Abstract

Single-nucleotide polymorphisms (SNPs) in CDKAL1 have been associated with the development of type 2 diabetes (T2D). CDKAL1 catalyzes 2-methylthio modification of adenosine at position 37 of tRNALys(UUU). A deficit of this modification causes aberrant protein synthesis, and is associated with impairment of insulin secretion in both mouse model and human. However, it is unknown whether the T2D-associated SNPs in CDKAL1 are associated with downregulation of CDKAL1 by regulating the gene expression. Here, we report a specific splicing variant of CDKAL1 termed CDKAL1-v1 that is markedly lower in individuals carrying risk SNPs of CDKAL1. Interestingly, CDKAL1-v1 is a non-coding transcript, which regulates the CDKAL1 level by competitive binding to a CDKAL1-targeting miRNA. By direct editing of the genome, we further show that the nucleotides around the SNP regions are critical for the alternative splicing of CDKAL1-v1. These findings reveal that the T2D-associated SNPs in CDKAL1 reduce CDKAL1-v1 levels by impairing splicing, which in turn increases miRNA-mediated suppression of CDKAL1. Our results suggest that CDKAL1-v1-mediated suppression of CDKAL1 might underlie the pathogenesis of T2D in individuals carrying the risk SNPs.

INTRODUCTION

Recent advances in genome-wide association studies have successfully revealed a number of loci associated with susceptibility to type 2 diabetes (T2D) (1–4). Among these risk loci, the Cdk5 Regulator Subunit Associated Protein 1-Like 1 (CDKAL1) locus is one of the most reproducible loci in European and Asian populations (5). Individuals carrying T2D-associated single-nucleotide polymorphisms (SNPs) in CDKAL1 show impaired insulin secretion and up to a 2-fold increase in the risk of T2D (1). We showed that Cdkal1 is a mammalian methylthiotransferase that catalyzes the 2-methylthio (ms2) modification of N6-threonyl-carbamoyladenosine (t6A) to produce 2-methylthio-N6-threonyl-carbamoyladenosine (ms2t6A) at position 37 of tRNALys(UUU) (6). The ms2-modification of tRNALys(UUU) is critical for the accurate decoding of the lysine codons AAA and AAG by stabilizing the codon–anticodon interaction (7). The precise decoding of the lysine codon by ms2-modification was particularly important for proinsulin synthesis, because a lack of Cdkal1 significantly compromised the proper translation and processing of proinsulin (7). As a result, the Cdkal1 knockout mice showed impaired insulin secretion and glucose metabolism. Furthermore, the downregulation of 2-methylthio modification level was also associated with impaired insulin secretion and T2D risk in human (8). Taken together, these results suggest that 2-methylthio modification in tRNALys(UUU) by Cdkal1 is critical for precise protein synthesis and the development of T2D.

In contrast to the discovery of molecular function of CDKAL1 and its pathological relevance, there remains fundamental question whether there is a correlation between the T2D-associated SNPs in CDKAL1 with the regulation of the CDKAL1 gene. Like most of disease-related SNPs, T2D-associated SNPs in CDKAL1 are located in deep intronic region (9). These intronic regions are neither conserved across species, nor do these regions contain predictable regulatory elements (1). Because of these obstacles, how the T2D-associated SNPs in CDKAL1 would affect gene function and lead to the development of T2D remains unknown.

In the present study, we investigated the functional role of T2D-associated SNPs in CDKAL1 and showed that the SNPs were actively involved in the regulation of cellular CDKAL1 levels through a unique post-transcriptional mechanism.

RESULTS

Associations of a specific CDKAL1 splicing variant (CDKAL1-v1) with T2D-associated SNPs and insulin secretion

The T2D-associated SNPs of CDKAL1 are associated with impaired insulin secretion as well as a decrease in CDKAL1-mediated ms2-modification in tRNALys(UUU) (8,10). To investigate the correlation of T2D-associated SNPs with the gene function, we purified DNA and RNA from peripheral blood samples from non-diabetic volunteers and compared the expression level of full-length CDKAL1 (NM_017774.3) in individuals carrying risk SNPs. We examined rs10946398 because the SNP has been extensively studied in T2D and are in high linkage disequilibrium with other SNPs such as rs7756992 and rs7754840 (11). The expression level of full-length CDKAL1 in carriers of the risk allele C/C of rs10946398 was not significantly different from that in carriers of the non-risk alleles A/A and A/C (Fig. 1A), although we observed a trend showing a decrease in CDKAL1 mRNA levels in carriers of risk alleles (P = 0.06, 87% for C/C carriers compared with A/A carriers, Fig. 1A).

Figure 1.

Decreased expression of CDKAL1-v1 in carriers of CDKAL1 risk alleles. (AC) Comparison of the CDKAL1 (A), CDKAL1-v1 (B) or CDKAL1-v2 (C) levels in groups of homozygous carriers of non-risk alleles (A/A), heterozygous carriers of a non-risk and a risk allele (C/A) and homozygous carriers of risk alleles (C/C) for rs10946398. n = 30 for A/A, n = 40 for C/A, n = 18 for C/C. (D) Correlation between the ms2-modification level of tRNALys(UUU) and the CDKAL1-v1 level in individuals with mixed genotypes. n = 94. (E) Correlation of the corrected insulin response with the CDKAL1-v1 level in individuals with mixed genotypes. n = 32. Error bars represent mean ± SEM.

Figure 1.

Decreased expression of CDKAL1-v1 in carriers of CDKAL1 risk alleles. (AC) Comparison of the CDKAL1 (A), CDKAL1-v1 (B) or CDKAL1-v2 (C) levels in groups of homozygous carriers of non-risk alleles (A/A), heterozygous carriers of a non-risk and a risk allele (C/A) and homozygous carriers of risk alleles (C/C) for rs10946398. n = 30 for A/A, n = 40 for C/A, n = 18 for C/C. (D) Correlation between the ms2-modification level of tRNALys(UUU) and the CDKAL1-v1 level in individuals with mixed genotypes. n = 94. (E) Correlation of the corrected insulin response with the CDKAL1-v1 level in individuals with mixed genotypes. n = 32. Error bars represent mean ± SEM.

Interestingly, we noticed there are two short splicing variants of human CDKAL1 in the NCBI database: AK310219 and BC064145, hereinafter referred as CDKAL1-v1 and CDKAL1-v2, respectively. We then examined the expression levels of CDKAL1-v1 and CDKAL1-v2 in carriers of risk alleles. CDKAL1-v1, but not CDKAL1-v2, was markedly decreased in carriers of the risk alleles for rs10946398 (Fig. 1B and C). The significant decrease in CDKAL1-v1 was also observed in carriers of the risk allele for rs7756992 (Supplementary Material, Fig. S1A and B). In addition to peripheral blood samples, the decrease in the CDKAL1-v1 level was also confirmed in human embryonic fibroblast cell lines carrying the risk alleles for rs10946398 and rs7756992 (Supplementary Material, Fig. S1C and D). Furthermore, to investigate the specificity of the correlation of CDKAL1-v1 with CDKAL1 SNPs, we examined the expression level of CDKAL1-v1 in carriers of the T2D-associated risk alleles of KCNQ1, which is also one of the most reliable T2D-associated genes (12,13). There was no significant change in CDKAL1-v1 level in carriers of the KCNQ1 risk allele (C/C) compared with carriers of non-risk alleles (T/T and C/T) (Supplementary Material, Fig. S2), suggesting that a decrease in CDKAL1-v1 is solely dependent on the risk alleles of CDKAL1.

Cdkal1-dependent 2-methylthio (ms2-) modification in tRNALys(UUU) is decreased in individuals carrying the risk alleles of the CDKAL1 SNPs and is correlated with insulin secretion (7,8). Given that CDKAL1 SNPs were associated with CDKAL1-v1 levels, it is speculated that CDKAL1-v1 level might also correlate with the ms2-modification level and insulin secretion. The ms2-modification and CDKAL1-v1 level were then subjected to quantitative analysis in total RNA isolated from peripheral blood specimens. As expected, there was a direct correlation between CDKAL1-v1 and ms2-modification level (r = 0.23, P = 0.0205, Fig. 1D). Next, we performed oral glucose tolerance tests in non-diabetic individuals and investigated the correlation of CDKAL1-v1 with insulin secretion. There was a direct correlation between CDKAL1-v1 level and insulin secretion (r = 0.39, P = 0.02, Fig. 1E). These results suggest that the decrease in CDKAL1-v1 may contribute to the development of T2D through impairment of CDKAL1 function in individuals carrying the risk alleles in CDKAL1.

Regulation of protein and mRNA levels of CDKAL1 by CDKAL1-v1

The CDKAL1-v1 transcript contains exons 1–4 and a portion of 3′UTR derived from intron 4 (Fig. 2A). Because the open-reading frame (ORF) of CDKAL1-v1 lacks the entire radical SAM domain and tRNA binding domain, which are absolutely required for methylthiotransferase activity (6), CDKAL1-v1 is not a functional tRNA-modifying enzyme. Thus, the direct correlation of CDKAL1-v1 with ms2-modification level could not be explained by a change in the enzyme activity of CDKAL1-v1. We then speculated that CDKAL1-v1 might modulate the ms2-modification level by regulation of the cellular CDKAL1 level. To investigate whether CDKAL1-v1 plays a direct role in the regulation of CDKAL1 levels, CDKAL1-v1 was acutely silenced by specific small interfering RNAs (siRNAs) targeting the 3′UTR in HEK293 cells. Knockdown of CDKAL1-v1 markedly reduced the CDKAL1 protein level (Fig. 2B). Accordingly, the ms2-modification level was significantly decreased in CDKAL1-v1-silenced cells (Fig. 2C), suggesting that CDKAL1-v1 regulates CDKAL1 activity in a translation-dependent manner. Interestingly, treatment with CDKAL1-v1-specific siRNAs resulted in a moderate but significant decrease in CDKAL1 mRNA (83% decrease by #1 siRNA, and 70% decrease by #2 siRNA compared with control, Fig. 2D). Conversely, treatment of CDKAL1-specific siRNA resulted in a moderate but significant decrease in CDKAL1-v1 (79% decrease by siCdkal1, Fig. 2E). The co-regulation of CDKAL1 and CDKAL1-v1 was also observed in other cell lines such as HeLa cells as well as human islet carcinoma QGP-1 cells (Supplementary Material, Fig. S3). These results prompted us to examine the correlation between CDKAL1 and CDKAL1-v1 levels in individuals regardless of genotype. Indeed, we observed a significant correlation between CDKAL1-v1 and CDKAL1 in individuals (r = 0.25, P = 0.016, Fig. 2F). The expression levels of CDKAL1 and CDKAL1-v1 were also examined in normal human tissues and cell lines. There was a direct correlation between CDKAL1 and CDKAL1-v1 (r = 0.5385, P = 0.0038, Fig. 2G). Taken together, these results suggest that there is a post-transcriptional regulation between CDKAL1-v1 and CDKAL1.

Figure 2.

Regulation of CDKAL1 level by CDKAL1-v1. (A) Structures of CDKAL1 and CDKAL1-v1. The 5′UTR and 3′UTR of CDKAL1-v1 are different from those of CDKAL1. The difference is highlighted by a red line in CDKAL1 and a green line in CDKAL1-v1. (B) Decrease in the CDKAL1 level in HEK293 cells transfected with CDKAL1-v1-targeting siRNAs (siCDKAL1-v1 #1 and siCDKAL1-v1 #2). The CDKAL1-targeting siRNA (siCDKAL1) was used as positive control. (C) Decrease in ms2-modification in HEK cells transfected with CDKAL1-v1-targeting siRNAs. n = 4. (D) Downregulation of CDKAL1 by CDKAL1-v1-targeting siRNAs. n = 4. (E) Downregulation of CDKAL1-v1 by CDKAL1-targeting siRNAs. n = 4. (F) Correlation of CDKAL1 and CDKAL1-v1 level in individuals with mixed genotypes. (G) Correlation of CDKAL1 and CDKAL1-v1 level in human tissues and cell lines. Inserted graph in the upper left corner shows linear regression of CDKAL1 versus CDKAL1-v1. Error bars represent mean ± SEM.

Figure 2.

Regulation of CDKAL1 level by CDKAL1-v1. (A) Structures of CDKAL1 and CDKAL1-v1. The 5′UTR and 3′UTR of CDKAL1-v1 are different from those of CDKAL1. The difference is highlighted by a red line in CDKAL1 and a green line in CDKAL1-v1. (B) Decrease in the CDKAL1 level in HEK293 cells transfected with CDKAL1-v1-targeting siRNAs (siCDKAL1-v1 #1 and siCDKAL1-v1 #2). The CDKAL1-targeting siRNA (siCDKAL1) was used as positive control. (C) Decrease in ms2-modification in HEK cells transfected with CDKAL1-v1-targeting siRNAs. n = 4. (D) Downregulation of CDKAL1 by CDKAL1-v1-targeting siRNAs. n = 4. (E) Downregulation of CDKAL1-v1 by CDKAL1-targeting siRNAs. n = 4. (F) Correlation of CDKAL1 and CDKAL1-v1 level in individuals with mixed genotypes. (G) Correlation of CDKAL1 and CDKAL1-v1 level in human tissues and cell lines. Inserted graph in the upper left corner shows linear regression of CDKAL1 versus CDKAL1-v1. Error bars represent mean ± SEM.

CDKAL1-v1 is a non-coding RNA

The 5′UTR of CDKAL1-v1 is ∼100 bases longer than that of CDKAL1, and contains three upstream ATG codons (Fig. 3A). Given the potential translational suppression by the upstream ORFs and the truncation of tRNA-modifying domains in CDKAL1-v1, we suspected that CDKAL1-v1 may not be translated and may instead regulate protein levels as a non-coding transcript. To confirm this speculation, we conjugated the 5′UTR or putative ORF of CDKAL1-v1 to Luciferase and examined the expression level (Fig. 3B and C). While the 5′UTR of full-length CDKAL1 did not affect Luciferase activity, the 5′UTR of CDKAL1-v1 significantly attenuated Luciferase activity (48% compared with Control 1). The putative ORF of CDKAL1-v1 markedly reduced Luciferase activity (4.8% compared with Control 1). Combining both the 5′UTR and the putative ORF of CDKAL1 almost eliminated Luciferase activity (2.8% compared with Control 1). In contrast to the luciferase activities, the mRNA levels of the CDKAL1- and CDKAL1-v1-derived luciferase reporters were not significantly different from the mRNA level of the control reporter. These results suggest that CDKAL1-v1 is not actively translated into functional protein. Furthermore, we examined the expression of CDKAL1 using a specific antibody targeting the N-terminus region of CDKAL1, which is encoded by both CDKAL1 and CDKAL1-v1. The antibody recognized the 61 kDa full-length CDKAL1 protein in extracts of HEK cells, which disappeared following the knockdown of both CDKAL1 and CDKAL1-v1 (Fig. 3D and E). In contrast, there was no detectable band of ∼11 kDa, which is thought to be the molecular weight of translated CDKAL1-v1, in extracts from either the control or CDKAL1/CDKAL1-v1 knockdown cells. These results suggest that CDKAL1-v1 regulates CDKAL1 level through a coding-independent mechanism. Because CDKAL1-v1 is a short splicing variant that contains a premature stop codon, there is a concern that CDKAL1-v1 may be subjected to nonsense-mediated mRNA decay (NMD) (14). CDKAL1-v1 levels were examined in cells treated with siRNA targeting UFP1, which is a key component of the cellular NMD system (Fig. 3F). However, the CDKAL1-v1 level in UFP1-silenced cells was not different from that in control-siRNA treated cells. Furthermore, we performed in situ hybridization in HEK cells to examine the cellular location of CDKAL1-v1 transcripts. While the probe targeting both CDKAL1 and CDKAL1-v1 showed the most intense signals in the cell bodies, the CDKAL1-v1-specific probe also resulted in moderate signals at the same locations (Fig. 3G), suggesting that CDKAL1-v1 is a functional non-coding transcript in human cells.

Figure 3.

CDKAL1-v1 is a non-coding transcript. (A) Alignment of 5′UTR of CDKAL1 and CDKAL1-v1. ATG codons in upstream ORF are underlined. (B) Constructs for the luciferase-based reporter assay with which to examine the 5′UTR and ORF of CDKAL1-v1. (C) Suppression of luciferase activity by the 5′UTR and ORF of CDKAL1-v1 is translation-dependent but not transcription-dependent. (D) Knockdown of both CDKAL1 and CDKAL1-v1 in HEK293 cells by a siRNA targeting both transcripts. n = 4. (E) Western blotting to examine the translated product of CDKAL1-v1. The arrow indicates 61 kDa, full length, CDKAL1. (F) Knockdown of UPF1 in fibroblast cells had no effect on CDKAL1 or CDKAL1-v1 level. n = 4. (G) Detection of cellular localization of CDKAL1-v1 by in situ hybridization. The upper illustration shows the regions to which the probes hybridize. The lower images show the cellular localization of probes (Digoxigenin). The nucleus is detected by DAPI staining. Bars = 100 μm. Error bars represent mean ± SEM.

Figure 3.

CDKAL1-v1 is a non-coding transcript. (A) Alignment of 5′UTR of CDKAL1 and CDKAL1-v1. ATG codons in upstream ORF are underlined. (B) Constructs for the luciferase-based reporter assay with which to examine the 5′UTR and ORF of CDKAL1-v1. (C) Suppression of luciferase activity by the 5′UTR and ORF of CDKAL1-v1 is translation-dependent but not transcription-dependent. (D) Knockdown of both CDKAL1 and CDKAL1-v1 in HEK293 cells by a siRNA targeting both transcripts. n = 4. (E) Western blotting to examine the translated product of CDKAL1-v1. The arrow indicates 61 kDa, full length, CDKAL1. (F) Knockdown of UPF1 in fibroblast cells had no effect on CDKAL1 or CDKAL1-v1 level. n = 4. (G) Detection of cellular localization of CDKAL1-v1 by in situ hybridization. The upper illustration shows the regions to which the probes hybridize. The lower images show the cellular localization of probes (Digoxigenin). The nucleus is detected by DAPI staining. Bars = 100 μm. Error bars represent mean ± SEM.

Regulation of cellular Cdkal1 level by a CDKAL1-v1-targeting miRNA

Because CDKAL1-v1 was positively correlated with the CDKAL1 protein level as well as the CDKAL1 mRNA level, it is unlikely that CDKAL1-v1 acts as an anti-sense non-coding RNA for CDKAL1. Interestingly, recent studies have revealed a unique regulatory mechanism by which some transcripts could regulate the cognate genes through a coding-independent mechanism (15–18). In this model, non-coding RNA indirectly regulates its cognate gene by the competitive binding of common miRNAs at the 3′UTR region. We speculated that CDKAL1-v1 might regulate CDKAL1 levels through the same mechanism. To investigate whether the 3′UTR of CDKAL1-v1 is capable of regulating CDKAL1 levels, the 3′UTR region of CDKAL1-v1 or the entire CDKAL1-v1 was transfected into primary human fibroblast cells that carried the risk alleles and had low CDKAL1 protein levels as well as CDKAL1-v1 level (HE37 in Supplementary Material, Fig. S4). Expression of the 3′UTR of CDKAL1-v1 or the entire CDKAL1-v1 transcript induced expression of the CDKAL1 protein (Fig. 4A). The 3′UTR of CDKAL1-v1 also increased the CDKAL1 mRNA level (121% compared with control, Fig. 4B). These results suggest that the 3′UTR of CDKAL1-v1 is capable of controlling CDKAL1 levels.

Figure 4.

Control of CDKAL1 level by CDKAL1-v1-targeting miRNA. (A) Increase in the CDKAL1 level following overexpression of full-length CDKAL1-v1 and 3′UTR of CDKAL1-v1. The arrow indicates bands corresponding to CDKAL1. (B) Increase in the CDKAL1 level following overexpression of the 3′UTR of CDKAL1-v1. n = 4. (C) Binding of has-miR-494 to both CDKAL1 and CDKAL1-v1. The solid lines represent canonical pairing. The dotted lines represent non-canonical pairing. The underlined UUU were mutated to AAA to make miR-494-insensitive constructs. (D and E) miR-494 suppressed the luciferase gene-carrying wild-type (WT) 3′UTR of CDKAL1-v1 (D) or CDKAL1 (E) but not the luciferase gene-carrying miR-494-insensitive (Mutant) 3′UTR of CDKAL1-v1 (D) or CDKAL1 (E). n = 4. (F) Reduced Cdkal1 protein level in response to miR-494. (G) Reduced CDKAL1-v1 and CDKAL1 levels in response to miR494. (H) Derepression of CDKAL1-v1 and CDKAL1 by a miR-494 inhibitor (494i). n = 4. (I) Increase in the CDKAL1 protein level in response to a miR-494 inhibitor (494i). The arrow indicates bands corresponding to CDKAL1. Error bars represent mean ± SEM.

Figure 4.

Control of CDKAL1 level by CDKAL1-v1-targeting miRNA. (A) Increase in the CDKAL1 level following overexpression of full-length CDKAL1-v1 and 3′UTR of CDKAL1-v1. The arrow indicates bands corresponding to CDKAL1. (B) Increase in the CDKAL1 level following overexpression of the 3′UTR of CDKAL1-v1. n = 4. (C) Binding of has-miR-494 to both CDKAL1 and CDKAL1-v1. The solid lines represent canonical pairing. The dotted lines represent non-canonical pairing. The underlined UUU were mutated to AAA to make miR-494-insensitive constructs. (D and E) miR-494 suppressed the luciferase gene-carrying wild-type (WT) 3′UTR of CDKAL1-v1 (D) or CDKAL1 (E) but not the luciferase gene-carrying miR-494-insensitive (Mutant) 3′UTR of CDKAL1-v1 (D) or CDKAL1 (E). n = 4. (F) Reduced Cdkal1 protein level in response to miR-494. (G) Reduced CDKAL1-v1 and CDKAL1 levels in response to miR494. (H) Derepression of CDKAL1-v1 and CDKAL1 by a miR-494 inhibitor (494i). n = 4. (I) Increase in the CDKAL1 protein level in response to a miR-494 inhibitor (494i). The arrow indicates bands corresponding to CDKAL1. Error bars represent mean ± SEM.

Next, we searched potential CDKAL1-v1- and CDKAL1-targeting miRNAs that are also expressed in human pancreatic islets (19,20). We found miR-494 as a potential candidate from the miRNA database (Fig. 4C). To validate our finding, we transfected miR-494 and 3′UTR of CDKAL1-v1 or CDKAL1 fused with Luciferase and monitored Luciferase activity as a readout for miR-494-mediated suppression (Fig. 4D and E). Expression of miR-494 significantly suppressed Luciferase activity. Furthermore, miR-494 no longer suppressed Luciferase activity when the seed sequences of the miR-494-targeting site in the 3′UTR of CDKAL1-v1 or CDKAL1 were mutated (Fig. 4D and E). We also examined the direct effect of miR-494 on endogenous CDKAL1-v1 or CDKAL1 levels. In HEK293 cells, overexpression of miR-494 markedly reduced CDKAL1 protein levels as well as CDKAL1-v1 and CDKAL1 mRNA levels (Fig. 4F and G), suggesting that CDKAL1-v1 and CDKAL1 are common targets for miR-494. Furthermore, introducing miR-494 inhibitor induced both CDKAL1-v1 and CDKAL1 levels (Fig. 4H). Accordingly, the inhibitor of miR-494 enhanced CDKAL1 protein levels (Fig. 4I). These results indicate that CDKAL1-v1 and CDKAL1 are subject to miR-494-mediated post-transcriptional regulation. Thus, a decrease in CDKAL1-v1 would conversely result in an increase in miR-494-mediated suppression of CDKAL1, as observed in carriers of risk alleles.

A T2D-associated SNP is directly involved in alternative splicing of CDKAL1-v1

Finally, we attempted to answer the question of why a single-nucleotide substitution in intron 5 of CDKAL1 results in a drastic decrease in the expression level of CDKAL1-v1. We hypothesized that the risk SNPs of CDKAL1 may have an inhibitory effect on the alternative splicing. To test this hypothesis, we utilized an AMO that specifically binds to the 5′ end of the U1 small nuclear ribonucleic acid (snRNA) (U1 AMO) to inhibit splicing (21,22). In cells carrying non-risk SNPs, inhibition of splicing is predicted to result in a decrease of CDKAL1. However, in cells carrying risk SNPs, inhibition of splicing should not affect the CDKAL1-v1 level, because the generation of CDKAL1-v1 by splicing has already been impaired. As expected, in HEK293 cells that carry non-risk SNPs, the U1 AMO significantly decreased CDKAL1-v1 (Fig. 5A). In contrast, in HeLa cells that are homozygous for risk SNPs, the U1 AMO had no effect on CDKAL1-v1 level (Fig. 5B). Moreover, the U1 AMO significantly decreased CDKAL1 levels in both HEK293 cells and HeLa cells. The same results were also confirmed in primary human fibroblast cell lines (Supplementary Material, Fig. S5). These results suggest that the risk SNPs are specifically associated with impaired alternative splicing to generate CDKAL1-v1.

Figure 5.

Modulation of splicing by SNPs in CDKAL1. (A) Inhibition of U1 snRNA by U1 AMO suppressed both the CDKAL1 and CDKAL1-v1 level in HEK293 cells carrying A/A alleles (non-risk/risk) for rs10946398. (B) U1 AMO suppressed CDKAL1 levels but not CDKAL1-v1 levels in HeLa cells carrying C/C alleles (risk/risk). (C) Genomic region (nucleotides in red) targeted by TALEN constructs. The underlined C is the SNP site for rs10946398. (D) Genomic sequences surround the SNP site in isolated HeLa cell lines transfected with TALEN. Dashed lines represent deleted nucleotides. Dashed rectangles show the position of the SNP site. (E and F) Expression level of CDKAL1-v1 (E) and CDKAL1 (F) in cell lines with an edited SNP site. n = 4. Error bars represent mean ± SEM.

Figure 5.

Modulation of splicing by SNPs in CDKAL1. (A) Inhibition of U1 snRNA by U1 AMO suppressed both the CDKAL1 and CDKAL1-v1 level in HEK293 cells carrying A/A alleles (non-risk/risk) for rs10946398. (B) U1 AMO suppressed CDKAL1 levels but not CDKAL1-v1 levels in HeLa cells carrying C/C alleles (risk/risk). (C) Genomic region (nucleotides in red) targeted by TALEN constructs. The underlined C is the SNP site for rs10946398. (D) Genomic sequences surround the SNP site in isolated HeLa cell lines transfected with TALEN. Dashed lines represent deleted nucleotides. Dashed rectangles show the position of the SNP site. (E and F) Expression level of CDKAL1-v1 (E) and CDKAL1 (F) in cell lines with an edited SNP site. n = 4. Error bars represent mean ± SEM.

To further investigate the direct involvement of SNPs in the regulation of CDKAL1-v1, we utilized transcription activator-like effector nucleases (TALEN) to directly edit the SNP site in CDKAL1. Specific TALEN constructs targeting the SNP (rs10946398) site were transfected in combination with template DNA into HeLa cells (Fig. 5C). Although there were no HeLa cells with single-nucleotide substitutions, we established five cell lines containing 3–19 nucleotide deletions around the SNP site (Fig. 5D). Subsequent analysis of CDKAL1-v1 and CDKAL1 levels in these cell lines revealed that deletion of nucleotides in the 3′ direction from the SNP site (Cell line #11 and #38) enhanced the expression of both CDKAL1-v1 and CDKAL1, while deletion of nucleotides in the 5′ direction from the SNP site (Cell line #74) slightly decreased the expression (Fig. 5E and F). Furthermore, the expression levels of CDKAL1 and CDKAL1-v1 were not altered in cell line containing three nucleotide deletions but without alteration of SNP site (Cell line #23). Taken together, these results suggest that SNPs in CDKAL1 are critical for alternative splicing to generate CDKAL1-v1.

DISCUSSION

In the present study, we discovered a novel molecular mechanism underlying the development of T2D in individuals carrying risk SNPs in CDKAL1. A human-specific splicing variant of CDKAL1, CDKAL1-v1, was significantly correlated with T2D-associated risk alleles of CDKAL1. Furthermore, the CDKAL1-v1 level was directly correlated with CDKAL1-mediated ms2-modification and insulin secretion. Functional analysis revealed that CDKAL1-v1 is a non-coding transcript and that it indirectly regulates the CDKAL1 level by acting as a decoy for a CDKAL1-targeting miRNA. In addition, we also demonstrated that the nucleotides surrounding risk SNPs were critical for alternative splicing to generate CDKAL1-v1. Our results suggest that in individuals carrying the risk SNPs in CDKAL1, impaired alternative splicing causes a dramatic decrease in the CDKAL1-v1 level, which then results in a shift of miRNA targeting from CDKAL1-v1 toward CDKAL1. Subsequently, miRNA-mediated suppression of cellular CDKAL1 levels causes a decrease in ms2-modification in tRNALys(UUU), which ultimately contributes to the dysfunction of pancreatic β-cells as revealed in Cdkal1 knockout mice (7).

Twelve SNPs in CDKAL1, including the three SNPs examined in the present study, have been associated with T2D [rs4712523 (23); rs4712524 (13); rs10440833 (24); rs6931514 (25); rs9465871 (26); rs9295474 (27); rs9356744, rs7773318, rs9465994 (28)]. To date, there is no general consensus about the causal SNP in these 12 SNPs. In addition, clear evidence to show the correlation between these SNPs and CDKAL1 level has been insufficient. Interestingly, all SNPs except two SNPs (rs7773318 and rs9465994) locate in the intron 5, suggesting genetic variations in this particular intron might have profound effect on CDKAL1 level. Previous studies have shown that the nucleotides that surround the SNPs occasionally act as cis- or trans-regulatory elements and regulate the expression of genes in the vicinity and at a distance (29,30). A conserved element near risk SNPs in Cdkal1 was shown to affect the expression of the nearby gene Sox4 in mice and zebrafish (31). Because Sox4 has been implicated in the development of the pancreas, there is a possibility that the risk SNPs in CDKAL1 may contribute to T2D through regulation of the expression of SOX4. However, the SOX4 expression level in carriers of risk alleles of CDKAL1 was not significantly different from that in carriers of non-risk alleles (Supplementary Material, Fig. S6). Given the extent of the reduction of the CDKAL1-v1 level in the carriers of risk alleles and the significant association of the CDKAL1-v1 level with insulin secretion, it is most likely that the risk SNPs in intron 5 are the critical elements for controlling expression of CDKAL1-v1 and ultimately contribute to the development of T2D.

The T2D-associated SNPs in CDKAL1 are not located at critical sites for splicing such as 3′- and 5′-splicing sites, polypyrimidine tract or branch sites. Nevertheless, several pieces of evidence have shown that such intronic SNPs might be capable of affecting splicing patterns (32–34). The SNPs in TCF7L2 have been extensively associated with T2D (35–37). These T2D-associated SNPs have also been found to be associated with differential splicing patterns (32,33), which might affect the function of TCF7L2 and lead to the development of T2D. Similar to these results, we demonstrated that the inhibition of splicing by the U1 AMO specifically inhibits the production of CDKAL1-v1 in cells carrying non-risk alleles but not in cells carrying homozygous risk alleles. These results not only support our hypothesis that the SNPs in CDKAL1 are actively involved in splicing, but also provide the molecular basis that these SNPs may affect splicing through interactions with U1 small nuclear RNP (snRNP). Recent studies have found that U1 snRNP is critical for the determination of mRNA length and the regulation of isoform expression in addition to its classic role as a member of a spliceosome, through direct interaction with nucleotides in the intron (38). Interestingly, such interaction does not necessarily require direct base pairing (39,40). We thus propose that the risk SNPs in CDKAL1 may impair the interaction between the U1 snRNP and the nucleotides near SNPs, which causes aberrant processing of the nascent transcription of CDKAL1 and results in a decrease in CDKAL1-v1. Moreover, given the similar effects of SNPs in CDKAL1 and TCF7L2, our results thus suggest the broad relevance of regulation of splicing pattern in the pathogenesis of T2D-associated SNPs.

In the present study, we utilized the TALEN method for direct editing of the genome and showed that SNP sites play an active role in the regulation of gene expression. We successfully obtained cell lines with deletion around the SNP site and demonstrated that the CDKAL1-v1 level was altered in these cell lines. Interestingly, regulation of CDKAL1-v1 seems to depend on the degree and direction of the deletion around the SNP site. It is currently unclear whether the nucleotide in the SNP site or the surrounding nucleotides including SNP site are responsible for the subsequent effects on gene expression. Furthermore, the incomplete recovery of the CDKAL-v1 expression level after alteration of the risk SNP site (rs10946398) suggest that other risk SNP sites might also contribute the suppression of CDKAL1-v1. Future studies utilizing a more efficient genome-editing tool such as CRISPR (41,42) may achieve single-nucleotide substitution in all risk SNP sites and answer these questions. Nevertheless, a previous study has found that the intronic SNPs might cause different structural folds in the nascent transcripts (43). Therefore, it is likely that the risk SNPs in CDKAL1 may affect the secondary structure of pre-mRNA by changing the chemical interaction with surrounding nucleotides so that splicing is not be properly achieved.

A unique finding in the present study is that CDKAL1-v1 is a non-coding transcript and acts as a decoy for CDKAL1-targeting miRNA to modulate the cellular Cdkal1 level. A similar regulatory mechanism has been revealed in the regulation of cancer-related genes, such as PTEN and KRAS, by the cognate pseudogenes PTENP1 and KRAS1P (15). The so-called competing endogenous RNA (ceRNA) mechanism expands the classic role of non-coding RNA as a suppressor for its anti-sense target gene, and provides a novel function for non-coding RNA as an active regulator of its cognate gene through its abilities to compete for miRNA binding at the 3′UTR (17). The observation of the miRNA-mediated correlation of CDKAL1-v1 with CDKAL1 in human tissues and various cell lines nicely fits CDKAL1-v1 into the ceRNA theory. In addition to CDKAL1-v1, previous studies have also found that some splicing variants of TCF7L2 are correlated with full-length TCF7L2 in the same direction (33). These results suggest that the ceRNA theory may apply to a variety of disease including cancer and T2D. Nevertheless, further study using primary human pancreatic islets is needed to elucidate the direct contribution of CDKAL1-v1 to T2D.

In the present study, we identified miR-494 as a candidate miRNA that targets both CDKAL1 and CDKAL1-v1. Previous studies showed that miR-494 was expressed in various human tissues with the highest expression in pancreatic β-cells (19,20). Interestingly, overexpression of miR-494 impaired mitochondrial function by down-regulating the expression of some genes involved in mitochondrial transcription as well as the respiratory chain (44,45). Since mitochondrial function is crucial for insulin secretion from pancreatic β-cells, it is conceivable that the decrease of CDKAL1-v1 in risk SNP carriers would enhance miR-494-mediated suppression of mitochondrial functions as well as CDKAL1-dependent tRNA modification, ultimately leading to the dysfunction of β-cells.

A limitation of the functional study of T2D-associated intronic SNPs is that most of the intronic SNPs are not conserved in mouse and other animal models. The risk SNPs in CDKAL1 are not conserved in the mouse genome, nor is murine CDKAL1-v1 found in the current databases. However, we found some truncated transcripts in murine mRNA NCBI database (AK169761 and AK158546) that contain an upstream ORF and share common miRNA targeting sequences with Cdkal1 in 3′UTR. Thus, it is conceivable that the ceRNA-mediated regulatory mechanism might be evolutionarily conserved, while the intermediate miRNA might be different between species.

In conclusion, T2D-associated SNPs in CDKAL1 was associated with marked decrease of the CDKAL1-v1 expression level, possibly by influencing proper splicing. The reduction in the level of CDKAL1-v1 causes a decrease in the CDKAL1 level by enhancing miRNA-mediated suppression of CDKAL1, which ultimately contributes to the development of T2D. Our results present a clear correlation between T2D-associated SNPs and CDKAL1 gene expression for the first time, and reveal a novel molecular mechanism underlying the development of T2D. Moreover, our results suggest that restoration of the homeostasis between miRNA and its cognate transcripts may be a new approach for treatment of T2D.

MATERIALS AND METHODS

Participant recruitment

Blood samples were obtained from non-diabetic volunteers at Kumamoto University. Informed consent was obtained before collecting blood. The Ethics Committee of Kumamoto University approved the experiments regarding the analysis of human genetic materials and the collection of human blood (approval ID: Genome 159).

Oral glucose tolerance test

Individuals were fasted beginning at 21:00 the night before the test day. These individuals were then given a 75 g glucose load (Ajinomoto) anytime between 9:00 and 11:00. Peripheral blood samples were obtained when the individuals were fasting and 30 min after the glucose load. Plasma glucose and serum insulin concentrations were measured by enzyme-linked immunosorbent assay (SRL Inc.). Corrected insulin response (CIR30) was calculated as I30/[G30 × (G30–70)] as described previously (1).

Genotyping of SNPs

Genomic DNA was purified from 200 μl of peripheral human blood using the QIAamp DNA Blood Mini Kit (Qiagen) and adjusted to 10 ng/μl with distilled water. The SNPs (rs10946398, rs7756992 and rs7754840) in CDKAL1 and the SNP in KCNQ1 (rs2237892) were examined using the TaqMan SNP Genotyping Assay kit (Life Technologies).

Isolation of RNA and gene expression assay

Total RNA fractions of cultured cell lines were isolated using TRIzol reagent (Life Technologies) according to the manufacturer's protocol. For isolation of total RNA from peripheral blood, 1.5 ml fresh peripheral blood was immediately mixed with the erythrocyte lysis buffer provided in the QIAamp RNA Blood Mini Kit (Qiagen). Total RNA fractions of leukocytes were isolated according to the manufacturer's protocol. Total RNA of normal human tissues was derived from commercial available Human Total RNA Master Panel II (TAKARA). A PrimerScript RT reagent kit (TAKARA) was used to generate cDNA. Quantitative real-time PCRs were performed using either a Taqman Gene Expression Kit (Applied Biosystems) or SYBR Premix Ex Taq (TAKARA). The results were normalized to the level of GAPDH or 18S rRNA. Information regarding the probes and primers used in this study is described in Supplementary Material, Table S1.

Measurement of ms2-modification

Measurement of ms2-modification in total RNA isolated from peripheral blood specimens or cultured cell lines was performed using the qPCR-MtR method as described previously (8). Total RNA was adjusted to 50 ng/μl in RNase-free water. Two microliters (100 ng) of RNA were digested by five units of DNase I (Roche Applied Science) in a 20 μl reaction at 37°C for 20 min. DNase I was inactivated by heat-inactivation at 75°C for 10 min. After the DNase treatment, 2.5 μl of digested RNA was subjected to reverse transcription (Transcriptor, Roche Applied Science) using reverse primer r1 (CCTGGACCCTCAGATTAAAA) or reverse primer r2 (GAACAGGGACTTGAACCCTG). Two microliters of synthesized complementary DNA (cDNA), forward primer (GTCGGTAGAGCATCAGACTT) and reverse primer r1 were subjected to quantitative PCR using the SYBR Premix Ex Taq Kit (Takara) and the ABI PRISM 7300 Real-time PCR System (Life Technologies) according to the manufacturer's protocol.

Relative modification was calculated as described previously (8). Briefly, the threshold cycle number (referred to as CTr1 or CTr2) was obtained from quantitative PCR using a cDNA template that was generated by primer r1 or primer r2. The difference between CTr1 and CTr2 (hereafter dCTr2r1) was calculated by subtracting CTr1 from CTr2. The dCTr2r1 number for individual samples was used to compare the relative modification rate between samples.

MicroRNA selection

To search the common miRNA that can target both 3′UTR of CDKAL1 and 3′ UTR of CDKAL1-v1, an online search program (www.microrna.org) was utilized with the most stringent criteria (conserved and with good mirSVR scores). The candidate miRNAs were further validated using another online search program (www.targetscan.org). The common candidate miRNA (miR-494) was selected for experiments.

Cell culture

HeLa cells and HEK293 cells were cultured in Dulbecco's modified Eagle's medium (Life Technologies) containing 10% fetal bovine serum. QGP-1 cells were obtained from the Japanese Collection of Research Bioresources Cell Bank and cultured according to a previous report (46). Human embryonic primary fibroblast cell lines were provided by the RIKEN BRC through the National Bio-Resource Project of the MEXT, Japan. The fibroblast cells were cultured in minimal essential medium containing 10% fetal bovine serum, 1 mM sodium pyruvate, 0.2 mM serine and 0.2 mM aspartate. Plasmids and siRNAs were transfected using Lipofectamine 2000 (Life Technologies) according to the manufacturer's protocol.

For western blotting, cells were lysed in lysis buffer containing 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, pH 7.2, 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid, 1% NP-40, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate, protease inhibitor cocktail (Roche). The protein concentration was adjusted to 1 mg/ml by bicinchoninic acid reagent (Piece). CDKAL1 was detected by homemade antibody targeting the N-terminus of human CDKAL1.

For in situ hybridization, DNA regions of putative ORF (208–542 of Ak310219), regions flanking ORF and the 3′UTR (428–943) and the regions of 3′UTR of CDKAL1-v1 (555–952) were cloned into pCRII-TOPO. A sense probe and an antisense probe were transcribed using either the T7 promoter or the Sp6 promoter using PCR DIG Labeling Mix (Roche). Hybridization was performed according to the manufacturer's protocol (Roche). Digoxigenin (DIG) was detected by anti-digoxigenin antibody (Roche) followed by Alexa488 anti-mouse secondary antibody. Images were acquired using Fluoview 300 (Olympus).

Plasmids and siRNA

All siRNAs were purchased from Life Technologies. For knockdown of CDKAL1, an siRNA targeting exon 16 in CDKAL1 was selected. For knockdown of CDKAL1-v1, siRNAs targeting the 3′UTR of CDKAL1-v1 were selected. For knockdown of both CDKAL1 and CDKAL1-v1, an siRNA targeting exon 4 was selected. A morpholino antisense oligo targeting U1 snRNA (U1 AMO) was designed as described previously (19) and purchased from Gene Tool, LLC. Human miR-494 (miRIDIAN Mimic) was purchased from Thermo Scientific. Inhibitor for miR-494 (mirVana miRNA inhibitor) was purchased from Life Technologies. The information regarding siRNAs or U1 AMO is provided in Supplementary Material, Table S1.

Luciferase assay

For the Luciferase-based assay, the 5′UTR or ORF regions of CDKAL1-v1 and CDKAL1 were cloned into multi-cloning sites in pGL4.14 (Promega). For the 3′UTR assay, 3′UTR of CDKAL1-v1 or CDKAL1 was cloned into the 3′ downstream of Luciferase in pGL4.14. Luciferase reporter vectors (100 ng) and the Renilla luciferase reporter vector (10 ng) were co-transfected into HEK293 cells for 16 h. Relative Luciferase activity to Renilla Luciferase activity was examined using the dual Luciferase reporter assay system (Promega). For examining the mRNA level of Luciferase-conjugated CDKAL1 or CDKAL1-v1 transcripts, report vectors were transfected into HEK cells, and total RNAs were purified by TRIzol (Invitrogen). To eliminate contamination with the plasmid vector, 2 μg total RNA was digested with DNase I (Roche) for 30 min and then subjected to quantitative PCR. The relative mRNA levels of reporters were examined by normalizing the level of Luciferase to that of Renilla Luciferase.

Genome editing by TALEN

Specific TALEN constructs for editing nucleotides in SNP sites (rs10946398) were designed and purchased from Cellectis Bioresearch. A DNA template containing 1.5 kb genomic regions from the 5′ direction and the 3′ direction of the SNP site (3 kb in total) was amplified from human DNA carrying risk alleles (C/C) or non-risk alleles (A/A), and cloned into pCRII-TOPO vector (Life Technologies). HeLa cells were seeded in 35 mm dishes at a density of 3 × 105 cells/ml. The next day, 2 μg of each TALEN construct and 4 µg of DNA template were co-transfected in HeLa cells using Lipofectamine 2000. Four days after transfection, HeLa cells were trypsinized and diluted to 10 cells/ml. One hundred microliters of cell suspension was transferred to each well of the 96-well plate. Colony grown from a single cell was isolated and expanded. To confirm genomic editing, the genomic region containing the SNP site was amplified by PCR followed by DNA sequencing. The expression levels of CDKAL1 and CDKAL1-v1 were examined as described above.

Statistics

Statistical analyses were performed using Prism 6 Software (GraphPad Software). Student's t-test was used to test the difference between two groups. Analysis of variance (one-way ANOVA) was used to test the difference among multiple groups followed by a post hoc examination of the P-value between two groups. A two-tailed P-value of 0.05 was considered statistically significant.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported by a Grant-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, by the Japan Society for the Promotion of Science (JSPS) through its “Funding Program for Next Generation World-Leading Researchers”, and by the Takeda Science Foundation.

ACKNOWLEDGEMENTS

We thank Y. Ueda and N. Maeda for technical assistance.

Conflict of Interest Statement. None declared.

REFERENCES

1
Steinthorsdottir
V.
Thorleifsson
G.
Reynisdottir
I.
Benediktsson
R.
Jonsdottir
T.
Walters
G.B.
Styrkarsdottir
U.
Gretarsdottir
S.
Emilsson
V.
Ghosh
S.
, et al.  . 
A variant in CDKAL1 influences insulin response and risk of type 2 diabetes
Nat. Genet.
 , 
2007
, vol. 
39
 (pg. 
770
-
775
)
2
Saxena
R.
Voight
B.F.
Lyssenko
V.
Burtt
N.P.
de Bakker
P.I.
Chen
H.
Roix
J.J.
Kathiresan
S.
Hirschhorn
J.N.
Daly
M.J.
, et al.  . 
Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels
Science
 , 
2007
, vol. 
316
 (pg. 
1331
-
1336
)
3
Scott
L.J.
Mohlke
K.L.
Bonnycastle
L.L.
Willer
C.J.
Li
Y.
Duren
W.L.
Erdos
M.R.
Stringham
H.M.
Chines
P.S.
Jackson
A.U.
, et al.  . 
A genome-wide association study of type 2 diabetes in Finns detects multiple susceptibility variants
Science
 , 
2007
, vol. 
316
 (pg. 
1341
-
1345
)
4
Zeggini
E.
Weedon
M.N.
Lindgren
C.M.
Frayling
T.M.
Elliott
K.S.
Lango
H.
Timpson
N.J.
Perry
J.R.
Rayner
N.W.
Freathy
R.M.
, et al.  . 
Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes
Science
 , 
2007
, vol. 
316
 (pg. 
1336
-
1341
)
5
Dehwah
M.A.
Wang
M.
Huang
Q.Y.
CDKAL1 and type 2 diabetes: a global meta-analysis
Genet. Mol. Res.
 , 
2010
, vol. 
9
 (pg. 
1109
-
1120
)
6
Arragain
S.
Handelman
S.K.
Forouhar
F.
Wei
F.Y.
Tomizawa
K.
Hunt
J.F.
Douki
T.
Fontecave
M.
Mulliez
E.
Atta
M.
Identification of eukaryotic and prokaryotic methylthiotransferase for biosynthesis of 2-methylthio-N6-threonylcarbamoyladenosine in tRNA
J. Biol. Chem.
 , 
2010
, vol. 
285
 (pg. 
28425
-
28433
)
7
Wei
F.Y.
Suzuki
T.
Watanabe
S.
Kimura
S.
Kaitsuka
T.
Fujimura
A.
Matsui
H.
Atta
M.
Michiue
H.
Fontecave
M.
, et al.  . 
Deficit of tRNALys(UUU) modification by Cdkal1 causes the development of type 2 diabetes in mice
J. Clin. Invest.
 , 
2011
, vol. 
121
 (pg. 
3598
-
3608
)
8
Xie
P.
Wei
F.Y.
Hirata
Y.
Kaitsuka
T.
Suzuki
T.
Suzuki
T.
Tomizawa
K.
Quantitative PCR measurement of tRNA 2-methylthio modification for assessing Type 2 diabetes risk
Clin. Chem.
 , 
2013
, vol. 
59
 (pg. 
1604
-
1612
)
9
McCarthy
M.I.
Genomics, type 2 diabetes, and obesity
N. Engl. J. Med.
 , 
2010
, vol. 
363
 (pg. 
2339
-
2350
)
10
Chistiakov
D.A.
Potapov
V.A.
Smetanina
S.A.
Bel'chikova
L.N.
Suplotova
L.A.
Nosikov
V.V.
The carriage of risk variants of CDKAL1 impairs beta-cell function in both diabetic and non-diabetic patients and reduces response to non-sulfonylurea and sulfonylurea agonists of the pancreatic KATP channel
Acta Diabetol.
 , 
2011
, vol. 
48
 (pg. 
227
-
235
)
11
Rong
R.
Hanson
R.L.
Ortiz
D.
Wiedrich
C.
Kobes
S.
Knowler
W.C.
Bogardus
C.
Baier
L.J.
Association analysis of variation in/near FTO, CDKAL1, SLC30A8, HHEX, EXT2, IGF2BP2, LOC387761, and CDKN2B with type 2 diabetes and related quantitative traits in Pima Indians
Diabetes
 , 
2009
, vol. 
58
 (pg. 
478
-
488
)
12
Yasuda
K.
Miyake
K.
Horikawa
Y.
Hara
K.
Osawa
H.
Furuta
H.
Hirota
Y.
Mori
H.
Jonsson
A.
Sato
Y.
, et al.  . 
Variants in KCNQ1 are associated with susceptibility to type 2 diabetes mellitus
Nat. Genet.
 , 
2008
, vol. 
40
 (pg. 
1092
-
1097
)
13
Unoki
H.
Takahashi
A.
Kawaguchi
T.
Hara
K.
Horikoshi
M.
Andersen
G.
Ng
D.P.
Holmkvist
J.
Borch-Johnsen
K.
Jørgensen
T.
, et al.  . 
SNPs in KCNQ1 are associated with susceptibility to type 2 diabetes in East Asian and European populations
Nat. Genet.
 , 
2008
, vol. 
40
 (pg. 
1098
-
1102
)
14
Isken
O.
Kim
Y.K.
Hosada
N.
Mayeur
G.L.
Hershey
J.W.
Maquat
L.E.
Upf1 phosphorylation triggers translational repression during nonsense-mediated mRNA decay
Cell
 , 
2008
, vol. 
133
 (pg. 
314
-
327
)
15
Poliseno
L.
Salmena
L.
Zhang
J.
Carver
B.
Haveman
W.J.
Pandolfi
P.P.
A coding-independent function of gene and pseudogene mRNAs regulates tumour biology
Nature
 , 
2010
, vol. 
465
 (pg. 
1033
-
1038
)
16
Tay
Y.
Kats
L.
Salmena
L.
Weiss
D.
Tan
S.M.
Ala
U.
Karreth
F.
Poliseno
L.
Provero
P.
Di Cunto
F.
, et al.  . 
Coding-independent regulation of the tumor suppressor PTEN by competing endogenous mRNAs
Cell
 , 
2011
, vol. 
147
 (pg. 
344
-
357
)
17
Salmena
L.
Poliseno
L.
Tay
Y.
Kats
L.
Pandolfi
P.P.
A ceRNA hypothesis: the Rosetta Stone of a hidden RNA language?
Cell
 , 
2011
, vol. 
146
 (pg. 
353
-
358
)
18
Hirotsune
S.
Yoshida
N.
Chen
A.
Garrett
L.
Sugiyama
F.
Takahashi
S.
Yagami
K.
Wynshaw-Boris
A.
Yoshiki
A.
An expressed pseudogene regulates the messenger-RNA stability of its homologous coding gene
Nature
 , 
2003
, vol. 
423
 (pg. 
91
-
96
)
19
Klein
D.
Misawa
R.
Bravo-Egana
V.
Vargas
N.
Rosero
S.
Piroso
J.
Ichii
H.
Umland
O.
Zhijie
J.
Tsinoremas
N.
, et al.  . 
MicroRNA expression in alpha and beta cells of human pancreatic islets
PloS One
 , 
2013
, vol. 
8
 pg. 
e55064
 
20
van de Bunt
M.
Gaulton
K.J.
Parts
L.
Moran
I.
Johnson
P.R.
Lindgren
C.M.
Ferrer
J.
Gloyn
A.L.
McCarthy
M.I.
The miRNA profile of human pancreatic islets and beta-cells and relationship to type 2 diabetes pathogenesis
PLoS One
 , 
2013
, vol. 
8
 pg. 
e 55272
 
21
Kaida
D.
Berg
M.G.
Younis
I.
Kasim
M.
Singh
L.N.
Wan
L.
Dreyfuss
G.
U1 snRNP protects pre-mRNAs from premature cleavage and polyadenylation
Nature
 , 
2010
, vol. 
468
 (pg. 
664
-
668
)
22
Berg
M.G.
Singh
L.N.
Younis
I.
Liu
Q.
Pinto
A.M.
Kaida
D.
Zhang
Z.
Cho
S.
Sherrill-Mix
S.
Wan
L.
, et al.  . 
U1 snRNP determines mRNA length and regulates isoform expression
Cell
 , 
2012
, vol. 
150
 (pg. 
53
-
64
)
23
Rung
J.
Cauchi
S.
Albrechtsen
A.
Shen
L.
Rocheleau
G.
Cavalcanti-Proença
C.
Bacot
F.
Balkau
B.
Belisle
A.
Borch-Johnsen
K.
, et al.  . 
Genetic variant near IRS1 is associated with type 2 diabetes, insulin resistance and hyperinsulinemia
Nat. Genet.
 , 
2009
, vol. 
41
 (pg. 
1110
-
1115
)
24
Voight
B.F.
Scott
L.J.
Steinthorsdottir
V.
Morris
A.P.
Dina
C.
Welch
R.P.
Zeggini
E.
Huth
C.
Aulchenko
Y.S.
Thorleifsson
G.
, et al.  . 
Twelve type 2 diabetes susceptibility loci identified through large-scale association analysis
Nat. Genet.
 , 
2010
, vol. 
42
 (pg. 
579
-
589
)
25
Zeggini
E.
Scott
L.J.
Saxena
R.
Voight
B.F.
Marchini
J.L.
Hu
T.
de Bakker
P.I.
Abecasis
G.R.
Almgren
P.
Andersen
G.
, et al.  . 
Meta-analysis of genome-wide association data and large-scale replication identifies additional susceptibility loci for type 2 diabetes
Nat. Genet.
 , 
2008
, vol. 
40
 (pg. 
638
-
645
)
26
Wellcome Trust Case Control Consortium
Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls
Nature
 , 
2007
, vol. 
447
 (pg. 
661
-
678
)
27
Sim
X.
Ong
R.T.
Suo
C.
Tay
W.T.
Liu
J.
Ng
D.P.
Boehnke
M.
Chia
K.S.
Wong
T.Y.
Seielstad
M.
, et al.  . 
Transferability of type 2 diabetes implicated loci in multi-ethnic cohorts from Southeast Asia
PLoS Genet.
 , 
2011
, vol. 
7
 pg. 
e1001363
 
28
Kuo
J.Z.
Sheu
W.H.
Assimes
T.L.
Hung
Y.J.
Absher
D.
Chiu
Y.F.
Mak
J.
Wang
J.S.
Kwon
S.
Hsu
C.C.
, et al.  . 
Trans-ethnic fine mapping identifies a novel independent locus at the 3′ end of CDKAL1 and novel variants of several susceptibility loci for type 2 diabetes in a Han Chinese population
Diabetologia
 , 
2013
, vol. 
56
 (pg. 
2619
-
2628
)
29
Pagani
F.
Baralle
F.E.
Genomic variants in exons and introns: identifying the splicing spoilers
Nat Rev Genet.
 , 
2004
, vol. 
5
 (pg. 
389
-
396
)
30
Visel
A.
Rubin
E.M.
Pennacchio
L.A.
Genomic views of distant-acting enhancers
Nature
 , 
2009
, vol. 
461
 (pg. 
199
-
205
)
31
Ragvin
A.
Moro
E.
Fredman
D.
Navratilova
P.
Drivenes
O.
Engstrom
P.G.
Alonso
M.E.
de la Calle Mustienes
E.
Gomez Skarmeta
J.L.
Tavares
M.J.
, et al.  . 
Long-range gene regulation links genomic type 2 diabetes and obesity risk regions to HHEX, SOX4, and IRX3
Proc. Natl. Acad. Sci. U S A.
 , 
2010
, vol. 
107
 (pg. 
775
-
780
)
32
Mondal
A.K.
Das
S.K.
Baldini
G.
Chu
W.S.
Sharma
N.K.
Hackney
O.G.
Zhao
J.
Grant
S.F.
Elbein
S.C.
Genotype and tissue-specific effects on alternative splicing of the transcription factor 7-like 2 gene in humans
J. Clin. Endocrinol. Metab.
 , 
2010
, vol. 
95
 (pg. 
1450
-
1457
)
33
Pang
D.X.
Smith
A.J.
Humphries
S.E.
Functional analysis of TCF7L2 genetic variants associated with type 2 diabetes
Nutr. Metab. Cardiovasc. Dis.
 , 
2013
, vol. 
23
 (pg. 
550
-
556
)
34
Dayeh
T.A.
Olsson
A.H.
Volkov
P.
Almgren
P.
Ronn
T.
Ling
C.
Identification of CpG-SNPs associated with type 2 diabetes and differential DNA methylation in human pancreatic islets
Diabetologia
 , 
2013
, vol. 
56
 (pg. 
1036
-
1046
)
35
Florez
J.C.
Jablonski
K.A.
Bayley
N.
Pollin
T.I.
de Bakker
P.I.
Shuldiner
A.R.
Knowler
W.C.
Nathan
D.M.
Altshuler
D.
Diabetes Prevention Program Research Group.
TCF7L2 polymorphisms and progression to diabetes in the Diabetes Prevention Program
N. Engl. J. Med.
 , 
2006
, vol. 
355
 (pg. 
241
-
250
)
36
Groves
C.J.
Zeggini
E.
Minton
J.
Frayling
T.M.
Weedon
M.N.
Rayner
N.W.
Hitman
G.A.
Walker
M.
Wiltshire
S.
Hattersley
A.T.
, et al.  . 
Association analysis of 6,736 U.K. subjects provides replication and confirms TCF7L2 as a type 2 diabetes susceptibility gene with a substantial effect on individual risk
Diabetes
 , 
2006
, vol. 
55
 (pg. 
2640
-
2644
)
37
Grant
S.F.
Thorleifsson
G.
Reynisdottir
I.
Benediktsson
R.
Manolescu
A.
Sainz
J.
Helgason
A.
Stefansson
H.
Emilsson
V.
Helgadottir
A.
, et al.  . 
Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk of type 2 diabetes
Nat. Genet.
 , 
2006
, vol. 
38
 (pg. 
320
-
323
)
38
Wahl
M.C.
Will
C.L.
Luhrmann
R.
The spliceosome: design principles of a dynamic RNP machine
Cell
 , 
2009
, vol. 
136
 (pg. 
701
-
718
)
39
Patel
S.B.
Novikova
N.
Bellini
M.
Splicing-independent recruitment of spliceosomal small nuclear RNPs to nascent RNA polymerase II transcripts
J. Cell Biol.
 , 
2007
, vol. 
178
 (pg. 
937
-
949
)
40
Spiluttini
B.
Gu
B.
Belagal
P.
Smirnova
A.S.
Nguyen
V.T.
Hebert
C.
Schmidt
U.
Bertrand
E.
Darzacq
X.
Bensaude
O.
Splicing-independent recruitment of U1 snRNP to a transcription unit in living cells
J. Cell Sci.
 , 
2010
, vol. 
123
 (pg. 
2085
-
2093
)
41
Mali
P.
Yang
L.
Esvelt
K.M.
Aach
J.
Guell
M.
DiCarlo
J.E.
Norville
J.E.
Church
G.M.
RNA-guided human genome engineering via Cas9
Science
 , 
2013
, vol. 
339
 (pg. 
823
-
826
)
42
Cong
L.
Ran
F.A.
Cox
D.
Lin
S.
Barretto
R.
Habib
N.
Hsu
P.D.
Wu
X.
Jiang
W.
Marraffini
L.A.
, et al.  . 
Multiplex genome engineering using CRISPR/Cas systems
Science
 , 
2013
, vol. 
339
 (pg. 
819
-
823
)
43
Shen
L.X.
Basilion
J.P.
Stanton
V.P.
Jr.
Single-nucleotide polymorphisms can cause different structural folds of mRNA
Proc. Natl. Acad. Sci. U S A.
 , 
1999
, vol. 
96
 (pg. 
7871
-
7876
)
44
Yamamoto
H.
Morino
K.
Nishio
Y.
Ugi
S.
Yoshizaki
T.
Kashiwagi
A.
Maegawa
H.
MicroRNA-494 regulates mitochondrial biogenesis in skeletal muscle through mitochondrial transcription factor A and Forkhead box j3
Am. J. Physiol. Endocrinol. Metab.
 , 
2012
, vol. 
303
 pg. 
E1419–1427
 
45
Bandiera
S.
Rüberg
S.
Girard
M.
Cagnard
N.
Hanein
S.
Chrétien
D.
Munnich
A.
Lyonnet
S.
Henrion-Caude
A.
Nuclear outsourcing of RNA interference components to human mitochondria
PLoS One
 , 
2011
, vol. 
6
 pg. 
e20746
 
46
Iguchi
H.
Hayashi
I.
Kono
A.
A somatostatin-secreting cell line established from a human pancreatic islet cell carcinoma (somatostatinoma): release experiment and immunohistochemical study
Cancer Res.
 , 
1990
, vol. 
50
 (pg. 
3691
-
3693
)

Author notes

These authors contributed equally to this work.

Supplementary data