Disruption of Gen1 Causes Congenital Anomalies of the Kidney and Urinary Tract in Mice.

Congenital anomalies of the kidney and urinary tract (CAKUT) are among the most common developmental defects in humans. Despite of several known CAKUT-related loci (HNF1B, PAX2, EYA1, etc.), the genetic etiology of CAKUT remains to be elucidated for most patients. In this study, we report that disruption of the Holliday Junction resolvase gene Gen1 leads to renal agenesis, duplex kidney, hydronephrosis, and vesicoureteral reflux (VUR) in mice. GEN1 interacts with SIX1 and enhances the transcriptional activity of SIX1/EYA1, a key regulatory complex of the GDNF morphogen. Gen1 mutation impairs Grem1 and Gdnf expression, resulting in excessive ureteric bud formation and defective ureteric bud branching during early kidney development. These results revealed an unidentified role of GEN1 in kidney development and suggested its contribution to CAKUT.

Introduction

Congenital anomalies of the kidney and urinary tract (CAKUT) occur in three to six per 1000 live births and serve as the leading cause of renal failure in infants and children 1-3. CAKUT cover a wide range of structural malformations of the kidney and/or the urinary tract, such as renal agenesis, duplex kidneys or ureters, vesicoureteric reflux, pelviureteric junction obstruction, and ureterovesical junction obstruction. Monogenic causes of CAKUT are supported by familial clustering as well as mouse genetic models 4. To date, more than twenty disease-causing genes have been identified for human CAKUT and mutations in 12 dominant CAKUT genes are reported to be responsible for over 6% of families with isolated CAKUT 5. However, most CAKUT patients still lack any molecular diagnosis 5-7.

CAKUT result from abnormal nephrogenesis. In mammals, kidney and urinary tract development is initiated upon the specification of metanephric mesenchyme (MM) in the nephrogenic cord (NC), followed by reciprocal interactions between MM and the adjacent nephric duct (ND) 8. A single ureteric bud (UB) is firstly induced on ND by signals from MM, then invades into and branches in MM to form the future collecting duct system. At the same time, MM cells are specialized by signals from UB tips to form future nephrons and fuse with the collecting duct 9. The formation and initial branching of UBs are critical for the development of the kidney and urinary tract 9. Blockage of either process could result in solitary kidney or renal agenesis, while extra UB formation usually leads to duplex kidney or ureters 3, 9.

During the early stages of kidney development, the secreted growth factor GDNF is expressed by MM to interact with its cognate receptor RET/GFRα-1 on ND to induce UB formation and branching 10, 11. GDNF expression is positively regulated by a group of transcription factors such as SIX1, SIX2, SIX4, PAX2, and EYA1. Disruption of any of these genes would block kidney development in mice 12-15. Meanwhile, GDNF expression are restricted by SLIT2/ROBO2 pathway that regulates the separation between ND and NC, such that supernumerary kidneys were observed in Robo2 mutants 16, 17. GDNF signal is transduced by the negative regulator Spry1, which acts downstream of Ret 18, 19. In addition to GDNF, Spry1 also modulates FGF signals. Deletion of Fgf10, which is expressed during the period of UB formation, caused less severe renal agenesis than Gdnf or Ret mutations. As expected, the renal phenotype in Fgf10 mutants could be rescued by heterozygous Spry1 deletions 19. BMPs provide another type of signals to UB development. In particular, BMP4 is expressed in stromal cells enveloping ND to inhibit UB branching but promote ureteric stalk elongation 20. BMP4 activity is locally restricted by GREMLIN1 (GREM1), a DAN domain protein that preferentially binds to BMP2/4 21. GREM1-mediated reduction of BMP4 activity is essential to initiate UB outgrowth 22, 23. Loss of Grem1 causes arrested kidney development prior to UB invasion into MM 22.

The Rad2/XPG family member GEN1 was initially identified as a Holliday junction resolvase in mammalian cells 24. It has been shown that GEN1 is involved not only in DNA homologous recombination, but also in maintaining centrosome integrity 24-26. To explore the physiological role of Gen1, we analyzed mice carrying an insertion mutation in this gene. Here we report that disruption of Gen1 affects the expression of both Gdnf and Grem1 in mice, and caused multiple congenital anomalies of the kidney and urinary tract. These results revealed an unidentified role of GEN1 in kidney development and suggested its contribution to CAKUT.

Results

Gen1 mutants display multiple congenital defects in the kidney and urinary tract

We identified a Gen1 mutation (Gen1) in a large-scale mutagenesis project in mice 27, 28. This allele carries a piggyBac (PB) transposon insertion in the second intron (Fig. 1A). Real-time RT-PCR detected severely compromised gene expression in the mutant mice. Gen1 transcription in homozygous (Gen1) and heterozygous (Gen1) embryos was reduced to 11.6% and 62.6% of that in the wild type, respectively (Fig. 1B).

Figure 1

Kidney and urinary tract defects in (A) Genomic structure of the Gen1 allele. A PB[Act-RFP] transposon was inserted in the second intron (solid lines) of Gen1. Open boxes, exons; black arrow, transcription direction; green arrows, PB termini; red arrow, RFP expression cassette. (B) Real-time PCR analysis of Gen1 expression in E12.5 embryos. Data is shown as the mean ± s.e.m. *** P < 0.001 determined by t test. (C) Percentages of newborn mice with normal number of kidneys (NK), unilateral renal agenesis (URA), or duplex kidneys (DK). Percentages of mice having accompanying hydronephrosis (HN) are indicated at the top of each column. (D-G) Representative images of the urinary system in newborn mice. URA (E), unilateral DK (F,G), severe HN and hydroureter (G) are presented. (H-O) Hematoxylin-eosin (H&E) staining of the paraffin sections of kidneys in newborn mice. A wild type kidney (H), URA (I), DK (J), and severe HN (K) are presented. Details within the boxes are enlarged in (L-O), respectively. DK is defined by double renal pelvises (arrowheads). Well differentiated glomeruli (arrows) could be observed in the wild type, URA, and DK. In contrast, tubular elements and glomeruli are almost missing in severe HN (O). Scale bars, 2 mm in D-G; 250 μm in H-K; 50 μm in L-O.

Gen1 mice are viable and fertile, but frequently developed kidney and urinary tract defects (Fig. 1C-O). We detected thirty-one abnormal individuals from sixty newborn pups by ultrasonic imaging. Varied and sometimes overlapped morphological defects were further revealed by following anatomical and histological analysis (Fig. 1C). We found unilateral duplex kidneys in fifteen mutants (Fig. 1F, J, N), bilateral duplex kidneys in one mutant, unilateral renal agenesis in eleven mutants (Fig. 1E, I, M), and unilateral hydronephrosis in eight mutants (Fig. 1G, K, O). The duplex kidneys in Gen1 mice are bi-papillary (Fig. 1J). Tubular elements and glomeruli were readily detected in solitary or duplex kidneys, but not in severe hydronephrotic kidneys. In addition to these morphological changes, we also observed two cases of unilateral vesicoureteric reflux (VUR) in nine Gen1 mice at weaning stages (Supplementary Fig. 1A and B). Gen1 mice have similar but weaker malformations (Fig. 1C). We only detected eleven unilateral duplex kidneys from forty-nine newborn Gen1 pups. No renal agenesis or hydronephrosis were observed. We also checked the renal function by measuring blood urea nitrogen (BUN) and creatinine concentration in the serum of four-month Gen1 mice. No significant differences were observed for both parameters (Supplementary Fig. 1C and D).

Other than the kidney and urinary tract defects, Gen1 mice have kinky tails as a result of delayed neural tube closure (Supplementary Fig. 2). All congenital defects were caused by the PB insertion. Generating revertant animals by removing PB in Gen1 could result in normal animals with no kidney and urinary tract defects or delayed neural tube closure (Supplementary Fig. 3).

Excessive UB formation accompanied by altered Grem1 and Gdnf expression in Gen1 mutants

The congenital kidney and urinary tract malformations observed in Gen1 mutants were reminiscent of the defects in CAKUT patients. To identify the developmental abnormalities that may cause the kidney and urinary tract malformations, we followed the formation of ureteric buds (UB), a key event that leads to the development of the renal collecting system and renal vesicles 8. In mice, UB formation starts around embryonic day 10.5 (E10.5) in mice. At E11.0, a single UB is readily observed on each side in wild type embryos (Fig. 2A). We examined eleven Gen1 embryos at this stage (Table 1). Surprisingly, nine of the mutant embryos (81.8%) possessed an ectopic UB on one side (Fig. 2B). The others developed single UBs in a more anterior position to the end of the nephric duct on one or both sides (Fig. 2C). Gen1 mice exhibit less severe defects, as only three of twenty-one (14.3%) Gen1embryos at E11.0 were detected to have an ectopic UB. Thus, disruption of Gen1 leads to excessive UB formation in mice.

Figure 2

(A-C) Ectopic UBs detected in E11 embryos by c-Ret RNA in situ hybridization. (D-F) Gdnf RNA in situ hybridization in E10.5 embryos. Homozygous embryos (E,F) express less Gdnf than the wild type (D). (G) Real-time RT-PCR of Grem1 from E10.5 wild type (n=4) and homozygous mutant (n=6) kidney primordia. Dashed lines indicate UB. Scale bars, 0.2 mm. * P<0.05.

Table 1

Defective UB formation cases in E11 The percentage of the embryos with indicated phenotype is shown in brackets. WT, wild type.

	Total embryos	Normal	Embryos with ectopic buds	Embryos with anterior buds
WT	8	8 (100%)	0	0
PB/+	21	18 (85.7%)	3 (14.3%)	0
PB/PB	11	0	9 (81.8%)	2 (19.2%)

Gen1 is ubiquitously expressed in the mesoderm of E10.5 embryos, including MM and UB regions (Supplementary Fig. 4). To explore if Gen1 mutation alters the signals controlling UB formation, we examined the expression of genes belong to the GDNF, FGF, SLIT, and BMP pathways in E10.5 kidney primordia (Supplementary Table 1) 8, 9. Among them, the BMP4 antagonist Grem1 is the only gene that was significantly upregulated, as revealed by RNA-seq. Real-time RT-PCR of E10.5 kidney primordia confirmed significant Grem1 upregulation in Gen1 embryos (Fig. 2G). In addition to Grem1, RNA-seq revealed significantly lower expression of GDNF pathway genes Eya1, Pax2, Pax8, Sall1, Six2, and Wt1 (Supplementary Table 1). We thus examined the expression of Gdnf in Gen1 mutants by in situ hybridization, despite the fact that Gdnf expression alteration was not recognized by RNA-seq. We detected reduced Gdnf expression in six of eleven (54.5%) homozygous embryos (Fig. 2D-F).

Defective UB branching with decreased GNDF signal in Gen1 mutants

Excessive UB formation usually leads to duplex kidneys. To explore the reason for frequently observed renal agenesis in Gen1 mutants, we further examined the branching process after UB formation. In wild type mice, UB always branches in the metanephric mesenchyme (MM) to form a T-shaped bifurcation at E11.5 (Fig. 3A). In contrast, Gen1 embryos are often affected by UB branching failure at this stage (Table 2). In fact, we observed eleven branching failure events in forty-seven UBs from nineteen E11.5 homozygotes (Fig. 3B). Of eleven embryos without ectopic UBs, three had unilateral branching failures. In seven embryos with a unilateral ectopic UB, we observed three embryos with one unbranched UB and two with both on the side of two UBs, respectively (Fig. 3B). The only embryo with bilateral ectopic UBs also had branching failure in one of the UBs. UB branching defects were weaker in Gen1 mice. In twenty-five E11.5 heterozygous embryos, no branching failure was observed in eighteen embryos without ectopic UB, while one out of seven embryos with a unilateral ectopic UB had one UB failed to branch on that side. These results suggest that UB branching failure as the reason for renal agenesis in Gen1 mutants.

Figure 3

Ureteric branching defects in (A) c-Ret RNA in situ hybridization at E11.5 showed normal branch in wild type embryos. (B) Both UBs at the same side failed to branch in some E11.5 Gen1 embryos. (C) Gdnf RNA in situ hybridization showed Gdnf expression around the UB branches in E11.5 wild type embryos. (D) No Gdnf expression was detected around the un-branched UBs in E11.5 Gen1 embryos. (E) Quantitative RT-PCR showed decreased Gdnf expression in MM of E11.5 Gen1 embryos. Dashed lines indicate UB. Scale bars, 0.2 mm.

Table 2

Defective UB branching cases in E11.5 The percentage of the embryos with indicated phenotype is shown in brackets. *, The only Gen1 embryo with one unbranched UB has two buds on that side.

	Total embryos	Normal	Embryos with one unbranched UB	Embryos with two unbranched UBs
WT	14	14 (100%)	0	0
PB/+	25	24 (96%)	1 (4%)*	0
PB/PB	19	10 (52.6%)	7 (36.8%)	2 (10.5%)

In wild type embryos, presence of GDNF in the condensed mesenchyme surrounding the tips of UB is essential for branching (Fig. 3C) 8. In E11.5 Gen1 embryos, Gdnf expression was present surrounding the tips of all thirty branched UBs but was undetectable in any of the seven unbranched UBs (Fig. 3D). Quantitative RT-PCR also detected approximately 50% drop of Gdnf transcription in Gen1 MM than that in wild type embryos (Fig. 3E). These results suggest that lower Gdnf expression caused by the Gen1 mutation may critically contribute to UB branching defects in the mutants.

GEN1 binds with SIX1 and regulates the transcriptional activity of SIX1/EYA1 complex

We next explored the molecular mechanism through which Gen1 regulates Gdnf during UB formation and branching. As a HJ resolvase, GEN1 was originally identified from nuclear extracts 24. It has been shown that approximately 20% of GEN1 proteins exist in the nuclear fraction during interphase 29. Since several transcription factors have been identified as regulators of Gdnf expression during early kidney development 12-14, 30-34, we hypothesize that GEN1 may interact with the upstream transcription factors to regulate Gdnf expression and tested potential interactions among them by Co-IP. We found that GEN1 could interact with SIX1, but not with its molecular partner EYA1, nor with SIX2, SIX4, PAX2, or FOXC1 in HEK293T cells (Fig. 4A, B and supplementary Fig. 5). Thus, GEN1 may interact with SIX1 to affect Gdnf expression.

Figure 4

GEN1 binds with SIX1 and enhances transcriptional activity of the SIX1/EYA1 complex. (A) HA-tagged SIX1 can be pulled down by FLAG-tagged GEN1. (B) GFP-tagged GEN1 can be pulled down by FLAG-tagged SIX1. (C) Luciferase activity of a SIX1/EYA1 reporter. Supplemented by SIX1, EYA1, and GEN1, the luciferase became more active than under any other environment. * P < 0.05, ** P < 0.01, *** P < 0.001 determined by t test. (D,E) In situ hybridization at E10.5 revealed lower Six2 expression in homozygous (E) than in wild type embryos (D). Dashed lines indicate UB. Full-length blots are presented in Supplementary Figure 6. Scale bar, 0.1 mm.

SIX1 was proposed to form a functional complex with EYA1 in regulating Gdnf during early kidney development 9. We thus examined if GEN1 could alter the transcription activity of SIX1/EYA1 complex. Consensus SIX1 binding sites have been identified in the intron of Gdnf 35. We found the expression of a luciferase reporter driven by a minimal promoter and consensus SIX1 binding sites (Gdnf-luc2) remained at the basal level in HEK293T cells when GEN1 was provided alone. The presence of both SIX1 and EYA1 increased luciferase expression by about six times, while GEN1, SIX1 and EYA1 together enhanced the transcription by approximately twenty folds (Fig. 4C). Metanephric mesenchyme expression of Six2 was also reduced in half (3/6) of the E10.5 Gen1 embryos (Fig. 4D, E). Taken together, these results suggested that GEN1 could enhance Gdnf expression by stimulating the transcriptional activity of the SIX1/EYA1 complex during early kidney development.

Previous studies have identified two conserved glutamic acid residues (E134/E136) required for GEN1 nuclease activity 24. We found the presence of a nuclease-inactive mutant, GEN1-2A (E134A/E136A), enhanced the expression of luciferase to similar levels as the wild type did (Fig. 5A, B). Thus, the resolvase activity of GEN1 is not required for GEN1 to promote SIX1/EYA1-mediated transcription. We also tested truncated GEN1 protein using the same assay. We found that a truncated GEN1 fragment (GEN1-N1, residues 1-585) that contains the potent resolvase fragment 24, 36 is able to enhance the activity of the SIX1/EYA1 complex, while two shorter fragments (GEN1-N2, residues 1-399; GEN1-N3, residues 1-319) could not (Fig. 5C). Co-IP experiment also proved impaired interaction between GEN1-N2/N3 and SIX1 (Fig. 5D). These data suggests that amino acids 400-585 are important for the transcriptional stimulation activity of GEN1.

Figure 5

Impaired interaction between mutant GEN1 and SIX1. (A) Schematic view of mutant GEN1 proteins. (B) Luciferase assay showed potent activity of GEN1-2A in enhancing the transcriptional activity of SIX1/EYA1 complex. (C) Luciferase assay showed impaired activity of truncated GEN1s in enhancing the transcriptional activity of SIX1/EYA1 complex. (D) Co-IP experiments showed impaired binding activity between truncated GEN1s and SIX1. Full-length blots are presented in Supplementary Figure 7. XPG-N, N-terminal nuclease domain; XPG-I, internal nuclease domain; H3TH, helix-hairpin-helix DNA binding domain. FL, full length; 2A, E134A/E136A.

Discussion

Initially purified as a mammalian Holliday junction resolvase, GEN1 was later reported necessary for centrosome integrity in mammalian cells 24-26, 29, 37. Here we found that Gen1 is required for kidney and urinary tract development and neural tube closure in mice. These are the first evidence showing that Gen1 is involved in developmental regulation. During UB formation, GEN1 not only affects the expression of BMP antagonist GREM1, but also stimulates the transcription of UB promoting signal GDNF, which may imply its role as a coordinator of both pathways to ensure normal nephrogenesis.

Intracellular localization was proposed as a critical factor for GEN1 activity. GEN1 was reported to be spatially regulated during cell cycle, such that its resolvase activity is restricted until mitosis when the nuclear membrane breaks down. With the strong nuclear exporting signal, it has been shown that approximately 80% of GEN1 proteins are presented in the cytoplasm 29. We have shown that GEN1 proteins could bind with the transcription factor SIX1 and enhance the transcriptional activity of the SIX1/EYA1 complex. Consistently, Gen1 mutations led to decreased expression of SIX1 downstream genes during kidney development. These results suggest that the apparently minor nuclear fraction of GEN1 may be critical for developmental regulation.

Not all Gen1 mice display kidney and urinary tract abnormalities. This may be due to the fact that Gen1 is a partial loss-of-function allele, such that significant amount of potent GEN1 still exist in the homozygous mutant mice. In fact, incomplete penetrance is not uncommon for mutations that affect kidney development. Heterozygotes of Six1, Eya1, or Foxc1 knock out mutant mice all have varied kidney abnormalities 13, 15, 34. In addition, varied kidney and urinary tract abnormalities in Gen1 mice may reflect the complexity of the regulatory signals during early kidney development.

During UB formation, we have detected down-regulation of Gdnf signals but up-regulation of Grem1. Gdnf is usually considered as a UB promoting signal since complete ablation of Gdnf would lead to failed UB formation in mice 38. However, it needs to be pointed out that UB formation is sensitive to the dosage of Gdnf signals. In Six1 and Sall1 mutant embryos, Gdnf expression was partially decreased, which did not interfere with normal UB formation, but caused failure of the first UB branching 13, 39. On the other hand, GREM1-mediated BMP antagonism is essential to initiate UB formation 23. Grem1 knockout mice exhibited disrupted metanephric development at the stage of UB formation; while additional recombinant GREM1 protein could induce additional budding of the wild type nephric duct 23, 40. In Gen1 mutants, Gdnf expression was partially reduced (Fig. 2E and F), while Grem1 transcription was increased to four folds of wild type embryos (Fig. 2G). It is likely the decreased Gdnf signal did not affect UB formation in E10.5 Gen1embryos, while the extra UB formation was due to the elevated Grem1 expression. Thus, varied Grem1 expression may lead to partially unobserved ectopic UB formation. During UB branching, ectopic UB in a single nephric duct may not always develop into a duplex kidney due to the lack of adequate GDNF signals so that less duplex kidney mutants were observed after birth. Unlike kidney defects, kinky tails were consistently observed in all Gen1 homozygotes. This suggests that a different threshold of GEN1 activity or a different mechanism is required for neural tube development.

The trunk of the initial UB differentiates into ureter, connecting the kidney with the bladder 41. Higher or lower UB formation on the nephric duct causes abnormal positioning of the ureter orifice in the trigone, leading to obstruction or reflux 42. We have observed single UBs at more anterior position in two Gen1 embryos (Fig. 2C). These abnormally positioned ureters may join the bladder at a position more lateral or anterior to the proper site in the trigone. This could shorten the intra-vesicular ureteral segment and disrupt the valve mechanism that prevents the urine backflow to the kidney, thus resulted in VUR. On the other hand, although we didn't observe proximal ureteral atresia or PUJO in Gen1 mutants (Fig.1), we could not exclude the possible contribution of potential ureteral atresia to the severe hydronephrotic phenotype. Similarly, our study did not exclude the possibility that Gen1 mutation may cause delayed kidney development. To distinguish this event, global quantification of tissue dynamics can be applied to monitor the developmental process of nephron number and branching at different stages in Gen1 mutant mice 43.

Our study raises the possibility of GEN1 as a disease gene of human CKAUT. To date, more than 90% of CAKUT patients still lack any molecular diagnosis 5-7. We have detected CAKUT phenotypes in both heterozygous and homozygous Gen1 mutant mice and shown that GEN1 proteins containing the N-terminal 399 amino acids could not function properly in regulating the transcriptional activity of SIX1/EYA1. At the same time, thirty-seven loss-of-function variants have been reported in the Exome Aggregation Consortium (ExAC) database (http://exac.broadinstitute.org), all of which are presented in heterozygosity and predicted to encode truncated GEN1 proteins with less than 400 residues in the N-terminus. Although incomplete penetrance, as implied by the mouse data, may bring complications in human genetic analysis, it is still worth to explore the genetic burden of GEN1 mutations in CAKUT patients.

Methods

Mice

All animal experiments were performed in accordance with protocols approved by the Animal Care and Use Committee of the Institute of Developmental Biology and Molecular Medicine (IDM) at Fudan University. The Gen1 mutant strain (081125049-HLA) was established and maintained on the FVB/N background. In the Gen1 allele, the PB insertion was mapped in the second intron of Gen1 (Chr:12.11268138, Ensembl release 54).

PCR

Genotyping PCR was performed with a PB specific primer LB2 (5'-CTGAGATGTCCTAAATGCACAGCG-3') and two flanking genomic primers 081125049-L1A (5'-TAGTGTGGGCACATGCAAGC-3') and 081125049-R1A (5'-CCCCTTTGCCTGTTCTTAACCTC-3'). Real-time RT PCR was performed with SYBR Green (Agilent) according to the manufacturer's instruction. Gapdh was selected as the internal control for normalization. The primers used for RT-PCR were: Gen1-RT-F1 in exon 2: 5'-GCACAGACAGTGAAGAAAATG-3'; Gen1-RT-B1 in exon 3: 5'-TGCTTATGACATCAGCTTTCAG-3'; PCR product size 151 bp. Grem1-RT-F: 5'-CCTTTCAGTCTTGCTCCTTCTGC-3'; Grem1-RT-R: 5'-TTCTTCTTGGTGGGTGGCTGTAGC-3'; PCR product size 94 bp. Gapdh-RT-F1: 5'-TGTTCCTACCCCCAATGTGTCC-3'; Gaphd-RT-B1: 5'-GGAGTTGCTGTTGAAGTCGCAG-3'. PCR product size 169 bp.

Histology

Kidneys and the urinary tract were dissected and embedded with paraffin according to the standard protocol 31. Hematoxylin-eosin (H&E) staining was then performed on the sections of 7 μm as previously described 31.

RNA-seq

Total RNA from kidney primordia of E10.5 embryos was extracted by TRIzol (Invitrogen) and GlycoBlue (Ambion) coprecipitation. We combined two mice to obtain 0.5 μg of total RNA for either wild type or Gen1 samples. Strand-specific RNA-seq libraries starting with 0.5 μg of total RNA were constructed as previously described 44. Two biological replicates of RNA-seq for each group (wild type or mutants) were included to analyze differentially expressed genes. The RNA-seq libraries were sequenced on the Illumina HiSeq2000 platform (Genergy Biotechnology, Shanghai). FastQC was used to evaluate the quality of high-throughput sequencing data (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc/). Paired-end reads were mapped to mm9 genome using TopHat2 with default parameters except -g -r and -mate-std-dev 45. In specificity, -g was set to 1 to ensure that there is only one alignment for a given read 45. Bowtie was used to get a good approximation of -r and -mate-std-dev, which is the expected inner distance between mate pairs and the standard deviation for the distribution on inner distances between mate pairs, respectively 46. The known transcripts supplied to TopHat2 were based on the Ensemble database and the NONCODE database 47. Cufflinks was used to assemble transcripts and estimates their abundances for each sample and Cuffdiff was used to get the differentially expressed genes 48.

RNA in situ hybridization

The coding sequences of Ret, Gdnf, and Six2 were cloned from E12.5 embryos through RT-PCR. Digoxigenin-labeled anti-sense probes were then produced by the DIG RNA Labeling kit (Sp6/T7) (Roche). Whole-mount in situ hybridization was performed as previously described 49. Frozen sections of 50 μm were then used for signal detection.

Co-immunoprecipitation (Co-IP) and luciferase assay

HEK293T cells were transfected with Lipofectamine 2000 (Invitrogen) for both co-immunoprecipitation and luciferase assay. Co-IP and immunoblot were performed as previously described 50. The antibodies used include mouse anti-HA (Covance), rabbit anti-HA (Abcam), mouse anti-FLAG (Sigma), and rabbit-anti-GFP (Cell signaling). For luciferase assay, each transfection was triplicated with 400 ng wild type or mutant Gen1, 100 ng pCX-EGFP, 200 ng HA-Six1, 200 ng HA-Eya1, and 200 ng Gdnf-luc2. Luciferase activity was scored 48 hours after transfection and calibrated by the percentage of GFP positive cells (determined by FACS). Each experiment was performed at least twice independently.

Statistics

All data are presented as mean ± SEM. The comparison of two groups was performed with unpaired two-tailed t-test. The threshold for significance was set at P < 0.05.

Data Availability

All data generated or analyzed during this study are included in this published article and its Supplementary Information files.

Supplementary figures and tables.

Click here for additional data file.