Structural basis of nucleosome deubiquitination by the bidentate Calypso/Asx complex

TL;DR

The study reveals how a bidentate Calypso/Asx complex interacts with and removes ubiquitin from nucleosomes, enabling processive deubiquitination along chromatin.

Contribution

The cryo-EM structure of the bidentate Calypso/Asx complex bound to a nucleosome is presented, revealing asymmetric binding and a spreading mechanism.

Findings

Cryo-EM shows asymmetric binding of the bidentate Calypso/Asx complex to ubiquitinated nucleosomes.

The positively charged C terminus of Calypso is crucial for chromatin engagement and spreading.

The bidentate complex enables processive deubiquitination along chromatin via alternating engagement.

Abstract

The Polycomb repressive complex 1 (PRC1) and PR-DUB constitute a canonical pair of histone-modifying enzymes that deposit and remove monoubiquitinated H2A at lysine 119 (H2AK119ub1), serving as a model of dynamic epigenetic regulation. In humans, PR-DUB, composed of BAP1 and ASXL1, functions as a monomeric complex, while the Drosophila homolog Calypso/Asx forms a bidentate dimer (Calypso2: Asx2) with an unclear chromatin engagement mechanism. Here, we present its cryo-EM structure bound to a nucleosome, revealing the molecular basis of interaction. Surprisingly, only one Calypso/Asx unit engages the nucleosome in a conformation similar to human BAP1/ASXL1, while the second remains disengaged. Structural and biochemical analysis of the positively charged Calypso C terminus suggests a “spreading” potential of the bidentate complex along chromatin, which was validated in vitro using nucleosome arrays. These findings support a model in which the bidentate Calypso/Asx complex enables processive deubiquitination along chromatin via alternating or cooperative engagement.

•Cryo-EM reveals asymmetric bidentate Calypso/Asx binding to ubiquitinated nucleosome•The positively charged tail of Calypso is a key determinant of chromatin engagement•A bidentate Calypso/Asx complex enables spreading along chromatin

Cryo-EM reveals asymmetric bidentate Calypso/Asx binding to ubiquitinated nucleosome

The positively charged tail of Calypso is a key determinant of chromatin engagement

A bidentate Calypso/Asx complex enables spreading along chromatin

Natural sciences; Biological sciences; Structural biology

Figures

Calypso/Asx binds to the nucleosome substrate in a 2:2 bidentate form(A) Domain organization of Calypso and Asx. Domain names and residue numbers marking their boundaries are labeled. The underlines denote the fragments used in this study.(B) Coomassie-stained SDS-PAGE analysis of PR-DUB complex. Molecular weight marker values are labeled on the left.(C) Western blot detection of deubiquitination by the Calypso/Asx (1–340) complex using nucleosome substrates, monoubiquitinated at H2AK119. The top and bottom blots used anti-H2AK119ub1 and anti-H3 antibodies, respectively. H3 lanes were run on t

The bidentate Calypso/Asx architecture enables efficient deubiquitination along the chromatin fiber(A) Fluorescence polarization measurements of the binding affinities of the indicated wild-type and mutant Calypso/Asx (1–340) complexes (designated as dimer and monomer) to wild-type nucleosomes and H2AC119ub1-modified nucleosomes. Dissociation constants (Kd) were obtained by fitting the data to a specific binding model with a variable Hill slope using GraphPad Prism 9 (*n* = 3 independent biological experiments; data are presented as mean ± SEM).(B) Fluorescence polarization measurements of the

The working model of the bidentate Calypso/Asx complex on the chromatin fiber, along with a comparison to BAP1(A) A cartoon model illustrating the proposed bidentate Calypso/Asx deubiquitination mechanism on chromatin. The different copies of Calypso/Asx are color-coded as in the previous figures. The red curve represents the positively charged tail of Calypso. The red plus sign represents the positive charge carried by positively charged amino acids, while the blue minus sign represents the negative charge on nucleosome DNA.(B) Domain organization of Calypso and BAP1. Domain names and residue

Full text

Introduction

H2A monoubiquitination (H2Aub1) is a highly conserved histone mark across eukaryotes, occurring at lysine 119 in vertebrates, K118 in Drosophila, and K121 in Arabidopsis.1^,^2 It is catalyzed by Polycomb repressive complex 1 (PRC1), whereas its removal is mainly carried out by the Polycomb repressive deubiquitinase (PR-DUB) complex.3 Studies in both Drosophila and mammals have demonstrated that the antagonistic interplay between PRC1 and PR-DUB is essential for ensuring the spatial precision, reversibility, and long-term stability of Polycomb-mediated gene repression.4^,^5 Loss of PR-DUB deubiquitinase activity results in a genome-wide increase of H2AK119ub1, including at loci that are not bona fide Polycomb targets.6^,^7 Overall, PR-DUB maintains Polycomb occupancy at target gene loci and ensures transcriptional repression by removing H2AK119ub1 from inappropriate sites.5

PR-DUB is a complex composed of the catalytic subunit BAP1 in humans (Calypso in Drosophila)3^,^8 and the regulatory subunit ASXL1/2/3 in humans (Asx in Drosophila).3^,^9^,^10^,^11 The core deubiquitinase subunits BAP1 and Calypso are highly conserved, each containing a UCH (ubiquitin C-terminal hydrolase) domain that serves as the catalytic domain and shares over 70% sequence similarity, including the catalytic triad (Cys-His-Asp) and the ubiquitin-binding site.12 The deubiquitinase activity of BAP1 and Calypso is activated by binding to the DEUBAD (deubiquitinase adapter domain) of their respective regulatory subunits, ASXL1/2/3 in mammals, or Asx in Drosophila.3^,^12 In mammals, BAP1 participates in a wide range of biological processes by forming distinct complexes with various cofactors such as FOXK1/2,13 HCF-1,14 and UBE2O,15 These interactions establish a multilayered regulatory network involved in tumor suppression,16^,^17 DNA damage repair,18 and metabolic control.19 By contrast, the Drosophila PR-DUB is primarily dedicated to regulating gene expression, participating in development and cell fate determination.3^,^20^,^21

The crystal structure of the Drosophila PR-DUB complex revealed a distinct 2:2 bidentate configuration, in which two Calypso subunits associate with two Asx subunits to form a dimeric assembly.22 Functional assays demonstrated that this bidentate architecture is critical for enzymatic activity, as disrupting the dimerization markedly impairs the complex’s ability to deubiquitinate nucleosomal substrates.22 In contrast, the cryo-EM structure of the human BAP1/ASXL1 complex bound to a monoubiquitinated nucleosome (NCP-ub), resolved at high resolution, revealed a 1:1 stoichiometry.23^,^24 In this configuration, BAP1 engages the nucleosome in a monomeric form, with its catalytic site precisely positioned to access the H2AK119ub1 modification. Despite their high conservation, BAP1/ASXL1 and Calypso/Asx adopt distinct oligomeric states and modes of nucleosome engagement. The structural and functional implications involved remain to be further elucidated. Understanding how this dimeric architecture contributes to chromatin recognition and catalytic activity is essential for reconciling the apparent discrepancy in PR-DUB mechanisms across species.

To bridge this mechanistic gap, we determined the cryo-EM structure of the Calypso/Asx complex bound to a NCP-ub, revealing its substrate-engaged conformation. We found that only one copy appears to be functionally active. Based on the overall impact of the positively charged C-terminal tail of Calypso and its spatial positioning, we speculated that it likely acts across nucleosomes and proposed a spreading model on chromatin. Finally, functional assays using 4∗NCP and 1∗NCP substrates in vitro validated the spreading mechanism of the bidentate Calypso/Asx complex along the chromatin fiber. These findings not only clarify a long-standing question regarding the function of the dimeric Calypso/Asx complex but also provide a structural framework for understanding the dynamic balance between ubiquitination and deubiquitination in Polycomb-mediated gene repression.

Results

Calypso/Asx binds to the nucleosome substrate in a 2:2 bidentate form

To reconstitute a catalytically active Drosophila PR-DUB complex, we co-expressed full-length Calypso and a truncated form of Asx (1–340) in insect cells (Figures 1A, 1B and S1A). This truncated Asx fragment retains the DEUBAD domain required for Calypso activation and has been shown to confer deubiquitinase activity toward H2AK119ub1 on nucleosomes without hydrolyzing H2Bub.3 To validate the enzymatic activity of the recombinant complex, we performed a deubiquitination assay using nucleosomes bearing H2AK119ub1 as substrate (Figures S2A and S2B). As expected, the Calypso/Asx (1–340) complex efficiently removed ubiquitin from H2AK119ub1 nucleosomes (Figure 1C), confirming its catalytic competence. For structural analysis, we prepared a chemically defined, homogeneous H2AK119ub1 nucleosome by site-specific conjugation of ubiquitin (G76C) to histone H2AK119C via a nonhydrolyzable 1,3-dichloroacetone linkage, yielding a stable mimic of the native ubiquitinated nucleosome substrates25 (Figures S1B, S1C, S2C, and S2D). We first examined the Calypso/Asx/NCP-ub ternary complex obtained using this crosslinking method by negative-stain electron microscopy, confirming the stability and homogeneity of the sample (Figure S1D). Subsequently, we applied the Grafix method to crosslink the same sample and isolated the most homogeneous peak fraction (Figure S1E). We assembled the Calypso/Asx (1–340) complex with this ubiquitinated nucleosome and subjected the ternary complex to single-particle cryo-electron microscopy. Cryo-EM reconstruction yielded a Calypso/Asx/Ubiquitin portion at ∼6.7 Å resolution and a nucleosome substrate map at 3.29 Å resolution, respectively (Figures S3A–S3E, and Table S1). We found that the Calypso/Asx portion undergoes continuous rotation relative to the nucleosome, which increased the difficulty of structural determination (Figure S3C). Therefore, we performed 3D variability analysis on the Calypso/Asx/Ub region, which clearly revealed its rotational states on the nucleosome (Video S1). Although the resolution of the complex was relatively low, the previously determined crystal structure of the Calypso/Asx complex allowed reliable docking into the obtained cryo-EM density map (Figure 1D). Upon fitting the first Calypso/Asx complex, we observed an additional density positioned between the nucleosome and the Calypso/Asx complex, which could be well accommodated by a ubiquitin (Figures 1D and 1E). We therefore interpret this density as the ubiquitin conjugated to H2A, in agreement with the observations reported in BAP1/ASXL1/NCP-H2AK119ub123 (Figure 2A), indicating a conserved catalytic mechanism between species. Furthermore, we detected an extra density distal to the nucleosome that could be readily fitted with a second Calypso molecule together with part of the Asx structure. Based on these observations, we hypothesize that Calypso/Asx engages the nucleosome as an asymmetric dimer of heterodimers, consistent with a 2:2 stoichiometry, in which one Calypso/Asx complex directly interacts with the ubiquitin modification (validated below) (Figures 1D, 1E, and 2A).Figure 1. Calypso/Asx binds to the nucleosome substrate in a 2:2 bidentate form(A) Domain organization of Calypso and Asx. Domain names and residue numbers marking their boundaries are labeled. The underlines denote the fragments used in this study.(B) Coomassie-stained SDS-PAGE analysis of PR-DUB complex. Molecular weight marker values are labeled on the left.(C) Western blot detection of deubiquitination by the Calypso/Asx (1–340) complex using nucleosome substrates, monoubiquitinated at H2AK119. The top and bottom blots used anti-H2AK119ub1 and anti-H3 antibodies, respectively. H3 lanes were run on the same gel as sample-processing controls.(D) The published Calypso/Asx structural model and nucleosome model (PDB:6CGA; PDB:8H1T; PDB:7OHC) were docked into the map, with two views of the composed density map of the Calypso/Asx/NCP-ub complex. The color code is used throughout.(E) Atomic model of the complex structure and a schematic diagram.Figure 2. The second Calypso/Asx heterodimer facilitates efficient nucleosome processing(A) Comparison of the electron density maps of Calypso/Asx/NCP-ub and BAP1/ASXL1/NCP-ub, with the nucleosome positioned at the same angle. The resolution of BAP1/ASXL1/NCP-ub has been low-pass filtered.(B) Previously proposed ubiquitin-binding site and key residues are shown (doi, https://doi.org/10.1038/s41467-018-06186-1).(C) A cartoon representation illustrates that, building upon the established binding mode of Calypso-1/Asx-1 with nucleosome-ub-1, a second nucleosome-ub-2 was modeled onto the catalytic center of Calypso-2/Asx-2 using the binding configuration observed for Calypso-1/Asx-1 with nucleosome-ub-1. This modeling results in pronounced steric clashes in the nucleosomal region (indicated by red dashed lines). Subunit coloring is maintained consistently with the previous figures.(D) A close-up view showing the key amino acids of the two Calypso coiled-coil hairpins, with the model referenced from PDB:6CGA.(E) SEC-MALS analysis of wild-type Calypso/Asx (1–340) (dimer) and mutant Calypso/Asx (1–340) (monomer), with the following mutations in Calypso: M288R, N292R, L340R. The gel filtration column is Superdex 200 5/150.(F) Michaelis-Menten analysis of wild-type and mutant Calypso/Asx proteins cleavage of ubiquitin-AMC. Error bars indicate ±SEM (n = 3 independent experiments).(G) Activity assays comparing the ability of wild-type (dimer), mutant (monomer) Calypso/Asx complexes to remove ubiquitin from nucleosomes mono-ubiquitinated at H2AK119. The H3 band was run on the same gel as a control.

Video S1. Three-dimensional variability analysis of the Calypso/Asx-nucleosome complex, related to Figure S3C

The docked Calypso/Asx model revealed that the complex engages the nucleosome asymmetrically, with only one Calypso/Asx heterodimer (designated Calypso-1/Asx-1) adopting a catalytically competent configuration, while the second heterodimer (Calypso-2/Asx-2) is positioned peripherally. The catalytic domain of Calypso-1 is oriented directly over the ubiquitinated H2A (Figures 1D and 1E), which is consistent with the previously proposed model of the Calypso/Asx/ubiquitin (Figure 2B).12^,^22 This arrangement places the catalytic triad of Calypso-1 in position to access the isopeptide bond between H2AK119 and ubiquitin. In contrast, Calypso-2/Asx-2 is oriented away from H2A-ub and does not engage the second H2A-ub modification within a single nucleosome simultaneously (Figures 1D and 1E). Structural modeling further indicates that when Calypso-1/Asx-1 is engaged with the nucleosome and ubiquitin, Calypso-2/Asx-2 cannot simultaneously accommodate another ubiquitinated nucleosome, as dual engagement of a second nucleosome is sterically unfavorable (Figure 2C). Together, these observations indicate that the second Calypso/Asx molecule is structurally disengaged and may serve a non-catalytic or regulatory function. The functional significance of this architectural asymmetry remains to be elucidated.

The second Calypso/Asx heterodimer facilitates efficient nucleosome processing

We next sought to explore the overall impact of disrupting Calypso/Asx dimerization. We designed a mutant version of Calypso in which three key residues required for dimerization were replaced with arginine (M288R/N292R/L340R), disrupting the dimerization interface without affecting the catalytic domain (Figure 2D).22 The resulting Calypso (M288R/N292R/L340R)/Asx (1–340) complex was purified using the same protocol as the wild-type complex. SDS-PAGE and SEC-MALS analysis confirmed that dimerization was abolished, with the mutant complex existing predominantly in a monomeric form (Figures 2E and S2E).

We compared the catalytic activity of the monomeric and dimeric Calypso/Asx complexes using two complementary assays. First, an Ub-AMC fluorescence assay was performed to determine whether disruption of dimerization affected the intrinsic activity of the catalytic site. Both complexes exhibited comparable Km values (Figure 2F), indicating that the monomeric form retains the same basal catalytic capacity as the wild-type dimer. Second, deubiquitination activity was measured using nucleosomes modified with H2AK119ub1, to assess catalytic efficiency within a chromatin context. Under this condition, the monomeric complex showed a marked reduction in activity compared to the wild-type dimer (Figures 2G, S2A and S2B). These results are consistent with previous studies, which suggested that the higher activity of the dimeric complex arises from increased substrate affinity.22 This observation prompted further investigation into the contribution of the second Calypso/Asx heterodimer to substrate recognition.

The C-terminal DNA-binding tail of Calypso contributes almost entirely to its nucleosome-binding affinity

Fluorescence polarization assays comparing monomeric and dimeric Calypso/Asx complexes revealed that the dimeric form exhibits markedly higher binding affinity toward nucleosomes (Figure 3A). Notably, both monomeric and dimeric Calypso/Asx complexes associated with wild-type and H2AC119ub1-modified nucleosomes in a similar affinity, indicating that ubiquitin recognition contributes minimally to the overall affinity of PR-DUB to nucleosomes. Given this observation, we asked whether additional regions within Asx might contribute to enhanced affinity. We compared Calypso/Asx (1–340) with a longer construct, Calypso/Asx (1–821), in binding assays against ubiquitinated nucleosomes. The results showed no appreciable difference in affinity between the two complexes (Figure 3B), suggesting that the extended region of Asx outside the DEUBAD domain plays little to no role in nucleosome binding under these conditions.Figure 3. The bidentate Calypso/Asx architecture enables efficient deubiquitination along the chromatin fiber(A) Fluorescence polarization measurements of the binding affinities of the indicated wild-type and mutant Calypso/Asx (1–340) complexes (designated as dimer and monomer) to wild-type nucleosomes and H2AC119ub1-modified nucleosomes. Dissociation constants (Kd) were obtained by fitting the data to a specific binding model with a variable Hill slope using GraphPad Prism 9 (n = 3 independent biological experiments; data are presented as mean ± SEM).(B) Fluorescence polarization measurements of the binding affinities of the indicated wild-type Calypso/Asx (1–340) and Calypso/Asx (1–821) to H2AC119ub1-modified nucleosomes, respectively.(C) A close-up view of the positively charged tail amino acids at the C terminus of Calypso.(D) Fluorescence polarization measurements of the binding affinity of different truncations of the positively charged C-terminal tail of Calypso to H2AC119ub1-modified nucleosomes, respectively.(E) Fluorescence polarization measurements of the binding affinity of different truncations of the positively charged C-terminal tail of Calypso to a random 20-bp DNA sequence fragment, respectively.(F) Michaelis-Menten analysis of ubiquitin-AMC cleavage by Calypso (C-terminal truncation)/Asx complex. Error bars indicate ±SEM (n = 3 independent experiments).(G) A close-up view of the position of the C-terminal flexible region (385–471 aa) of Calypso-2 in the structural model, indicated by a red dashed line.(H) Western blot was used to examine the deubiquitination rate of the dimeric Calypso/Asx complex on mono-nucleosome and 4∗nucleosome arrays. The H4 band was run on the same gel as a control.(I) Western blot was used to examine the deubiquitination rate of the monomeric Calypso/Asx complex on mono-nucleosome and 4∗nucleosome arrays. The H4 band was run on the same gel as a control.For (B, D, and E), fluorescence polarization data were quantified and analyzed in the same manner as described in (A).

We next turned our attention to the C-terminal tail of Calypso, which is predicted to be positively charged (Figure 3C) and was previously shown to be involved in DNA binding.12 We generated three truncated variants of Calypso: Calypso (1–466), Calypso (1–461), and Calypso (1–458)—reconstituted each variant with Asx (1–340). Fluorescence polarization assays demonstrated that Calypso (1–466) retained nucleosome binding affinity comparable to that of the wild-type complex, indicating that residues beyond position 466 have negligible impact on nucleosome binding affinity. In contrast, further truncations at residues 461 and 458 resulted in a sustained and pronounced decrease in binding affinity: compared with the wild-type, the 1–461 aa truncation showed an approximately 20-fold reduction, while the 1–458 aa truncation exhibited about a 290-fold reduction (Figure 3D). The same samples were further tested with DNA fragments to assess DNA-binding affinity, which yielded similar results (Figure 3E). Subsequent analysis of the ubiquitin hydrolysis activity of the C-terminally truncated mutant showed no substantial change compared with the full-length protein (Figures 2F and 3F). These findings indicate that the C-terminal fragment engages nucleosomal DNA through strong electrostatic interactions, independent of ubiquitin hydrolysis activity.

The bidentate Calypso/Asx architecture enables efficient deubiquitination along the chromatin fiber

Overall, the major contribution to nucleosome binding arises from the positively charged C-terminal segment of Calypso. Structural analysis further reveals that only one heterodimer within the 2:2 complex adopts a catalytically competent conformation, while the second heterodimer extends away from the nucleosome. Notably, the C-terminal tail of the disengaged Calypso subunit remains flexible and projects toward the solvent-exposed DNA, placing it in proximity to both the currently engaged nucleosome and potential neighboring chromatin regions (Figure 3G). Moreover, the continuous rotation of the Calypso/Asx portion relative to the nucleosome observed in our previous structural analysis is likely to confer a greater range of motion to the C-terminus of Calypso copy-2 (Figure S3C and Video S1). This spatial arrangement may serve as an adaptation to the polynucleosome environment along the chromatin fiber.

To test our hypothesis, we reconstituted in vitro mononucleosomes and nucleosome arrays (4∗NCP) carrying the H2AK119ub1 modification (Figure S2F) and compared the deubiquitination rates of dimeric and monomeric Calypso/Asx under these different substrate contexts. The nucleosome concentration and the level of H2AK119ub1 modification were carefully matched across all reaction mixtures. In the case of the dimeric complex, deubiquitination was significantly faster on 4∗NCP arrays than on mononucleosomes (Figure 3H). In contrast, the monomeric complex showed no enhancement in activity on the 4∗NCP arrays, and its overall reaction rate was markedly reduced (Figure 3I). These results reveal the potential of the bidentate Calypso/Asx complex to mediate cross-nucleosome deubiquitination on chromatin, consistent with our earlier hypothesis. In addition, we considered that extending the linker of the 4∗NCP to 40 bp would theoretically increase the difficulty for the positively charged C-terminal tail of Calypso-2 to engage. Consequently, the deubiquitination rate of the bidentate Calypso/Asx on 4∗NCP (40-bp linker) would be significantly reduced. As expected, the experimental results show that the deubiquitination rate on 4∗NCP (40-bp linker) is significantly lower compared to 4∗NCP (20-bp linker) (Figure S2G). This provides additional support for the plausibility of the bidentate Calypso/Asx “spreading” model on chromatin proposed in our study.

The ability to efficiently carry out deubiquitination along the chromatin chain appears to be essential for Calypso/Asx complex. In Drosophila studies, the H2AK118ub1 modification (homologous to H2AK119ub1 in humans) is found throughout the genome, not just at Polycomb target genes.7^,^26 Drosophila PR-DUB needs to regulate the global level of H2AK118ub1 to maintain repression at Polycomb sites. Given the abundance of H2AK118ub1 across the genome, it is reasonable that PR-DUB employs a spreading mechanism along chromatin to efficiently achieve genome-wide deubiquitination. This dynamic interplay may be particularly important during developmental transitions or in response to environmental cues.

Discussion

The Polycomb repression system regulates gene expression through epigenetic modifications, forming a dynamic regulatory network that ensures cells maintain their specific identity and function during development. This regulation is crucial for cell differentiation and developmental processes. The modification of H2AK119ub1 serves as a pivotal “switch” within this system, and the addition and removal of ubiquitin at this site have long been a major focus of research. Our study focuses on the deubiquitination mechanism of H2AK119ub1 by the bidentate Drosophila Calypso/Asx complex. Here, we have resolved the structure of the ternary complex comprising the bidentate Calypso/Asx and the H2AC119ub1-modified nucleosome, providing insights into how Calypso/Asx interacts with its nucleosome substrate. Based on this structural framework, we compared the binding mode of the human BAP1/ASXL1/NCP complex, assessed the deubiquitination efficiency of Calypso/Asx in dimeric and monomeric mutant forms, measured binding affinities under various conditions, and evaluated the impact of the positively charged C-terminal tail of Calypso on nucleosome affinity. From these results, we propose a potential working model for the bidentate Calypso/Asx complex on chromatin (Figure 4A). In this model, the second copy of Calypso/Asx recognizes the N+1 nucleosome DNA through the positively charged C-terminal amino acids of Calypso. Following the completion of H2AK119ub1 hydrolysis, the entire complex rapidly transitions to the N+1 nucleosome via the positively charged tail of the second Calypso copy, enabling Calypso/Asx to move efficiently along the chromatin fiber, much like the left and right feet of a human “walking” along the chromatin fiber.Figure 4. The working model of the bidentate Calypso/Asx complex on the chromatin fiber, along with a comparison to BAP1(A) A cartoon model illustrating the proposed bidentate Calypso/Asx deubiquitination mechanism on chromatin. The different copies of Calypso/Asx are color-coded as in the previous figures. The red curve represents the positively charged tail of Calypso. The red plus sign represents the positive charge carried by positively charged amino acids, while the blue minus sign represents the negative charge on nucleosome DNA.(B) Domain organization of Calypso and BAP1. Domain names and residue numbers marking their boundaries are labeled. The number of amino acids inserted in the coiled-coil regions of Calypso and BAP1 is additionally marked in red.(C) Schematic diagram of the domains and binding partners of Drosophila and human PR-DUB complexes. The illustration is adapted from the article Nat Commun 11, 5947 (2020).

This model closely aligns with the proposed “read” and “erase” mechanisms of PRC1 across nucleosomes on chromatin. Studies have shown that the ncPRC1^RYBP^ complex exhibits the ability to recognize inherited “old” H2AK119ub1 through the NZF domain of RYBP, thereby allowing ncPRC1 to deposit new H2AK119ub1 either on the opposite side of the same nucleosome (intra-nucleosome) or on a neighboring nucleosome (inter-nucleosome).27^,^28 Our study further visualizes, at the molecular level, how PR-DUB and PRC1 functionally antagonize each other on chromatin to ensure precise Polycomb-mediated gene repression.5^,^29 Additionally, structural studies of PRC2 and nucleosomes have shown that PRC2 forms dimers and, through its ability to bind across nucleosomes, can compact dispersed chromatin.30 These results reveal the “inter-nucleosome” capabilities of Polycomb (PcG) proteins and their underlying mechanisms, indicating that the strongly related polycomb deubiquitinase complex, Calypso/Asx, is also likely to possess inter-nucleosome activity.

Building on this, we compared the structural, functional, and mechanistic differences between BAP1 and Calypso. Sequence alignment between BAP1 and Calypso revealed a major difference within the coiled-coil domain. While Calypso contains a short 29-amino-acid insertion, BAP1 harbors a much larger 365-amino-acid insertion in the corresponding region (Figure 4B). This extensive insertion in BAP1 provides binding sites for several interaction partners, including HCF-1,14^,^31 FOXK1/2^13^, and OGT32 (Figure 4C). The structural and functional versatility of BAP1 complexes is closely associated with this large insertion. However, this is likely to require the BAP1/ASXL1 complex to possess higher substrate specificity, rather than enhancing its chromatin deubiquitination capacity through a bidentate configuration as observed in Calypso/Asx.

Limitations of the study

Here, we compare the three-dimensional structures of the Drosophila Calypso/Asx/NCP-ub complex and the human BAP1/ASXL1/NCP-ub complex. Our comparison clearly indicates that the Calypso/Asx complex is present as two copies bound to the nucleosome, occupying a similar overall location. However, the resolution of our Calypso/Asx/NCP-ub reconstruction is relatively low and does not allow for accurate interpretation of the structural details. Therefore, we can only propose, based on complementary biochemical evidence, a working hypothesis regarding the biological significance of this two-copy assembly.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Jun He (he_jun@gibh.ac.cn).

Materials availability

This study did not generate new unique reagents. All materials generated during this study are available from the lead contact upon request.

Data and code availability

• •Original western blot images have been deposited at Mendeley at https://doi.org/10.17632/nctzwh3cp7.1 and are publicly available as of the date of publication.
• •Microscopy data reported in this study will be shared by the lead contact upon request.
• •The cryo-EM density maps and corresponding atomic coordinates have been deposited in the Electron Microscopy DataBank (EMDB) and Protein DataBank (PDB) under accession numbers EMD-64748 and PDB-9V33 for the Calypso/Asx/NCP-ub complex.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We thank the cryo-EM Center, Guangzhou Institutes of Biomedicine, and Health Chinese Academy of Sciences and Advanced Bio-imaging Technology Platform of Guangzhou Laboratory for cryo-EM beamtime and all staff members for their assistance with data collection.

This work was supported by grants from the 10.13039/501100001809National Natural Science Foundation of China (10.13039/501100001809NSFC) (32361163669, 32241021 to J.H.) and partially supported by 10.13039/501100012245Science and Technology Planning Project of Guangdong Province, China (2023B1212060050, 2023B1212120009), as well as the Science and Technology Program of Guangzhou, China (2024A04J6348). J.H. acknowledges start-up grants from the 10.13039/501100002367Chinese Academy of Sciences. The 10.13039/501100002858China Postdoctoral Science Foundation (CPSF) (2023M743512 to F.S.), 10.13039/501100010031Postdoctoral Research Foundation of China (GZC20232691 to F.S.), and the 10.13039/501100021171Guangdong Basic and Applied Basic Research Foundation (2023A1515110923 to F.S.).

Author contributions

J.H. and C.W. designed the project. C.W. constructed the expression plasmids and prepared protein complexes involved in the research. C.W., F.S., J.G., N.Z., Y.Z., and W.S. prepared the nucleosomes involved in the research and reconstituted protein-nucleosome complexes for the cryo-EM studies. C.W. performed biochemical assays. C.W. conducted cryo-EM experiments and determined the structures. C.W. and H.Z. prepared the structures and figures. C.W., H.Z., and J.H. prepared the manuscript with input from all authors. J.H. supervised and directed the overall research.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesMouse monoclonal anti Strep-tag IIThermo FisherCat#MA5-37747Rabbit monoclonal anti Ubiquityl-Histone H2A (Lys119)SelleckCat#F0313Rabbit monoclonal anti Histone 3AbcamCat#Ab1791Rabbit monoclonal anti Histone 4AbclonalCat#A23000Bacterial and virus strainsDH5αInternal stockN/ABL21(DE3)-RosettaInternal stockN/ADH10BacInternal stockN/AChemicals, peptides, and recombinant proteinsUb-AMCJieXinBioCat#JX-DT02991,3-DichloroacetoneSigma-AldrichCat#168548DethiobiotinAladdinCat#R303913TCEPAladdinCat#T107252Page RulerThermo scientificCat#26616Isopropyl beta-D-thiogalactoside(IPTG)AladdinCat#I104812ChloramphenicolAladdinCat#C100333KanamycinSolarbioCat#K8020AmpicillinMacklinCat#A800429GlutaraldehydeSigma-AldrichCat#G7776Tris(hydroxymethyl)aminomethaneAladdinCat#T434093Sodium chlorideAladdinCat#C111533Potassium chlorideAladdinCat#P112133SF900 II mediumGibcoCat#12658-027FuGENE HDPromegaCat#E2311Critical commercial assaysPlasmid Miniprep KitTIANGENCat#DP103-03Gel Extraction KitCWBIOCat#CW2302MClonExpress° l One Step Cloning KitVazymeCat#C112-02Deposited dataCryo-EM structure Calypso/ASX/NCP complexThis paperPDB-9V33; EMD-64748Experimental models: Cell linesSf9ATCCCRL-1711High FiveInternal stockN/AOligonucleotidesPrimer: Forward: DNA-147-FMA 5′-FAM-ACAGGATGTATATATCTGACACG-3′AzentaN/APrimer: Forward: DNA-147 5′- ACAGGATGTATATATCTGACACG-3′AzentaN/APrimer: Reverse: DNA-147 5′- TATCGATGTATATATCTGAC -3′AzentaN/APrimer: Forward: DNA-187 5′-GGACCCTATACGCGGCCGCCCTGGAG-3′AzentaN/APrimer: Reverse: DNA-187 5′- GGTCGCTGTTCAATACATGCTATCGAT -3′AzentaN/ARecombinant DNAPlasmid: pFUGG-StrepII-Calypso/Asx(1-340aa)This paperN/APlasmid: pETDuet-1-H2A/H2B/H3/H4This paperN/APlasmid: pGEM-3z/601-AddgeneAddgene_26656Software and algorithmsGraphPad Prism 9.0.0Graphpadhttps://www.graphpad.com/ASTRA7Wyatthttps://www.astra7.de/enUCSF ChimeraXMeng et al.33https://www.cgl.ucsf.edu/chimerax/cryoSPARCPunjani et al.34https://cryosparc.com/PHENIXAfonine et al.35https://phenix-online.org/documentation/reference/validation_cryo_em.htmlRELIONScheres et al.36https://www3.mrc-lmb.cam.ac.uk/relion//index.php/Main_PageCootEmsley et al.37https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/EPUThermo Fisher ScientificN/AOtherStrepTrap columnCytivaCat#28913630Resource QCytivaCat#17117701HiTrap Heparin HP columnCytivaCat#17040703Superdex 200 Increase 3.2/300CytivaCat#28990946Superdex 200 Increase 10/300 GLCytivaCat#28990944HiLoad 16/600 Superdex 200 pgCytivaCat#28989335Superose 6 Increase 3.2/300CytivaCat#29091598Superdex 200 Increase 5/150CytivaCat#28990945

Experimental model and study participant details

Bacteria culture

A total of 1 μL Histone octamer–pETDuet-1-H2A/H2B/H3/H4 plasmid (approximately 100 ng) was added to 100 μL thawed E. coli Rosetta competent cells and gently mixed. The mixture was incubated on ice for 30 min and then heat-shocked in a 42°C water bath for 60 s. The cells were subsequently placed on ice for 2 min. In a biosafety cabinet, 800 μL LB medium was added, and the cells were allowed to recover at 37°C with shaking at 220 rpm for 1 h. The recovered culture was centrifuged at 3000 rpm for 5 min, 500 μL of the supernatant was discarded, and the pellet was resuspended. A 200 μL aliquot of the transformed cells was spread onto LB agar plates containing kanamycin and incubated upside down at 37°C overnight. The next day, a single colony was inoculated into 10 mL TB medium and grown for 6 h. The preculture was then inoculated into a 2 L baffled flask at a 1:1000 ratio and grown at 37°C with shaking at 180 rpm. After approximately 4 h, when the OD_600_ reached 0.4-0.8, protein expression was induced with 0.4 mM IPTG for 14 h. Cells were harvested by centrifugation at 5000 rpm for 10 min in a pre-chilled centrifuge (4°C), and the cell pellets were stored at -80°C until use.

Insect cell culture

The obtained bacmids coexpressing wild-type Calypso/Asx (1-340 aa) and the mutant Calypso (M288R, N292R, L340R)/Asx (1-340 aa)—in which a double-StrepII tag was engineered at the N-terminus of Calypso—were transfected into Sf9 cells to produce recombinant baculovirus. Sf9 cells (3 mL per well in a 6-well plate) were transfected with 1 μg bacmid DNA and 6.6 μL FuGENE HD (Promega, Cat#E2311) at 27°C and incubated for 96 h. The resulting virus-containing supernatant was collected and used to infect 40 mL Sf9 cells, which were cultured at 27°C and 130 rpm for 48 h, followed by harvesting of the supernatant to obtain the viral stock. A total of 10 mL of the amplified virus stock was added to 1 L of High Five cells and incubated at 27°C and 130 rpm for protein expression. When cell viability decreased to ∼80% (approximately 48 h), the cells were harvested by centrifugation at 1000 g for 10 min. The resulting cell pellets were stored at -80°C.

Method details

Preparation and purification of Calypso/Asx complex

The coding sequences of wild-type (WT) or mutant Calypso and Asx (1-340) were individually amplified by PCR and subsequently cloned into modified pFUGG vectors.33 To facilitate affinity purification, a double-StrepII tag was engineered at the N-terminus of Calypso. Recombinant viruses harboring both subunit genes were mixed and used to infect High Five insect cells, followed by 48 h co-expression period. Insect cells were harvested by centrifugation and lysed using a high-pressure homogenizer in buffer A (25 mM Tris-HCl, pH 7.5, 600 mM NaCl, 0.5 mM TCEP, 10% glycerol, 1 mM PMSF, protease inhibitor cocktail and Supernuclease). The resulting lysate was clarified by ultracentrifugation at 22,000 rpm for 1 hour. The protein complex was then captured on a StrepTrap column (Cytiva), washed, and eluted using buffer A supplemented with 2.5 mM desthiobiotin. For further purification, the eluted protein complex was subjected to size-exclusion chromatography on Superdex 200 16/600 column (Cytiva) using buffer B (25 mM HEPES, pH 7.5, 150 mM NaCl, and 0.5 mM TCEP). Peak fractions were pooled, concentrated, and aliquoted for storage at -80°C.

Preparation and purification of octamer-WT and octamer-H2AK119C

Octamer-WT and octamer-H2AK119C were prepared using the same method. The modified pETDuet-1 vector carrying Xenopus laevis histone genes enables the polycistronic co-expression of histones in their octameric form.34 Escherichia coli Rosetta (DE3) cells transformed with this vector were cultured to an OD600 of 0.4-0.5, at which point protein expression was induced with 0.4 mM IPTG. The culture was then incubated at 37°C with shaking at 170 rpm for an additional 20 hours. Following expression, bacterial cells were harvested and lysed in a buffer containing 20 mM Tris pH 8.0, 500 mM NaCl, 0.1 mM EDTA, and 0.5 mM TCEP. The clarified lysate was applied to a Heparin affinity column (Cytiva) and subjected to a stepwise wash with buffer containing 500 mM NaCl, followed by a gradual elution using a salt gradient ranging from 500 mM to 2 M NaCl. The collected samples were further subjected to HiLoad 16/600 Superdex 200 pg (Cytiva) in 2 M NaCl, 20 mM HEPES to remove residual nucleic acid contamination.

Expression and purification of ubiquitin-WT and ubiquitin G76C

Ubiquitin-WT and ubiquitin-G76C were prepared using the same method. The pGEX 6P-1 plasmid vector was engineered to express GST fusion protein, followed by the human rhinovirus 3C protease cleavage site. The plasmid containing the ubiquitin gene was transformed into BL21 (DE3) cells. The cells were grown in an LB medium containing 50 μg/mL kanamycin to OD600 of 1.5 and then induced by 0.4 mM IPTG overnight at 20°C. Cells were resuspended in lysis buffer (1 x PBS, pH 7.5, 1 mM PMSF, 5 mM DTT) and lysed by ultra high-pressure homogenizer. After centrifugation at 22,000 rpm for 60 min at 4°C. The supernatant was loaded onto a 10 mL GST resin pre-equilibrated in the GST buffer (1 x PBS, pH 7.5, 5 mM DTT) and further washed 20 column volumes (CV) by GST buffer. Then the 3C-GST was loaded onto GST resin to cleave the GST tag. SDS-PAGE was used to assess the extent of digestion after 24 h. The concentrated sample was then loaded onto a Superdex 75 16/600 column (Cytiva). The peak fractions were pooled, concentrated and flash-frozen in liquid nitrogen and stored in the refrigerator at -80°C for long-term storage.35

Cross-linking and purification of the octamer-H2AC119ub1(for Cryo-EM)

All reagents were cooled to 4°C before the reaction. In a typical reaction, H2AK119C histone octamer (100 μM, 14 mg) and 10 equivalents of ubiquitin (1 mM, 10 mg) were mixed and 10 equivalents of 1,3-dichloroacetone in DMF (500 mM, 0.58 μL) added and the solution incubated at 4°C for 3 hours. The reaction was quenched using 50 mM β-mercaptoethanol. After cross-linking, the reaction mixture was purified by gel filtration on a Superdex 200 column (Cytiva) to remove excess ubiquitin and histone octamers. The purified products were subsequently subjected to HisTrap affinity chromatography (Cytiva). The bound proteins were eluted by increasing the imidazole concentration from 0 mM to 500 mM linearly over 30 CV. Each fraction was analyzed by SDS-PAGE. The peak fractions were pooled according to SDS-PAGE analysis and concentrated. Then the protein was aliquoted and flash-frozen in liquid nitrogen and stored in the refrigerator at -80°C for long-term storage.25

Nucleosome reconstitution

Nucleosome-WT and nucleosome-H2AC119ub1 were prepared using the same method.36 The nucleosome assembly was performed using the “double bag” dialysis method. The DNA used for NCP-mono was based on the classic 601-147 sequence.37 For the 4∗NCP, the DNA consisted of four tandem repeats of the 601-147 sequence, each separated by a 20 bp random linker (20N20N20N20N20). DNA was first resuspended in a high-salt buffer containing 20 mM Tris pH 8.0, 2 M NaCl, 0.5 mM EDTA, and 0.5 mM TCEP. The histone octamer, prepared in the same high-salt buffer, was then added at a molar ratio of 1.1:1 (histone octamer to DNA). To facilitate controlled reconstitution, dialysis buttons (0.2 mL capacity) containing the histone octamer-DNA mixture were placed inside a dialysis bag filled with 50 mL of high-salt buffer. The bag was subsequently submerged in 1 L of dialysis buffer (20 mM Tris, pH 8.0, 1 M NaCl, 0.5 mM EDTA, 0.5 mM TCEP) and incubated overnight at 4°C. After 12 hours, the dialysis bag (still containing 50 mL of 1 M salt buffer) along with the dialysis buttons was transferred into 1 L of low-salt buffer (20 mM Tris pH 8.0, 50 mM NaCl, 0.5 mM EDTA, 0.5 mM TCEP) and dialyzed for an additional 5-6 hours. Finally, the dialysis buttons were subjected to a final dialysis step in the low-salt buffer for 3-4 hours to ensure complete nucleosome reconstitution.

Ubiquitin-AMC assays

The deubiquitinase activity of wild-type Calypso/Asx (1-340) (dimer), mutant Calypso (M288R; N292R; L340R)/Asx (1-340) (monomer) and Calypso-C^Δ^ was assessed by monitoring the release of fluorescent 7-amido-4-methylcoumarin (AMC) from the quenched Ubiquitin-AMC substrate, providing a direct measurement of enzymatic activity. Purified Ubiquitin-AMC powder was initially dissolved in pure DMSO to a concentration of 20 mg/mL and subsequently diluted with milliQ water to a final concentration of 0.5 mg/mL, ensuring that the residual DMSO concentration in the reaction not exceeding 2.5%. Reactions were conducted in an assay buffer containing 25 mM HEPES pH 7.5, 150 mM NaCl, 5 mM DTT, and 0.05% Tween-20 at 25°C. Fluorescence was measured every 30 seconds for 10 minutes with excitation and emission wavelengths of 380 nm and 460 nm, respectively. To determine kinetic parameters, 5 nM of the indicated Calypso/Asx (1-340) complexes were incubated with a range of Ubiquitin-AMC concentrations (0, 0.3, 0.625, 1.25, 2.5, 5, and 10 μM) under the same reaction conditions. Fluorescence intensity was converted to AMC concentration using a free AMC standard (Sigma-Aldrich). Initial reaction rates were determined from the linear portion of the reaction curve and plotted against substrate concentration. Kinetic parameters were derived by fitting the data to the Michaelis–Menten equation using non-linear regression analysis in Prism 9 (GraphPad).

Nucleosome DUB assays

Mono-ubiquitination of nucleosomes at Lys119 on histone H2A was achieved through an in vitro enzymatic reaction. Specifically, recombinant nucleosomes (400-800 nM) were incubated in an assay buffer (20 mM Tris-HCl pH 7.5, 50 mM NaCl, 5 mM ATP, 2 mM MgCl_2_, and 2 mM DTT) with 21 nM hUbe1, 150 nM UbcH5b, 40 nM Ring1B (1-159)/PCGF4, and 2 μM wild-type ubiquitin. The reaction was carried out at 25 °C for 2 hours with 100 μL system. The reaction mixture was subsequently subjected to gel filtration on a Superdex 200 10/300 column (Cytiva) to isolate nucleosomes, followed by western blot analysis to verify H2AK119ub1 modification using a rabbit monoclonal anti-ubiquityl-histone H2A (Lys119) antibody (Selleck). For deubiquitination activity assays, reactions were performed in a total volume of 10 μL at 25 °C. Assay conditions included 10nM of wild-type and mutant Calypso/Asx (1-340), along with 500 nM H2AK119ub1 nucleosomes. The reaction was terminated at 5 min, 20 min, and 40 min by the addition of SDS-PAGE loading buffer, the reaction mixtures were then separated by 15% SDS-PAGE and analyzed by western blot. In the comparative experiment of deubiquitination rates on nucleosome arrays, the concentrations of the two types of H2AK119ub1-modified nucleosomes (1∗NCP and 4∗NCP) were 500 nM and 125 nM, respectively. Wild-type and mutant Calypso/Asx (1-340) were applied at a concentration gradient of 5 nM, 10 nM, 15 nM, and 20 nM. After a 20-minute reaction at 25 °C, SDS-loading buffer was added and the samples were boiled at 98 °C for 5 minutes to terminate the reaction. The reaction mixtures were then separated by 15% SDS-PAGE and analyzed by western blot.

SEC-MALS analysis

80 μL PR-DUB sample (10 μM) was loaded onto a Superdex 200 5/150 column (Cytiva) with a buffer containing 20 mM Tris-HCl pH 7.5, 150 mM NaCl. A static light-scattering detector and a differential refractive index detector (Wyatt) were connected to the analytical gel filtration chromatography system. Data were analyzed with ASTRA7 provided by Wyatt.

Fluorescence polarization assay

Binding affinity was determined by incubating 20 nM of 5′-FAM-labeled nucleosomes, either unmodified or mono-ubiquitinated at H2AK119C, with a two-fold serial dilution of PR-DUB components (ranging from 1 to 0.000488 μM) in buffer containing 20 mM Tris pH 7.5, 100 mM NaCl, 1 mM EDTA, 2 mM DTT, and 5% glycerol, 0.01% NP-40, 0.01% CHAPS, and 0.1 mg/mL BSA. The mixtures were incubated at room temperature for 30 minutes, and 40 μL of each reaction was transferred to a 384-well low-volume black round-bottom polystyrene non-binding surface microplate (Corning). Fluorescence polarization was measured using a BioTek Synergy Neo2 plate reader equipped with a fluorescence filter (excitation: 485/20 nm, emission: 528/20 nm) and a polarizer filter. The gain was set to 50 for all experiments. Three technical replicates were performed for each independent experiment unless otherwise specified, and fluorescence polarization (mP) values from two scans were averaged. The dissociation constant (Kd) was determined using a specific binding model with the Hill slope in Prism 9 (GraphPad).38

Sample preparation of EM

The gradient master was used to prepare the gradient by mixing the top solution and bottom solution. The top solution (20 mM HEPES pH 7.5, 50 mM NaCl, 10% glycerol) and bottom solution (20 mM HEPES pH 7.5, 50 mM NaCl, 30% glycerol, 0.125% glutaraldehyde) were prepared. The protein complex samples (150 μL at ∼10 mg/mL) were added above the solution level of centrifuge tubes and spun down for 14 hours at 35,000rpm and 4°C in the SW41-Ti centrifuge rotor (Beckman). The independent fraction separator gently sucked out around 30 fractions from each tube.39 SDS-PAGE or native gels were used to detect these fractions. The cryo-EM grids (Quantifoil, Au, R1.2/R1.3 300 mesh) were treated in a glow discharge system (GloQube) and the cryo-EM samples were prepared by using the Vitrobot (Thermo Fisher Scientific). In the environment of 100% humidity and at 4°C, 3.5 μL samples were added to the grids, and the grids were blotted for 2 s with force 4 and then inserted into liquid ethane for quick freezing. The girds were screened or stored in liquid nitrogen.

Cryo-EM data collection and image processing

The datasets were collected by 300 kV Titan Krios G4 electron microscope (Thermo Fisher Scientific) and automatically collected using the EPU software. Movies were recorded by Falcon4 direct electron detector equipped with a SelectrisX energy filter with a 10 eV slit width. In electron event representation (EER) mode, movies were recorded at a nominal magnification of 165,000× with raw pixel size of 0.71 Å on the image plane. The movies were recorded in a –0.8 μm to −2.2 μm defocus range, with an electron dose rate of 26 e–/Å2/s and a total dose of 50 e–/Å2. All the EER movies were processed by CryoSPARC software,40 and CryoSPARC Live preprocessed the initial motion correction and CTF estimate. After several rounds of 2D classification, the average particles from the good class were submitted to Ab-Initio reconstruction. Particles containing Calypso/Asx/Ub density on the nucleosome were selected from the Ab-Initio reconstruction, first extracted with Bin 2, and then subjected to Non-Uniform refinement. Based on the Non-Uniform refined map, the nucleosome density was subtracted, followed by 3D classification. The particles showing relatively complete Calypso/Asx/Ub density were then subjected to local refinement. based on the local refinement map, an additional mask targeting the Calypso/Asx region was generated, followed by an additional round of 3D classification. Subsequent local refinement yielded a reconstruction at 7.14 Å resolution from 27,764 particles. Based on this map, the densities corresponding to Calypso-1/Asx-1/Ub and Calypso-2/Asx-2 were subtracted separately, followed by additional rounds of local refinement for each component. Subsequently, the corresponding particles were re-extracted, and the Calypso/Asx/Ub density was subtracted using a mask. Local refinement was then performed to calculate the nucleosome density. Finally, the three maps (nucleosome; Calypso-1/Asx-1/Ub; Calypso-2/Asx-2) were combined to generate the final map.41^,^42

Model building

For the model building of Calypso/Asx/NCP-ub complexes, the nucleosome models were based on the available crystal structure (PDB:7OHC). The ubiquitin model was built by fitting the previous structure (PDB:8H1T) and then manually edited in COOT.43 The Calypso/Asx models were built by fitting the model PDB:6CGA and then manually edited in COOT. All above models were subjected to PHENIX for several rounds of real-space refinement and obtained validation finally.44 UCSF ChimeraX were used for the generation of figures.45

Quantification and statistical analysis

Quantification and statistical analyses were performed as described below, and detailed statistical information, including exact values of n, is provided in the corresponding figure legends.

Deubiquitinase activity was quantified by monitoring the release of 7-amido-4-methylcoumarin (AMC) from Ubiquitin-AMC. Fluorescence intensity was converted to AMC concentration using a free AMC standard curve. Initial velocities were calculated from the linear phase of the reaction progress curves and plotted against substrate concentration. Kinetic parameters (Km and Vmax) were derived from nonlinear regression fitting to the Michaelis–Menten equation in GraphPad Prism 9. All Ub-AMC kinetic measurements were performed in three independent biological replicates.

Fluorescence polarization (FP)-based binding assays were quantified by measuring millipolarization (mP) values from 5′-FAM-labeled nucleosomes incubated with serially diluted PR-DUB complexes. For each reaction, two consecutive plate scans were acquired and averaged. Three independent biological replicates were performed for all FP experiments, each containing three technical replicates unless otherwise stated. Technical replicates were averaged prior to statistical analysis. Dissociation constants (Kd) were obtained by fitting the FP data to a specific binding model with a variable Hill slope in GraphPad Prism 9.

In all experiments, n represents the number of independent biological experiments performed with independently prepared samples. Quantitative data are presented as mean ± SEM, where the center value represents the mean and the dispersion measure represents the standard error of the mean. No hypothesis-testing–based statistical comparisons were performed unless otherwise indicated. No data points were excluded, and no statistical methods were used to predetermine sample size.