Phosphorylation of Barrier-to-Autointegration Factor (BAF) regulates anchoring of centromeric heterochromatin to the nuclear envelope (NE)

Abstract

In eukaryotes, the spatial segregation of heterochromatin and euchromatin is key for the structural organization and function of the genome. Heterochromatin interacts with the nuclear envelope (NE) and occupies a more peripheral position than euchromatin. However, the mechanisms that govern tethering of heterochromatin to the NE are not fully understood. Here, we report that Barrier-to-Autointegration Factor (BAF), a highly conserved NE-associated protein, interacts with centromeric heterochromatin and regulates its anchoring to the NE in a phosphorylation-sensitive manner. We show that impaired BAF phosphorylation leads to its persistent association with centromeric heterochromatin and reinforced anchoring. We also show that, concomitant with reinforced anchoring of centromeric heterochromatin to the NE, impaired BAF phosphorylation has important functional consequences, compromising both NE integrity and heterochromatin coalescence, and disturbing mitotic progression. Altogether, our results suggest that anchoring of centromeric heterochromatin to the NE is a highly dynamic process regulated through BAF phosphorylation, and reveal the deleterious functional consequences of perturbing this dynamic regulation.

Graphical Abstract

Graphical Abstract.

Introduction

A conserved feature in eukaryotes is the spatial segregation of heterochromatin and euchromatin inside the nucleus (reviewed in [1–5]. Heterochromatin preferentially locates at the nuclear periphery and associates with the nuclear envelope (NE), while euchromatin tends to occupy a more central nuclear location, away from the NE. This subnuclear compartmentalization is key for the regulation of 3D genome organization and function. However, the mechanisms driving heterochromatin anchoring to the NE are not yet fully understood.

The NE consists of a double lipid bilayer interconnected by nuclear pore complexes. The outer nuclear membrane faces the cytoplasm and is continuous with the endoplasmic reticulum (ER), while the inner nuclear membrane (INM) is decorated with specialized proteins. In higher eukaryotes, the INM organizes the nuclear lamina (NL), a dense scaffold of intermediate filaments that provides stiffness to the NE and determines the spherical shape of the nucleus (reviewed in [6–10]). Lamin proteins are the principal components of the NL. They constitute a proteinaceous platform for the binding of numerous functionally distinct NL-associated proteins and tethers large transcriptionally inactive chromatin regions (Lamina-Associated Domains or LADs) (reviewed in [4, 11–13]). NL-associated proteins include integral INM transmembrane proteins that play important roles in chromatin tethering. For instance, Lamin B receptor has been shown to contribute to heterochromatin anchoring both directly, through the recognition of H4K20 methylation [14], and indirectly, by binding to heterochromatin protein 1 (HP1) [15]. Similarly, the INM transmembrane LEM-domain proteins have been shown to mediate chromatin anchoring at LADs [4, 11–13].

Barrier-to-Autointegration Factor (BAF) is a highly conserved NL-associated protein in metazoans (reviewed in [16–18]). BAF depletion is lethal during embryogenesis in Caenorhabditis elegans and Drosophila [19, 20], while, in humans, BAF mutations are associated with Nestor–Guillermo Progeria Syndrome (NGPS), an early-onset aging condition [21, 22], and with a dominant motor neuropathy [23]. Originally identified as a factor inhibiting autointegration of retroviruses, BAF is involved in multiple cellular functions, from chromatin organization and transcription regulation to NE reassembly after mitosis and the repair of NE ruptures (reviewed in [16, 17, 18, 24, 25]). Central to BAF’s function is its ability to bridge chromatin to the NE. At the NL, BAF interacts with LEM-domain proteins and Lamins [26–30]. Structural studies showed that BAF is an obligate homodimer, to which LEM-domain proteins and Lamins bind through opposite clefts formed at the dimer interface [31–34]. It has also been shown that, through the helix–hairpin–helix DNA-binding domain of each monomer, BAF homodimers can bind two DNA strands in a sequence-independent manner [19, 35, 36]. The structure of a BAF–LaminA/C IgFold complex bound to nucleosomes has been recently solved, showing that BAF binds the two DNA strands at the nucleosome dyad [37]. The dual binding of BAF to DNA/nucleosomes and LEM-domain/Lamin proteins accounts for its reported contribution to chromatin anchoring at LADs (reviewed in [4, 11–13]). Noteworthily, BAF mutations associated with human diseases impair its ability to bridge chromatin to the NE, reducing the affinity for Lamins and/or DNA [23, 34, 38].

Although generally concentrated at the NL, BAF is also detected in the nucleoplasm and the cytoplasm. Phosphorylation has an important contribution to the regulation of BAF localization and function. Phosphorylation promotes BAF localization to the cytoplasm [39–42], while in vitro studies have shown that it inhibits binding to DNA [39–42, 43]. Vaccinia-related kinases VRK1 and VRK2 phosphorylate BAF at conserved S/T residues in its N-terminal region [39, 42, 44]. Depletion of VRK1 impairs post-mitotic NE assembly and leads to BAF retention on mitotic chromosomes [40, 45]. The protein phosphatases PP2A and PP4 have been shown to revert BAF phosphorylation [46–48].

Here, we report that BAF dynamically regulates anchoring of centromeric heterochromatin to the NE in a phosphorylation-sensitive manner. Mutant BAF forms with impaired phosphorylation accumulate at centromeric heterochromatin and reinforce its anchoring to the NE, which disturbs mitotic progression and compromises nuclear stability.

Materials and methods

DNA constructs

Complementary DNA (cDNA) encoding Drosophila BAF was obtained from Drosophila Genomics Resource Center (clone GH06291). Plasmids expressing YFP::BAF (N-BAF) and BAF::YFP (C-BAF) under the control of the endogenous BAF promoter are described in [49]. These plasmids were used to perform PCR-directed mutagenesis in which T4 and/or S5 were replaced by A in N-BAF (N-BAF^T4A, N-BAF^S5A, and N-BAF^T4A/S5A) or by E in C-BAF (C-BAF^T4E, C-BAF^S5E, and C-BAF^T4E/S5E). Flag-tagged BAF constructs were generated by replacing the YFP sequences in the previous constructs with the Flag sequence through site-directed substitution using oligonucleotides carrying the Flag coding sequence (Flag::BAF: Fwd 5′-ATGAGCGGAACCAGCCAAAA-3′, Rev 5′-CTTGTCGTCATCGTCTTTGTAGTCCATGGCTAGCGT TTGTAGTTTGCT-3′; BAF::Flag : Fwd 5′-GGATCCACCGGATCTAGATAACTG-3′, Rev 5′-GCTAGCTCACTTGTCATCGTCTTTGTAGTCC AGAAATTCTTCACACCAGTC-3′). To generate the plasmid expressing mCherry::BAF^T4A/S5A, the BAF^T4A/S5A construct was subcloned into plasmid act5Stable2Blast/ΔGFP (Addgene 208034). To generate the plasmid expressing mCherry::HP1a, the Drosophila HP1a coding sequence was obtained from plasmid pCopia-mCherry::HP1a (VectorBuilder) and subcloned to plasmid act5Stable2Blast/ΔGFP (Addgene 208034).

Fly stocks and genetic procedures

Actin-GAL4 and nub-GAL4 flies were obtained from Bloomington Stock Center. baf^RNAi corresponds to 102013 stock from the Vienna Drosophila RNAi Center. Transgenic flies carrying the various UAS-BAF constructs described in Supplementary Fig. S4 were obtained by site-directed integration of the corresponding pUASTattb plasmids into chromosome 3 using 3R-86Fb embryos.

For experiments with knockdown baf^RNAi flies, crosses were left at 25°C until third-instar larvae stage. For overexpression experiments, homozygous transgenic flies carrying the corresponding UAS-constructs were crossed to heterozygous Actin-GAL4/Tb, when viability was determined, or homozygous nub-GAL4 flies, when wing defects were determined. To analyze the effects on wing development, flies were kept in 75% ethanol, 25% glycerol solution for at least 24 h at room temperature and washed in phosphate buffered saline (PBS). Then, wings were dissected and immediately mounted in Fauré’s medium under gentle pressure. Images were acquired using a 4× objective lens on a Nikon E-600 microscope equipped with an Olympus DP72 camera and CellF software.

Stable Drosophila S2 cell lines

Cultured cells used in these experiments were Schneider’s Drosophila Line 2 [D. Mel. (2), SL2] (ATCC® CRL-1963™). To obtain stable cell lines, Drosophila S2 cells were grown under standard conditions [25°C in Schneider’s medium (Sigma) supplemented with 10% FBS (Gibco), 100 mg/ml Streptomycin, and 100 mg/ml Penicillin] and transfected with the appropriate constructs by the calcium phosphate method as described in [50]. After 48 h of transfection, 0.8 mg/ml of G418 and/or 0.02 mg/ml of Blasticidin were added for selection. After selection, the levels of expression of the corresponding constructs were determined by RT-qPCR and/or WB (Supplementary Fig. S1A–D). For fluorescence visualization, to ameliorate the observed differences in expression levels of N-BAF and C-BAF, N-BAF expressing cells were sorted by FACS to select those showing low expression (Supplementary Fig. S1E and F). For transient expression experiments, the indicated constructs were transfected by the calcium phosphate method as described in [50]. Cells were collected 48 h post-transfection, and the levels of expression determined by WB (Supplementary Fig. S1G).

Antibodies

Rabbit polyclonal αBAF and rat polyclonal αCENP-C antibodies are described in [49]. Rat polyclonal αHP1a antibodies are described in [51]. The rest of antibodies used are commercially available: mouse monoclonal αHP1a (DSHB C1A9), mouse monoclonal αLaminDm0 (DSHB ADL67.10), rabbit polyclonal αGFP (Invitrogen A11122), mouse polyclonal αGFP (Roche 118144600), rabbit polyclonal αFlag (Merck Life Science F7425), rabbit polyclonal αH3 (Cell Signaling 9715), rabbit polyclonal αH3K9me3 (Millipore 67–442), rabbit polyclonal αH3PS10 (Cell Signaling 9701), and mouse monoclonal αTubulin (MAB3408 Millipore).

Western blot analysis

Western blot analyses were performed according to standard procedures using the following antibody dilutions: αBAF (1:2500), rat αHP1a (1:1000), αH3 (1:2500), αTubulin (1:5000), mouse αGFP (1:2000), αCENP-C (1:3000), αFlag (1:2500), and αLaminDm0 (1:5000).

RT-qPCR

When the levels of N-BAF and C-BAF messenger RNA (mRNAs) were determined by RT-qPCR, total RNA from N-BAF or C-BAF expressing cells was isolated using RNazol (Merck) in combination with the RNeasy Mini Kit (Qiagen), according to manufacturers’ instructions. First-strand cDNA synthesis was performed with the Transcriptor First Strand cDNA Synthesis Kit (Roche) using oligo(dT) primers. qPCR was conducted using the standard curve method. Transcript levels were normalized to Actin5C mRNA levels. The following primers were used: EGFP-Fw: TATATCATGGCCGACAAGCA/EGFP-Rv: GTTGTGGCGGATCTTGAAGT for N-BAF and C-BAF, and Actin-Fw: CACCAAATCTTACAAAATGTGTGAC/Actin-Rv: CATCGTCTCCGGCAAATC for Actin5C.

Co-immunoprecipitation experiments

For co-immunoprecipitation (co-IP) experiments, GFP-Trap Magnetic agarose beads (Chromotek gtma-20) were used to immunoprecipitate YFP-tagged BAF forms following manufacturer’s instructions. Briefly, cells were incubated in ice cold 300 mM NaCl RIPA buffer in the presence of Protease Inhibitor Cocktail and 0.1 mM PMSF for 30 min. After centrifugation, 1 volume of 150 mM NaCl, 10 mM Tris–HCl (pH 7), 0.5 mM ethylenediaminetetraacetic acid (EDTA) supplemented with Protease Inhibitor Cocktail and 0.1 mM PMSF was added to the cleared lysate. Then, GFP-Trap beads were added and incubated at 4°C for 2 h. After washing with 100 mM Tris–HCl (pH 7.5), 150 mM NaCl, 0.05% NP-40, 0.5 mM EDTA, beads were eluted with 2× sodium dodecyl sulfate (SDS) PLB, 10% β-mercaptoethanol.

Analysis of BAF phosphorylation

To analyze BAF phosphorylation, total cell protein extracts were obtained in 150 mM NaCl, 50 mM Tris–HCl (pH 8), 10% glycerol, 0.1% SDS, 1% NP40, 1 mM PMSF, Protease Inhibitor Cocktail, and phosphatase inhibitors 50 mM NaF, 2 mM Na₃VO₄, and 10 mM glycerol phosphate. Phosphorylated BAF forms were resolved by Phos-tag gel electrophoresis [52], as described in [49], and analyzed by WB. Where indicated, cells were treated with 20 nM okadaic acid for 3 h to increase BAF phosphorylated forms.

Immunostaining experiments

Immunostaining experiments with αCENP-C (1:500), rat αHP1a (1:200), αFlag (1:500), αH3PS10 (1:500), or αLaminDm0 (1:1500) antibodies were performed as previously described [49] in cells seeded in Concanavalin A-coated cover slides or, for mitotic chromosome spreads, in impacted cells. For visualization, slides were mounted in Mowiol (Calbiochem-Novabiochem) containing 0.2 ng/ml DAPI (Sigma) and imaged in a Leica TCS/SPE confocal microscope equipped with LAS/AF software or a Zeiss confocal LSM880 microscope equipped with Airyscan. Images were processed using ImageJ (http://imagej.nih.gov/ij/) [53]. Mean gray intensities were calculated on thresholded DAPI regions of interest running Analyze particles plugin on the FeatureJ Laplacian (http://imagescience.org/meijering/software/featurej/).

When the localization of HP1a foci relative to Lamin was determined (see Fig. 6A and B), a specific Z-plane was identified for each nucleus, where protein signals were strongest in both intensity and size. The obtained segmentation mask determined the nuclear region. To ensure a robustly well-defined nucleus segmentation through different αLamin immunostaining intensities, a deep learning pretrained model, cellpose CPx [54], was used. Then, the distance transform of the nuclear mask was normalized so that each pixel inside the nucleus had a value between 0 and 1, with 0 at the boundary and 1 at the center. We divided a nucleus region into three equally spaced band regions according to the distance values: periphery, intermediate, and center regions. For each HP1a foci, identified by a segmented area, its location within a nucleus was characterized by three values representing its distribution across the three defined nuclear regions.

Figure 6.

Impaired BAF phosphorylation reinforces heterochromatin anchoring. (A) Immunostaining with αLamin antibodies (in cyan) in stable S2 cells co-expressing mCherry::HP1a (in red) and either N-BAF, C-BAF, or N-BAF^T4A/S5A. mCherry::HP1a signal is direct fluorescence. Scale bars are 2 μm. (B) Quantification of the results shown in panel (A). The proportion of HP1a foci in the peripheral, intermediate, and central nuclear regions is shown for cells expressing N-BAF, C-BAF, or N-BAF^T4A/S5A (see “Materials and methods” section for details). As a control, a similar quantification was performed in S2 cells after co-immunostaining with αLamin and αHP1a antibodies. Results are the sum of two independent experiments. (N = 91 for N-BAF, 93 for C-BAF, and 111 for N-BAF^T4A/S5A; P-values with respect to control: ns > 0.05, ***** < 0.00001; two-tailed paired Student’s t-test). (C) Quantification of the results shown in panel (A). The number of HP1a foci per nucleus is shown for cells expressing N-BAF, C-BAF, or N-BAF^T4A/S5A. Results are the sum of two independent experiments. Error bars are SD. (N = 35 for N-BAF, 34 for C-BAF, and 33 for N-BAF^T4A/S5A; P-values with respect to control: ***** < .00001; two-tailed paired Student’s t-test). (D) ChIP-seq genomic profiling of endogenous BAF, N-BAF, C-BAF, and N-BAF^T4A/S5A. For endogenous BAF, ChIP was performed with αBAF antibodies in Drosophila S2 cells. For N-BAF, C-BAF, and N-BAF^T4A/S5A, ChIPs were performed with αGFP antibodies in stable S2 cells expressing the corresponding BAF forms. The IP/Input coverage ratio is presented for two representative genomic regions and compared to Lamin (GEO GSM509086) [61]. LADs are indicated in gray. The Spearman correlation coefficients respect to Lamin for the whole genome are also indicated.

When the αLamin intensity was measured at NE regions showing foci of the indicated YFP-tagged BAF forms (see Fig. 8E and F), the relative αLamin intensity was used. It is defined as the ratio of the median αLamin intensity at YFP-tagged BAF foci regions to the median αLamin intensity across the entire NE. First, both 3D Lamin and YFP-tagged BAF foci regions were identified. Those YFP-tagged BAF foci regions in contact with Lamin were kept and used to obtain the relative ⍺Lamin intensity for each foci.

Figure 8.

Non-phosphorylatable BAF forms compromise NE integrity. (A) Images of metaphase cells expressing N-BAF, C-BAF, or N-BAF^T4A/S5A. N-BAF, C-BAF, and N-BAF^T4A/S5A signals are direct fluorescence (in green). DNA was stained with DAPI (in gray). Scale bars are 2 μm. (B) In vivo time-lapse recordings of a stable S2 cell co-expressing N-BAF (in green) and mCherry::N-BAF^T4A/S5A (in red) that is undergoing mitosis. Time is in minutes. Scale bars are 5 μm. (C) Immunostaining with αLamin antibodies (in cyan) of stable S2 cells co-expressing mCherry::HP1a (in red) and either N-BAF (left), C-BAF (center), or N-BAF^T4A/S5A (right). mCherry::HP1a signal is direct fluorescence. Scale bars are 2 μm. (D) Quantification of the results shown in panel (C). The proportion of cells showing NE defects is presented for cells expressing the indicated BAF constructs. (N = 323 for N-BAF, 360 for C-BAF, and 341 for N-BAF^T4A/S5A; P-values with respect to N-BAF expressing cells: ***** < .00 001; Fisher’s exact test). (E) Immunostaining with αLamin antibodies (in red) of stable S2 cells expressing N-BAF (top), C-BAF (center), or N-BAF^T4A/S5A (bottom). Representative images of NE regions showing foci of the corresponding BAF forms are presented. N-BAF, C-BAF, and N-BAF^T4A/S5A signals (in green) are direct fluorescence. Scale bars are 0.5 μm. (F) Quantification of the results shown in panel (E). Box plots showing median αLamin intensity at NE regions showing N-BAF, C-BAF, or N-BAF^T4A/S5A  foci relative to the median αLamin intensity across the entire NE (see “Materials and methods” section for details). Boxes represent the median and IQR, whiskers are 1.5 × IQR. (N = 168 for N-BAF, 119 for C-BAF, and 149 for N-BAF^T4A/S5A; P-values: ***** < .00 001; two-tailed paired Student’s t-test).

Scripts used for the analyses described earlier, example test data and detailed usage instructions together with parameter settings could be found on Zenodo (https://zenodo.org/records/15490539).

Immuno-proximity ligation assays

Immuno-proximity ligation assays (Immuno-PLA) was performed by using Duolink^R  in situ detection Red kit (DUO92008). First, cells were seeded on Concanavalin A-coated multichambered slices for 24 h. After fixation with 4% PFA (Paraformaldehyde), cells were washed with PBS and permeabilized in 0,5% Triton X-100 in PBS. Then, samples were incubated with Duolink blocking solution in a preheated humid chamber for 1 h at 37°C. After blocking, cells were incubated with primary antibodies diluted in Duolink antibody diluent overnight at 4°C at the following dilutions: rabbit polyclonal αGFP (1:1000), mouse monoclonal αHP1a (1:100), rat monoclonal αHP1a (1:200), rabbit polyclonal αBAF (1:300), and rabbit polyclonal αH3K9me3 (1:100). Next, MINUS and PLUS secondary PLA probes were added and incubated for 1 h at 37°C. Subsequently, ligation and amplification steps were carried out following manufacturer’s instructions. Finally, Cy5 αrat secondary antibody diluted in PBS was incubated for 45 min at room temperature. Slides were mounted in Mowiol (Calbiochem-Novabiochem) containing 0.2 ng/ml DAPI (Merck Life Science). Images were acquired in a Leica TCS/SPE confocal microscope equipped with LAS/AF software and analyzed and processed using ImageJ (http://imagej.nih.gov/ij/).

FRAP experiments

For fluorescence recovery after photobleaching (FRAP) experiments, stable cell lines expressing YFP-tagged BAF forms were grown in 35 mm μ-dishes (ibidi) coated with concanavalin A the day before imaging. Cells were imaged using a 60 × 1.42 oil immersion objective (Plan Apo N) on an Olympus IX81 inverted confocal spinning disk microscope from Andor. YFP fluorescence was photobleached using 488 nm light from a 488 diode laser (20%) in a defined region (10 × 10 pixels) of each cell. The mean intensity of three regions was quantified over time, both pre-bleaching (20 frames at 0.142 ms time frequency) and after bleaching (300 frames at the same time frequency as above). Recovery dynamics were normalized for different starting expression levels and for background noise according to [55], and analyzed using the easyFRAP curve fitting web tool (https://easyfrap.vmnet.upatras.gr/) with a double exponential model to extract the mobile fraction and halftime of recovery.

Flow cytometry

Samples were analyzed using a Gallios multi-color flow cytometer (Beckman Coulter, Inc, Fullerton, CA) set up with the 3-laser, 10-color standard configuration. Excitation was done using a blue (488 nm) laser. Forward scatter, side scatter, red (620/30 nm) fluorescence emitted by propidium iodide was collected. Red fluorescence was projected on a 1024 monoparametric histogram. Aggregates were excluded by gating single cells on their area versus peak fluorescence signal. Cell cycle analysis was done by measuring DNA content of GFP positive cells using FlowJo™ v10.10 Software (BD Life Sciences).

ChIP-seq

For ChIP experiments, cells (14 × 10⁶) were crosslinked with 1.8% formaldehyde for 10 min at room temperature, and chromatin was prepared as described in [56]. Chromatin samples were sonicated using a Bioruptor Pico (Diagenode) to obtain fragments between 200–500 bp and, after sonication, lysates were adjusted to 1% Triton X-100, 0.1% sodium deoxycholate (DOC), 140 mM NaCl, incubated for 10 min on a rotating wheel at 4°C, and chromatin was recovered by centrifugation. For each experiment, 400 μl of chromatin were used for IP, and 40 μl were retained as input. Immunoprecipitation (IP) was carried out in RIPA buffer (140 mM NaCl, 10 mM Tris–HCl, pH 8.0, 1 mM EDTA, 1% Triton X-100, 0.01% SDS, and 0.1% DOC). For endogenous BAF, lysates were pre-cleared with 30 μl of 50% (v/v) Protein A-Sepharose CL-4B beads (GE Healthcare, 17-0780-01) pre-blocked with RIPA + 1% BSA and, after pre-clearing, 10 μl of αBAF antibodies were added, and samples were incubated overnight at 4°C with rotation. Then, IPs were performed by adding 40 μl of 50% (v/v) Protein A-Sepharose CL-4B beads (pre-blocked with RIPA + 1% BSA) and incubating for 3 h at 4°C on a rotating wheel. For N-BAF, C-BAF, and N-BAF^T4A/S5A samples, 40 μl of GFP-Trap Magnetic Agarose beads (Chromotek) were added, and samples were incubated for 3 h at 4°C with rotation. After IP, beads from all conditions were washed five times for 5 min with 1 ml of RIPA buffer, once with high-salt buffer [250 mM LiCl, 10 mM Tris–HCl (pH 8.0), 1 mM EDTA, 0.5% NP-40, 0.5% DOC], and twice with TE buffer for 5 min. Beads were resuspended in 40 μl of TE, and RNase A (DNase-free) was added to a final concentration of 0.25 μg/ml to both IP and input samples, and samples were incubated for 30 min at 37°C. To purify DNA, samples were adjusted to 1% SDS, 100 mM NaHCO₃, 0.2 mg/ml Proteinase K, and incubated overnight at 65°C. DNA was purified using phenol-chloroform extraction.

For NGS, libraries were generated using NEBNext Ultra II DNA Library Prep Kit (Illumina) and sequenced by 2 × 150 paired-end reads on a NovaSeq S4 (Illumina) at the CNAG (Barcelona). FastQ files were aligned to the Drosophila melanogaster dm6 genome using Bowtie2 v2.5.1 [57] with options -N 1 -k 1. Duplicated reads for generation of TDF tracks were identified and removed with sambamba v0.5.1 [58]. TDF tracks were generated with IGVTools2 [59]. Quality control of FastQ and BAM files was done with FastQC v0.11 and MultiQC. Protein occupancy was assessed by calculating the normalized log₂ CPM fold change (IP/Input) using deepTools [60]. For comparative analyses, genome-wide Lamin-DamID data from Kc cells (GEO GSM509086) [61] were aligned to dm6. Lamin-DamID signal was divided into deciles based on intensity, and the genome coverage was ranked accordingly. Log₂ CPM fold-change values from the ChIP-seq IP/Input datasets were also divided into deciles, and comparisons between the Lamin-DamID and ChIP-seq datasets were performed using Spearman correlation analyses.

AlphaFold predictions

For structural prediction, AlphaFold v3 [62] was used. BAF and NHK-1 protein sequences were retrieved from the UniProt database (accession numbers: Q9VLU0 and Q7KRY6, respectively). Structural models with the highest pTM and ipTM scores were selected [63, 64] and visualized using ChimeraX. The Prodigy web server [65] was used to calculate dissociation constants (Kd) as described in [66, 67].

Statistical analysis

For each experiment, the number of independent biological replicates, sample sizes, and statistical test used are indicated in the corresponding figure legend.

Results

BAF associates with centromeric heterochromatin

In Drosophila, nuclear BAF is enriched around the NE, forming a perinuclear ring, and is also detected surrounding the nucleolus and throughout the nucleoplasm [20, 68, 69]. Perinuclear BAF reflects its interaction with NL components [26–30] and has been shown to be involved in chromatin anchoring at LADs [4, 11–13]. However, whether BAF interacts with centromeric heterochromatin and contributes to its anchoring to the NE remains unknown. To address this question, we performed Immuno-PLA in Drosophila S2 cells. In these experiments, we assessed the association of BAF with centromeric heterochromatin from the PLA signal obtained using αBAF and αHP1a antibodies, and cells were simultaneously immunostained with αHP1a antibodies. We observed strong αBAF/αHP1a PLA signal at HP1a-positive regions in a large proportion of cells (Fig. 1A, top and B). These results closely resembled those obtained using αHP1a and, as a positive control, antibodies against H3K9me3, an epigenetic modification that marks centromeric heterochromatin and is recognized by HP1a (reviewed in [70]) (Fig. 1A, center and B). Notice that no substantial PLA signal was detected when αGFP antibodies were used as a negative control (Fig. 1A, bottom and B).

Figure 1.

BAF associates with centromeric heterochromatin. (A) Immuno-PLA with αHP1a antibodies and either αBAF (top), αH3K9me3 (center), or αGFP (bottom) antibodies in S2 cells. The PLA signal is shown in red. Immunostaining with αHP1a antibodies is shown in cyan. DNA was stained with DAPI (in gray). In the top and center panels, enlarged images of representative nuclei show the overlap of PLA and αHP1a signals. Scale bars are 10 μm (in the top, center, and bottom panels) and 2 μm (in the enlarged images). (B) Quantification of the results shown in panel (C). The percentages of PLA-positive cells are presented for assays performed with αHP1a antibodies and either αGFP, αH3K9me3, or αBAF antibodies. Results are the sum of two independent experiments. (N = 224 for αGFP, 228 for αH3K9me3, and 230 for αBAF; P-values: ns > 0.05; *** < 0.001; Fisher’s exact test).

Next, we performed direct fluorescence visualization experiments using stable cell lines expressing either N- or C-terminally tagged BAF-YFP forms, hereinafter referred to as N-BAF and C-BAF, respectively. We found that N-BAF displays a broad nuclear distribution, prominently localizing along the perinuclear ring, with additional enrichment at centromeric heterochromatin (Fig. 2A). In contrast, C-BAF exhibits a more restricted nuclear distribution, with markedly reduced localization at the perinuclear ring and preferential clustering at centromeric heterochromatin (Fig. 2B and C). Immuno-PLA experiments performed with αGFP antibodies, which specifically detect the YFP-tagged forms, and αHP1a antibodies confirmed the association of both N-BAF and C-BAF with centromeric heterochromatin (Fig. 2D and E). Noteworthily, we observed that the proportion of PLA-positive cells is significantly higher for cells expressing C-BAF than for N-BAF-expressing cells.

Figure 2.

The localization patterns of N-BAF and C-BAF. (A) The pattern of localization of N-BAF is determined by direct fluorescence in stable S2 cells co-expressing N-BAF (in green) and mCherry::HP1a (in red). Scale bars are 2 μm. (B) As in panel (A), but for C-BAF in stable S2 cells co-expressing C-BAF (in green) and mCherry::HP1a (in red). Scale bars are 2 μm. (C) Quantification of the results shown in panels (A) and (B). Box plots of the fluorescence intensity at the perinuclear ring relative to centromeric heterochromatin are presented for N-BAF and C-BAF expressing cells. Boxes represent the median and interquartile range (IQR), whiskers are 1.5 × IQR. Results are the sum of three independent experiments. (N = 42 for N-BAF and C-BAF; P-value: **** < .0001; unpaired Student’s t-test). (D) Immuno-PLA with αHP1a antibodies and αGFP antibodies in stable S2 cells expressing N-BAF (top) or C-BAF (bottom). The PLA signal is shown in red. Immunostaining with αHP1a antibodies is shown in cyan. DNA was stained with DAPI (in gray). Scale bars are 10 μm. (E) Quantification of the results shown in panel (D). The percentages of PLA-positive cells are presented for assays performed with αHP1a antibodies and αGFP in stable S2 cells expressing N-BAF (green) or C-BAF (gray). The percentage of αHP1a/αGFP PLA-positive cells shown in Fig. 1D is included for comparison. Results are the sum of two independent experiments. (N = 379 for N-BAF and 266 for C-BAF; P-value: *** < .001; Fisher’s exact test).

Next, we performed FRAP experiments. We observed that N-BAF is exchanged at a similarly rapid rate at both the perinuclear ring and centromeric heterochromatin (Fig. 3). In contrast, C-BAF, which is largely absent from the perinuclear ring (Fig. 2C), shows an extremely slow exchange rate at centromeric heterochromatin (Fig. 3).

Figure 3.

C-BAF is slowly exchanged at centromeric heterochromatin. In the top, recordings of FRAP experiments performed in stable S2 cells expressing N-BAF or C-BAF. Squares indicate the bleached regions. For N-BAF expressing cells, bleaching was performed both at the perinuclear ring and at centromeric heterochromatin (HET); two examples are shown at HET. In the bottom, quantification of the results showing FRAP for N-BAF, at the perinuclear ring (in black) and at centromeric heterochromatin (HET) (in yellow), and C-BAF (in red) expressing cells. The t-half and proportion of mobile fraction are indicated. Results are expressed as mean ± SD. (N = 34 and 15 for N-BAF at the perinuclear ring and centromeric heterochromatin, respectively; N = 36 for C-BAF). (see “Materials and methods” section for details).

Altogether, these results indicate that BAF associates with centromeric heterochromatin and that the association of C-BAF is markedly different in some fundamental aspects, resulting in slow exchange and preferential accumulation at centromeric heterochromatin.

Phosphorylation regulates BAF dynamics at centromeric heterochromatin

The preferential association of C-BAF with centromeric heterochromatin was unexpected. It is well established that phosphorylation negatively regulates the binding of BAF to DNA [39–42, 43, 71]. In this context, it is noteworthy that, as previously reported [49], C-BAF exhibits impaired phosphorylation. In Drosophila, BAF is phosphorylated by the VRK1 homolog NHK1 at the conserved T4 and S5 residues in its N-terminus [39, 40, 42, 72]. Accordingly, Phos-tag gel electrophoresis analysis of endogenous BAF detects mono- (1P) and di-phosphorylated (2P) species [49] (Supplementary Fig. S2A, left). This phosphorylation pattern is preserved in N-BAF (Supplementary Fig. S2A, center). However, phosphorylation of C-BAF is profoundly altered [49] (Supplementary Fig. S2A, right).

Co-immunoprecipitation experiments have shown that NHK1 interacts with BAF [72]. However, little is known about the molecular basis of this interaction. Since no direct structural data are available, we modeled the interaction of BAF with NHK1 using AlphaFold3 [62] and found that it is predicted to involve a shallow cleft at the dimer interface, formed by the C-terminal α-helices of the monomers (Fig. 4A and D, and Supplementary Movie S1A). A similar region has been shown to mediate the interaction of BAF with the LaminA/C IgFold domain [34]. These results suggest that BAF homodimerization is crucial for NHK1 binding and, consequently, for its phosphorylation. Notably, AlphaFold3 models for N-BAF predict normal homodimerization (Fig. 4B) and stable interaction with NHK1 (Fig. 4E and Supplementary Movie S1B). In contrast, models for C-BAF show impaired BAF homodimerization (Fig. 4C). Furthermore, although models of the C-BAF–NHK1 complex suggest potential BAF homodimerization, the interaction with NHK1 appears to be displaced and of reduced affinity (Fig. 4F and Supplementary Movie S1C). Consistent with impaired dimerization, we observed that, in comparison to N-BAF, co-immunoprecipitation of C-BAF with endogenous BAF is markedly reduced (Fig. 4G and H). Altogether, these findings suggest that impaired phosphorylation of C-BAF results from altered dimerization.

Figure 4.

C-BAF has impaired dimerization and phosphorylation. (A-C) AlphaFold3 structural models for BAF (A), N-BAF (B), and C-BAF (C). Only structures with the highest ranking-confidence scores are presented. BAF monomers are shown in yellow and red. YFP is shown in green. The C- and N-terminus of BAF are indicated. (D-F) AlphaFold3 structural models for the complex of NKH1/VRK1 with BAF (D), N-BAF (E), and C-BAF (F). Only structures with the highest ranking-confidence scores are presented. NHK1/VRK1 is shown in magenta, BAF is shown in yellow, N-BAF is shown in light blue, and C-BAF is shown in red. The YFP domain is shown in gray. Dissociation constants (K_d) of the predicted complexes calculated as described in [66, 67] are indicated. (G) Co-immunoprecipitation of endogenous BAF with N-BAF and C-BAF. IPs were performed with GFP-Trap beads in total extracts obtained from S2 cells expressing N-BAF or C-BAF, and from control S2 cells. WBs were performed with αBAF antibodies. Input lanes correspond to 3% of the extracts used for IP. The positions of endogenous BAF, and the YFP-tagged N-BAF and C-BAF forms are indicated. The position of Mw (kDa) markers are also indicated. (H) Quantification of the results shown in panel (G). The ratios of endogenous BAF versus N-BAF and C-BAF after immunoprecipitation with αGFP antibodies are presented. Results are the average of three independent experiments. Error bars are SD. (P-value is indicated; two-tailed paired Student’s t-test).

Next, we addressed whether the accumulation of C-BAF at centromeric heterochromatin is a consequence of its impaired phosphorylation. For this purpose, we analyzed the behavior of phospho-mutant N-BAF forms in which T4, S5, or both are replaced by A. As shown by Phos-tag gel electrophoresis analysis, T4A and S5A mutants show only 1P species (Supplementary Fig. S2B, lanes T4A and S5A), whereas the double T4A/S5A mutant is not phosphorylated (Supplementary Fig. S2B, lanes T4A/S5A). Notably, we observed that the uniform pattern of localization around the NE of N-BAF becomes constrained to centromeric heterochromatin when phosphorylation is impaired in the phospho-dead N-BAF^T4A/S5A mutant that, like C-BAF, is largely absent from the perinuclear ring (Fig. 5A and B). This change in localization is most clearly demonstrated in a stable cell line co-expressing both N-BAF and an N-terminally tagged mCherry::BAF^T4A/S5A form. In this cell line, while N-BAF is uniformly distributed around the NE with some enrichment at centromeric heterochromatin, mCherry::BAF^T4A/S5A is predominantly localized at centromeric heterochromatin (Fig. 5C). Moreover, immuno-PLA experiments confirmed the association of N-BAF^T4A/S5A with centromeric heterochromatin (Fig. 5D and E), and FRAP experiments showed that, though faster than C-BAF, N-BAF^T4A/S5A is exchanged to a markedly slower rate than N-BAF (Fig. 5F). These results indicate that impairing N-BAF phosphorylation restricts its localization to centromeric heterochromatin. As a matter of fact, the proportion of cells showing constrained localization to centromeric heterochromatin increases progressively as phosphorylation is impaired in N-BAF (Fig. 5G). This proportion is higher in cells expressing N-BAF^T4A/S5A than in cells expressing N-BAF^S5A, while in cells expressing N-BAF^T4A is only slightly higher than in N-BAF expressing cells (Fig. 5G). We also observed that, vice versa, constrained localization of C-BAF to centromeric heterochromatin is reduced in the phospho-mimetic C-BAF^T4E/S5E and C-BAF^S5E mutants, though not in the C-BAF^T4E mutant (Fig. 5G). As a matter of fact, the extent of perinuclear localization of the phospho-mimetic C-BAF^T4E/S5E form is similar to that of N-BAF (Fig. 5B).

Figure 5.

The phospho-dead N-BAF^T4A/S5A mutant mimics C-BAF. (A) The pattern of localization of N-BAF^T4A/S5A is determined by direct fluorescence in stable S2 cells co-expressing N-BAF^T4A/S5A (in green) and mCherry::HP1a (in red). Scale bars are 2 μm. (B) Box plots showing the fluorescence intensity at the perinuclear ring relative to centromeric heterochromatin for cells expressing N-BAF^T4A/S5A (in red) or C-BAF^T4E/S5E (in yellow). Results are the sum of 3 and 2 independent experiments, respectively. Results in Fig. 2C for N-BAF (in green) and C-BAF (in gray) expressing cells are included for comparison. Boxes represent the median and IQR, whiskers are 1.5 × IQR. (N = 40 for N-BAF^T4A/S5A and 31 for C-BAF^T4E/S5E; P-value: **** < .0001; unpaired Student’s t-test). (C) The patterns of localization of N-BAF and N-BAF^T4A/S5A are determined by direct fluorescence in stable S2 cells co-expressing N-BAF (in green) and N-BAF^T4A/S5A (in red). Scale bars are 2 μm. (D) Immuno-PLA with αHP1a antibodies and αGFP antibodies in stable S2 cells expressing N-BAF^T4A/S5A. The PLA signal is shown in red. Immunostaining with αHP1a antibodies is shown in cyan. DNA was stained with DAPI (in gray). Scale bars are 10 μm. (E) Quantification of the results shown in panel (D). The percentage of PLA positive cells is presented for assays performed with αHP1a antibodies and αGFP in stable S2 cells expressing N-BAF^T4A/S5A (red). For comparison, the percentages of αHP1a/αGFP PLA positive cells shown in Fig. 2E for N-BAF (green) and C-BAF (gray) expressing cells are included. Results are the sum of two independent experiments. (N = 173; P-values: * < .05, *** < .001; Fisher’s exact test). (F) In the top, recordings of FRAP experiments in stable S2 cells expressing N-BAF^T4A/S5A. Squares indicate the bleached regions. In the bottom, quantification of the results showing FRAP for N-BAF^T4A/S5A. The t-half and proportion of mobile fraction are indicated. Results are expressed as mean ± SD. Quantification of the results obtained for N-BAF and C-BAF shown in Fig. 3 are included for comparison. (N = 33 for N-BAF^T4A/S5A). (see “Materials and methods” section for details). (G) The proportion of cells expressing the indicated BAF constructs that show perinuclear (in green) or centromeric heterochromatin (in red) localization of the corresponding BAF protein. Results are the sum of two independent experiments (N > 180; P-values of the N-BAF/C-BAF mutated forms respect to the corresponding wild-type form: ns > 0.5; **** < .0001; Fisher’s exact test).

We also analyzed the behavior of Flag-tagged BAF forms, where the tag is substantially smaller than YFP. Consistent with observations from YFP-tagged constructs, N-terminally tagged Flag::BAF remains phosphorylatable, whereas C-terminally tagged BAF::Flag exhibits severely impaired phosphorylation (Supplementary Fig. S3A). Notice that, in these experiments, cells were treated with okadaic acid to inhibit protein phosphatases and enhance BAF phosphorylation. Concomitant with impaired phosphorylation, C-terminally tagged BAF::Flag predominantly localizes to centromeric heterochromatin (Supplementary Fig. S3B), whereas N-terminally tagged Flag::BAF displays a uniform nuclear distribution (Supplementary Fig. S3C), with noticeable enrichment at the perinuclear ring.

Expression of non-phosphorylatable BAF is deleterious

Notably, in flies, ubiquitous overexpression of the non-phosphorylatable C-BAF form results in high lethality, with flies dying at the pupal stage, whereas overexpression of N-BAF or untagged BAF do not affect viability (Supplementary Table S1). In these experiments, males homozygous for the appropriate UAS-BAF constructs were crossed with Actin5C-GAL4/Tb females to induce ubiquitous overexpression of the corresponding BAF form in the offspring. Progeny were scored based on the presence or absence of the Tb marker: non-Tb individuals overexpressed the BAF form, while Tb individuals did not. We found that the observed numbers of flies overexpressing N-BAF, or untagged BAF, were close to those expected (Supplementary Table S1). Similar N-terminally tagged gfp::baf alleles were previously reported to rescue baf loss-of-function in both Drosophila and C. elegans [73, 74]. In contrast, no flies overexpressing C-BAF were detected (Supplementary Table S1).

Furthermore, as previously reported [49], the non-phosphorylatable C-BAF form acts as a dominant-negative mutation since its specific overexpression in the wing imaginal disc of larvae using a nub-GAL4 driver mimics BAF depletion, resulting in the absence of wings in the adult (Supplementary Fig. S4A, B, and D). In contrast, overexpression of N-BAF or untagged BAF has no detectable wing phenotypes (Supplementary Fig. S4C and E).

In line with these observations, we noted that, in stable S2 cell lines, C-BAF is expressed at much lower levels than N-BAF (Supplementary Fig. S1A and B). Notably, this regulation takes place at the mRNA level (Supplementary Fig. S1C), which suggests that low expression of C-BAF is likely due to its deleterious dominant-negative nature, such that only cells with low expression levels are viable and selected during stable cell line generation. Similarly, the phospho-dead N-BAF^T4A/S5A mutant also shows low expression (Supplementary Fig. S1D), which is suggestive of a dominant-negative behavior. As a matter of fact, the levels of expression of the different N-BAF and C-BAF phospho-mutants correlate well with the extent to which phosphorylation is impaired or mimicked, respectively (Supplementary Fig. S1D). Likewise, the non-phosphorylatable C-terminally tagged BAF::Flag construct is expressed at markedly lower levels than its phosphorylatable N-terminally tagged Flag::BAF counterpart (Supplementary Fig. S3D and E).

These results indicate that impaired BAF phosphorylation has important deleterious effects. Below, we examine these functional consequences in more detail.

Non-phosphorylatable BAF forms reinforce heterochromatin anchoring

Results reported above suggest that impairing BAF phosphorylation might reinforce anchoring of centromeric heterochromatin to the NE. Indeed, centromeric heterochromatin occupies a more peripheral location, in contact with the NE, in cells expressing C-BAF or N-BAF^T4A/S5A than in N-BAF expressing cells (Fig. 6A and B). In these experiments, we performed αLamin immunostaining in stable S2 cell lines expressing mCherry::HP1a and either N-BAF, C-BAF, or N-BAF^T4A/S5A, and in control S2 cells co-immunostained with αHP1a antibodies. We observed that, in comparison to N-BAF expressing cells or control S2 cells, the proportion of HP1a foci in the peripheral region was higher in cells expressing C-BAF or N-BAF^T4A/S5A (Fig. 6B). Concomitantly, while centromeric heterochromatin normally coalesce into few foci, we observed that cells expressing C-BAF or N-BAF^T4A/S5A have a higher number of foci than N-BAF expressing cells (Fig. 6A and C). These results suggest that impairing BAF phosphorylation reinforces anchoring of centromeric heterochromatin to the NE and disrupts its structural properties.

The association of BAF with chromatin is not restricted to centromeric heterochromatin. In fact, genomic profiling experiments have shown that BAF localizes across chromatin, being enriched at LADs [75], and a general contribution to chromatin anchoring at these sites has been proposed [4, 11–13]. Notably, ChIP-seq experiments showed strong co-localization of C-BAF and N-BAF^T4A/S5A with endogenous BAF and Lamin across chromatin (Fig. 6D), indicating that non-phosphorylatable BAF forms preserve the association with LADs. On the other hand, we found that N-BAF is rather uniformly distributed across chromatin (Fig. 6D). Altogether, these results suggest that impairing BAF phosphorylation likely reinforces chromatin anchoring at LADs as well.

Non-phosphorylatable BAF forms delay mitosis

Previous studies indicate that BAF plays a role in cell cycle regulation [45, 49, 76]. In particular, it has been reported that, in mammalian cells, VRK1 depletion delays mitosis [45], suggesting that BAF phosphorylation is crucial for mitotic progression. To directly address this question, we performed transient expression experiments of C-BAF and N-BAF in S2 cells. In these experiments, both forms were expressed at comparable levels (Supplementary Fig. S1G), in contrast to the situation observed in the stable cell lines, where C-BAF is expressed at a markedly low level (Supplementary Fig. S1A–C). FACS analysis reveals an increased proportion of G2/M cells upon expression of C-BAF (Fig. 7A). A similar increase is detected upon expression of the phospho-dead N-BAF^T4A/S5A mutant (Fig. 7A), whereas expression of the phospho-mimetic C-BAF^T4E/S5E mutant reverts this phenotype (Fig. 7A). Next, we performed immunostaining with αH3PS10 antibodies to identify mitotic cells. Compared to mock-transfected control cells or cells expressing N-BAF, the proportion of αH3PS10-positive cells increased in C-BAF-expressing cells (Fig. 7B and C), with a similar increase observed in cells expressing the phospho-dead N-BAF^T4A/S5A mutant (Fig. 7B and C). Altogether, these results further confirm that impaired BAF phosphorylation delays mitotic progression.

Figure 7.

Non-phosphorylatable BAF forms delay mitosis. (A) The proportion of G2/M cells determined by FACS is shown for S2 cells transiently expressing N-BAF, C-BAF, N-BAF^T4A/S5A, or C-BAF^T4E/S5E, and for control mock transfected S2 cells. Results are from a single experiment (N = 3818 for N-BAF, 6565 for C-BAF, 6764 for N-BAF^T4A/S5A, 818 for C-BAF^T4E/S5E, and 59 729 for control; P-values: ns > .05; * < .05; **** < .0001; Chi-square test). (B) Immunostaining with αH3PS10 (in red) of cells transiently expressing N-BAF, C-BAF, or N-BAF^T4A/S5A. N-BAF, C-BAF, and N-BAF^T4A/S5A signals (in green) are direct fluorescence. DNA was stained with DAPI (in gray). Scale bars are 10 μm. (C) Quantification of the results shown in panel (B). The proportion of αH3PS10 positive cells is shown for S2 cells transiently expressing N-BAF, C-BAF, or N-BAF^T4A/S5A, and for control mock transfected S2 cells. (N = 389 for N-BAF, 275 for C-BAF, 235 for N-BAF^T4A/S5A, and 748 for control; P-values with respect to control: **** < .0001; Fisher’s exact test).

In this context, it is noteworthy that, as previously reported [49], expression of C-BAF reduces centromeric levels of the constitutive centromere-associated protein CENP-C, which is essential for centromere function and kinetochore assembly, and, consequently, for mitotic progression (reviewed in [77]). Notably, we observed that expression of the phospho-dead N-BAF^T4A/S5A form reduces centromeric CENP-C to a similar level as C-BAF expression (Supplementary Fig. S5A and B). Furthermore, low centromeric CENP-C levels in C-BAF expressing cells are rescued in cells expressing the phospho-mimetic C-BAF^T4E/S5E mutant (Supplementary Fig. S5A and B). WB analysis showed only minor differences in total CENP-C levels (Supplementary Fig. S6A and B). Altogether, these results indicate that impaired BAF phosphorylation interferes with centromeric CENP-C deposition and, thus, functional centromere assembly.

Non-phosphorylatable BAF forms compromise NE integrity

During mitosis, BAF is released from chromatin, which helps to resolve chromatin anchoring to the NE and is a necessary step in NE breakdown [40, 42, 46, 72]. Previous studies have shown that BAF is retained on mitotic chromosomes upon depletion of VRK1/NHK1 [40, 45], suggesting that its release requires phosphorylation. Consistent with this, we observed that the non-phosphorylatable C-BAF and N-BAF^T4A/S5A forms remain associated with chromosomes during mitosis, in contrast to N-BAF, which is undetectable on mitotic chromosomes (Fig. 8A). This differential behavior is most clearly illustrated in the cell line co-expressing both YFP N-BAF and mCherry N-BAF^T4A/S5A, in which N-BAF^T4A/S5A remains bound to chromosomes throughout mitosis, while N-BAF does not (Fig. 8B and Supplementary Movie S2).

Upon exiting mitosis, BAF plays a central role in NE reassembly [29, 40, 46, 47, 78, 79]. At late anaphase, and before the recruitment of any other NE-associated protein tested, BAF is detected at the core regions that nucleate NE reassembly. Subsequently, BAF spreads around the bulk of decondensing chromosomes, ER vesicles are recruited, and the NE reforms. Notably, we observed that, while N-BAF localizes to core regions at late anaphase, the non-phosphorylatable BAF forms remain associated with the bulk of chromosomes with no detectable enrichment at core regions (Fig. 8B, white arrowheads), suggesting that NE reassembly is affected.

Indeed, compared to cells expressing N-BAF, those expressing C-BAF or N-BAF^T4A/S5A show a higher incidence of NE defects (Fig. 8C and D; see also Fig. 6A). These defects include morphological abnormalities, increased nuclear blebbing, micronuclei formation, and Lamin discontinuities. Interestingly, Lamin discontinuities preferentially occur at regions enriched in C-BAF or N-BAF^T4A/S5A (Fig. 8E). In fact, relative αLamin intensity at BAF-enriched regions is lower in C-BAF and N-BAF^T4A/S5A expressing cells than in cells expressing N-BAF (Fig. 8F). We also observed that total Lamin levels are reduced in C-BAF expressing cells, and to a lesser extent in N-BAF^T4A/S5A expressing cells (Supplementary Fig. S6A and C).

Altogether, these results indicate that impairing BAF phosphorylation prevents its release from chromosomes at mitosis, thereby interfering with NE breakdown and reassembly, which compromises NE integrity and, ultimately, leads to nuclear instability.

Discussion

Here, we report that the highly conserved NL-associated protein BAF associates with centromeric heterochromatin and plays a central role in regulating its anchoring to the NE. BAF’s role in connecting chromatin to the NL is well established, as it has been shown to interact with both chromatin [19, 35–37] and NL components [26–30]. However, its specific contribution to the anchoring of centromeric heterochromatin has remained unknown. BAF is distributed around the entire NE, forming a perinuclear ring beneath the NL. Unexpectedly, we found that, while an N-terminal YFP fusion (N-BAF) faithfully recapitulates the localization pattern of endogenous BAF, a similar C-terminal fusion (C-BAF) accumulates at centromeric heterochromatin and is undetectable at the perinuclear ring. We also show that, at centromeric heterochromatin, C-BAF exchanges at an extremely slow rate, resulting in reinforced anchoring to the NE.

Constrained localization of BAF to centromeric heterochromatin and its slow exchange dynamics are associated with impaired phosphorylation since C-BAF is not phosphorylatable, and, notably, the phospho-dead N-BAF^T4A/S5A mutant also accumulates at centromeric heterochromatin and exhibits slow exchange. This is consistent with in vitro studies demonstrating that phosphorylation negatively regulates BAF’s binding to DNA [39– 43, 71]. More specifically, Marcelot et al. [43, 71] have recently shown that phosphorylation increases the rigidity of the BAF’s N-terminal region, which is involved in DNA binding, thereby impairing its interaction with DNA. However, in vivo evidence regarding the effects of BAF phosphorylation on chromatin binding has been limited, largely inferred from studies involving depletion of VRK1/NHK1, the main kinase that phosphorylates BAF [39, 42, 44]. These studies have shown that, while BAF is normally released from chromatin during mitosis [40, 42, 46, 72], it remains bound to mitotic chromosomes when VRK1/NHK1 is depleted [40, 45]. Here, we directly confirm this hypothesis since the non-phosphorylatable C-BAF and N-BAF^T4A/S5A forms are, indeed, retained on mitotic chromosomes in wild-type cells. Our findings eliminate potential confounding indirect effects arising from VRK1/NHK1 depletion and support a direct role for phosphorylation in regulating BAF dissociation from chromatin in vivo. We noted that, at centromeric heterochromatin, C-BAF is exchanged at a slower rate than N-BAF^T4A/S5A, suggesting that additional factors may contribute to its persistent interaction with centromeric heterochromatin. In this regard, it is worth noting that C-BAF is predicted to dimerize via the YFP moiety (see Fig. 4), leaving the BAF fold available to mediate dimer-dimer interactions and promote the formation of higher-order oligomers. Such oligomerization could potentially account for the markedly slow exchange dynamics of C-BAF.

Our results further indicate that phosphorylation plays a pivotal role in regulating the subnuclear localization of BAF, as the non-phosphorylatable C-BAF and N-BAF^T4A/S5A forms are not detectable by immunostaining at the perinuclear ring beneath the NL. Notably, the majority of perinuclear BAF does not appear to participate in chromatin binding, since chromatin association of both C-BAF and N-BAF^T4A/S5A is readily detected by ChIP-seq. This observation suggests that perinuclear BAF localization primarily depends on interactions with NL components, such as LEM domain proteins and Lamins [26–30]. Interestingly, recent work shows that these interactions are largely preserved upon BAF phosphorylation [39, 42, 43]. The redistribution of non-phosphorylatable BAF forms likely reflects a reduced affinity for NL components compared to centromeric heterochromatin, or chromatin in general, which, given their low expression levels, may favor their accumulation at centromeric heterochromatin. In this context, it is noteworthy that non-phosphorylated BAF has been shown to form hexamers in the presence of DNA in vitro [19], a property that may enhance its affinity for chromatin. Intriguingly, ChIP-seq analysis failed to detect N-BAF enrichment at LADs. Notably, Phos-tag gel electrophoresis reveals a higher proportion of phosphorylated forms for N-BAF compared to endogenous BAF (see Supplementary Fig. S2A). In line with this, due to additional contacts mediated by the YFP moiety, the predicted Kd for the N-BAF-NHK1/VRK1 complex is several orders of magnitude lower than that of endogenous BAF (see Fig. 4). Together, these observations suggest that N-BAF is more extensively phosphorylated than endogenous BAF, potentially resulting in a more dynamic and transient interaction with chromatin. Further work will be required to clarify these questions.

Results reported here and elsewhere [49] indicate that impaired BAF phosphorylation is deleterious to cells. Here, we have identified multiple defects associated with the expression of non-phosphorylatable BAF forms. Notably, their expression correlates with enhanced anchoring of centromeric heterochromatin to the NE and delayed mitosis. Chromatin-NE attachments are normally resolved at mitosis [80, 81]. While multiple pathways contribute to this regulation, our findings highlight a critical role for BAF phosphorylation in the release of centromeric heterochromatin from the NE, thereby facilitating progression into mitosis. Additionally, we show that expression of non-phosphorylatable BAF forms leads to reduced levels of centromeric CENP-C, impairing functional centromere assembly and disrupting mitotic progression. This defect is likely linked to the reinforced anchoring of centromeric heterochromatin. As centromeres are embedded within centromeric heterochromatin, they tend to cluster and anchor at the NE. In Drosophila S2 cells, centromeric CENP-C deposition occurs primarily during interphase [82]. Interestingly, several observations suggest a functional link between centromeres and nucleoli. Centromeres frequently localize to the nucleolar periphery [83, 84], and key centromeric proteins, including CENP-C and the centromere assembly factor HJURP/CAL1, localize to both centromeres and nucleoli [85–89]. Moreover, in Drosophila, the nucleolar protein Modulo interacts with HJURP/CAL1 and is required for the deposition of both HJURP/CAL1 and CENP-A at centromeres [90]. Although speculative, these observations raise the possibility that centromere-nucleolus interactions may be required for proper CENP-C deposition. Reinforced anchoring of centromeric heterochromatin to the NE could interfere with this interaction. In this context, BAF phosphorylation appears crucial for releasing centromeres from the NE, thereby facilitating centromere-nucleolus interactions and promoting efficient CENP-C loading.

Our results also show that impaired BAF phosphorylation compromises NE integrity and nuclear stability. Several factors likely account for these effects. First, chromatin-NE interactions are known to contribute to the mechanical properties of the nucleus (reviewed in [91]). Notably, altering heterochromatin has been shown to drive nuclear softening, thereby protecting the genome from mechanical stress [92]. Reinforced anchoring and fragmentation of centromeric heterochromatin when BAF phosphorylation is impeded likely affects nuclear stiffness, increasing rigidity of chromatin-NE interactions and, thus, nuclear fragility. On the other hand, expression of non-phosphorylatable BAF forms disrupts the organization and definition of the core regions, which are critical for post-mitotic NE reassembly [78, 93], and for chromatin compaction, reorganization, and anchoring to the NE after mitosis [16, 29, 78]. Actually, alterations in core/non-core organization have been shown to result in aberrant/unstable NE structures, ultimately compromising nuclear integrity [29, 93].

Interestingly, we observed that expression of non-phosphorylatable BAF forms increases the frequency of Lamin discontinuities, indicative of a high incidence of NE ruptures. Previous studies have shown that an early event in the repair of NE ruptures involves the recruitment of unphosphorylated BAF, which binds to exposed chromatin and prevents its leakage into the cytoplasm [24, 25]. Consistent with this, the Lamin discontinuities we observed are enriched in non-phosphorylatable BAF forms, suggesting that these forms are indeed recruited to rupture sites. However, our results further suggest that non-phosphorylatable BAF forms are defective in mediating NE repair. This raises the possibility that successful repair requires dynamic BAF-chromatin interactions, regulated by phosphorylation. Further work is required to answer these questions.

In summary, our results unveil the crucial contribution of BAF phosphorylation to the dynamic regulation of heterochromatin-NE interactions. Cycles of BAF phosphorylation and dephosphorylation appear essential for proper anchoring of centromeric heterochromatin to the NE during interphase. Disruption of this regulatory mechanism leads to reinforced anchoring, delayed mitosis and nuclear instability. The resulting nuclear defects resemble those in BAF-related NGPS [94] and, more broadly, in laminopathies [95], suggesting that impaired BAF phosphorylation may represent a more widespread pathogenic mechanism in these disorders than previously recognized.

Notes

Present address: Department of Genetics, IMO Miranza Group, Josep María Lladó, 3, 08035 Barcelona, Spain

Present address: Biophysics and Bioengineering Unit, Biomedical Department, University of Barcelona, 08014 Barcelona, Spain

Present address: Barcelona Supercomputing Center (BSC-CNS), 08034 Barcelona, Spain

Present address: Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, BC V6T1Z3 Vancouver, Canada

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

MICIN/AEI [10.13039, PGC2018-094538-B-100, PID2021-123303NB-I00] and Generalitat de Catalunya [SGR2017-475]. Funding to pay the Open Access publication charges for this article was provided by MICIN/AEI [PID2021-123303NB-I00].

Data availability

The accession number of the ChIP-seq data generated in this work is GEO (GSE298230). Scripts for the image analyses, together with example data, usage instructions, and parameter settings, have been deposited at https://zenodo.org/records/18172795. The rest of the data underlying this article will be shared on reasonable request to the corresponding authors.