Transcription Factor Hepatocyte Nuclear Factor-1β (HNF-1β) Regulates MicroRNA-200 Expression through a Long Noncoding RNA*

Background: Transcription factor HNF-1β regulates epithelia-specific gene expression. Results: Mutations of HNF-1β down-regulate the miR-200b/200a/429 miRNA cluster and increase expression of Zeb2 and Pkd1. Conclusion: HNF-1β directly activates the transcription of a lncRNA encoding miR-200b/200a/429. Significance: A novel mechanism whereby mutations of HNF-1β produce kidney cysts and epithelial-mesenchymal transition was identified. The transcription factor hepatocyte nuclear factor-1β (HNF-1β) regulates tissue-specific gene expression in the kidney and other epithelial organs. Mutations of HNF-1β produce kidney cysts, and previous studies have shown that HNF-1β regulates the transcription of cystic disease genes, including Pkd2 and Pkhd1. Here, we combined chromatin immunoprecipitation and next-generation sequencing (ChIP-Seq) with microarray analysis to identify microRNAs (miRNAs) that are directly regulated by HNF-1β in renal epithelial cells. These studies identified members of the epithelial-specific miR-200 family (miR-200b/200a/429) as novel transcriptional targets of HNF-1β. HNF-1β binds to two evolutionarily conserved sites located 28 kb upstream to miR-200b. Luciferase reporter assays showed that the HNF-1β binding sites were located within a promoter that was active in renal epithelial cells. Mutations of the HNF-1β binding sites abolished promoter activity. RT-PCR analysis revealed that a long noncoding RNA (lncRNA) is transcribed from the promoter and encodes the miR-200 cluster. Inhibition of the lncRNA with siRNAs decreased the levels of miR-200 but did not affect expression of the Ttll10 host gene. The expression of the lncRNA and miR-200 was decreased in kidneys from HNF-1β knock-out mice and renal epithelial cells expressing dominant-negative mutant HNF-1β. The expression of miR-200 targets, Zeb2 and Pkd1, was increased in HNF-1β knock-out kidneys and in cells expressing mutant HNF-1β. Overexpression of miR-200 decreased the expression of Zeb2 and Pkd1 in HNF-1β mutant cells. These studies reveal a novel pathway whereby HNF-1β directly contributes to the control of miRNAs that are involved in epithelial-mesenchymal transition and cystic kidney disease.

whereby HNF-1␤ directly contributes to the control of miRNAs that are involved in epithelial-mesenchymal transition and cystic kidney disease.
In humans, mutation of HNF1␤ produces cystic kidney disease and early-onset diabetes mellitus, a syndrome called renal cysts and diabetes (RCAD) (19). Mutations or deletions of HNF1␤ have also been linked to renal hypoplasia, multicystic * This work was supported, in whole or in part, by National Institutes of Health (NIH) Grants R37DK042921 (to P. I.), P30DK079328 (to P. I.), K08DK084311 (to V. P.), NIH Training Grant T32DK007257 (to P. I.). The authors declare that they have no conflicts of interest with the contents of this article. 1 3 The abbreviations used are: HNF-1, hepatocyte nuclear factor-1␤; EMT, epithelial-mesenchymal transition; miRNA, microRNA; pri-miRNA, primary miRNA; Q-PCR, quantitative real-time PCR; lncRNA, long noncoding RNA; ChIP-Seq, chromatin immunoprecipitation and high-throughput sequencing.
renal dysplasia, and glomerulocystic kidney disease. Inactivation of HNF-1␤ or expression of dominant-negative mutant HNF-1␤ in the mouse kidney produces renal cysts. HNF-1␤ controls the transcription of cystic disease genes, including PKHD1, the gene mutated in autosomal recessive polycystic kidney disease (ARPKD), and PKD2, which is mutated in autosomal dominant polycystic kidney disease (13,20,21). Decreased expression of PKHD1, PKD2, and other cystic disease genes may contribute to renal cyst formation in individuals carrying HNF1␤ mutations. In addition, down-regulation of HNF-1␤ has been implicated in epithelial-mesenchymal transition (EMT) and renal fibrosis (22). MicroRNAs (miRNAs) are short, noncoding RNAs that regulate post-transcriptional gene expression. MicroRNAs are encoded by genomic sequences located in the introns of protein-coding host genes (intragenic) or present as independent transcriptional units (intergenic). The biosynthesis of miRNAs initially involves transcription of a primary miRNA (pri-miRNA), which can be several kilobases in length (23). The pri-miRNA is processed in the nucleus by Drosha to form a pre-miRNA, which then enters the cytoplasm where it is processed by Dicer to produce the mature 20 -22-nucleotide miRNA. The mature miRNA associates with members of the Argonaute family to form the miRNA-induced silencing complex, which binds to complementary sequences located primarily within the 3Ј-UTR of target mRNAs. Binding of the miRNAinduced silencing complex to a target mRNA leads to degradation of the transcript, inhibition of translation, and/or sequestration in P bodies (24,25). Because individual miRNAs can target multiple different transcripts in the same pathway, miRNAs can provide coordinated regulation of biological processes.
Aberrant up-regulation or down-regulation of specific miRNAs has been observed in polycystic kidney diseases (26,27). Inactivation of Dicer in the kidney leads to cyst formation, suggesting that miRNAs are essential for normal tubular homeostasis (28). Increased expression of miR-17ϳ92 is observed in multiple genetic models of PKD. Moreover, overexpression of miR-17ϳ92 in renal tubules is sufficient to produce kidney cysts, whereas kidney-specific inactivation of miR-17ϳ92 inhibits cyst growth (26). These findings suggest that miRNAs may play a pathogenic role in PKD. However, the mechanisms that regulate the expression of miRNAs in the kidney are poorly understood. Here, we used genome-wide sequencing to identify miRNAs that are regulated by HNF-1␤ in kidney cells. We found that HNF-1␤ directly regulates the expression of the kidney-enriched miR-200 family through a novel long noncoding RNA.
Cells and Animals-Mouse renal epithelial cells (53A cells) that express dominant-negative mutant HNF-1␤ (DN-HNF-1␤) have been previously described (29). mIMCD3 cells, 53A cells, and HeLa cells were grown in DMEM low glucose medium supplemented with 10% FBS and maintained in 5% CO 2 atmosphere at 37°C. Mice with kidney-specific inactivation of Hnf1␤ have been previously described (13). All experiments involving animals were conducted under the auspices of the UT Southwestern Institutional Animal Care and Use Committee.
Chromatin Immunoprecipitation-ChIP assays were performed using ChIP-IT High Sensitivity kits (Active Motif TM ) or EZ ChIP TM kits (EMD Millipore) according to the manufacturer's protocol. mIMCD3 cells were cross-linked with 1% formaldehyde for 15 min at room temperature. Immunoprecipitation was performed with 5 g of rabbit anti-HNF-1␤ antibody, sc-22840 (Santa Cruz), or rabbit IgG, sc-2027 (Santa Cruz), as a negative control. Purified genomic DNA was amplified using specific primers and visualized on 1% agarose gels. The products were normalized relative to 10% of the input DNA.
ChIP Sequencing-Two control (anti-IgG) and two experimental (anti-HNF-1␤) samples from the immunoprecipitated DNA from mIMCD3 cells were sent for sequencing using a Solexa platform. Quality assessment of the raw sequencing data was done using FASTQC. Reads with Ͼ70% bases with quality lower than Phred score of 20 were removed. Quality score filtered reads were then aligned to the mouse reference genome NCBI37 (mm9) using Burrows-Wheeler aligner. Identification of transcription binding sites was performed using the peak calling algorithm, QuEST. A -fold-change cutoff of three was applied to identify peaks in the experimental ChIP that were significantly above background. The identified peaks were annotated using HOMER. Peaks were mapped to miRNAs if they were located upstream or upstream of an miRNA locus (independent of distance) and no intervening gene was present between the miRNA and the binding site.
MicroRNA Microarrays-53A cells were incubated for 48 h with mifepristone to induce the expression of dominant-negative mutant HNF-1␤ or vehicle as a control. Total RNA was isolated with miRNeasy Mini kits, and miRNA microarray analysis was performed by LC Sciences (Houston, TX) as previously described (28).
Plasmids-Mouse genomic DNA was amplified using primers 5Ј-TAGAATACGCGTACAAAGCGGTTCTTGAC-AGG-3Ј and 5Ј-TAGAATCTCGAGGCTCAGCAGAAAGG-GAGAGA-3Ј, where the underlined sequences are MluI and XhoI sites, respectively. The 2.5-kb PCR product was cloned into pGL3-Basic (Promega) that was previously digested with MluI and XhoI to generate the pGL3-S1S2 plasmid. The HN-F-1␤ sites were mutated to produce the pGL3-mutS1S2 plasmid. The sequences of the primers used to introduce the S1 mutation were 5Ј-CTGTTTTTTCTCAGAGCAGAAGTC-GGCAGCCGGTCAACACAAAATTCCTGGGCT-3Ј and 5Ј-AGCCCAGGAATTTTGTGTTGACCGGCTGCCGAC-TTCTGCTCTGAGAAAAAACAG-3Ј. The sequences of the primers used to introduce the S2 mutation were 5Ј-CTCTTGTGCTTTTCCAAAGGGTCGGCAGCCGCCTT-TCTCTTCTTGCTTGC-3Ј and 5Ј-GCAAGCAAGAAGAGA-AAGGCGGCTGCCGACCCTTTGGAAAAGCACAAGAG-3Ј. The underlined sequences represent the HNF-1␤ binding site, and the italicized sequences denote the mutated bases. Sitedirected mutagenesis was performed as previously described (28). pGL3-FLP S1S2 was constructed by ligating a 2.5-kb mouse genomic PCR product that includes the HNF-1␤ binding sites into pGL3-Basic that was previously digested with MluI and XhoI. The sequences of the PCR primers were 5Ј-TAGAATCTCGAGACAAAGCGGTTCTTGACAGG-3Ј and 5Ј-TAGAATACGCGTGCTCAGCAGAAAGGGAGAGA-3Ј, where the underlined sequences denote the XhoI and MluI sites, respectively.
Luciferase Reporter Assays-53A cells expressing dominantnegative HNF-1␤ or HeLa cells were plated in 6-well plates (2 ϫ 10 5 cells/well) and co-transfected with 0.6 g of the pGL3 plasmids encoding Photinus luciferase and 0.02 g of pRL-CMV encoding Renilla luciferase. Cells were harvested 48 h after transfection and lysed in 250 l of passive lysis buffer. 40 l of the lysate was added to 96-well plates. Photinus and Renilla luciferase activities were measured using the Dual Luciferase Reporter Assay system (Promega) according to the manufacturer's directions.
Overexpression of miR-200 -53A cells expressing dominantnegative mutant HNF-1␤ were stably transfected with plasmids that overexpress mouse miR-200a (catalog #SC400901) and miR-200b (catalog #SC400902) or a control plasmid (catalog #PCMVMIR) using Lipofectamine 2000. Twenty-four hours after transfection, the medium was supplemented with G418 (0.8 g/ml), and cells were allowed to grow for 12-14 days. The surviving cells were expanded, and the expression of the mature miR-200a and miR-200b transcripts was measured to confirm overexpression.
Western Blots-Total protein was extracted from HNF-1␤ knock-out and control kidneys. 20 g of protein was loaded on a 4 -15% SDS-polyacrylamide gel, and the proteins were transferred to a nitrocellulose membrane. The membrane was blocked in 5% powdered milk and probed overnight at 4°C with a primary rabbit anti-Zeb2 antibody (1:1000). Goat-anti rabbit HRP-conjugated IgG was used as a secondary antibody, and the blot was developed using the SuperSignal West Dura Extended Duration substrate from Pierce. The protein bands were quantified using Quantity One imaging software from Bio-Rad.
RNA Isolation and Quantitative RT-PCR-Total RNA was isolated from cultured cells or mouse kidneys using miRNeasy Mini kits (Qiagen). First-strand cDNA was synthesized from mRNA using SuperScriptIII Reverse Transcriptase from Life Technologies. The Universal cDNA Synthesis kit from Exiqon was used for first-strand synthesis from miRNA. Quantitative real-time PCR was performed by loading samples in triplicate on a CFX Connect TM Real-time PCR detection system. 18S and 5S RNA were used to normalize expression of mRNA and miRNA, respectively. Data were analyzed using the Bio-Rad CFX software. For semi-quantitative PCR, products were loaded on a 1% agarose gels and stained with ethidium bromide. The images were visualized and analyzed on a Gel Doc TM XRϩ system (Bio-Rad). The sequences of the PCR primers are shown in Table 1.
Statistical Analysis-Data are shown as the mean Ϯ S.E. or mean Ϯ S.D. as indicated in the figure legends. Statistical analysis was performed using Student's t tests. Analysis of variance followed by Dunnett's post hoc test was used for multiple comparisons.

HNF-1␤ Regulates Expression of miRNAs in Renal Epithelial
Cells-To identify miRNAs that are directly regulated by HNF-1␤, we combined chromatin immunoprecipitation and highthroughput sequencing (ChIP-Seq) with miRNA expression profiling. We reasoned that miRNAs that are located near HNF-1␤ binding sites and show altered expression in cells expressing mutant HNF-1␤ are likely to represent direct targets, whereas miRNAs that change expression but are not located near binding sites may represent indirect targets. Chromatin from mouse inner medullary collecting duct cells (mIMCD3) was cross-linked and immunoprecipitated with an antibody against HNF-1␤. The cross-links were reversed, and the immunoprecipitated DNA was sequenced by next-generation sequencing. Genomic sequences that were enriched by immunoprecipitation with anti-HNF-1␤ antibody compared with control IgG were mapped on the mouse genome to identify HNF-1␤ binding sites. The locations of the binding peaks were compared with the locations of miRNAs in the mouse genome. MicroRNAs were selected for further analysis if no other gene was present between the HNF-1␤-bound genomic region and the miRNA, unless the miRNA was intragenic. Based on this criterion, we identified 85 miRNAs that were potential HNF-1␤ targets. HNF-1␤ binding near representative miRNA targets was verified by chromatin immunoprecipitation (Fig. 1B).
To identify miRNAs whose expression is regulated by HNF-1␤, we utilized renal epithelial cells expressing an HNF-1␤ mutant that lacks the C-terminal transactivation domain. Previous studies showed that this protein functions as a dominantnegative mutant (DN-HNF-1␤) (29). Cells were treated with mifepristone to induce expression of the DN-HNF-1␤ mutant or vehicle as a negative control. miRNA microarray analyses identified 22 miRNAs whose expression increased and 30 miRNAs whose expression decreased in cells expressing DN-HNF-1␤ (Fig. 1C). The changes in expression of representative miRNA targets were verified by quantitative real-time PCR (Q-PCR) (Fig. 1D). Combining the results from ChIP-Seq and expression profiling identified 22 miRNAs that are directly regulated by HNF-1␤, of which 12 were up-regulated and 10 were down-regulated in HNF-1␤ mutant kidney cells (Fig. 1A).

miR-200 Levels Are Reduced in HNF-1␤ Mutant Cells and HNF-1␤
Knock-out Kidneys-Among the 22 miRNAs that were directly regulated by HNF-1␤, three miRNAs (miR-200b, miR-200a, and miR-429) belonged to the miR-200 cluster on mouse chromosome 4 (Fig. 1C, arrows). Like HNF-1␤, the miR-200 cluster has been implicated in PKD and EMT and was, therefore, selected for further study. Q-PCR analysis confirmed that the levels of miR-200b, miR-200a, and miR-429 were reduced 2-fold in cells expressing the DN-HNF-1␤ mutant (Fig. 2B). To verify whether expression changed in vivo, Q-PCR was performed on kidneys from mutant mice in which HNF-1␤ was deleted from renal tubules by Cre/loxP recombination (13). The levels of miR-200b, miR-200a, and miR-429 were significantly reduced in HNF-1␤ mutant kidneys (Fig. 2C). The residual expression likely reflects incomplete deletion of HNF-1␤ and the contribution of other transcription factors. In contrast, the levels of miR-200c and miR-141, which are located on a different chromosome, were unchanged in DN-HNF-1␤-expressing cells and HNF-1␤ mutant kidneys (Fig. 2, B and C). The miR-200 cluster containing miR-200b, miR-200a, and miR-429 is located in intron 1 of a host gene, Ttll10. However, the expression of Ttll10 exon 1 and exon 3 was unchanged in DN-HNF-1␤-expressing cells and HNF-1␤-deficient kidneys (Fig. 2D). These results indicate that expression of mutant HNF-1␤ specifically inhibits the expression of miR-200b, miR-200a, and miR-429 and does not affect the expression of the Ttll10 host gene.

Primer ID Sequence Product
HNF-1␤ Binds to a Region Located 28 kb Upstream of the miR-200 Locus-ChIP-Seq analysis of chromatin from mIMCD3 renal epithelial cells revealed that HNF-1␤ binds to a region located 28 kb upstream to miR-200b (Fig. 3B). This region contains two evolutionary conserved HNF-1␤ binding sites, S1 and S2, which are separated by ϳ1.5 kb (Fig. 3, B and   C). Quantitative ChIP analysis revealed that HNF-1␤ binding is enriched ϳ3-fold at the S1 site and 6-fold at the S2 site (Fig. 3D). ChIP performed using an antibody that recognizes the phosphorylated C-terminal domain of RNA polymerase II showed enrichment at the S1 and S2 sites, indicating that this region may be transcriptionally active (Fig. 3E). Analysis of the mouse

JOURNAL OF BIOLOGICAL CHEMISTRY 24797
ENCODE database indicated that epigenetic marks of active transcription, including H3K4me3, H3K27ac, and H3K4me1, co-localized with the HNF-1␤ binding sites in adult kidney (Fig.  3B). Taken together, these results indicate that the genomic region containing the HNF-1␤ binding sites is bound by RNA polymerase II and is transcriptionally active.
The Ϫ28-kb Site Exhibits HNF-1␤-dependent Promoter Activity-To determine if the HNF-1␤ binding sites at Ϫ28 kb are functional, a 2.5-kb genomic fragment containing the S1 and S2 sites was cloned into a promoterless pGL3-Basic luciferase reporter plasmid (Fig. 4A). The recombinant plasmid was transfected into HeLa cells, which lack endogenous HNF-1␤ activity, together with an HNF-1␤ expression plasmid. HNF-1␤ transactivated the luciferase reporter plasmid containing the Ϫ28 kb sites but had no significant effect on empty pGL3-Basic (Fig. 4C). Mutation of the S1 and S2 binding sites (Fig. 4B) prevented transactivation by HNF-1␤ (Fig. 4C). When the 2.5-kb genomic DNA fragment was cloned into pGL3-Basic in the reverse orientation, HNF-1␤ failed to stimulate luciferase activity. These results indicate that the 2.5-kb genomic fragment contains an orientation-dependent promoter. To confirm these findings, the luciferase reporter plasmids were transfected into renal epithelial cells expressing DN-HNF-1␤. Transfection of the wild-type plasmid into control (uninduced) cells stimulated luciferase activity, and induction of the DN-HNF-1␤ mutant inhibited promoter activity (Fig. 4D). This   (H3K4m1, H3K4m3, H3K27ac) and binding of RNA polymerase II (Pol2) in chromatin from mouse kidney from the ENCODE database (54). Lower panel, alignment with long RNA transcripts in the indicated mouse tissues from ENCODE Cold Spring Harbor Laboratory Long RNA-seq (55). C, the S1 and S2 binding sites contain evolutionarily conserved HNF-1␤ consensus recognition sequences. D, validation of the HNF-1␤ binding at the S1 and S2 sites by chromatin immunoprecipitation. The arrows denote the PCR primers. The histograms indicate the -fold enrichment of HNF-1␤ binding to the S1 and S2 sites compared with IgG control as determined by Q-PCR. E, -fold enrichment of the binding of the active form of RNA polymerase II at the S1 and S2 sites compared with control IgG determined by Q-PCR.

JOURNAL OF BIOLOGICAL CHEMISTRY 24799
inhibition was blocked by mutation of the HNF-1␤ binding sites. Transfection of the reporter plasmid containing the genomic fragment in the reverse orientation did not stimulate luciferase activity compared with empty pGL3-Basic. Taken together, these results indicate that the Ϫ28-kb region contains a functional promoter that is activated by HNF-1␤ binding to the S1 and S2 sites.
miR-200 Cluster Is Encoded by a Long Noncoding Transcript-Because the Ϫ28-kb region functions as a gene promoter, we investigated whether the transcript that is made from this region encodes the miR-200 cluster. PCR primers were designed to amplify the region between the HNF-1␤ binding sites and the miR-200b genomic locus. RT-PCR of kidney cDNA produced overlapping PCR products extending from the HNF-1␤ binding sites to miR-200b (Fig. 5A, upper panel). As a negative control, primers designed to amplify a region 2-kb upstream to the HNF-1␤ binding sites did not produce a PCR product. RT-PCR using a forward primer that is upstream to the Ttll10 promoter and a reverse primer within the miR-200b locus produced a 4.7-kb product in RNA from mIMCD3 cells that was not detected in the absence of reverse transcriptase (Fig. 5A, lower panel). Collectively, these results identify a long noncoding RNA (lncRNA) transcript extending from the Ϫ28-kb site to miR-200b. RT-PCR analysis of various mouse tissues indicated that the expression of the lncRNA transcript was specific to the kidney (Fig. 5B).
To determine if HNF-1␤ regulates the expression of the lncRNA, we designed RT-PCR primers along the length of the transcript and measured the levels in HNF-1␤-deficient kidneys. The levels of the lncRNA were decreased 2-fold, indicating that HNF-1␤ regulates the expression of the lncRNA (Fig.  5C). To determine if the lncRNA encodes the miR-200 cluster, four siRNAs that targeted the lncRNA at locations 25, 14, 10, and 6.4 kb upstream to the miR-200 locus were transfected in mIMCD3 cells. RT-PCR analyses showed that the levels of the lncRNA were reduced by 30 -40% in mIMCD3 cells transfected with the siRNAs (Fig. 5D). Inhibition of the expression of the lncRNA resulted in a corresponding 20 -40% reduction in the levels of miR-200b, miR-200a, and miR-429 (Fig. 5E). In contrast, the levels of miR-200c, which is located on a different chromosome, were unchanged. Similarly, the levels of Ttll10 exon1 and exon3 were unchanged by knockdown of the lncRNA (Fig. 5F). Taken together, these results indicate that the lncRNA encodes the miR-200b/a/429 cluster but does not encode the Ttll10 host gene.
We have previously shown that Pkd1, which is mutated in autosomal dominant polycystic kidney disease, is post-transcriptionally regulated by miR-200 in kidney cells (28). Therefore, we measured the levels of Pkd1 mRNA in HNF-1␤ mutant cells and kidneys. The levels of Pkd1 mRNA were increased in cells expressing DN-HNF1-␤ and in HNF-1␤-deficient kidneys (Fig. 7A). Reexpression of miR-200 in cells expressing FIGURE 4. The HNF-1␤ binding sites are located within a functional promoter. A, schematic diagram of the luciferase reporter plasmids. A 2.5-kb genomic region (orange bar) containing the HNF-1␤ sites was cloned into pGL3-Basic to generate the pGL3-S1S2 plasmid. The red triangles denote the conserved HNF-1␤ binding sites (S1 and S2) located 28 kb upstream to miR-200b. The HNF-1␤ sites were mutated (white triangles) to create the pGL3-mutS1S2 plasmid, and the orientation of the region was reversed in the pGL3-FLP S1S2 plasmid (inverted red triangles). B, the sequences of the wild-type and mutated S1 and S2 sites are shown. DN-HNF-1␤ decreased the levels of Pkd1 (Fig. 7B). These findings indicate that HNF-1␤ controls the expression of Zeb2 and Pkd1 through its regulation of the miR-200 cluster.

Discussion
HNF-1␤ is a tissue-specific transcription factor that regulates the expression of membrane transporters and other markers of epithelial differentiation in the mammalian nephron. HNF-1␤ plays a central role in cystic kidney disease genes through its regulation of PKD genes such as Pkd2 and Pkhd1.
Here, we identify a novel role of HNF-1␤ in the regulation of miRNA expression in the kidney. We utilized ChIP-Seq combined with microarray analysis to identify 22 miRNAs that are directly regulated by HNF-1␤. Among the direct targets, miR-200b, miR-200a, and miR-429 were of particular interest as they constitute a cluster of miRNAs that is associated with EMT and renal cancer (30 -33), both of which have been linked to HNF-1␤ (22,34). Like HNF-1␤, the expression of miR-200b, miR-200a, and miR-429 is epithelia-specific. In the kidney, miR-200 cluster members are expressed in renal tubules where they are involved in tubular maturation (28). The overlapping patterns of expression suggest that the epithelial tissue-specific Studies on the mechanism of HNF-1␤-dependent regulation of the miR-200 cluster unexpectedly revealed that HNF-1␤ does not bind near the miR-200 cluster or the promoter of its Ttll10 host gene. Instead, HNF-1␤ binds to two evolutionarily conserved sites located 28 kb upstream where it regulates the transcription of a lncRNA that extends to the miR-200b/a/429 cluster. The Ϫ28-kb site contains chromatin marks of transcriptional initiation regions of lncRNAs (35) and exhibits HNF-1␤-dependent transcriptional activity. Based on our experimental results and data available from the ENCODE consortium, the lncRNA is at least 28 kb in length and is expressed only in the kidney (Figs. 3B and 5B). This is the first report showing that HNF-1␤ regulates the expression of a lncRNA. Some tissue-specific lncRNAs have been shown to interact with chromatin-modifying complexes and thereby epigenetically regulate genes involved in cell differentiation (36). Further studies will be needed to determine whether the lncRNA regulated by HNF-1␤ has similar functions in the kidney.
The long transcript regulated by HNF-1␤ represents an example of a lncRNA that functions as a pri-miRNA. Supporting this conclusion, the levels of the lncRNA and miR-200b/a/ 429 cluster are coordinately reduced in HNF-1␤ mutant cells and kidneys, and siRNA knockdown of the lncRNA leads to a corresponding reduction in the levels of mature miR-200b, miR-200a, and miR-429. Moreover, RT-PCR using a forward primer that is upstream to the Ttll10 promoter and a reverse primer within the miR-200b locus confirmed the existence of a continuous RNA transcript extending from upstream to the Ttll10 promoter to the miR-200 cluster. These findings indicate that the lncRNA functions in part as a pri-miRNA that directly encodes the miR-200 cluster. Additional pri-miRNAs encoded by the Ttll10 host gene may contribute to basal expression of miR-200 but do not appear to be involved in HNF-1␤-dependent expression. Recent studies have identified other lncRNAs that are processed to produce miRNAs. For example H19  encodes miR-675-3p and miR-675-5p, linc-MD1 encodes miR-133b, lnc-31 encodes miR-31, and Pvt1 encodes miR-1207-50 (37)(38)(39)(40). The studies described here add miR-200b/a/429 to the miRNAs that can be produced from a lncRNA.
Besides the miR-200b/a/429 cluster on chromosome 4, the miR-200 family includes the miR-200c/141 cluster located on chromosome 6. In the kidney the transcriptional regulation by HNF-1␤ appears to be specific to the miR-200b/a/429 cluster as the levels of the miR-200c/141 cluster are unchanged in HNF-1␤ mutant cells or kidneys. The promoters of the human miR-200b/a/429 and miR-200c/141 clusters have been characterized (41,42), and transcription factors that regulate expression have been identified, including Sp1, the p53/p63/p73 family, and c-Myb (43)(44)(45). These transcription factors may contribute to the residual expression of the miR-200 cluster observed in HNF-1␤ mutant cells and kidneys (Fig. 2). The human miR-200b/a/429 cluster is located upstream of a protein coding gene, TTLL10 (tubulin tyrosine ligase-like, member 10), whereas the mouse cluster is located within the first intron of Ttll10 (Fig. 3A). Although the mouse miR-200b/a/429 cluster is intragenic,itsHNF-1␤-dependentexpressionappearstobeindependent of the host gene. In contrast to the miR-200 cluster, the expression of Ttll10 is unchanged in cells expressing DN-HNF-1␤ and in HNF-1␤-deficient kidneys, and knockdown of the lncRNA does not affect the levels of Ttll10 transcripts. These findings indicate that HNF-1␤ regulates the expression of the miR-200 cluster but does not regulate the expression of Ttll10.
The miR-200 cluster has been implicated in EMT. Downregulation of miR-200 in epithelial cells induces EMT, and decreased expression of miR-200 in cancer cells has been linked to increased tumor metastases. Conversely, overexpression of miR-200 members can induce EMT (41,46). HNF-1␤ also regulates mesenchymal-to-epithelial transitions. The promoter of the HNF-1␤ gene (HNF1B) contains binding sites for the transcriptional repressor Snail (22). Overexpression of Snail downregulates HNF-1␤, which induces EMT possibly through decreased expression of its target gene, Ksp-cadherin (Cadherin-16). Our studies reveal an alternative mechanism whereby down-regulation of HNF-1␤ induces EMT. Loss of HNF-1␤ reduces expression of the lncRNA encoding miR-200 and leads to increased expression of miR-200 targets, including Zeb2. Zeb2 is a transcriptional repressor that is sufficient to induce EMT (47). Epithelial cells normally express low levels of Zeb2; however HNF-1␤-deficient kidneys and cells expressing DN-HNF-1␤ exhibit increased levels of Zeb2. Importantly, overexpression of miR-200a/b in cells expressing DN-HNF-1␤ restores normal expression of Zeb2, indicating that the increased expression is dependent on miR-200a/b.
We have also uncovered a novel mechanism that links mutations in HNF-1␤ to cyst formation in the kidney. Previous studies have shown that the expression of Pkd1, which is mutated in autosomal dominant polycystic kidney disease, is tightly regulated. Alterations in the levels of Pkd1, either transgenic overexpression or expression of hypomorphic alleles, produce kidney cysts (48 -52). We found that the expression of Pkd1 is increased in HNF-1␤ mutant cells and in kidneys from HNF-1␤ knock-out mice. However, Pkd1 does not appear to be a direct transcriptional target of HNF-1␤. Rather, HNF-1␤ regulates the expression of miR-200, which we previously demonstrated controls the post-transcriptional expression of Pkd1 (28). In support of this mechanism, reexpression of miR-200a/b in HNF-1␤ mutant cells reduces expression of Pkd1. Taken together, our studies identify a novel mechanism by which HNF-1␤ regulates the expression of Pkd1 via an lncRNA encoding the miR-200 cluster. Therapeutic approaches that increase the levels of the lncRNA may correct some of the phenotypic abnormalities caused by mutations of HNF-1␤.