The N-terminal tail of C. elegans CENP-A interacts with KNL-2 and is essential for centromeric chromatin assembly.

Centromeres are epigenetically defined by the centromere-specific histone H3 variant CENP-A. Specialized loading machinery, including the histone chaperone HJURP/Scm3, participates in CENP-A nucleosome assembly. However, Scm3/HJURP is missing from multiple lineages, including nematodes, with CENP-A-dependent centromeres. Here, we show that the extended N-terminal tail of Caenorhabditis elegans CENP-A contains a predicted structured region that is essential for centromeric chromatin assembly; removal of this region prevents CENP-A loading, resulting in failure of kinetochore assembly and defective chromosome condensation. By contrast, the N-tail mutant CENP-A localizes normally in the presence of endogenous CENP-A. The portion of the N-tail containing the predicted structured region binds to KNL-2, a conserved SANTA domain and Myb domain-containing protein (referred to as M18BP1 in vertebrates) specifically involved in CENP-A chromatin assembly. This direct interaction is conserved in the related nematode Caenorhabditis briggsae, despite divergence of the N-tail and KNL-2 primary sequences. Thus, the extended N-tail of CENP-A is essential for CENP-A chromatin assembly in C. elegans and partially substitutes for the function of Scm3/HJURP, in that it mediates a direct interaction between CENP-A and KNL-2. These results highlight an evolutionary variation on centromeric chromatin assembly in the absence of a dedicated CENP-A-specific chaperone/targeting factor of the Scm3/HJURP family.

INTRODUCTION

Centromeres are specialized chromosomal loci that direct chromosome segregation. In most species, active centromeres are defined by the presence of CENP-A, a histone variant that replaces histone H3 in centromeric nucleosomes (Kixmoeller ; Mitra ). CENP-A provides the physical foundation for assembly of the kinetochore, a multiprotein complex mediating spindle microtubule attachment to chromosomes (Musacchio and Desai, 2017). The cues leading to the centromere-restricted localization of CENP-A are being actively investigated. The underlying centromeric DNA is not conserved and, with the exception of budding yeasts, neither necessary nor sufficient to propagate CENP-A chromatin (Allshire and Karpen, 2008; McKinley and Cheeseman, 2016).

A segment of the histone fold domain of CENP-A, known as the CENP-A targeting domain (CATD), when transferred into canonical histone H3, is sufficient to confer centromere localization (Black ). The CATD of CENP-A interacts with a CENP-A–specific histone chaperone, known as olliday junction repair protein (HJURP) in vertebrates and Scm3 in fungi (Dunleavy ; Foltz ). This interaction is essential for CENP-A centromere targeting during mitotic exit (Jansen ; Foltz ). However, HJURP/Scm3 is not conserved in all species that build centromeres using CENP-A, including insects and nematodes (McKinley and Cheeseman, 2016). In Drosophila melanogaster, there is compelling evidence that the unrelated protein Cal1 acts as a functional homologue of HJURP/Scm3 (Erhardt ; Mellone ; Chen ). By contrast, in C. elegans, no HJURP/Scm3-like activity has been identified to date.

In addition to promoting CENP-A nucleosome assembly, HJURP/Scm3 proteins target the CENP-A/H4-HJURP/Scm3 prenucleosomal complex to the specific location of the centromere by interaction with centromeric DNA/chromatin-bound targeting factors. In budding yeast, the CBF3 complex specifically recognizes centromeric DNA and its subunit Ndc10 interacts with HJURP/Scm3 of the prenucleosomal complex to localize new CENP-A nucleosome assembly (Cho and Harrison, 2011). Outside of budding yeasts, where the CBF3 complex is not present, the Mis18 complex is the primary candidate for recognizing existing centromeric chromatin domains and targeting the deposition of new CENP-A via an interaction with HJURP/Scm3. The Mis18 complex is composed of Mis18 and/or KNL-2/M18BP1, depending on the species: Mis18α, Mis18β, and M18BP1 in humans (Fujita ); Mis18 only in Schizosaccharomyces pombe (Hayashi ; Pidoux ; Williams ); KNL-2 only in C. elegans (Maddox ), and Arabidopsis (Lermontova ). S. pombe Mis18 and human Mis18α/β interact with HJURP/Scm3 in vitro (Pidoux ; Wang ; Nardi ; Pan ) and Mis18 complex-mediated CENP-A recruitment can be bypassed by artificial tethering of HJURP/Scm3 to chromatin (Foltz ; Barnhart ; Ohzeki ). In human cells, centromere localization of the Mis18 complex precedes that of the CENP-A-H4-HJURP prenucleosomal complex during mitotic exit-coupled new CENP-A chromatin assembly (Jansen ). The Mis18 complex is proposed to recognize existing centromeric chromatin at least in part by binding to CENP-C, the reader of CENP-A nucleosomes that directs kinetochore assembly (Moree ; Kato ). However, in C. elegans, KNL-2 localizes to chromatin independently of CENP-C (Maddox ) and, even in nonmammalian vertebrates, the Mis18 complex directly recognizes CENP-A nucleosomes (French ; Hori ). KNL-2/M18BP1 family proteins contain a conserved Myb-like DNA-binding domain and a SANTA domain whose functions independent of CENP-C association are unclear (Zhang ; Maddox ; Ohzeki ; French and Straight, 2019). Interestingly, in fungi where KNL-2/M18BP1 proteins are absent, Myb domains can be found in HJURP/Scm3 proteins (Sanchez-Pulido ), suggesting potential fusion of multiple functions within a single polypeptide.

Here, we investigate CENP-A chromatin in C. elegans, which requires KNL-2/M18BP1 for its assembly but lacks an HJURP/Scm3 family member. C. elegans is holocentric, with condensed mitotic chromosomes having two “stripes” of CENP-A chromatin—one per sister chromatid—on geometrically opposite surfaces (Maddox ; Melters ). Genomic approaches have localized CENP-A and KNL-2 to broad permissive domains in the C. elegans genome that are transcriptionally inactive (Gassmann ), although restriction to specific sites has also been suggested (Steiner and Henikoff, 2014). We focused on the unusually long amino-terminal tail (N-tail) of C. elegans CENP-A, which, unlike the N-termini of CENP-A from other species, is predicted to harbor a region with α-helical secondary structure. Analysis of the divergent N-tails of CENP-A family members have implicated them in kinetochore assembly and epigenetic stability of centromeric chromatin (Ravi ; Fachinetti ; Folco ). Employing single-copy, targeted transgene insertion to replace endogenous CENP-A, we find that the N-tail of C. elegans CENP-A is essential for CENP-A loading, and we link this essential function to a direct interaction between the N-tail and the loading factor KNL-2/M18BP1. Interaction of the extended N-tail of C. elegans CENP-A to KNL-2/M18BP1 represents an evolutionary variation to HJURP/Scm3-mediated targeting of CENP-A to centromeres.

RESULTS

The C. elegans CENP-A N-tail has a predicted structured region that is essential for viability

The C. elegans CENP-A N-tail is unusually long at 189 amino acids and, based on computational analysis (performed using PSIPRED; http://bioinf.cs.ucl.ac.uk/psipred/), is predicted to have α-helical character in the first 100 amino acids (Figure 1A). The prediction of a structured region in the N-tail is unexpected as the CENP-A tail is often short (e.g., in fission yeast or humans, where it is 20 and 39 aa, respectively) and, even in other species with an extended CENP-A N-tail, such as Drosophila melanogaster (123 aa) or S. cerevisiae (130 aa), is not predicted to have any secondary structure (Figure 1A).

FIGURE 1:

The extended N-tail of C. elegans CENP-AHCP-3 contains a predicted structured region that is essential for viability. (A) Secondary structure predictions of CENP-A from different model organism species. Secondary structure predictions were generated using PsiPred. Predicted α-helical segments are indicated as boxes. The histone fold domain (HFD) is marked in gray. (B) Schematic of RNAi-resistant gfp::CENP-A single-copy transgene insertions on chromosome II. The three variants of CENP-AHCP-3 expressed from single-copy transgene insertions are indicated below. (C) Embryo viability analysis for the indicated conditions. N refers to the number of worms and n to the total number of embryos scored. Error bars are the SEM. (D) Anti-CENP-AHCP-3 immunoblot performed using an antibody raised to the linker region (aa 105–183) showing expression levels of WT GFP-CENP-AHCP-3 and the ∆109 N-tail truncation mutant in the presence and absence of endogenous CENP-AHCP-3 (∆ indicates homozygous CENP-A deletion mutant; (RNAi) indicates CENP-A). Asterisk (*) marks a background band that serves as a loading control.

To test the functional significance of the predicted structured region of the C. elegans CENP-A N-tail, we developed a transgene-based system to replace endogenous CENP-A (named HCP-3 and referred to here as CENP-AHCP-3) in C. elegans with engineered N-tail mutants. In brief, the nucleotide sequence of the CENP-A coding region was altered to maintain the native amino acid sequence while enabling selective RNAi-mediated depletion of endogenous CENP-AHCP-3; in addition, an N-terminal GFP tag was added to monitor localization (Supplemental Figure S1A). The wild-type (WT), as well as two mutant transgenes (∆109, which removes the predicted structured region, and ∆184, which removes the majority of the N-tail), were inserted in single copy at a fixed genomic location harboring a Mos transposon insertion (Figure 1B and Supplemental Figure S1A). The conditions we employ for RNAi penetrantly and selectively deplete the endogenous protein loaded into oocytes (Oegema and Hyman, 2006); in conjunction with RNAi-resistant transgenes, this approach enables replacing any targeted protein with a transgene-encoded version in the one-cell embryo. The reencoded gfp::CENP-A transgene rescued embryonic lethality observed following depletion of endogenous CENP-AHCP-3 by RNAi as well as the lethality of a deletion mutant (hcp-3(ok1892), referred to as CENP-A; Figure 1C and Supplemental Figure S1B). By contrast, deletion of the predicted structured region (∆109) and of the majority of the N-tail (∆184) resulted in fully penetrant embryonic lethality (Figure 1C). Immunoblotting with an antibody raised to the unstructured linker (amino acids 105–183 of the N-tail) indicated that the ∆109 mutant was expressed similarly to endogenous CENP-AHCP-3 (Figure 1D); thus, the observed lethality is not because the ∆109 N-tail mutant is not expressed. We conclude that the predicted structured region of the N-tail of CENP-A is essential for viability of C. elegans embryos.

The CENP-A HCP-3 N-tail deletion mutant exhibits a kinetochore-null phenotype in one-cell embryos

We next assessed the phenotype observed when endogenous CENP-AHCP-3 was replaced by the ∆109 N-tail mutant. We crossed an mCherry::H2b marker into strains harboring single-copy transgenes expressing WT or ∆109 GFP::CENP-AHCP-3, depleted endogenous CENP-AHCP-3 by RNAi, and imaged one-cell embryos. As a control, we also depleted CENP-AHCP-3 in the absence of any transgene. Depletion of CENP-A resulted in the characteristic kinetochore-null phenotype, with two clusters of chromatin–-one from each pronucleus–-instead of a metaphase plate, and a failure of segregation (Figure 2A; Oegema ; Desai ). This severe phenotype was fully rescued by transgene-encoded RNAi-resistant WT GFP::CENP-AHCP-3. By contrast, the observed phenotype for the ∆109 N-tail mutant was similar to that of removal of CENP-AHCP-3 (Figure 2A). Thus, deletion of the first 109 amino acids of the N-tail of CENP-AHCP-3 results in a chromosome segregation phenotype that is equivalent to CENP-AHCP-3 removal in the C. elegans embryo.

FIGURE 2:

Deletion of the predicted α-helical region of the CENP-AHCP-3 N-tail results in a kinetochore-null phenotype and failure to accumulate on mitotic chromatin. (A) mCherry::H2b images from time-lapse sequences for the indicated conditions in metaphase- and anaphase-stage one-cell embryos. Similar results were obtained in at least 10 embryos per condition. Scale bar, 5 µm. (B) Images of WT and ∆109 GFP::CENP- HCP-3∆ in metaphase-stage embryos expressing mCherry-H2b in the presence (left set of panels) or absence (right set of panels) of endogenous CENP-AHCP-3. Scale bar, 2.5 µm. (C) Quantification of integrated chromosomal GFP intensity in metaphase-stage embryos for the indicated conditions. t tests were used to assess if indicated pairwise comparisons were significantly different. Error bars are the SD. (D) Images of KNL-1::mCherry, expressed from an integrated single-copy RNAi-resistant transgene, in metaphase-stage one-cell embryos for the indicated conditions; note that endogenous KNL-1 was depleted in all cases. Scale bar, 5 µm. (E) Quantification of integrated KNL-1::mCherry kinetochore intensity in metaphase-stage embryos for the indicated conditions. Error bars are the SD.

The CENP-A HCP-3 N-tail deletion mutant does not accumulate on chromatin and fails to support kinetochore assembly

Stable incorporation of CENP-A into chromatin in yeast and humans involves a region of the histone fold, referred to as the CATD, which is specifically bound by the chaperone Scm3/HJURP (Black ; Cho and Harrison, 2011; Hu ). In these species, the N-tail is not essential for CENP-A centromere targeting, and alterations of the N-tail do not phenocopy loss of CENP-A (Chen ; Fachinetti ; Folco ). The similar phenotypes observed for CENP-AHCP-3 removal and for the ∆109 N-tail mutant of C. elegans CENP-A suggested that the ∆109 N-tail mutant, in contrast to the N-tail mutants in other species, may not accumulate on centromeric chromatin. To test this idea, we imaged WT and ∆109 GFP::CENP-AHCP-3 in a strain coexpressing mCherry::H2b, and quantified the GFP signal on metaphase chromosomes. In the presence of endogenous CENP-AHCP-3, both WT and ∆109 GFP::CENP-AHCP-3 localized to the diffuse kinetochores on the poleward faces of the holocentric mitotic chromosomes (Figure 2B) and quantification of fluorescence intensity indicated equivalent localization of both (Figure 2C). However, in the absence of endogenous CENP-AHCP-3, localization of ∆109 GFP::CENP-AHCP-3 was greatly reduced relative to WT GFP::CENP-AHCP-3 (Figure 2, B and C). Thus, ∆109 GFP::CENP-AHCP-3 fails to localize to chromatin on its own.

The absence of the ∆109 CENP-AHCP-3 mutant in chromatin should result in a kinetochore assembly defect. To confirm that this was indeed the case, we analyzed the localization of an RNAi-resistant mCherry-fusion of KNL-1, an outer kinetochore scaffold protein (Desai ; Espeut ). We introduced the transgene expressing this fusion into the strain expressing either WT or ∆109 GFP::CENP-AHCP-3, depleted endogenous KNL-1 and CENP-AHCP-3, and imaged and quantified the KNL-1::mCherry signal. This analysis revealed that loss of KNL-1::mCherry localization in the ∆109 GFP::CENP-AHCP-3 mutant was analogous to the CENP-AHCP-3 depletion (Figure 2, D and E). Thus, the ∆109 CENP-AHCP-3 mutant fails to support kinetochore assembly.

The failure of the CENP-A HCP-3 N-tail deletion mutant to accumulate on chromatin is not due to a kinetochore assembly defect

CENP-AHCP-3 is essential for kinetochore assembly (Oegema ). However, preventing kinetochore assembly by depleting CENP-CHCP-4 or KNL-1, both of which result in a similar kinetochore-null phenotype as CENP-AHCP-3 depletion (Oegema ; Desai ), does not affect GFP::CENP-AHCP-3 chromatin accumulation (Figures 2B and 3, A and B). Thus, the absence of ∆109 CENP-AHCP-3 on chromatin cannot be a secondary consequence of defective kinetochore assembly.

FIGURE 3:

Inability of ∆109 CENP-AHCP-3 to accumulate on chromatin is not due to failure of kinetochore assembly. (A) Images of WT GFP::CENP-AHCP-3 in metaphase-stage embryos also expressing mCherry::H2b that were depleted of endogenous CENP-AHCP-3 and KNL-1 (top) or CENP-CHCP-4 (bottom). Scale bar, 5 µm. (B) Quantification of integrated chromosomal WT GFP::CENP-A HCP-3 intensity in metaphase-stage embryos for the indicated conditions. The CENP-A alone value is the same as in Figure 2C. Error bars are the SD. t tests were employed to assess statistical significance of indicated pairwise comparisons. (C, D) Images of mCherry::H2b in sperm pronuclei from time-lapse sequences for the indicated conditions. Times are in seconds after nuclear envelope breakdown (NEBD). Similar results were observed in at least 10 embryos filmed per condition. Scale bar, 5 μm.

CENP-AHCP-3 depletion, in addition to the kinetochore-null phenotype, also leads to defects in condensation of the holocentric C. elegans chromosomes (Maddox , 2007). By contrast, preventing kinetochore formation by depletion of CENP-CHCP-4 does not result in a similarly severe condensation defect (Maddox ). We therefore compared chromosome condensation in CENP-AHCP-3 and CENP-CHCP-4 depletion to that in the ∆109 CENP-AHCP-3 tail mutant (Figure 3C); we additionally assessed if WT GFP::CENP-AHCP-3 rescued the condensation defect observed in the CENP-AHCP-3 depletion (Figure 3D). This analysis focused on sperm pronuclei as they are formed before injection of the dsRNA employed to deplete endogenous CENP-AHCP-3 and are therefore free of potential meiotic defects (Maddox ). Chromosome condensation in the ∆109 CENP-AHCP-3 mutant resembled that resulting from CENP-AHCP-3 depletion and not CENP-CHCP-4 depletion (Figure 3C); in addition, transgene-encoded WT CENP-AHCP-3 rescued the condensation defect of CENP-AHCP-3 depletion (Figure 3D). These results provide additional support for the conclusion that the ∆109 mutant is compromised for its loading onto chromatin. When considered in light of prior work showing that CENP-AHCP-3 is loaded de novo on chromatin in one-cell C. elegans embryos (Gassmann ), the above results indicate that the absence of ∆109 CENP-AHCP-3 on chromatin is due to a defect in its loading after fertilization.

The predicted α-helical region of the CENP-A tail interacts with the CENP-A loading factor KNL-2

The above results implicate the predicted structured region of the N-tail of C. elegans CENP-AHCP-3 in its loading on chromatin. In species with Scm3/HJURP, a key step in the loading reaction is the interaction of the Scm3/HJURP-CENP-A complex with the Mis18 complex, which includes the Myb domain–containing protein KNL-2 (also known as M18BP1 or Mis18BP1, based on its association with Mis18α/β in human cells; Fujita ; Maddox ; Wang ; French ; Pan ; to date, a Mis18α/β homologue has not been identified in C. elegans). In addition, Scm3/HJURP functions as a chaperone for assembly of CENP-A nucleosomes (Dunleavy ; Foltz ; Pidoux ; Williams ; Shuaib ).

To determine how the N-tail of CENP-AHCP-3 contributes to its loading, we tested if it interacts with KNL-2. KNL-2 family proteins are characterized by a conserved Myb-like DNA-binding domain and a predicted folded N-terminal domain referred to as the SANTA domain (Figure 4A; Zhang ; Maddox ); in addition, they possess an acidic/aromatic tail at the C-terminus (Supplemental Figure S2A). Using yeast two-hybrid analysis, we observed a robust interaction between a segment of the middle region of KNL-2 (residues 267–470; predicted to be unstructured) and the predicted α-helical region of the N-tail of CENP-AHCP-3 (Figure 4B); an interaction between KNL-2 and CENP-AHCP-3 was also reported in a large-scale two-hybrid screen of proteins essential for C. elegans embryogenesis (Boxem ). This interaction was not observed with full-length KNL-2 but, as no interaction has been observed with this fusion, this may be a false negative due to the full-length protein not being properly expressed/folded in yeast; a similar caveat also applies to the SANTA domain- and C-terminal KNL-2 fragments that did not show any interaction. Importantly, an interaction between similar regions of CENP-AHCP-3 and KNL-2 was also observed for the Caenorhabditis briggsae proteins (Figure 4B), despite primary sequence divergence (23.7% identity/43.5% similarity for CENP-AHCP-3 N-tail and 42.9% identity/55.1% similarity for KNL-2 middle region; Supplemental Figure S2B). The interaction was species specific, as the C. briggsae CENP-AHCP-3 N-tail did not interact with C. elegans KNL-2 middle region and vice versa (Figure 4B).

FIGURE 4:

The N-tail of CENP-AHCP-3 interacts with an unstructured middle region of KNL-2. (A) Domain structure of KNL-2. The presence of the SANTA domain (PF09133) and the Myb domain (also known as the SANT domain; PF00249) is conserved among the KNL-2/M18BP1 protein family. No secondary structure elements are predicted in the middle region of KNL-2. (B) Yeast two-hybrid analysis of CENP-AHCP-3 N-tail and KNL-2. The bait CENP-AHCP-3 N-tail fusions are listed on top and the prey KNL-2 fusions are listed on the left. (C–E) Biochemical analysis of the CENP-AHCP-3-KNL-2 interaction. Nickel-immobilized recombinant CENP-AHCP-3(1–109)-MBP-His6 was used to pull down indicated reticulocyte lysate-expressed S35-labeled MBP-KNL-2 fragments. In C, S35-autoradiogram (top) shows Input (I) and bead-bound (B) KNL-2 fragments; Coomassie staining (bottom) shows input lysate and CENP-AHCP-3(1–109)-MBP-His6 bait. In D, fragments of the 301–500 amino acid region of KNL-2 tested for binding to CENP-AHCP-3(1–109) are schematized on the left. S35-autoradiogram (top) shows reticulocyte lysate-expressed MBP-KNL-2 fragments; S35-autoradiogram (middle) shows bound KNL-2 fragments; Coomassie staining (bottom) shows CENP-AHCP-3 N-tail bait. In E, indicated KNL-2 fragments were tested for binding to control (MBP-His6) and CENP-AHCP-3(1–109)-MBP-His6 baits. S35-autoradiogram (top) shows input and bound fragments; Coomassie staining (bottom) shows input lysates and baits.

To confirm the two-hybrid interaction, we performed pull-down assays with a purified MBP-His6 fusion of the CENP-AHCP-3 N-tail immobilized on nickel agarose and in vitro–translated MBP-KNL-2 fragments. We first screened a series of overlapping fragments of KNL-2 and found that, consistent with the yeast two-hybrid results, a fragment containing residues 301–500 interacted with the N-tail (Figure 4C); we note that this KNL-2 fragment exhibited variable background interaction with the MBP-only control (Figure 4E); however, the interaction with CENP-AHCP-3 N-tail MBP fusion was consistently above background. We next analyzed a series of truncated fragments in this region and found that a central region of 50 amino acids (376–425) was essential for the interaction (Figure 4, D and E). However, this 50–amino acid region on its own did not interact with the CENP-AHCP-3 N-tail (Figure 4E), suggesting that residues on either side are important for the observed interaction. We conclude that the predicted α-helical region of the N-tail of CENP-AHCP-3 that is important for chromatin loading in vivo interacts directly with the CENP-AHCP-3 loading factor KNL-2 in vitro and that this interaction is conserved in a related nematode species with a significantly diverged N-tail sequence.

DISCUSSION

Here, we investigated how CENP-A chromatin is assembled in the absence of a HJURP/Scm3 family protein in C. elegans. We found that the unusually long N-tail of C. elegans CENP-A, which contains a predicted α-helical region, is required for CENP-A loading onto chromatin. By contrast, the divergent N-tails of CENP-A family members have been proposed to contribute to kinetochore assembly and to epigenetic stability of centromeric chromatin (Ravi ; Fachinetti ; Folco ), but have not been implicated in assembly of CENP-A chromatin. By comparing deletion of the predicted structured region of the CENP-A N-tail to two other perturbations that prevent kinetochore assembly, we show that the absence of N-tail-mutant CENP-AHCP-3 on chromatin is due to a failure in loading and not a consequence of defective kinetochore assembly. In addition, as the N-tail-mutant CENP-AHCP-3 localizes normally in the presence of endogenous CENP-AHCP-3, the absence of localization cannot be attributed to misfolding or inability to interact with histone H4. Thus, the N-tail effectively acts as an intramolecular-targeting signal, analogous to the CATD in HJURP/Scm3-containing species.

CENP-AHCP-3 is loaded de novo in the one-cell C. elegans embryo (Gassmann ). Thus, whether the N-tail-KNL-2 interaction reported here continues to be required to load CENP-AHCP-3 after de novo loading in the first embryonic division is an important question. A recent study reported that embryos homozygous mutant for N-tail deleted CENP-AHCP-3, derived from a heterozygous parent and thus having maternally loaded full-length CENP-AHCP-3, developed into adults; by contrast, embryos homozygous null mutant for CENP-AHCP-3 derived from a heterozygous parent did not (Prosée ). This observation indicates that the N-tail-deleted CENP-AHCP-3 supports development following depletion of maternally loaded full-length CENP-AHCP-3, suggesting that the N-tail may not be essential for loading later in embryogenesis. However, an alternative possibility is raised by the observation that N-tail deleted CENP-AHCP-3 localizes to centromeres in the presence of full-length CENP-AHCP-3, likely due to the presence of 2 CENP-As in octameric centromeric nucleosomes. Thus, N-tail-deleted CENP-AHCP-3 may stabilize/enhance the functionality of maternally loaded full-length CENP-AHCP-3, and thereby lead to a less severe developmental phenotype. Development of acute means of removing maternally loaded CENP-AHCP-3 after the initial de novo loading in the early embryo will be necessary to address whether the N-tail is required for loading throughout embryogenesis or whether its role in loading is restricted to early embryonic divisions.

HJURP/Scm3 has two roles: one is to act as a chaperone promoting assembly of CENP-A nucleosomes, and the second is to target this assembly reaction to centromeric chromatin through an interaction with centromere recognition factors (Ndc10 in budding yeast, Mis18 in fission yeast, KNL-2 in C. elegans and plants, and Mis18 complex in vertebrates). Through two-hybrid and in vitro biochemical assays, we provide evidence that the C. elegans CENP-A N-tail possesses the latter activity; it interacts directly with the middle region of KNL-2 and this interaction is preserved in a species-specific manner in C. briggsae, despite significant primary sequence divergence (especially in the N-tail sequence). A recent study independently described a CENP-AHCP-3 N-tail–KNL-2 interaction that is consistent with what we report here (Prosée ). To date, we have been unable to selectively mutate this interaction and assess the consequences in vivo; this will be important to pursue in the future. An obvious question emerging from our results is whether the N-tail of CENP-AHCP-3 also exhibits chaperone activity, analogous to HJURP/Scm3. In preliminary work, we have not observed an interaction between the N-tail and the histone fold of CENP-AHCP-3 in two-hybrid and in vitro binding assays, which would argue against the presence of chaperone activity. However, significantly more effort needs to be placed on reconstitutions with purified components to address whether these negative results are indeed due to an absence of chaperone activity. RNAi experiments have implicated the C. elegans orthologue of the histone-binding WD40 domain chaperone RbAp46/48, LIN-53, in CENP-AHCP-3 chromatin assembly (Lee ), suggesting that it may work together with the CENP-AHCP-3 N-tail–KNL-2 interaction described here to assemble centromeric chromatin.

In conclusion, we provide evidence for an evolutionary variation on CENP-A chromatin assembly in which the N-tail of CENP-A has acquired part of the function of the specialized chaperone/targeting factor HJURP/Scm3 and become essential for loading into chromatin. This represents a distinct solution from Drosophila, which also lacks HJURP/Scm3 but appears to have convergently evolved an HJURP/Scm3-like chaperone called Cal1 (Erhardt ; Phansalkar ; Chen ; Medina-Pritchard ). Understanding how different species build CENP-A chromatin at restricted genomic locations should provide insight into the general principles by which the epigenetic state of centromeric chromatin is defined and propagated.

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

C. elegans strains

C. elegans strains (genotypes in Supplemental Table S1) were maintained at 20°C (Edgley ). Engineered GFP::CENP-AHCP-3 transgenes were cloned into pCFJ151 and injected into strain EG4322. The KNL-1::mCherry transgene was cloned into pCFJ178 and injected into EG6700 (Frokjaer-Jensen ). The amplified hcp-3 genomic locus was flanked on the 5′ end by 5′-GACGACGCTCCGAATCATTTGGGAG-3′ and on the 3′ end by 5′-CTATTTGTCAAATAATAAAGATTCATTACTTGTAAATGAGAACATTTTATTTAA-3′. For the GFP:: CENP-AHCP-3 transgenes the GFP sequence was inserted following the start codon and preceded by a GGRAGSGGRAGSGGRAGS linker. All exons were reencoded in CENP-A to allow RNAi-mediated depletion of endogenous CENP-A without affecting the introduced transgene. Single-copy insertion was confirmed by PCR. Transgenic strains were crossed into various marker or deletion strains using standard genetic procedures.

RNA-mediated interference (RNAi)

Double-stranded RNAs were generated using oligos (Supplemental Table S2) to amplify regions from N2 genomic DNA or cDNA. PCR reactions were used as templates for in vitro RNA production (Ambion), and the RNA was purified using a MegaClear kit (Ambion). Eluted RNA from the T3 and T7 reactions were mixed together, combined with 3× soaking buffer (32.7 mM Na2HPO4, 16.5 mM KH2PO4, 6.3 mM KCl, 14.1 mM NH4Cl), and annealed (68°C for 10 min, 37°C for 30 min). dsRNA was injected into L3/L4 hermaphrodite worms 38–42 h before imaging. For double depletions dsRNAs were mixed in equal amounts (≥1–3 mg/ml for each RNA).

Immunoblotting

For immunoblotting a mixed population of worms growing at 20°C on an NGM+OP50 agar plate were collected with M9+0.1% Triton X-100, pelleted, and washed. Worms were vortexed in a mix of 100 μl M9+0.1% Triton X-100, 50 μl 4× sample buffer, and 100 μl glass beads and boiled, then vortexed and boiled again. Samples were run on an SDS–PAGE gel, transferred to a polyvinylidene fluoride membrane, probed with 1 µg/ml affinity-purified anti-CENP-AHCP-3 (rabbit; antigen was CENP-AHCP-3[105-183]::6xHis) and detected using an horseradish peroxidase–conjugated secondary antibody (rabbit or mouse; GE Healthcare). For antibody production CENP-AHCP-3(105–183)::6xHis was expressed in Escherichia coli, purified, and injected into rabbits (Covance). Serum was affinity purified on a HiTrap NHS column to which CENP-AHCP-3(105–183)::6xHis was covalently coupled.

Yeast two-hybrid screens

Yeast two-hybrid analysis was performed according to the manufacturer guidelines (Matchmaker; Clontech Laboratories). Genes of interest were cloned from WT (N2) C. elegans or C. briggsae cDNA.

Imaging and quantification

For live imaging of one-cell embryos, gravid hermaphrodite adult worms were dissected into M9 buffer, embryos were manually transferred to 2% agarose pads, and overlaid with a coverslip. For all experiments, except the one shown in Figure 3D, a 5 × 2 μm z-series was collected every 10–15 s using an inverted Zeiss Axio Observer Z1 system with a Yokogawa spinning-disk confocal head (CSU-X1), a 63×, 1.4 NA Plan Apochromat objective, and a QuantEM 512SC EMCCD camera (Photometrics). For the experiment shown in Figure 3D, 6 × 2 μm z-series were acquired every 10 s using an Andor Revolution XD Confocal System (Andor Technology), with a CSU-10 spinning-disk confocal scanner unit (Yokogawa) mounted on a TE2000-E inverted microscope (Nikon), 60×, 1.4 NA Plan Apochromat lenses, and a back-thinned EMCCD camera (iXon, Andor Technology). Environmental temperature during image acquisition was 19°–20°C.

All images and movies were processed, scaled, and analyzed using ImageJ (Fiji) and Photoshop (Adobe). Quantification of CENP-AHCP-3 and KNL-1 kinetochore localization during metaphase of one-cell embryos was performed on maximum intensity projections. A rectangle was drawn around the fluorescence signal and average pixel intensity was measured. The rectangle was expanded on all sides and the difference in integrated intensity between the expanded rectangle and the original rectangle was used to define the background intensity per pixel. Integrated fluorescence was then calculated for the original rectangle after background subtraction (Moyle ).

Protein purification

CENP-AHCP-3(1–109)-MBP-6xHis was cloned into pET21a. HCP-3(1–109)-MBP-6xHis and MBP::6xHis were expressed in BL21(DE3). E. coli cultures were grown to OD600 0.6–0.8 and induced with 0.1 mM IPTG (isopropyl-β-d-thiogalactoside) for 6 h at 20°C. Induced BL21(DE3) cells were lysed in lysis buffer (20 mM Tris, pH 7.5, 300 mM NaCl, 20 mM imidazole, 8 mM β-mercaptoethanol [BME]) and clarified at 40,000 × g for 45 min at 4°C. Ni-NTA agarose (Qiagen) was incubated with clarified lysates for 45 min, washed with wash buffer (20 mM Tris, pH 7.5, 300 mM NaCl, 50 mM imidazole, 8 mM BME), and eluted with 20 mM Tris, pH 7.5, 300 mM NaCl, 300 mM imidazole, 8 mM BME. The eluted protein was fractionated using a Superose10 gel filtration column (GE Healthcare). Protein concentrations were determined using a NanoDrop 1000 spectrophotometer (Thermo Scientific).

In vitro translation and binding assay

KNL-2-MBP fragments were [35]S labeled using a TnT Quick Coupled Transcription/Translation System (Promega). In vitro translation lysate (10 µl) was incubated for 1 h at 4°C with 50 µg CENP-AHCP-3(1–109)-MBP-His (in 20 mM Tris, pH7.5, 300 mM NaCl, 0.05% NP40, 10 mM imidazole) in a final volume of 50 µl, mixed with 25 µl of a 1:1 nickel agarose slurry equilibrated with the binding buffer for an additional hour at 4°C. Beads were washed three times with binding buffer, eluted using sample buffer, and the elution analyzed by SDS–PAGE and autoradiography.

Click here for additional data file.