Dual Roads of CENP‐A for Centromere Identification

ABSTRACT

Faithful chromosome segregation requires the kinetochore, a macromolecular protein complex that assembles on centromeric chromatin. In vertebrates, centromere identity is epigenetically defined by the histone H3 variant CENP‐A, whose nucleosomes are replenished during early G1 phase through a cell cycle‐regulated deposition mechanism. The CENP‐A chaperone HJURP is recruited to centromeres via the Mis18 complex to enable CENP‐A incorporation (Mis18C pathway). Recent genetic analyses in chicken cells, however, revealed an additional, independent recruitment route via direct HJURP–CENP‐C interaction (CENP‐C pathway). Structural studies show that CENP‐C and the Mis18C subunit KNL2/M18BP1 engage pre‐existing CENP‐A nucleosomes at different surfaces of the Constitutive Centromere‐Associated Network (CCAN), constraining the spatial coordinates where new CENP‐A is incorporated. This review summarizes current understanding of HJURP recruitment mechanisms for vertebrate CENP‐A deposition and proposes that this structure‐defined deposition geometry underlies epigenetic self‐propagation of centromere positional information by maintaining proper CCAN spacing across cell divisions.

The Mis18C and CENP‐C pathways are structurally distinct in centromeres; however, each pathway acts on a pre‐existing CENP‐A nucleosome and recruits HJURP to deposit a new CENP‐A nucleosome nearby during the early G1 phase. This dual, structurally guided mechanism helps maintain centromere identity in chicken cells throughout cell divisions.

1. Introduction

Chromosome segregation is a critical event that occurs during cell division. It ensures that the genetic information replicated during S phase is faithfully transmitted to daughter cells. During this process, a large multiprotein complex called the kinetochore connects chromatin to microtubules. Chromosome segregation then occurs as spindle microtubules generate tension that pulls the chromosomes apart. The chromosomal region upon which the kinetochore assembles is called the centromere. Its genomic position is stably maintained across multiple cell divisions. The centromeric regions of vertebrates contain highly repetitive DNA sequences that are unique to each species. The human centromeric region, for example, consists of a 171‐bp α‐satellite unit, repeated thousands of times and spanning several megabases in total. Similarly, the mouse centromeric region contains a repetitive array of minor satellite units. Given this, it is easy to imagine a model in which such centromere‐specific DNA structures determine the position of the centromere. However, the discovery of neocentromeres—cases in which a noncentromeric chromosomal region suddenly becomes activated and acquires centromere function, despite the original centromeric DNA sequence remaining intact elsewhere in the genome (Figure 1a)—clearly demonstrates that centromere function is determined by a mechanism independent of DNA sequence [1, 2]. These findings led to the current model in which centromere identity is epigenetically specified in vertebrates. A unique centromere‐specific histone H3 variant called CENP‐A serves as the primary epigenetic mark within a specialized nucleosome. Studies conducted in vertebrate cells, including human, chicken, and Xenopus cells, have established a model in which new CENP‐A nucleosomes are deposited near pre‐existing CENP‐A nucleosomes, thereby maintaining centromere position across cell divisions [3, 4, 5]. During the G1 phase, the Mis18 complex (Mis18C, consisting of KNL2/M18BP1, Mis18α, and Mis18β) assembles at centromeres [6, 7, 8]. This assembly is dependent on several kinetochore proteins and occurs in the vicinity of pre‐existing CENP‐A nucleosomes [9, 10, 11]. The pre‐nucleosomal CENP‐A and histone H4 dimer forms a complex with its chaperone HJURP (HJURP‐C). Then, HJURP‐C recognizes Mis18C at centromeres, leading to its own centromeric recruitment. This leads to the deposition of new CENP‐A nucleosomes (Mis18C pathway). Recent genetic analyses using chicken DT40 cells have revealed that, in addition to the Mis18C pathway, a pathway dependent on the centromere protein CENP‐C (CENP‐C pathway) also leads to CENP‐A incorporation independently of the Mis18C pathway [12, 13]. Multiple lines of evidence indicate that both the Mis18C and CENP‐C pathways also operate in Xenopus cells [4, 14, 15, 16]. Comparative analyses of the conserved and divergent features of centromere specification mechanisms across species are expected to illuminate the fundamental principles governing centromere epigenetics.

FIGURE 1.

Current models of CENP‐A incorporation in vertebrate cells. (a) Neocentromeres form at noncentromeric loci despite an intact original centromere, demonstrating DNA sequence‐independent centromere specification. (b) Schematic model of CENP‐A loading into centromeres in human cells. Just before the G1 phase begins, the Mis18 complex (Mis18C) assembles and localizes to the centromere by interacting with CENP‐C and several CCAN proteins. In the early G1 phase, the CENP‐A‐H4/HJURP pre‐deposition complex is recruited to the centromere via recognition of Mis18C. Then, CENP‐A is loaded as a nucleosome near the old CENP‐A nucleosome (Mis18C pathway). (c) In the case of chickens and Xenopus, the CENP‐A‐H4/HJURP pre‐deposition complex localizes to the centromere by binding to either CENP‐C (CENP‐C pathway) or Mis18C (Mis18C pathway). Then, CENP‐A loading occurs via both pathways at the beginning of the G1 phase.

CENP‐A assembles into chromatin as a nucleosome, forming an octamer together with two copies each of the histones H4, H2A, and H2B [17]. Comparing nucleosome structures containing either canonical H3 or CENP‐A reveals that CENP‐A nucleosomes have unique structural features, including CENP‐A‐specific tail sequences and an extended L1 loop that forms the RG loop, which are absent from H3 nucleosomes [17]. These features are recognized by CENP‐C and CENP‐N, which serve as a scaffold for assembling multiple factors within the CENP‐A chromatin domain. This ultimately leads to kinetochore formation [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28]. Interestingly, in addition to CENP‐C and CENP‐N, KNL2/M18BP1, a component of Mis18C, also directly binds to the CENP‐A nucleosome in most vertebrates except mammals [3, 4, 14, 29, 30].

The kinetochore is a macromolecular complex, composed of more than a hundred proteins at each centromere. In vertebrates, 16 conserved proteins assemble on centromeric chromatin to form a protein network known as the Constitutive Centromere‐Associated Network (CCAN) [31, 32, 33, 34, 35]. When cells enter M phase, three additional protein complexes (the KNL1 complex: KNL1C, the Mis12 complex: Mis12C, and the Ndc80 complex: Ndc80C) further assemble to form a mitosis‐specific network called the KMN network [35, 36, 37]. The CCAN and KMN together constitute the functional structural unit essential for accurate chromosome segregation during M phase. To understand the function of the kinetochore at the molecular level, significant effort by many groups has gone into determining the structures of its constituent proteins and their complexes. In particular, the structures of each KMN subcomplex have been reported, as have models of the binding between microtubules and the Ndc80 complex [36, 38, 39, 40, 41, 42, 43]. More recently, cryogenic electron microscopy (cryo‐EM)‐based approaches have resolved the structures of the CCAN proteins in complex with DNA or the CENP‐A nucleosome [44, 45, 46, 47, 48, 49, 50] as well as the KMN network [51, 52]. These advances have made it possible to explain the molecular details of centromeric chromatin architecture or kinetochore‐microtubule attachment. Recent studies have also made progress in describing the interactions between Mis18C and the CCAN complex in the context of CENP‐A deposition [11]. By modeling the interactions among the CENP‐A nucleosome, the factors that directly bind it (CENP‐C and Mis18C), and HJURP‐C, all of which are centered on the CCAN, it may be possible to create a spatial blueprint of the chromatin architecture into which new CENP‐A nucleosomes are incorporated. In this review, we aim to integrate these findings into a model of centromere epigenetics, identify key conceptual considerations, and highlight remaining questions.

2. CENP‐A Deposition Pathways Vary Among Species

In vertebrates, the position of the centromere, which is the platform upon which the kinetochore assembles, is determined by the presence of CENP‐A (Figure 1a). Like canonical histone H3, CENP‐A forms nucleosomes. Due to its exceptional stability within the genome, CENP‐A allows for the epigenetic inheritance of centromere positional information across multiple cell divisions. In humans, chickens, and Xenopus, CENP‐A is deposited during the early G1 phase, which occurs immediately after cell division. This replenishes the pool of CENP‐A, which is halved during the S phase. Thus, both the quantity and position of this epigenetic mark are maintained [53, 54, 55]. Prior to incorporation into chromatin, the CENP‐A‐histone H4 dimer forms a complex with the dedicated chaperone HJURP (HJURP‐C) [56, 57]. In human cells, the centromeric targeting of HJURP‐C during the G1 phase proceeds through three sequential steps: (1) assembly of Mis18C (Mis18α, Mis18β, and KNL2/M18BP1); (2) recruitment of Mis18C to the centromere through its interactions with multiple CCAN components; and (3) binding of HJURP‐C to the assembled Mis18C, resulting in centromeric localization (Figure 1b).

Studies in human cells have shown that each of these steps is regulated by the activities of cyclin‐dependent kinase (CDK) and Polo‐like kinase 1 (PLK1). From late G1 through M phase, CDK‐mediated phosphorylation of KNL2/M18BP1 at multiple sites prevents the formation of Mis18C and inhibits its centromeric localization [7, 8, 58, 59]. Upon mitotic exit, the decline in CDK activity leads to the dephosphorylation of KNL2/M18BP1, which is likely mediated by PP2A [60]. This allows KNL2/M18BP1 to assemble with Mis18α/β and localize to centromeres. Then, PLK1 phosphorylates Mis18α via its interaction with KNL2/M18BP1. This induces a conformational change in the N‐terminal region of Mis18α, enabling binding to HJURP‐C [61, 62, 63]. PLK1 has also been shown to interact directly with Mis18α and HJURP [62, 63]. It remains to be established whether PLK1 activity is required solely for HJURP recruitment or if it also plays an essential role in the subsequent incorporation of CENP‐A into chromatin.

In contrast, in chickens and Xenopus, the recruitment of HJURP‐C employs two distinct pathways: The Mis18C pathway, in which the complex recognizes Mis18C and localizes to the centromere, is similar to the pathway in humans. The other one is the CENP‐C pathway, in which CENP‐C directly binds to HJURP‐C, which then localizes to the centromere (Figure 1c) [4, 12, 14]. In both pathways, once HJURP‐C is recruited to the centromere, it subsequently deposits new CENP‐A nucleosomes into neighboring chromatin (Figure 1c). CDK activity also participates in regulating CENP‐A deposition in chickens and Xenopus [15, 16, 58]. In chicken DT40 cells, inhibiting CDK1 and CDK2 results in ectopic CENP‐A deposition during the G2 phase, when deposition does not normally occur [58]. In Xenopus, HJURP phosphorylation at S220 inhibits its interaction with CENP‐C. Dephosphorylation of this site permits binding. This G1‐specific activation of the HJURP–CENP‐C interaction parallels the cell cycle‐dependent regulation observed in human cells [15]. Since inhibiting PLK1 in Xenopus blocks CENP‐A deposition [64], a regulatory mechanism similar to that in human cells may also exist in this system. However, the precise molecular details of how CDK and PLK1 coordinate the two deposition pathways in nonmammalian vertebrates remain to be investigated. According to this model, a single deposition unit in human cells introduces one new CENP‐A nucleosome at a single site (Figure 1b). In contrast, in chickens and Xenopus, two independent deposition units (one driven by the Mis18C pathway and the other by the CENP‐C pathway) each contribute to new CENP‐A incorporation (Figure 1c). Nevertheless, the key mechanistic steps that govern CENP‐A deposition after HJURP recruitment to the centromere are not well understood. Questions that await future investigation include whether additional factors are required, whether CENP‐A is incorporated into nucleosome‐free chromatin, and whether it is introduced through exchange with specialized placeholder nucleosomes, such as those containing H3.3 [65, 66].

3. CENP‐A Binding Proteins Promote CENP‐A Loading

CENP‐A nucleosomes define the position of functional centromeres and are recognized by multiple centromere‐associated proteins. This recognition is believed to read the epigenetic mark and trigger a series of events that lead to kinetochore assembly. The mechanisms by which CENP‐A nucleosomes are distinguished from the abundance of canonical H3 nucleosomes throughout the genome have been characterized through biochemical and structural approaches. In humans, CENP‐A has distinctive tail sequences at both its N‐ and C‐termini that are clearly different from those of canonical H3. The C‐terminal tail comprises six CENP‐A‐specific residues and is recognized by CENP‐C [18, 19, 20, 21, 22, 23, 24]. The N‐terminal tail is also recognized by CENP‐C, but indirectly via CENP‐B [20]. Possessing at least one of these two tail‐recognition activities is sufficient for kinetochore assembly, indicating that the two interactions are functionally redundant [19]. Furthermore, structural analyses of reconstituted CENP‐A nucleosomes using recombinant proteins revealed a distinctive protruding loop structure arising from a unique two‐amino acid insertion (Arg‐Gly, called the RG loop) within the L1 region of CENP‐A. This feature is absent from H3 nucleosomes and represents a defining structural difference between the two [17]. Structural studies have shown that CENP‐N recognizes and interacts with the RG loop through its N‐terminus [23, 25, 26, 27, 28]. However, structural analyses of the intact CCAN complex suggest that the CENP‐N subunit would sterically clash with the CENP‐A nucleosome upon direct contact [44, 45, 48, 50]. The interaction between the recombinant CENP‐A nucleosome and CENP‐N is disrupted when recombinant CENP‐C is added [26]. Whether this structural incompatibility reflects the existence of a CENP‐N recognition mode that operates independently of the CCAN under specific functional contexts remains an open question [67]. Structural and biochemical studies in chicken and Xenopus cells have demonstrated that CENP‐C recognizes and directly binds CENP‐A nucleosomes [26, 30, 68]. In chickens, the evolutionarily conserved CENP‐C motif and the downstream region of CENP‐C mediate binding to CENP‐A nucleosomes through the C‐terminal tail and the unique RG loop of CENP‐A, respectively (Figure 2a) [26]. Intriguingly, the phosphorylation of the neighboring threonine residue (T651) in CENP‐C during the M phase has been shown to strengthen this interaction [26, 69]. CENP‐C also directly binds to Mis12C, which is part of the KMN outer kinetochore network, through its N‐terminus (Figure 2a). Whether this tight binding to CENP‐A nucleosomes during M phase is required to withstand the tensile forces on chromatin from microtubule pulling remains an open question. However, cells expressing a CENP‐C mutant lacking the N‐terminal region required for Mis12C binding remain viable in vertebrate species, indicating that microtubule attachment via CENP‐C is not essential [70, 71]. Moreover, at least in cultured cells, the CENP‐C motif itself is dispensable for cell viability [69]. The reinforced binding of CENP‐C to CENP‐A nucleosomes during M phase may instead serve to guide the deposition of new CENP‐A in the subsequent G1 phase. The biological significance of the CENP‐C binding to the CENP‐A nucleosome remains to be determined.

FIGURE 2.

Key CENP‐A binding proteins. (a) CENP‐C domain structures in humans, chickens, and Xenopus. CENP‐C is a conserved CENP‐A‐binding protein that possesses several key biological domains. The well‐conserved Mis12C‐binding and Cupin domains are located at the N‐terminal tip and C‐terminal end, respectively. Another conserved CCAN‐binding domain is on the C‐terminal side of the Mis12C‐binding domain. In humans and Xenopus, CENP‐C has two CENP‐A‐binding regions, the central domain and the CENP‐C motif. In chickens, however, only the CENP‐C motif is present; the central domain is absent. The AF3‐predicted complex structure of the chicken CENP‐C motif of CENP‐C643‐683 bound to the CENP‐A nucleosome is shown on the right, and its schematic model is also shown in the upper right. (b) KNL2/M18BP1 domain structure in humans, chickens, and Xenopus. KNL2 is a conserved component of Mis18C and possesses several key biological domains. The Mis18C‐binding, SANTA, and SANT domains are functionally conserved and located at the N‐terminal tip, middle, and C‐terminal side, respectively. Chickens and Xenopus have a unique domain called the CENP‐C‐like motif, which has sequence similarity to the CENP‐C motif of CENP‐C and is responsible for CENP‐A binding. However, humans do not possess this motif. Notably, Xenopus KNL2 has two isoforms, L and S. The AF3‐predicted complex structure of the chicken CENP‐C‐like motif of KNL2^517‐560 bound to the CENP‐A nucleosome is shown on the right, and its schematic model is also shown in the upper right.

Chicken and Xenopus KNL2/M18BP1 contain a CENP‐C‐like motif that is closely related in sequence to the CENP‐C motif (Figure 2b). Structural analyses have revealed that the CENP‐C‐like motif mediates direct binding to CENP‐A nucleosomes [3, 4, 14, 29, 30]. Unlike CENP‐C, chicken KNL2 does not have any identified phosphorylation sites that regulate its binding to CENP‐A nucleosomes. In contrast to humans, chicken KNL2 localizes to centromeres throughout the cell cycle [3]. In addition, genetic analyses have shown that the centromeric localization of KNL2 depends entirely on CENP‐A nucleosomes. This suggests that chicken cells lack a cell‐cycle‐dependent regulatory mechanism that controls KNL2 localization [3, 29]. In Xenopus, two isoforms of KNL2/M18BP1 (M18BP1‐L and M18BP1‐S) display subtly distinct cell‐cycle‐dependent localization behaviors (Figure 2b) [10, 14, 16]. M18BP1‐S localizes to centromeres throughout the cell cycle [16], whereas M18BP1‐L is centromeric in interphase but loses centromeric localization in M phase [10]. A phosphorylation event that governs the cell‐cycle‐regulated localization of M18BP1‐L has recently been described. Phosphorylation of a serine residue within the CENP‐C‐like motif of M18BP1‐L during M phase disrupts its binding to CENP‐A nucleosomes, leading to the loss of centromeric localization [14]. M18BP1‐S similarly dissociates from CENP‐A nucleosomes in M phase; however, in this case CDK phosphorylation of T166 triggers a switch in binding partners from CENP‐A nucleosomes to CENP‐C [16]. This binding partner switch in M18BP1‐S during M phase acts to inhibit interactions between HJURP and CENP‐C or Mis18C. In Xenopus, a model has been proposed in which these regulatory mechanisms restrict new CENP‐A deposition in interphase.

CENP‐C and KNL2 are two proteins that directly bind to CENP‐A nucleosomes and play critical roles in CENP‐A deposition. Recognizing pre‐existing CENP‐A nucleosomes with the CENP‐A loading machinery is a logical, mechanistic strategy for faithfully inheriting centromere positional information. However, a notable divergence exists: unlike their chicken and Xenopus counterparts, human KNL2/M18BP1 lacks a CENP‐C‐like motif and does not directly bind CENP‐A nucleosomes (Figure 2b). Instead, it localizes to pre‐existing CENP‐A nucleosomes through CENP‐C and other CCAN components. Since KNL2 orthologs in plants and nearly all vertebrates except mammals retain a CENP‐C‐like motif, it is likely that human KNL2 has lost this motif during evolution [72, 73]. To compensate, human cells have acquired a centromeric targeting system that relies on CENP‐C, which recognizes CENP‐A nucleosomes together with several CCAN components. This system preserves the ability to deposit new CENP‐A in proximity to pre‐existing CENP‐A nucleosomes.

4. Epigenetic Maintenance of CENP‐A Stoichiometry at the Centromere

The position of the centromere is maintained with remarkable stability in vertebrates. However, centromere drift, a phenomenon in which the centromeric locus shifts by several to tens of kilobases along the genome, has been reported in multiple organisms [74, 75, 76]. In addition, major chromosomal rearrangements, such as chromosomal breaks and fusions, can lead to de novo centromere formation at entirely novel genomic loci. This phenomenon is known as neocentromere formation (Figure 1a) [1, 2, 77, 78, 79, 80]. Thus, the centromere possesses a degree of plasticity that enables it to adapt to changes in genomic context and consequently reposition its identity. However, under normal circumstances, a robust positive feedback cycle operates to stably maintain centromere identity at the initially established locus. Consistent with this, observations across several organisms indicate that the size of the core centromeric chromatin domain—the foundation on which the kinetochore assembles—is kept constant. While the size of the core centromeric chromatin differs by about twofold between maize and oats, most maize chromosomes are eliminated in hybrid plants derived from crosses of the two species. The few surviving maize chromosomes’ centromeres then expand to match the size of oat centromeres [81]. Similarly, multiple neocentromeres that were experimentally induced in chicken DT40 cells were each found to span a region (∼30 kb) that is comparable in size to the endogenous centromere, even though they formed at distinct chromosomal positions [78]. The reason why each organism harbors a species‐specific centromere size and the molecular mechanisms that specify that size remain unclear. A super‐resolution microscopy study examining the intranuclear organization of the centromere proteins in human cells revealed that CENP‐A forms a ∼300 nm rosette‐like structure surrounding HJURP [82]. It is intriguing to speculate that if these structures are intrinsically species‐specific in size, this structural organization could be the size of the organism‐specific centromeric chromatin.

In any case, CENP‐A is the epigenetic mark that specifies both the position and the size of the centromere. The key deposition units responsible for CENP‐A incorporation recognize pre‐existing CENP‐A nucleosomes on chromatin. Therefore, newly deposited CENP‐A is invariably placed in close proximity to existing CENP‐A, ensuring the faithful inheritance of positional information. Is the copy number of CENP‐A nucleosomes maintained during multiple cell divisions? Quantitative, cell‐cycle‐resolved analyses in human cells have shown that the amount of CENP‐A per centromeric chromatin domain peaks immediately after CENP‐A deposition during the G1 phase. Then, it is halved as the genome replicates in the S phase, and the CENP‐A nucleosomes are distributed between sister chromatids [53, 83, 84]. Thus, the copy number is maintained through iterative cycles of replenishment in G1 and halving in S phase. In human cells, Mis18C is recruited to centromeres through interactions with CCAN components, including CENP‐C, which directly binds the CENP‐A nucleosome (Figure 3a) [11]. New CENP‐A is then deposited by HJURP‐C, which associates with Mis18C. In human cells, one deposition unit corresponds to one CCAN unit: each CCAN unit engages a single pre‐existing CENP‐A nucleosome and deposits one new CENP‐A nucleosome (Figure 3a). Immediately after deposition, each CCAN unit is therefore associated with two CENP‐A nucleosomes. Upon S phase progression, these nucleosomes are distributed between sister chromatids, returning each CCAN unit to a state containing a single CENP‐A nucleosome and thereby preserving copy number across cell divisions (Figure 3a).

FIGURE 3.

Epigenetic stoichiometry to maintain centromere chromatin. a) Models showing the number of CENP‐A nucleosomes in each CCAN unit or the CENP‐A loading machinery at various stages of the cell cycle, including G2/M, G1, and post‐replication G2 in human cells. b) Model showing the number of CENP‐A nucleosomes in each CCAN unit or each CENP‐A loading machinery (Mis18C and CENP‐C) at various stages of the cell cycle, including G2/M, G1, and G2 in chicken and Xenopus cells. Mis18C and CENP‐C bind to the CENP‐A nucleosome independently and promote CENP‐A loading.

In chickens and Xenopus, two distinct CENP‐A deposition pathways operate at centromeres, driven by CENP‐C and Mis18C (KNL2/M18BP1), respectively. In contrast to humans, the two deposition units in these organisms together constitute a single CCAN unit, with each deposition unit independently engaging a separate pre‐existing CENP‐A nucleosome (Figure 3b). Cytogenetic analyses in chicken cells indicate that the Mis18C and CENP‐C pathways each independently deposit equivalent amounts of CENP‐A; consistent with this, each pathway is predicted to deposit one new CENP‐A nucleosome per cell cycle [12]. Under this model (Figure 3b), immediately after deposition, each CCAN unit would contain four CENP‐A nucleosomes (two pre‐existing plus two newly deposited). Following DNA replication and entry into G2, these would be distributed to yield two CENP‐A nucleosomes per CCAN unit, thereby preserving copy number. Taken together, these observations suggest a general principle by which each deposition unit deposits a number of new CENP‐A nucleosomes proportional to the number of pre‐existing CENP‐A nucleosomes it recognizes, ensuring faithful copy number maintenance across cell divisions. A systematic analysis of CENP‐A copy number at defined cell cycle stages will be essential to verify this model.

5. Is the Structurally Defined CENP‐A Loading Position Essential for Centromere Epigenetics?

CENP‐C and Mis18C are two key factors responsible for CENP‐A deposition in chickens and Xenopus. Therefore, a complete structural understanding of the mechanisms underlying their functions is essential to providing a detailed molecular blueprint of how CENP‐A is incorporated into centromeric chromatin. The sites at which CENP‐C and Mis18C bind to HJURP‐C directly influence the position of new CENP‐A deposition on chromatin. Along with the interfaces through which CENP‐C and Mis18C bind to pre‐existing CENP‐A nucleosomes, these interactions define the spatial coordinates of new CENP‐A incorporation into centromeric chromatin (Figure 2a, b). Recent studies using the chicken model have employed AlphaFold‐based structure prediction and in vitro pull‐down analyses. These studies revealed that the hydrophobic surface of the CENP‐C Cupin dimer interacts directly with a hydrophobic region located C‐terminal to the CENP‐A‐binding domain of HJURP (Figure 4a) [12]. This region is referred to as the CENP‐C‐binding domain (Figure 4a). In G1, two HJURP‐C molecules are recruited via the Cupin domains of the CENP‐C homodimer (Figure 4b left). Consequently, the two sets of CENP‐A:H4 supplied by these two HJURP molecules give rise to a single new CENP‐A nucleosome (Figure 4b left). A region corresponding to the CENP‐C‐binding domain of chicken HJURP appears to be conserved in humans; however, significant binding between this region and the CENP‐C Cupin domain has not been detected by AlphaFold prediction or in vitro binding assays [12]. Nevertheless, an in vitro binding assay has demonstrated human HJURP interaction with the C‐terminus of CENP‐C [85], suggesting that a CENP‐C‐dependent HJURP recruitment pathway may also operate in human cells through a distinct interface.

FIGURE 4.

HJURP domain structure and its two centromeric contact sites. (a) HJURP domain structure in chickens, Xenopus, and humans. HJURP possesses several key biological domains. The well‐conserved CENP‐A‐binding domain is located at the N‐terminal tip. The Mis18C‐binding domain is shown in humans and chickens. The CENP‐C‐binding domain is shown in chickens and Xenopus. The predicted structures, generated using AF3, are also shown: the interaction between chicken HJURP100–150 and a dimer of chicken CENP‐C677–864 (lower left); and the interaction between chicken HJURP493–523 or HJURP546–576 and the Mis18α/β complex, which contains two Mis18α160–204 molecules and one Mis18β160–213 molecule (lower right). (b) Schematic models illustrate CENP‐A loading through the CENP‐C (left) and Mis18C (right) pathways. In the upper panel, during the M phase, a dimerized CENP‐C or two KNL2 molecules bind to a CENP‐A nucleosome. Next, in the G1 phase (lower), the CENP‐C dimer recruits two CENP‐A‐H4/HJURP complexes, which cooperatively load one CENP‐A nucleosome. In the case of the Mis18C pathway, two KNL2 molecules form a complex with Mis18 (4α and 2β) and recruit one CENP‐A‐H4/HJURP complex, followed by the loading of one CENP‐A nucleosome.

In the G1 phase, Mis18C localizes to centromeres as a hetero‐octamer comprising two KNL2/M18BP1 molecules, four Mis18α molecules, and two Mis18β molecules. One unit of HJURP‐C binds to the Mis18α/β side of this octamer via the Mis18C‐binding domain of HJURP (Figure 4a, b right) [7, 8, 86]. Unlike the CENP‐C pathway, the Mis18C pathway supplies only one set of CENP‐A:H4 in a single step. This implies that an additional set of CENP‐A:H4 must be supplied to complete the assembly of one new CENP‐A nucleosome. It is unclear whether this second set is delivered by HJURP recruited via a neighboring CENP‐C or a distinct Mis18C unit, or whether it is re‐supplied from the same Mis18C unit through successive rounds of HJURP‐C turnover. In human cells, a model has been proposed in which HJURP forms a homodimer through its C‐terminal region. In this scenario, a single HJURP homodimer associated with Mis18C could supply enough CENP‐A:H4 to assemble one new CENP‐A nucleosome in a single deposition event [87].

CENP‐C interacts with CCAN components, primarily CENP‐L/‐N, via its CCAN‐binding domain (Figure 4b left, 5a) [88, 89]. This interaction determines the spatial arrangement of the CCAN on centromeric chromatin and constitutes a critical contact that influences where new CENP‐A is deposited. Beyond CENP‐C, a recent study has further demonstrated that Mis18C interacts with a specific surface of the CCAN in the human system. According to the report [11], the SANTA domain of M18BP1 and the ∼60 amino acid pre‐SANTA region flanking it N‐terminally mediate centromeric localization by contacting CENP‐H, CENP‐K, and CENP‐C within the CCAN (Figure 4b right, 5a). The amino acid residues in KNL2/M18BP1 that are critical for these interactions are relatively well conserved in chicken and Xenopus, suggesting that analogous contacts may exist in these species (Figure 5a). Recent advances in cryo‐EM‐based structural analysis have made it possible to solve structures encompassing nearly all subunits of the CCAN bound to a nucleosome or linker DNA [44, 45, 46, 47, 48, 49, 50, 90]. Based on these structures, the relative positioning of each component can be considered. CENP‐C and Mis18C each adopt a fixed orientation through their contacts with a specific surface of the CCAN (Figure 5a, b). From these positions, they interact with CENP‐A nucleosomes on chromatin. In other words, the spacing of CENP‐A nucleosomes on centromeric chromatin is constrained by the structural geometry of the CCAN–CENP‐C and CCAN–Mis18C interfaces. Therefore, new CENP‐A nucleosomes are deposited at a position dictated by the spatial arrangement of CENP‐C and Mis18C. This process propagates a characteristic CENP‐A nucleosome spacing that is permissive of these structural interactions (Figure 5b). A recently reported cryo‐EM structure of an in vitro reconstituted complex, in which the CCAN is assembled onto a dimeric α‐satellite DNA harboring two CENP‐A nucleosomes, has provided a structural model of how the kinetochore is built on CENP‐A nucleosome arrays [90]. It will be interesting to examine the extent to which this in vitro model, derived from cryo‐EM analysis, is consistent with the spatial arrangement of the CCAN predicted by the in vivo CENP‐A deposition mechanism described above.

FIGURE 5.

Does the structure of the kinetochore unit define the CENP‐A loading points for each pathway? (a) The AF3‐predicted complex structure of chicken CCAN proteins with linker DNA, together with the pre‐SANTA domain of KNL2301‐343 and the CCAN‐binding domain of CENP‐C166‐227, is shown. Binding sites for the pre‐SANTA domain of KNL2 and the CCAN‐binding domain of CENP‐C are indicated by arrows on the CCAN structure. (b) Models of CENP‐A loading via the CENP‐C and Mis18C pathways, together with CCAN. The upper model shows that CENP‐C or Mis18C (KNL2) interacts with one CCAN unit. In the lower model, CENP‐C or Mis18C (KNL2) makes contact with two neighboring CCAN units as shown in Figure 4b. The organization of the contacts between CCAN and CENP‐C or Mis18C (KNL2) defines the spacing of the CCAN units along the centromeric chromatin, which fixes the CENP‐A deposition sites.

The linear arrangement of CENP‐A nucleosomes and CCAN units on centromeric chromatin is organized in 3D on the surface of mitotic chromosomes, promoting the proper capture of spindle microtubules [91]. In interphase nuclei, however, a distinct higher‐order organization of CCAN‐associated chromatin may be established. This creates a specialized nuclear environment that promotes proper CENP‐A deposition, including the spatial definition of the chromosomal domain where new CENP‐A is incorporated [82].

Consistent with this model, centromere drift, a phenomenon in which the genomic position of the core centromere shifts during cell divisions, has been reported at an elevated frequency in chicken cells lacking CENP‐U or CENP‐S. These proteins are members of the CENP‐O and CENP‐T complexes, respectively, and are subunits of the CCAN [74]. While CENP‐U and CENP‐S are dispensable for chicken cell viability, knocking out either subunit perturbs stable CCAN assembly and leads to CCAN destabilization [74, 92, 93]. Therefore, this centromere drift phenotype may reflect a scenario in which the destabilization of the CCAN structure—which provides the proper structural geometry for CENP‐A deposition—ultimately compromises the fidelity of CENP‐A incorporation site specification. Cells that have proliferated for many generations are in a state in which centromeric chromatin with the correct CENP‐A nucleosomes and CCAN arrangement can autonomously and epigenetically self‐propagate. During DNA replication, this spatial arrangement is faithfully transmitted to the next cell generation. It is of particular interest whether this model is correct, and specifically, whether the positions at which new CENP‐A is incorporated exhibit a defined spatial arrangement consistent with CCAN spacing.

6. Biological Significance of the Dual CENP‐A Deposition

CENP‐A deposition mechanisms fall into two vertebrate categories: humans use only the Mis18C pathway, while chickens and Xenopus additionally employ a CENP‐C pathway. Genetic analyses of chicken DT40 cells have demonstrated that these two pathways are not redundant; rather, each pathway is required for cell proliferation [12]. What is the biological significance of utilizing two distinct pathways for CENP‐A deposition? One possibility is that a single pathway is insufficient to maintain the quantity of CENP‐A required for centromere function. This is consistent with the hypothesis, discussed in Section 4, that each pathway contributes independently to CENP‐A deposition and, consequently, to maintaining CENP‐A levels. Alternatively, the dual‐pathway system may be important for maintaining centromere positional information more robustly. CENP‐B, an α‐satellite DNA‐binding protein that recognizes the CENP‐B box, is absent in chickens and Xenopus but may supplement the single‐pathway system in humans by providing an additional mechanism for maintaining centromere position.

Regarding the regulatory mechanisms that restrict both pathways to the G1 phase, CDK and PLK1 activities may also be involved in chicken cells. This is because a similar mechanism has recently been proposed for Xenopus in a preprint [64]. However, the precise molecular details of this regulation have yet to be determined. If the two pathways are governed by a common regulatory mechanism, coordinating the timing of CENP‐A deposition through both pathways would be straightforward. Conversely, if each pathway is independently regulated, one pathway could be selectively controlled over the other, raising the intriguing possibility that the relative contribution of each pathway could be modulated, for example, in a cell lineage‐dependent manner. It remains an open question whether this mechanistic divergence leads to biologically significant differences, such as cell lineage‐dependent functional modulation of the centromere, between organisms that rely on a single pathway and those that employ a dual pathway. Furthermore, the possibility that organisms currently considered to use a single pathway, such as humans, might operate through an as‐yet‐unrecognized dual‐pathway mechanism cannot be entirely ruled out. Therefore, it is worth reconsidering whether a dual‐pathway mechanism exists in mammals, and particularly in humans. In chickens, Mis18C and CENP‐C each localize independently, allowing the two pathways to be detected separately. In humans, however, Mis18C localization depends on CENP‐C, so distinguishing the two pathways and establishing their independent existence will require careful investigation. Further clarifying these detailed mechanisms, including whether a dual‐pathway system operates in mammals, could contribute to our understanding of chromosome segregation and its associated diseases. It could also lead to the development of corresponding therapeutic strategies. The functional significance of the dual‐pathway system and the possibility of a hidden dual‐pathway mechanism in mammals are important goals for future investigation.

As discussed in Section 5, it is important to experimentally test the possibility that the distinctive spatial arrangement of CENP‐A deposition is linked to the correct positioning of CCAN units and to the assembly of functional centromeric chromatin. If genetic systems capable of selectively inactivating the CENP‐C or Mis18C pathway could be established, the resulting changes in CENP‐A nucleosome positioning could potentially be detected by applying techniques such as DeMeLo‐seq. This method enables the mapping of epigenetic positions on individual alleles across long chromatin stretches [94]. It could be used to compare pathway‐specific mutant cells with controls. Although detecting such positional changes may be technically challenging, given reports of nonspecific CENP‐A incorporation into chromatin [78, 95], high‐resolution analysis of CENP‐A distribution would provide important indirect insight into the spatial arrangement of CCAN units on centromeric chromatin. Additionally, reconstituting and structurally analyzing higher‐order complexes comprising poly‐nucleosomal chromatin containing CENP‐A and canonical H3 nucleosomes in defined ratios, along with the CCAN and the CENP‐A deposition machinery, would be a critical step toward creating a definitive structural model of centromere epigenetic inheritance.

7. Conclusions and Prospects

In this review, we focused on the mechanisms governing HJURP localization—the key step in CENP‐A deposition—and their structural basis. In contrast, the subsequent molecular events, from CENP‐A loading onto chromatin to nucleosome formation, remain poorly understood, both in terms of the factors involved and the reactions that occur. Recently, ChIP‐CryoEM studies captured structural intermediates of RNA polymerase progressing along nucleosome‐containing template DNA, using immunopurified chromatin from human cell nuclei [96]. Applying similar structural approaches, combined with reconstituting or purifying relevant complexes from in vivo sources, would be a powerful way to capture intermediates along the CENP‐A nucleosome assembly pathway. Such efforts should reveal the in vivo molecular progression of CENP‐A nucleosome assembly, a process that remains largely uncharacterized mechanistically.

Finally, centromere organization strategies vary remarkably across the tree of life, reflecting diverse evolutionary solutions to the shared challenge of faithful chromosome segregation. Organisms with holocentromeres, where kinetochore activity spans the entire chromosome, either retain CENP‐A, as in Caenorhabditis elegans, or assemble kinetochores without it, as in the silkworm Bombyx mori [97]. Conversely, some organisms with regional centromeres similar to humans and chickens, such as Trypanosoma brucei and Mucor lusitanicus, determine centromere position independently of CENP‐A [98, 99]. In T. brucei, a set of kinetochore proteins with no clear homology to known kinetochore components has been identified, and functional analyses of these proteins are ongoing [100, 101, 102, 103]. The mechanisms of centromere specification in these divergent organisms remain unknown, and elucidating them is an important future challenge. With this broader landscape in mind, the vertebrate centromere, with its CENP‐A‐based epigenetic memory and structure‐guided deposition geometry, represents a well‐characterized solution to a universal biological problem. Continued investigation of its mechanisms will reveal principles relevant far beyond the vertebrate lineage.

Author Contributions

Tetsuya Hori: conceptualization, figure design, and writing – original draft and editing. Tatsuo Fukagawa: writing – review and editing.

Conflicts of Interest

The authors declare no competing interests.

Data Availability Statement

Data sharing is not applicable to this article as no unique datasets were generated for this review article.