This article is mentioned in:

Introduction

Genome-wide association studies have identified numerous single nucleotide polymorphisms (SNPs) that are associated with human diseases, including type 2 diabetes mellitus (T2DM) (1). However, the majority of these SNPs are located in the non-coding regions of the genome and the mechanisms by which they affect disease risk are poorly understood. Potassium voltage-gated channel subfamily Q member 1 (KCNQ1) is a gene that is significantly associated with T2DM in different ethnic groups, particularly in patients of East Asian origin (1–3). Numerous T2DM-associated SNPs, including rs2237892, rs2237895, rs151290, rs2283228 and rs2074196, have been located in a KCNQ1 intron (2–4). A recent study has also indicated that increased KCNQ1 protein expression can limit insulin secretion from pancreatic β cells by regulating potassium channel currents (5). However, to the best of our knowledge, there are currently no data regarding the association between disease-associated variants located in a KCNQ1 intron and the expression of KCNQ1.

Studies from the Encyclopedia of DNA Elements (ENCODE) project have confirmed that disease-associated variants are generally enriched in regulatory DNA, and that promoters and distal elements engage in multiple long-range interactions to form complex networks (6–9). For example, obesity-associated variants within the α-ketoglutarate dependent dioxygenase gene form long-range functional connections with the iroquois homeobox 3 gene as a distal enhancer (8). Therefore, SNPs in the KCNQ1 gene may affect the expression of nearby or remote genes. Among the diabetes-associated candidate genes located close to KCNQ1, cyclin-dependent kinase inhibitor 1C (CDKN1C) is notable. Loss of this gene in focal hyperinsulinism-associated lesions is correlated with an increased proliferation of pancreatic β cells (10) and targeting CDKN1C promotes adult human β cell replication in vitro (11). Furthermore, the long non-coding RNA (lncRNA), KCNQ1 opposite strand/antisense transcript 1 (KCNQ1OT1), regulates the expression of CDKN1C via imprinting (12–14). A mouse model has also demonstrated that a paternal allelic mutation of the KCNQ1 locus reduces pancreatic β cell mass by epigenetic modification of CDKN1C through KCNQ1OT1 (15).

In order to analyze the functional impact of SNPs located in non-coding regions of the genome, our group recently developed a comparative method using a novel nanobead system. It was successfully demonstrated that this method was effective for the identification of allele-specific DNA-binding proteins and may provide important information to understand the functional impact of SNPs (16). Our group demonstrated that nuclear transcription factor Y (NF-Y) specifically bound to a genomic region containing the non-risk allele for T2DM of the KCNQ1 SNP rs2074196. NF-Y also stimulated transcriptional activity in an allele-specific manner in an artificial promoter containing SNP rs2074196 (16). These results suggest that SNP rs2074196 modulates the binding activity of the locus for NF-Y and may induce subsequent changes in gene expression. However, in this previous study, candidate target genes that may be regulated by the SNP rs2074196 locus and NF-Y were not identified. Furthermore, our group detected allele-specific binding proteins for other intronic SNP (rs2237892, rs151290 and rs2283228) regions of the KCNQ1 gene, which revealed that it is possible that other regions may also be functional and contribute to T2DM susceptibility (16).

The ENCODE project has also characterized genome-wide histone modification patterns, including the acetylation of lysine 27 of histone H3 (H3K27Ac), which is often found in histones that flank active promoters and enhancers (17). Among the T2DM-associated SNPs in the KCNQ1 locus, only KCNQ1 SNP rs163184 exists in a genomic region that co-localizes with H3K27Ac in human skeletal muscle myoblasts, normal human lung fibroblasts, human adipose tissue and human pancreatic islets, as demonstrated in the University of California, Santa Cruz Genome Browser (genome.ucsc.edu/). Covalent modifications of histones are known to maintain an appropriate balance in stability and reversibility, and to influence biological processes in the context of development and cellular responses (18). Therefore, the SNP rs163184 locus can be considered to be a functional genomic region; although, it is unclear whether the locus positively or negatively regulates gene expression. Furthermore, the SNP rs163184, and widely studied KCNQ1 SNPs such as rs2237892 and rs2237895, are in moderate linkage disequilibrium with each other (2). This information prompted the speculation that certain proteins or protein complexes may bind to the SNP rs163184 locus in an allele-specific manner and may change the transcriptional regulation of target genes. Therefore, the present study used our group's previously developed comparative nanobead method to investigate SNP rs163184 and identify allele-specific binding proteins for this region. In addition, potential target genes that are regulated by SNP rs163184 and its specific binding proteins were investigated.

Materials and methods

Cells

The rat pancreatic β cell line, INS-1 (19), was kindly provided by Dr C.B. Wollheim and Dr N. Sekine (University of Geneva, Geneva, Switzerland). The INS-1 cells were cultured at 37°C with 5% CO2 atmosphere in RPMI-1640 medium (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany), supplemented with 10% fetal bovine serum (FBS; Biowest, Nuaille, France), 10 mM HEPES, 1 mM sodium pyruvate and 50 mM 2-mercaptoethanol. KP-3, KP-4, QGP-1 and HuH-7 cells were obtained from the Japan Health Science Research Resources Bank (Osaka, Japan). HeLa, 293T, MCF-7 and Caco-2 cells were sourced from the American Type Culture Collection (Manassas, VA, USA). KP-3 and QGP-1 cells were cultured in RPMI-1640 medium supplemented with 10% FBS. KP-4 cells were cultured in Iscove's modified Dulbecco's medium (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% FBS. HuH-7, HeLa and 293T cells were cultured in Dulbecco's modified Eagle's medium (Sigma-Aldrich; Merck KGaA) supplemented with 10% FBS. MCF-7 and Caco-2 cells were cultured in Eagle's minimum essential medium (Sigma-Aldrich; Merck KGaA) supplemented with 10% FBS. These cell lines (KP-3, KP-4, QGP-1, HuH-7, HeLa, 293T, MCF-7 and Caco-2) were cultured at 37°C in an atmosphere with 5% CO2.

Preparation of DNA-immobilized nanobeads

The oligonucleotides used in the present study were synthesized by Eurofins Genomics (Tokyo, Japan). These oligonucleotides contained either risk or non-risk alleles of the KCNQ1 gene via single-nucleotide substitutions. DNA-immobilized nanobeads were prepared as previously described (16). Briefly, two complementary oligonucleotides were phosphorylated, annealed and ligated to produce oligomers that ranged from 100-3,000 base pairs. These oligomers, containing tandemly repeated SNP regions, were applied to a NICK column (cat. no. 17-0855-02; GE Healthcare, Chicago, IL, USA), according to the manufacturer's protocol. The second fraction (100 µg prepared DNA) from each NICK column was mixed with 1 mg FG beads (Tamagawa Seiki Co., Ltd., Nagano, Japan), then a coupling reaction was conducted at 50°C for 24 h.

Affinity purification

Nuclear extracts were prepared from INS-1 cells according to the method described by Dignam et al (20). The DNA-immobilized nanobeads or negative control nanobeads without immobilized DNA (200 µg) were equilibrated using binding buffer [20 mM HEPES-NaOH (pH 7.9), 10% glycerol, 100 mM KCl, 1 mM MgCl2, 0.2 mM CaCl2, 0.2 mM EDTA, 0.1% NP-40, 1 mM DTT and 0.2 mM PMSF]. INS-1 nuclear extracts (200 µg) were mixed with 100 µg of single-stranded DNA (cat. no. AM9680; Ambion; Thermo Fisher Scientific, Inc.) and 10 µg of poly(dI-dC):poly(dI-dC) (Sigma-Aldrich; Merck KGaA). These solutions were then mixed with equilibrated DNA-immobilized nanobeads or negative control nanobeads without immobilized DNA, and incubated for 4 h at 4°C with gentle rotation using a RT-50 rotator (Taitec Corp., Saitama, Japan). After washing with binding buffer three times, bound proteins were eluted with binding buffer supplemented with 1 M KCl.

Mass spectrometry analysis

Affinity purified proteins were separated using SDS-PAGE (5–20% gradient gel) and subjected to silver staining using a protocol described by Shevchenko et al (21). Specific protein bands were excised and mass spectrometry analysis was performed, as previously described (22). Band slices were reduced with 10 mM DTT for 1 h at 56°C and then alkylated with 55 mM iodoacetamide for 45 min at room temperature. The bands were next digested overnight at 37°C with sequencing grade modified trypsin (12 mg/ml; Promega Corp., Madison, WI, USA) and desalted with ZipTip C18 (EMD Millipore, Billerica, MA, USA). Nano liquid chromatography-tandem mass spectrometry analysis was performed as previously described (23). Extracted peptides were separated via nanoflow liquid chromatography (Paradigm MS4; Michrom BioResources, Tokyo, Japan) using a reverse phase C18 column (L-Column Micro; Chemicals Evaluation and Research Institute, Tokyo, Japan). The liquid chromatography eluent was coupled to a nano-ionspray source attached to a QSTAR Elite Q-TOF mass spectrometer (Sciex, Framingham, MA, USA). Peptide identification was performed by matching raw spectra data against rat proteins from the International Protein Index database (24) (version 3.87) using ProteinPilot software (version 3.0; Sciex) and the Paragon algorithm with a confidence threshold of 95% (ProteinPilot unused score >1.3).

Immunoblotting

Antibodies directed against transcription factor Sp1 (Sp1; cat. no. 07-645), transcription factor Sp3 (Sp3; cat. no. 07-107), myelin transcription factor 1 (Myt1; cat. no. ABN21), histone deacetylase 1 (Hdac1; cat. no. 06-720) and REST corepressor 1 (Rcor1; cat. no. MABN486) were obtained from EMD Millipore. Antibodies directed against Myt1-like protein (Myt1l; cat. no. ab139732), REST corepressor 2 (cat. no. ab113826) and NF-YA (cat. no. ab6558) were supplied by Abcam (Cambridge, MA, USA). Antibodies directed against lysine-specific histone demethylase 1A (Lsd1/Kdm1a; cat. no. 2184), Hdac2 (cat. no. 2540) and Myc proto-oncogene protein (Myc)-tagged proteins (Myc-tag; cat. no. 2276) were sourced from Cell Signaling Technology, Inc. (Danvers, MA, USA). The antibody directed against nucleolar protein 4 (Nol4; cat. no. 14802-1-AP) was purchased from ProteinTech Group, Inc. (Chicago, IL, USA). The antibody directed against the FLAG-tag (cat. no. F1804) was purchased from Sigma-Aldrich.

Nuclear extracts (10 µg) from INS-1 cells and affinity-purified fractions were resolved by SDS-PAGE (5–20% gradient gel) and electro-transferred to a polyvinylidene difluoride membrane (EMD Millipore). The membrane was then blocked with 5% non-fat milk for 1 h at room temperature and then incubated overnight at 4°C with the primary antibodies as previously described (16). This was followed by incubation with a horseradish peroxidase-conjugated secondary antibody [anti-mouse-immunoglobulin G (IgG), cat. no. 7076; or anti-rabbit-IgG, cat. no. 7074; Cell Signaling Technology, Inc.] for 1 h at room temperature. Visualization was accomplished using Immobilon Western Chemiluminescent HRP Substrate (cat. no. WBKLS0500; EMD Millipore) and a ChemiDoc XRS Plus system (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

Preparation of radiolabeled recombinant proteins

Total RNA was isolated from HeLa cells using an RNeasy mini kit (cat. no. 74106; Qiagen GmbH, Hilden, Germany) according to the manufacturer's protocol. Complementary DNA (cDNA) was synthesized from the total RNA using the SuperScript III First-Strand Synthesis system (cat. no. 18080-051; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. cDNA for MYT1 and MYT1L were amplified using PrimeSTAR HS DNA Polymerase (cat. no. R010A; Takara Bio, Inc., Otsu, Japan) according to the manufacturer's protocol. The primers used for amplification were as follows: MYT1 forward, 5′-GCCGCCGCGATCGCCATGAGCTTAGAAAATGAAGACAAGC-3′ and reverse, 5′-GCTCGAGCGGCCGCGTGACCTGGATGCCCCTCACAGC-3′; and MYT1L forward, 5′′-GCCGCCGCGATCGCCATGGAGGTGGACACCGAGGAG-3′ and reverse, 5′-GCTCGAGCGGCCGCGTGACCTGAATTCCTCTCACAGCC-3′. The thermocycling conditions were as follows: 98°C for 1 min; followed by 30 cycles of 98°C for 10 sec, 60°C for 5 sec and 72°C for 4 min; and 72°C for 5 min.

The cDNA were inserted into a pCMV6-Entry vector (OriGene Technologies, Inc., Rockville, MD, USA) and sequenced with an ABI 3700 sequence analyzer (Applied Biosystems; Thermo Fisher Scientific, Inc.). Expression plasmids for SP1 (cat. no. RC209154), SP3 (cat. no. RC222027), LSD1/KDM1A (cat. no. RC230555) and PHD finger protein 21A (PHF21A; cat. no. RC213929) were obtained from OriGene Technologies, Inc. These are mammalian expression vectors that express the cloned cDNA (SP1, SP3, LSD1/KDM1A and PHF21A, respectively) with an Myc-FLAG-tag. The subsequent preparation of radiolabeled recombinant protein was performed as previously described (22). T7 promoter-tagged DNA fragments of these cDNA (MYT1, MYT1L, SP1, SP3, LSD1/KDM1A and PHF21A, respectively) were amplified using PrimeSTAR HS DNA Polymerase (Takara Bio, Inc.) according to the manufacturer's protocol. The primers used for amplification were as follows: Forward, 5′-CACCGGTACCTAATACGACTCACTATAGGGAATACAAGCTACTTGTTCTTTTTGCAAGATCTGCCGCCGCGATCGCC-3′ and reverse, 5′-TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTAAACCTTATCGTCGTCATCC-3′. The aforementioned thermocycling conditions were used. These amplified DNA fragments were used to synthesize 35S-radiolabeled recombinant proteins in a coupled transcription/ translation system (TNT T7 Quick for PCR DNA; cat. no. L5540; Promega Corp.), following the manufacturer's protocol.

Binding assay

DNA-immobilized nanobeads or negative control nanobeads without immobilized DNA (200 µg) were equilibrated with binding buffer, as previously described, and incubated with 200 µl of each of the radiolabeled proteins (Sp1, Sp3, Lsd1/Kdm1a, Phf21a, Myt1 and Myt1l) at 4°C for 4 h using a RT-50 rotator. After washing with binding buffer, bound proteins were eluted by boiling for 5 min at 98°C in SDS-PAGE sample buffer [125 mM Tris-HCl (pH 6.8), 250 mM 2-mercaptoethanol, 4% SDS, 0.2% bromophenol blue and 20% glycerol]. Eluates and inputs were subjected to SDS-PAGE (5–20% gradient gel). Gels were dried and auto-radiography was used to visualize the radiolabeled proteins.

Luciferase assay

To construct reporter plasmids that contained a luciferase gene under the control of the SNP rs163184 region, complementary oligonucleotides containing three copies of the SNP region were annealed and cloned into the pGL4.23 vector (Promega Corp.). The cloned sequences were verified as correct using an ABI sequence analyzer 3700 (Thermo Fisher Scientific, Inc.).

The constructed luciferase reporter plasmids (100 ng) and the control Renilla luciferase reporter vector pGL4.74 (10 ng; Promega Corp.), were transfected into 293T and HuH-7 cells (2×105 cells/well) with the mammalian expression vectors (SP1, SP3, LSD1/KDM1A, PHF21A, MYT1 or MYT1L) using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). Cells were seeded into 24-well plates and 12 h after transfection, the culture medium was changed. Cell extracts were prepared using Reporter Lysis Buffer (cat. no. E3971; Promega Corp.), according to the manufacturer's protocol, 48 h after transfection, and the Firefly and Renilla luciferase activities were determined using a Dual-Glo luciferase assay system (Promega Corp.) with a CentroPro LB 962 luminometer (Berthold Technologies GmbH & Co. KG, Bad Wildbad, Germany). Luciferase activities were calculated in relative light units as a ratio of Firefly luciferase activities to Renilla luciferase activities.

The Lsd1/Kdm1a inhibitor 2-PCPA and the monoamine oxidase A (Mao) inhibitor pargyline were obtained from Sigma-Aldrich (Merck KGaA). These reagents (5–500 µM) were added to the culture when the medium was changed.

Chromatin immunoprecipitation (ChIP) assay

Several cell lines (HeLa, MCF-7, KP-3, KP-4, QGP-1, 293T and Caco-2; 6×106 cells/dish) were transfected with 16 µg of the expression vector for Sp3-Myc-FLAG (cat. no. RC222027; OriGene Technologies, Inc.) using Lipofectamine 2000. A total of 48 h after transfection, ChIP sample preparation was performed using a Magna ChIP G kit (EMD Millipore) according to the manufacturer's protocol. Samples were sonicated on ice for 10 rounds of 30 pulses each (0.5 sec on and 0.5 sec off) with intervals of 10 min using a Branson 450D Sonifier (Emerson, Danbury, CT, USA) at a maximum amplitude of 15%. The sonicated samples were immunoprecipitated overnight at 4°C with 2.5 µg anti-Myc-tag antibody. After immunoprecipitation, DNA fragments were separated by 2% agarose gel electrophoresis and fragments between 200 and 1,000 base pairs were isolated using a NucleoSpin Gel and PCR clean-up kit (cat. no. 740609; Macherey-Nagel GmbH, Düren, Germany). The amount of DNA in the input and immunoprecipitated fractions was quantified by quantitative PCR (qPCR) using a Fast SYBR-Green Master mix and a StepOne Real-Time PCR system (both from Applied Biosystems; Thermo Fisher Scientific, Inc.). The thermocycling conditions were as follows: 95°C for 20 sec; followed by 50 cycles of 95°C for 3 sec and 60°C for 30 sec. The primers used for qPCR were as follows: 5′-TTGCTCAGTAACGGACTGGAC-3′ and 5′-GTCAATGTGGGGAGGTGGTA-3′ for the SNP rs163184 region; 5′-ATGGAGAGGGCTGAACACG-3′ and 5′-CCCTTGCCTCACATCCTTT-3′ for the 3,000-base pair region upstream of SNP rs163184; and 5′-AGCCACCAGGAGCAGTTGT-3′ and 5′-TCTGGGAGTTGATGAGGGAAG-3′ for the 3,000-base pair region downstream from SNP rs163184. The primers for upstream and downstream regions of SNP rs163184 were designed as reference regions. Bound to input ratios were determined for each amplicon [cycle threshold (Ct) values of immunoprecipitated samples were normalized to Ct values obtained from input samples] (25).

RNA interference and gene expression analysis

Several cell lines (HeLa, MCF-7, KP-3, KP-4, 293T and Caco-2) were seeded (2×105 cells/well) into 24-well plates and trans-fected with either a control small interfering RNA (siRNA) (10 nM; cat. no. 4390844; Thermo Fisher Scientific, Inc.) or siRNA (10 nM) targeting human SP3 (cat. no. s13324; Silencer Select siRNA; Thermo Fisher Scientific, Inc.) using Lipofectamine 2000.

A total of 48 h after transfection, total RNA was isolated from cells using an RNeasy mini kit (Qiagen) according to the manufacturer's protocol. For RT-qPCR analysis, cDNA was synthesized from the total RNA using the SuperScript III First-Strand Synthesis system (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. qPCR was then performed using the TaqMan Gene Expression Master mix (Applied Biosystems; Thermo Fisher Scientific, Inc.) for each TaqMan probe according to the manufacturer's protocol. The thermocycling conditions were as follows: 95°C for 20 sec; followed by 40 cycles of 95°C for 1 sec and 60°C for 20 sec. Relative mRNA abundance was calculated using the 2−ΔΔCq method (25) and normalized to the corresponding values for β-actin (ACTB).

TaqMan Gene Expression assays for SP3 (UniGene ID, Hs01595811_m1), CD81 molecule (CD81; UniGene ID, Hs01002167_m1), tumor suppressing subtransferable candidate 4 (TSSC4; UniGene ID, Hs00185082_m1), transient receptor potential cation channel subfamily M member 5 (TRPM5; UniGene ID, Hs00175822_m1), KCNQ1 (UniGene ID, Hs00165003_m1), KCNQ1OT1 (UniGene ID, Hs03665990_s1), CDKN1C (UniGene ID, Hs00175938_m1), solute carrier family 22 member 18 (SLC22A18; UniGene ID, Hs00180039_m1), SLC22A18 anti-sense (SLC22A18AS; UniGene ID, Hs00757934_m1), pleckstrin homology like domain family A member 2 (PHLDA2; UniGene ID, Hs00169368_m1), nucleosome assembly protein 1-like 4 (NAP1L4; UniGene ID, Hs00924275_m1), cysteinyl-tRNA synthetase (CARS; UniGene ID, Hs01000965_m1), oxysterol binding protein like 5 (OSBPL5; UniGene ID, Hs00957760_m1), MAS related GPR family member G (MRGPRG; UniGene ID, Hs01010229_s1) and ACTB (UniGene ID, 4352935) were obtained from Applied Biosystems (Thermo Fisher Scientific, Inc.).

Immunoprecipitation and immunoblotting analysis

The expression vectors (4 µg each) for Sp3-Myc-FLAG (cat. no. RC222027; OriGene Technologies, Inc.) and NF-Ys (NF-YA, NF-YB-HA and NF-YC-Myc) were transfected into 293T cells (6×106 cells/ dish) using Lipofectamine 2000. The constructions of the expression vectors for NF-Ys (NF-YA, NF-YB-HA and NF-YC-Myc) were previously described (16). A total of 48 h after transfection, cells were lysed in lysis buffer [50 mM Tris-HCl (pH 8.0), containing 150 mM NaCl and 1.0% NP-40] with protease inhibitor cocktail (cat. no. 25955-11; Nacalai Tesque, Inc., Kyoto, Japan). Immunoprecipitation from transfected cell lysates was performed using anti-FLAG M2 Magnetic beads (cat. no. M8823; Sigma-Aldrich; Merck KGaA), and the beads were washed three times with lysis buffer. Immunoprecipitates or total cell lysates were analyzed by immunoblotting using anti-FLAG-tag antibody (1:1,000; cat. no. F1804; Sigma-Aldrich; Merck KGaA) and anti-NF-YA antibody (1:1,000; cat. no. ab6558; Abcam), using the aforementioned protocol.

Bioinformatics

Image data involving the information on the genomic positions of H3K27Ac mark and SNPs in the KCNQ1 locus were downloaded from the University of California, Santa Cruz Genome Browser (genome.ucsc.edu/).

Statistical analysis

All data are expressed as the mean ± standard deviation from at least three independent experiments. Statistical analysis was performed using an unpaired two-tailed Student's t-test for comparisons of two groups or by two-way non-repeated measures analysis of variance, followed by the Bonferroni post hoc test, for the comparisons of multiple groups. These statistical analyses were performed using Microsoft Excel 2011 for Mac (Microsoft Corporation, Redmond, WA, USA) with the add-in software, Excel Statistical Program File ystat2008 (Igakutosho-shuppan, Ltd., Tokyo, Japan). P<0.01 was considered to indicate a statistically significant difference.

Results

Purification of allele-specific DNA-binding proteins using magnetic nanobeads

One of the SNPs that is most highly associated with T2DM is the KCNQ1 SNP rs163184 (2). The position of this SNP overlaps with a H3K27Ac, which has been shown by the ENCODE project to mark active regulatory elements (Fig. 1A). The present study aimed to identify proteins that bind the genomic region that encompasses the SNP rs163184 region (Fig. 1B). To identify allele-specific binding proteins, DNA-immobilized beads were prepared by binding each SNP rs163184-containing allele onto magnetic nanobeads (Fig. 2A). Using these nanobeads, DNA-binding proteins were purified from the nuclear extract of INS-1 pancreatic β cells and bound proteins were analyzed using SDS-PAGE (Fig. 2B). This revealed that the binding activities of several proteins differed between the alleles for the region containing SNP rs163184. Notably, in the sample using oligo 3, many extra bands were detected, which were not observed, or less observed, in the sample using oligos 4 or 5 (Fig. 2B and C).

Figure 1 Schematic of the experimental design. (A) Screenshots from the University of California, Santa Cruz Genome Browser indicating the human KCNQ1 gene locus. The position of the SNP rs163184 examined is indicated by arrows. (B) Workflow for the purification of the nuclear extracts from INS-1 cells using nanobead technology and subsequent analyses. KCNQ1, potassium voltage-gated channel subfamily Q member 1; SNP, single nucleotide polymorphism; NR, non-risk.

Figure 2 Purification and identification of allele-specific proteins bound to the region containing SNP rs163184. (A) Oligos immobilized on the nanobeads. Oligos 6 and 7 were designed for the risk and the NR alleles of SNP rs2074196, respectively, as described in our previous study (16), and were included to demonstrate the reproducibility of the purification system using nuclear extracts from INS-1 cells. Oligo 5 was designed against the predicted PPARA binding site, which was also included in oligos 3 and 4. However, further analysis identified no binding between PPARA and this region. (B) Aliquots of purified fractions were separated using SDS-PAGE (5–20% gradient gel) and visualized using silver staining. (C) The same samples (panel B, lanes 3 and 4) were separated using longer SDS-PAGE (5–20% gradient gel) and visualized using silver staining to allow for the excision of the specific protein bands. The names and locations on the gel of polypeptides that were analyzed using ion-spray mass spectrometry are indicated by corresponding numbers. Polypeptide band 4 was broad and part of the band was also present in the oligo 4 (risk) lane, designated as polypeptide band 4′ and used in further analysis. SNP, single nucleotide polymorphism; oligo, oligonucleotides; NR, non-risk.

Following mass spectrometry analysis, the proteins described in Fig. 2C were identified, including Myt1/Myt1l/St18, Sp3, Lsd1, Nol4/Nol4l/Sp1, Nol4/Phf21a, Hdac1/Hdac2/Rcor1/Rcor2/ Rcor3 and Nol4/Nol4l. Immunoblotting with specific antibodies directed against these proteins confirmed that Myt1, Myt1l, Sp1, Sp3, Lsd1/Kdm1a, Hdac1, Hdac2 and Rcor1 were purified in an allele-specific manner (Fig. 3A). It was not possible to obtain a specific antibody directed against rat Phf21a that worked effectively. Nol4 only exhibited non-specific binding activity against the SNP rs163184 genomic region. Sp3 exhibited the highest recovery efficiency among these proteins.

Figure 3 Evaluation of the binding activities of the identified proteins. (A) INS-1 cell nuclear extracts and purified fractions were subjected to immunoblotting with specific antibodies. Except for Sp3, 10% of the nuclear extract used for purification were loaded onto a gel. For Sp3, 50% of the nuclear extract was used for electrophoresis. (B) In vitro radiolabeled proteins were individually mixed with DNA-immobilized nanobeads of the single nucleotide polymorphism rs163184 region (risk or NR) or (−). Input and eluate fractions were separated using SDS-PAGE and visualized by autoradiography. Except for Sp3, 10% of the samples used for binding assay were loaded onto a gel. For Sp3, 50% of the sample was subjected to electrophoresis. NR, non-risk; (−), negative control nanobeads without immobilized DNA.

Next, the in vitro DNA-binding activities of Myt1, Myt1l, Sp1, Sp3, Lsd1/Kdm1a and Phf21a were examined using recombinant proteins. Myt1, Myt1l, Sp3 and Phf21a each exhibited higher specific binding activity against the non-risk allele of the genomic region that includes SNP rs163184 (Fig. 3B). The DNA-binding specificities of recombinant Myt1, Myt1l, and Sp3 against the non-risk allele were consistent with those observed in the immunoblotting of purified proteins from INS-1 nuclear extracts (Fig. 3). Lsd1/Kdm1a exhibited no DNA binding activity (Fig. 3B). This indicates that Lsd1/Kdm1a is recruited to the rs163184 region by other proteins, such as Sp3, which can bind to the region in an allele-specific manner. Unexpectedly, the transcription factor Sp1 also exhibited no observable binding activity for this SNP region (Fig. 3B). This suggests that Sp1 is also recruited to the SNP 163184 region by other proteins, or due to post-translational modifications that are not present in the recombinant Sp1 synthesized in the cell-free system and are required for binding to this SNP region. Sp3 exhibited the highest DNA-binding activity among the proteins examined for this SNP region.

SNP rs163184 modulates Sp3-dependent transcriptional activity

The transcriptional activities of Myt1, Myt1l, Sp1, Sp3, Phf21a and Lsd1/Kdm1a were investigated using a luciferase reporter gene assay. As illustrated in Fig. 4A, two types of reporter constructs that were either risk or non-risk alleles were prepared. The basal luciferase activities, mediated by endogenous transcription factors (that may include Sp3), were markedly different between the two types of constructs (Fig. 4A). The relative luciferase activities (RLUs) of the non-risk allele reporter construct were higher than those of the risk allele reporter construct in both 293T and HuH-7 cells.

Figure 4 Effects of the identified proteins on transcriptional activity under the control of the genomic region containing single nucleotide polymorphism rs163184. (A) Construction of Firefly luciferase reporter vectors. (B) 293T and HuH-7 cells were transfected with a Firefly luciferase reporter construct and a control Renilla luciferase reporter vector, in addition to a mammalian expression vector of individual identified proteins. (C) 293T cells were transfected with a Firefly luciferase reporter construct and a control Renilla luciferase reporter vector. At 12 h post-transfection, the medium was changed and inhibitors (2-PCPA or pargyline) were added (5-500 µM). *P<0.01 vs. the (−) group; #P<0.01 vs. the untreated group of each reporter construct. NR, non-risk; (−), without exogenous effectors.

When the expression construct containing SP3 was cotransfected with each of the luciferase reporter plasmids under the control of the SNP rs163184 region, RLUs of the non-risk allele reporter construct were increased (RLU, 6.16 vs. 59.17 in 293T cells; RLU, 1.68 vs. 27.79 in HuH-7 cells) (Fig. 4B). The RLUs of the risk allele reporter construct were also increased; however, these were still much lower than the RLUs of the non-risk allele reporter constructs. In HuH-7 cells, Sp3-induced a larger fold increase in luciferase activities in non-risk allele constructs (RLU, 1.68 vs. 27.79; fold change, 16.57) compared to those in risk allele constructs (RLU, 1 vs. 5.32; fold change, 5.32). These results suggest that Sp3 may modulate the transcription of some specific genes using the genomic region encompassing KCNQ1 SNP rs163184 as a cis-regulatory element, particularly in the non-risk allele.

When the expression construct containing MYT1 was cotransfected, the RLUs of each reporter construct were decreased (Fig. 4B). This suggests that Myt1 downregulates transcription controlled by the genomic region flanking SNP rs163184.

When the expression construct containing KDM1A was cotransfected with that of SP3, Sp3-dependent luciferase activity was decreased compared to that in cells transfected with SP3 alone (Fig. 4B). This indicates that Lsd1/Kdm1a negatively regulates Sp3-dependent transcription under the control of the SNP rs163184 region.

In order to confirm the suppressive effect demonstrated by Lsd1/Kdm1a, the Lsd1/Kdm1a inhibitor 2-PCPA, which reduces Lsd1/Kdm1a, Mao-A and Mao-B activity, was used. The Mao inhibitor pargyline, which inhibits Mao-A and Mao-B, but is less potent against Lsd1/Kdm1a, was also used. The Lsd1/Kdm1a inhibitor 2-PCPA enhanced the luciferase activities of the non-risk allele reporter construct (RLU, 8.53 vs. 22.32) (Fig. 4C). However, the Mao inhibitor pargyline had little effect on the RLUs of the non-risk allele reporter construct (RLU, 8.53 vs. 8.58) (Fig. 4C). In addition, 2-PCPA enhanced the RLUs of the risk allele reporter construct (RLU, 1 vs. 2.98), although it was still much lower than that of the non-risk allele reporter construct (Fig. 4C).

SNP rs163184 modulates the binding activity of the locus for Sp3

The DNA-binding activities of Sp3 were further examined using a ChIP assay in HeLa, MCF-7, KP-3, KP-4, QGP-1, 293T and Caco-2 cells. HeLa, MCF-7 and KP-3 cells were identified to be homozygous for the non-risk allele, while KP-4 and QGP-1 cells were heterozygous, and 293T and Caco-2 cells were homozygous for the risk-associated allele. These cell lines were transfected with expression vectors containing Sp3-Myc-FLAG. ChIP analysis revealed that the genomic region containing the KCNQ1 SNP rs163184 was highly enriched for Sp3 in MCF-7 and KP-3 cells, moderately enriched in QGP-1 cells, slightly enriched in HeLa, KP-4 and Caco-2 cells, and not significantly enriched in 293T cells (Fig. 5). Taken together with the data from the in vitro binding assay, these results suggest that Sp3 preferentially binds the non-risk allele of the endogenous KCNQ1 intronic region that contains SNP rs163184. However, the possibility that these cell lines may have additional genetic variants that impact on the recruitment of Sp3 to the SNP rs163184 region cannot be excluded.

Figure 5 Allele-specific binding activity of Sp3 for the genomic region containing SNP rs163184. Cells were transfected with expression vectors for Sp3-Myc-FLAG. DNA-protein complexes were then immunoprecipitated with the anti-Myc-tag antibody. Enrichment of the target genomic regions was estimated by quantitative polymerase chain reaction analysis, and the fold enrichments of the SNP rs163184 region vs. the region 3,000 base pairs upstream and the region 3,000 base pairs downstream were determined. *P<0.01 vs. the upstream group. NR, non-risk; SNP, single nucleotide polymorphism.

In addition, in order to examine the interaction between Sp3 and NF-Y, which binds the SNP rs2074196 region in the KCNQ1 locus and possibly induces subsequent change in gene expression, immunoprecipitation and immunoblotting experiments were performed. These results demonstrated possible interaction between Sp3 and NF-Y (data not shown).

CDKN1C is a candidate target gene regulated by the SNP rs163184 locus

In order to identify target genes that are regulated by Sp3 and the locus encompassing the KCNQ1 SNP rs163184, the expression of genes flanking KCNQ1 after siRNA-mediated knockdown of endogenous SP3 was examined in HeLa, MCF-7, KP-3, KP-4, 293T and Caco-2 cells. QGP-1 cells were excluded from this analysis as endogenous SP3 could not be sufficiently reduced (data not shown). In the luciferase reporter gene assay, it was demonstrated that Sp3 upregulated the expression of the reporter gene under the control of the KCNQ1 SNP rs163184 region and that Lsd1/Kdm1a downregulated Sp3-dependent transcription under the control of the SNP rs163184 region. The present study indicated that Lsd1/Kdm1a itself does not exhibit DNA-binding activity, at least to the SNP rs163184 region, and seems to be recruited to this genomic region by other transcription factors such as Sp3. Therefore, certain genes are upregulated by Sp3 alone and others are suppressed by Sp3 and Lsd1/Kdm1a, which are possible targets of Sp3.

The expression levels of genes following SP3 knockdown, relative to cells transfected with control siRNA, are presented in Fig. 6. The downstream target gene was chosen depending on the following criteria: i) Gene expression was upregulated or downregulated after SP3 knockdown in the cell lines that possessed one or two non-risk alleles (HeLa, MCF-7, KP-3 and KP-4); or ii) gene expression was hardly changed after SP3 knockdown in the cell lines that only possessed the risk allele (293T and Caco-2). The expression of CDKN1C after SP3 knockdown was significantly upregulated in MCF-7 (fold change, 2.14), KP-3 (fold change, 2.02) and KP-4 (fold change, 2.19) cells. By contrast, the expression of CDKN1C revealed little observable change in 293T and Caco-2 cells. None of the other genes in close proximity that were examined fit these criteria, despite certain genes being upregulated or downregulated depending on the particular cell line.

Figure 6 Effect of SP3 knockdown on the expression of KCNQ1 and nearby genes. Cell lines were transfected with control siRNA or siRNA that targeted human SP3. Gene expression levels were determined by reverse transcription-quantitative polymerase chain reaction analysis and normalized to β-actin. The expression level of each gene using negative control siRNA was designated as 1.0 in each cell line. The data for TRPM5 and SLC22A18AS from all cell lines, as well as the data for KCNQ1 and OSBPL5 from Caco-2 cells, have been excluded because of the limit of detection. *P<0.01 vs. negative control siRNA samples. siRNA, small interfering RNA; NR, non-risk.

These results indicate that the SNP rs163184 locus exerts a negative impact on CDKN1C expression through regulation by Sp3 and Lsd1/Kdm1a, suggesting that CDKN1C is a downstream target gene of the SNP rs163184 locus.

Discussion

Previous genome-wide association studies have indicated that SNPs in the KCNQ1 locus are associated with development of T2DM (2–4). As the majority of these SNPs are located in the intronic regions of KCNQ1 and are associated with impaired insulin secretion, it was hypothesized that several of these SNP-containing regions participate in regulating gene expression in pancreatic β cells. These may include cis-regulatory elements, such as enhancers; SNPs may also affect associated activity at the KCNQ1 locus. In a previous comparative analysis between alleles using DNA-immobilized nanobeads, our group systematically identified allele-specific DNA-binding proteins in several SNP regions (rs2237892, rs151290, rs2283228 and rs2074196) (16). In addition, it was demonstrated that the KCNQ1 SNP rs2074196 modulates the binding activity of the locus for NF-Y, which may in turn induce changes in gene expression (16).

In the present study, it was indicated that the function of the SNP rs163184 region was allele-specific and context-sensitive (Fig. 7). It was demonstrated that, for several proteins, binding activities for the SNP rs163184 region in the KCNQ1 gene locus differed between risk and non-risk alleles. Sp3 preferentially bound to the non-risk allele of the KCNQ1 SNP rs163184 region and stimulated transcriptional activity of an artificial promoter containing the SNP rs163184 region. Lsd1/Kdm1a also preferentially bound to the non-risk allele of the SNP rs163184 region indirectly, possibly via the formation of complexes with other proteins, such as Sp3. This suppressed the Sp3-dependent transcriptional activity of the artificial promoter containing the SNP rs163184 region. Furthermore, CDKN1C was indicated to be a functional target gene regulated by Sp3, Lsd1/Kdm1a and the locus containing SNP rs163184.

Figure 7 Proposed mechanisms of gene regulation mediated by single nucleotide polymorphism rs163184, Sp3 and Lsd1/Kdm1a.

CDKN1C is a cyclin-dependent kinase inhibitor and a negative regulator of cellular proliferation, which is located in close proximity to KCNQ1 and SNP rs163184. CDKN1C has been demonstrated to regulate pancreatic β cell proliferation, and the loss of CDKN1C in focal hyperinsulinism-associated lesions is associated with an increased proliferation of pancreatic β cells (10). Additionally, targeting CDKN1C promotes adult human β cell replication in vitro (11). A recent study has also indicated that a novel variant of CDKN1C is associated with early-adulthood-onset diabetes, as well as intrauterine growth restriction and a short stature (26). Therefore, overexpression of CDKN1C may lead to reduced insulin production and the onset of diabetes due to decreased proliferation of β cells during development. In the present study, the risk-associated allele of SNP rs163184 may have induced overexpression of CDKN1C because of decreased binding activity for Sp3 and Lsd1/Kdm1a at that locus. Therefore, it was hypothesized that SNP rs163184 in the KCNQ1 locus increases T2DM susceptibility, at least in part, through the modulation of CDKN1C expression.

Several mechanisms may contribute to the regulation of CDKN1C gene expression. Firstly, the activity of the CDKN1C promoter is regulated by the transcription factors Sp1 and early growth response protein 1 (27,28). Secondly, the lncRNA KCNQ1OT1 regulates the expression of CDKN1C via imprinting (12–14). Finally, the present study revealed that the KCNQ1 intron 15 region is also involved in regulation of CDKN1C expression via the recruitment of Sp3 and Lsd1/Kdm1a. Lsd1/ Kdm1a is a component of the B-Raf proto-oncogene serine/threonine kinase-HDAC repression complex, which includes Hdac1, Hdac2, Rcor1, Phf21a and high mobility group 20B (29). This multimeric complex regulates histone acetylation and methylation (29). In human breast cancer cells, depletion of Lsd1/Kdm1a by siRNA enhances CDKN1C gene expression (30). Therefore, Lsd1/Kdm1a is a negative regulator of CDKN1C expression. Although the factor that recruits Lsd1/Kdm1a to genomic regions remains unknown, Sp3 and Phf21a are considered potential candidates. Phf21a was reported to form a protein complex with Lsd1/Kdm1a (29), although the binding activity of Phf21a for this SNP region appears weak. Sp3 has a high binding activity for this SNP region and it was reported that SUMOylation of Sp3 switched its function from an activator to a repressor via an association with Lsd1/Kdm1a (31,32). Although the function of Sp3 in pancreatic β cells is not well understood, several studies have identified Sp3-regulating genes in pancreatic β cells, including pancreatic and duodenal homeobox 1, glucagon receptor, potassium voltage-gated channel subfamily J member 11 and gastric inhibitory polypeptide receptor (33–36). In addition to these observations, the present study revealed that Sp3 and Lsd1/ Kdm1a regulated the expression of CDKN1C, suggesting that Sp3 participates in the regulation of β cell development and β cell function.

It is also necessary to consider several additional points. Firstly, it is possible that other genomic regions in KCNQ1 harboring intronic SNPs, such as rs2237892, rs2237895 and rs2074196, are also functional and contribute to T2DM susceptibility. As our groups has previously revealed, SNP rs2074196 modulates the binding activity of the locus for NF-Y and possibly induces subsequent changes in gene expression (16). In the present study, our group has also demonstrated a possible interaction between Sp3 and NF-Y. One explanation for this is that the KCNQ1 SNP rs163184 and KCNQ1 SNP rs2074196 regions recruit and interact with Sp3 and NF-Y, respectively, and that each have functional target genes. Another possible interpretation is that CDKN1C expression may be cooperatively regulated by the rs163184 and rs2074196 SNP regions by recruiting and interacting with Sp3 and NF-Y together. Furthermore, our group has already detected allele-specific binding proteins for other KCNQ1 SNP regions, including rs2237892, rs151290 and rs2283228, that are also highly associated with T2DM (16). Secondly, one of the major limitations of the present study is that the data examining CDKN1C expression were obtained from human cell lines, which were not derived from pancreatic β cells. For the purification of proteins binding to the SNP rs163184 region, nuclear extracts prepared from rat INS-1 cells were used, which have been widely used for the study of islet biology. However, this cell line could not be used for the analysis of the functional impact of the intrinsic SNP region (ChIP assay and endogenous gene expression analysis), as the sequence homology for the genomic region including SNP rs163184 is very low between humans and rodents. In addition, there are very few human islet cell lines established for research. Thirdly, we have examined only genes close to KCNQ1 as potential candidate targets. The ENCODE project has revealed that there can be long-range interactions between gene promoters and disease-associated variants (7). We cannot, therefore, exclude the possibility that remote genes may also be functional targets. Further experiments employing each allele-specific binding protein to each T2DM-associated KCNQ1 SNPs and examining the interactions between these SNPs using chromosome conformation capture assays should clarify any molecular relationship between KCNQ1 SNPs and T2DM susceptibility.

Although recent genome-wide association studies have been extremely successful, it remains a challenge to functionally annotate SNPs that localize to non-coding regions in susceptibility genes. In order to understand the functional impact of individual disease-associated SNPs, it is essential to identify DNA-binding proteins that recognize the nucleotide sequence surrounding these disease-associated variants. The comparative analysis conducted in the present study between alleles using DNA-immobilized nanobeads has proven to be a useful and powerful tool that can be used to investigate the molecular mechanisms that link non-coding genetic variants to the risks of developing common diseases.

Acknowledgments

The authors would like to thank Dr Wataru Nishimura, Dr Miho Kawaguchi, Mr. Dai Suzuki, Ms. Kazuko Nagase and Ms. Keiko Kano (National Center for Global Health and Medicine, Tokyo, Japan) for their technical assistance. The present study was supported by the Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (grant no. 22510216; to M.H.), the Japan Diabetes Foundation (to M.H.), Suzuken Memorial Foundation (to M.H.), the Leading Project of Ministry of Education, Culture, Sports, Science and Technology, Japan (to K.Y.), National Institute of Biomedical Innovation (to K.Y.) and National Center for Global Health and Medicine (to K.Y.).

Abbreviations:

References