Chromatin Remodelers Interact with Eya1 and Six2 to Target Enhancers to Control Nephron Progenitor Cell Maintenance

Significance Statement

Understanding how the precise gene expression states that define nephron progenitor cell identity are established and maintained is crucial for developing approaches to repair or regenerate the damaged nephron. Combination of Eya1-Six1/2 has been explored in inducing nephron progenitor–like cell reprogramming. This study uncovers a critical role for chromatin-remodeling SWI/SNF complex through interactions with Eya1-Six2 in nephron fate induction and maintenance and identifies critical factors, including Pbx1, as direct targets of SWI/SNF and Eya1-Six2. The properties of chromatin remodelers as transcriptional coregulators and Pbx1 in hematopoietic stem cell self-renewal suggest that they may be combined with Eya1-Six1/2 in reprogramming nephrons.

Abstract

Background

Eya1 is a critical regulator of nephron progenitor cell specification and interacts with Six2 to promote NPC self-renewal. Haploinsufficiency of these genes causes kidney hypoplasia. However, how the Eya1-centered network operates remains unknown.

Methods

We engineered a 2×HA-3×Flag-Eya1 knock-in mouse line and performed coimmunoprecipitation with anti-HA or -Flag to precipitate the multitagged-Eya1 and its associated proteins. Loss-of-function, transcriptome profiling, and genome-wide binding analyses for Eya1's interacting chromatin-remodeling ATPase Brg1 were carried out. We assayed the activity of the cis-regulatory elements co-occupied by Brg1/Six2 in vivo.

Results

Eya1 and Six2 interact with the Brg1-based SWI/SNF complex during kidney development. Knockout of Brg1 results in failure of metanephric mesenchyme formation and depletion of nephron progenitors, which has been linked to loss of Eya1 expression. Transcriptional profiling shows conspicuous downregulation of important regulators for nephrogenesis in Brg1-deficient cells, including Lin28, Pbx1, and Dchs1-Fat4 signaling, but upregulation of podocyte lineage, oncogenic, and cell death–inducing genes, many of which Brg1 targets. Genome-wide binding analysis identifies Brg1 occupancy to a distal enhancer of Eya1 that drives nephron progenitor–specific expression. We demonstrate that Brg1 enrichment to two distal intronic enhancers of Pbx1 and a proximal promoter region of Mycn requires Six2 activity and that these Brg1/Six2-bound enhancers govern nephron progenitor–specific expression in response to Six2 activity.

Conclusions

Our results reveal an essential role for Brg1, its downstream pathways, and its interaction with Eya1-Six2 in mediating the fine balance among the self-renewal, differentiation, and survival of nephron progenitors.

The nephron in the kidney develops from the metanephric mesenchyme (MM), which forms from the intermediate mesoderm at approximately embryonic day 10.5 (E10.5) in mice.¹ Upon induction by the ureteric bud (UB), the MM aggregates around the UB tip to form the cap-like mesenchyme (CM). The nephron progenitor cells (NPCs) in the CM either self-renew to replenish or differentiate into pretubular aggregates (PTAs) to form renal vesicles (RVs), which eventually differentiate into mature nephron tubules.^2,3 Disruption of any of these steps can lead to renal dysplasia.

MM specification depends on transcription factors (TFs) Eya1, Six1, Six4, and Osr1, as this structure is absent in Eya1^−/−, Six1^−/−;Six4^−/−, and Osr1^−/− mice.^4–6 Although Eya1, Six1, and Six4 are expressed in nephrogenic mesenchyme, Osr1 is expressed in both the mesenchyme and nephric duct. Several MM-specific TFs are required for NPC maintenance and differentiation. Among them, Six2 is essential for promoting self-renewal as Six2^−/− MM undergoes premature epithelialization.⁷ Myc is necessary for renewing the NPC pool.⁸ Although Eya1 acts upstream of Six2, Eya1 protein interacts with Six2 and Myc.⁹ Published studies have demonstrated that the Eya proteins as transcriptional coactivators interact with the Six/So proteins to regulate gene expression in Drosophila eye formation¹⁰ or cochlear explant systems.^11–13 However, it is unclear how gene expression states regulated by Eya1-Six/So are coupled with transcriptionally linked chromatin modifications.

Recently, genome-wide chromatin immunoprecipitation followed by sequencing (ChIP-seq) in E16.5 mouse kidneys has identified Six2-binding sites.^14,15 As Eya1 does not bind to DNA directly, in an effort to identify additional Eya1-interacting proteins, we previously performed a yeast two-hybrid screen and identified Myc, Six, and Sox families.^9,16 In addition, we isolated BAF155, a subunit of the ATP-dependent chromatin-remodeling complex BAF (SWI/SNF). Mammalian BAF complexes contain one of two ATPases, Brm or Brg1 (Brm-related gene 1), and use energy derived from ATP hydrolysis to alter chromatin structures and regulate nuclear processes, such as transcription, proliferation, and DNA repair.^17,18 Eya1 interacts with exogenous Brg1-BAF170/155,¹⁶ but whether these endogenous proteins form functional complexes in promoting NPC self-renewal remains unclear.

A recent study has shown that Brg1 knockout using Six2-Cre results in smaller kidneys.¹⁹ An in vitro pull-down assay revealed the physical association of Sall1 with Brg1-BAFs, and Brg1 ChIP-seq was performed in E16.5 kidneys.¹⁹ However, there is no evidence to support that these endogenous proteins interact in the kidney or that Brg1 binding sites are capable of mediating NPC-specific expression. At present, the regulation and functional specificity of Brg1-BAFs in driving MM formation and NPC maintenance are largely unexplored.

To gain more insight into the Eya1-centered protein network, we performed mass spectrometry using purified 2xHA-3xFlag-Eya1 and identified more subunits of the SWI/SNF. Coimmunoprecipitation (coIP) of nuclear extracts from kidneys of the 2× HA-3× Flag-Eya1 (Eya1^HA-Flag) knock-in mice demonstrated a physical interaction between Eya1-Six2 and Brg1-BAF155-BAF60a/b/c but not with BAF170. Temporal deletion of Brg1 using Eya1^CreER/+ results in failure of MM formation or premature epithelialization and depletion of NPCs. Transcriptional profiling of FACS-purified cells and ChIP-seq revealed Brg1's downstream pathways and its cooperation with the NPC-specific master regulators Eya1/Six2 in maintaining the NPC identity.

Methods

Animals and Tamoxifen Treatment

Brg1^flox,²⁰ Eya1^CreER,⁹ Eya1^flox,⁹ and Six2^+/−7 mice were maintained on a 129/Sv and C57BL/6J mixed background. All animal experiments were approved by the Animal Care and Use Committee of the Icahn School of Medicine at Mount Sinai (#06–822). Mice were bred using timed mating, and noon on the day of vaginal plug detection was considered E0.5.

For induction of the CreER protein, tamoxifen (T5648; Sigma) was dissolved in corn oil (C8267; Sigma) and administrated (1–1.5 mg/10 g body weight) by oral gavage. Observed variations among Brg1 mutants are likely due to preexisting developmental variation between embryos when tamoxifen was given.

Generation of the Eya1^HA-Flag Knock-In Allele

The Eya1^HA-Flag knock-in allele was created by replacing the two endogenous start codons with a promoterless Escherichia coli ATG-2×HA-3×Flag-Eya1-polyA and the PGK-neo gene, similar to the generation of Eya1^lacZ or Eya1^Six1 mice.^21,22 Mice carrying the Eya1^HA-Flag allele were obtained using gene-targeting technology. Southern blot using 420 bp of 5′-flanking probe detected the wild-type (WT) allele of 20.2 kb and the knock-in allele of 13.7 kb. A set of primers (forward primer 5′-CAAGCAAAACCAAATTAAGGG-3′ and reverse primer 5′-GGAATGTCTGATGTATCTGAG-3′) was used to detect the knock-in allele of 320 bp, and a set of primers (forward primer 5′-GTTAAAAGTGAGCATTGTAGG-3′ and reverse primer 5′-CGTCACCTGTGTCATTTTAATT-3′) was used to detect the Eya1 WT allele of 450 bp.

CoIP Analysis

Nuclear extracts were prepared from E13.5 Eya1^HA-Flag kidneys, and coIP was performed as described previously.²³ Briefly, kidneys were homogenized and lysed. After removal of the cytoplasmic fraction, the crude nuclei pellet was lysed in nuclear extraction buffer (20 mM HEPES, pH 7.9, 1.5 mM MgCl₂, 420 mM NaCl, 0.2 mM EDTA, 25% glycerol, 1 mM dithiothreitol, and protease and phosphatase inhibitors cocktail). The extracted nuclear proteins were diluted with immunoprecipitation (IP) buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% NP-40, and 10% glycerol) and precleared with protein A/G followed by IP. The immunocomplexes were separated in SDS-PAGE and detected with primary antibodies and HRP-conjugated secondary using the enhanced chemiluminescence method (WBKLS0500; Millipore).

In Vitro Pull-Down Assay

The HEK293 cells were transfected with HA-Flag-Eya1/pcDNA3, Brg1-Flag (Addgene plasmid 19148), or Six2/pcDNA3 constructs. Transfected cells were harvested and lysed in RIPA buffer (20 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1% Triton, 0.5% sodium deoxycholate, and 0.1% SDS) with protease inhibitor (Sigma; P8340). Individual protein was purified in RIPA buffer and washed with RIPA buffer three times after IP. This stringent buffer with a high detergent concentration in IP can prevent those interacting proteins from being pulled down together to ensure the purity of purified proteins. The Brg1-Flag or HA-Flag-Eya1 was pulled down using Flag beads (Sigma; A2220) and eluted with Flag peptide (Sigma; F4799).

For the Six2 protein, anti-Six2 (Sigma; HPA001893) with protein A/G beads (Santa Cruz; sc-2003) was used for purification. As HEK293 cells endogenously express Brg1, we used anti-Brg1 with protein A/G beads for purifying Brg1 beads. After wash, Six2 beads, Brg1 beads, or beads alone as a negative control were mixed with purified individual Brg1-Flag, HA-Flag-Eya1, or both together and incubated in IP buffer (20 mM Tris-Cl, pH 7.5, 50 mM KCl, 1.25 mM MgCl₂, 0.5 mM EDTA, and 0.1% NP40) overnight. Beads were washed four times with the IP buffer, and the immunocomplexes were analyzed by western blot.

Primary Antibodies

Anti-Brg1 (ab110641; Abcam), anti-BAF170 (sc-17838; Santa Cruz), anti-BAF155 (sc-48350; Santa Cruz), anti-BAF60a (sc-135843; Santa Cruz), anti-BAF60b (sc-101162; Santa Cruz), anti-BAF60c (ab171075; Abcam), anti-Flag (F7425 and A2220; Sigma), anti-HA (ab9110; Abcam and H9658; Sigma), anti-Six2 (66347–1-Ig and 11562–1-AP; Proteintech), anti-Mycn (198912; Abcam), anti-Pbx1 (HPA003505; Sigma), anti-Wt1 (sc-192; Santa Cruz Biotechnology), and anti-H3K27ac (ab4792; Abcam).

Histology, In Situ Hybridization, and Immunostaining

Histologic examinations were performed as described previously.⁹ Whole-mount or section in situ hybridization (ISH) and immunostaining were performed according to standard procedures. Cy3-, Cy2-, Cy5-, and FITC-conjugated secondary antibodies were used. Hoechst 3342 was used for nuclear staining. Probes for ISH were reported previously. At least six embryos for each genotype at each stage were analyzed for each staining.

Proliferation and Terminal Deoxynucleotidyl Transferase–Mediated Digoxigenin-Deoxyuridine Nick-End Labeling Assays and Quantification

The 5-ethynyl-2’-deoxyuridine (EdU) assay was performed using a kit (C10640; Life Technologies) following the manufacturer’s instructions, and embryos were collected 3 hours after injection. The terminal deoxynucleotidyl transferase–mediated digoxigenin-deoxyuridine nick-end labeling (TUNEL) assay was performed using the ApopTag kit for apoptosis fluorescein detection (S7110; Millipore-Sigma). TUNEL⁺ cells from nine sections per E12.5 kidney and three kidneys for each sample were counted. Values represent the average number of TUNEL⁺ cells (± SDs) per section (6 μm).

Kidney Single-Cell Isolation and FACS

Kidneys were isolated from approximately E12.5 to E12.75 Eya1^CreER/+;R26-tdTomato or Eya1^CreER/+;Brg1^fl/fl;R26-tdTomato embryos, minced, and digested with dispase (1 mg/ml) and collagenase IV (0.7 mg/ml) in PBS for 20 min at 37°C. After passing through a 40-μm strainer (352235; BD Biosciences) to eliminate cell clumps, the cell suspension was centrifuged, washed twice in PBS to remove fragments, then resuspended in a freezing medium, and stored in a liquid N2 tank. Twelve kidneys from six embryos were collected from four different pregnant females. After genotyping, we combined cell suspensions for FACS to isolate tdTomato⁺ cells.

RNA-seq Analysis

For each sample, we used approximately 5000 cells for total RNA preparation in replicates using the NucleoSpin RNA XS kit (740902; TakaraBio). mRNA was purified from total RNA using magnetic beads (S1550S; New England Biolabs). Random priming strand-specific RNA-seq libraries were generated using the SMARTer stranded RNA-seq kit (634839; TakaraBio), and single-end 75-bp sequencing was performed using Illumina NextSeq500. Raw sequencing reads were mapped to mm10 using HISAT2.²⁴ Counts of reads mapping to genes were obtained using HTSeq-count against Ensembl v90 annotation.²⁵ Differential expression was done using the DESeq2 package. A volcano plot of global changes between control and Brg1^cKO/cKO cells was generated using the volcano plot tool in Galaxy. Gene ontology (GO) and Panther/reactome pathway enrichment analyses were performed using geneontology.org. Heat map, K means, and Kyoto Encyclopedia of Genes and Genomes analyses were performed using http://bioinformatics.sdstate.edu/idep/.²⁶

RT-PCR and Quantitative Real-Time PCR

Eya1^CreER/+ control and Brg1^cKO/cKO kidneys collected from E12.5 to E12.75 embryos were used for total RNA extraction using Trizol Reagents (15596026; Invitrogen). In total, 100–500 ng of total RNA was treated with RNase-Free DNase I (79254; QIAGEN) and then used for reverse transcription with a SuperScript IV Reverse Transcription (18090010; Thermo Fisher Scientific) for first-strand cDNA synthesis. Quantitative real-time PCR (qPCR) was performed using SYBR Green Master Mix (4309155; Applied Biosystems) and StepOnePlus Real-Time PCR Systems. Expression levels of each transcript were normalized using β-actin as an internal control. Each set of experiments was repeated three times, and the DDCT relative quantification method was used to evaluate quantitative variation. All PCR primers are listed in Supplemental Table 2.

ChIP-seq and Chromatin ChIP-qPCR

For Brg1 or Six2 ChIP-seq, 40 E13.5 kidneys were used, whereas 16 E13.5 kidneys were used for H3K27ac ChIP-seq, respectively. The kidneys were crosslinked, homogenized, and lysed, and chromatin was sonicated as described.²³ Sonicated chromatin was cleared by pelleting insoluble material at 13,000 rpm at 4°C, followed by preclear with protein A/G beads and incubation with approximately 1–2 μg anti-Brg1, -Six2, or -H3K27ac overnight or approximately 1–2 μg rabbit IgG as a negative control. The pull-down and input control sequencing libraries were generated using the ThruPLEX DNA-seq Kit (R400429; Rubicon Genomics) and sequenced on Illumina NextSeq500.

For chromatin immunoprecipitation (ChIP)-qPCR, the ChIPed DNAs were subjected to qPCR amplification as described previously.²³ The enrichment fold of IP over mock IgG IP was calculated using the comparative threshold cycle method. Data were normalized with inputs, and the enrichment of mock IgG IP was considered to be one-fold. This experiment was repeated three times, and each qPCR was performed in triplicate. The primers of qPCR are listed in Supplemental Table 2.

Peak Calling and Annotation

Quality controls using FastQC (v0.11.2) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc) were generated, and raw sequencing reads were then aligned to the mouse mm10 genome using default settings of Bowtie (v2.2.0).²⁷ Peak calling was performed using MACS (v2.1.1)²⁸ with various P value cutoffs. The peak bed files were generated from peak calling against genomic input control or IgG control with the default setting (10⁻⁵ cutoff). The common peaks from these two bed files were used for subsequent analyses. The overlapping peaks of bed files were identified using the bedtools from Galaxy(https://usegalaxy.org/).²⁹ Motif enrichment analysis was performed using the Homer package (v4.8.3).³⁰ The peak annotation and GO analysis was performed using the GREAT program³¹ and the Panther classification system.³² The bamCoverage, computeMatrix, and plotHeatmap tools from Galaxy platform²⁹ were used for a general comparison of overall peaks/signal via the normalized coverage comparisons.

Transient (G0) Transgenic Analysis of Enhancer Activity

The Hsp68 minimal promoter driving LacZ or eGFP expression reporter flanked by the H19 insulators²³ was used for inserting individual enhancer elements upstream of the minimal promoter. Pronuclear injection was performed at our Mouse Genetics and Gene Targeting facility. Transgene expression was analyzed in G0 embryos at indicated stages.

Cell Culture, Transfection, ChIP, and Reporter Assay

HEK293 cells were cultured in DMEM+10% FBS, and transfection was performed as described previously.^9,33 For ChIP-qPCR, cells were fixed and lysed for chromatin preparation 2 days post-transfection. Anti-Six2 and IgG were used for IP, and the ChIPed DNAs were subjected to qPCR as described above.

For reporter assay, each GFP reporter was cotransfected with empty vector, Six2, Eya1, Eya1/Six2, Eya1/Brg1/BAF155, or Eya1/Brg1/BAF155/Six2 expression plasmids. The RFP construct was used as an internal control. The fluorescence intensities were detected by the SYNERGY ¹H microplate reader (BioTek) 36 hours post-transfection. The GFP fluorescence was detected with excitation/emission at 485/512, and the RFP was detected with excitation/emission at 555/585. The assays were performed three times in triplicates, and GFP intensities were normalized by RFP.

Statistical Analyses and Data Plot

To determine the P values for all of the experiments performed, a two-tailed t test was performed. A P value of 0.05 was assumed as statistically significant. P<0.05, P<0.01, and P<0.001 are represented AQ12 with single, double, and triple asterisks, respectively. All statistical analyses were performed with Microsoft Excel.

Column scatterplots were generated using Prism.

Results

Eya1-Six2 Interact with the Endogenous Brg1-Based SWI/SNF Complex in the Developing Kidney

To gain more insight into the transcriptional network in which Eya1 operates, we characterized the Eya1-centered protein interaction network by performing mass spectrometry using purified 2×HA-3×Flag-Eya1 from HEK293 cells (J. Li, C. Cheng, T. Zhang, J. Xu, and P.-X. Xu, unpublished manuscript). We identified more proteins of the SWI/SNF, including BAF60B, SNF2H, SNF2L, and BAF155 (Supplemental Figure 1, A and B).

To understand the importance of Eya1 interaction with its partners during kidney development, it is essential to validate these interactions between endogenous proteins.

We, therefore, engineered the 2× HA-3× Flag-Eya1 knock-in mouse line (Eya1^HA-Flag) (Figure 1A, Supplemental Figure 1C), which allows anti-HA or -Flag for IP of the multitagged Eya1. Eya1^HA-Flag homozygotes are viable with normal kidneys, and immunostaining with anti-Flag or -HA confirmed its restricted distribution in the CM and RVs (Figure 1B), recapturing the endogenous Eya1 expression.⁹ IP of HA-Flag-Eya1 with anti-Flag using nuclear extracts of Eya1^HA-Flag kidneys confirmed the complex formation of Eya1 with Brg1, BAF155, and all three variants of BAF60 but not BAF170 (Figure 1C). As expected, anti-Brg1 precipitated HA-Flag-Eya1 and BAF170/155/60, whereas Six2 was coprecipitated by anti-Flag or -Brg1. Immunostaining for Brg1—the ATPase subunit in the SWI/SNF chromatin-remodeling machinery—confirmed its broad expression in the kidney, including the NPC lineages (Figure 1D). The in vitro pull-down assay using purified proteins confirmed that Eya1 or Six2 directly interacts with Brg1 (Supplemental Figure 1D). Thus, on the basis of the physical interaction between Eya1-Six2 and BAFs and their overlapping expression in the NPCs, we hypothesized that the Eya1-Six2 function might depend on the activity of these interacting chromatin remodelers.

Figure 1.

Eya1 physically interacts with Brg1-BAFs in developing kidneys. (A) Generation of Eya1^HA-Flag knock-in mice. Southern blot using a 5′-flanking probe detected the WT allele of 20.2 kb and the targeted knock-in allele of 13.7 kb. (B) Coimmunostaining of E13.5 kidney from Eya1^HA-Flag with anti-Flag/Six2 or -HA/Six2. Note that Eya1 expression in RVs (arrows) is higher than Six2, consistent with our previous observation by X-gal staining of Eya1^LacZ reporter.⁹ (C) CoIP analysis using nuclear extracts from E13.5 Eya1^HA-Flag kidneys. The input was 5% of the amount used for IP. Anti-Flag was used for IP and anti-HA was used for western blot detection of the multitagged Eya1. Other antibodies used for IP or western blot are indicated. (D) Immunostaining with anti-Brg1 or anti-Brg1/HA on E14.5 WT or Eya1^HA-Flag kidney sections. ES, embryonic stem cells; CB, comma-shaped body; SB, S-shaped body. Scale bars: 30 μm.

Brg1 Plays a Role in the Formation of MM

To investigate whether BAF complex may regulate MM formation, we used Eya1^CreER/+ and Brg1^fl mice^9,20 to knockout the ATPase subunit Brg1 in the nephrogenic progenitors from E9.0 to E9.5. Brg1 is ubiquitously distributed in the MM, nephric duct, and UB, and its expression was detectable in the residual MM cells of Eya1^CreER/+; Eya1^fl/fl (referred as Eya1^cKO/cKO hereafter) at E10.5 (Figure 2A). The presence of some MM cells in Eya1^cKO/cKO is likely due to the incomplete deletion of Eya1, as CreER may not be efficiently activated in all cells simultaneously and shown previously that peak of CreER activation occurs over a subsequent period of 12–24 hours.^34,35 Consistent with this view, the MM is not formed in Eya1−/− embryos.⁴ All Eya1^CreER/+; Brg1^fl/fl (referred as Brg1^cKO/cKO hereafter) at E16.5 lacked kidneys bilaterally (Figure 2B), whereas six of ten Brg1^cKO/+ had smaller kidneys approximately 25%±3% (P=0.01) shorter in length than their WT littermates (Figure 2B). Similar to Eya1^cKO/cKO (Figure 2A), Brg1^cKO/cKO at E10.5 displayed some residual MM cells approximately 36 hours after initiating tamoxifen treatment (Figure 2C), some of which expressed Wt1 (Figure 2C, Supplemental Figure 2B), but there were no identifiable MM cells at E11.5 (Supplemental Figure 2A). ISH found that the expression of Eya1 in the Brg1^cKO/cKO MM region was significantly reduced compared with Eya1^CreER/+ littermates (Figure 2D), suggesting that the expression of Eya1 in MM depends on Brg1. ISH for c-Ret confirmed the lack of UB outgrowth (Figure 2E). The cell proliferation assay using the mitotic tracer EdU in E10.5 embryos found that the number of EdU⁺ cells in the MM region was markedly reduced in Brg1^cKO/cKO compared with Eya1^CreER/+ (Supplemental Figure 2B). In contrast, abnormal apoptosis was increased in the entire nephrogenic cord at E10.0 detected by the TUNEL assay (Supplemental Figure 2C). These defects were also observed in Eya1 null.^4,5 Thus, loss of Brg1 leads to kidney agenesis due to the lack of MM, which is associated with the loss of Eya1 expression.

Figure 2.

Ablation of Brg1 using Eya1^CreER/+ from E9.0 to E9.5 disrupts MM formation and results in kidney agenesis. (A) Immunostaining showing Brg1 expression in the MM, UB, and nephric duct (ND) in E10.5 Eya1^CreER/+ (n=6) and Eya1^cKO/cKO (Eya1^CreER/+;Eya1^fl/fl; n=6) littermates (tamoxifen administration from approximately E9.0–E9.5). (B) Kidney agenesis in Brg1^cKO/cKO embryo at E16.5 (n=8 for each genotype). a, adrenal gland; k, kidney; o, ovary. (C) Hematoxylin&Eosin-stained (upper panels) and Wt1-immunostained (lower panels) sections showing condensed MM and outgrowth of UB in Eya1^CreER/+ but only residual MM cells (arrows), some of which are positive for Wt1, in Brg1^cKO/cKO (n=6 for each experiment). (D) Medial view of whole-mount embryos hybridized for Eya1 riboprobe. Dorsal is up, and the arrow points to the residual signal of Eya1 in the presumptive MM region (n=6). (E) Medial view of whole-mount embryos hybridized with c-Ret riboprobe (n=6). Dorsal is down, and the arrow points to no UB outgrowth in Brg1^cKO/cKO. Scale bars: 30 μm.

Brg1 Is Required for NPC Maintenance

Ablation of Brg1 after UB outgrowth at approximately E10.5–E10.75 resulted in smaller kidneys approximately 51%±2% (P=0.01) the size of those from controls at E14.5 but only residual kidney structures by E17.5 (Figure 3A). Although it is unclear why the kidney regresses by E17.5, anti-Wt1 staining revealed that Brg1^cKO/cKO at E14.5 lacked not only CM in the outermost region of the kidney but also, maturing glomeruli (Figure 3B), thus suggesting that mature nephron structures fail to form in Brg1^cKO/cKO. Histologic analysis confirmed the lack of CM in E13.5 Brg1^cKO/cKO (Figure 3C). Next, we examined whether NPCs undergo premature differentiation to form ectopic RVs. At E12.5, Brg1^cKO/cKO UB development was arrested at the initial branching stage with surrounding vesicle-like structures (Figure 3D), and the NPCs appeared as clustered PTAs from E11.5 (Figure 3E). Wnt4 marks PTAs and RVs on the medullary side of the branching UB (Figure 3, F and G). In Brg1^cKO/cKO, ectopic Wnt4 expression was observed in the entire mesenchyme surrounding the UB branching, indicating premature epithelization of the NPCs. Similar results were obtained for Pax8 (Figure 3, H and I). Thus, depletion of Brg1 induces NPCs to undergo early differentiation.

Figure 3.

Temporal deletion of Brg1 in the induced MM leads to premature epithelialization and depletion of the nephron progenitors. The schedule for tamoxifen treatment is outlined. (A) Kidneys (K) at E14.5 and E17.5 of Eya1^CreER/+ and Brg1^cKO/cKO littermates (n=8). The arrow points to residual kidney tissue. a, adrenal gland. (B) Immunostaining for Wt1 on E14.5 kidney sections (n=8). (C–E) Hematoxylin&Eosin-stained kidney sections of Eya1^CreER/+ and Brg1^cKO/cKO kidneys at (C) E13.5, (D) E12.5, and (E) E11.5 (n=8). The arrow in (C) indicates depletion of the condensed CM, and arrows in (D) and (E) indicate ectopic vesicle–like structures on the outmost region of the kidney. UR, ureter. (F–I) ISH showing (F and G) Wnt4 and (H and I) Pax8 expression in (F and H) PTAs at E11.5 and (G and I) RVs at E12.5 in Eya1^CreER/+ and Brg1^cKO/cKO kidneys. (J and K) ISH showing expression of (J) Eya1 and (K) Six2 (n=6 for each stage). Scale bars: 60 μm.

ISH showed loss of both Eya1 and Six2 expression in Brg1^cKO/cKO (Figure 3, J and K). Additionally, Mycn and Myc expression was not maintained (Supplemental Figure 3A). EdU labeling and TUNEL assays found a marked decrease in EdU⁺ cells but significantly increased apoptosis in E12.5 kidneys (Supplemental Figure 3, B–D). To further confirm the deletion of Brg1 in Brg1^cKO/cKO, we performed coimmunostaining for Brg1/Six2. We observed Brg1 signal in some residual MM cells 24–48 hours after initiating tamoxifen administration, some of which showed weak Six2 signal (Supplemental Figure 3, E and F). Together, these results indicate that Brg1 is essential for maintaining the Six2⁺ NPCs. The premature differentiation and depletion of NPCs observed in Brg1^cKO/cKO may be attributed to the loss of Six2, Eya1, and Myc expression.

Transcriptome Profiling Shows Downregulation of Lin28, Nfatc1, Pbx1, and Dchs1-Fat4 Signaling but Upregulation of Podocyte Lineage and Cell Death–Inducing Factors in Brg1^cKO/cKO

To gain molecular insights into how the loss of Brg1 compromises the NPC development, we bred Eya1^CreER/+ mice with the Cre-dependent R26-tdTomato reporter³⁶ and performed RNA-seq of FACS-purified tdTomato⁺ cells of control (Eya1^CreER/+;R26-tdTomato) and Brg1^cKO/cKO (Eya1^CreER/+;Brg1^fl/fl;R26-tdTomato) kidneys at approximately E12.5–E12.75, respectively (Supplemental Figure 4A). The transcriptome analysis identified 3383 differentially expressed genes (DEGs), with 1952 downregulated and 1431 upregulated in Brg1^cKO/cKO (Figure 4A, Supplemental Figure 4, B and C, Supplemental File 1). Although heatmap analysis showed reduced expression of Six2, Meox1, Myc, and Eya1 (Figure 4B), changes in their expression were not found to be statistically significant due to variation between replicates and were excluded in Supplemental File 1. This may reflect the sensitivity of deep sequencing as Brg1 expression may not be depleted in all cells simultaneously by tamoxifen administration. Statistical power analysis using powsimR³⁷ indicated that the true positive rate for DEGs (P adjusted =0.05) is close to 40% for n=2 per group, which increases to a value of approximately 40% or 62% for three or five replicates per group, respectively, whereas the false discovery rate is close to 50% for n=2 per group and decreases to approximately 40% or approximately 25% for three or five replicates per group, respectively (Supplemental Table 1).

Figure 4.

Transcriptome profiling analyses reveal the requirement of Brg1 for NPC maintenance. Biologic replicates (samples “1” and “2”) with each RNA-seq library prepared from different 5000-cell populations FACS sorted from control or Brg1^cKO/cKO (cKO) at the same time were sequenced. (A) Volcano plot showing transcripts differentially responding to depletion of Brg1. Genes with P adjusted >0.05 and log₂ fold change <0.58 are displayed in gray. (B) Heat map showing 33 selected DEGs in each sample. Genes in black were found to be statistically insignificant (P adjusted >0.05, in gray in [A]) due to variation between replicates and are excluded from the 3383 DEG list (Supplemental File 1). Blue and red indicate down- and upregulated genes, respectively, in Brg1^cKO/cKO. Genes underlined represent oncogenic and apoptotic factors. (C) Quantitative RT-PCR of Eya1^CreER/+ and Brg1^cKO/cKO kidneys. qPCR was performed in triplicate and repeated three times. *P<0.05; **P<0.01; ***P<0.001.

Notably, the ligand for the c-Met receptor tyrosine kinase Hgf, the noncanonical calcium/NFAT Wnt signaling molecule Nfatc1,³⁸ and the RNA-binding protein Lin28a displayed significant downregulation. Multiple NPC-stromal regulators, including Dchs1-Fat4 (CM-stromal signaling essential for kidney development),³⁹ Pbx1/Meis1, Cfh, Penk, and Dcn, were also conspicuously downregulated (Figure 4B). Additionally, cell cycle regulators, such as the Myc targets Myct1/Ccnd2 and the Cdkn2a-interacting protein Cdkn2aip that regulates DNA damage response,^40–42 were downregulated. In contrast, consistent with the premature differentiation of Brg1^cKO/cKO NPCs, the negative regulator of cell proliferation Cdkn1c and genes associated with differentiating nephron precursors, such as Wnt4/Jag1/Pou3f3, were upregulated (Figure 4B). Especially, genes for podocyte lineages, including key TFs Foxc1/Foxc2/Efnb2 essential for podocyte differentiation^43,44 and Mafb/Magi2/Nphs1–2/Plat/Synpo/Thsd7a/Zbtb7c/Clic5/Podxl/Ptpro/Tgfbr3, were noticeably upregulated (Figure 4B). Interestingly, the expression of Rasd2, whose downregulation was implicated in podocyte differentiation,⁴⁵ was dramatically decreased (Figure 4A), but the apoptotic chromatin condensation inducer Acin1, the programmed cell death ligand-1 (PD-L1/Cd274), and the oncogenic TF Runx3³⁹ were significantly increased (Figure 4B). Although excluded from the DEG list due to the variation between replicates, upregulation of the apoptotic factor Map3k5 (known as apoptotic signal-regulating kinase 1) in Brg1^cKO/cKO was apparent, as shown in the heat map (Figure 4B). We also noted marked upregulation of olfactory receptors and TFs involved in brain neurogenesis, such as En1/Otx2/Lmx1a/Otp (>100-fold) and Nkx2.1/Nkx2.3/Phox2b (>80-fold). We performed quantitative RT-PCR using RNA isolated from whole kidneys and confirmed upregulation of Map3k5/Enfb2/Mafb/Nphs1 and downregulation of Eya1/Six2/Pbx1/Dchs1/Lin28a/Nftac1 in Brg1^cKO/cKO (Figure 4C).

GO analysis of DEGs listed in Supplemental File 1 displayed enrichment of downregulated genes related to cell migration/adhesion/proliferation/communication, tube morphogenesis, response to stress, and nephron development, whereas neurogenesis/brain development/glomerular development/nephron differentiation was enriched in upregulated genes (Supplemental Figure 4, C–E). Pathway analysis identified multiple signaling pathways, including cadherin/Wnt/integrin signaling in downregulated genes, and Met activates PTKs signaling in upregulated genes. Together, these analyses indicate the necessity of this chromatin remodeling factor in the expression of many genes and suggest that depletion of Brg1 may provide a permissive environment, leading to the imbalance of multiple pathways and thereby, triggering growth arrest and cell death.

Genome-Wide Analysis of Brg1 Occupancy and Its Co-Occupancy with Six2

BAFs are critical transcriptional coregulators by directly interacting with histone modification enzymes or TFs at gene promoters and cis-regulatory regions (CREs).^46–48 Given the physical association of Eya1-Six2 with Brg1-BAFs, we hypothesized that these two factors might regulate the functional specificity of Brg1-BAFs to target CREs with essential regulatory roles for NPC maintenance. To test this, we investigated genome-wide occupancy of Brg1 by ChIP-seq using chromatin from E13.5 kidneys. We chose this relatively earlier stage to avoid maturing nephron structures due to the wide expression of Brg1. As Eya1 does not bind to DNA directly, we in parallel performed ChIP-seq for Six2 and the histone mark H3K27ac, which is associated with active promoters and enhancers.⁴⁹ In peak calling against both genomic input and IgG controls, we generated a bed file for 8437 Brg1 regions (4108 genes by assigning peaks to their single nearest genes within 500 kb of the nearest gene's transcription start sites) from three datasets for further analysis (Supplemental File 2). Notably, a large fraction of Brg1 sites overlapped with H3K27ac deposition (Figure 5A). Although almost all Brg1 regions near transcription start sites had H3K27ac deposition, approximately 67.2% of Brg1 intronic and intergenic regions were accompanied by H3K27ac (Figure 5B), thus representing an active class of CREs. GO analysis revealed strong enrichment of Brg1 peaks associated with genes related to chromatin organization, cell cycle/cell number maintenance, p38MAPK/TGFβ/apoptotic/Hippo/BMP signaling, and UB/metanephric nephron development (Supplemental Figure 5A). Comparison with the RNA-seq dataset identified 541 DEGs targeted by Brg1, including Hgf, Pbx1-Fat4/Dcn, Cited2, Ccnd/Cdkn2aip/Cdkn1c, and Foxc1/Efnb2/Mafb (Supplemental File 3). Pathway analysis of these genes identified enrichment of multiple signaling pathways, including signaling transduction/MET activates PTK2 signaling/collagen formation/PI3K-Akt/MAPK/Ras signaling (Supplemental Figure 5B).

Figure 5.

ChIP-seq reveals Brg1 and Six2 genome-wide occupancy in the E13.5 kidney. (A) Venn diagram indicating overlap of Brg1 peaks and H3K27ac deposition. (B) Genomic distribution of Brg1 peaks and its overlapping peaks with H3K27ac. UTR, untranslated region. (C) Venn diagram indicating the overlap of Brg1- and Six2-binding sites or targeted genes. (D) Genomic browser visualization of Brg1 peaks at the loci of Fgfr1 and Tcf21 and distal regions co-occupied by Brg1/Six2 (boxed areas). The direction of transcription is shown by the arrow beginning at the transcription start site. Six2E16.5 is from the public database (GSE39837)¹⁴ for comparison. (E) Sequence logos of the significantly enriched top motifs from Homer known motif analysis; letter size indicates nucleotide frequency. Numbers of target sites in Brg1 peaks with the significance of motif occurrence (P value) are indicated. (F) Venn diagrams indicate overlap of Brg1 sites in WT and Six2^−/− kidneys and 596 of 919 sites cobound by Brg1/Six2 absent in Six2^−/− kidney. (G) Heat maps showing Brg1 peaks within a −3-/+3-kb window centered on all Brg1 peaks in WT or Six2^−/− kidneys.

For Six2 ChIP-seq, we identified 8375 peaks (4908 genes) from two datasets (Supplemental File 4), and GO analysis displayed a significant enrichment related to MM development (Supplemental Figure 5A). Among them, 919 regions were co-occupied by Six2/Brg1 (Supplemental File 5), and 1936 genes were cotargeted by Brg1/Six2 (Figure 5C, Supplemental File 6). These common targets include Pbx1/Fat4/Dcn/Eya1/Six2/Meox1/Cited2/Sall1/Tcf/Sox/Efnb2/Foxc1/Notch1/Notch2/Gdnf/Ccnd/Mycn/Myc/Cdkl1/Cdkn1/Bmp7/Fgfr1–3/Map3k5 (Figure 5D, Supplemental Figure 5C). Motif analysis revealed enrichment for the Six2 motif at Brg1 sites (Figure 5E), whereas the most enriched motif at Six2 peaks matched to Six2-binding consensus (Supplemental Figure 5D). Other highly enriched motifs at the Brg1 sites include zinc finger and homeobox proteins, E2F, and ETS (Figure 5E). To a lesser degree, motifs for TEAD, Forkhead, and SOX proteins were also found at the Brg1 sites (Supplemental Figure 5E). TFs possessing these motifs are known to act as Brg1’s transcriptional coregulators.^47,50–55 Thus, this genome-wide analysis provides insight into potential DNA-binding TFs that Brg1 interacts with to target CREs.

Next, we sought to investigate whether access of Brg1 to CREs depends on Six2 activity by examining Brg1 binding in Six2^−/− kidneys. Peak calling identified 7690 Brg1 peaks in E13.5 Six2^−/− kidneys (Figure 5F, Supplemental File 7). Normalized coverage analysis showed stronger Brg1 signals in Six2^−/− than in WT (Figure 5G), likely due to experimental variation between different samples. Among the Brg1 peaks, 2792 peaks were common in both WT and Six2^−/− (Figure 5F). Of 919 peaks co-occupied by Brg1/Six2, peak calling failed to detect Brg1 enrichment at 596 sites (445 genes) in Six2^−/− (Figure 5F, Supplemental File 8), suggesting that Brg1 binding to these sites depends on Six2 activity. These targets include Bmp7, Fgfr1/3, Meox1, Mycn, Ccnd1, Pbx1, Cdkl1/Cdkn1c, Smad2/3/6/7, Tcf21/7l1, genes involved in histone deacetylation complex (Cbx5, Hdac2, Hint1, Mbd2, Nacc2, Rere, Sall1, Tbl1xr1, and Zfp217), and genes involved in transcriptional repressor complex (Bahcc1, Ddx20, Hdac2, Sall1, and Tnrc18). Pathway analysis of these genes identified enrichment of the Hippo signaling pathway, microRNAs in cancer, the TGFβ signaling pathway, transcriptional misregulation in cancer, and cell cycle (Supplemental Figure 6A).

Cooperative Interaction between Brg1, Six2, and Eya1 in the Regulation of Pbx1 Mediated through Two Distal-Intronic Enhancers Co-Occupied by Six2/Brg1

To further understand the synergistic mechanism of Brg1 and Six2-Eya1 in the maintenance of NPCs, we focused on the CREs co-occupied by Brg1/Six2 in genes critical for kidney development and whose expression in Brg1^cKO/cKO was downregulated. Our RNA-seq identified significant downregulation of multiple NPC-stromal genes in Brg1^cKO/cKO, including Pbx1 (Figure 4), a common target of Brg1/Six2. Pbx1 is considered a stromal regulator because it is highly expressed in stromal cells but low in NPCs.^56–58 Previous studies have demonstrated that Pbx1 is essential for kidney development as deletion of Pbx1 in mice leads to renal dysplasia,^56,57 and haploinsufficiency of PBX1 in humans causes congenital anomalies of the kidney and urinary tract.^58,59 However, whether the low level of Pbx1 expression plays a cell-autonomous role in the NPCs and how Pbx1 expression is differentially regulated in distinct mesenchymal cell populations during kidney are not understood.

At the Pbx1 locus, we identified two Brg1/Six2 peaks at highly conserved distal-intronic regions with H3K27ac deposition (+49 and +39 kb) (Figure 6A). Notably, Brg1 binding to both elements was disrupted in Six2^−/− kidneys, which was confirmed by ChIP-qPCR (Figure 6B). To begin to understand the potential role of Pbx1 in the maintenance of NPCs as a target of Six2/Brg1, we first confirmed that Pbx1 is expressed at lower levels in the Six2⁺ CM lineage (Figure 6E). Next, to obtain in vivo evidence that the two Brg1/Six2-bound Pbx1 CREs are functional in the NPCs, we inserted 510 bp of Pbx1+49000 and 907 bp of Pbx1+39000 into an LacZ reporter driven by the Hsp68 minimal promoter flanked by the insulator H19 and used an established mouse transient transgenic enhancer reporter assay. Notably, both elements were active only in the NPC lineage but not in the stromal cells (Figure 6, F and G). The activity of Pbx1+49000 was observed in NPCs, PTAs, and RVs, whereas Pbx1+39000 was more active in committing progenitors to form PTAs and RVs than in the NPCs. Remarkably, however, mutation of both predicted Six2 motifs in either element (Figure 6C) abolished Six2 binding as confirmed by ChIP assay (Figure 6D) and disrupted enhancer activity in the kidney (Figure 6, H and I). Thus, these enhancers mediate NPC-specific expression in response to Six2 activity.

Figure 6.

Brg1 occupancy to two distal-intronic enhancer elements of Pbx1 depends on Six2 activity, and both enhancers govern NPC-specific expression in response to binding to Six2. (A) Genome browser visualization of overlapping occupancy of Brg1and Six2 to the Pbx1 gene. Two conserved regions approximately +49 and +39 kb with strong H3K27ac deposition are co-occupied by Brg1/Six2 in the kidney and by Six1 in the E10.5 embryo, and Brg1 enrichment at these regions was reduced in Six2^−/− kidney. (B) ChIP-qPCR in the E13.5 kidney confirming disruption of Brg1 enrichment at both enhancer regions in the Six2^−/− kidney. The enrichment of mock IP was considered one-fold. ***P<0.001. (C) Both enhancer regions contain two SIX motifs, and each mutant reporter transgene was generated by introducing mutations into both predicted SIX motifs. (D) ChIP-qPCR confirming disruption of these SIXmt mutations to Six2 binding assessed by using chromatin prepared from HEK293 cells cotransfected with His-Six2 expression plasmid and each WT or mutant reporter. The enrichment of mock IP was considered one-fold. **P<0.001. (E) Coimmunostaining for Pbx1 (red)/Six2 (green) in the E16.5 kidney showing Pbx1 expression in stromal cells (SCs), CM, and RV. (F–I) G0 Tg analysis of the LacZ transgene driven by 51 -bp of Pbx1+49000, 907 bp of Pbx1+39000, or each SIXmt reporter in the E16.5 kidney. (F) β-Gal⁺ cells were detected in CM and RVs of transgenic kidneys driven by the enhancer region at +49 kb (n=7/7 Tg lines), whereas (G) the enhancer at +39 kb drove transgene expression with more β-Gal⁺ cells around the UB tip and migrating toward ventral side to the UB branching and fewer in the CM peripheral to the UB branching (n=2/2 Tg lines). (H and I) X-gal–stained kidney sections of each transgene driven by each SIXmt enhancer (n=8/8 in [H] or n=10/10 in [I]). Scale bars: 30 μm.

We further examined if Eya1 or Eya1 together with Brg1-BAF155 cooperates with Six2 in regulating this reporter expression in HEK293 cells and found that coexpression of Eya1 and Six2 did not lead to an apparent increase in the expression of either reporter (Supplemental Figure 6B). However, coexpression of Eya1-Brg1-BAF155 and Six2 led to a slight but synergistic increase in reporter expression, which was not observed with each mutant reporter. Together, these results support the notion that Brg1-BAFs and Eya1 are integral components of the NPC-restricted transcriptional control of Pbx1 mediated through two distal-intronic enhancers via interaction with Six2/Brg1 (Supplemental Figure 6C).

Coregulation of Mycn by Six2/Brg1 Mediated through a Proximal-Promoter Region

Eya1 interacts with Myc proteins, which play a role in NPC renewal.^8,60 However, how Myc expression is regulated is not understood. Our data show that the expression of Myc genes is not maintained in Brg1^cKO/cKO (Supplemental Figure 3A) and that Brg1 and Six2 cobound to a proximal-promoter region of Mycn (−440 bp) but that Brg1 binding to this region was reduced in Six2^−/− (Figure 7, A and B). To determine whether this region mediates Mycn expression regulated by Six2/Brg1, we assayed this fragment (631 bp; −333 to −962 bp) in transgenic embryos. This CRE drove GFP reporter expression in Mycn-expressing NPCs and RVs (Figure 7E), whereas a mutant reporter in which all three presumptive Six2-binding sites were mutated (Figure 7C) abolished Six2 binding in cultured cells (Figure 7D) and had no activity in vivo (Figure 7E). Thus, this suggests that this proximal CRE mediates Mycn expression coregulated by Brg1/Six2 in NPC self-renewal maintenance.

Figure 7.

Enhancers/CREs of Mycn and Eya1 loci occupied by Brg1 and/or Six2 drive expression in MM lineage. (A) Genome browser visualization of overlapping occupancy of Brg1 and Six2/Six1 to the Mycn promoter (red blocks; −333 to −962 bp). Peak calling did not detect Brg1 binding to this region in the Six2^−/− kidney. The arrow indicates the direction of transcription beginning at the transcription start site. Black asterisks indicate nonspecific peaks as seen in the IgG control. (B) ChIP-qPCR confirmed disruption of Brg1 binding to the Mycn promoter region in the Six2^−/− kidney. (C) The Mycn proximal-promoter fragment (631 bp; −333 to −962 bp) contains three SIX motifs, and a mutant reporter transgene was generated by introducing mutations into these predicted SIX motifs. (D) ChIP-qPCR confirming disruption of Six2 binding to the mutant reporter assessed using chromatin prepared from HEK293 cells cotransfected with His-Six2 expression plasmid and reporter Mycn -631-EGFP and Mycn-631SIXmt-EGFP. The enrichment of mock IP was considered one-fold. ***P<0.001. (E) Anti-Mycn immunostaining and G0 transgenic analysis of EGFP transgene driven by Mycn-631 and SIXmt in the E16.5 kidney showing EGFP activity in the NPCs and RVs (n=2/2 transgenic lines), but SIXmt abolished EGFP expression (n=12/12 Tg lines). (F) Genomic browser visualization showing comparison of Brg1, Six2 (2012, GSE39837¹⁴ and 2016, GSE73867¹⁵), and Six1 in E10.5 embryo (GSE108130)²³ and H3K27ac and H3K27me3 enrichment (GSE166588) at three distal regions of the Eya1 locus. The left panel shows the Six2-bound peak approximately +520 kb (red asterisks), which is also bound by Six1 in E10.5 embryo. The center panel shows the Brg1-bound −195-kb region with H3K27me3 at E10.5 but with H3K27ac at E13.5. The red asterisks indicate weak Six2 enrichment and reduced Brg1 binding in Six2^−/−. The right panel shows two Six2-bound regions without Brg1 peak or H3K27ac deposition. The one approximately −316 kb (double gray asterisks) is untested, whereas the other approximately −325 kb (double black asterisks) was previously shown to drive transgene expression in the NPCs.¹⁴ G0 transgenic analysis of transgenes driven by (G) a 1043-bp fragment approximately +520 kb bound by Six2 and Six1 (Eya1 + 520kb-GFP) and (H) a 2263-bp fragment approximately −195 kb occupied by Brg1 (Eya1–195kb-LacZ) in mouse embryos. For Eya1 + 520kb-GFP, all Tg lines showed consistent expression in uninduced MM progenitors at E10.5 (n=3) and in CM, PTA, and RV at E14.5 (n=3). For Eya1–195kb-LacZ, five Tg embryos at E14.5 were analyzed, and all showed consistent expression restricting to the NPCs. ND, nephric duct. Scale bars: 30 μm.

Two Distal Elements Occupied by Brg1 or Six2 at the Eya1 Locus Drive Nephron Progenitor–Specific Expression

Eya1 expression is reduced in Brg1^cKO/cKO, and Brg1 targets Eya1 (Supplemental File 2). Although Eya1 expression is Six2 independent,⁹ Six2 is likely involved in upregulating Eya1 in the NPCs because Six2 binds to several sites of Eya1 in E16.5 kidneys.^14,15 However, how Eya1 expression is upregulated and maintained throughout nephrogenesis is largely unknown. At the Eya1 locus, although an Six2-bound element approximately −325 kb, previously shown to be active in the NPCs,¹⁴ had no enrichment for Brg1/H3k27ac (Figure 7F), peak calling identified Brg1 binding to the promoter (Supplemental Figure 7, A and B) and an intergenic region approximately −195 kb with H3K27ac deposition (Figure 7F). Interestingly, a comparison with E10.5 found strong deposition of the repressive histone mark H3K27me3 at −195 kb. In Six2^−/− kidney, Brg1 binding to this region appeared to be altered at a subregion with some enrichment for Six2 (Figure 7F), confirmed by ChIP-qPCR (Supplemental Figure 7C). There are two Six2-bound intergenic regions >+500 kb with weak H3K27ac or Brg1 enrichment, one of which (+520 kb) was also bound by Six1 in E10.5 embryo.²³ Additional sites bound by Six2 in both proximal and distal regions of the Eya1 were also identified (Figure 7F, Supplemental Figure 7A).

To begin to understand how Brg1 and Six2 might cooperate to maintain the high level of Eya1 expression, we examined whether Brg1- or Six2-bound CREs are active in the NPCs. In transgenic embryos, Eya1+ 520 kb (1047 bp) drove expression in both uninduced and induced NPCs (Figure 7G), whereas this element was also active in nascent nephrons where Eya1 expression is downregulated. This regional expansion of transgene expression is likely due to the absence of repressive elements that are typically present in the Eya1 locus. In fact, when individual CREs are assayed in vivo, they are unable to precisely mimic endogenous gene expression due to a lack of cooperative interaction with enhancers or repressors associated with the intact gene locus.

In contrast, Eya1 -195 kb (2263 bp) drove reporter expression in NPCs (Figure 7H) but not in uninduced MM (n=4/4 Tg lines). Because this element had strong H3K27me3 deposition at E10.5, it may only direct Eya1 expression in the induced MM. Motif analysis of this region failed to identify the Six2-binding sites but found other motifs for TFs, such as C/EBP/TBP/GATA1—known to recruit Brg1^61–63 and WT1. Although it is unclear if Eya1 interacts with Six2 to autoregulate itself, our data suggest that the temporal expression of Eya1 is mediated through a combination of multiple CREs occupied by Six1/2, Brg1-BAFs and other factor(s) (Supplemental Figure 7D).

Discussion

Chromatin remodelers promote nucleosome depletion to grant TFs access to their cognate binding sites, and their tissue-specific functions depend on interaction with tissue-specific TFs. However, how chromatin remodelers interact with lineage-specific TFs to establish de novo enhancers to activate lineage-specific gene programs during kidney development remains largely unexplored. Here, we provide evidence that Eya1 and Six2 collaborate with Brg1-based SWI/SNF to regulate a set of regulators to maintain the NPCs.

Eya1 is a critical regulator for MM specification and acts together with the Six/So protein family.^4,5,9,64 Other than the Six/So family, very little is known about the Eya1-centered protein network. Furthermore, how Eya1 is activated in the initial events of MM fate induction is not understood. The knockout of Brg1 from E9.0 provides the first evidence for its involvement in establishing the MM and Eya1 expression. The timing of Eya1 activation may require a unique BAF complex in response to external signals to promote an MM fate within the intermediate mesoderm. Although future work is necessary to investigate how Brg1 depletion alters the accessibility landscape of enhancers with regulatory roles in establishing the MM fate, we speculate that (1) Eya1 expression in uninduced and induced progenitors is mediated through distinct CREs and that (2) Brg1 occupancy to multiple regions of Eya1 is essential to introduce local changes in chromatin structure that then activate expression of this key regulator to promote an MM lineage. This process likely involves TFs expressed in the intermediate mesoderm during the transition to an Eya1⁺ fate. Eya1 and Six/So may crossregulate and cooperate with BAFs to establish a functional MM. This explains the absence of MM in Six1/4 null and Six1/2 binding to Eya1+ 520 kb, which was active in both uninduced and induced progenitors. In the induced MM, Brg1-BAFs likely interact with not only Six2 but also, other NPC-specific factors to maintain the high levels of Eya1 mediated through additional CREs (Supplemental Figure 7D). Consistent with this, Brg1 binds to Eya1− 195 kb, which was associated with H3K27me3 at E10.5, was inactive in uninduced MM in transgenic embryos, and did not contain the Six2-binding sites.

The physical association of Eya1-Six2/Brg1-BAFs and CREs or genes co-occupied by Brg1/Six2 suggests that Eya1-Six2 interactions may cooperate to regulate the functional specificity of BAFs, and depletion of either one may compromise Brg1 recruitment to their target loci. This could explain why despite Brg1 expression, Eya1^cKO/cKO or Six2^−/− shows a similar phenotype. On the basis of our genetic and molecular analyses, we propose that Eya1 as a coactivator may bridge Six2 and Brg1-BAFs to enable DNA bending to form a compact active complex capable of inducing a robust transcriptional activation (Supplemental Figure 6C). As Eya1 interacts with all three variants of the BAF60 subunit, which are known to recruit TFs^65,66 or nuclear receptors^67,68 to the SWI/SNF complex, these BAF60 variants may bridge Eya1-Six2 to the SWI/SNF complexes to their target sites. Thus, the alternative usage of specific SWI/SNF variants and the selective assembly of SWI/SNF with BAF60a-, BAF60b-, or BAF60c-Eya1 may promote distinct expression programs of MM-affiliated genes to renew and maintain the NPCs.

Basta et al.¹⁹ reported that Brg1 knockout using the Six2-Cre leads to smaller kidneys associated with reduced proliferation, but Six2 expression is present. This difference may be caused by the use of different Cre lines. Compared with the BAC transgenic Six2-Cre,⁶⁹ Eya1^CreER/+ as a knock-in allele may enhance the kidney phenotype of Brg1 knockout if Eya1 and Brg1 genetically interact, which may cause further reduction of Six2 expression. Basta et al.¹⁹ also reported ChIP-seq data at E16.5 and RNA-seq using whole kidneys. Once those sequencing data become publicly available, bioinformatic comparison may provide useful information to help explain. Nonetheless, our analyses clearly indicate that Brg1 deletion leads to downregulation of Six2 and premature differentiation and depletion of the NPCs. Wnt4 is an early marker for NPC differentiation, and its expression was increased in Brg1^cKO/cKO. Our ChIP-seq failed to identify Brg1 binding to the Wnt4, but Six2 binds to three distal regions at the Wnt4 locus, one intronic and two upstream approximately −60 and 65 kb (Supplemental Figure 7E). Thus, Brg1 may not interact with Six2 to repress Wnt4, and the premature differentiation observed in Brg1^cKO/cKO could be due to the loss of Six2.

Our RNA-seq data suggest that some genes may not be sensitive to Brg1 deletion, as many Brg1 ChIP-seq targets did not show significant changes in their expression. We found changes in the expression of some Brg1 targets, including Eya1, Six2, Map3k5, Lin28b, and Wnt4, to be quantitatively insignificant due to the variation between replicates (P adjusted >0.05) (Figure 4B). As we confirmed their changes by ISH or quantitative RT-PCR, such variation between replicates has likely resulted from uneven deletion of Brg1 between different FACS-purified cells. Consistent with this, we observed high Brg1 in some cells and uneven Six2 levels 24–48 hours after initiating tamoxifen treatment. We also observed some variation in the number of NPCs among Brg1^cKO/cKO embryos, likely due to preexisting developmental variation between embryos when tamoxifen was given. Nonetheless, as Brg1 binds to Map3k5, it may directly regulate Map3k5 expression to inhibit apoptosis. Because abnormal apoptosis was also observed in Eya1-⁹ or Six2-null⁷ kidneys and Six2 also binds to Map3k5, these proteins may interact to regulate NPC survival.

Multiple cell cycle regulators were downregulated in Brg1^cKO/cKO, and of particular interest is the proto-oncogene Pbx1. Although haploinsufficiency of PBX1 causes congenital anomalies of the kidney and urinary tract^58,59 and Pbx1 knockout results in kidney agenesis or hypoplasia,^56,57 no studies have been carried out to address if it has a role in the NPCs and how its expression in different cell populations is differentially regulated. In hematopoietic stem cells, Pbx1 supports self-renewal by promoting TGFβ signaling as Pbx1 deficiency induces cell cycle activity and exhaustion of hematopoietic stem cells.⁷⁰ We speculate that the relatively low level of Pbx1 in the NPCs may have a role in cell cycle progression. Although NPC-specific deletion of Pbx1 is currently underway in our laboratory, the identification of Pbx1 as a direct target of Six2/Brg1 together with the downregulation of Lin28a and the CM-stromal signaling Dchs1-Fat4 in Brg1^cKO/cKO provide potential mechanisms into how the Brg1-based BAFs regulate the fine balance between cell cycle regulation and maintenance.

Disclosures

P.-X. Xu reports scientific advisor or membership with Frontiers in Bioscience–Landmark. All remaining authors have nothing to disclose.

Funding

This work was supported by National Institutes of Health grant RO1 DK064640 (to P.-X. Xu).

Data Sharing Statement

All data needed to evaluate the conclusion are presented in the paper, Supplemental Figures 1–7, and Supplemental Tables 1 and 2. Brg1, Six2, and H3K27ac ChIP-seq and RNA-seq data reported in this paper have been deposited to the Gene Expression Omnibus under accession number GSE185050 (ChiP-seq) and GSE159483 (RNA-seq). Protocols and materials are available upon request.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2021040525/-/DCSupplemental.

Supplemental File 1. 3383 differentially expressed genes.

Supplemental File 2. Peak positions and associated genes of 8437 Brg1 peaks.

Supplemental File 3. 541 common genelist to Brg1 ChIP-seq and RNA-seq.

Supplemental File 4. Peak positions and associated genes of 8375 Six2 peaks.

Supplemental File 5. Peak positions and associated genes of 919 Brg1-Six2 cobound regions.

Supplemental File 6. 1936 common genelist between brg1 and six2.

Supplemental File 7. Peak positions and associated genes of 7690 Brg1 peaks in Six2mt kidney.

Supplemental File 8. Peak positions and associated genes of 596 Brg1-Six2-WT unique regions.

Supplemental Figure 1. Interaction of HA-Flag-Eya1 with Brg1-BAFs and generation of the Eya1HA-Flag knock-in mouse line.

Supplemental Figure 2. Deletion of Brg1 using Eya1CreER/+ before UB outgrowth leads to a lack of MM formation.

Supplemental Figure 3. Deletion of Brg1 in MM progenitors after UB outgrowth leads to depletion of the nephron progenitors.

Supplemental Figure 4. Analysis of differentially expressed genes identified by RNA-seq.

Supplemental Figure 5. ChIP-seq analysis and comparison with RNA-seq data.

Supplemental Figure 6. GO and pathway enrichment analysis of Brg1/Six2 peaks and enhancer-mediated Pbx1 expression by Brg1-BAFs-Eya1/Six2 in cultured cells.

Supplemental Figure 7. Genomic browser visualization showing the comparison of Brg1, Six2, and H3K27ac enrichment at the Eya1 and Wnt4 loci.

Supplemental Table 1. Power analysis of RNA-seq using powsimR.

Supplemental Table 2. Primers used for RT-qPCR and ChIP-qPCR.