HeLa cells obtained from ATCC were cultured in Dulbecco’s Modified Eagle Medium (DMEM) from Invitrogen (11965092, Carlsbad, CA, USA), enriched with 1% penicillin/streptomycin and 10% fetal bovine serum (FBS) sourced from Gibco (10091148, Auckland, New Zealand), and were kept at maintained at 37 °C with 5% CO₂. HeLa cells stably expressing hemagglutinin (HA)-ATRX (wild type (WT) or mutants) were selected and kept with 1.0 µg/mL or 0.5 µg/mL puromycin (NSC-3055, TargetMol, Shanghai, China). Clones 2A4 and 3A2 of HP1 DKO cells were previously described [18]. All cell lines were authenticated before use via short tandem repeat (STR) profiling to ensure their authenticity and reliability.　Additionally, all cell lines were confirmed to be mycoplasma-free using the GMyc-PCR Mycoplasma Detection Kit (40601, YEASEN, Shanghai, China).

GST (glutathione S-transferase)-HP1$\alpha$, GST-HP1$\alpha$-I165E, GST-HP1$\alpha$-W174A, HP1$\alpha$-Flag-6xHis, HP1$\beta$-Flag-6xHis, HP1$\gamma$-Flag-6xHis, and pBos-centromere protein B (CB)-HP1$\alpha$-GFP were previously described [18]. HA-ATRX was kindly provided by professor Fangwei Wang (Zhejiang University). GST-HP1$\beta$, GST-HP1$\gamma$, GST-ATRX-L1 (561-610aa), GST-ATRX-L2 (1951-2000aa), GST-ATRX-L3 (171-220aa), GST-ATRX-L4 (1601-1650aa) were generated by inserting polymerase chain reaction (PCR) fragments into the pGEX-4T1-GST vector. HA-ATRX-$\mathrm{\Delta }$P1 and HA-ATRX-$\mathrm{\Delta }$PxVxL were created using Mut Express MultiS Fast Mutagenesis Kit (C215-01, Vazyme, Nanjing, China). Every plasmid was sequenced to ensure that no unwanted changes existed and that only desired mutations were found.

siATRX duplexes and control small interfering RNA (siRNA) (RiboBio, Guangzhou, China): siATRX#1 (5^′-GAGGAAACCUUCAAUUGUAUU-3^′), siATRX#2 (5^′-GUGGGCUGAAGAAUUUAAUdTdT-3^′), siWapl (5^′-CGGACUACCCUUAGCACAAdTdT-3^′). Plasmid transfections employed Polyethyleneimine (19850, Polysciences, Warrington, PA, USA) or Fugene 6 (E2691, Promega, Madison, WI, USA), siRNA transfections employed Oligofectamine (12252011, Invitrogen, USA) and Lipofectamine RNAi MAX (13778150, Invitrogen, Carlsbad, CA, USA). 1.0 µM nocodazole (R17934, MCE, Monmouth Junction, NJ, USA), 10 µM MG132 (HY-13259, MCE, Monmouth Junction, NJ, USA) are used in this study. Selective detachment with “shake-off” facilitated the collection of mitotic cells.

Primary antibodies: rabbit polyclonal antibodies against ATRX (HPA001906, Sigma, St. Louis, MO, USA), Wapl (A300-268A, Bethyl, Montgomery, TX, USA), GFP (A11122, Invitrogen, USA), ACTB (PAB13193, abnova, Taiwan, China). Mouse monoclonal antibodies against HP1$\alpha$ (MAB3584 for immunostaining; MAB3446 for immunoblotting, Millipore, Bedford, MA, USA), HP1$\beta$ (MAB3448, Millipore, USA), HP1$\gamma$ (MAB3450, Millipore, USA), HA (16B12, Biolegend, Beijing, China), Flag-tag (M2, Sigma, USA), GAPDH (YM3029, Immunoway, Plano, TX, USA). Guinea pigs polyclonal antibodies against centromere protein C (CENP-C) (PD030, MBL, Beijing, China).

Secondary antibodies: goat anti-rabbit IgG-HRP or horse anti-mouse (7074P2 or 7076P2, CST, Boston, MA, USA). Anti-mouse IgG-Alexa Fluor 488 or 555 (A-21202 or A-31570, Invitrogen, USA), anti-rabbit IgG-Alexa Fluor 488 or 555 (A-21206 or A-31572, Invitrogen, USA), anti-guinea pig IgG-Alexa Fluor 555 (A-21435, Invitrogen, USA), all for immunostaining.

Single-guide RNA (sgRNA) was procured from Qingke (Beijing, China), then annealed and inserted into pX330 or pX462 plasmid from Addgene (110403 or 48141, Boston, MA, USA). After plasmids transfected HeLa cells for 48 hours, cells were individually split to establish clonal cell lines, puromycin was used to the selection process. HP1 triple knockout (TKO) clone H32-21 was obtained with sgRNA sequences, 5^′-ACGTGTAGTGAATGGGAAAGtgg-3^′ target to HP1$\gamma$, 5^′-CTTTGCCCTTTACCACTCGAcgg-3^′ and 5^′-GGAGTACCTCCTAAAGTGGAagg-3^′ target to HP1$\beta$, 5^′-AGAACTGTCAGCTGTCCGCTtgg-3^′ and 5^′-GGAGGAGTATGTTGTGGAGAagg-3^′ target to HP1$\alpha$. For H23-19 clone, 5^′-CTTTGCCCTTTACCACTCGAcgg-3^′ and 5^′-GGAGTACCTCCTAAAGTGGAagg-3^′ target to HP1$\beta$, 5^′-TTTTCCACGACAAATTCTTCagg-3^′ and 5^′-CTAGATCGACGTGTAGTGAAtgg-3^′ target to HP1$\gamma$, 5^′-AGAACTGTCAGCTGTCCGCTtgg-3^′ and 5^′-GGAGGAGTATGTTGTGGAGAagg-3^′ target to HP1$\alpha$. HP1 knock out clones were identified by immunostaining and immunoblotting. pBluescript II KS (-) was used to subclone genomic DNA PCR fragments, which were then transformed into E. coli. To confirm the gene disruption, a select number of bacterial colonies were subsequently sequenced.

Cells were first fixed in 2% Paraformaldehyde (PFA, E672002, Sangon, Shanghai, China) in PBS for a duration of 20 minutes, then treated with 0.5% Triton X-100 (A110694, Sangon, Shanghai, China) in PBS for 5 minutes. For shake-off procedure, HeLa cells were re-attached to glass coverslips at 1500 rpm for 5 minutes by Cytospin. For chromosome spreads, mitotic cells were incubated in 75 mM KCl (A610440, Sangon, Shanghai, China) for 10 minutes, then reattaching them to glass coverslips using Cytospin, then fixed them like before. After fixation, cells were treated with 3% BSA (V900933, Sigma, USA) in PBS at room temperature for two hours with primary antibodies and one hour with secondary antibodies. Using DAPI (C1005, Beyotime, Shanghai, China), DNA staining was done for 5 minutes.

A fluorescent microscope (DM3000, Leica, Shanghai, China) was used to capture pictures. The centromere marker CENP-C was used to measure the inter-kinetochore (KT) distance, where over 20 kinetochores per cell were analyzed across a minimum of 15 cells. The distance was computed using the Leica LAS Live Measurement imaging software (Leica, Wetzlar, Germany). Images acquired under consistent illumination conditions, and fluorescent intensity was quantified of using ImageJ (NIH, Bethesda, Maryland, USA). Specifically, within chromosome spreads, the average pixel intensity for ATRX and HP1$\alpha$ staining at the inner centromeres, identified as areas within a circle of 3 pixels in diameter surrounding paired centromeres, was evaluated. The intensity of CENP-C was determined within a 9-pixel diameter circle that included individual centromere spots. The ratio of ATRX/anti-centromere antibody (ACA), ATRX/CENP-C, HP1$\alpha$/CENP-C, and HA-ATRX/CENP-C intensity was calculated for each centromere after background correction. Cohesion loss was defined as the separation of over 20% of sister-chromatid pairs within a single cell. Image editing was performed using Adobe Photoshop (Adobe, San Jose, CA, USA) and Illustrator (Adobe, USA). GraphPad Prism 10 (GraphPad Software, San Diego, CA, USA) was employed to do statistical analysis using unpaired t-tests.

Cell lysates were generated using either standard SDS sample buffer (B548118, Sangon, Shanghai, China) or P150 buffer intended for immunoblotting procedures. For co-immunoprecipitation purposes, cells underwent lysis in P150 buffer supplemented with 25 mM Tris-HCl (pH 7.5) (B548124, Sangon, Shanghai, China), 2 mM MgCl₂ (B601194, Sangon, Shanghai, China), 150 mM NaCl (A610476, Sangon, Shanghai, China), 1 mM dithiothreitol (DTT) (A620058, sangon, Shanghai, China), cocktail (P1005, Beyotime, Shanghai, China), 10 mM NaF (A500850, Sangon, Shanghai, China), 1 mM phenylmethanesulfonyl fluoride (PMSF) (ST505, Beyotime, Shanghai, China), 0.1 µM okadaic acid (S1786, Beyotime, Shanghai, China) and 20 mM $\beta$-glycerophosphate (ST637, Beyotime, Shanghai, China). After high-speed centrifugation, GammaBind G Sepharose (28411301, GE Healthcare, Chicago, IL, USA) was used for lysates preclear. The lysates were then mixed with antibodies for 3 hours at 4 °C, followed by the addition of GammaBind G Sepharose for an additional hour. After being washed three times with the P150 lysis buffer, the beads were boiled with standard SDS sample buffer prior to immunoblotting analysis.

Plasmids that encode proteins fused to GST were introduced into BL21 cells (TSV-A09, Qingke, Beijing, China). These cells were cultivated until the optical density at 600 nm reached between 0.6 and 0.7. Protein expression was then initiated by adding 0.1 to 0.4 mM isopropyl $\beta$-D-thiogalactoside (IPTG, A100487, Sangon, Shanghai, China) and incubating at 16 °C for 16 hours. Following this, the cells were disrupted using sonication in a buffer solution containing 20 mM Tris-HCl (pH 8.0), 1 mM EDTA (B540625, Sangon, Shanghai, China), 100 mM NaCl, and 1% Triton X-100. The lysate was clarified through centrifugation and subsequently mixed with Glutathione Sepharose 4B (17075601, NEB, cytiva, Washington DC, USA) in the lysis buffer. The resin was then washed with lysis buffer.

For the pulldown of ATRX from lysates using GST-HP1$\alpha$ (WT or mutants), mitotic HeLa cells expressing exogenous ATRX and arrested with nocodazole were lysed in a P150 buffer. The lysates were precleared with glutathione Sepharose 4B beads, followed by incubation with GST fusion proteins attached to the beads for 2 hours at 4 °C. The beads underwent three washes with the same buffer before being analyzed by immunoblotting or Coomassie Brilliant Blue (CBB) staining.

In mammals, HP1 comprises three HP1 isoforms, HP1$\alpha$, HP1$\beta$, and HP1$\gamma$. Upon mitotic entrance, most of HP1 proteins are shifted from chromosome arms, while a fraction remains localized at the heterochromatic centromere region [39, 40]. Our prior investigations revealed that both HP1$\alpha$ and HP1$\gamma$ localize at the inner centromere, where they serve redundant functions in safeguarding centromeric cohesion in human cells [18]. According to immunofluorescence microscopy, ATRX localizes to heterochromatin during interphase and subsequently relocates to centromeres in mitotic HeLa cells. Notably, ATRX colocalizes with HP1$\alpha$ during both mitosis and interphase (Fig. 1A). Interestingly, in HP1$\alpha$ and HP1$\gamma$ double knockout (HP1 DKO) cell lines, the presence of centromeric ATRX was diminished by 30%–40%, with residual localization observed at the centromere (Fig. 1B,C).

Fig. 1.

HP1 promotes centromeric localization of ATRX. (A) Asynchronous HeLa cells were immunostained for interphase cells. Mitotic cells were collected by treated with nocodazole for 3 hours, then underwent cytospin for immunostaining. (B,C) Cells were treated with nocodazole for a duration of 3 hours, after which the shake-off mitotic cells were subjected to immunostaining. The relative enrichment of ATRX was measured in across 10 cells (B) (analyzed using an unpaired t-test). Example images are shown in (C). (D) Immunoblot analysis was performed on asynchronous HeLa and the relevant HP1-TKO clones. (E,F) The stable cell lines specified underwent treatment with nocodazole for a duration of 3 hours, after which the shake-off mitotic cells were subjected to immunostaining. The relative enrichment of ATRX was measured in across 10 cells (E) (analyzed using an unpaired t-test). Representative images are supplied in (F). (G,H) HeLa cells were transiently transfected with either centromere pro- tein B (CB)-GFP or CB-HP1$\alpha$-GFP, and the shake-off mitotic cells arrested with nocodazole were subjected to immunostaining. The ratio of centromeric ATRX/GFP immunofluorescence intensity was determined in across 10 different cells (G) (analyzed using an unpaired t-test), and example images are given (H). Data details: The means and standard deviations (SDs) are presented (B,E,G). Scale bars: 10 µm. For further details, refer to Supplementary Fig. 2. ATRX, $\alpha$ thalassemia/mental retardation syndrome X-linked; HP1, heterochromatin protein 1; TKO, triple-knockout; ACA, anti-centromere antibody; GFP, green fluorescent protein.

Using CRISPR-associated protein 9 (CRISPR/Cas9) system [41, 42, 43], we generated two stable HP1 triple-knockout (HP1 TKO) clones, denoted as H23-19 and H23-21, in which different single-guide RNAs (sgRNAs) were used to individually knock out HP1$\alpha$, HP1$\beta$, and HP1$\gamma$ [44]. Immunoblotting analysis has confirmed the depletion of the HP1 protein (Fig. 1D). Immunofluorescence microscopy revealed a more pronounced decrease in mitotic centromeric ATRX in HP1 TKO cells than in HP1 DKO cells, resulting in severe elimination of centromeric localization (Fig. 1E,F). In contrast to HeLa cells, ATRX localization at heterochromatin almost completely disappeared throughout interphase in HP1 TKO cells (Supplementary Fig. 1A). Additionally, endogenous (Fig. 1G,H) and exogenous (Supplementary Fig. 1B,C) ATRX were recruited to the centromere by transient expression of CB-HP1$\alpha$-GFP but not CB-GFP. Furthermore, the knockout of HP1$\alpha$, HP1$\beta$, or HP1$\gamma$ alone did not compromise the centromeric localization of ATRX (Supplementary Fig. 1D,E). Meanwhile, the total ATRX protein level remained unaltered in both the HP1 DKO and TKO cell lines (Supplementary Fig. 1F).

These findings indicate that the localization of ATRX at inner centromeres during mitosis and heterochromatin foci in interphase relies on the presence of HP1$\alpha$, HP1$\beta$, and HP1$\gamma$. Importantly, ATRX knockdown did not affect the localization of HP1 proteins at centromeres (Supplementary Fig. 1G–I).

Given that ATRX is unable to localize to the centromere in HP1 TKO cells, an interaction between HP1 and ATRX may be implied. In mitotic HeLa cell lysates, we showed that GST-HP1$\alpha$ can effectively pull down endogenous ATRX (Fig. 2A). Furthermore, in nocodazole-arrested mitotic 293T cell lysates, a reciprocal co-immunoprecipitation (co-IP) was observed between HP1$\alpha$-Flag-6xHis, which was expressed exogenously, and HA-ATRX (Fig. 2B).

Fig. 2.

The N-terminal end of ATRX interacts with HP1$\alpha$, HP1$\beta$, and HP1$\gamma$. (A) Nocodazole-arrested mitotic HeLa cell lysates were pulled down using GST or GST-HP1$\alpha$, and then Coomassie Brilliant Blue staining and anti-ATRX immunoblotting were performed. (B) Lysates of 293T cells arrested in mitosis by nocodazole and transiently expressing HP1$\alpha$-Flag-6xHis and HA-ATRX were immunoprecipitated using anti-HA antibodies, then analyzed by immunoblotting. (C) Lysates from nocodazole-arrested mitotic HeLa cells transiently expressing HA-ATRX and HP1-Flag-6xHis (including HP1$\alpha$, HP1$\beta$, and HP1$\gamma$) were immunoprecipitated using anti-Flag antibodies, which was subsequently followed by immunoblotting with either anti-FLAG or anti-HA antibodies. (D) Mitotic HeLa cell lysates that were arrested with nocodazole underwent immunoprecipitation using anti-ATRX antibodies, and the corresponding antibodies were then used for immunoblotting. (E) Lysates from mitosis HeLa cells, which were treated with nocodazole and transiently expressing SFB-ATRX, were subjected to pulldown assays. This was followed by immunoblotting and staining with Coomassie Brilliant Blue. (F) Lysates from HeLa cells arrested in mitosis with nocodazole were processed through pulldown assays. Following this, immunoblotting and Coomassie Brilliant Blue staining were conducted. (G) Lysates from 293T cells arrested in mitosis with nocodazole and transiently expressing HP1$\alpha$-Flag and HA-ATRX (both wild type and $\mathrm{\Delta }$P1 mutant) were subjected to immunoprecipitation. This was followed by immunoblotting. (H) Lysates from HeLa cells were transiently expressing HP1$\alpha$-Flag and HA-ATRX (including both wild type and $\mathrm{\Delta }$PxVxL mutant) underwent immunoprecipitation. This was followed by immunoblotting. (I) Lysates derived from 293T cells arrested in mitosis with nocodazole were transiently expressing HA-ATRX (either wild type or mutants) were subjected to pulldown assays. Following this, Coomassie Brilliant Blue staining and immunoblotting were conducted. Data information: Please refer to Supplementary Fig. 2 for additional details. GST, Glutathione S-transferase; CBB, Coomassie Brilliant Blue; IP, Immunoprecipitation; HA, hemagglutinin; WT, wild type.

We subsequently investigated the interaction of ATRX with various HP1 isoforms (HP1$\alpha$, HP1$\beta$, and HP1$\gamma$) in mammalian cells. In mitotic HeLa cell lysates, we showed that HP1$\alpha$, HP1$\beta$, and HP1$\gamma$were linked to both exogenous and endogenous ATRX through co-immunoprecipitation and GST-pulldown assays (Fig. 2C–E). Conversely, endogenous ATRX is hardly to be extracted from mitotic cell lysates by the GST-HP1$\alpha$ mutants I165E or W174A (Fig. 2F).

The impairment of ATRX in the co-immunoprecipitation with the HP1$\alpha$-W174A mutant strongly suggests that ATRX may harbor a PxVxL motif that engages with the hydrophobic pocket of the CSD dimer. Earlier research indicated that HP1$\alpha$ has interactions with the ATRX region encompassing 586-LYVKL-590aa (P1 short) [34, 38], our investigations revealed that the deletion of residues P1 short did not abolish the interaction between ATRX and HP1$\alpha$ (Fig. 2G–I). Subsequent analysis identified three additional conserved potential PxVxL motifs: P2 (1977-PSVSL-1981aa), P3 (192-LQVLI-196aa), and P4 (1623-LVVCP-1627aa) in ATRX (Supplementary Fig. 2A). To investigate these motifs, we generated GST-fused fragments encompassing 50 aa of ATRX, which included the conserved P1, P2, P3, P4 motif, individually. These fragments were named GST-L1, GST-L2, GST-L3, and GST-L4. Our results revealed that GST-L1 and GST-L3 effectively pulled down exogenous HP1$\alpha$-Flag in mitotic HeLa cell lysates, but not GST-L2 or GST-L4 (Supplementary Fig. 2B–E). We subsequently constructed HA-ATRX-$\mathrm{\Delta }$PxVxL, in which both P1 and P3 of ATRX were deleted. This mutant exhibited a deficiency in co-immunoprecipitating with HP1$\alpha$-Flag (Fig. 2H). Additionally, we confirmed that HA-ATRX, but not the HA-ATRX-$\mathrm{\Delta }$PxVxL mutant was pulled down by GST-fused HP1$\alpha$ (Fig. 2I). In summary, our data revealed that HP1$\alpha$, HP1$\beta$, and HP1$\gamma$ interact with the N-terminus of ATRX through the two PxVxL motifs (586-LYVKL-590aa and 192-LQVLI-196aa).

ATRX directly interacts with HP1, and HP1 facilitates the centromeric localization of ATRX. The loss of ATRX is correlated with impaired centromeric cohesion and increased chromosome segregation errors. This observation led us to explore the potential involvement of the HP1-ATRX interaction in both the centromeric localization of ATRX and its function in regulating sister chromatid cohesion.

To determine the functional importance of the HP1-ATRX interaction, we generated an ATRX siRNA-resistant plasmid and subsequently expressed wild-type HA-ATRX (1A2 and 1C3) and the HP1-binding-deficient mutant HA-ATRX-$\mathrm{\Delta }$PxVxL (3A1) in HeLa cells (Fig. 3A).

Fig. 3.

Disruption of the HP1- ATRX interaction delocalizes ATRX from mitotic centromeres, resulting in increased centromeric cohesion defects. (A) Immunoblotting was performed on lysates derived from asynchronous HeLa cells along with the specified stable cell lines. (B,C) The specified stable cell lines as described in A underwent treatment with MG132, with or without the inclusion of endogenous ATRX RNAi cells. Cells were fixed at the given time intervals for DNA staining, and about 200 cells were counted (n = 2). (D) HeLa along with the referenced stable cell lines as described in A were treated with MG132 for 8 hours. The fraction of cells exhibiting cohesion loss was assessed in approximately 200 cells (n = 2) via mitotic chromosome spreads. Sample pictures are shown in Supplementary Fig. 3C. (E,F) The specified cell lines underwent a 3-hour treatment with nocodazole. The immunostaining of mitotic chromosome spreads was performed, and the ratio of immunofluorescence intensity for centromeric HA/CENP-C was measured in across 10 cells (E) (analyzed using an unpaired t-test). Example images are depicted in (F). Data information: Mean values and standard deviations (SDs) are displayed (B,C,E). Scale bar: 10 µm. Refer to Supplementary Fig. 3 for more information. WT, wild type; siControl, small interfering RNA Control.

Endogenous ATRX protein was knocked down by siRNA duplexes. Following MG132-induced metaphase arrest, cells lacking ATRX presented pronounced deficiencies in maintaining correct chromosomal biorientation and sister chromatid cohesion. Importantly, these defects were efficiently rescued by the presence of HA-ATRX, but not by HA-ATRX-$\mathrm{\Delta }$PxVxL (Fig. 3B–D, Supplementary Fig. 3A–C).

Immunofluorescence microscopy additionally demonstrated that the exogenous expression of HA-ATRX, while not HA-ATRX-$\mathrm{\Delta }$PxVxL, promoted the centromeric localization of HA-ATRX when endogenous ATRX was absent (Fig. 3E,F). These findings emphasize the importance of the HP1-ATRX interaction in preserving centromeric cohesion during mitosis and in promoting the appropriate localization of ATRX.

ATRX RNAi triggers a loss of cohesion, these findings underscore the significance of ATRX-interacting proteins in the protection of cohesion mediated by ATRX. Considering that the disruption of ATRX localization weakens sister chromatid cohesion, we hypothesized that the abnormal increase in local Wapl activity, resulting from the delocalization of centromeric ATRX, which we speculated would be a contributing factor to the centromeric cohesion abnormalities seen in ATRX-RNAi cells. siRNA duplexes were employed to knock down the endogenous ATRX protein, following MG132-induced metaphase arrest, our findings indicated that Wapl knockdown by siRNA significantly prevented the misalignment of metaphase chromosomes in cells expressing HA-ATRX-$\mathrm{\Delta }$PxVxL (Fig. 4A,B). Similarly, these mutant cells demonstrated the ability to sustain sister chromatid cohesion even without Wapl (Fig. 4C,D). Hence, the depletion of Wapl eliminates the need for HP1-ATRX interactions to maintain centromeric cohesion.

Fig. 4.

Depletion of Wapl abolishes the need for the HP1-ATRX interaction in safeguarding centromeric cohesion. (A) Cell lines were transfected with the specified siRNAs were subjected to immunoblotting. (B) The selected cell lines expressing the indicated siRNAs were treated with MG132, fixed at predetermined time points, and analyzed in approximately 200 cells (n = 2). Means and standard deviations (SDs) were recorded. (C,D) HeLa cells and the specified cell lines received MG132 treatment for 8 hours. By utilizing mitotic chromosome spreads, the fraction of cells with cohesion loss was assessed in around 100 cells (C). Example images of the mitotic chromosome spreads are provided (D). Scale bar, 10 µm. (E) Model for HP1 in localizing ATRX at mitotic centromeres, which inhibits Wapl’s release of cohesin from mitotic centromeres.