Dysregulation of WTI (−KTS) is Associated with the Kidney-Specific Effects of the LMX1B R246Q Mutation

Mutations in the LIM homeobox transcription factor 1-beta (LMX1B) are a cause of nail patellar syndrome, a condition characterized by skeletal changes, glaucoma and focal segmental glomerulosclerosis. Recently, a missense mutation (R246Q) in LMX1B was reported as a cause of glomerular pathologies without extra-renal manifestations, otherwise known as nail patella-like renal disease (NPLRD). We have identified two additional NPLRD families with the R246Q mutation, though the mechanisms by which LMX1BR246Q causes a renal-specific phenotype is unknown. In this study, using human podocyte cell lines overexpressing either myc-LMX1BWT or myc-LMX1BR246Q, we observed dominant negative and haploinsufficiency effects of the mutation on the expression of podocyte genes such as NPHS1, GLEPP1, and WT1. Specifically, we observed a novel LMX1BR246Q-mediated downregulation of WT1(−KTS) isoforms in podocytes. In conclusion, we have shown that the renal-specific phenotype associated with the LMX1BR246Q mutation may be due to a dominant negative effect on WT1(−KTS) isoforms that may cause a disruption of the WT1 (−KTS):(+KTS) isoform ratio and a decrease in the expression of podocyte genes. Full delineation of the LMX1B gene regulon is needed to define its role in maintenance of glomerular filtration barrier integrity.

Steroid resistant nephrotic syndrome (SRNS) is a condition characterized by defects in the glomerular filtration barrier (GFB) which leads to proteinuria, hypoalbuminemia, edema and an inability to achieve clinical remission following standard corticosteroid therapy. SRNS is associated with a number of renal pathologic changes, the most common being focal segmental glomerulosclerosis (FSGS). Globally, SRNS is a major cause of end stage kidney disease (ESKD) requiring dialysis and kidney transplantation. The pathogenesis of SRNS is still being unraveled, however evidence from studies of the familial form of the disease points to defects in glomerular podocytes as central to disease pathogenesis. Familial forms of SRNS are genetically heterogeneous and are associated with mutations in over 25 genes 1 . One of the genes previously associated with familial SRNS is LIM homeobox transcription factor 1-beta (LMX1B) [2][3][4][5] .
LMX1B is a transcription factor in the LIM-homeodomain family of proteins. The LIM-homeodomain proteins are vital for the normal development of dorsal limb structures, components of the GFB, and the anterior segment of the eyes. Mutations in LMX1B are a cause of nail patellar syndrome (NPS) (OMIM 161200), a condition that is characterized by skeletal dysplastic changes, glaucoma and FSGS with type III collagen fibrils in the glomerular basement membrane 2 . It has previously been observed that patients with NPS caused by a LMX1B mutation located in the homeodomain showed significantly more frequent and more severe renal involvement compared with patients with mutations outside of this domain 3 . More recently, a renal specific missense mutation (R246Q) in the homeobox domain of LMX1B (LMX1B R246Q ) has been reported by different groups in France, United Kingdom and Japan [6][7][8][9] . In each of these families, all of the affected individuals presented with glomerular pathologies without skeletal or other extra-renal manifestations. However, the full spectrum of renal phenotypes associated with the LMX1B R246Q mutation is unknown and the reason why the mutation causes kidney-limited disease has not been fully elucidated. As part of our effort to fully characterize genes that are important for maintaining the functional integrity of the GFB, we carried out genome-wide linkage studies (GWLS) and whole-exome sequencing (WES) in families with familial SRNS. We identified two families with the renal-limited LMX1B R246Q mutations with widely variable renal phenotypes. We uncovered patterns of expression that suggest differential effects of the mutation on a panel of podocyte-expressed genes by comparing immortalized human podocyte control cells to immortalized lines stably overexpressing either wild-type LMX1B (LMX1B WT ) or LMX1B R246Q . We observed reduced transcription of WT1, TRPC6, NPHS1, and GLEPP1 in LMX1B R246Q -expressing cells in a pattern suggesting a dominant negative effect by the R246Q mutation on WT1 gene expression and a haploinsufficiency effect on TRPC6, NPHS1, and GLEPP1 gene expression. Additionally, we discovered a novel LMX1B R246Q -mediated downregulation of WT1(− KTS) isoforms in podocytes which may result in an imbalance of the WT1(−KTS) isoforms and subsequently this renal-limited phenotype. This study expands our knowledge of the phenotypes associated with the LMX1B R246Q mutation and provides further insight into the effects of the mutation on the transcriptional regulation of key podocyte genes.

Results
Families and linkage analysis. We identified six families with familial SRNS with segregation patterns consistent with autosomal dominant (AD) inheritance. The majority of the affected individuals were diagnosed with FSGS on renal biopsy, and others with findings such as membranous nephropathy, mesangial proliferative glomerulonephritis and minimal change disease. We performed genome wide linkage studies (GWLS) on the six families, the analysis did not yield conclusive evidence of linkage (logarithm of the odds (LOD) score > 3) for any individual family. Two families (DUK35705 and DUK34319) however had suggestive logarithm of the odds (LOD) scores of 1.7 and 1.4 on chromosome 9q respectively (Supplementary Figure 1a,b). The pedigrees of these two families are shown in Fig. 1a,b. To identify the smallest candidate region, we performed combined GWLS analysis on the two families and identify a minimal candidate region on chromosome 9q region that spans a physical distance of 58.4 Mb between rs966097 and rs3811133 (Supplementary Figure 2a). Table 1. Most of the affected individuals presented with early onset kidney disease with a median age of 12 (range: [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] years. Three of the eight affected individuals in Family DUK35705 presented with early onset FSGS while kidney biopsy in the affected individuals in Family DUK34319 showed a wide variety of morphological changes including C1q nephropathy, immune complex glomerulonephritis, and focal global glomerulosclerosis among others (Table 1). In addition to renal diseases, two individuals in DUK35705 have a history of glaucoma and autism spectrum disorder (Table 1). Five out of the thirteen affected individuals in the two families developed ESKD before the age of 25 years. None of the affected individuals who received kidney transplantation had recurrence of the primary kidney disease after transplantation (Table 1).

Phenotypes of the two families with linkage to Chromosome 9q locus. A summary of clinical findings in affected individuals is shown in
Whole exome sequencing (WES) and targeted sequencing. The DNA from three affected individuals in Family DUK35705 were subjected to whole-exome sequencing (WES) using the Illumina TruSeq platform. After excluding variants with minor allele frequencies > 1% in the dbGaP database, 68 shared variants remained between the three individuals, of which one variant is the missense variant c.737G > A p.R246Q in LIM homeobox transcription factor 1-β (LMX1B) gene (Supplementary Figure 2b)  sequenced exons 4 to 6 of LMX1B that encode for the DNA binding domain of LMX1B protein in 130 additional families with non-syndromic FSGS, we did not find additional families with mutations in LMX1B, thus mutations in exons 4 to 6 of LMX1B are responsible for 1-2% of AD FSGS in our cohort. We did not find any other segregating variants in the chromosome 9q region (Supplementary Figure 2a).

Modifier genes.
Since it has been suggested that variants in other genes may be responsible for the variability in renal phenotype associated with LMX1B mutations 5 , we searched for presumed pathogenic rare variants (MAF < 1%) in all known FSGS genes, COL4A3, COL4A4 and COL4A5 in the exomes of three individuals in 35705. We found a rare variant G545A in the COL4A4 gene in one individual in Family DUK35705. Direct sequencing of all other affected individuals in this family showed that only one individual carried this variant. This individual was reported as having membranous nephropathy on light microscopy. There was no obvious difference in the phenotype of this individual and the other members of the family, therefore it appears that modifier genes play a limited (if any) role in these families.
Phenotypes associated with kidney specific LMX1B mutations. We conducted an extensive literature search to identify studies describing the renal limited phenotype associated with LMX1B mutations [6][7][8][9] . We identified four studies reporting three mutations in the LMX1B DNA binding homeodomains, these studies reported on six families with LMX1B mutations. Four out of the six families had the R246Q mutation while the other two families carried R246P and R249Q mutations. These four studies reported a diversity of renal biopsy findings including minimal change disease (MCD), FSGS, mesangial proliferative glomerulonephritis, C1q nephropathy among others (Table 2).

R246Q mutation and the expression of key glomerular filtration barrier (GFB) genes.
To characterize the functional effects of the LMX1B R246Q mutation, we examined the changes in LMX1B transcriptional regulation induced by the R246Q mutation. To accomplish this, we generated conditionally immortalized human podocyte cell lines stably overexpressing either myc-LMX1B WT or myc-LMX1B R246Q via lentiviral infection. Total RNA was collected from both LMX1B-overexpressing lines and from differentiated controls (i.e. uninfected podocytes and tGFP-expressing podocytes). The infection was verified by immunoblot analysis for myc-tag expression and quantification of LMX1B mRNA levels in LMX1B-overexpressing podocytes. The levels of LMX1B overexpression were similar between the myc-LMX1B WT and myc-LMX1B R246Q infected cells ( Fig. 2a and Supplementary Figure 3). We did not detect any obvious differences in the morphology of the F-actin cytoskeleton in the myc-LMX1B WT and myc-LMX1B R246Q podocytes by immunofluorescence examination. In addition, subcellular localization of Nephrin, CD2AP, GLEPP1 and WT1 were similar between the two cell lines (Supplementary Figure 4).
We evaluated for alterations in the transcription of several key podocyte genes in the myc-LMX1B WT and myc-LMX1B R246Q -overexpressing podocytes by quantitative Real Time PCR (qPCR). NPHS1, TRPC6, and GLEPP1 gene expression was upregulated in both myc-LMX1B WT and myc-LMX1B R246Q -expressing podocytes. However, the expression of NPHS1, TRPC6 and GLEPP1 was significantly reduced in myc-LMX1B R246Q -expressing podocytes compared to myc-LMX1B WT -expressing podocytes, suggesting that the mutation may exert a haploinsufficiency effect on the transcriptional regulation of these genes. (Fig. 2b and Supplementary Figure 5). Conversely, no significant changes in gene expression were observed in INF2, CD2AP, SYNPO or COL4A4 between myc-LMX1B WT and myc-LMX1B R246Q -overeexpressing podocytes. (Fig. 2b and Supplementary Figure 5). The reduction in NPHS1 expression and lack of change in CD2AP, SYNPO and CD2AP was confirmed at the protein level by immunoblot analysis (Fig. 2c). There was no difference in expression of GLEPP1 between the wild type and the mutant cell line (Fig. 2c).
R246Q mutation and the expression of WT1 gene. A different pattern of gene expression was uncovered for the regulation of the Wilms' Tumor 1 gene (WT1). In myc-LMX1B WT -expressing podocytes, WT1 was modestly increased compared to the control cell line (RQ of 1.26 vs 1.00, p = 0.0032) (WT1 probe 1 Fig. 3). Conversely, in myc-LMX1B R246Q -expressing podocytes, WT1 was significantly decreased relative to wild-type (RQ of 0.53 vs 1.26, p = 0.0002) and controls (RQ of 1.00 vs 0.53, p < 0.00001) (WT1 probe 1 Fig. 3) suggesting a dominant negative effect of the mutation on the regulation of WT1 gene expression. This reduction of WT1 expression in myc-LMX1B R246Q -expressing podocytes compared to myc-LMX1B WT -expressing podocytes was confirmed at the protein level by immunoblot analysis (Fig. 3).
In a previous study, we identified synaptopodin (SYNPO) as a target gene of WT1D, a + KTS isoform of WT1 10 . Despite the apparent dominant negative effect of LMX1B R246Q on WT1 gene expression, there was no effect on SYNPO mRNA expression. This finding suggests that LMX1B may selectively modulate WT1-mediated gene transcription via the isoform-specific regulation of WT1 expression in podocytes. To test this hypothesis, we performed qPCR using an assay designed to detect only the + KTS isoforms (WT1 probe 2 Fig. 3). The expression of + KTS isoforms was increased in myc-LMX1B WT and myc-LMX1B R246Q expressing podocytes relative to controls and did not show the dramatic LMX1B R246Q induced reduction that was observed in the initial WT1 screen. (WT1 probe 1 Fig. 3). Taken together, these findings suggest that LMX1B regulates WT1 expression in an isoform-specific manner to selectively influence the expression profile of key podocyte genes (Fig. 4).

Discussion
Advances in genome science and the availability of powerful next-generation sequencing tools for genomic research have accelerated the pace of gene discovery in hereditary nephrotic syndromes. To date more than 25 single genes have been linked to the development of SRNS 1 . Their discovery not only improves our understanding of disease mechanisms, but also provides data allowing better classification and treatment of diseases 11 . Mutations in LMX1B was previously identified as a cause of NPS. More recently, missense mutations in the homeodomain of LMX1B have been shown to cause a kidney-limited phenotype without extra renal manifestations. So far, twenty-seven cases from six families have been reported with kidney-limited phenotypes due to mutations in  exons of LMX1B that encode for the homeodomain of the protein. In the present study, we report two additional families with 18 affected individuals. The predominant LMX1B mutation in the present report and the six families in previous reports is the R246Q mutation (6 of 8 families). Families with this mutation are predominantly of European descent, suggesting that this may be a founder mutation in this population. In our cohort of patients with non-syndromic AD SRNS, mutations in the LMX1B homeodomain are responsible for up to 2% of cases, similar to the report of Boyer et al. 6 .
The phenotypes associated with mutations of the homeodomain have been described using varied terminology including the term nail patella-like renal disease (NPLRD) 8 . NPLRD is associated with classic transmission electron microscopy findings of diffuse irregular thickening of the glomerular basement membrane with patchy electron lucent areas and irregular deposition of type III collagen fibrils. However, because light microscopy is a more widely available modality for examining renal biopsy especially in resource poor countries, it is important to report the spectrum of light microscopy findings associated with these mutations. In this study different light microscopic findings, including FSGS, MCD, membranous nephropathy and immune complex glomerulonephritis, were described in our patients. Reasons for this observation are not known but it may be due to factors such as the timing of kidney biopsy and/or the presence of modifier genes affecting phenotypes. In line with this speculation, we searched for rare variants in key podocyte and GBM genes that have been associated with SRNS. We identified a single rare variant in COL4A4 (G545A) in only one individual with no significant phenotypic variability from other affected family members without the variant. Further studies in a larger cohort of patients are needed to further elucidate the role of modifier genes.
The mechanisms by which mutations in the homeodomain region of LMX1B induce renal specific phenotypes have not been fully investigated. Using luciferase assay, Isojima et al. suggested that the mutation resulted in LMX1B haploinsufficiency 8 . However, a comprehensive survey of the LMX1B gene regulon in podocytes has not been reported. In this study we generated human podocyte cell lines stably expressing myc-LMX1B WT and myc-LMX1B R246Q to evaluate the functional effects of the R246Q mutation on LMX1B-mediated transcription on collagen alpha 4 and a panel of key podocyte genes. We showed a reduced expression of NPHS1, TRPC6, and GLEPP1 in immortalized podocyte cell line expressing the myc-LMX1B R246Q compared to the myc-LMX1B WT , suggesting that the mutation may exert a haploinsufficiency effect on the expression of these genes. We were able to confirm this LMX1BR246Q induced reduction in NPHS1 and WT1 expression at the protein level by immunoblot however, there was no change observed in GLEPP1 protein expression. The reason for the disparity in GLEPP1 expression is unclear, but may be due to a prolonged half-life of the GLEPP1 protein in podocytes. Conversely, we did not identify any significant differences in RNA or protein expression between the myc-LMX-1B WT and myc-LMX1B R246Q overexpressing cell lines in other podocyte genes such as CD2AP, SYNPO, and INF2. Our expression findings are consistent with the observations by Isojima et al. 8 . Notably, the pattern of WT1 gene regulation is different from all the other genes examined. Specifically, the LMX1B R246Q mutation exerts a haploinsufficiency effect on TRPC6, NPHS1, and GLEPP1 expression while for WT1(− KTS) isoforms, the LMX1B R246Q mutation exerts a clear dominant negative effect.
The alternate splicing of leucine, threonine and serine (KTS) amino acids in a critical area between the third and fourth zinc finger domains of WT1 is a highly evolutionarily conserved splicing event [12][13][14] . However, this archetypal program of translational processing may be insufficient to explain the transcriptional effects of the LMX1B R246Q mutation on WT1 expression. A possible explanation for LMX1B-mediated isoform-specific WT1 gene regulation is the use of alternative promoters. These cis-regulatory elements facilitate diversification of gene expression through influences on the efficiency of transcription initiation, pre-mRNA processing, transcript composition, and mRNA turnover 15 . Additionally, these promoter elements have been shown to influence tissue-specific gene expression and developmental gene activity 16,17 . It is estimated that 53% of human genes contain alternative promoters and Pal et al. have demonstrated that alternative transcription exceeds alternative splicing in generating transcriptome diversity in the developing and adult cerebellum 18,19 . Although intragenic alternative promoters have been identified within the WT1 gene, the role of LMX1B in the activation of these regulatory elements is unknown 12 .
Previous studies have shown distinct functional roles and transcriptional targets for WT1(+ KTS) and WT1(− KTS) isoforms [20][21][22][23][24] . The WT1(− KTS) isoforms have been shown to act as traditional regulators of transcription for genes such as NPHS1, NPHS2, and SYNPO 23,[25][26][27] . Alternatively, the WT1(+ KTS) isoforms display reduced DNA binding potential and have been shown to be required for transcription of other podocyte genes such as SYNPO, MAGI2 and SCRIB and may also participate in RNA processing 10,[21][22][23][24][25][26][27] . It is clear that the ratio of WT1(− KTS) and WT1(+ KTS) isoforms is crucial for proper kidney development and function, as a decrease in the + KTS isoform is believed to be responsible for the development of FSGS in Frasier syndrome 28,29 . Frasier syndrome is characterized by male pseudohermaphroditism and progressive glomerulopathy due to a mutation in intron 9 of the WT1 gene that results in reduced WT1 (+ KTS) splicing events. In mouse models of KTS isoform specific knockdown, elimination of either isoform leads to a severe disruption of glomerular development and podocyte differentiation 23,24 . These findings suggest that an altered WT1 isoform ratio may be one of the mechanisms by which the LMX1B R246Q mutation produces a kidney-specific effect in NPLRD patients (Fig. 4). Additional studies will be required to further examine the potential alterations in KTS isoform ratio and to unravel which of these podocyte gene expression changes are due to direct alterations in LMX1B transactivation potential versus alterations in other transcriptional regulators such as WT1.
Contrary to previous reports, we found some individuals with extra renal pathology such as autism and glaucoma in one of the two families. Glaucoma is one of the manifestation of classical NPS, but we are not aware of the report of an association of autism with LMX1B mutation in NPS. An association between autism and common variants in LMX1B has been reported in a small cohort of patients 30 . It is therefore likely that the finding of autism in our patients may represent effects of other variants in LMX1B and other modifier genes. Alternatively, it is Scientific RepoRts | 7:39933 | DOI: 10.1038/srep39933 possible that the renal-specific mutations may dysregulate the transcription of genes that are important during the development of the podocytes and basement membranes in the kidney, eyes and brain.
In conclusion, we reported two additional families with renal specific mutations in LMX1B and showed wide variability in renal pathologic findings within family and between families. Based on these series of experiments, we proposed that the constellation of different glomerular changes in a family with hereditary kidney disease may be a pointer to a diagnosis of kidney specific LMX1B mutation. We provide evidence that the renal specific phenotype associated with this mutation may be due to dominant negative effect on WT1(− KTS) specific isoforms causing a disruption of the WT1 isoform ratio and subsequent decrease in key podocyte related genes. These findings suggest that a comprehensive delineation of the LMX1B regulon in podocytes may uncover novel transcriptional regulatory patterns and improve our understanding of podocyte LMX1B transcriptional activity in health and disease.

Materials and Methods
Institutional review board approval and participants enrollment. Institutional Review Board approval was obtained from the Duke University Medical Center (Durham, NC, USA) and informed consent was obtained from all subjects. All methods were performed in accordance with the relevant guidelines and regulations. Inclusion criteria and determination of affection status are as previously reported 10,31,32 . Linkage analysis. A genome-wide linkage scan was performed using the informative markers in the Illumina HumanOmniExpress-24 genotyping beadchip assay (Illumina Inc., San Diego, California). The list of the informative markers is available on request. Genotyping was performed on 60 individuals from six families comprising of 36 affected individuals and 24 unaffected individuals (unaffected and unknown status). Two-point and multipoint LOD scores were calculated for all the informative SNP markers. A rare dominant model was assumed and a conservative "affecteds-only" analysis was performed.
Whole-exome sequencing. Whole exome sequencing was performed on three affected individuals in family Duke 35705 using standard protocol. We used the Agilent All Exon 50MB kit and sequenced to at least 75x coverage using one lane of a HiSeq 2000 sequencer. Reads were aligned to the Human Reference genome (HG 18) using the BWA software 33 . Single nucleotide variants were called usingSAMtools 34 . The variants were annotated to Ensembl 50_36 l using SequenceVariantAnalyzer (SVA) and were analyzed using the ATAV software 33 .
Sanger sequencing. Potential disease causing variants identified by whole-exome sequencing were confirmed by Sanger sequencing. All sequences were analyzed with the Sequencher software (Gene Codes Corp Ann Arbor, MI).
Conditionally Immortalized Human Podocyte Culture. Conditionally immortalized human podocytes were cultured as described previously 10 . mRNA Extraction and Quantitative Real Time PCR. Total RNA was manually extracted from differentiated podocyte cells using an RNeasy Mini kit (Qiagen, Valencia, CA). Subsequently, 0.5 μg of total RNA was reverse transcribed into cDNA utilizing the RT system (Promega Corporation, Madison, WI) with oligo(dT) primers, according to the manufacturer's protocol. Quantification of mRNA by real-time PCR was performed using the ABI VIAA7 system (Applied Biosystems, Foster City, CA). PCR reactions were performed in a final volume of 10 μl, consisting of 2 μl cDNA, 2.5 μl RNAse and DNAse free water, 0.5 μl of 20× TaqMan Gene Expression Assays, and 5 μl of TaqMan 2× PCR Master Mix (both Applied Biosystems). Probe sequences are listed in Supplementary Table 3. Each individual experiment was performed at least three times, in triplicate or quadruplicate. Relative expression of the target genes was analyzed by normalizing to the housekeeping gene β-Actin. Lentiviral Constructs and Infection. Standard molecular cloning methods were used to replace the ubiquitin-EGFP of FUGW 19 with CMV-turboGFP, CMF-myc-hLMX1B WT and CMV-LMX1B R246Q . Lentivirus was made and conditionally immortalized human podocytes were transduced as described previously 32 . Infected and uninfected controls were treated with 2 ug/ml puromycin to select for cells expressing the transgene construct.
Immunofluorescence. Conditionally immortalized human podocytes were differentiated per established protocols on collagen I-coated coverslips (BD Biosciences) and processed as described previously 10 . Cells were incubated with mouse monoclonal WT1 antibody (Santa Cruz), mouse monoclonal nephrin antibody (Santa Cruz), or polyclonal rabbit CD2AP (Abcam) overnight at 4 °C. Cells were then washed with ice-cold PBS and secondary Alexa Flora 488 antibody was applied (Invitrogen) at a concentration of 1:1000 for 1 hour at room temperature. Cells were then washed four times with PBS at room temperature before addition of phalloidin alexa-568 (Invitrogen) for one hour at RT followed by two PBS washes and then 4′ ,6-diamidino-2-phenylindole