All animal protocols were approved by Animal Care and Use Committee of the Icahn School of Medicine at Mount Sinai (protocol #06-0822).

The Eya1^Six1/+ (Nie et al., 2010) and Six2 ^+/− (Self et al., 2006) were maintained at the Icahn School of Medicine at Mount Sinai Animal Facility. Mice were bred using timed mating, and noon on the day of vaginal plug detection was considered as E0.5.

Histology, immunohistochemistry, and in situ hybridization were carried out according to standard procedures. Briefly, kidneys were fixed in 4% paraformaldehyde (PFA) for overnight at 4°C, dehydrated, and embedded in wax or OCT. Paraffin or frozen sections were generated at 6 μm of thickness. We used six embryos for each genotype at each stage for each probe and the result was consistent in each embryo. Probes for ISH were reported previously (Xu et al., 2014). Cy3-, Cy2-, Cy5-and FITC-conjugated secondary antibodies were used and Hoechst was used for nuclear counter-staining.

Primary antibodies: Anti-Six1 (12,891, Cell Signaling), -Six2 (MBS610128, MyBiosource), anti-Wt1 (sc-192, Santa Cruz Biotechnology), -Eya1 (25-067, Prosci Inc. and MABE1047, Sigma), and -PH3 (ab10543, Abcam).

The TUNEL assay was performed using the Apop Tag kit for in situ apoptosis fluorescein detection (S7110, Millipore-Sigma) following the manufacturer’s instructions.

TUNEL- or PH3-positive MM progenitor cells on the peripheral side of branching UB were counted from 28 to 35 sections of 5 different kidneys and values represent average number of TUNEL⁺ or PH3⁺ cells (±standard deviations) per section (6 μm). Two-tailed Student’s t test was used for statistical analysis.

For Six1 ChIP, we used 40 kidneys from ∼E13.5 Eya1^Six1 embryos. ChIP-seq was performed according to previous protocols with some modifications. Briefly, the kidneys were cross-linked with 1% formaldehyde at room temperature for 30 min and then homogenized and lysed in cold lysis buffer (50 mM HEPES–KOH, pH 7.5, 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP–40, 0.25% Triton X-100, 1 × protease inhibitors), followed by spinning at 2000 g at 4°C, resuspending in cold wash buffer (10 mM Tris–HCl, pH 8.0, 200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 1 × protease inhibitors) and spinning at 2000 g at 4°C in a benchtop centrifuge. Samples were then resuspended in 1 ml cold sonication buffer (10 mM Tris–HCl, pH 8.0, 2 mM EDTA, 0.1% SDS) and sonicated to 200–500 bp fragments using a Covaris S220 Focused-ultrasonicator. Sonicated chromatin was cleared by pelleting insoluble material at 13,000 RPM at 4°C, followed by preclearing with protein A/G beads and incubation with 1–2 µg antibody overnight (anti-Six1, Cell Signaling #12891) or 1–2 µg rabbit IgG as a negative control. Chromatin–antibody complexes were precipitated with protein A/G beads at 4°C for another 5 h. Immunoprecipitated complexes were subjected to series of wash steps with low salt buffer (20 mM Tris–HCl 8.0, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 0.1% SDS), high salt buffer (20 mM Tris–HCl pH 8.0, 500 mM NaCl, 2 mM EDTA, 1% Triton X-100, 0.1% SDS), LiCl wash buffer (10 mM Tris–HCl pH 8.0, 250mMLiCl, 1mMEDTA, 1%NP-40, 1% sodium deoxycholate) and TE plus NaCl, followed by elution and reverse crosslinking overnight at 65°C. The quality control of ChIPed DNA was performed with Qubit 2.0 Fluoremeter using dsDNA HS assay Kit (Q32854, ThermoFisher Scientific) and Agilent 2200 TapeStation System using High Sensitivity D1000 Reagents (5067–5585, Agilent). The pulldown and input control sequencing libraries were generated using the ThruPLEX DNA-seq Kit (R400429, Rubicon Genomics) and sequenced on Illumina NextSeq 500.

Quality controls using FastQC (v0.11.2) (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). Peak-calling was performed using MACS (v2.1.1) 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/).

Motif enrichment analysis was performed using the Homer package (v4.8.3) (Heinz et al., 2010). The peak annotation and gene ontology analysis was performed using GREAT program (great.stanford.edu). The bamCoverage and coverage heatmap were generated by centering and scaling corrected ChIP-seq coverage using mean and standard deviation, and plotting normalized coverage across 3-kb centered on the Six1-enriched region.

ChIP-seq data sets have been submitted to the NCBI Gene Expression Omnibus (GSE189131).

The nephron progenitors on the peripheral side of the branching UB tip self-renew to replenish to generate a sufficient number of nephrons. In contrast, the progenitors on the ventral side of the branching UB tip undergo a mesenchyme-to-epithelial transition to form pretubular aggregates (PTAs), which then epithelialize to form renal vesicle (RVs)—the primordia of nephron tubules. Six2 is highly expressed in the multipotent nephron progenitors, and Six2 knockout leads to rapid RV formation and depletion of the nephron progenitors, resulting in hypoplastic kidneys (Self et al., 2006). To examine the functional equivalence between Six1 and Six2 during nephrogenesis, we tested whether forced expression of Six1 in the nephron progenitors can rescue Six2-deficient kidney phenotype by crossing Six2 ^+/− mice with a Six1 knockin mouse model expressing Six1 under Eya1 transcriptional regulatory control (Eya1^Six1) (Nie et al., 2010).

Eya1 is coexpressed with Six1 in the uninduced MM (Xu et al., 2003; Xu and Xu 2015) and with Six2 in the CM, and it is upstream of Six2 (Xu et al., 2014). We previously reported that renal agenesis associated with Six1-deficiency can be rescued in Eya1 ^Six1/+ ;Six1 ^−/− mice (Nie et al., 2010). However, to our surprise, forced expression of Six1 failed to rescue Six2-deficient hypoplastic kidneys at E14.5–17.5 (n = 8, Figures 1A,B). Quantitative analysis of kidney size showed that the length of Eya1^Six1/+ kidney was ∼104 ± 5% (n = 6, p = 0.0375) at E14.5 and ∼101 ± 3% (n = 8, p = 0.0418) at E17.5 of the wild-type, while the length of Eya1 ^Six1/+ ;Six2 ^−/− kidney was ∼106 ± 5% (n = 8, p = 0.0357) at E14.5 and ∼114 ± 7% (n = 8, p = 0.0208) of Six2 ^−/− littermate. Histological analysis and immunostaining for Wt1, which is expressed in the CM, PTAs, RVs and differentiating podocytes, revealed that both Eya1 ^Six1/+ ;Six2 ^−/− and Six2 ^−/− mice lacked the nephron progenitors in the outmost nephrogenic zone where the UB branching morphogenesis normally takes place (Figures 1C,D). Similar observation was obtained when we used Eya1^CreER to induce Six1 expression in the MM by tamoxifen administration (Eya1 ^CreER ;R26-Six1mCherry) and tested its ability to restore kidney development in Six2 ^−/− (data not shown). Thus, these findings demonstrate that the expression of Six1 in the MM cells under the Eya1 transcriptional regulatory control cannot functionally substitute for Six2 to maintain the nephron progenitor cells during mouse kidney development.

We next examined whether there is a partial rescue at earlier stages, from the first “T” bud stage at E11.5. As reported previously (Self et al., 2006), the progenitors within the entire MM surrounding the branching UB underwent premature differentiation to form PTA- or RV-like structures in Six2 ^−/− embryos at E11.5 (Figure 2D) and RV-like structures surrounded the entire UB at E12.5 (Figures 2H,L). In contrast, Eya1 ^Six1/+ ;Six2 ^−/− MM progenitors surrounding the branching UB appeared less dense than those in Eya1^Six1/+ or wild-type littermate controls at E11.5 (Figures 2A–C), but they did not commit to an overall RV fate, as shown by histological analysis and anti-Wt1 immunostaining (Figures 2C,G,K). This observation was consistent with the expression pattern of Wnt4, which is one of the earliest markers labeling differentiating nephron structure PTAs at E11.5 and RVs at E12.5 on the ventral side of the branching UB (Figures 2M,N,Q,R). Differing from the relatively uniform expression of Wnt4 throughout the entire MM in Six2 ^−/− embryos (Figures 2P,T), Eya1 ^Six1/+ ;Six2 ^−/− embryos did not display a global activation of Wnt4 in the entire MM and higher levels of Wnt4 expression was detected on the ventral side of the branching UB at E11.5 (Figure 2O). However, the pattern of Wnt4 expression appeared abnormal at E12.5 (Figure 2S) compared to those in Eya1^Six1/+ littermate controls (Figures 2Q,R). By E12.5, the UB typically has completed second branching in control embryos (Figures 2E,F,I,J). In Six2 ^−/− embryos, the first branching T-shaped UB appeared as a single tube, and the UB tips were not induced to undergo second branching due to the depletion of the MM progenitors (Figures 2D,H,L,P,T). In contrast, Eya1 ^Six1/+ ;Six2 ^−/− UB tips appeared to be induced and expanded to invade the MM to initiate (Figure 2K) or undergo second branching (Figure 2G), but the second branching was either incomplete or abnormal. Thus, while the UB development in Eya1 ^Six1/+ ;Six2 ^−/− is arrested at the second branching, Six1 expression in Six2 ^−/− MM appears to prevent premature mesenchyme-to-Epithelial Transition

As we noticed that the MM in Eya1 ^Six1/+ ;Six2 ^−/− is not as dense as in Eya1^Six1/+, we sought to examine whether Six1 can rescue cell survival or proliferation defects in Six2 ^−/− embryos. TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) assay revealed a noticeable degree of apoptosis throughout the MM areas, as a marked increase in the number of TUNEL⁺ cells in the MM was detected in both Eya1 ^Six1/+ ;Six2 ^−/− and Six2 ^−/− (Figures 3A,C). It should be noted that increased apoptosis was also observed in the branching UB of Six2 ^−/− embryos at E11.5 (Figure 3A). In contrast, anti-PH3 (phosphohistone H3)-labeled mitotic cells in the MM were decreased in Eya1 ^Six1/+ ;Six2 ^−/− and more decreased in Six2 ^−/− littermate embryos (Figures 3B,D). These data suggest that while Six1 appears to prevent premature epithelialization of the progenitors, it is not able to substitute the essential role of Six2 in nephron progenitor renewal and survival, thus leading to depletion of the nephron progenitors and arrest of nephrogenesis during second UB branching in Eya1 ^Six1/+ ;Six2 ^−/− embryos.

Loss of either Eya1 or Six2 in the nephron progenitors leads to increased cell death and reduced proliferation (Figure 3) (Self et al., 2006; Xu et al., 2014). We therefore performed ISH to confirm Six1 expression in the MM progenitors on the peripheral side of the branching UB in Eya1^Six1/+ at E11.5, compared to no expression in wild-type or Six2 ^−/− littermates (Figure 4A). In Eya1 ^Six1/+ ;Six2 ^−/− , the MM appeared smaller in size as outlined by the Six1 expression (Figure 4A). We previously reported that although Eya1 mRNA is expressed in Six2 ^−/− MM progenitors, Eya1 protein is localized in the cytoplasm of Six2 ^−/− MM progenitors and its nuclear translocation depends on Six2 activity (Xu et al., 2014). In Eya1 ^Six1/+ ;Six2 ^−/− embryos, the Eya1⁺ domain was comparable to that seen in Six2 ^−/− littermates, both of which were smaller in size than in control littermates (Figure 4B). Notably, however, anti-Eya1 immunostaining revealed nuclear localization of Eya1 in Eya1 ^Six1/+ ;Six2 ^−/− MM progenitors, in contrast to its cytoplasmic localization in Six2 ^−/− MM cells (Figure 4C). Thus, Six1 is also able to mediate nuclear translocation of Eya1 in the progenitors in the absence of Six2. This observation is consistent with the previous finding that Six1 or Six2 but not Six3 can mediate nuclear translocation of Eya1 when coexpressed in HEK293 cells (Buller et al., 2001).

We further examined whether Eya1 expression can be maintained in the MM progenitors of Eya1 ^Six1/+ ;Six2 ^−/− embryos at later stages. As shown by whole-mount ISH (Figure 5A), Eya1 expression was strongly detected in the MM progenitors of wild-type or Eya1^Six1/+ littermate kidneys at E12.5, but only residual Eya1 transcripts were observed in Eya1 ^Six1/+ ;Six2 ^−/− kidneys. Consistent with our previous observation (Xu et al., 2014), anti-Eya1 immunostaining revealed higher levels of Eya1 protein expression in the CM and lower levels in the differentiating RVs of wild-type or Eya1^Six1/+ kidneys (Figure 5B). Co-immunostaining with a Six1-specific antibody confirmed co-expression of Six1 with Eya1 in Eya1^Six1/+ but not in wild-type kidneys (Figure 5B). However, in Eya1 ^Six1/+ ;Six2 ^−/− kidneys, neither Eya1 nor Six1 protein expression was observed. Thus, Six1 cannot replace Six2 to maintain Eya1 expression in the MM progenitors in Eya1 ^Six1/+ ;Six2 ^−/− at E12.5.

Genome-wide ChIP-seq analysis identifies selective binding of Six1 to a subset of Six2 targets essential for nephron progenitor cell maintenance but no co-occupancy of Six1 with Six2 to genes involved in nephron differentiation such as Wnt4, Fgf8, and Pax8.

Our data suggest that Six1 cannot drive developmental programs regulated by Six2 to expand and maintain the nephron progenitors, thus implying that Six1 cannot target Six2-occupied sites in the nephron progenitors. In the human kidney, while ChIP-seq identified more SIX2 binding sites than SIX1, ∼81% (1307 of 1610 peaks) of SIX1 peaks overlapped with SIX2 peaks (∼20.8% of 6276 SIX2 peaks) (O'Brien et al., 2016). However, the functional cooperation and relative contributions of SIX1 versus SIX2 to progenitor cell maintenance and differentiation remain unknown. To directly address the potential functional differences between the mouse Six1 and Six2, we performed ChIP-seq to map Six1-bound regions genome-wide in Eya1^Six1/+ kidneys. Immunohistochemistry with a Six1-specific antibody confirmed Six1 protein expression in the CM of Eya1^Six1/+ at E16.5 (Figure 6A), indicating the stability of the overexpressed Six1 protein in the nephron progenitors. We then used kidneys ∼ E13.5 for Six1 ChIP-seq to avoid differentiating structures, and the peak bed files were generated from MACS peak calling against both genomic input DNA and IgG ChIP-seq controls with the default setting (10⁻⁵ cutoff). Peak calling identified 148 and 2527 Six1 peaks respectively from two different datasets, and we focused on the 2527 peaks (2206 genes) for subsequent analyses and comparison with our recent Six2 ChIP-seq data sets on wild-type kidneys at the same developmental stage (Li et al., 2021) (Figures 6B,C, Supplementary Tables S1,S2). Only 190 Six1 regions (∼7.5%) overlapped with Six2 peaks (Supplementary Table S3), but ∼41% of putative Six1 target genes (906 out of 2206 genes) were shared with Six2 (Supplementary Table S4). While a significant fraction of Six1 peaks were located near the promoter, ∼29% were found between 5–500 kb of TSSs (transcriptional start sites) and the majority of these peaks were distributed within intronic and intergenic regions (Figure 6D). Homer de novo motif search revealed that Six1 or Six2 motifs were not among the top mostly enriched motifs in the Six1 peaks (Figure 6E). Motifs for transcription regulators RELB, Stat3, Tcf21, Osr2, and Smad4 are among the top five enriched motifs. Gene Ontology (GO) analyses using GREAT did not identify nephron-specific terms, but revealed overrepresentation of genes associated with chromatin modification, histone modification, and transition of mitotic cell cycle (Figure 6F).

Eya1 is a critical factor for the maintenance of the nephron progenitors and interacts with Myc and Six2 (Xu et al., 2014), but Six2 is likely involved in maintaining high levels of Eya1 expression as Six2 binding to multiple regions at the Eya1 locus were identified in E13.5 and E16.5 kidneys (Park et al., 2012; O'Brien et al., 2016; Li et al., 2021). Our data suggest that Six1 cannot replace Six2 to maintain Eya1 expression in the progenitors (Figure 5), we therefore asked if Six1 can target the Six2-bound regions at the Eya1 locus to replace the Six2 function in enhancing Eya1 regulatory expression. However, peak calling did not identify Six1 occupancy to the Eya1 locus (Supplementary Table S2). Among the 190 regions shared between Six1 and Six2, Six1 enrichment was identified at ∼112-kb upstream of the Six2 promoter (Figure 6G), representing one of the several Six2 targeted sites with the active histone mark H3K27ac deposition. Similarly, Six1 enrichment was identified at a distal region ∼38-kb downstream of the Mycn promotor (Figure 7A), which is one of the several Six2-bound sites at the Mycn locus and is a highly conserved region associated with H3K27ac-deposition. Thus, Six1 appears to selectively target a small subset of Six2-bound sites at genes essential for nephron progenitor maintenance, which may explain why the number of mitotic MM progenitor cells was more reduced in Six2 ^−/− than in Eya1 ^Six1/+ ;Six2 ^−/− (Figure 3D). The lack of Six1 co-occupancy at Six2-target sites of the Eya1 locus may at least partially lead to the disappearance of Eya1 expression in Eya1 ^Six1/+ ;Six2 ^−/− .

Previous studies have suggested a role for Six2 in inducing nephron differentiation as Six2 targets regions associated with H3K27ac-deposition at genes essential for nephron differentiation, such as Fgf8, Pax8, and Wnt4 (Park et al., 2012; O'Brien et al., 2016). In contrast, Six1 binding to these putative Six2 target genes was not observed (Supplementary Table S2). Interestingly, however, co-occupancy of Six1 with Six2 was identified at the promotor regions of the enhancer of zeste homolog Ezh1 and Ezh2, which had H3K27ac-deposition (Figure 7A). Ezh1 and Ezh2 are polycomb histone methyltransferases, and both are required for the renewal potential of nephron progenitors and co-regulate chromatin accessibility (Liu et al., 2020). Since inactivation of both Ezh1 and Ezh2 triggers unscheduled activation of Wnt4-driven differentiation and results in early termination of nephrogenesis (Liu et al., 2020), Six1 binding to these two genes could maintain the expression of both Ezh1 and Ezh2 in the absence of Six2, which in turn prevents ectopic activation of Wnt4 in the nephron progenitors in Eya1 ^Six1/+ ;Six2 ^−/− kidneys.

We further compared with human SIX1 ChIP-seq data in the fetal kidneys and found only 58 (∼2.3%) of mouse Six1 peaks overlapped with human SIX1-binding sites and ∼27% of putative Six1 target genes shared with human SIX1 (Figure 7B, Supplementary Table S5,S6), despite the high conservation between Six1 and SIX1 (Figure 7C). Previous studies identified only ∼8% of mouse Six2-binding sites but ∼50% of putative Six2 target genes overlapped with human SIX2 targets (O'Brien et al., 2016). Sequence comparison between these proteins revealed that the C-terminal domain is quite divergent in length between Six1 and Six2 with a low degree of sequence similarity, and the C-terminus of human SIX2 is shorter than the mouse Six2 (Figure 7C). This structural difference may affect the stability of forming complexes with specific partner proteins, thus resulting in Six1 selectively binding to only a small subset of Six2 target sites.