E3 Ligase for CENP-A (Part 2)

Abstract

Centromeric CENP-A, a variant of histone H3, plays a central role in proper chromosome segregation and its function is highly conserved among different species. In most species with regional centromeres, an active centromere relies not on defined DNA sequences, but on the presence of CENP-A proteins in centromeric nucleosomes. CENP-A is proposed to be the non-DNA indicator (epigenetic mark) that defines proper centromere assembly and function. Recently, many post-translational modifications (PTMs) of CENP-A and their functions have been reported. They revealed the importance of the functions of CENP-A PTMs in CENP-A deposition at centromeres, proteolysis/protein stability, and recruitment of other centromere-kinetochore proteins. Ubiquitylation and sumoylation by E3 ligases regulate multiple functions, including proteolysis and signaling, and play important roles in the cell cycle and mitotic control. Recently, the function of E3 ligase that ubiquitylates/sumoylates and controls CENP-A protein has emerged as an important regulatory paradigm in different species. Many have reported the importance of CENP-A ubiquitylation and sumoylation in CENP-A deposition at centromeres and for protein stability, which is regulated by specific E3 ligases. Therefore, here we summarize what is known about the E3 ligases for CENP-A ubiquitylation and sumoylation and their biological functions and significance in different species.

1. Introduction

During cell division, proper chromosome segregation must be achieved to avoid unequal distribution of chromosomes to daughter cells. Spindle microtubules must attach to a single region of each chromosome, termed the “centromere,” in most eukaryotes. The kinetochore is a complex of proteins that are located at the centromere. Defects in the centromere-kinetochore and spindle check point functions lead to aneuploidy and cancer and are often associated with a poor prognosis. Therefore, it is highly important to study the spatiotemporal regulation and the structures of centromere and kinetochore proteins to understand chromosome instability (CIN) during development and cancer progression. The key question is how the chromosomal location and function of a centromere (i.e., centromere identity) are determined and thus participate in accurate chromosome segregation. In most species with regional centromeres (see the previous chapter for an exception of the budding yeast Saccharomyces cerevisiae that has genetically defined point centromeres), centromere identity relies not on a defined DNA sequence, but on the presence of a special nucleosome that contains a specific histone-like protein called CENP-A. CENP-A is proposed to be the non-DNA indicator (epigenetic mark) of centromere identity. CENP-A partially replaces histone H3 in the centromeric regions. CENP-A-containing nucleosomes are the basis for kinetochore formation and are the most important marker for centromere function in eukaryotes [1].

The structure of CENP-A-containing nucleosomes is more compact than H3-containing nucleosomes [2, 3, 4]. Although it is commonly reported that CENP-A-containing nucleosomes are formed with the canonical histones H2A, H2B, and H4 at the active centromeres, their structure remains controversial among different research groups [5]. CENP-A is at the top of a hierarchy of the pathway that determines the assembly of kinetochore components [6], and how CENP-A defines the position of the centromere in humans is the fundamental question. While the function of CENP-A protein is highly conserved among most eukaryotes, its protein sequence has apparently undergone both convergent and divergent evolution [7], and the centromere DNA repeats with which the CENPA-containing nucleosome interacts are also highly diverged. The architectures of CENP-A chromatin with quantified numbers of CENP-A (CenH3) molecules (e.g., ~400 molecules of human CENP-A/kinetochore) have been reported using fluorescence microscope assays among different species [8, 9, 10, 11]. CENP-A is also called CenH3 (centromere-specific histone H3). Its homologs in different species are summarized in Table 1.

Species	CENP-A homolog	E3 ligase (ubiquitylation or sumoylation)	Function	Preceding PTMs before ubiquitylation or sumoylation	Another proposed factor relevant to the E3 function
Saccharomyces pombe	Cnp1/SpCENP-A	N.D.	Proteasomal degradation to remove non-centromeric Cnp1	N.D.	N-terminal domain of Cnp1, Overexpression of H3/H4
Drosophila melanogaster	CID/Cid	CUL3/RDX (ubiquitylation)	Interacts with CAL3 and promotes CAL3 function, loading and stabilizing (maintenance) of CID protein at centromeres (proteasomal independent mechanism)	N.D.	N.D.
		SCF^Ppa (ubiquitylation)	Prevents the promiscuous incorporation of CID across chromatin during replication, (targeting CID that is not in complex with CAL1)	S20 phosphorylation	S20 phosphorylation
		APC/C^Cdh1 (ubiquitylation)	Degradation of the CAL1-CID complex (likely regulates centromeric CID deposition)	N.D.	N.D.
Homo sapiens	CENP-A	CUL4A/RBX1/COPS8	Facilitate interaction of CENP-A with HJURP through CENP-A ubiquitylation, CENP-A deposition at the centromere (proteasomal independent mechanism)	N.D.	COPS8 as an adaptor, heterodimerization of CENP-A, SUGT1-HSP90
Arabidopsis thaliana (CENH3 was expressed in Nicotiana tabacum)	AtCENH3	N.D. (VHHGFP4-human SPOP as synthetic E3 ligase expressed in Nicotiana tabacum)	Proteasomal degradation of AtCENH3	N.D.	N.D.

Table 1.

E3 ligases for CENP-A in species with regional centromeres.

Note: E3s of some species (e.g., Caenorhabditis elegans, Xenopus laevis, zebrafish Danio rerio, chicken Gallus domesticus DT40 cells, Mus musculus, etc.) are not discovered, and so are not described in this table. N.D. = not determined; PTMs = post-translational modifications.

CENP-A contains a short centromere targeting domain (CATD) within the histone fold region [2] in the C-terminus. Replacement of the corresponding region of histone H3 with the CATD is sufficient to direct histone H3 to the centromere [2], and this chimeric histone can rescue the viability of CENP-A-depleted cells [2, 12]. The CENP-A C-terminus contains another tail domain that recruits CENP-C to promote centromere and kinetochore assembly [13, 14]. CENP-N was also identified as the first protein to selectively bind CENP-A nucleosomes but not H3 nucleosomes during centromere assembly [15].

Meanwhile, the functions of the N-terminal CENP-A are also reported for some species [16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27] (see also previous chapter, Sections 2.1, 2.2.2, 2.4.1, 2.9 and this chapter, Sections 2.1, 2.3, 2.4, 3.1, 4.1, 4.6, and 5.1). Loading of CENP-A at centromeres and its incorporation/deposition and maintenance in centromeric chromatin is cell cycle-regulated. In cells overexpressing CENP-A, the ectopic protein incorporates throughout the chromatin in interphase [28]. By the next G1, however, mis-incorporated CENP-A seems to have been cleared from chromatin by a mechanism that likely involves ubiquitin-mediated proteolysis, as suggested by experiments in yeast and Drosophila cells [28]. Importantly, the timing of deposition of newly synthesized CENP-A within the cell cycle may be variable not only among different species but also different developmental stages within the same species. Yeast suppressor of chromosome missegregation protein 3 (Scm3) [29] (previous chapter, Figure 1; Table 1) is a distant counterpart of human Holliday junction recognition protein (HJURP) (Figures 2 and 3; Table 1), and they are CENP-A (CenH3)-specific chromatin assembly factors [29, 41, 42, 43]. The incorporation of newly synthesized CenH3 (CENP-A) into centromeric nucleosomes depends on Scm3/HJURP [41, 42, 43] in budding, fission yeasts, and humans. In addition, other components and factors that contribute to CENP-A deposition, maintenance, and inheritance in centromeric nucleosomes have been reported [28, 44].

Figure 1.

Mechanistic scheme for Drosophila melanogaster CENP-A^CID pathways. (Right) (a) Bade et al. proposed the role of the E3 ligase CUL3/RDX in CUL1-dependent ubiquitylation of CID [30, 31]. CUL3/RDX activity, which is presumably in dimer form [30], leads to the monoubiquitylation of CID through an interaction with the RDX-binding sites (RBS) of CAL1. This monoubiquitylation of CID is proteasomal-independent but is required for stable localization of CID and CAL1 to the centromere. The ubiquitin-conjugating enzyme (E2) bound on RBX1 is omitted for simplicity. (b) In the absence of RDX, CENP-A is not monoubiquitylated by CUL3, and both CENP-A and CAL1 are subjected to proteasome-dependent degradation, but presumably ubiquitylated by (c) SCF^Ppa or (d) APC/C^Cdh1 as Moreno-Moreno proposed [32] (see below). Furthermore, the absence of RDX results in cell death and severe chromosomal aberration (e.g., chromosome fragmentation), some of which may be attributed to the loss of CENP-A and CAL1 from centromeric regions (not shown in this cartoon). Moreno-Moreno et al. suggest that (c) whereas SCF^Ppa targets the fraction of CID that is not in complex with CAL1, (d) APC/C^Cdh1 contributes to the degradation of the CAL1-CID complex and, thus, likely regulates centromeric CID deposition [32] as previously proposed [30]. Huang et al. proposed that phosphorylation of CID of serine 20 (S20) regulates both protein turn-over and centromere-specific loading [33] (see also left). The CID S20 phosphorylation renders CID a substrate for ubiquitylation by SCF^Ppa, thereby regulating the abundance of free pre-nucleosomal CID through subsequent proteasomal degradation (see also left (g)). (Left) (e) The role of DLP/DAXX in CID deposition into ectopic nucleosomes through CID ubiquitylation as proposed in a human cell model [34] has yet to be confirmed experimentally. (f) CTCF occlusion by the aberrant nucleosome of heterotypic tetramer consisting of CENP-A-H4 with H3.3-H4 as proposed in human cell models [34] has not been confirmed in D. melanogaster. (g) Huang et al. observed that CID S20 is phosphorylated by casein kinase II (CK2) not only insoluble but also chromatin-bound CID, and this phosphorylation also facilitates removal of CID from ectopic but not from centromeric sites in chromatin [33]. (h) Factors/components that stabilize ectopically incorporated CID and are required for neocentromere formation and its maintenance are not yet clear. The status of overall PTMs, including polyubiquitylation of CID, especially in the ectopic nucleosome, has yet to be elucidated.

Figure 2.

Mechanistic scheme for human CENP-A pathways. (Right) In normal conditions, CUL4A-RBX1-COPS8 E3 ligase activity is required for CENP-A mono- or di-ubiquitylation on lysine 124 (K124) and CENP-A centromere localization [35]. CENP-A K124 mono- or di-ubiquitylation is required for CENP-A’s interaction with the chromatin assembly factor HJURP and CENP-A deposition at the centromere. The CUL4A complex targets CENP-A through the adaptor COPS8/CSN8 that has WD40 motifs. In non-canonical CRL4 machinery, CUL4/RBX1/COPS8 may dimerize as a CUL4/DACAF1 complex [36, 37], but the dimerization unit remains unknown [31]. Here only the CUL4/RBX1/COPS8 monomer is shown for simplicity. Upstream, the SGT1-HSP90 complex is required for the composition of the CUL4A complex and recognition of COPS8 to target CENP-A. Therefore, the SGT1-HSP90 complex is also required for CENP-A ubiquitylation and localization of CENP-A to centromeres. “CA” refers to the CENP-A monomer. “Ace” refers to the putative acetylated lysine 124 (K124) previously reported by Bui et al. that is concurrent with the structural transitions of CENP-A-containing nucleosomes through the cell cycle [38]. Their computational modeling suggests that acetylation of K124 causes tightening of the histone core and hampers accessibility to its C-terminus, which in turn reduces CENP-C interaction [39] (not shown in this cartoon). However, its precise function and relationship with K124 ubiquitylation remain to be studied. (Left) When human CENP-A is overexpressed, CENP-A is incorporated into ectopic nucleosomes consisting of a heterotypic tetramer that contains CENP-A-H4 with H3.3-H4 [34]. This ectopic localization of this particle (aberrant nucleosome) depends on the H3.3 chaperone DAXX rather than the centromeric CENP-A-specific chaperone HJURP. (a) Post-translational modifications of human CENP-A, especially before recognition by DAXX and after incorporation into the ectopic nucleosome, have yet to be elucidated. (b) CTCF occlusion by the aberrant nucleosome of a heterotypic tetramer consisting of CENP-A-H4 with H3.3-H4 was also proposed in a human cell model [34], but specific DAXX localization on these CTCF sites under CENP-A overexpression has not been confirmed experimentally. (c) Factors/components that stabilize ectopically incorporated CENP-A and are required for neocentromere formation and its maintenance are not yet clear. In addition, the status of overall post-translational modifications, including polyubiquitylation of CENP-A, especially in the ectopic nucleosome, is unknown. (d) Currently, the proteolysis mechanism for mis-incorporated human CENP-A and its E3 ligase is not yet clear. Note that histone H4 and phosphorylation of human CENP-A are omitted for simplicity.

Figure 3.

Models of epigenetic inheritance of CENP-A ubiquitylation through heterodimerization. In the octamer model, two CENP-A dimers in one nucleosome are split/diluted between the two daughter centromere-DNA sequences, and one CENP-A molecule replaces with one H3 molecule or leaves a molecule-free space during the replication/S phase. HJURP (Holliday junction recognition protein) predominantly interacts with ubiquitylated, preassembled “old” CENP-A, which resides mostly in nucleosomes. A non-ubiquitylated newly synthesized (“new”) CENP-A monomer targets ubiquitylated centromeric CENP-A through preassembled HJURP. Note that histone H4 is omitted for simplicity. (a) New CENP-A is appropriately ubiquitylated in a heterodimerization-dependent manner (i.e., dimers of old CENP-A with new CENP-A). In this way, both ubiquitylation and the location of the centromere are inherited epigenetically. (b) If K124 ubiquitylation does not occur on new CENP-A, the non-ubiquitylated CENP-A nucleosome distributed during the S phase does not recruit HJURP to the centromere because the affinity of non-ubiquitylated new CENP-A to HJURP is low. Subsequently, this loss of localization of HJURP at the centromere leads to the lack of new CENP-A targeting to ubiquitylated centromeric CENP-A through HJURP, and eventually to the lack of new CENP-A deposition. This figure is partly adapted from Niikura et al. [40].

Recently, many post-translational modifications of CENP-A and their functions have been reported [45]. They revealed the importance of these changes in CENP-A deposition at centromeres, proteolysis/protein stability, and recruitment of the CCAN (constitutive centromere-associated network) proteins [45]. Thus, here we focus on E3 ligase activities (i.e., on ubiquitylation and sumoylation) of CENP-A and summarize these functions for each species with regional centromeres in the following sections.

2. E3 ligase for fission yeast (Schizosaccharomyces pombe) CENP-A^Cnp1 and its function

2.1 Overview of CENP-A^Cnp1

Fission yeast (Schizosaccharomyces pombe) centromeres consist of large (40–100 kb) inverted repeats that display heterochromatic features. Therefore, fission yeast provides a good model for higher eukaryotic centromeres. The mechanistic processes to establish centromeric chromatin of fission yeast and its structures have been reviewed [46, 47]. This section focuses on the E3 ligase(s) for fission yeast CENP-A^Cnp1 and its function through ubiquitylation, although endogenous E3 ligase for S. pombe CENP-A^Cnp1 is not yet identified and its specific regulation is still unclear.

In fission yeast, the recruitment of the CENP-A-specific chaperone to the centromere is an essential step in epigenetic inheritance. The fission yeast Scm3 could be functionally homologous to HJURP. It interacts with CENP-A, localizes to centromeres during most of the cell cycle (except in mitosis), and is required for CENP-A deposition [48, 49]. Sequence analysis revealed a shared common domain in Scm3 and HJURP proteins [29]. Dunleavy et al. identified another chaperone known as Sim3 (start independent of mitosis 3) in fission yeast [50, 51]. Sim3 is homologous to known histone binding proteins NASP (human) and N1/N2 (xenopus) and aligns with Hif1 (S. cerevisiae), defining the SHNi-TPR family [51]. Sim3 is distributed throughout the nucleoplasm, yet it associates with CENP-A^Cnp1 and also binds H3. It interacts also with non-chromosomal CENP-A and is required for its incorporation in S. pombe. These results are consistent with those in Arabidopsis thaliana [19] (see also Section 5.1). Sim3 also has been proposed to share some common roles with the histone chaperone Asf1, mutations in which cause a defect in overall chromatin structure [52, 53]. It has been suggested that Sim3 could function as an escort chaperone, handing off CENP-A to Scm3, a role that human HJURP may accomplish by itself [43, 48, 50].

Mis16 (human homologs of Mis16 are RbAp46 and RbAp48) and Mis18 (human homologs of Mis18 are Mis18α and Mis18β) are required for loading of newly synthesized Cnp1/CENP-A into centromeric chromatin [54, 55], but are absent from organisms with point centromeres [44] (see also previous chapter, Section 2.3.3 and this chapter, Sections 3.1 and 4.1). Mis16 and Mis18 are also required for the maintenance of the hypoacetylation of histone H4 specifically within the central domain of the centromere [55], and Mis16 homologs are components of several histone chaperon complexes [56]. Moreover, acetylation of histone H4 lysine 5 and 12 (H4K5ac and H4K12ac) within the pre-nucleosomal CENP-A-H4-HJURP complex mediated by the RbAp46/48-Hat1 complex is required for CENP-A deposition into centromeres in chicken and humans [57], consistent with Hat1’s role in Drosophila melanogaster [58] (see also Sections 3.1 and 4.1). In mouse studies, Mis18α interacts with DNMT3A/3B, and this interaction is required to maintain DNA methylation [59]. Mis18α deficiency leads to not only the reduction of DNA methylation, but altered histone H3 modifications, and uncontrolled noncoding transcripts in the centromere region (see also Section 4.1). It is an interesting model that Mis16 and Mis18 complexes “prime centromeres” affect post-translational modifications of histone H3/H4 proteins and centromeric DNA in advance of CENP-A incorporation. How such chromatin structures feedback with the regulation of E3 ligases of CENP-A has not yet been reported, which could be important.

In S. pombe, spMis16, and spMis18 mutants eliminate Cnp1 incorporation to centromeres and Mis18 directly interacts with Scm3 in vitro, suggesting they cooperate to assemble Cnp1 into centromeric chromatin [48]. S. pombe lacks the vertebrate Mis18BP1 ortholog, and the Mis18BP1 function in S. pombe is replaced by the Eic1 protein (a.k.a Mis19) [44, 60, 61]. While Eic2 (a.k.a Mis20) is dispensable for the recruitment of Cnp1 to the centromere, Eic1 is required for the recruitment of the Mis18, Mis16, and Scm3 proteins to the centromere and Cnp1 incorporation. Both of the Eic1 and Eic2 proteins co-purify with the spMis18 and exhibit a similar centromeric localization throughout the cell cycle [60, 61]. Taken together, these data suggest that Eic1 is functionally analogous to the Mis18BP1 subunit [60, 61]. However, Eic1 is evolutionarily distinct and no homolog of Mis19 has been found in the human genome, and Eic1 does not share any apparent sequence homology to Mis18BP1 [60, 61]. Centromere localization and function of Mis18 require Yippee-like domain-mediated oligomerization [62]. Furthermore, there are at least two mechanisms to restrict the assembly of CENP-A nucleosomes in G1—disruption of Mis18 multimerization by HJURP-Mis18 interaction, and ubiquitylation and degradation of Mis18β through SCF^βTrCP E3 ligase [44].

Domain-specific function, such as the N-terminal function, of fission yeast Cnp1/CENP-A is also reported as budding yeast Cse4 [24, 25] (see also previous chapter, Section 2.4). Folco et al. demonstrated that alteration of the Cnp1 N-tail does not affect Cnp1 loading at centromeres, outer kinetochore recruitment, or spindle checkpoint signaling but significantly increases chromosome loss [17]. On the other hand, their N-tail mutants exhibit centromere inactivation enhanced by an altered centromere. The N-tail mutants specifically reduced localization of the CCAN proteins CENP-T^Cnp20 and CENP-I^Mis6, but not CENP-C^Cnp3. Therefore, these authors suggest that the Cnp1 N-tail maintains the epigenetic stability of centromeres in fission yeast, at least in part via assembly of the CENP-T branch of the CCAN. Tan et al. identified a proline-rich “GRANT” (Genomic stability Regulating site within CENP-A N-Terminus) motif that is essential for Cnp1 centromeric targeting [24]. They showed that especially GRANT proline-15 (P15) undergoes cis-trans isomerization to drive proper chromosome segregation. This cis-trans isomerization appears to be carried out by two FK506-binding protein (FKBP) family prolyl cis-trans isomerases. In addition, they identified Sim3 as a Cnp1 NTD interacting protein that is dependent on GRANT proline residues. Together, they suggest cis-trans proline isomerization of Cnp1 is required for precise propagation of centromeric integrity in fission yeast, presumably via targeting Cnp1 to the centromere. Thus, the requirement of cis-trans proline isomerization of CenH3^Cnp1 in fission yeast studies appears to be consistent with the one of CenH3^Cse4 proposed in budding yeast studies [63] (see also previous chapter, Section 2.2.3). However, they suggest that the GRANT-prolines of Cnp1 do not coordinate proteolysis of the SpCENP-A protein as do proline residues in the budding yeast Cse4 NTD. In addition, Tan et al. showed that sequential truncation of the NTD did not improve the stability of the protein, suggesting that the NTD of Cnp1 does not regulate the turnover of the protein [25]. Instead, they proposed that heterochromatin integrity may contribute to Cnp1 stability and promote its chromatin incorporation.

Compared to the studies of budding yeast and some of the other species, currently, there are few studies on post-translational modifications and domain-specific functions of fission yeast CenH3/Cnp1. Further research is required on the relationships among Cnp1 post-translational modifications, structural change, interaction with its chaperones (e.g., Scm3 and Sim3), and surrounding heterochromatin regulation.

2.2 Dos1/2-Cdc20 complex

In S. cerevisiae, all pre-existing CENP-A is replaced by newly synthesized CENP-A during the S phase [64], whereas in S. pombe, two pathways of CENP-A deposition exist at the S and G2 phases of the cell cycle [50, 65]. Parental CENP-A is deposited at centromeres during the S phase, whereas newly synthesized CENP-A is deposited during later stages of the cell cycle [66]. The mechanisms involved in the deposition of CENP-A at centromeres during the S phase remain poorly understood [66]. In S. pombe, the GATA-like transcription factor Ams2, a key factor in CENP-A deposition during the S phase, appears to work, at least in part, through the regulation of transcription of core histones [65].

Li et al. reported that the DNA polymerase (Pol) epsilon catalytic subunit A (pol2), Cdc20, interacts with the Dos1-Dos2 silencing complex to facilitate heterochromatin assembly and inheritance of H3K9 methylation during the S phase [67]. We note that fission yeast S. pombe Cdc20 (UniProtKB—P87154) is not the ortholog of human CDC20 (cell division cycle protein 20 homolog, UniProtKB—Q12834), but of human POLE (UniProtKB—Q07864). Gonzalez et al. showed that the Dos1/2-Cdc20 complex is also required for localization of Cnp1 at centromeres at this stage [66]. Disruption of Dos1 (also known as Raf1/Clr8/Cmc1), Dos2 (also known as Raf2/Clr7/Cmc2), or Cdc20, a DNA polymerase epsilon subunit, leads to delocalization of CENP-A from centromeres and mislocalization of the protein to ectopic (non-centromeric) sites. All three mutants of Dos1, Dos2, and Cdc20 exhibit spindle disorganization and mitotic defects. Inactivation of Dos1 or Cdc20 also results in the accumulation of noncoding RNA transcripts from centromeric cores, a feature common to mutants affecting kinetochore integrity. These authors found that Dos1 physically associates with Ams2 and contributes to the interaction of Ams2 with centromeric cores during the S phase. They further showed that Dos2 associates with centromeric cores during the S phase and that its recruitment to centromeric cores depends on Cdc20. Together, this study identifies a physical link between DNA replication and the CENP-A assembly machinery and provides mechanistic insight into how CENP-A is faithfully inherited during the S phase.

It is important to clarify how exactly the Dos1-Dos2-Cdc20 complex contributes to the inheritance of preexisting Cnp1 during centromere replication [66]. Interestingly, Rik1 is a component of silencing factors. The heterochromatic methylation of histone H3-K9 by Clr4 is promoted by silencing factors: Dos1-Dos2-Rik1-Lid2 [67]. Horn et al. reported that subunits of a cullin-dependent E3 ubiquitin ligase interact with Rik1 and Clr4, and Rik1-TAP preparations exhibit robust E3 ubiquitin ligase activity [68]. They also demonstrated that the expression of a dominant-negative allele of the Pcu4 cullin subunit (the human Cullin-4 homolog) disrupts the regulation of K4 methylation within heterochromatin. Hong et al. also reported a novel complex that associates with the Clr4 methyltransferase, termed the CLRC (CLr4-Rik1-Cul4) complex using affinity purification of Rik1, and found that Rik1 interacts with the fission yeast Cullin4 (Cul4, encoded by cul4⁺), the ubiquitin-like protein, Ned8, and two previously uncharacterized proteins, designated Cmc1 and Cmc2 [69]. They also demonstrated a defect in the processing of noncoding RNA to small RNA caused by the defective Clr4-Rik1-Cul4 complex, suggesting that the components of the Clr4-Rik1-Cul4 complex collaborate at an early step in heterochromatin formation. Unlike the studies of CUL3/RDX in fruit flies (Figure 1, right; see also Section 3), the function of Cul4 E3 ligase targeting non-centromeric CENP-A^Cnp1 and the mechanism of its proteolysis are not yet studied in fission yeast. In fission yeast, there is no report about the involvement of Cul4 E3 ligase in CENP-A^Cnp1 deposition at the centromere, unlike in humans (Figure 2, right; see also Section 4). On the other hand, it would be interesting to test if the Cul4 E3 function for heterochromatin assembly is conserved in other species, including humans.

2.3 Assembly of Cnp1 at non-centromeric chromatin

Consistent with the results in budding yeast Cse4 [23, 70, 71] (see also the previous chapter, Section 2.1), Gonzalez et al. reported that the overexpression of fission yeast Cnp1 results in the assembly of Cnp1 at non-centromeric chromatin during mitosis and meiosis [18]. The non-centromeric Cnp1 is preferentially recruited near heterochromatin and is able to recruit kinetochore components, and Cnp1 overexpression leads to severe chromosome missegregation and spindle microtubule disorganization. Moreover, ectopic Cnp1-containing chromatin is inherited over multiple generations using pulse induction of Cnp1 overexpression. Interestingly, ectopic assembly of Cnp1 is suppressed by overexpression of histone H3 or H4 (Table 1), as other groups suggest that the balance between histones H3 and H4 and CENP-A is important for centromeric chromatin assembly [72, 73]. Further, Gonzalez et al. demonstrated that deletion of the N-terminal domain of Cnp1 results in an increase in the number of ectopic CENP-A sites, suggesting that the N-terminal domain of CENP-A prevents CENP-A assembly at ectopic loci via the ubiquitin-dependent proteolysis [18].

However, it is not yet clear by which E3 ligase the exogenous Cse4 expressed in the fission yeast S. pombe is targeted, and a budding yeast Psh1 homolog is not yet identified in fission yeast. Further study is required to elucidate how the activity of a specific E3 ligase targeting endogenous Cnp1 is regulated in fission yeast.

2.4 Heterochromatin and RNAi regulate centromeres by protecting Cnp1 from ubiquitin-mediated degradation

In most eukaryotes, the centromere is flanked and bordered by the epigenetically distinct heterochromatin domain. The establishment of centromeric heterochromatin profoundly correlates to centromere function, but the precise role of heterochromatin in centromere specification and activation is not yet clear. The transition between point centromeres (e.g., budding yeast S. cerevisiae) and regional centromeres (e.g., fission yeast S. pombe) is considered one of the most substantial centromere evolutionary events.

Yang et al. demonstrated that budding yeast Cse4 can localize to centromeres in fission yeast and partially substitute fission yeast Cnp1, however, overexpressed Cse4 localizes to heterochromatin regions [26]. Cse4 undergoes efficient ubiquitin-dependent degradation in S. pombe, and its N-terminal domain contributes to its centromere distribution via ubiquitination. Importantly, their results showed that GFP-Cse4 fails to localize at centromeres without heterochromatin and RNA interference (RNAi) using Clr4 mutant (clr4Δ) and dicer mutant (dcr1Δ), respectively. Therefore, they showed that RNAi-dependent heterochromatin is required for centromeric localization of Cse4 and protects Cse4 from ubiquitin-dependent degradation. Heterochromatin is also required for the deposition of native Cnp1 at the centromere via the same mechanism. Together, they suggest that protection of CENP-A from degradation by heterochromatin is a conserved mechanism used for centromere assembly and provided novel insights into centromere evolution from point centromere to regional centromere.

However, E3 ligase targets endogenous Cnp1 is still unclear, and its degradation mechanism through heterochromatin and RNAi machinery in fission yeast is still elusive. Further study is required to elucidate how E3 ligase activity is involved in RNAi-dependent heterochromatin formation and maintenance in fission yeast.

3. E3 ligase for fruit fly (Drosophila melanogaster) CENP-A^CID and its function

3.1 Overview of CENP-A^CID

Fruit fly (Drosophila melanogaster) centromeres extend for 200–420 kb and contain repetitive DNA that is interspersed with transposable elements (TEs) [74]. TEs are sequences that have the capacity to move other chromosomal locations and are a component of the “interspersed repeat” fraction of most genomes [75]. In fruit flies and other species (e.g., plants, wallabies, humans), the significance and function of these TEs in centromeric DNA remain to be studied. In plants, Jiang et al. suggest that the retention of active transcriptional machinery within the long terminal repeat may promote demarcation of the active centromere [76] (see also Section 5). The importance of centromeric long noncoding RNA (cenRNA) for centromere integrity has been suggested in various species [77, 78, 79]. In humans, a cenRNA is required for targeting CENP-A to the centromere [80] (see also Section 6). Arunkumar and Melters hypothesize that loading of both CENP-A and CENP-C could be one major function of centromeric transcripts, and RNA-DNA triplexes (e.g., R-loops) could be involved in loading both proteins; thereby, one may elucidate the role of RNA-DNA triplexes in both CENP-A and CENP-C loading [77].

The mechanism of heterochromatin silencing in fruit flies has been reported [81], including the position-effect variegation [82], histone modification [83], and the RNAi machinery [84]. Recently, a PIWI-interacting RNAs (piRNAs) system has been implicated in heterochromatin formation [85, 86, 87, 88], and the mechanism of heterochromatic piRNA production is being elucidated in Drosophila [89]. Ectopic CID/the Drosophila CENP-A homolog is prone to localize at euchromatin-heterochromatin boundaries, and this observation suggests that CID chromatin is likely to localize right next to a heterochromatin domain [90]. Kwenda et al. showed that RNA polymerase I transcription is required for efficient CID assembly in meiosis, as well as centromere tethering to nucleoli [91]. Recent work in mammalian and fruit fly cell lines showed that chemical inhibition of activated RNA polymerase 2 (RNAP2) resulted in the loss of centromeric CID chromatin [80, 92]; and the elongation factor Spt6 facilitates maintenance of centromeric CID [93]. These reports strongly suggest that transcription and RNA production are involved in CID incorporation.

The timing of CID incorporation occurs during metaphase/anaphase in D. melanogaster [74]. In human cells, the incorporation of newly synthesized CENP-A occurs in telophase/early G1 [94, 95]. Similar to humans, in the fast cycles of Drosophila syncytial embryos, CID incorporates in anaphase [96]. However, in Drosophila Kc cells, GFP-tagged CID was detected in metaphase cells 2 h after induction of its expression, implying that incorporation occurred at some point between the preceding G2 and metaphase [97]. In S2 cells, newly synthesized CAL1 is deposited at centromeres in prophase, preceding CID loading in metaphase [98]. Based on this observation, CAL1, like the Mis18 complex in humans, was suggested to prime the centromere before assisting in CID loading [98, 99]. In somatic tissues of Drosophilalarvae, centromeric CID deposition initiates at late telophase and continues during G1 when APC/C^Cdh1 is active [32, 100].

In Drosophila, Erhardt et al. performed an RNAi-based genome-wide search and identified CAL1 and CENP-C for CID localization determinants [101]. CID, CAL1, and CENP-C co-immunoprecipitate and are mutually dependent for centromere targeting and function. However, the molecular mechanism underlying these dependencies remains to be clarified. No homologs for CAL1 have been reported in other organisms. They also proposed that the mitotic cyclin A (CYCA) localizes at the centromere, and CYCA and RCA1/Emi1 couple centromere assembly to the cell cycle through regulation of the fizzy-related/CDH1 subunit of the APC [101], while Moreno-Moreno et al. proposed that APC/C^Cdh1 contributes to the degradation of the CAL1-CID complex [32] (see also Section 3.3). Consistent with the role of histone H4 acetylation in chickens and humans [57], Boltengagen et al. showed that the histone acetyltransferase Hat1 contributes to the CID/CENP-A assembly pathway in D. melanogaster [58] (see also previous chapter, Section 2.3.3 and this chapter, Sections 2.1 and 4.1).

Recently, there have been more reports published on the mechanism of how these three proteins (CID, CAL1, and CENP-C) work in CID incorporation. Chen et al. showed that the constitutive centromere protein CENP-C is required for recruitment of the Drosophila melanogaster (mel) CAL1 protein to existing centromeres [102]. Rosin and Mellone showed that exogenously expressed CAL1 from two different Drosophila species was efficiently recruited to D. melanogaster endogenous centromeres [103]. The CENP-C interaction with CAL1 is conserved across the Drosophila phylogeny. Whereas the coordinated evolutionary changes between CAL1 and CID prevent the recruitment of Drosophila species bipectinata (bip) CID to melanogaster centromeres, the CAL1 proteins showed no species specificity in their recruitment. The importance of the CENP-C protein for recruiting the CENP-A deposition machinery is shared in the fly and human centromere assembly pathways. However, in humans, the Mis18 complex, which is absent in Drosophila, interacts with CENP-C to recruit HJURP and CENP-A to existing centromeres [104] (see also Section 4.1).

CENP-A is maintained to mark paternal centromeres, whereas most histones are removed from mature sperm. In Drosophila males, Kwenda et al. showed that the centromere assembly factors CAL1 and CENP-C are required for meiotic chromosome segregation, CID assembly and maintenance on sperm, and fertility [91]. They showed that CID accumulates with CAL1 in nucleoli in meiosis, and CENP-C normally limits the release of CAL1 and CID from nucleoli for proper centromere assembly in meiotic prophase I. Pauleau et al. found that overexpression of CAL1 is associated with increased CID levels at centromeres and uncouples CID loading from mitosis [105]. CID levels inversely correlate with mitosis, and mitosis length is influenced by the spindle assembly checkpoint. They found that CAL1 interacts with the SAC protein and RZZ complex component Zw10 and thus constitutes the anchor for the recruitment of RZZ. Demirdizen et al. showed that the N-terminus of CID contributes to nuclear localization and protein stability [106]. While co-expression of mutant CID with RbAp48 leads to exclusive non-centromeric CID incorporation, co-expression with CAL1 leads to exclusive centromere loading of CID, suggesting that CID-associated proteins, rather than CID itself, determine its localization. Their further analysis revealed that NuRD is required for ectopic CID incorporation. The interaction of the NuRD complex with CENP-A is mediated by RbAp48 and MTA1-like (i.e., a subunit of NuRD complex), which binds specifically to the N-terminal region of CENP-A. Roure et al. showed a positive feedback loop between CID, CENP-C, and CAL1 [107], and Medina-Pritchard et al. showed that CAL1 binds both CID and CENP-C without the requirement for the Mis18 complex, using X-ray crystallography [108].

Studies of the neocentromere have also been performed in Drosophila [109, 110, 111], and the requirements, mechanism, and transmission for the neocentromere are actively under study. Two groups independently reported overexpressed CID mislocalization and ectopic incorporation into non-centromeric chromatin [112, 113]. Heun et al. demonstrated that overexpressed CID is mislocalized into normally non-centromeric regions in Drosophila tissue culture cells (S2 cells) and animals and induces severe mitotic defects [113]. These CID mis-incorporated regions display the presence of microtubule motors and binding proteins, and spindle attachments. Moreno-Moreno et al. showed that centromeric localization of transiently expressed CID is impaired in the presence of the proteasome inhibitor MG132 in Kc cells, and mislocalization of CID affects cell cycle progression with strong mitotic defects [112]. Recently, Palladino et al. used a LacO/LacI ectopic centromeric chromatin assembly system and showed that multiple genomic locations can acquire centromere activity. In addition, they demonstrated that these de novo centromeres can be transmitted and maintained epigenetically in mitotic tissues [114]. Together, their data suggest that proteolysis-mediated regulation of ectopic CenH3^CID is also present in fruit flies as in other species. Further mechanisms of CID protein degradation, including the identification of E3 ligase, are described in the following Sections 3.2 and 3.3.

3.2 CUL3/RDX E3 Ligase

In Drosophila, CENP-A^CID deposition to centromeres depends on a specialized loading factor that is called CAL1 [30]. Bade et al. showed that CAL1 directly interacts with RDX, an adaptor for CUL3/RDX-mediated ubiquitylation, through the two conserved RDX-binding sites (RBSs) of CAL1 [30] (Figure 1a; Table 1). However, CAL1 is not a substrate of the CUL3/RDX ligase but functions as an additional substrate-specifying factor for the CUL3/RDX-mediated ubiquitylation of CID. It is noteworthy that this fly CID ubiquitylation is proteasomal independent—ubiquitylation of CID by CUL3/RDX does not trigger its degradation but stabilizes CID and CAL1. Loss of RDX leads to rapid degradation of CAL1 and CID and to massive chromosome segregation defects during development (Figure 1b). Therefore, they suggest a proteolysis-independent role of ubiquitin conjugation in centromere regulation that is essential for the maintenance of the centromere-defining protein CID and its loading factor CAL1. Bade et al. proposed that this CID ubiquitylation event induces a conformational change within the CAL1/CID complex, or alternatively, increases the affinity toward centromeric chromatin, where it is protected from proteasomal degradation. The data of Bade et al. support a dual role of CAL1 in both loading and stabilizing CID protein (Figure 1a). Interestingly, their proposed “proteasomal-independent” mechanism of CUL3/RDX-mediated fly CID ubiquitylation is consistent with one of CUL4-mediated human CENP-A ubiquitylation found independently by our group [35, 115, 116, 117] (Figure 2, right; Table 1; see also Section 4.2). In humans, our group speculates that CENP-A mono- or di-ubiquitylation might sterically affect the overall conformational change, L112 residue, or C-terminal portion of the CATD on which HJURP recognition is mainly dependent (see also Section 4.2).

In humans, ectopic localization of CID depends on the H3.3 chaperone DAXX rather than the centromeric CENP-A specific chaperone HJURP [34] (Figure 2, left). This human CENP-A-containing ectopic nucleosome involves a heterotypic tetramer that contains CENP-A-H4 with H3.3-H4 [34] (Figure 2, left). Cells overexpressing human CENP-A are more tolerant of DNA damage induced by camptothecin or ionizing radiation, and both the survival advantage and CTCF occlusion by the aberrant nucleosome of heterotypic tetramer in these human cells are dependent on DAXX [34] (Figure 2, left). Although D. melanogaster has a DAXX ortholog, Daxx-like protein (DLP), the role of DLP/DAXX in CID deposition into ectopic nucleosomes through CID ubiquitylation (Figure 1e) and the CTCF occlusion by the aberrant nucleosome (Figure 1f) must be confirmed experimentally in D. melanogaster.

3.3 The E3-ligases SCF^Ppa and APC/C^Cdh1 co-operate to regulate CID expression across the cell cycle

Moreno-Moreno et al. reported that the F box protein partner of paired (Ppa), which is a variable component of an SCF E3-ubiquitin ligase complex, controls CenH3^CID stability in Drosophila [44, 118] (Figure 1b; Table 1). They showed that Ppa depletion results in increased CenH3^CID levels, and Ppa physically interacts with CenH3^CID through the CATD^CID and regulates CenH3^CID stability in Drosophila [44, 118]. Their results showed that most known SCF complexes are inactive at mitosis when newly synthesized CenH3^CID is deposited at centromeres. Therefore, they suggest that CenH3^CID deposition and proteolysis are synchronized events in Drosophila. They further reported that, in Drosophila, CID expression levels are regulated throughout the cell cycle by the combined action of SCF^Ppa and APC/C^Cdh1 [32] (Table 1). They showed that SCF^Ppa regulates CID expression in G1. Importantly, in S phase SCF^Ppa prevents the promiscuous incorporation of CID across chromatin during replication. In the G1 phase, CID expression is also controlled by APC/C^Cdh1. They also showed that CAL1, the specific chaperone that deposits CENP-A^CID at centromeres, protects CID from SCF^Ppa-mediated degradation but not from APC/C^Cdh1-mediated degradation. Together, they suggest that, whereas SCF^Ppa targets the fraction of CID that is not in complex with CAL1 (Figure 1c; Table 1), APC/C^Cdh1 contributes to the degradation of the CAL1-CID complex and, thus, likely regulates centromeric CID deposition (Figure 1d; Table 1).

3.4 Phosphorylation of Drosophila CID on serine 20 regulates protein turnover and centromere-specific loading

Huang et al. showed that CID is phosphorylated at serine 20 (S20) by casein kinase II (CK2) and that the phosphorylated form is enriched on chromatin during mitosis [33] (Figure 1c and g; Table 1). Their results revealed that S20 phosphorylation regulates the turnover of prenucleosomal CID through the SCF^Ppa-proteasome pathway (Figure 1c; Table 1) and that phosphorylation facilitates removal of CID from ectopic but not from centromeric sites in chromatin (Figure 1g and h; Table 1). They provided multiple lines of evidence for an essential role of S20 phosphorylation in regulating restricted incorporation of CID into centromeric chromatin, suggesting that modulation of the phosphorylation state of S20 may lead to fine-tuned control of CID levels to prevent malignant incorporation into non-centromeric chromatin.

On the other hand, factors/components that stabilize ectopically incorporated CID and are required for neocentromere formation and its maintenance are not clear in D. melanogaster (Figure 1h). The status of overall post-translational modifications, including polyubiquitylation of CID, especially in ectopic nucleosomes, remains to be elucidated.

4. E3 ligase for human CENP-A its function

4.1 Overview of human CENP-A

In most eukaryotes, including humans, the centromere has no defined DNA sequence but is associated with large arrays of repetitive DNA; in humans, this sequence is a 171-bp alpha-satellite DNA, although several other sequence types are found in this region. CENP-A-containing nucleosomes are formed with canonical histones H2A, H2B, and H4 at the active centromeres [5]. CENP-A nucleosomes localize to the inner plate of mammalian kinetochores [119] and bind to the 171-bp alpha-satellite DNA. Recently, the importance of centromeric cis-element, transcription, and centromeric long noncoding RNA (cenRNA) for centromere integrity has been suggested in various species, including humans [77, 78, 79] (see also Sections 3 and 5). Interestingly, when the CENP-B box DNA sequence is located proximal to the CENP-A nucleosome, CENP-B forms a more stable complex with the CENP-A nucleosome through specific interactions with CENP-A [120]. In humans, a centromeric long noncoding RNA (cenRNA) is required for targeting CENP-A to the centromere [80].

Currently, it is commonly reported that CENP-A-containing nucleosomes are formed with canonical histones H2A, H2B, and H4 at the active centromeres, however, their structure remains controversial among different research groups [5]. Bui et al. suggest that CENP-A nucleosomes alter from tetramers to octamers before replication and revert to tetramers after replication, using combinatory methods, including atomic force microscopy [38]. It is noteworthy that reversible chaperone binding, chromatin fiber folding changes, and CENP-A K124 acetylation (K124ac) and H4 K79 acetylation (K79ac) are concurrent with these structural transitions. Further computational modeling suggests that acetylation of K124 causes tightening of the histone core and hampers accessibility to its C-terminus, which in turn reduces CENP-C interaction [39] (see also the following paragraph about the function of histone H4 acetylation). Further study, including the solution of real-time post-translational modifications or the 3D structure of free Cse4 complexes, is required to determine how different chaperons recognize Cse4/CENP-A-H4 for incorporation into different locations of chromatin.

CENP-A contains a short centromere targeting domain (CATD) within the histone fold region [2]. Replacement of the corresponding region of H3 with the CATD is sufficient to direct H3 to the centromere [2], and this chimeric histone can rescue the viability of CENP-A-depleted cells [2, 12]. On the other hand, Logsdon et al. found contributions from small portions of the N-terminal tail and the CATD in the initial recruitment of CENP-C and CENP-T, using a LacO/LacI ectopic centromeric chromatin assembly system [20]. Jing et al. reported that deletion of the first 53 but not the first 29 residues of CENP-A from the N-terminus, resulted in its cytoplasmic localization [121]. They identified two motifs for CENP-A nuclear accumulation and one motif involved in the centromeric accumulation of CENP-A, as well as the interaction of CENP-A with core histone H4 and CENP-B.

Early studies in human cells showed that CENP-A mRNA and protein start to accumulate in the mid-S phase and peak in G2 [122, 123], however, further cell type-specific regulation of human CENP-A mRNA and protein remains to be studied.

In human cells, the incorporation of newly synthesized CENP-A occurs in telophase/early G1 [94, 95]. The incorporation of newly synthesized CENP-A into centromeric nucleosomes depends on the HJURP, which is a CENP-A-specific chromatin assembly factor [41, 42, 43]. Like CENP-A, HJURP is also assembled during early G1 to centromeres [42, 43, 94, 96]. The primary structural analysis demonstrated that human HJURP is a distant counterpart of Scm3, which is required to deposit centromeric nucleosomes in yeast [29]. CENP-A interacts with HJURP as a soluble pre-nucleosomal complex, and the unique structural dynamics of HJURP together with CENP-A/H4 heterodimer/tetramer (pre-nucleosomal CENP-A-H4-HJURP complex) have been reported [3, 124, 125, 126, 127, 128, 129, 130, 131, 132]. HJURP recruitment to centromeres depends on the activity of the Mis18 complex [41, 104], which affects the histone modification and DNA methylation status of centromeres [54, 59]. The human proteins hMis18 and M18BP1/KNL2 are recruited to the centromere at telophase/G1, suggesting that the hMis18 complex and RbAp46/48 (homologs of Mis16) prime the centromere for CENP-A localization [54, 133]. Moreover, acetylation of histone H4 lysine 5 and 12 (H4K5ac and H4K12ac) within pre-nucleosomal CENP-A-H4-HJURP complex mediated by the RbAp46/48-Hat1 complex is required for CENP-A deposition into centromeres in chickens and humans [57], consistent with the role of Hat1 shown in D. melanogaster [58] (see also Section 3.1). In mouse studies, Mis18α interacts with DNMT3A/3B, and this interaction is required to maintain DNA methylation [59]. Mis18α deficiency leads to not only the reduction of DNA methylation, but altered histone H3 modifications, and uncontrolled noncoding transcripts in the centromere region. Faithful CENP-A deposition requires integrated signals from Plk1 and cyclin-dependent kinase (CDK), with Plk1 promoting the localization of the Mis18 complex, and CDK inhibiting Mis18 complex assembly [134]. Moreover, the remodeling and spacing factor complex is required for the assembly of CENP-A chromatin [135], and the CENP-A licensing factor M18BP1/KNL2 and the small GTPases-activating protein MgcRacGAP cooperate to maintain the stability of newly loaded CENP-A at centromeres [136, 137].

Currently, the proteolysis mechanism for mis-incorporated human CENP-A and its E3 ligase is not yet clear (Figure 2d), and there are no reports to date on proteasome-mediated degradation of human CENP-A [138]. We reported that mono- or di-ubiquitylation of CENP-A K124 is required for CENP-A deposition at the centromere [35] (Figure 2, right). However, the stability of endogenous CENP-A is not affected by CUL4A or RBX1 depletion, and the stability of exogenous CENP-A K124R is the same as in wild-type cells. Rather, overexpression of a monoubiquitin-fused CENP-A mutant induces neocentromere formation, suggesting that signaling CENP-A mono- or di-ubiquitylation determines centromere location and activity [115] (see also Sections 4.2 and 4.3). Future studies are required to reveal how ectopic CENP-A is degraded and removed from the non-centromeric chromosome, and/or how the neocentromere established through CENP-A ubiquitylation is deactivated in humans (Figure 2c and d). This proteolysis could be initiated on chromatin and the machinery involved could be specifically excluded from centromeric regions. Alternatively, mis-incorporated CENP-A nucleosomes may dissociate more easily than those properly localized and be subsequently degraded in the nucleoplasm [139]. Obuse et al. performed chromatin immunoprecipitation with an anti-CENP-A monoclonal antibody using HeLa interphase nuclei and systematic identification of its interactors by mass spectrometric analyses [140]. They identified UV-damaged DNA binding protein 1 (DDB1) as a component of the CEN complex and BMI-1 that is transiently co-localized with the centromeric region in interphase.

RbAp46 forms a complex with the CRL4 ubiquitin ligase and DDB1 protein (where DDB1 mediates the association of CUL4 with its substrate-specific receptor—RbAP46) [141, 142]. RbAp46 is required for stabilizing CENP-A protein levels and the CRL4-RbAp46 complex activity promotes efficient new CENP-A deposition in humans [142]. This is in contrast to studies in yeast and fruit flies, where the association of CENP-A with the SCF E3-ubiquitin ligase complex leads to CENP-A degradation. However, our group showed that CUL4A-RBX1-COPS8 E3 ligase activity is required for CENP-A mono- or di-ubiquitylation on lysine 124 (K124) and CENP-A centromere localization, although our results suggest that DDB1 is not required for CENP-A recruitment to centromeres [35] (Figure 2, right; see also Sections 4.2–4.5). In humans, soluble CENP-A is associated with the centromeric CENP-A specific chaperone HJURP (see also Introduction). Depletion of HJURP leads to a significant decrease in CENP-A levels, suggesting that HJURP protects the fraction of CENP-A that will be incorporated at the centromere in G1 while remaining “free” CENP-A will be degraded to prevent its incorporation into non-centromeric chromatin [42, 43]. Our results also support this model, because CENP-A ubiquitylation enhances the affinity between HJURP with ubiquitylated CENP-A [35] (see also Sections 4.2–4.5).

One question is also generated about the function of H3.3 histone chaperone proteins, HIRA and DAXX, which were previously reported to promote ectopic CENP-A deposition in human cancer cells [34, 143]. Lacoste et al. found that CENP-A overexpression in human cells leads to ectopic enrichment at sites of active histone turnover involving a heterotypic tetramer that contains CENP-A-H4 with H3.3-H4 [34] (Figure 2, left). Ectopic localization of this particle (aberrant nucleosome) depends on the H3.3 chaperone DAXX rather than the centromeric CENP-A specific chaperone HJURP (Figure 2, left). Cells overexpressing CENP-A are more tolerant of DNA damage induced by camptothecin or ionizing radiation, and both the survival advantage and CTCF occlusion by the aberrant nucleosome of heterotypic tetramer in these cells are dependent on DAXX (Figure 2, left). However, post-translational modifications of human CENP-A, especially before recognition by DAXX and after incorporation into the ectopic nucleosome, must be elucidated (Figure 2a), and specific DAXX localization on these CTCF sites under CENP-A overexpression has to be confirmed experimentally (Figure 2b).

Shrestha et al. showed that mislocalization of CENP-A to chromosome arms is one of the major contributors to CIN, as depletion of histone chaperone DAXX prevents CENP-A mislocalization and rescues the reduced interkinetochore distance and CIN phenotype in CENP-A-overexpressing cells [144]. Nye et al. reported that in human colon cancer cells, the H3.3 chaperones HIRA and DAXX promote ectopic CENP-A incorporation [143]. They found that a correct balance between levels of the centromeric chaperone HJURP and CENP-A is required to prevent ectopic assembly by H3.3 chaperones. Their results also suggest that CENP-A occupancy at the 8q24 locus is significantly correlated with amplification and overexpression of the MYC gene within that locus. Together, CENP-A mislocalization into non-centromeric regions resulting from its overexpression leads to chromosomal segregation aberrations and genome instability [145]. Overexpression of CENP-A is a feature of many cancers and is likely associated with malignant progression and poor outcomes [146, 147, 148]. CENP-A overexpression is often accompanied by overexpression of its chaperone HJURP, leading to “epigenetic addiction” in which increased levels of HJURP and CENP-A become necessary to support rapidly dividing p53-deficient cancer cells [149]. In addition, the functional roles of DAXX and HIRA in the development of cancer and other diseases have been described [150, 151, 152, 153]. Elucidation of the proper mechanism of H3.3 incorporation into chromatin through DAXX and HIRA may also lead to proper CENP-A incorporation at centromeres as well as an effective disease (e.g., cancer) therapy.

Recently, the importance of the site-specific posttranslational modifications of human CENP-A and their biological functions has been reported [44, 45]. The functional roles of phosphorylation at CENP-A-Ser68 are still under active investigation [124, 125, 154, 155, 156]. How the defects of CENP-A PTMs and the dysfunction of centromere contribute to the generation and the development of cancer is an unsolved question. Takada et al. demonstrated that CENP-A Ser18 hyperphosphorylation by cyclin E1/CDK2 occurred upon loss of FBW7, a tumor suppressor whose inactivation leads to CIN [157]. This CENP-A Ser18 hyperphosphorylation reduced the CENP-A centromeric localization, increased CIN, and promoted anchorage-independent growth and xenograft tumor formation. Defects of CENP-A PTMs are significantly associated with chromosome segregation errors and CIN [149].

4.2 CENP-A K124 ubiquitylation is required for CENP-A deposition at the centromere

In budding yeast, Scm3 and Psh1 might compete for binding to Cse4. Cse4 that is not associated with Scm3 may be targeted by Psh1 for proteolysis, but Cse4 in a complex with Scm3 may be protected [71] (see also previous chapter, Section 2.1). On the other hand, in D. melanogaster it was proposed that CENP-A^CID ubiquitylation induces a conformational change within the CAL1/CENP-A complex, or alternatively, increases the affinity toward centromeric chromatin, where it is protected from proteasomal degradation [30] (see also Section 3.2).

In humans, our group found that CUL4A-RBX1-COPS8 E3 ligase activity is required for CENP-A mono- or di-ubiquitylation on lysine 124 (K124) and CENP-A centromere localization [35] (Figure 2, right). CUL4A complex targets CENP-A through the adaptor COPS8/CSN8 that has WD40 motifs in non-canonical CRL4 machinery (Figure 2, right). A mutation of CENP-A, K124R, reduces interaction with HJURP and abrogates localization of CENP-A to the centromere. The addition of monoubiquitin is sufficient to restore CENP-A K124R to centromeres and the interaction with HJURP, indicating that “signaling” ubiquitylation is required for CENP-A loading at centromeres (Figure 2, right).

However, one question remains—how does such mono- or di-ubiquitylation of CENP-A facilitate the interaction of CENP-A with HJURP? The CENP-A K124 site and its proximal residues might not directly affect CENP-A-HJURP interaction in the crystal structure of the HJURP-CENP-A-histone H4 complex, since we did not detect defects in CENP-A dimerization of K124R mutant (Figure 3; see also Section 4.3) or any ubiquitin interacting motif in HJURP. Therefore, we speculate that CENP-A mono- or di-ubiquitylation might sterically affect the overall conformational change, L112 residue (the closest CENP-A’s residue to K124 out of the seven residues reported to be important for appropriate interaction with HJURP), or C-terminal portion of the CATD on which HJURP recognition is mainly dependent. In addition, acetylated lysine 124 (K124) was previously reported by Bui et al. [38], but the functional role of K124 acetylation and its relationship with K124 ubiquitylation remains to be studied (Figure 2, right). Moreover, currently, the proteolysis mechanism for mis-incorporated human CENP-A and its E3 ligase is not clear, and there are no reports to date regarding proteasome-mediated degradation of human CENP-A [138] (Figure 2d). Future studies are required to reveal how ectopic CENP-A is degraded and removed from the non-centromeric chromosome (Figure 2c and d).

4.3 CENP-A ubiquitylation is inherited through dimerization between cell divisions

The mechanism by which centromere inheritance occurs is largely unknown. Gassmann et al. suggested that in Caenorhabditis elegans, pre-existing CENP-A^HCP−3 nucleosomes are not necessary to guide the recruitment of new CENP-A nucleosomes [158]. In contrast, in Drosophila melanogaster, CENP-A^CID is present in mature sperm, and the amount of CID that is loaded during each cell cycle appears to be determined primarily by the pre-existing centromeric CID, a finding that is consistent with a “template-governed” mechanism [159]. However, in humans, it is unclear how CENP-A works as the epigenetic mark at the molecular level.

Our group showed that pre-existing ubiquitylated CENP-A is necessary for the recruitment of newly synthesized CENP-A to the centromere and that CENP-A ubiquitylation is inherited between cell divisions (Figure 3). In vivo and in vitro analyses using dimerization mutants and dimerization domain fusion mutants revealed that the inheritance of CENP-A ubiquitylation requires CENP-A dimerization. Therefore, we propose models in which CENP-A ubiquitylation is inherited and centromere location is determined through dimerization (Figure 3).

Numerous studies have found that CENP-A can be experimentally mistargeted to non-centromeric regions of chromatin and that this mistargeting leads to the formation of ectopic centromeres in model organisms [160]. Chromosome engineering has allowed the efficient isolation of neocentromeres on a wide range of both transcriptionally active and inactive sequences in chicken DT40 cells [57]. More than 100 neocentromeres in human clinical samples have been described [161]. They form on diverse DNA sequences and are associated with CENP-A localization, but not with alpha-satellite arrays; thus, these findings provide strong evidence that human centromeres result from sequence-independent epigenetic mechanisms. However, neocentromeres have not yet been created experimentally in humans; overexpression of CENP-A induces mislocalization of CENP-A, but not the formation of functional neocentromeres [162].

Our group demonstrated that overexpression of a monoubiquitin-fused CENP-A mutant induces neocentromeres at non-centromeric regions of chromosomes, and this result further supports our model in which CENP-A ubiquitylation is inherited and determines centromere location through dimerization (Figure 3). Our assay using the LacO/LacI ectopic centromeric chromatin assembly system clearly revealed that CENP-A ubiquitylation contributes to the recruitment of CENP-A chaperones (HJURP and DAXX) and outer kinetochore components (HEC1 and SKA1). It is possible that ubiquitylation of CENP-A contributes to maintain and stabilize ectopic neocentromeres in humans (Figure 2c).

However, it remains unclear how the neocentromere established through CENP-A ubiquitylation is deactivated. Future studies are required to reveal the mechanism of site-specific (centromeric and/or non-centromeric) deubiquitylation CENP-A and subsequent proteolysis in humans (Figure 2c and d). In this context, it would be interesting to test if the Ubp8-driven deubiquitylation mechanism in budding yeast [163] (see also previous chapter, Section 2.7) is conserved in humans.

4.4 SGT1-HSP90 complex is required for CENP-A loading at centromeres

The mechanism that controls the E3 ligase activity of the CUL4A-RBX1-COPS8 complex remains obscure. Our group found that the SGT1-HSP90 complex is required for recognition of CENP-A by COPS8 [164] (Figure 2, right). SGT1/SUGT1, a co-chaperone of HSP90, is involved in multiple cellular activities, including cullin E3 ubiquitin ligase activity [165]. The SGT1 gene was originally isolated as a dosage suppressor of the skp1–4 mutant in yeast S. cerevisiae, which causes defects in yeast kinetochore function, but also as a novel subunit of the Skp1-Cullin-F-box (SCF) ubiquitin ligase complex [166]. In both yeast and humans, the interaction between SGT1 and heat shock protein 90 (HSP90) is also required for kinetochore assembly [167, 168, 169]. In humans, cancer cells utilize Hsp90 as a chaperone to promote the folding and function of mutated or overexpressed oncoproteins, because aberrant oncoproteins are unstable [170, 171]. SGT1 contributes to cancer development by stabilizing oncoproteins, and the SGT1-HSP90 complex is a potential target for therapies aimed at cancer, brain, and heart disease [165].

Our group initially applied RNA interference (RNAi)-mediated SGT1 and/or HSP90 depletion in HeLa cells and found that the SGT1-HSP90 complex is required for CENP-A ubiquitylation in vivo and CENP-A deposition at centromeres [164] (Figure 2, right). Moreover, our group and others demonstrated in vivo interactions of SGT1A with CUL4A [164] and HSP90 with CUL4 [172], respectively (Figure 2, right). Previously, we had also reported that the CUL4A complex targets CENP-A through the adaptor COPS8/CSN8 that has WD40 motifs in non-canonical CRL4 machinery [35] (Figure 2, right; see also Section 4.2). Therefore, we hypothesized that depletion of SGT1 or HSP90 protein promotes loss of interaction among components of the CUL4A complex. Indeed, SGT1 or HSP90 siRNA disrupted interactions of COPS8 with CENP-A and CUL4A. These results suggest that the SGT1-HSP90 complex is required for the composition of the CUL4A complex as well as recognition of CENP-A by COPS8 (Figure 2, right). Thus, we clarified how the SGT1-HSP90 complex contributes to the E3 ligase activity of the CUL4A complex in CENP-A ubiquitylation (Figure 2, right).

In our study, SKP1 siRNA treatment did not lead to any signal reduction of CENP-A at centromeres [164]. Therefore, we proposed that the SGT1-HSP90 complex is involved in CENP-A deposition at centromeres in an SKP1-independent and/or SCF-independent manner. This conclusion is consistent with our previous report that the CUL4A-RBX1 complex, which does not require SKP1 to function, contributes to CENP-A deposition at centromeres [35]. Because our results suggest that SKP1 is not required for the recruitment of CENP-A to centromeres, it is unlikely that SKP1 activity affects the CENP-A loading pathway. Because CENP-A is at the top of a hierarchy of the pathway that determines the assembly of kinetochore components [6], destabilization of the MIS12 complex at the kinetochore was observed by Davies et al. [173] could be partially due to the defect in CENP-A recruitment. This idea is supported by our results demonstrating that SGT1 siRNA treatment did not significantly change the recruitment of endogenous MIS12, HEC1, and SKA1 proteins in LacO arrays after ectopic loci were determined through LacO-LacI-CENP-A interactions. Collectively, these data suggest that the losses of immunofluorescence signals of the central-outer kinetochore proteins at the kinetochore caused by SGT1 siRNA defects, including ones reported previously [174], are explained by CENP-A mislocalization caused by SGT1 siRNA defects.

4.5 CENP-A ubiquitylation is indispensable to cell viability

Our group reported that CENP-A K124 ubiquitylation, mediated by the CUL4A-RBX1-COPS8 complex, is essential for CENP-A deposition at the centromere [35] (Figure 2, right; see also Section 4.2). On the other hand, Fachinetti et al. reported that CENP-A K124R mutants show no defects in centromere localization and cell viability [156]. However, there are substantive problems with their experiments that yielded these results. We reported our response describing potential issues with the results and their conclusions [117]. A major caveat is that they used a fusion protein much larger molecular size than CENP-A. In their RPE-1 CENP-A^−/F knockout system, the enhanced yellow fluorescent protein (EYFP) is approximately 30 kDa, and endogenous CENP-A is about 16 kDa. Fachinetti et al. also used SNAP-tags, and they found that SNAP-CENP-A K124R showed no defects in centromere deposition. Because the SNAP-tag (20 kDa) is also a larger tag than CENP-A (approximately 16 kDa) and has 10 lysines, SNAP-CENP-A K124R, presumably, is ubiquitylated at a site different than K124. One possibility is that the tagging of a large protein may endogenously lead to ubiquitylation at an amino acid other than K124 in the CENP-A K124R mutant protein, and this ubiquitylation at another site could suppress the mutant phenotype as a compensatory mechanism. Therefore, our group hypothesized that the presence of a large fusion protein promotes ubiquitylation at a different lysine in the CENP-A K124R mutant protein.

Indeed, our group found that EYFP tagging induces additional ubiquitylation of EYFP-CENP-A K124R, which allows the mutant protein to bind to HJURP [116]. Our immunoprecipitation mass spectrometry analysis showed that lysine 306 (K306) in the EYFP-CENP-A K124R mutant is ubiquitylated in vivo. This site corresponds to lysine 56 (K56) in CENP-A. These data suggest that once EYFP is tagged to a K124R mutant, another ubiquitylation occurs at a different site than K124 as endogenous compensatory machinery. Using a previously developed conditional CENP-A knockout system and our CENP-A K124R knockin mutant created by the CRISPR-Cas9 system, we show that the small size Flag-tagged or untagged CENP-A K124R mutant is lethal. This lethality is rescued by monoubiquitin fusion, indicating that CENP-A ubiquitylation is essential for viability. Therefore, our group suggests a caveat in the use of GFP/EYFP as a tool to analyze the function of a protein, and our data still support that the CENP-A ubiquitylation is indispensable to cell viability.

4.6 Hypothetical regulation of human CENP-A through sumoylation

In budding yeast, CENP-A^Cse4 is sumoylated on its N-terminal tail by Siz1/Siz2 SUMO E3 ligases [22] (previous chapter, Figure 1a and b) (see also previous chapter, Section 2.4.1). Cse4 is poly-sumoylated at K65 in its N-terminal domain, which recruits the yeast SUMO-targeted ubiquitin ligase (STUbl) Slx5, leading to the polyubiquitination of poly-sumoylated Cse4 and its subsequent degradation [21]. Cse4 K215/216 sumoylation in C-terminus also controls its interaction with the histone chaperones Scm3 and CAF-1, facilitating the deposition of overexpressed Cse4 into CEN and non-CEN regions, respectively [175] (previous chapter, Figure 1) (see also previous chapter, Section 2.4.2).

In humans, depletion of the human Slx5 homolog ring finger protein 4 (RNF4) contributes to SUMOylation-dependent degradation of the CCAN protein CENP-I, while SENP6 stabilizes CENP-I by antagonizing RNF4 [176]. SENP6 is a desumoylation enzyme as well as a member of a large family of Sentrin-specific protease enzymes (SENP1–7) [138, 177]. In budding yeast, two SUMO proteases are known, ubiquitin-like protease 1 and 2 (Ulp1 and 2); in mammalian cells, these have diverged into the SENP family. SENP1–5 is evolutionarily conserved to Ulp1, while the more divergent SENP6 and SENP7 belong to the Ulp2 group. Depletion of SENP6 in HeLa cells leads to the loss of the CENP-H/I/K complex from the centromeres, but not an apparent reduction in centromeric CENP-A/B/C levels recognized by CREST sera [176].

Liebelt et al. identified a protein group de-modification by SENP6, including the constitutive centromere-associated network (CCAN), the CENP-A loading factors Mis18BP1 and Mis18A, and DNA damage response factors [178]. SENP6-deficient cells are severely compromised for proliferation, accumulate in the G2/M phases, and frequently form micronuclei. Centromeric assembly of CENP-T, CENP-W, and CENP-A is impaired in the absence of SENP6. However, the increase of SUMO chains is not required for ubiquitin-dependent proteasomal degradation of the CCAN subunits. Therefore, their results indicated that SUMO polymers can act in a proteolysis-independent manner and consequently, have a more diverse signaling function than previously expected. On the other hand, Mitra et al. identified the SUMO-protease SENP6 as a key factor, not only controlling CENP-A stability but virtually the entire centromere and kinetochore using a genetic screen coupled to pulse-chase labeling [179]. Loss of SENP6 results in hyper-sumoylation of CENP-C and CENP-I, but not CENP-A itself. SENP6 activity is required throughout the cell cycle, suggesting that a dynamic SUMO cycle underlies continuous surveillance of the centromere complex that in turn ensures stable transmission of CENP-A chromatin. Mitra et al. and other groups did not detect sumoylation of CENP-A, suggesting that CENP-A is not a direct substrate of SENP6 [138, 179]. However, the effect of SENP6 depletion on CENP-A stability is much greater than observed on depletion of CENP-C or -B alone [179]. This result suggests that there may be other components required for the SENP6-mediated stabilization of centromeric chromatin [138].

5. E3 ligases for plant CENP-A (CENH3) its function

5.1 Overview of plant CENP-A (CENH3)

Studies of E3 ligases at plant centromeres-kinetochores are not as advanced as those in model animal species. The structure and organization of plant centromeric DNA have been described, and satellite repeats associated with centromeres have been reported in many plant species [76]. Plant centromeres also have mega-base-sized arrays of tandem repetitive DNA sequences, as in centromeres of humans and other mammals, and transposable elements are abundant in centromeric and paracentromeric regions [76, 180]. In early studies, Jiang et al. suggest that the retention of active transcriptional machinery within the long terminal repeat may promote demarcation of an active centromere [76]. A Ty3/gypsy class of centromere-specific retrotransposons, the centromeric retrotransposon (CR) family, was discovered in the grass species. Highly conserved motifs were found in the long terminal repeat of the CR elements from rice, maize, and barley [181]. The CR elements are highly enriched in chromatin domains associated with CENH3/CENP-A, the centromere-specific histone H3 variant. CR elements as well as their flanking centromeric satellite DNA are actively transcribed in maize. These data suggest that the deposition of centromeric histones might be a transcription-coupled event. The importance of centromeric transcription and centromeric long noncoding RNA (cenRNA) for centromere integrity has been suggested in various species, including plants [77, 78, 79] (see also Sections 3 and 4). Moreover, in maize, CENP-C binding to centromeric DNA is associated with small RNA [182], whereas in humans CENP-A loading is linked to lncRNAs [80]. It is not yet known whether the same transcript can recruit and stabilize both CENP-A and CENP-C at centromeric chromatin [77].

Plant CENH3/CENP-A and other centromere-kinetochore proteins have been reported showing high conservation among species. On the other hand, DNA sequences of plant centromeres, of which loci are determined epigenetically by centromeric histone 3 (CENH3), have revealed high structural diversity, ranging from the canonical monocentric form seen in vertebrates, to polycentric and holocentric forms [183, 184]. Plant centromeres can change position over evolutionary time or upon genomic stress, such as in McClintock’s genome shock [185] or physically damaged or broken chromosomes [183]. Jiang et al. suggested that the centromeric state is reinforced and maintained by the tension applied during spindle attachment [76]. The chromatin damaged by such mechano-force could then be marked for repair by the replication-independent mechanism similar to the one originally incorporated in CENH3. Indeed, human centromere-kinetochore proteins, including CENP-A, are involved in DNA damage/repair [186], and the incorporation of newly synthesized CENP-A occurs “right after mitosis” (i.e., telophase/early G1) [94, 95]. However, the model of CenH3 (CENP-A) incorporation upon mechano-force-induced DNA damage/repair is not yet experimentally demonstrated, and its precise mechanism needs to be elucidated. Meanwhile, there is evidence of divergent evolution originating in CenH3 in plants [187, 188] and Drosophila [189]. The CenH3 (CENP-A) has apparently undergone both convergent and divergent evolution [7]. Nagaki et el. and others described that the centromere DNA repeats with which CENH3-containing nucleosome interacts are also highly diverged, proposing an “arms race” hypothesis where centromere DNA repeats are changing and expanding to increase their segregation properties, while CENH3 is changing to curb this process and keep segregation frequencies equal to avoid fixing traits [180, 184, 189, 190].

Plant studies of dicentric centromeres and neocentromeres have been described along with those of other eukaryotes [180, 183]. The active state of one of the two centromeres on the wheat dicentric chromosome can be epigenetically silenced [180], as in the human dicentric chromosome [191]. Neocentromeres have been described extensively in human and fruit fly chromosomes as well as in some plant species, such as barley, maize, and rice [114, 184, 192]. In D. melanogaster, Palladino et al. showed that multiple genomic locations can acquire centromere activity, using a LacO/LacI ectopic centromeric chromatin assembly system. In addition, they demonstrated that these de novo centromeres can be transmitted and maintained epigenetically in mitotic tissues [114]. Although studies of human neocentromeres have indicated that they are generated at new positions in a single step; the barley neocentromere appears to have shifted several times along the chromosomal arm region during the deletion steps to finally reach the observed position [180]. The emergence of new centromeres was also observed in hybrid conditions [183, 184], and Wang et al. described the proposed model for hybrids between maize and oat [193]. The “centromere repositioning” then generates neocentromeres; the establishment of a new centromere does not require specific DNA composition in the target loci [76, 194]. Most new centromeres have no satellite DNA [195]. However, most mature centromeres are overwhelmingly composed of repetitive DNA, especially satellite DNA [76, 194]. One hypothesis to explain this apparent contradiction was described by Oliveira et al. as “satellite DNA invasion mechanism”—a new satellite repeat or one already present in other centromeres may invade and occupy the CenH3 domain of the new centromere [184]. The satellite DNA invasion mechanism is still elusive, and the retrotransposons would be the main source for the origin of new repeats [184, 196].

Plant studies of minichromosomes and artificial chromosomes also have been reported, as in other eukaryotes [180, 183]. The main issues of these studies are what are the size and factors required for the maintenance and stability of such special chromosomes during cell division. Harrington et al. constructed human artificial minichromosomes [197], and Ananiev et al. artificially generated minichromosomes in maize by introducing the DNA molecule containing native centromere segment, ori, and telomere repeats [198]. These studies suggested that repetitive DNA may play an important but unknown role in centromere function. The repetitive centromeric DNA may be still important, although it is not essential for centromeric function, since plant centromeric DNA does not generate functional centromeres when reintroduced into plant cells [199] and new centromeres are functional even if located in loci with non-centromeric DNA [161].

In terms of the plant CENH3 recruitment mechanism to centromeres, most CENP-A is loaded in G2 by a replication-independent mechanism in Arabidopsis thaliana [200]. However, in plants as in other species, the timing of deposition of newly synthesized CENP-A within the cell cycle may be variable—not only among different plant species but also different developmental stages within the same species. Le Goff et al. reported that the H3 histone chaperone NASP^SIM3 escorts CENH3 in Arabidopsis [19]. They showed that the Arabidopsis ortholog of the mammalian nuclear autoantigenic sperm protein (NASP) and S. pombe histone chaperone Sim3 is a soluble nuclear protein that interacts with CENH3 and influences its abundance at the centromeres [19]. NASP^SIM3 is co-expressed with Arabidopsis CENH3 in dividing cells and binds directly to both the N-terminal tail and the C-terminal histone fold domain of non-nucleosomal CENH3. Reduced NASP^SIM3 expression by NASP^SIM3 knockdown impairs CenH3 deposition. Thus, they identified NASP^SIM3 as a CenH3 histone chaperone as demonstrated in fission yeast (see also Section 2.1).

5.2 Engineered degradation of EYFP-tagged CENH3 in plants

Currently, an endogenous E3 ligase for plant CENP-A (CENH3) is not yet identified. Sorge et al. developed a synthetic biology approach to degrade plant CENP-A using E3-ligase adapter protein SPOP (Speckle-type POZ adapter protein) with a specific anti-GFP nanobody (VHHGFP4) [201] (Table 1). To determine the function of proteins, CRISPR/Cas9-based methods and antisense/RNAi strategies are commonly used to remove the selected protein from all organs in a cell- and tissue-specific manner. However, CRISPR/Cas9 and antisense/RNAi strategies are still error-prone and can show off-target effects [202]. Classical genetic strategies to knock out/down protein function in plants still have problems, such as the time-consuming process of generating homozygous transgenic lines or the risk of lethal phenotypes at early developmental stages.

Sorge et al. attempted to solve these problems by utilizing the synthetic E3 ligase activity in protein ubiquitylation and degradation pathway. They successfully recruited the 26S proteasome pathway to directly degrade CENP-A of A. thaliana via replacement of the interaction domain of the E3 ligase adaptor protein SPOP (Speckle-type POZ adapter protein) with a specific anti-GFP nanobody (VHHGFP4). They proved that the target protein CENH3 of A. thaliana fused to EYFP was subjected to nanobody-guided proteasomal degradation in planta. Thus, their results show the potential of the modified E3-ligase adapter protein VHHGFP4-SPOP to induce nucleus-specific protein degradation in plants. However, further studies are required to identify endogenous E3 ligase for plant CENP-A (CENH3) and determine the function of the plant CENP-A (CENH3) proteolysis or deposition at centromeres.

6. Conclusions

Each species reviewed in our articles, including the previous chapter has advantages and disadvantages for research. For example, the centromere sequence size of the budding yeast is small and the sequences can be easily mutated to identify the important functional regions [203]. Techniques such as ChIP are also possible, which cannot be easily performed on highly repetitive centromeres in other organisms. Moreover, the centromere can be shifted to other genomic regions, allowing the construction of artificial chromosomes and plasmids as well as tools, such as conditional centromeres. Fission yeast and fruit fly models have progressed more than others in studies of heterochromatin regulation and gene silencing. Plant models have advanced more in evolutionary studies of centromeric DNA structures, including CR family comparisons among different plant species.

On the other hand, in fission yeast and plant species, the E3 ligase of CENP-A (CenH3) and its specific regulation and/or function are not yet identified. The E3 ligase of CENP-A is unknown in multiple species (e.g., Caenorhabditis elegans, Xenopus laevis, zebrafish Danio rerio, chicken Gallus domesticus DT40 cells, Mus musculus, etc.) or this research is sparse in these species compared with others. At present, the most common species studied and reported in the past for E3 ligase of CenH3 (Cse4) is budding yeast. However, in other species, much is not understood, particularly about control of the balance between E3 ligase and deubiquitylase and the balance among SUMO E3 ligase, the desumoylation enzyme, SUMO proteases (e.g., SENP6). Future research into the E3 ligases of CENP-A will elucidate the regulation and mechanisms of these subtle balances in each species and human diseases.

Studying the mechanisms of formation and maintenance of neocentromeres will deepen our understanding of the centromere-kinetochore formation and promote the building and establishment of artificial chromosomes. Such studies will lead to the construction of artificial cells and tissues that can be controlled by DNA levels through chromosome dynamics. As a result, the function of E3 ligase can be artificially adjusted, which will increase the effectiveness of future gene therapies. Minichromosomes generated to date suggest that the repetitive centromeric DNA may be still important, although perhaps, it is not essential for centromeric function. In addition, it is unclear whether there is causality or feedback between cenRNA transcription and overall transcriptional change after chromosome missegregation and CIN. As of now, we have little understanding of the effects of these cenRNAs on the E3 ligase of CENP-A, including how these transcriptional changes and regulation are related to the function of E3 ligase.

Although our group showed that ubiquitylation occurs at a different site than CENP-A K124 as endogenous compensatory machinery, the compensatory machinery of post-translational modifications in endogenous conditions is poorly understood. This machinery can be incorporated in a process of disease progress or development. For example, suppose a post-translational modification is required for host cancer cell development but its activity can be blocked by cancer drugs. However, another site’s post-translational modification could compensate for that change, so that host cancer cells can survive, proliferate, and eventually metastasize. For cell proliferation and differentiation in general, such compensatory machinery could be a versatile backup system. However, such backup systems may not have been detected experimentally due to our limited technology or brief experimental periods. Thus, many E3 ligases may work in similar signal pathways (see also the previous chapter, Conclusion), or the function of a post-translational modification in one site may be compensated for or complemented by another site, but it is currently unknown how likely such complementary machineries would be. Research to predict such compensatory systems and resilience could be expected as future directions to study the spatiotemporal regulation of E3 ligase of CENP-A.

Ultimately, studies of E3 ligase in CENP-A in higher mammals or humans are essential for translational research and informing future therapy. Overexpression and mislocalization of human CENP-A are presumably features of cancer development, however, the detailed mechanisms for cancer development and possible therapies still remain unclear. In addition to cancer, translational studies of CENP-A and its E3 ligase could be beneficial for CREST autoimmune diseases and other diseases. Centromere proteins, including CENP-A, have been identified as antigens from CREST patients [204, 205], but the mechanism that causes CREST syndrome and how CENP-A and other centromere-kinetochore proteins are involved is unknown. Observations of neocentromeres were also reported in patients with other developmental diseases [206], but research has been limited, in part because of the relatively smaller number of patients.

Defects in ubiquitin E3 ligases promote the pathogenesis of several human diseases, including cancer, and CRL4 [207], a well-defined E3 ligase, has been reported to be upregulated and is proposed to be a potential drug target in cancers [208]. However, the biological functions of CRL4 and the underlying mechanism regulating cancer chemoresistance are still largely elusive. In humans, proteolysis activity of CRL4 ubiquitin ligase targeting CENP-A has not been observed so far, and other E3 ligases that function in CENP-A proteolysis are unidentified (Figure 2d). It is also important to determine if ubiquitylation or sumoylation-related enzymes, including E3 ligases, can be druggable targets.

Tumors develop in complex tissue microenvironments, where they depend on for sustained growth, invasion, and metastasis [209]. We could be at a turning point to fill the gap between the detailed intracellular mechanisms of CENP-A function studied in the past and its mechanism in complex tissue microenvironments. Thus, cell type and/or tissue-specific CENP-A function involved in different types of cancer in different organs is a likely focus for future research. There are many unknowns about whether the function of E3 ligase of CENP-A represents a cell or tissue-specific difference, or whether the cell or tissue completely replaces E3 ligase itself. The utilization and application of organoid, spheroid, and coculture systems may reduce the effort, time, and cost that is required to answer these questions and ultimately yield better therapies.

Acknowledgments

We thank past and current researchers at Model Animal Research Center, School of Medicine, Nanjing University, Greehey Children’s Cancer Research Institute at UT Health Science Center San Antonio, the Research Institute at Nationwide Children’s Hospital, and St. Jude Children’s Research Hospital for their helpful discussions. Y.N. was supported by Jiangsu Province “Double-First-Class” Construction Fund, Jiangsu Province Natural Science Fund (BK20191252), Jiangsu Province 16th Six Big Talent Peaks Fund (TD-SWYY-001), Jiangsu Province “Foreign Expert Hundred Talents Program” Fund (BX2019082), and National Natural Science Foundation in China (31970665). KK was supported by the National Science Foundation under Grant No.1949653 (KK) and a Mays Cancer Center Pilot Award CCSG P30 CA054174.

Conflict of interest

The authors declare no conflict of interest.