right branches of the ABaraap cell lineage (Nakano et al., 2010
). This initial asymmetry is transduced into the MI mother cell by a CEH-36/NGN-1/HLH-2 transcriptional cascade that acts through multiple rounds of cell divisions. The CEH-36 protein is present in the MI grandmother cell and promotes expression of two proneural bHLH proteins, NGN-1 and HLH-2. The NGN-1 and HLH-2 proteins form a dimer and are present in the MI mother cell. Given previous reports that mammalian homologs of these bHLH proteins regulate histone modifications (Beck et al., 2009
; Pujadas et al., 2011
), we suggest that the NGN-1/HLH-2 complex recruits histone-modifying enzymes that act on the nucleosomal arrays assembled by CAF-1 to generate epigenetic marks necessary for the specification of the MI neuron (). The molecular memory of the developmental cue established by the CEH-36/NGN-1/HLH-2 pathway might be transmitted to the post-mitotic MI neuron by CAF-1-mediated replication-coupled chromatin assembly in the MI mother cell, generating a stable mark of the MI-neuronal chromatin state. In addition, because expression of NGN-1 and HLH-2 is left-right asymmetric and these proteins are present in the MI mother cell but not in the e3D mother cell (Nakano et al., 2010
), CAF-1 and the NGN-1/HLH-2 complex likely establish a left-right asymmetry in chromatin states to generate the MI-e3D asymmetry.
Model for the Specification of MI by the CAF-1-PCN-1 Complex
The effect of the gain-of-function mutations in histone H3 genes is strikingly specific to MI cell-fate specification, which suggests that a mechanism of specifying the MI neuron is particularly sensitive to histone H3 gain-of-function mutations. One possible mechanism could be the generation of differing nucleosome densities at a locus important for MI cell-fate specification between two sister chromatids during S phase of the MI mother cell (). Specifically, it has been proposed that human CAF-1 and PCNA can generate a difference in the density of newly synthesized nucleosomes at a given locus between two sister chromatids (Shibahara and Stillman, 1999
). Upon selective segregation of chromatids, such distinct chromatin complexes would generate a difference between sister cells during development. MI fate specification could involve such a mechanism, so that during S phase of the MI mother cell, PCN-1 and CAF-1 generate differing nucleosome densities between two sister chromatids at loci required to specify the MI neuronal fate. We propose that such newly synthesized nucleosomes are modified by the NGN-1/HLH-2 complex, leading to the generation of differing densities of modified histones between two sister chromatids. There would then be a selective inheritance of these distinct chromatin structures to the MI neuron and its sister m1DR muscle cell (). In this way, both NGN-1/HLH-2-mediated histone modifications and CAF-1-mediated specific nucleosome density would be required for expression of target genes necessary to specify the MI neuronal fate.
An alternative possibility is that the NGN-1/HLH-2 complex and CAF-1 could act on distinct sets of target genes. In this case, the NGN-1/HLH-2 complex might modify histones at one locus that are deposited by a replication-independent pathway, and CAF-1/PCN-1 might generate differing nucleosome densities at another locus between sister chromatids. Nonetheless, the simplest hypothesis is that both act on the same target genes to together control distinct aspects of a left-right asymmetric epigenetic regulation that specifies the MI bilateral asymmetry.
Such a mechanism of a selective inheritance of sister chromatids might also be involved in left-right axis determination in mammals. inversus viscerum
) mutant mice were originally reported to show randomized directionality of their visceral asymmetries (Hummel and Chapman, 1959
). The iv
locus encodes a microtubule-based motor protein, called left-right dynein (LRD) (Supp et al., 1997
), that is required for nodal cilia motility and nodal flow (Okada et al., 1999
). Interestingly, LRD has been implicated to play a role in the biased chromatid segregation of mouse chromosome 7 in embryonic stem cells, endoderm and neuroectoderm (Armakolas and Klar, 2007
), and recently it was reported that the iv
mutant hippocampus exhibits a loss of a bilateral asymmetry rather than randomized directionality of the asymmetry (Kawakami et al., 2008
). Our findings provide a link between these observations and suggest that LRD might promote a nervous system bilateral asymmetry through biased segregation of chromatids in a manner independent of nodal signaling. We suggest that selective segregation of chromatids marked with distinct chromatin states could be a conserved epigenetic mechanism that generates bilateral asymmetries in the nervous system in organisms as distant as nematodes and mammals.