The Zinc-Finger Protein POGZ Associates with Polycomb Repressive Complex 1 to Regulate Bone Morphogenetic Protein Signaling During Neuronal Differentiation

Abstract

Polycomb Repressive Complex 1 (PRC1) is a group of epigenetic regulatory complexes critical for mammalian development. Elucidating PRC1 composition and function across cell types and developmental stages is key to understanding the epigenetic regulation of cell fate determination. In this study, we discovered POGZ, a prominent autism spectrum disorder (ASD) risk factor, as a novel component of PRC1.6, forming the PRC1.6-POGZ complex. Functional assays revealed that POGZ elicits transcriptional repression that is dependent on RING1B expression. Analysis of publicly available data showed that POGZ highly colocalizes with RING1B and HP1g, two PRC1.6 components, at genes involved in multiple aspects of transcriptional regulation in embryonic mouse cortical cells. Although Pogz knockout (KO) does not compromise stem cell pluripotency, Pogz ablation in neuronal progenitor cells (NPCs) led to widespread transcriptomic dysregulation with failed activation of key neuronal genes. Finally, we demonstrated that PRC1.6-POGZ regulates neuronal differentiation by repressing bone morphogenetic protein (BMP) signaling. These findings reveal a mechanism by which PRC1 and POGZ coordinate transcription during neuronal differentiation and demonstrate that disrupting this complex impairs BMP signaling, potentially contributing to neurodevelopmental disorders such as ASD.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12015-025-11028-x.

Introduction

Polycomb group (PcG) proteins are key epigenetic modifiers that regulate transcription by modulating chromatin structure and catalyzing histone modifications [1]. These functions are essential for critical processes in mammalian development, such as stem cell self-renewal and differentiation [2–6]. PcG proteins assemble into two major protein complexes named Polycomb Repressive Complex 1 (PRC1) and 2 (PRC2) [7], which regulate gene expression through distinct repressive chromatin modifications. PRC1 deposits histone H2A mono-ubiquitination (H2AK119ub1), while PRC2 catalyzes H3 lysine 27 mono-, di-, and tri-methylation (H3K27me1/2/3) [8, 9].

Mammalian PRC1 complexes display extensive subunit diversity, arising from multiple homologs of core proteins initially identified in Drosophila melanogaster [10–12]. This diversity includes two E3 ligases (RING1A/B), five chromodomain proteins (CBX2/4/6/7/8), three PH homologs (PHC1-3), and six Polycomb group RING finger proteins (PCGF1-6) [7, 13]. PRC1 is further characterized into six major subcomplexes (PRC1.1–1.6), each defined by the exclusive association of one of PCGF1-6 and RING1A/B [14]. Additionally, canonical PRC1 complexes associate with CBX and PHC proteins, while non-canonical complexes incorporate RYBP/YAF2 [14–17]. Although previous studies laid the groundwork for understanding PRC1 assembly, the distinct roles of subcomplexes in development and differentiation remain relatively unclear.

Specific PRC1 subcomplexes have previously been shown to play unique roles in neurodevelopment and differentiation. For instance, AUTS2, a gene often disrupted in individuals suffering from neurological disorders [18, 19], was identified as a component of the PRC1.5-AUTS2 complex in HEK 293 T-REx cells [14]. Subsequent studies discovered that PRC1.5-AUTS2 acts as a transcriptional activator in both in vitro and in vivo models [20–22], thus broadening the repertoire of traditional PRC1 function. Despite these advancements, further investigation is needed to dissect the potential contributions of other PRC1 subcomplexes in neuronal cells.

Disruptions in POGZ (POGO-transposable element with ZNF domain) are strongly associated with ASD and White-Sutton syndrome [23–26]. POGZ has been shown to play a role in neurite development in primary cortical neurons [27] and in neuronal maturation of iPSC-derived neurons [28]. POGZ is also proposed to have transcriptional activity due to its multiple predicted zinc finger domains [29] and sequence-specific DNA-binding properties [30]. Indeed, recent research has highlighted a dichotomy in POGZ’s transcriptional roles, demonstrating its influence on both transcriptional activation and repression [31–34]; however, the underlying mechanisms remain unclear.

In this study, we uncovered the surprising interaction between POGZ and PRC1 in primary murine neuronal cells. Biochemical analyses revealed that POGZ specifically associates with the PRC1.6 complex, prompting us to name this complex PRC1.6-POGZ. Functionally, POGZ exhibited transcriptional repression activity that is dependent on RING1B expression. The formation of the PRC1.6-POGZ complex was further supported by genomic colocalization of POGZ, RING1B, and HP1g, key components of PRC1.6, in embryonic cortical cells. Interestingly, while Pogz is not required for stem cell pluripotency, its loss impairs differentiation to NPCs. Furthermore, during neuronal differentiation, Pogz loss leads to activation of the BMP pathway. These findings uncover an exciting new link between POGZ, a putative ASD risk gene, and PRC1, key epigenetic regulators, in controlling a critical signaling pathway for neurodevelopment.

Materials and Methods

Preparation of Cortical Neurons and Neuronal Progenitors

Primary cortical neurons were cultured as previously described [35]. Briefly, pregnant female mice were euthanized on E16.5. Embryonic cortices were harvested, dissociated, and plated (~ 6 × 10⁶ cells) onto poly-D-lysine (0.1 mg/mL, Sigma, P6407) coated 10 cm cell culture plate (Corning, 353003) and incubated in a cell culture incubator with 5% CO2, at 37 °C. Media was replaced with lysine-free Neurobasal media (AthenaES, 0500 − 118) supplemented with L-Lysine-13C6,15N2 hydrochloride at 73 mg/500 mL after one hour of undisturbed incubation in plating media. On DIV 2, cells were treated with cytosine ß-D-arabinofuranoside (ara-c, Sigma, C1768) at a final concentration of 0.5 µM. On DIV 3, one-half of the media was changed for fresh media. Media was changed every three days until collection at DIV17. The purity of the primary cortical neurons is approximately 90% [35]. Neuronal progenitor cells were generated in vitro using the differentiation process described in the Neuronal differentiation section below.

Neuronal Differentiation

Neuronal differentiation was performed as previously described [36, 37]. ESCs cultured on mouse embryonic fibroblasts were split into gelatin-coated plates for two passages before differentiation. The differentiation medium contained DMEM medium, 15% FBS, non-essential amino acids, β-mercaptoethanol, L-glutamine, penicillin/streptomycin, and sodium pyruvate. On Day 0, ESCs differentiated using the hang-drop method. Each 20 µL droplet contained ~ 1500 ESCs. The hanging drop plates were placed in a 37 °C incubator with 5% CO₂ content. On day 2, the ESCs will be differentiated into EBs. EBs were then transferred to a suspension plate containing differentiation medium for two additional days. On Day 4, 5µM retinoic acid was added to the culture medium to trigger differentiation towards the neuronal lineage. By Day 8, NPC differentiation yields approximately 60% purity [36]. NPCs were then washed with 1X PBS, pelleted, and collected for various downstream analyses.

RING1B Immunoprecipitation (IP) Sample Preparation and Nano LC-MS/MS Analysis

To conjugate antibodies to beads, a 3:1 Protein A/G mixture was washed using 1X PBS. Beads were then incubated with 100 µg of antibody for 3 h at RT or overnight (O/N) at 4 °C. The antibodies used are RING1B (Bethyl, cat. no. A302-869 A-T) and IgG (Cell Signaling, cat. no. 2729 S). Beads were then washed twice with 0.2 M Sodium borate buffer, pH 8.0. Crosslinking was performed using 20 mM DMP (prepared in 0.2 M sodium borate buffer), followed by rotation at RT for 2 h. Beads were centrifuged and washed three times with 0.2 M ethanolamine-HCl, pH 8.0, to stop the reaction. Next, 0.1 M glycine, pH 2.5, was added to the beads, followed by washes with 1 M Tris, pH 8.0. Beads were washed three times with PBS.

To perform the IP, E17.5 brain, cortical neurons, and neuronal progenitor cells were extracted from mice and processed using the nuclear extraction protocol below. 1 mg of nuclear extract and beads were incubated O/N at 4 °C with rotation. The lysate-bead mix was then centrifuged and washed with Buffer C three times with rotation at 4 °C. For the last wash, the beads were resuspended in 1X PBS and transferred to a Micro-spin column (Pierce, cat no. 89879). Samples were centrifuged, then eluted with 0.1 M glycine, pH 2.5, and neutralized with 10 µL of 1 M Tris, pH 9.5. Each elution is performed with an equivalent volume of glycine to beads. Elution was repeated three times.

Immunoprecipitated samples were digested with trypsin (sequencing grade, Thermo Scientific, Cat# 90058) overnight at 37 °C per manufacturer’s instructions. Peptides were extracted, washed twice with a solution of 5% formic acid in 60% acetonitrile, and dried under vacuum. The samples were then analyzed using nano-liquid chromatography coupled to tandem mass spectrometry (nano LC-MS/MS) on a Q-Exactive HF mass spectrometer (Thermo Scientific) connected to an Ultimate 3000 RSLCnano chromatography system (Thermo Scientific).

Peptides were initially loaded onto a fused silica trap column (Acclaim PepMap 100, 75 μm x 2 cm, Thermo Scientific) and washed for 5 min with 0.1% trifluoroacetic acid (TFA) at a flow rate of 5 µL/min. The trap column was then switched in-line with an analytical column (Nanoease MZ peptide BEH C18, 130Å, 1.7 μm, 75 μm x 250 mm, Waters) for LC-MS/MS separation. Peptides were fractionated using a linear gradient with Solvent A (0.2% formic acid) and Solvent B (0.16% formic acid in 80% acetonitrile) at a flow rate of 300 nL/min, using the following scheme: 4–15% Solvent B over 30 min, 15–25% Solvent B over 40 min, 25–50% Solvent B over 44 min, and 50–90% Solvent B over 11 min.

Mass spectrometry data were acquired using a data-dependent acquisition method. An MS1 scan (resolution 120,000) was followed by MS/MS scans (resolution 30,000; HCD collision energy 27%) on the 20 most intense ions. Dynamic exclusion was set to 20 s. Peak lists (MGF files) were generated using Thermo Proteome Discoverer (version 2.1) and searched against the UniProt database and a common contaminant database (cRAP) using a local installation of X! Tandem (GPM Furry). Search parameters included protein and peptide log10 expectation scores of less than − 4 and − 2, respectively, fragment mass error of 20 ppm, and parent mass error of ± 7 ppm. The peptide false-positive rate was 0.4%. The IgG and RING1B IP peaks were evaluated with peptides that had 2 or more counts in the IgG excluded. Peptides were then ranked and compared across RING1B samples to find peptides common in the NSC, neurons, and embryonic brain samples.

Nuclear Extraction

The IP and glycerol gradient experiments described herein were performed using nuclear extracts (NE). NEs were prepared as previously described [14]. Briefly, the buffers used for nuclear extraction are Buffer A (10 mM Tris-HCl, pH 7.9, 1.5 mM MgCl₂, 10 mM KCl) and Buffer C (20 mM Tris-HCl, pH 7.9, 1.5 mM MgCl₂, 420 mM NaCl, 0.2 mM EDTA, 25% glycerol). The following protease inhibitors were added before every experiment to protect samples from proteolysis: 0.2 mM PMSF, 1 µg/ml Pepstatin A, 1 µg/ml Leupeptin, and 1 µg/ml Aprotinin. Cell pellets were first resuspended using Buffer A, incubated on ice for 15 min, and homogenized using a 2mL dounce homogenizer for 15 strokes. After 30 min rotation at 4 °C, the samples are centrifuged at 15,000 rpm for 15 min at 4 °C. The supernatant (cytosolic fraction) was removed, and the remaining pellet was resuspended in Buffer C. After resuspension, the lysate was homogenized, rotated at 4 °C for 30 min, and centrifuged at 15,000 rpm for 15 min. The resulting supernatant, containing the NE, was then subjected to various downstream analyses.

Cell lines

HEK 293 T-REx cells stably expressing pINTO-NFH-tagged proteins were generated and maintained as previously described [14]. Briefly, pINTO-N-FH-fusion proteins or vector control (pINTO-NFH) were transfected into HEK 293 T-REx cells using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s protocol. Cells were selected with 200 µg/mL zeocin and 10 µg/mL blasticidin. After selection, cells were maintained in medium containing 50–100 µg/mL zeocin and 10 µg/mL blasticidin. Luciferase reporter cell lines expressing GAL4-fusion proteins or vector control (pINTO-GAL4) were generated and maintained as previously described [14]. Briefly, GAL4-fusion proteins or vector control (pINTO-GAL4) were transfected into HEK 293 T-REx -luciferase cells containing a stably integrated 5XGal4RE-tk-Luc-neo construct and selected with 200 µg/ml zeocin, 10 µg/ml blasticidin, and 400 µg/ml G418 [38]. After selection, cells were maintained in medium containing 50–100 µg/mL zeocin and 10 µg/mL blasticidin. All HEK 293 T-REx cell lines were maintained in standard DMEM medium containing 10% FBS (Atlanta Biologicals, Cat# S11050), L-glutamine, and penicillin/streptomycin.

Immunoprecipitation

Immunoprecipitation experiments were performed as described previously [14]. To perform immunoprecipitation, NEs are incubated with pre-washed FLAG M2 beads (Sigma, cat. no. A2220) O/N at 4 °C. After the O/N rotation, the beads were washed five times with Buffer W (1/3 volume of Buffer A, 2/3 volume of Buffer C, 0.02% IGEPAL, 0.2 mM PMSF, 1 µg/ml Pepstatin A, 1 µg/ml Leupeptin, and 1 µg/ml Aprotinin). To elute, 40 µL of Buffer C and 10 µL of 5X SDS were added to each sample, and the mixture was boiled for 5 min at 95 °C. Samples were centrifuged, and the supernatant was collected for immunoblotting. All blots were developed on a ChemiDoc MP Imaging System (BioRad, cat. no. 12003154) using SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific, cat. no. 34094).

Proximity Ligation Assay

For PLA assays, HeLa cells were seeded into 8-chamber slides. Cells were permeabilized with 0.5% Triton for 10 min at 4 °C, washed with PBS, fixed at room temperature with 3% paraformaldehyde in PBS for 10 min, washed again in PBS, and then blocked in Duolink blocking solution (Millipore Sigma DUO82007) for 1 h at 37 °C, and incubated overnight at 4 °C with primary antibodies. Antibodies used were: POGZ (Abcam, cat. no. ab171934) and RING1B (MBL, cat. no. D1393). Samples were then subjected to a proximity ligation reaction using the Duolink kit (Millipore Sigma DUO92008) according to the manufacturer’s instructions. Slides were imaged using a confocal microscope (Leica SP5), and images were analyzed using ImageJ 1.53a software.

Glycerol Gradient Analysis

Glycerol gradient analysis was performed as previously described [14]. To perform glycerol gradient analysis, 30–15 cm plates per HEK 293 T-REx cell line were treated with 1 µg/mL doxycycline for 24 h. The resulting pellet was processed to obtain the NE (see above). Volumes for Buffer A and Buffer C were adjusted to 15 mL and 13.5 mL, respectively. 12 ml NE was mixed with 3 ml Buffer A, 0.02% NP-40, and 200µL of pre-washed FLAG M2 beads. After rotation at 4 °C O/N, the M2 beads were washed five times with Buffer W and then eluted with 500 µL of 250 µg/ml FLAG peptide in Buffer W by rotating at 4 °C for 1 h. The M2 eluate was added on top of a 4.5 ml 15–35% glycerol gradient and centrifuged in an SW60Ti rotor (Beckman) at 18,000xg at 4 °C for 16 h. The resulting gradient was then fractionated every 180 µl. Odd-numbered fractions were then subjected to SDS-PAGE and immunoblotting.

Luciferase Reporter Assay

To perform the luciferase assays, cell lines were plated in 6-well plates in sextuplicate at 300,000 cells per well on Day 0. Three wells were treated with either 1 µg/mL doxycycline or vehicle on Day 1 for 24 h. The cells were then assayed on Day 2 using Promega’s Luciferase Assay System (cat. no. E1500) per the manufacturer’s instructions, except that the luciferase assay reagent was added manually. Luminescence was measured using the Promega GloMax 96 Microplate Luminometer (cat. no. E6501). Luciferase results were normalized to protein concentration and then analyzed by comparing luminescence fold changes between induced and non-induced conditions per cell line. Fold changes were then compared by a two-sample t-test.

RNA Interference

Cells were transfected with siRNAs using Lipofectamine RNAiMAX (Life Technologies) according to the manufacturer’s protocol. Human siRNAs used in this study are si-ctrl (Qiagen, AllStars Negative Control siRNA, cat no. SI03650318), si-RING1B (Qiagen, cat no. SI00095543), and si-SETDB1 (Horizon, L-020070-00-0010).

Cut&RUN/TAG Analysis

Publicly available sequencing datasets for POGZ, HP1g, and RING1B were used in this study. The POGZ WT/KO and HP1g CUT&RUN data sets were generated from dissected E13.5 telencephalon (NCBI GEO Series GSE187010) [31]. The RING1B CUT&Tag data set was generated from isolated E11 neural precursor cells from the mouse cortex (DNA Data Bank of Japan Sequence Read Archive, accession no. DRA010296) [39]. Sequencing results were mapped using the mm10 genome using BWA [40]. Duplicated reads were removed using Samtools [41]. Data sets were normalized by library size. BamCoverage and computeMatrix from deepTools were used to generate the normalized BigWig files and ChIP-Seq heatmaps, respectively [42]. Macs2 callpeak was used to generate narrowPeak files [43]. Bedtools Intersection was used to identify PRC1.6-POGZ, POGZ & RING1B, POGZ & HP1g, and POGZ-only bound loci [44]. ChIP-seq read density files were generated using igvtools and viewed in Integrative Genomics Viewer (IGV) [45]. Genomic peak annotation was performed using the R package ChIPseeker [46]. Gene annotations were obtained using the genomic regions enrichment of annotations tool (GREAT) [47]. The single nearest gene within 100 kb was used for gene annotation. GO analysis was performed using DAVID [48].

CRISPR/Cas9-Mediated Gene Editing

Oligos corresponding to candidate sgRNA sequences were purchased from IDT and cloned into pX458 (Addgene) as previously described [49]. The sgRNA sequences used were: sgRNA1 (5’-ACTTGTGGGACGCCAACTGTCGG-3’) and sgRNA2 (3’-CTAGTTTGGGATTCGAGGTCTGG−5) (PAM trinucleotides are underlined). Plasmids carrying sgRNAs were transfected into ESCs using Lipofectamine (Life Technologies). Two days after transfection, cells were sorted into 96-well plates coated with gelatin at one cell per well. Clones were tested by genomic PCR using primers: 5′- TTCCTTTTGTTCAGTGGACAGC-3′, and 3′-AAGGAGGGAGCTACATTGACC-5′. Positive clones were further validated by Sanger sequencing and immunoblotting using a POGZ antibody (Bethyl, cat# A302-510 A).

ESC Culturing

ESCs and CRISPR-engineered ESCs were cultured in embryonic stem cell medium comprised of DMEM medium, supplemented with 15% FBS (ES certified, Atlanta Biologicals, cat. no. S10250), LIF, non-essential amino acids, β-mercaptoethanol, L-glutamine, penicillin/streptomycin, sodium pyruvate, and two small-molecule kinases (MEK and GSK3) inhibitors (PD0325901, Cayman Chemical, cat. no. 13034, and CHIR99021, Cayman Chemical, cat. no. 13122). When revived from liquid nitrogen, ESCs are initially plated on mitotically arrested mouse embryonic fibroblast feeder cells. ESCs are split every 48 h on gelatin-coated plates.

RT-qPCR

Total RNA was extracted with TRIzol and used to synthesize cDNA with the SuperScript III system (Invitrogen, Cat# 18080-044). The resulting cDNA was mixed with Brilliant III Ultra-Fast SYBR QPCR master mix (Agilent Cat#600883) and primers. The mixture was then run on a Biorad CFX Connect real-time PCR detection system. The RT-qPCR primers used in this study are listed in Table S4.

Immunofluorescence

Differentiated NPCs were fixed with 4% paraformaldehyde at room temperature for 20 min, followed by PBS washes. NPCs were incubated in 10%, 20%, and 30% sucrose solutions (diluted in 1X PBS) sequentially with overnight incubation at 4 °C. NPCs were embedded in O.C.T compound (Sakura) and cryosectioned at 5 µM thickness. For immunofluorescence imaging, sections were washed 3 times with 1X PBS, incubated in 50 µM glycine for 15 min at RT, then washed 3 times with 1X PBS. Sections were then blocked and permeabilized in 0.5% Triton X-100 and 5% BSA in PBS for 1 h at RT. After three washes, sections were incubated overnight at 4 °C with a Nestin primary antibody (BD Transduction Laboratories, cat. No. 611658, 1:100 dilution) in 0.1% Triton X-100 and 1% BSA in PBS. After washing 3 times in PBS, sections were incubated with Alexa Fluor 488 anti-mouse antibody (Gibco, 1:500 dilution) for 1 h at room temperature. Slides were washed 3 times, stained with DAPI, and mounted for imaging. Images were captured using a Zeiss Axio Imager M2 microscope.

For image quantification, the fluorescence images were loaded and converted to greyscale in Image J. NPC sections were selected using drawing tools. The integrated fluorescence intensity of sections was measured by Image J. For each image, five random regions outside NPC sections were measured and averaged to obtain the background fluorescence intensity. The corrected fluorescence intensity of each NPC was equal to the integrated intensity minus the background intensity.

RNA-Seq and Analysis

RNA-Seq was performed on POGZ WT and KO ESCs and NPCs. Two biological replicates were used per genotype and cell type. RNA was extracted from 500,000 cells for ESC samples. For NPCs, 2 million cells were differentiated, and RNA was extracted from the resulting NPCs. For each sample, 500ng total RNA was used to prepare cDNA libraries with the NEXTflex™ Illumina Rapid Directional RNA-Seq Library Prep Kit (Illumina, cat. no. NOVA-5198-0) according to the manufacturer’s instructions. Libraries were loaded onto a TruSeq Rapid flow cell on an Illumina HiSeq 2500 (Genome Sciences Facility, Penn State College of Medicine). The samples were run for 50 cycles using either a single-read or pair-end recipe according to the manufacturer’s instructions. Gene expression abundance was quantified using Kallisto against the mm10 transcriptome reference [50]. Differentially expressed genes (DEGs) were identified using DESeq2 [51]. DEGs were identified using the following thresholds: adjusted p-value < 0.5 and absolute log₂ fold change > 1. The z-scores of the gene expression matrix were used to generate the heatmap. DEGs were sent for GO analysis, which was performed using DAVID [48].

Statistics

Statistical analyses were performed using an unpaired, two-tailed Student’s t-test where applicable for comparison between two groups. All data are presented as the mean ± standard deviation of the mean. Figure legends include the sample size for each experiment and indicate whether an alternative statistical test was used. Statistical significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant.

Supporting Information.

This article contains supporting information, including two figures and four tables.

Results

Identification of POGZ as a Novel PRC1 Subunit in Neuronal Cells

PRC1 complexes have a well-established role in developmental gene regulation [7, 52], but how these complexes are organized in specific lineages, such as neuronal cells, remains poorly understood. To explore PRC1 composition in neuronal cells, we employed an unbiased approach using RING1B, the standard component of PRC1 complexes, as the bait to identify novel PRC1 interactions. Because RING1B is shared across all PRC1 subcomplexes, this approach allows the identification of canonical and non-canonical PRC1 associations as well as potential PRC1-independent interactors. Immunoprecipitation (IP) with RING1B antibody-conjugated Protein A/G beads in nuclear extracts from E17.5 mouse brain, cortical neurons, and neuronal progenitor cells, followed by mass spectrometry, revealed a 50% overlap in the top 50 interactors across the neuronal cell types (Fig. 1a-b, Table S1). Among the common interactors were many known PRC1 components, including nearly every PCGF protein (PCGF1/2/3/4/6), multiple CBX proteins (CBX2/4/8), all PHC proteins (PHC1/2/3), RYBP, and additional subunits specific to each PRC1.1–1.6.6 complex (Fig. 1b) [14, 53]. Particularly interesting was the identification of POGZ, a high-confidence autism risk factor [24–26], as a novel PRC1 interactor (Fig. 1b). Previously, studies have demonstrated the direct association between POGZ and HP1 proteins [32, 33, 54]. Given that HP1γ is a known PRC1.6 subunit, we hypothesized that POGZ would associate with PRC1.6 among all the PRC1 subcomplexes and potentially influence PRC1 function. To determine which of the six PRC1 subcomplexes POGZ associates with, we performed an IP analysis using our previously generated HEK 293 T-REx cell lines stably expressing doxycycline-inducible N-terminal FLAG- and HA-fused PCGF1-6 (NFH-PCGF1-6) [14]. This inducible system offers the advantage of controlling the expression of tagged proteins to endogenous levels, thereby avoiding potential overexpression artifacts. Following doxycycline induction, pull-down with HA beads from each of the PCGF proteins revealed that POGZ specifically associates with the PRC1.6 complex (Fig. 1c). We then performed a reciprocal FLAG IP using the NFH-POGZ doxycycline inducible HEK 293 T-REx cells. The NFH-empty vector cell line was used as the negative control. This FLAG IP confirmed that multiple subunits of PRC1.6, including RING1B, HP1g, and L3MBTL2, co-immunoprecipitated with overexpressed NFH-POGZ (Fig. 1d). To strengthen our findings supporting this novel interaction, we performed a proximity ligation assay (PLA) using POGZ- and RING1B-specific antibodies in wild-type HeLa cells. In this experiment, we detected significant colocalization of endogenous POGZ with RING1B compared to the negative controls (Fig. 1e-f). Negative controls include single antibodies that showed minimal background signal, confirming the specificity of the results. Overall, these data provide evidence for the formation of a novel POGZ-containing PRC1.6 complex, which we name PRC1.6-POGZ.

Fig. 1.

Novel POGZ-Containing PRC1 Complex Identified in the Developing Nervous System. (a) Schematic displaying the RING1B IP experiment followed by mass spectrometry in (left to right) mouse embryonic brain, cortical neurons, and neuronal progenitors. (b) RING1B interactors identified in (a) are grouped by their association with specific PRC1 subcomplexes. (c) HA IP of NFH-PCGF1-6 HEK 293 T-REx cell lines. Bound proteins were resolved on SDS-PAGE and detected by immunoblotting for the indicated antigens (IN = Input). (d) FLAG IP of NFH-POGZ and NFH-Control HEK 293 T-REx cells. The bound proteins were resolved on SDS-PAGE and blotted for the indicated PRC1.6 components. (e) PLA experiments showing that POGZ and RING1B co-localize in HeLa cells. Single antibody conditions are shown as negative controls. At least 100 cells were quantified for each condition. Bars indicate the mean values, error bars represent standard errors of the mean, and asterisks indicate statistical significance (Mann-Whitney). Immunoblotting of POGZ and RING1B to demonstrate their expression in HeLa cells. (f) Representative micrographs of the PLA experiment, with scale bars indicating 10 μm

PRC1.6-POGZ is distinct from other POGZ-containing complexes.

To further examine the PRC1.6-POGZ complex, we performed a FLAG-IP using the inducible NFH-PCGF6 cell line followed by glycerol gradient analysis (Fig. 2a). Every other fraction was resolved on SDS–PAGE, followed by immunoblotting for POGZ and other PRC1.6 components. Our Immunoblotting results show that HA-PCGF6, RING1B, POGZ, HP1g, and RYBP are enriched in similar fractions, indicating that these factors form a stable complex (Fig. 2b). Previous reports have suggested that POGZ associates with other transcriptional regulatory complexes, including the ChAHP (CHD4, ADNP, HP1) complex and the esBAF (embryonic stem cell-specific BRG1-associated factor) complex. The ChAHP complex is a recently identified transcriptional repressor complex required for embryonic stem cell (ESC) maintenance [55]. The esBAF complex is a transcriptional activator complex in ESCs that maintains stem cell identity [30]. To explore POGZ association with epigenetic complexes other than PRC1, we performed glycerol gradient analysis using an NFH-HP1γ HEK293 TREx cell line, since HP1γ is a shared subunit of the PRC1.6-POGZ, ChAHP, and esBAF complexes. Our results revealed that HA-HP1g, RING1B, POGZ, and additional PRC1.6 specific components, L3MBTL2 and WDR5, were enriched in the same fractions, further supporting the formation of the novel PRC1.6-POGZ complex (Fig. 2c). In contrast, CHD4, a core subunit of the ChAHP complex, and BRG1, the core ATPase subunit of the esBAF complex, were enriched in earlier fractions compared to those corresponding to the PRC1.6-POGZ complex (Fig. 2c). Taken together, our data strongly suggests the existence of a new POGZ-associated complex that constitutes a subtype of PRC1 (Fig. 2d).

Fig. 2.

POGZ is a stable component of the PRC1.6 complex. a. Schematic of the glycerol gradient experimental workflow. b-c. Glycerol gradient analysis of FLAG-purified PCGF6 (b) and HP1g (c) complexes. Every other fraction from a 15–35% glycerol gradient was resolved on SDS-PAGE, followed by immunoblotting for the indicated antigens. Red boxes highlight fractions containing PRC1.6-POGZ subunits, and the green box corresponds to non-PRC1 POGZ-containing complexes. d. Schematic illustration of the proposed PRC1.6-POGZ complex. Blue subunits represent previously identified PRC1.6 subunits, and the orange subunit highlights POGZ as a novel PRC1.6 interactor

POGZ-Dependent Repression Requires RING1B

PRC1 complexes are widely regarded as transcriptional repressors; however, in certain circumstances, PRC1 has been demonstrated to possess transcriptional activation activity [20–22]. Thus, it is vital to determine the transcriptional function of novel PRC1 complexes such as PRC1.6-POGZ. To do this, we performed a luciferase assay using a stable HEK 293 T-REx cell line containing an integrated luciferase reporter with five consecutive GAL4 DNA binding sites (UAS) and doxycycline-inducible GAL4–POGZ (Fig. 3a). We also used previously established lines with GAL4 alone or GAL4-PCGF4 [20] as the negative and positive controls, respectively. Upon doxycycline induction, luciferase activity decreased in GAL4–PCGF4 cells (Fig. 3b), aligning with its function in transcriptional repression [20]. Consistent with previous reports, induced expression of GAL4-POGZ also led to a significant decrease in luciferase activity (Fig. 3b), supporting its role as a transcriptional repressor [33]. To further investigate whether POGZ association with PRC1 was essential for its transcriptional repression activity, we silenced RING1B through short interfering RNAs (siRNAs) in GAL4–POGZ cells (Fig. 3c-d). Interestingly, RING1B knockdown (KD) in GAL4–POGZ cells significantly reduced repression of reporter gene transcription (Fig. 3c-d). In contrast, knockdown of SETDB1, a component of a previously identified POGZ-containing complex [56], did not affect reporter gene expression (Fig. S1). These findings demonstrate that the transcriptional repressor function of POGZ is dependent on RING1B expression, highlighting a key mechanistic link between POGZ and PRC1 in transcriptional repression.

Fig. 3.

RING1B Activity Is Essential for POGZ-Mediated Transcriptional Repression. (a) Schematic of the luciferase reporter system used to examine the transcriptional activity of POGZ. (b) Fold change in luciferase activity in cells expressing GAL4 only, GAL4–PCGF4, and GAL4–POGZ after 24-hour doxycycline induction. Each value is the mean of three biological replicates, with error bars representing the standard deviation. ***p < 0.001 determined by one-way ANOVA followed by Tukey’s post hoc test, compared with the GAL4 control. (c) Fold change in luciferase activity in GAL4–POGZ cells upon KD of RING1B. Cells were transfected with control or RING1B siRNAs, and luciferase activity was measured after 24-hour doxycycline induction. Each value is the mean of three biological replicates, with error bars representing the standard deviation. **p < 0.01 by two-sided t-test, compared with the GAL4 control. (d) Immunoblotting of samples used for luciferase activity reporter assay in (c) to demonstrate NG4-POGZ WT induction and RING1B KD efficiency

POGZ Colocalizes with PRC1.6 at Neurodevelopmental Genes in Mouse Cortical Cells

With the identification of POGZ as a PRC1.6-associated subunit, we next sought to determine whether POGZ colocalizes with PRC1.6 in a developmentally relevant model. To address this, we analyzed publicly available CUT&RUN/Tag datasets for POGZ WT and KO, RING1B, and HP1g in embryonic cortical cells isolated from mice [31, 39]. Our analysis identified 2,927 total POGZ peaks, with 46.3% of peaks (n = 1358) overlapping with both RING1B and HP1g, therefore representing PRC1.6-POGZ target loci (Fig. 4a). We also found that 45.9% of POGZ peaks (n = 1344) overlapped with only HP1g, possibly representing loci targeted by non-PRC1.6-POGZ-containing complexes (Fig. 4a). Only a small percentage of peaks are co-occupied by both POGZ and RING1B (1.7%, n = 51) or POGZ alone (6%, n = 174) (Fig. 4a). The POGZ KO panel shows no detectable enrichment for POGZ which confirms the specificity of this approach and validation of the POGZ KO. Genomic annotation of PRC1.6-POGZ target loci revealed that this complex is predominantly detected near or within genes (85.9%), especially within ± 3 kb of transcriptional start sites (TSS) (70.8%) and occasionally in introns, exons, and untranslated regions (UTRs). Additionally, 14.1% of target loci are found in intergenic areas (Fig. 4b). Gene ontology (GO) analysis of PRC1.6-POGZ target genes revealed enrichment for terms relating to transcriptional regulation (Fig. 4c). Furthermore, we observed strong POGZ, RING1B, and HP1g localization at Klf6 and Myh9, PRC1.6-POGZ target genes that are implicated in various aspects in neurodevelopment (Fig. 4d) [57–60]. The enrichment of PRC1.6-POGZ binding at promoters of neurodevelopment regulators suggests that this complex may modulate the expression of neuronal regulatory genes. Together, these findings highlight the complexity of POGZ genomic occupancy and identify the genes that are targeted by the PRC1.6-POGZ complex.

Fig. 4.

PRC1.6-POGZ genomic localization within embryonic mouse cortical cells. (a) Heatmap showing enrichment of PRC1.6-POGZ, POGZ & RING1B, POGZ & HP1g, and POGZ only bound peaks centered around TSS (± 3 kb) regions across loci. (b) Pie chart showing the percentage of PRC1.6-POGZ target loci on each genomic region. (c) GO analysis of PRC1.6-POGZ target genes in mouse cortical cells. The x-axis (in logarithmic scale) corresponds to the binomial raw P values. (d) Representative genomic tracks showing the normalized tracks of POGZ, RING1B, and HP1g at the indicated loci

Pogz Deletion Leads to a Transcriptomic Alteration in NPCs

To investigate the impact of PRC1.6-POGZ on its target gene expression, we generated a Pogz KO ESC line (Fig. S2a). Deletion of Pogz was confirmed by Sanger sequencing, immunoblotting, and RT-qPCR (Fig. S2a-c). Examination of pluripotency markers Nanog and Oct4 at both protein and transcript levels showed no changes in expression following Pogz KO (Fig. S2b-c). Additionally, alkaline phosphatase (AP) staining of Pogz WT and KO ESCs revealed no noticeable difference in AP activity, further indicating that Pogz deletion does not affect pluripotency (Fig. S2d).

Using an established differentiation protocol [36, 37], Pogz WT and KO ESCs were differentiated into NPCs (Fig. 5a). To evaluate the impact of Pogz KO on neuronal differentiation, we examined the expression of NPC marker genes, including Pax6, Nes, and Sox1, using RT-qPCR. We found that these marker genes failed to be induced in Pogz KO NPCs, pointing to a disruption in the differentiation process (Fig. 5b). To support these findings, we also quantified immunofluorescence (IF) staining of the neuronal precursor marker Nestin in Pogz WT and KO NPCs. Consistent with our RT-qPCR results, we found that Nestin staining was significantly lower in Pogz KO NPCs compared to Pogz WT NPCs (Fig. 5c-d). To further examine the changes in gene expression upon loss of Pogz, we performed RNA-Seq in Pogz WT and KO ESCs and NPCs (Fig. 5e). Our RNA-Seq analysis identified many differentially expressed genes between Pogz WT and KO NPCs, with 1,139 downregulated and 1,973 upregulated genes identified in Pogz KO NPCs (Fig. 5e-f). GO analysis showed that the downregulated genes were enriched in terms related to transcriptional regulation, differentiation, and neurogenesis (Fig. 5f). GO analysis of the up-regulated genes showed enrichment in terms related to general metabolism, macromolecule transport, and angiogenesis (Fig. 5f). These results demonstrate that Pogz is required for ESC neuronal differentiation through establishing an NPC-specific transcriptome.

Fig. 5.

Pogz KO leads to dysregulation of the NPC transcriptome. (a) Brightfield images of Pogz WT and KO cells at various time points of the differentiation process. (b) RT-qPCR analysis of NPC marker gene expression in Pogz WT and KO NPCs. All mean values and standard deviations were calculated from three biological replicates. * p < 0.05 by two-sided t-test. (c) Immunofluorescence staining of Nestin (Green) in Pogz WT and KO NPCs. DAPI is in Blue. (d) Quantification of average Nestin immunofluorescence intensity measured by Image J. Each value is the mean of five independent measurements, with error bars representing standard error. *** indicates p < 0.001 by two-sided t-test. (e) Heatmap of the global transcriptomic profile in Pogz WT and KO ESCs and NPCs. Duplicate samples were subjected to RNA-seq analysis. TPM values for each gene were converted to z-scores and used to generate the heatmap. (f) GO analysis of down-regulated (top) and up-regulated (bottom) genes in Pogz KO NPCs compared to Pogz WT NPCs. The x-axis (in logarithmic scale) corresponds to the binomial raw P values

PRC1.6-POGZ Negatively Regulates the BMP Pathway During Neuronal Differentiation

To further understand the impact of Pogz KO on gene regulation during neuronal differentiation, we cross-referenced PRC1.6-POGZ target genes (Fig. 4a) with our RNA-Seq results. Our analysis revealed 68 up-regulated and 60 down-regulated PRC1.6-POGZ target genes (Fig. 6a, Table S2−3). Considering that POGZ is a known transcriptional repressor [33, 34] and that its repressive activity depends on RING1B (Fig. 3), we focused on the derepressed PRC1.6-POGZ target genes identified in Pogz KO NPCs. Among the de-repressed genes, our RNA-Seq data and RT-qPCR validation revealed that Klf6, an upstream regulator of Transforming Growth Factor beta (TGF-β) signaling [61], was significantly de-repressed in Pogz KO NPCs (Fig. 6a-b). Notably, Klf6 is also a direct PRC1.6-POGZ target gene identified in embryonic cortical cells (Fig. 4d), linking PRC1.6-POGZ occupancy to transcriptional regulation of signaling regulators in our model system.

Fig. 6.

Pogz KO leads to hyperactivation of BMP signaling. (a) Comparison of PRC1.6-POGZ target gene expression levels based on RNA-seq analysis from Pogz WT and KO NPCs. The x-axis is the log2 value of the average TPM value of a gene, and the y-axis is the log2 value of the fold changes of TPM of a gene between two groups. Triangles and circles represent data points with absolute log2 fold change greater than or less than 1, respectively. Pink data points highlight genes with an absolute log2 fold change > 1 and an adjusted p-value < 0.05. Grey indicates genes that do not meet both thresholds. (b) RT-qPCR analysis of PRC1.6-POGZ target gene expression in Pogz WT and KO NPCs. All mean values and standard deviations were calculated from three biological replicates. ** p < 0.01 by two-sided t-test. (c) Immunoblotting of Pogz WT and KO ESC and EB using the indicated antibodies. (d) Immunoblotting of the indicated antigens in Pogz WT and KO EBs treated with and without Noggin. (e) RT-qPCR analysis of NPC marker genes and Klf6 in Pogz WT and KO NPCs treated with or without Noggin. All mean values and standard deviations were calculated from three biological replicates. * p < 0.05, ** p < 0.01, n.s. not significant by two-sided t-test

The TGF-β and BMP signaling pathways must be inhibited during neuroectoderm differentiation, as they promote meso- and endodermal differentiation [62–64]. Activation of the TGF-β/BMP pathway leads to the phosphorylation of regulatory SMAD (R-SMAD) proteins, among which SMAD1/5/9 respond to BMP and SMAD2/3 respond to TGF-β [62]. Consistent with this mechanism, loss of Pogz led to increased phosphorylated SMAD1/5/9 at the EB stage compared with WT cells (Fig. 6c), indicating BMP pathway activation.

De-repression of Klf6 in Pogz KO NPCs led us to hypothesize that increased Klf6 expression alters TGF-β/BMP signaling pathways in these cells. Our immunoblotting experiments revealed that loss of Pogz led to an increase in phosphorylated SMAD1/5/9 (pSMAD1/5/9) at the EB stage compared to WT cells (Fig. 6c). Conversely, the TGF-β pathway does not play a significant role in the context of neuronal differentiation of Pogz KO cells, as there was little to no change in pSMAD2/3 levels in KO cells (Fig. 6c).

To test whether inhibiting the BMP signaling could restore pSMAD1/5/9 levels and rescue the expression of NPC marker genes, we treated our cells with Noggin, a BMP signaling inhibitor, during differentiation. Immunoblotting of day four EBs showed that Noggin treatment restored pSMAD1/5/9 levels in POGZ KO EBs (Fig. 6d) and rescued the expression of these NPC marker genes (Fig. 6e), indicating that increased BMP signaling in POGZ KO NPCs hinders neuronal differentiation. Notably, Klf6 transcript levels were reduced after Noggin treatment, likely through a feedback regulation of the BMP pathway. Ultimately, these findings suggest that Pogz is required for inhibiting BMP signaling through PRC1.6-POGZ-mediated repression of Klf6, which in turn contributes to the regulation of proper neuronal differentiation.

Discussion

Our investigation into the composition of PRC1 in neuronal cells led to the discovery of the PRC1.6-POGZ complex, consisting of a novel component, POGZ, an autism risk factor [24–26]. The existence of the PRC1.6-POGZ complex is further supported by genomic colocalization of POGZ, RING1B, and HP1g in primary murine cortical cells. We further demonstrated that RING1B mediates POGZ transcriptional repression and that Pogz KO NPCs exhibit de-repression of its target genes. Our results suggest a model where PRC1.6-POGZ controls proper neuronal differentiation by inhibiting BMP signaling.

Our discovery of POGZ as a novel interactor of PRC1.6 in neuronal cells is particularly intriguing, given the well-established role of PCGF6 as a master regulator of ESC pluripotency [65–68]. This finding raises the possibility of an alternative PRC1.6 complex composition in differentiated cell types. Previous studies have identified various DNA-binding factors as PRC1 recruiters in a cell-type-dependent manner [69–72]. PRC1.6 is known to be recruited to target loci through its interactions with a variety of chromatin-associated factors, including E2F6/DP1 and MGA/MAX [66, 73–76]. It is worth pointing out that although we identified MGA as a RING1B interacting protein in neuronal cells, E2F6, DP1, and MAX were not present in our MS analysis (Fig. 1b). It is highly likely that POGZ serves as a PRC1.6 recruiter to target genes in neuronal cells by leveraging its zinc finger domains [29] and sequence-specific DNA-binding properties [30] to mediate recruitment. Future studies, including loss-of-function experiments, will be needed to fully dissect the dynamics of PRC1.6-POGZ recruitment in neuronal cells.

TGF-β signaling plays a vital role in cell fate determination [62]. The differentiation of ESCs into the neuroectodermal lineage requires the absence of TGF-β and its family members, BMPs, whereas mesodermal and endodermal differentiation require this signaling [63]. Furthermore, inhibition of TGF-β and BMP signaling greatly enhances the efficiency of neuronal differentiation from ESCs [64]. Interestingly, previous studies on PRC1 and related complexes revealed their critical role in regulating these signaling pathways during neuronal differentiation [21, 53]. In our study, we discovered a novel role for PRC1.6-POGZ in repressing BMP signaling by targeting Klf6 (Figs. 4 and 6), a previously reported activator of the TGF-β pathway [61]. Although the exact mechanism by which PRC1.6-POGZ represses Klf6 remains to be fully understood, we propose that in WT cells undergoing neuronal differentiation, PRC1.6-POGZ targets the Klf6 loci, leading to its repression and inhibition of the BMP pathway. These studies establish a link between various PRC1 complexes and ASD risk factors and their impact on the deregulation of TGF-β/BMP pathways as a potential shared mechanism in neurological disorders.

Recent studies have highlighted the multifaceted transcriptional function of POGZ, emphasizing how its function is shaped by the complexes it engages with and the experimental context in which it is studied. While some reports describe POGZ as a transcriptional activator through interactions with the esBAF complex [30], others associate it with repression via the ChAHP [31, 55] or SETDB1/TRIM28 [56] complexes. In contrast, our data show that PRC1.6-POGZ is distinct from these complexes (Fig. 2c). Furthermore, although differences in ESC lines and culturing conditions across studies may account for these discrepancies, it is very likely that multiple mechanisms operate in concert to regulate neuronal fate transition. As POGZ continues to be studied as a component of various transcriptional regulatory complexes, future studies should aim to systematically dissect the genomic occupancy and functional impact of these complexes. Establishing a well-defined model would allow for a clearer understanding of how POGZ contributes to transcriptional regulation across distinct cell types and developmental stages.

Supplementary Information

Below is the link to the electronic supplementary material.

Footnotes

AP

Alkaline phosphatase

ASD

Autism spectrum disorder

DEG

Differentially expressed genes

ESC

Embryonic stem cell

EB

Embryoid body

GO

Gene Ontology

GREAT

Genomic regions enrichment of annotations tool

IP

Immunoprecipitation

KD

Knockdown

KO

Knockout

NE

Nuclear extract

NPC

Neuronal progenitor cell

O/N

Overnight

PcG

Polycomb group

POGZ

POGO-transposable element with ZNF domain

pSMAD

Phosphorylated SMAD

PRC1

Polycomb Repressive Complex 1

R-SMAD

Regulatory SMAD

RT

Room temperature

siRNAs

Short interfering RNAs

Author Contributions

J.C., T.W., C.E., Z.Geng, Y.T., and J.F. conducted the experiments; J.C. performed the bioinformatic analysis on deep sequencing data; Z. Gao and J.S. designed the experiments; J.C., J.S., and Z.Gao drafted the original manuscript. All authors contributed to the discussion of the manuscript.

Funding

This work was partially funded by NIGMS Early-Stage Investigator Maximizing Investigator’s Research Award (ESI MIRA, R35 GM133496) to Z.G., Diversity Supplement to ESI MIRA, and the GlaxoSmithKline Graduate Fellowship to J.C. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Data Availability

The raw mass spectrometry data reported in this paper have been deposited in the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the MassIVE private database, with the dataset identifier PXD062180. The NCBI GEO accession number for the RNA-seq data reported in this paper is GSE281010.

Code Availability

No custom code was developed for this study. All analyses were performed using publicly available tools and software, as detailed in the Methods section.

Declarations

Ethics Approval

Not applicable.

Consent to Participate

Not applicable.

Consent for Publication

Not applicable.

Competing interests

The authors declare no competing interests.