www.kidney-international.org basic research

Haploinsufficiency for the Six2 increases nephron progenitor proliferation promoting branching and nephron number Alexander N. Combes1,2, Sean Wilson2, Belinda Phipson2, Brandon B. Binnie3, Adler Ju3, Kynan T. Lawlor2, Cristina Cebrian4, Sarah L. Walton5, Ian M. Smyth6,7,8, Karen M. Moritz5, Raphael Kopan9, Alicia Oshlack2 and Melissa H. Little2,10

1Department of Anatomy and Neuroscience, University of Melbourne, Melbourne, Victoria, Australia; 2Murdoch Children’s Research Institute, Parkville, Victoria, Australia; 3Division of Cell Biology and Molecular Medicine, Institute for Molecular Bioscience, The University of Queensland, Brisbane, Queensland, Australia; 4Division of Gastroenterology, Department of Internal Medicine, University of Michigan, Ann Arbor, Michigan, USA; 5School of Biomedical Sciences and Centre for Children’s Health Research, The University of Queensland, Brisbane, Queensland, Australia; 6Department of Biochemistry and Molecular Biology, Monash University, Clayton, Melbourne, Australia; 7Monash Biomedicine Discovery Institute, Monash University, Clayton, Victoria, Australia; 8Department of Anatomy and Developmental Biology, Monash University, Clayton, Victoria, Australia; 9Division of Developmental Biology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio, USA; and 10Department of Paediatrics, The University of Melbourne, Melbourne, Victoria, Australia

The regulation of final nephron number in the kidney is KEYWORDS: genetics; imaging; kidney development; transcription poorly understood. Cessation of nephron formation occurs regulation ª when the self-renewing nephron progenitor population Copyright 2017, International Society of Nephrology. Published by Elsevier Inc. All rights reserved. commits to differentiation. Transcription factors within this progenitor population, such as SIX2, are assumed to control expression of promoting self-renewal such that homozygous Six2 deletion results in premature ephron number varies in both humans and mice, with commitment and an early halt to kidney development. In pathologic consequences for individuals on the lower N 1 contrast, Six2 heterozygotes were assumed to be end of the spectrum. Nephrons are induced to form unaffected. Using quantitative morphometry, we found a from a self-renewing nephron progenitor (NP) population fi that promotes branching in the adjacent ureteric tip and paradoxical 18% increase in ureteric branching and nal 2 nephron number in Six2 heterozygotes, despite evidence responds to tip-produced signals to form nephrons. Limiting the number of nephron progenitors by restricting fibroblast for reduced levels of SIX2 and transcript. This was 3 accompanied by a clear shift in nephron progenitor growth factor (FGF) signaling or ablating a portion of the 4 fi identity with a distinct subset of downregulated progenitor progenitor population reduces the nal number of nephrons. genes such as Cited1 and Meox1 while other genes were Conversely, preventing progenitor differentiation shortly after birth by chemical inhibition of SMAD signaling resulted in a unaffected. The net result was an increase in nephron 5 progenitor proliferation, as assessed by elevated EdU modest increase in nephron number. Thus, regulating the 0 balance of progenitor self-renewal and differentiation has a (5-ethynyl-2 -deoxyuridine) labeling, an increase in fi protein, and transcriptional upregulation of MYC target signi cant bearing on nephron endowment. genes. Heterozygosity for Six2 on an Fgf20-/- background The SIX2, which is expressed in the resulted in premature differentiation of the progenitor NP population of the developing kidney in both mice and fi humans, plays a central role in maintaining a functional pool population, con rming that progenitor regulation is 6,7 compromised in Six2 heterozygotes. Overall, our studies of self-renewing NPs. SIX2 is presumed to act by reveal a unique dose response of nephron progenitors to suppressing NP differentiation and driving self-renewal. Ho- mozygous loss of Six2 in mice causes a global and early-onset the level of SIX2 protein in which the role of SIX2 in 7 progenitor proliferation versus self-renewal is separable. loss of self-renewal and premature differentiation of NPs. This early loss of progenitors severely reduces organ size Kidney International (2017) -, -–-; https://doi.org/10.1016/ j.kint.2017.09.015 because it is also the NP population that drives ureteric branching via the production of factors such as glial cell- derived neurotrophic factor (GDNF).8 Conversely, over- expression of Six2 in mice prevents NP differentiation.9 Correspondence: Melissa Little, Murdoch Children’s Research Institute, Flemington Genome-wide binding studies in mice and humans have Road, Melbourne, Victoria 3052, Australia. E-mail: [email protected] Or Alex- been used to identify hundreds to thousands of genomic loci ander N. Combes. E-mail: [email protected] potentially regulated by SIX2.10,11 As these include genes Received 5 June 2017; revised 3 September 2017; accepted 7 expressed in NPs and resulting nephrons, the assumption is September 2017 that SIX2 can both activate and suppress transcription.

Kidney International (2017) -, -–- 1 basic research AN Combes et al.: Increased nephron number in Six2 heterozygotes

Evidence of binding of SIX2 to its own promoter also sug- genes previously associated with the most progenitor-like gested an autoregulatory feedback loop.12 SIX2Hi NP population. This shift in NP heterogeneity resul- Cellular heterogeneity within the NP population is evident ted in a global increase in progenitor proliferation accom- in variations in the level of expression of genes including panied by evidence of increased MYC protein and MYC þ þ Cited1 and Meox1, with an assumption that Cited1 Six2 NP pathway activity. As such, this suggests a bimodal response of cells are less committed than NP cells expressing Six2 the developing kidney to SIX2 protein levels and suggests a alone.13,14 Indeed, we have demonstrated that SIX2 levels previously unappreciated dose- and target-sensitive separa- within individual NP cells appears to correlate with variation tion between the role of SIX2 in self-renewal and progenitor in cell-cycle length; high SIX2 is associated with slower proliferation. cycling NP cells, whereas lower SIX2 levels are seen in faster 15 cycling NP cells closer to the site of nephron formation. RESULTS One interpretation of this observation is that the progenitor Increased branching, proliferation, and nephron number in population is gradually progressing from an uncommitted to Six2GCE/D mice þ a committed state as SIX2 protein levels change. Despite this, We first investigated whether the Six2GCE/ was hap- the Six2 heterozygous state has been regarded as having loinsufficient by quantitative polymerase chain reaction and normal kidney development. Indeed, the Six2GCE mouse line Western blotting of whole 15.5-day post coitum (dpc) þ (Six2GCE/ )6 is heterozygous for Six2 but has been used in a embryonic kidneys. This revealed a reduction of both Six2 þ number of studies to conditionally delete genes within NPs on transcript and SIX2 protein by w50% in Six2GCE/ mice þ þ the assumption that the background phenotype is wild compared with Six2 / controls (Figure 1a–f, Supplementary – type.10,16 21 Figure S1). Expression of Cited1, a marker of the uncom- þ Given the importance of Six2 in NP regulation, we used mitted NP subpopulation, was reduced by 70% in Six2GCE/ multiscale imaging and transcriptional profiling to more mice (Figure 1g). Comparison of a pure population of iso- þ þ þ þ carefully examine this Six2GCE/ mouse strain. As Six2 lated NPs from Six2GCE/ and Six2 / kidneys was not knockout mice show a complete collapse of kidney develop- feasible without introducing an additional reporter to label ment, we anticipated an intermediate phenotype between wild-type NPs, which may have had phenotypic conse- wild-type and Six2 null mice. Although reduced SIX2 protein quences. For example, another common NP reporter, the þ was evident in the kidneys of Six2GCE/ mice, contrary to Six2-TGC BAC transgenic reporter, has a distinct heterozy- expectations, branching and nephron number were increased. gous phenotype independent of the Six2 locus (A.N. Combes At the transcriptional level, no change was observed for some & A.N. Graham 2016, unpublished data). Based on the markers of NP identity, including Sall1, Pax2, Wt1, and Gdnf; phenotype of the total Six2 knockout, we speculated that however, there was a clear reduction in a distinct subset of NP reducing Six2 levels would result in increased differentiation

Figure 1 | Validation of haploinsufficiency and compromised progenitor state. (a) Six2 mRNA is reduced by 50% in Six2GCE/þ compared with Six2þ/þ at 15.5 days post coitum (dpc). Data are the average of 3 biological replicates from each genotype, P ¼ 0.0003. mRNA levels expressed relative to Six2þ/þ.(b) Western blot bands for SIX2, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and PAX2 at 15.5 dpc with 4 biological replicates per genotype. See Supplementary Figure S1 for full gels. (c) Densitometry analysis showing reduced SIX2 relative to GAPDH (P ¼ 0.014), no change in PAX2 relative to GAPDH (P ¼ 0.256), reduced SIX2 relative to PAX2 (P ¼ 0.049). Data represent the average of 4 biological replicates. (d,e) Pseudocolored map of SIX2 intensity in representative Six2þ/þ and Six2GCE/þ samples imaged and displayed on the same settings. Color scale indicates fluorescence intensity units. Bar ¼ 100 mm. (f) Box and whisker plot of mean SIX2 intensity per cap cluster for Six2þ/þ (N ¼ 104) versus Six2GCE/þ (N ¼ 55), 2-sample t test with Welch’s correction (P ¼ 6.4e-10). (g) Cited1 mRNA levels are dramatically reduced in the Six2GCE/þ compared with Six2þ/þ (P ¼ 0.00005). Error bars on all graphs represent SEM, P values from 2-tailed t tests with Welch’s correction. ns, not significant; Rel., relative; TBP, tata-box binding protein mRNA. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.

2 Kidney International (2017) -, -–- AN Combes et al.: Increased nephron number in Six2 heterozygotes basic research

and a decrease in branching. Potential changes in branching later in development. By P0, the number of NP cells per niche were assessed with optical projection tomography by and the total NP population was reduced by w10%, whereas measuring the number of NP fields (niche number), which tip cell number was unchanged (Figure 2c and e). The reports the amount of branching in the whole organ.15,22 decrease in NP cells per niche at this stage could reflect a Counter to expectations, niche number was increased in faster rate of NP differentiation, in line with reduced Cited1 þ þ þ þ Six2GCE/ compared with Six2 / kidneys (23% at 15.5 dpc, expression in Six2GCE/ ; however, the timing of cessation of 14% at 19.5 dpc/postnatal day [P] 0 and 18% at P2; Figure 2a nephron formation was not altered. NP cells were present at and b). Confocal analysis did not detect a significant reduc- P2 (Figure 2a) and had differentiated by P4 (data not shown) þ tion in the average number of NP and tip cells per niche at in Six2GCE/ and controls. 5-Ethynyl-20-deoxyuridine (EdU) 15.5 dpc (Figure 2c and d). Multiplying the average number incorporation at 13.5 and 18.5 dpc revealed a significant and of NPs per niche by niche number results in an w10% in- sustained increase in NP proliferation across development crease in the NP population at 15.5 dpc. We previously (Figure 2f). Unbiased stereology was used to determine characterized the rate of branching across time and found whether the increased branching and NP population size that the most rapid phase of branching occurs before 15.5 resulted in an increase in final glomerular number. At P21, þ dpc,15 so a transient increase in the progenitor population at total glomerular counts were increased by 18% in Six2 /GCE and before this time could drive a lasting increase in (Figure 2g). Because the number of nephrons per tip at P4 branching, even if the total number of progenitors decreases was unchanged (data not shown), the increased final number

Figure 2 | Increased branching, tip proliferation, and nephron endowment in Six2GCE/D. (a) Whole-organ optical projection tomography of SIX2-antibody stained Six2þ/þ and Six2GCE/þ kidneys at 15.5 days post coitum (dpc), 19.5 dpc, and postnatal day (P) 2. Bar ¼ 300 mm for all. Exposures were optimized for each sample; hence, signal intensity should not be compared between images. (b) Niche counts from optical projection tomography data for 15.5 dpc (N ¼ 9 Six2þ/þ,8Six2GCE/þ), 19.5 dpc (N ¼ 8 Six2þ/þ,8Six2GCE/þ), and P2 (N ¼ 8 Six2þ/þ,9Six2GCE/þ). (c) Maximum intensity projection confocal data from 15.5 dpc and 19.5 dpc Six2þ/þ and Six2GCE/þ kidneys stained with SIX2 (red) and CYTOK (white). Bars ¼ 30 mm. Exposures were optimized for each sample; hence, signal intensity should not be compared between images. (d) Tip and cap cell number at 15.5 dpc. Each data point represents the average cell number per niche in an individual sample; sample numbers ¼ 9 Six2þ/þ and 8 Six2GCE/þ.(e) Tip and cap cell number at 19.5 dpc. Points per part d, N ¼ 9 Six2þ/þ,8Six2GCE/þ.(f) Percentage of 5-ethynyl-20-deoxyuridine incorporation for the nephron progenitor population at 13.5 dpc and 18.5 dpc 30 minutes after exposure to 5-ethynyl-20-deoxyuridine. Each data point represents the percentage of incorporation per sample; 13.5 and 18.5 time points were separate experiments and relative levels of proliferation should not be compared between them. (g) Comparison of estimated glomerular number between P21 Six2þ/þ and Six2GCE/þ by stereology; N ¼ 10 Six2þ/þ, N ¼ 12 Six2GCE/þ. Error bars in all graphs represent SEM, P values determined by 1-tailed t test with Welch’s correction. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.

Kidney International (2017) -, -–- 3 basic research AN Combes et al.: Increased nephron number in Six2 heterozygotes

of nephrons was associated with increased NP proliferation change [FC]) and Crym (0.40 FC) were further down- and branching rather than an increase in the number of regulated in 11.5 dpc Six2GCE/GCE (Figure 3b). This represents nephrons forming around each tip. a dose-sensitive (haploinsufficient) response pattern and included Cited1 (0.33 FC at 15.5 dpc), Meox1 (0.70 FC), Crym – Transcriptional comparison between Six2D/D, Six2GCE/D, and (0.59 FC), Hoxd12 (0.72 FC), and Phf19 (0.74 FC).14,23 27 Six2GCE/GCE kidneys Most of these represent genes previously identified as The global transcriptional changes observed in the Six2 ho- marking a subset of the NP population regarded as the most mozygous or heterozygous state have not previously been uncommitted. Considering all differentially regulated genes, examined. To examine the transcriptional response to changes these were enriched for genes known to be expressed in NPs in Six2 levels, whole-kidney RNA-Seq was performed on and predicted to be regulated by SIX2. Of the 40 genes þ þ þ þ þ þ Six2GCE/ and Six2 / kidneys at 15.5 dpc as well as Six2 / , significantly downregulated at 15.5 dpc in the Six2GCE/ state, þ Six2GCE/ , and Six2GCE/GCE kidneys at 11.5 dpc 83% represented NP markers and 65% were associated with a (Supplementary Table S1). As expected, Six2 mRNA levels in previously described SIX2 binding site.10,11 þ Six2GCE/ were w50% of wild type. The top differentially analysis28 of downregulated genes revealed a link to meta- expressed genes from RNA-seq at 11.5 and 15.5 were cross- bolism, negative regulation of glycolysis, and muscle-related validated by real-time polymerase chain reaction terms, consistent with a change in proliferative capacity and þ (Supplementary Figure S2). The Six2GCE/ construct drives the previously described role for SIX in regulating the expression of GFP and Cre recombinase in place of the myogenesis29 (Supplementary Table S2). Of the 13 upregu- native Six2 transcript.6 To test the possibility that the tran- lated genes at 15.5 dpc, 9 were previously associated with a scriptional and phenotypic changes observed in this strain SIX2 ChIP peak and 7 were expressed in the NPs. A similar resulted from the expression of recombinant GFP-CreERT2 pattern was observed at 11.5 dpc (Figure 3b). within the NPs, quantitative polymerase chain reaction for Despite evidence of a clear transcriptional effect within the þ genes both up- and downregulated in Six2GCE/ kidneys was NP population, even in the heterozygous state, many NP performed in kidneys from GdnfGdnf-CreERT2 mice.4 These genes and previously proposed SIX2 target genes showed no þ genes were not differentially expressed in the GdnfGdnf-CreERT2 change in expression in the 15.5 dpc Six2GCE/ kidney. For line (Supplementary Figure S3), consistent with a specific example, Gdnf, Wnt4, and Fgf8, all previously proposed as – response to Six2 haploinsufficiency. direct SIX2 target genes,7,10 12 were not significantly changed, whereas Eya1 was only moderately upregulated (1.2 FC Six2GCE/D kidneys display altered expression in only a subset adjusted P ¼ 8.3E-6) (Figure 3c). In addition, although of NP markers and previously described SIX2 targets branching was increased, no substantial changes were A subset of NP marker genes were differentially expressed in observed in genes associated with the regulation of branching þ Six2GCE/ kidneys at 15.5 dpc (Figure 3a). At 11.5 dpc, morphogenesis or NP-stroma interactions (Figure 3c, and downregulated marker genes including Cited1 (0.38 fold data not shown). This would suggest that many genes affected

Figure 3 | changes in the Six2GCE/D kidney. (a) Genes differentially expressed (DE) at #0.4 or $0.4 logFC of Six2þ/þ values at 15.5 days post coitum (dpc) in Six2GCE/þ and #1or$1 logFC of wild type in 11.5 dpc Six2GCE/GCE; adjusted P < 0.05 for all. Key indicates fold change in expression compared with wild type for all parts. For all parts, W ¼ wild type/Six2þ/þ,H¼ heterozygous/Six2GCEþ,K¼ knockout/ Six2GCE/GCE.(b) Genes expressed at #0.4 or $0.4 logFC of wild-type values at 11.5 dpc in Six2GCE/þ and #1or$1 logFC of wild type in Six2GCE/GCE-adjusted P < 0.05. (c) Examples are given of gene expression for some established SIX2 targets and branching genes in Six2GCE/þ and Six2GCE/GCE; aside from Six2, none of these changes were statistically significant in the Six2GCE/þ. FC, fold change.

4 Kidney International (2017) -, -–- AN Combes et al.: Increased nephron number in Six2 heterozygotes basic research

þ by total deletion of Six2 were insensitive to this reduced level Six2GCE/ mice, Eya1 mRNA levels were only increased 1.2- of protein. fold at 15.5 dpc, and there was no difference in MYC T58 phosphorylation or the prevalence of T58 positive cells within Evidence of differential dose-sensitivity in Six2 target genes the NP by immunofluorescence (Supplementary Figure S4). Comparison of transcriptional changes in the homozygous This suggests a different mechanism for elevation of MYC state compared with wild type (Figure 4a) provided an protein. Investigation of the mTOR pathway and phosphati- 0 opportunity to validate a role for Six2 in the regulation of dylinositol-3 -kinase-p110-alpha protein levels failed to detect previously proposed target genes. Comparing all differentially any significant change (Supplementary Figure S4). Thus, the expressed genes in Six2GCE/GCE kidneys (adjusted P < 0.05, no changes in NP proliferation are associated with an increase in fold-change cutoff) with genes associated with SIX2 binding in MYC protein levels, either directly or indirectly, due to mouse10,11 revealed that only 8.8% of potential SIX2 targets changes in SIX2 protein levels. responded to complete loss of SIX2 protein in vivo. A complete description of the transcriptional response observed in the Six2 heterozygous NPs are compromised with respect to homozygous state can be found in the Supplementary Material. self-renewal GCE/þ þ/þ Indeed, by identifying the overlapping genes showing differ- Although Six2 kidneys branched more than Six2 GCE/þ ential regulation in both the heterozygous and homozygous kidneys, it remained possible that the Six2 NP domain states at 11.5 dpc, it is possible to identify those genes most represents a sensitized progenitor state. Fgf9 and Fgf20 are 3 sensitive to reduced levels of SIX2 protein. Again, this high- known to be critical for NP survival and proliferation. / lights those genes previously identified as marking the un- Kidneys from Fgf20 mice display reduced NP prolifera- committed NP domain. Although reduced expression of these tion and reduced niche thickness at birth but otherwise 3 genes is associated with the loss of NP self-renewal/increased appear normal. More careful analysis of kidney morpho- / NP commitment in the homozygote state, in the heterozy- genesis in the Fgf20 mice by optical projection tomogra- gous state, this was accompanied by an increased NP niche phy showed a 15% reduction in ureteric branching (Supplementary Figure S5). These mice were crossed with the number and increased ureteric branching. þ Six2GCE/ strain and the expression of SIX2 protein, and Six2 Increased CM proliferation is associated with elevated MYC and Cited1 mRNA examined across all potential genotypes protein levels in Six2 haploinsufficient kidneys (Supplementary Figure S5). Although removal of 1 or both þ As the Six2GCE/ phenotype results in increased branching copies of Fgf20 did not result in a further reduction in Six2 GCE/þ / with a more proliferative NP population, we looked for evi- levels in Six2 mice, Six2 levels were reduced in Fgf20 dence of a proliferative signature within the differentially kidneys at 12.5 dpc but not 15.5 dpc (Supplementary GCE/þ / expressed genes at 15.5 dpc. One of the genes most signifi- Figure S5). Six2 Fgf20 kidneys were severely hypo- cantly downregulated at both 15.5 dpc and 11.5 dpc (0.42 FC plastic and depleted of NPs by 15.5 dpc (Figure 5a and b). At GCE/þ / and 0.53 FC, respectively) was Actn3b. ACTN3B protein levels 12.5 dpc, Six2 Fgf20 kidneys had weak SIX2 expres- þ were also reduced by >50% at 15.5 dpc in Six2GCE/ sion and ectopic renal vesicles forming around most tips 7 (Supplementary Figure S2). This decrease in ACTN3B is (Figure 5c–h), reminiscent of the Six2 knockout. Although associated with an increase in metabolic efficiency, also likely FGF signaling plays an important role in NP maintenance, the GCE/ to influence progenitor proliferation.30 In addition, genes lack of further reduction of Six2 levels in the Six2 þ þ þ þ / differentially expressed between the Six2GCE/ versus Six2 / Fgf20 kidneys suggests that Fgf20 is not directly regu- kidney at 15.5 dpc were tested for enrichment of the Broad lating Six2 levels. Fgf20, on the other hand, is bound by 10 GCE/GCE Institute’s Hallmark gene sets using the CAMERA gene set SIX2 and substantially downregulated in Six2 kid- GCE/þ test.31 At 15.5 dpc, upregulated genes were significantly neys. These results primarily show that Six2 NPs are enriched in pathways associated with mTORC1 (mammalian sensitized to differentiation as Fgf20 deletion alone does not target of rapamycin complex 1) signaling, and MYC target result in this phenotype. This is functional evidence that the GCE/þ genes (Figure 4b). These results were confirmed at 11.5 dpc NP state is compromised in the Six2 kidneys and is an with an additional signature for phosphoinositide 3-kinase– important caution for future studies that may use this strain AKT-mTOR signaling (Figure 4b, Supplementary Table S2). as an inducible Cre, assuming that Six2 heterozygosity will Although Myc is associated with SIX2 binding and therefore not influence the experimental outcome. potentially regulated by SIX2,10 Myc mRNA was not upre- gulated in the whole-kidney data. However, using Western DISCUSSION blotting, we could confirm that MYC protein levels were This study not only reveals a kidney phenotype in Six2 het- þ þ þ increased by 19% in Six2GCE/ compared with Six2 / across erozygous mice, but a paradoxical increase in nephron number multiple litters (P ¼ 0.009) (Figure 4c). EYA1, SIX2, and and branching events despite clear evidence of a transcriptional MYC have been proposed to cooperatively regulate NP dysregulation of a subset of SIX2 target genes. This variation in expansion, with SIX2 mediating EYA1 nuclear translocation, phenotype from wild type to heterozygote to homozygote and EYA1 dephosphorylating MYC T58 to prevent protein suggests a dose-sensitive separation between the regulation of degradation and promote progenitor expansion.32 In NP proliferation versus self-renewal. What is evident is that as

Kidney International (2017) -, -–- 5 basic research AN Combes et al.: Increased nephron number in Six2 heterozygotes

SIX2 protein levels decrease, inhibition of NP proliferation is relaxed before a loss of NP self-renewal. The NP population expands as a consequence of increased proliferation associated with increased MYC protein and MYC pathway signaling. This expanded NP population drives increased branching, appar- ently without any requirement for an overall upregulation of Gdnf expression levels but simply an increase in the number of NP cells present. As SIX2 protein levels fall further, the loss of regulation of self-renewal is affected, tipping the NP population into nephron commitment.7 This immediately blocks branching by reducing the size of the NP population. Of in- terest, what we did not observe is a change in the timing of cessation of nephrogenesis in the Six2 heterozygous state. As is seen in the wild-type situation, the NPs eventually reduced in cell number per niche. The overall increase in nephron number in this genotype, therefore, suggests that each tip was able to generate a normal number of nephrons, supporting a genuine expansion of the progenitor pool without a change in the timing of cessation. þ Expansion of the progenitor pool in Six2GCE/ was clearly driven by an increase in NP proliferation, with evidence of increased EdU labeling, increased MYC protein levels, and hallmarks of increased MYC pathway activity. Even modest increases in MYC levels have previously been shown to pro- mote dose-dependent expansion of cell populations in the fly and mouse33,34 and to expand human pluripotent cardiac progenitors.35 Myc was previously shown to regulate NP turnover, and deletion of Myc within the NP domain reduces NP number and kidney volume.32,36 The increase in progenitor þ proliferation in Six2GCE/ may also be complemented by the de-repression or compensatory upregulation of Fgf1, Adra1, and Klhdc8a, all of which were upregulated in the heterozygous state by 15.5 dpc and all of which are associated with SIX2 binding10 and increased growth in other contexts.37,38 In particular, the role of Fgf family members regulating NP state and proliferation is well established. This study identified a high correlation between differen- þ tially expressed genes in the Six2GCE/ mouse strain with genes that are associated with SIX2 binding.10,11 However, it also revealed some unexpected findings. Established SIX2 targets Gdnf and Wnt4 were not affected by reduced SIX2 protein levels. Even in Six2 null kidneys, there was only a modest reduction in SIX2 target genes Gdnf and Eya1, indi- cating that SIX2 is not the only regulator of these genes. The Six2 locus itself has a documented autoregulatory – element,10 12 which did not maintain Six2 expression levels in the heterozygous state. Six2 is proposed to silence drivers of early nephrogenesis, including Wnt4 and Fgf8.11 Wnt4 was = (MYC targets V1 and V2), progenitor maintenance (mTOR [mamma- Figure 4 | Analysis of transcriptional changes. (a) Unsupervised lian target of rapamycin complex], mTORC1 [mTOR complex 1] clustering of the top 250 differentially expressed genes between 11.5 signaling), and differentiation (phosphoinositide 3-kinase signaling) in days post coitum (dpc) Six2GCE/GCE and Six2þ/þ. Each row represents a Six2GCE/þ. The pathways associated with MYC and mTORC1 are gene and each column an independent sample. Color indicates downregulated in Six2GCE/GCE.(c) MYC protein levels are significantly normalized expression (Z score): blue is low, red high. (b) Gene set increased in the Six2GCE/þ. Points represent average values from 3 testing reveals an upregulation of pathways associated with growth independent litters; unpaired t test.

6 Kidney International (2017) -, -–- AN Combes et al.: Increased nephron number in Six2 heterozygotes basic research

Figure 5 | Nephron progenitors lacking 1 copy of Six2 undergo premature differentiation in the absence of Fgf20. (a) Staining of the NP cells (SIX2, red), ureteric tree (ECAD protein, green), and nuclei (40,6-diamidino-2-phenylindole, blue) in a 15.5 days post coitum (dpc) Six2þ/ þFgf20/ kidney. Bar ¼ 200 mm. (b) Nephron progenitors are depleted before 15.5 dpc in Six2GCE/þFgf20/ kidneys. Dilated nephrons are observed (arrows), and the ureteric tree is underdeveloped. Staining and scale as in a.(c) Maximum intensity projection of the NPs (SIX2, green), tree (ECAD, gray), renal vesicles (JAG1 protein, red), and nuclei (40,6-diamidino-2-phenylindole, blue) in a 12.5-dpc Six2þ/þFgf20þ/ kidney; bar ¼ 50 mm. (d) Rendering of the tree (gray) and renal (red) vesicles in c; bar ¼ 50 mm. (e) Zoom of boxed region in e shows t-stage tip with 2 renal vesicles attached on the medullary side; bar ¼ 20 mm. (f) A 12.5-dpc Six2GCE/þFgf20/ kidney as in c; bar ¼ 50 mm. (g) Rendering from f; bar ¼ 50 mm. (h) Zoom of region in g showing t-stage tip with 5 renal vesicles, 2 in ectopic positions; bar ¼ 20 mm. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org. upregulated in Six2GCE/GCE, but Fgf8 was unchanged. In coexpressed in NPs.10 The data presented here would not contrast, multiple genes previously identified as WT1 tar- predict any pathogenicity from a heterozygous mutation in þ gets26 were among the top downregulated genes in the Six2. It should also be noted that the Six2GCE/ strain is also Six2GCE/GCE, consistent with a model in which NP mainte- used for conditional deletion of other genes and lineage – þ nance is regulated by multiple transcription factors working tracing.10,16 21 An interaction between the Six2GCE/ pheno- redundantly on a set of core target genes. type and experiments using this line should be considered The observed increase in nephron number in Six2 het- when interpreting these studies. As demonstrated with the þ erozygotes has implications for our understanding of Six2GCE/ Fgf20 / cross, Six2 heterozygosity has the capacity congenital anomalies of the kidney and urinary tract to exacerbate otherwise modest phenotypes. (CAKUT) in humans. Despite an initial association between In summary, we provide evidence here of distinct phe- heterozygous Six2 mutations and a renal phenotype,39 sub- notypes dependent on the amount of SIX2 protein present sequent analysis of large cohorts of CAKUT patients cast within the NPs of the developing kidney. As a result of dif- doubt on the clinical relevance of some reported mutations ferential target gene sensitivity, haploinsufficiency for Six2 and failed to identify any new associations with Six2.40,41 This caused a paradoxical increase in NP proliferation and is consistent with a high likelihood of functional redundancy increased nephron number in contrast to the loss of the NP between SIX1 and SIX2 in humans, which, unlike in mice, are population in the absence of Six2. Despite increased

Kidney International (2017) -, -–- 7 basic research AN Combes et al.: Increased nephron number in Six2 heterozygotes

proliferation, NPs heterozygous for Six2 are sensitized to The RNA-Seq data are publicly accessible from the Gene Expression differentiation as evidenced by reduced Cited1 expression and Omnibus under accession number GSE103219. þ progenitor depletion on the Six2GCE/ Fgf20 / background. Importantly, although the concept that reduced progenitor DISCLOSURE All the authors declared no competing interests. number results in fewer nephrons is accepted, we provide evidence that the converse is also true. A better understanding ACKNOWLEDGMENTS of how to positively control NP expansion in vivo may have This work was supported by the Australian Research Council therapeutic implications for improving developmental out- (DE150100652), the National Health and Medical Research Council of comes and generating renal tissue in vitro. Australia (APP1002748, APP1063696), the Human Frontiers in Science Program (RGP0039/2011). Microscopy was performed at the ACRF/ IMB Cancer Biology Imaging Facility at Monash University and the CONCISE METHODS Murdoch Children’s Research Institute. ANC holds a Discovery Early Mouse strains and embryo staging Career Researcher Award from the Australian Research Council. MHL In all experiments, noon of the day on which the mating plug was is a Senior Principal Research Fellow of the NHMRC. We thank the IMB observed was designated 0.5 dpc. For postnatal samples, P was Sequencing Facility for NGS services, David Ornitz for the Fgf20 mice, recorded with P0 equivalent to birth. In most instances, birth Kieran Short, Lynelle Jones, Shireen Lamande, Chantal Coles, and þ represented 19.5 dpc. Six2GCE/ mice were used, which carry a tar- Peter Houweling for technical assistance. MCRI is supported by the Victorian Government’s Operational Infrastructure Support Program. geted insertion of EGFP-CRE-ERT2 in the Six2 locus that generates a null allele (JAX stock number 009600).6 Fgf20-null mice ß-gal 42 AUTHOR CONTRIBUTIONS (Fgf20 ) were kindly provided by David Ornitz. All animal ex- SW, BB, AJ, KTL, and CC performed experiments and contributed to periments in this study were assessed and approved by the University data analysis and presentation. BP and AO performed and advised on of Queensland or Murdoch Children’s Research Institute Animal RNA-Seq analysis. SLW and KM performed and advised on nephron Ethics Committees and were conducted under applicable Australian counting. RK facilitated the Fgf20 studies, IMS facilitated access to laws governing the care and use of animals for scientific purposes. optical projection tomography. ANC conceived the study with MHL, Immunofluorescence and image analysis The following primary performed experiments, analyzed and presented the data, and wrote antibodies were used: mouse anti-calbindin D28K (C9848, Sigma- the manuscript. MHL supervised the study and substantially Aldrich, St Louis, MO), mouse anti-cytokeratin (Ab11213 and contributed to writing. All authors contributed to revising and editing Ab115959, Abcam, Cambridge, UK), rabbit anti-SIX2 (11562-1- the manuscript. AP, Proteintech, Manchester, United Kingdom), rabbit anti-JAG1 protein (ab7771, Abcam), mouse anti-SIX2 (H00010736-MO1, SUPPLEMENTARY MATERIAL Abnova, Taipei City, Taiwan), rat anti-Ecad (13-1900, Invitrogen, Table S1. Differential gene expression and gene ontology analysis. Carlsbad, CA). The mouse SIX2 primary was used in conjunction Table S2. Hallmark gene set results. D with an isotype-specific secondary antibody (A21121, Invitrogen) Figure S1. Reduced SIX2 protein levels in Six2GCE/ .(A) Full gels for to reduce background staining. Alexa Fluor-conjugated secondary Western blot results presented in Figure 1. SIX2 protein levels were antibodies (Invitrogen) were used to detect primary antibodies and compared with CM-expressed transcription factor PAX2 and control 40,6-diamidino-2-phenylindole (D8417, Sigma-Aldrich) was used protein glyceraldehyde-3-phosphate dehydrogenase among 4 þ/þ GCE/þ at 1:2000 to label nuclei. Whole-mount immunofluorescence, Six2 and 4 Six2 littermates (left to right on each gel). (B) Quantification of SIX2 levels per cell for 3 independent samples from confocal microscopy, and optical projection tomography was car- þ/þ GCE/þ ried out according to published protocols.22 Cell counts per niche Six2 and Six2 littermates (SIX2 levels per niche presented in Figure 1). Immunostaining and imaging parameters were kept (confocal) and niche counts (optical projection tomography) were consistent across the experiment. Each spot represents a SIX2þ cell, performed as reported.22 Quantitative analysis of SIX2 fluorescence with the color and position on the X axis indicating the SIX2 intensity intensity is detailed in Supplementary Methods. and the position on Y axis indicating the Z depth in the 3-dimensional images from which these data were derived. Note that SIX2 intensity þ þ Proliferation analysis values are consistently higher in the Six2 / samples compared with þ Relative amounts of proliferation were investigated by analysis of the Six2GCE/ littermates. The red line marks 20,000 fluorescence EdU incorporation in CM and ureteric tip, 30 minutes after exposure intensity units for comparison between genotypes. to EdU as assessed by whole-mount confocal imaging and quanti- Figure S2. Quantitative polymerase chain reaction (qPCR) validation þ þ þ tative analysis, as reported.15 A total of 8 Six2 / and 9 Six2 /GCE of gene expression changes and analysis of ACTN3 protein levels. kidneys were assessed at 13.5, and 7 versus 7 at 18.5, each kidney (A,B) Top up- and downregulated genes from RNA-Seq analysis were fi from an individual embryo, and each sample grouping including assessed by qPCR. Most changes were con rmed. Changes that failed kidneys from multiple timed pregnancies. to validate could have been due to substantial differences in gene coverage or sensitivity between qPCR and RNA-Seq. (C) A Western blot–validated antibody was available for ACTN3 protein, which was Glomerular estimation downregulated by 50% at the RNA level. ACTN3 protein levels Glomerular number was estimated using the physical dissector/ were <50% of wild-type values compared with glyceraldehyde- 43 fractionator method, as previously described ; further details are 3-phosphate dehydrogenase (GAPDH) or PAX2. provided in Supplementary Methods. Figure S3. Six2GCE/þ gene expression changes are not caused by expression of Cre-ERT2 in the CM cells. A subset of genes that were Transcriptional profiling using RNA-Seq and Western blotting differentially expressed in the Six2GCE/þ mice were tested for gene Experimental information and analysis workflow for RNA Seq and expression changes in whole kidneys of an unrelated mouse line that Western blotting is detailed in the Supplementary Methods. also expresses CreERT2 in the cap mesenchyme. None of the tested

8 Kidney International (2017) -, -–- AN Combes et al.: Increased nephron number in Six2 heterozygotes basic research

genes were significantly altered, suggesting that the changes in the 3. Barak H, Huh SH, Chen S, et al. FGF9 and FGF20 maintain the stemness of Six2GCE/þ line are not an artifact of Cre-ERT2 expression in the CM cells. nephron progenitors in mice and man. Dev Cell. 2012;22:1191–1207. ’ Figure S4. (A) Protein level analysis with the ProteinSimple WES 4. Cebrian C, Asai N, D Agati V, et al. The number of fetal nephron progenitor cells limits ureteric branching and adult nephron system for cMYC pT58 (Santa Cruz sc-135647), mammalian target of endowment. Cell Rep. 2014;7:127–137. rapamycin (mTOR), phosphorylated (p)-mTOR (Ser24488), phosphor- 5. Brown AC, Muthukrishnan SD, Oxburgh L. A synthetic niche for nephron ylated (p) 4EBP1 (Thr37/46), p-P70 S6 kinase (Ser371), and phos- progenitor cells. Dev Cell. 2015;34:229–241. phoinositide 3-kinase (PI3K) p110alpha. No significant change was 6. Kobayashi A, Valerius MT, Mugford JW, et al. Six2 defines and observed for any of these proteins or modifications. mTOR and regulates a multipotent self-renewing nephron progenitor population phosphoinositide 3-kinase–related antibodies were obtained from throughout mammalian kidney development. Cell Stem Cell. 2008;3: 169–181. Cell Signaling Technology (Danvers, MA) (nos. 9964S, 9862S, and 7. Self M, Lagutin OV, Bowling B, et al. Six2 is required for suppression of 9655S). (B) Representative images from a comparison of SIX2 (green), nephrogenesis and progenitor renewal in the developing kidney. EMBO ECAD (white), and cMYC-T58 (red, Santa Cruz sc-135647, Santa J. 2006;25:5214–5228. Biotechnology, Dallas, TX) protein levels and localization in the 8. Combes AN, Davies JA, Little MH. Cell-cell interactions driving kidney developing kidney at 15.5 days post coitum. Note the higher levels of morphogenesis. Curr Top Dev Biol. 2015;112:467–508. SIX2 protein in Six2þ/þ compared with Six2GCE/þ under the same 9. Chung E, Deacon P, Marable S, et al. Notch signaling promotes – imaging conditions. (C) Quantification of the number of T58- nephrogenesis by downregulating Six2. Development. 2016;143:3907 3913. 10. O’Brien LL, Guo Q, Lee Y, et al. Differential regulation of mouse and expressing cells per cap mesenchyme domain between genotypes þ/þ GCE/þ human nephron progenitors by the Six family of transcriptional (Six2 , N ¼ 132; Six2 , N ¼ 72; N represents the number of CM regulators. Development. 2016;143:595–608. domains) reveals no significant difference (t test with Welch’s 11. Park JS, Ma W, O’Brien LL, et al. Six2 and Wnt regulate self-renewal and correction P ¼ 0.774). Avg., average. commitment of nephron progenitors through shared gene regulatory – Figure S5. Branching is reduced in Fgf20 / ; SIX2 protein, Six2 and networks. Dev Cell. 2012;23:637 651. Cited1 mRNA expression vary with Fgf20 and Six2 genotypes. (A,B) 12. Brodbeck S, Besenbeck B, Englert C. The transcription factor Six2 /þ / activates expression of the Gdnf gene as well as its own promoter. Mech Three-dimensional reconstructions of Fgf20 (A) and Fgf20 (B) Dev. 2004;121:1211–1222. optical projection tomography data at 15.5 days post coitum (dpc) 13. Brown AC, Muthukrishnan SD, Guay JA, et al. Role for with CM domains labeled with SIX2 (red). Bars in both images are 300 compartmentalization in nephron progenitor differentiation. Proc Natl mm. (C) Niche number is reduced by 15% in Fgf20/ compared with Acad Sci U S A. 2013;110:4640–4645. Fgf20/þ. Data points represent biological replicates sourced from 2 14. Mugford JW, Yu J, Kobayashi A, et al. High-resolution gene expression fi litters. Bars represent SEM, P ¼ 0.048 using a 2-tailed t test with analysis of the developing mouse kidney de nes novel cellular compartments within the nephron progenitor population. Dev Biol. Welch’s correction. (D–F) Staining for SIX2 (NPs, red), ECAD (epithe- – 0 2009;333:312 323. lium, green), and 4 ,6-diamidino-2-phenylindole (nuclei, blue) across 15. Short KM, Combes AN, Lefevre J, et al. Global quantification of tissue Six2 and Fgf20 genotypes from 15.5 dpc kidneys (magnified views dynamics in the developing mouse kidney. Dev Cell. 2014;29: from Figure 5, with additional Six2GCE/þFgf20þ/ panel). Note the 188–202. decrease in SIX2 protein between Six2þ/þFgf20/ and Six2GCE/ 16. Taguchi A, Kaku Y, Ohmori T, et al. Redefining the in vivo origin of þFgf20þ/.(G) Six2 mRNA levels at 12.5 and 15.5 dpc with varied Six2 metanephric nephron progenitors enables generation of complex kidney structures from pluripotent stem cells. Cell Stem Cell. 2014;14:53–67. and Fgf20 gene dosage. Six2 mRNA is reduced at 12.5 in Fgf20 / but 17. Yuri S, Nishikawa M, Yanagawa N, et al. Maintenance of Mouse Nephron recovers by 15.5 in agreement with protein levels shown in D and E. Progenitor Cells in Aggregates with Gamma-Secretase Inhibitor. PLoS / Although Six2 levels are reduced in Fgf20 kidneys, introducing Six2 One. 2015;10:e0129242. heterozygosity does not further reduce the levels of Six2 mRNA on 18. Togel F, Valerius MT, Freedman BS, et al. Repair after nephron ablation reveals the Fgf20/ background. Six2 is very low at 15.5 dpc in the limitations of neonatal neonephrogenesis. JCI Insight. 2017;2:e88848. Six2GCE/þFgf20/ cross, which presumably reflects the loss of the CM 19. Xu J, Liu H, Park JS, et al. Osr1 acts downstream of and interacts < synergistically with Six2 to maintain nephron progenitor cells during at this stage. Bars represent SE from 3 biological replicates, *P 0.05 – ’ fi kidney organogenesis. Development. 2014;141:1442 1452. from 2-tailed t test with Welch s correction. Insuf cient 15.5 dpc 20. Huang L, Mokkapati S, Hu Q, et al. Nephron Progenitor But Not Stromal GCE/þ þ/ Six2 Fgf20 samples were obtained for inclusion in this analysis. Progenitor Cells Give Rise to Wilms Tumors in Mouse Models with beta- (H) Cited1 mRNA levels at 12.5 and 15.5 dpc at varying levels of Six2 Catenin Activation or Wt1 Ablation and Igf2 Upregulation. Neoplasia. and Fgf20 gene dosage. Cited1 levels are significantly reduced in 2016;18:71–81. Fgf20/ compared to wildtype mice at 12.5 and 15.5 dpc. Cited1 21. Combes AN, Lefevre JG, Wilson S, et al. Cap mesenchyme cell swarming fl levels are not significantly different between Six2þ/þFgf20/ mice during kidney development is in uenced by attraction, repulsion, and adhesion to the ureteric tip. Dev Biol. 2016;418:297–306. and different Fgf20 genotypes that are also heterozygous for Six2 at GCE/þ þ/ 22. Combes AN, Short KM, Lefevre J, et al. An integrated pipeline for the 12.5 dpc. 15.5 dpc Six2 Fgf20 per G. Rel., relative. multidimensional analysis of branching morphogenesis. Nat Protoc. Supplementary Results. Includes a description of results from 2014;9:2859–2879. profiling Six2-null kidneys at 11.5 dpc. 23. Boyle S, Shioda T, Perantoni AO, et al. Cited1 and Cited2 are differentially Supplementary Methods. Includes a description of methods used expressed in the developing kidney but are not required for fi nephrogenesis. Dev Dyn. 2007;236:2321–2330. for the quanti cation of SIX2 protein intensity by fi fl 24. Brunskill EW, Potter SS. RNA-Seq de nes novel genes, RNA processing immuno uorescence, glomerular estimation, RNA-Seq analysis, and patterns and enhancer maps for the early stages of nephrogenesis: Hox Western blotting. supergenes. Dev Biol. 2012;368:4–17. Supplementary material is linked to the online version of the paper at 25. Karner CM, Das A, Ma Z, et al. Canonical Wnt9b signaling balances www.kidney-international.org. progenitor cell expansion and differentiation during kidney development. Development. 2011;138:1247–1257. 26. Motamedi FJ, Badro DA, Clarkson M, et al. WT1 controls antagonistic FGF REFERENCES and BMP-pSMAD pathways in early renal progenitors. Nat Commun. 1. Hoy WE, Hughson MD, Bertram JF, et al. Nephron number, hypertension, 2014;5:4444. renal disease, and renal failure. J Am Soc Nephrol. 2005;16:2557–2564. 27. Rumballe BA, Georgas KM, Combes AN, et al. Nephron formation adopts 2. Kopan R, Chen S, Little M. Nephron progenitor cells: shifting the balance a novel spatial topology at cessation of nephrogenesis. Dev Biol. of self-renewal and differentiation. Curr Top Dev Biol. 2014;107:293–331. 2011;360:110–122.

Kidney International (2017) -, -–- 9 basic research AN Combes et al.: Increased nephron number in Six2 heterozygotes

28. Chen J, Bardes EE, Aronow BJ, et al. ToppGene Suite for gene list 36. Couillard M, Trudel M. C-myc as a modulator of renal stem/progenitor enrichment analysis and candidate gene prioritization. Nucleic Acids Res. cell population. Dev Dyn. 2009;238:405–414. 2009;37:W305–W311. 37. Mukasa A, Wykosky J, Ligon KL, et al. Mutant EGFR is required for 29. Relaix F, Demignon J, Laclef C, et al. Six homeoproteins directly activate maintenance of glioma growth in vivo, and its ablation leads to escape Myod expression in the gene regulatory networks that control early from dependence. Proc Natl Acad Sci U S A. 2010;107:2616–2621. myogenesis. PLoS Genet. 2013;9:e1003425. 38. Piascik MT, Perez DM. Alpha1-adrenergic receptors: new insights and 30. MacArthur DG, Seto JT, Raftery JM, et al. Loss of ACTN3 gene function directions. J Pharmacol Exp Ther. 2001;298:403–410. alters mouse muscle metabolism and shows evidence of positive 39. Weber S, Taylor JC, Winyard P, et al. SIX2 and BMP4 mutations associate selection in humans. Nat Genet. 2007;39:1261–1265. with anomalous kidney development. J Am Soc Nephrol. 2008;19:891–903. 31. Wu D, Smyth GK. Camera: a competitive gene set test accounting for 40. Faguer S, Chassaing N, Bandin F, et al. Should SIX2 be routinely tested in inter-gene correlation. Nucleic Acids Res. 2012;40:e133. patients with isolated congenital abnormalities of kidneys and/or urinary 32. Xu J, Wong EY, Cheng C, et al. Eya1 interacts with Six2 and Myc to tract (CAKUT)? Eur J Med. Genet. 2012;55:688–689. regulate expansion of the nephron progenitor pool during 41. Hwang DY, Dworschak GC, Kohl S, et al. Mutations in 12 known nephrogenesis. Dev Cell. 2014;31:434–447. dominant disease-causing genes clarify many congenital anomalies of 33. Claveria C, Giovinazzo G, Sierra R, et al. Myc-driven endogenous cell the kidney and urinary tract. Kidney Int. 2014;85:1429–1433. competition in the early mammalian embryo. Nature. 2013;500:39–44. 42. Huh SH, Jones J, Warchol ME, et al. Differentiation of the lateral 34. Moreno E, Basler K. dMyc transforms cells into super-competitors. Cell. compartment of the cochlea requires a temporally restricted FGF20 2004;117:117–129. signal. PLoS Biol. 2012;10:e1001231. 35. Birket MJ, Ribeiro MC, Verkerk AO, et al. Expansion and patterning of 43. Cullen-McEwen LA, Armitage JA, Nyengaard JR, et al. A design-based cardiovascular progenitors derived from human pluripotent stem cells. method for estimating glomerular number in the developing kidney. Am Nat Biotechnol. 2015;33:970–979. J Physiol Renal Physiol. 2011;300:F1448–F1453.

10 Kidney International (2017) -, -–-