Structural analysis of fungal CENP-H/I/K homologs reveals a conserved assembly mechanism underlying proper chromosome alignment

Abstract

The kinetochore is a proteinaceous complex that is essential for proper chromosome segregation. As a core member of the inner kinetochore, defects of each subunit in the CENP-H/I/K complex cause dysfunction of kinetochore that leads to chromosome mis-segregation and cell death. However, how the CENP-H/I/K complex assembles and promotes kinetochore function are poorly understood. We here determined the crystal structures of CENP-I N-terminus alone from Chaetomium thermophilum and its complex with CENP-H/K from Thielavia terrestris, and verified the identified interactions. The structures and biochemical analyses show that CENP-H and CENP-K form a heterodimer through both N- and C-terminal interactions. CENP-I integrates into the CENP-H/K complex by binding to the C-terminus of CENP-H, leading to formation of the ternary complex in which CENP-H is sandwiched between CENP-K and CENP-I. Our sequence comparisons and mutational analyses showed that this architecture of the CENP-H/I/K complex is conserved in human. Mutating the binding interfaces of CENP-H for either CENP-K or CENP-I significantly reduced their localizations at centromeres and induced massive chromosome alignment defects during mitosis, suggesting that the identified interactions are critical for CENP-H/I/K complex assembly at the centromere and kinetochore function. Altogether, our findings unveil the evolutionarily conserved assembly mechanism of the CENP-H/I/K complex that is critical for proper chromosome alignment.

Figure 1.

In vitro reconstitution of CENP-H/I/K complex from thermophilic fungi. (A) Schematic view of thCENP-H/I/K, hsCENP-H/I/K and ctCENP-I proteins. The colored regions denote the domains revealed in the structures. The numbered regions represent the constructs used in this study. It has been identified previously that N-terminus of hsCENP-I binds to CENP-H/K, and C-terminus interacts with CENP-M (39). (B) SEC elution profiles of thCENP-H/K, ctCENP-I^NT and ternary complex, respectively, from top to bottom. Peak fractions were resolved with SDS-PAGE and stained with Coomassie Blue. (C) SV AUC analysis of ctCENP-I^NT (blue), thCENP-H/K (green) and ternary complex (red). The molecular weights were calculated based on sedimentation coefficient, obtained with SEDFIT software using a continuous c(s) model.

Figure 2.

Crystal structures of ctCENP-I^NT alone and its complex with thCENP-H^CT/K^CT. (A and B) Cartoon view of the crystal structure of ctCENP-I^NT alone in side (A) and top (B) orientations. Residues 198–205 between α10 and α11 not traceable in the electron density map are shown as dotted lines. (C and D) Cartoon view of the crystal structure of fungal CENP-H/I/K complex in side (C) and top (D) orientations. The thCENP-K^CT, thCENP-H^CT and ctCENP-I^NT are colored yellow, magenta and cyan, respectively. ( E) Electron density map for thCENP-K^CT in the complex structure. The composite omit map was calculated with Phenix, contoured at 1.0σ. There is no electron density for residues 192–267 between KH1 and KH2 of thCENP-K. (F) Overlay of the ribbon diagrams of the ctCENP-I^NT alone (orange) and its complex with thCENP-H^CT/K^CT (colored yellow, magenta and cyan, respectively). (G) SEC elution profiles of thCENP-H^N164/CENP-K^N143 complex (left) and its mixture with ctCENP-I^NT (right). Peak fractions were resolved with SDS-PAGE and stained with Coomassie Blue. (H) The N-termini of hsCENP-K only bound to the N-terminal fragment of hsCENP-H (left panel), the C-termini of hsCENP-K only bound to C-terminal fragment of hsCENP-H (right panel). GST-hsCENP-K^N85 and GST-hsCENP-K^ΔN85 were immobilized on beads and incubated with sumo-hsCENP-H^N120 or sumo-hsCENP-H^ΔN120. The bead-bound proteins were resolved with SDS-PAGE and stained with Coomassie Blue.

Figure 3.

Validate the C-terminal binding interface between CENP-H and CENP-K. (A) Zoomed-in view of the binding interface between thCENP-H^CT and thCENP-K^CT in the crystal structure of ternary complex. The interacting residues were drawn in stick and marked in number. (B) The relative binding activity of thCENP-H (WT and mutants) bound to GST-thCENP-K^FL and GST-thCENP-K^ΔN75 was assessed using GST pull-down assays. Error bars represent standard deviations, which were obtained from three independent experiments. The representative results of the pull-down assays were also available in Supplementary Figure S6B and C. (C) Alignment of the C-terminus sequences of CENP-H orthologs across species using Clustal Omega Program. The conserved residues were colored. The interacting residues identified in our structures were marked with yellow dots (binding to thCENP-K^CT) and green dots (binding to ctCENP-I^NT). ( D) The relative binding activity of hsCENP-H (WT and mutants) bound to GST-hsCENP-K^FL and GST-hsCENP-K^ΔN85 was assessed using GST pull-down assays. Error bars represent standard deviations, which were obtained from three independent experiments. The representative results of the pull-down assays were also available in Supplementary Figure S7A and B.

Figure 4.

Validate the binding interface between CENP-H and CENP-I, and the assembly mode of CENP-H/I/K ternary complex. ( A) Zoomed-in view of the binding interface between thCENP-H^CT and ctCENP-I^NT. The interacting residues were drawn in stick and marked in number. (B) The relative binding activity of thCENP-H (WT and mutants) bound to GST-ctCENP-I^NT was assessed using GST pull-down assays. Error bars represent standard deviations, which were obtained from three independent experiments. The representative results of the pull-down assays were also available in Supplementary Figure S8A. (C) In vitro binding assays were performed to examine the binding activity among the ternary complex. Proteins with distinct combinations as indicated were mixed and GST pull-down assays were performed. The bead-bound proteins were resolved with SDS-PAGE and stained with Coomassie Blue. (D) In vitro binding assays were performed to assessing the effects of thCENP-H mutants on the formation of the fungal CENP-H/I/K ternary complex. Proteins with distinct combinations as indicated were mixed and GST pull-down assays were performed. The bead-bound proteins were resolved with SDS-PAGE and stained with Coomassie Blue. (E) Quantification of the relative binding activity between ctCENP-I^NT and thCENP-H (WT and mutants) in the presence of thCENP-K based on the results from lanes 5, 6 and 7 in (D). (F) Zoomed-in view of the binding interfaces of thCENP-H^CT in the complex structure. The interacting resides in thCENP-H were drawn in stick and marked with numbers in red. And the counterpart residues in hsCENP-H were shown with numbers in black. ( G) HeLa Tet-On cells transfected with GFP-hsCENP-H (WT or mutants), non-tagged hsCENP-K and MYC-hsCENP-I. Cell lysates were treated with anti-GFP antibody. Immunoprecipitated samples were resolved with SDS-PAGE and blotted with the indicated antibodies. ( H) Quantification of the relative binding activity between hsCENP-I and hsCENP-H WT or mutants (upper panel), and between hsCENP-K and hsCENP-H WT or mutants (lower panel) based on the results from Figure 4G.

Figure 5.

Mutating the interface residues of CENP-H abolished its centromeric localization and induced massive chromosome mis-alignment. (A and B) Representative images of interphase HeLa Tet-On cells treated with Luciferase and hsCENP-H siRNA. The centromeric localization of endogenous hsCENP-H (A) or hsCENP-I (B) was detected by indicated antibody. (C) Quantification of the centromeric intensities of hsCENP-H in (A) and hsCENP-I in (B) normalized to the ones of ACA (CREST). At least 10 cells (20 kinetochores per cell) were quantified for each condition. Mean ± SD (standard deviation) was shown here. (D) HeLa Tet-On cells transfected with RNAi-resistant GFP-hsCENP-I WT, L219AL233A and K234E/L238A were treated with hsCENP-H siRNA. Cells were briefly treated with MG132 before subjected to staining. Representative images of mitotic cells were shown here. (E and F) Quantification of the centromeric intensities of GFP-hsCENP-H (E) and hsCENP-I (F) normalized to the ones of ACA (CREST) kinetochore signal. At least 10 cells (20 kinetochores per cell) were quantified for each condition. Mean ± SD was shown here. (G) Quantification of mitotic cells with chromosome alignment defects for the experiments in (A, B and D). At least 50 cells were counted for each condition. Mean ± SD was shown here. (H) The assembly mode of CENP-H/I/K ternary complex among the CCAN. CENP-A directly recruits CENP-C and CENP-L/N to the centromere (25,26,29,32,43,44). Both CENP-C and CENP-L/N are required for kinetochore recruitment of CENP-/H//I/K/M (32,43). CENP-C directly interacts with CENP-H/K, not CENP-I (32). How CENP-L/N interacts with CENP-H/I/K/M remains unclear. CENP-H/K form a heterodimer through both N-termini and C-termini, and the C-termini of CENP-H/K heterodimer binds to N-terminus of CENP-I. CENP-M directly interacts with the C-terminus of CENP-I and integrate into the CENP-H/I/K/M complex (39).