Mapping of INS promoter interactions reveals its role in long-range regulation of SYT8 transcription.

Insulin (INS) synthesis and secretion from pancreatic β-cells are tightly regulated; their deregulation causes diabetes. Here we map INS-associated loci in human pancreatic islets by 4C and 3C techniques and show that the INS gene physically interacts with the SYT8 gene, located over 300 kb away. This interaction is elevated by glucose and accompanied by increases in SYT8 expression. Inactivation of the INS promoter by promoter-targeting siRNA reduces SYT8 gene expression. SYT8-INS interaction and SYT8 transcription are attenuated by CTCF depletion. Furthermore, SYT8 knockdown decreases insulin secretion in islets. These results reveal a nonredundant role for SYT8 in insulin secretion and indicate that the INS promoter acts from a distance to stimulate SYT8 transcription. This suggests a function for the INS promoter in coordinating insulin transcription and secretion through long-range regulation of SYT8 expression in human islets.

INTRODUCTION

The human genome is organized within the nucleus into a compact chromatin structure that facilitates many interactions between what are otherwise distal sites[1]. A few of these long-range interactions have been shown to impact gene transcription in human cells[2-5]. However, the function of chromatin interactions in the regulation of physiological processes and how environmental signals regulate these dynamic interactions in human tissues remain to be determined. We investigated the INS gene locus in this study both for its physiological importance, and as a model to study the role of long-range chromatin interactions in the establishment of regulatory networks. Most of our knowledge about insulin (INS) gene expression and regulation comes from studies done in rodents, especially the mouse. However, there exist two insulin genes in rodents and only one in human and it has become apparent recently that human pancreatic islets contain unique anatomical and functional features they do not share with rodent islets[6-9]. For these and other reasons we made use of intact human pancreatic islets, in which the INS gene is active.

The human INS gene, located on chromosome 11p15, is flanked by two clusters of imprinted genes. The DNA sequence and gene organization of this 1-Mb gene-dense region are conserved between mouse and human[10]. Insulin transcription and secretion from pancreatic β cells are tightly regulated in response to glucose[11,12]. Glucose markedly stimulates the human INS promoter activity in transfected primary islet cells[13,14], actively promotes open chromatin structure and RNA polymerase II recruitment on the endogenous INS promoter, and increases INS gene transcription in human islets[11,15,16]. The INS gene is transcribed at high levels only in pancreatic β cells[17,18]. The mouse Ins2 gene and its human ortholog physically interact with the distal H19 imprinted control region (ICR) and enhancers that are located over 100 kb away[19-21]. The H19 ICR has been the subject of intense study, and recent 4C (circular chromosome conformation capture) analyses have revealed its long-range interactions with other gene loci, in at least two cases affecting transcription of distant genes[22-24]. Although these data provide evidence for the existence of long-range physical contacts involving the INS gene, they are limited to the region near H19, in non-human tissues, and in cells not expressing insulin.

In this study, we undertook a series of 3C (chromosome conformation capture) and 4C studies[24,25] in human islets to determine patterns of interaction between the INS gene and other sites within the nucleus, and to explore possible connections with INS expression. We show that, in human islets, the INS promoter physically interacts with and positively regulates transcription of the SYT8 gene located over 300 kb away, and that this gene is a critical regulator of insulin secretion. These results suggest a regulatory function for the INS promoter in coupling insulin transcription with secretion through regulation of SYT8 gene expression in human islets.

RESULTS

4C-Seq analysis reveals SYT8–TNNI2 locus association with INS

Initial 3C studies confirmed that the INS gene and the H19 ICR interact with each other both in normal human primary fibroblasts (NHPF) and at a higher level in human islets (Supplementary Fig. 1). We explored more generally, using the 4C method[24,25], interactions in human islets between the INS gene and other genomic sites. This procedure (4C-Seq, Supplementary Fig. 2a) is done essentially as described[25] except for the last few steps.

We first performed 4C-Seq using a part of the INS promoter and gene body as the bait (see Supplementary Fig. 2b). This generated 1.8 million unique sequence reads; 29.6% of them were mapped within the 1-Mb region surrounding the INS gene. In islets treated with glucose for one hour, the INS gene interacted strongly with loci in this 1-Mb region (Fig. 1a, b). Multiple interacting sites were clustered in this region. The analysis was repeated with islets from another donor and using the other side of the BglII site as the bait (Supplementary Fig. 2b); similar contact patterns were observed for this 1-Mb region (P<10−90) (Supplementary Fig. 3a and Supplementary Notes). Thus, the 4C-Seq method is reproducible (see Supplementary Notes). Because we were focused on interactions within this 1-Mb locus, we did not attempt to comprehensively identify more distant regions on chromosome 11 or on any of other chromosomes (Fig. 1a and Supplementary Notes). Instead, we used the 4C data to provide partial information about significant long range contacts that could then be confirmed and explored by 3C methods.

Figure 1

4C-Seq analysis reveals the association of SYT8 with INS gene in human pancreatic islets. 4C-Seq analysis of INS-associated loci in the entire human chromosome 11 (a), the INS nearby region (b) and the SYT8–TNNI2 gene locus (c). 4C peaks are shown (a-c), and INS, SYT9, SYT8, SYT13 and SYT7 genes and the H19 ICR are located (a, b). The frequency of two 4C peaks marked with asterisks is larger than 25,000 and truncated. (c) The location of SYT8 and TNNI2 as well as CTSD and LSP1 genes, the 3C PCR primers and the corresponding BglII sites (vertical line) in the SYT8 and TNNI2 locus (lower). The 3C PCR primers are numbered. The exons are marked with solid bars and the transcription direction marked with arrows. Not all of the BglII sites in this region are shown and there is only one BglII site within the CTSD gene. (d) TaqMan quantitative 3C analysis of INS interactions with the SYT8-TNNI2 locus in glucose-treated islets, islets-derived hIPCs and primary human fibroblasts. Plotted are the relative interactions in arbitrary units of the INS gene with eight BglII sites (upper, not in scale), which are numbered as in c and whose location in the region shown in c. The mean ± SEM is shown (n = 8). The sites for the SYT8 promoter and its 3′ downstream site (or the TNNI2 promoter) are marked with arrows.

The INS-interacting loci in this 1-Mb region (Fig. 1b) are in a part of the genome that is evolutionarily conserved in DNA sequence and gene organization[10] and contain genes that play roles in diverse cellular processes including exocytosis, cellular proliferation and apoptosis, and cell-cell interactions. In this study we focus on one of these genes, synaptotagmin 8 (SYT8)[26,27], and its immediate neighbor, TNNI2. Other synaptotagmins have been shown to be Ca2+ sensors for neurotransmitter release in neurons[28], and thought to regulate insulin secretion in pancreatic β cells[26,27,29-31]. Syt8 could also be involved in this process because its C2AB domain peptide markedly decreases Ca2+–induced insulin secretion when introduced into primary rat islet cells[26]. However, earlier studies[27] of Syt8 concluded that it did not act as a calcium sensor in endocrine cells or neurons, and did not participate in the calcium-mediated regulation of evoked exocytosis. Thus, whether SYT8 plays a role in insulin secretion remains unclear.

Our analysis showed that the INS gene physically contacted both the promoter and 3′ downstream region of SYT8 (Fig. 1c). The second INS-interacting site is located less than 2 kb away from the transcription start site of TNNI2, the neighboring gene of SYT8 (see Fig. 1c for the location). The INS gene did not interact with the CTSD gene or the 3′ downstream region of the LSP1 gene, two flanking genomic regions both located about 60 kb away from the SYT8-TNNI2 locus (Fig. 1c); this was confirmed in further 3C experiments (see below). Thus, proximity alone is not sufficient for an interaction. Finally, 4C-Seq analysis showed that the INS gene did not interact with any of the other SYT gene family members such as SYT7, SYT9 or SYT13 that are also located on chromosome 11 (Fig. 1a, Supplementary Fig. 3b and data not shown).

3C analysis confirms SYT8–TNNI2 locus interaction with INS

To confirm the 4C results (Fig. 1c), we designed quantitative 3C experiments to detect specifically interactions between the INS bait and the region containing the SYT8 and TNNI2 genes in islets treated with glucose for 1 h. We focused on interactions of the INS gene with eight BglII sites in the region, representing the two sites at the SYT8 promoter and its 3′ downstream site (or the TNNI2 promoter), two immediately neighboring sites flanking the SYT8-TNNI2 locus, three sites covering the LSP1 gene and the only site within the CTSD gene (see Fig. 1c for the location). Quantitative 3C assays confirmed strong contacts between INS and the SYT8-TNNI2 locus in islets from a second donor (Fig. 1d). Weak contact between these sites occurred at 2- to 3-fold lower levels in human primary fibroblasts (Fig. 1d). Similar to the 4C-Seq analysis, this 3C assay also showed that the INS gene interacted strongly with the LSP1 gene but not with its 3′ downstream region or the CTSD gene in islets (Fig. 1d). Islets from the same donor were also used to generate islet-derived precursor cells (hIPC) that express, weakly if at all, the INS[32] or SYT8 gene (Supplementary Fig. 4a). After islets were dedifferentiated into hIPC, INS interactions with the SYT8-TNNI2 locus were decreased by about 3-fold while the LSP1-INS interaction in hIPCs, unlike in fibroblasts, was maintained (Fig. 1d). This confirmed the 4C-Seq results showing that the SYT8-TNNI2 locus interacts with the INS gene in islets and suggested that these intranuclear interactions might be regulated, at least partly, by an islet cell-specific mechanism. We note that although INS makes strong contacts with SYT8-TNNI2, it does not make contact with the nearby CTSD gene (Fig. 1d), which is also active in islets (see below), indicating that the contacts are not generated simply through active transcription.

Glucose stimulates INS–SYT8 interactions and SYT8 expression

To determine whether glucose could regulate INS-SYT8 interaction in islets, we measured its effect on 3C contacts. Before glucose addition, interactions of SYT8 and its nearby region with INS were readily detected; half an hour after glucose addition, INS interactions with the SYT8 promoter and its 3′ downstream region (or the TNNI2 promoter) were increased by 3–5 fold in islets from two donors (Fig. 2a). The INS-CTSD interaction, in contrast, was weak both before and after glucose treatment (Fig. 2a). One hour after glucose treatment, a 1.8- to 3.2-fold increase in INS-SYT8 interaction was also seen in islets from three donors (Fig. 2b). Time course analysis in one donor’s islets showed that this interaction increased five fold within 30 min of glucose treatment and then gradually decreased (Fig. 2c). We conclude that glucose dynamically regulates interactions of the SYT8-TNNI2 locus with INS in islets.

Figure 2

Glucose stimulates INS interactions with the SYT8-TNNI2 gene locus and increases SYT8 and TNNI2 gene expression in human islets. (a) TaqMan quantitative 3C analysis of the relative interaction of the INS gene with the SYT8 promoter and its 3′ downstream site (or the TNNI2 promoter), marked with arrows, in islets from two donors with and without glucose treatment for 30 min. The mean ± SEM is shown (n = 8). (b) TaqMan quantitative 3C analysis of the interaction of the INS gene with the SYT8 promoter in islets from three donors with and without glucose treatment for 1 h. The SYT8-INS interaction is shown relative to that in islets without glucose treatment. The mean ± SEM is shown (n = 8). (c) Time-course analysis of the interaction of the INS gene with the SYT8 promoter in islets from one other donor treated with glucose for the indicated times. (d) Quantitative real-time RT-PCR (qRT-PCR) analysis of SYT8, TNNI2, SYT7, SYT13 and INS gene expression in islets from three donors with and without glucose treatment for the indicated times. The RNA levels are normalized to those of HPRT1 and the mRNA levels relative to islets at t = 0 are plotted. The mean ± SEM is shown (n = 9).

Glucose markedly increases human INS promoter activity in primary cultured islet cells[13,14] and also stimulates endogenous INS gene transcription in human islets[15,16]. It has been reported that, within 30 min of addition, glucose markedly increases H4 acetylation and RNA polymerase II recruitment to the INS promoter in islets[11], while the levels of mature INS transcripts appear to be stable for at least 48 hours after glucose treatment; their long half-life leads to high levels of accumulation in β cells[15,16]. We confirmed this result: levels of mature INS transcript were stable (Fig. 2d). On the other hand, steady-state mRNA levels of SYT8 and TNNI2 were increased by 1.4- to 3.2-fold within 30 min of glucose treatment in islets from 3 donors (Fig. 2d). In contrast, SYT7 and SYT13 expression levels were almost unchanged (Fig. 2d). Thus, glucose specifically and rapidly increases SYT8 and TNNI2 gene expression in islets, with kinetics of mRNA accumulation similar to the EGR1 (ZIF268) gene, whose expression is also regulated by glucose in pancreatic β cells[33] (Supplementary Fig. 5). Therefore, SYT8 and TNNI2 increases in expression correlate in time with increases in INS interactions with the SYT8 and TNNI2 locus upon glucose treatment in islets.

The INS promoter positively regulates SYT8 and TNNI2 expression

The data presented so far are formally consistent with a model in which both INS and SYT8 expression are stimulated by glucose, and as a consequence are drawn into proximity at a shared transcription factory, as shown for the globin locus[34,35]. The data are also consistent with a distinct, but not incompatible, model in which insulin expression levels directly affect SYT8 expression. To investigate the role of the INS promoter in SYT8 and TNNI2 transcription, we treated islets with siRNAs that specifically target the INS promoter (Fig. 3a). Promoter-targeting siRNAs have been shown to inactivate endogenous promoters in dividing and non-dividing human cells[36,37]. Islets were treated with the INS promoter-targeting siRNA duplexes or control siRNA duplexes for four and a half days. To prevent effects of the potential insulin deficiency in siRNA-treated islets on gene expression[38], siRNA-treated islets were washed twice with fresh basal islet media and cultured in fresh supplemented islet media containing physiological levels of exogenous insulin for at least 6 h before total RNAs were prepared. Treatment of islets from two donors with these siRNAs led to 34–50% reduction in the INS preRNA levels (Fig. 3a, b), indicative of effective silencing of the INS promoter. The steady-state INS mRNA levels were stable or slightly decreased after this short-term siRNA treatment (Fig. 3a, b) probably because of their long half-life. This siRNA-induced inactivation of the INS promoter decreased SYT8 and TNNI2 gene expression by 58–83% and 43–66%, respectively, in islets from two donors (Fig. 3a, b) but did not alter expression of SYT13 or of CTSD and MRPL23, two genes located about 60 and 100 kb away, respectively, from each end of the SYT8-TNNI2 locus. In addition, this inactivation of the INS promoter had little effect on expression of KCNQ1 and TSSC4, which are also located in the 1-Mb region (Supplemental Fig. 6a). These results demonstrate that the INS promoter exerts long-range effects on the regulation of SYT8 and TNNI2 gene transcription in islets. Although it is likely that INS, SYT8 and TNNI2 do share a transcription factory, this is not sufficient to explain the dependence of SYT8 expression on the INS promoter (see Discussion). The data also indicate that long-range regulation of transcription by the INS promoter is not domain-wide but rather gene-specific. Since inactivation of the INS promoter has no effect on expression of MRPL23, a gene located between the SYT8 and INS loci and about 100 kb closer to INS than SYT8 and TNNI2, the INS promoter seems not to spread its effects on transcription by linear propagation of an activating signal, but more likely through higher-order chromatin organization[1-5,20,22,24,25].

Figure 3

The INS promoter positively regulates SYT8 and TNNI2 gene expression in human islets. (a) qRT-PCR analysis of INS preRNA E2I2 and mature transcript INS E2 (exon 2) and mature transcripts of SYT8, TNNI2, CTSD, MRPL23 and SYT13 genes in islets from two donors that were treated with non-targeting control siRNA (white bar) or one of the two siRNAs targeting to the INS promoter (black bar). The RNA levels are normalized to those of HPRT1. mRNA levels are plotted relative to control. The mean ± SEM is shown (n = 9). The INS exons are shown and the siRNA-targeting sites are marked with arrows (top). (b) qRT-PCR analysis as in a of the islets from one other donor that were treated separately with either of the two siRNAs targeting to the INS promoter.

The mechanism by which the INS promoter transactivates SYT8 remains to be determined. However, we found that the INS promoter, when coupled to the SYT8 promoter, could stimulate SYT8 promoter activity (Supplementary Fig. 6b). In a complementary experiment, inactivating the SYT8 promoter in islets had little effect on INS expression (Supplementary Fig. 7).

CTCF regulates SYT8 and TNNI2 expression and their contacts with INS

CTCF binds to DNA, largely in a cell-type-independent manner[39] and is a major regulator of chromatin architecture[20,22,24,39]. The presence of CTCF-binding sites at both SYT8 and TNNI2 promoters and at the 3′ downstream region of the INS gene[40] (see Supplementary Fig. 8a, b for the location) suggests its involvement in the regulation of INS-SYT8 interaction. Incubation of islets with CTCF-specific siRNAs reduced CTCF gene expression from four donors by 40–72% (Fig. 4a) while it caused a 46–76% decrease in SYT8 mRNA levels, but had little effect on SYT7 and INS gene expression (Fig. 4a). Similarly to SYT8, TNNI2 gene expression was also reduced by CTCF knockdown in islets from three of four donors (Fig. 4a). CTCF depletion reduced SYT8-INS interaction in the same islets described above (Fig. 4b and Supplementary Fig. 8c). The decrease in SYT8-INS interaction and SYT8 gene expression is correlated with the extent of CTCF knockdown (Fig. 4a, b), indicating that CTCF is important for maintenance of SYT8-INS interaction as well as SYT8 and TNNI2 gene expression in islets. The INS-SYT8 interaction is also detected, although at much lower levels (Fig. 1d and Supplementary Fig. 4c), in human primary fibroblasts that do not express SYT8 and TNNI2 genes at appreciable levels (Supplementary Fig. 4b). CTCF depletion in fibroblasts reduced CTCF binding at the SYT8 and INS loci and caused a 67% reduction in the SYT8-INS interaction (Fig. 4c-e), suggesting that CTCF binding at the SYT8 and INS loci also regulates long-range SYT8-INS interactions in human fibroblasts.

Figure 4

CTCF positively regulates SYT8 and TNNI2 gene expression in human islets and is important for the maintenance of the SYT8-INS interaction in human islets and human fibroblasts. (a) qRT-PCR analysis of CTCF, SYT8, TNNI2, SYT7 and INS gene expression in islets from four donors that were treated with non-targeting control or CTCF-specific siRNA. The RNA levels are normalized to those of HPRT1. Plotted are the mRNA levels relative to control. The mean ± SEM is shown (n = 9). (b) TaqMan quantitative 3C analysis of INS-SYT8 interaction in the two same siRNA-treated islets as shown in a. The SYT8-INS interaction is shown relative to control. ***P < 0.0001. (c) qRT-PCR analysis of CTCF gene expression in normal human primary fibroblasts that were treated with control (white bar) or CTCF-specific siRNA (black bar). Shown are the CTCF mRNA levels relative to control. (d) Quantitative ChIP analysis of CTCF occupancy at the SYT8 promoter and the INS 3′ downstream region in human fibroblasts treated as in c. Shown is the CTCF occupancy relative to control. (e) TaqMan quantitative 3C analysis of INS-SYT8 interaction in human fibroblasts treated as in c. Shown is the SYT8-INS interaction relative to control. ***P < 0.0001.

SYT8 is an important regulator of insulin secretion in islets

The existence of an actively-regulated mechanism for SYT8 gene expression (Fig. 2 and Fig. 3) and its potential involvement in insulin secretion[26] suggest a role of SYT8 in controlling glucose response in islets. Incubation of islets with an SYT8-specific siRNA caused a 77% reduction in SYT8 gene expression (Fig. 5a). Compared to islets treated with control siRNA, basal insulin secretion from SYT8-depleted human islets was decreased on average by 43% (P<0.0001) (Fig. 5b). SYT8 knockdown in islets from two donors also attenuated glucose- and arginine-induced insulin secretion by more than 50% (P<0.0001) (Fig. 5c, d) and 31% (P=0.0018) (Fig. 5d), respectively. Finally, SYT8 depletion impaired the first and second phase of insulin secretion in an insulin secretion assay (Fig. 5e). We conclude that, similar to Syt7 in mice[30], the SYT8 gene is important for both basal and glucose-stimulated insulin secretion in human islets. Consistent with this finding, the mouse Syt8 gene is found important for exocytosis of acrosomes in sperm[41].

Figure 5

SYT8 is an important regulator of insulin secretion in human islets. (a) qRT-PCR analysis of SYT8 gene expression in islets treated with control (white bar) or SYT8-specific siRNA (black bar). Total RNA was prepared directly from siRNA-treated islets without medium change. The mean ± SEM is shown (n = 9). (b) ELISA analysis of insulin levels in the medium for islets from 5 donors that were separately treated with siRNA as in a. The insulin levels are normalized to the amount of total RNAs made from the treated islets. As insulin levels vary among islets from different donors, the insulin levels relative to control are plotted. The mean ± SEM is shown (n = 20, 4 measures for each of 5 donors); ***P < 0.0001. (c) ELISA analysis of insulin levels in the medium of islets from one donor treated as in b and cultured with 30.5 mM glucose for 30 min. The insulin levels are normalized to the amount of islets necessary to produce 1μg total RNA; ***P < 0.0001. (d) ELISA analysis of insulin levels in the medium of islets from another donor that were treated as in c and cultured with 30.5 mM glucose or 20 mM L-arginine for 30 min. The insulin levels are normalized as in c.***P<0.0001; *P=0.0018. (e) Insulin secretion from siRNA-treated islets that were treated with 30.5 mM glucose and supernatants were collected every 5 min. The mean ± SD is shown (n = 4).

DISCUSSION

An important role for the INS promoter in long-range transcriptional regulation was evident from our observations that specific inhibition of INS transcription by promoter-targeting siRNA decreases SYT8 and TNNI2 gene expression by 40–80% in human islets. SYT8 and TNNI2 gene expression correlate positively with their interactions with the INS gene in islets treated with glucose (Fig. 2), indicating a positive role for intranuclear contact between the INS gene and the SYT8-TNNI2 locus in SYT8 and TNNI2 gene transcription. Glucose stimulates both INS promoter activity in primary islet cells[13,14] and interactions of the INS gene with the SYT8-TNNI2 locus in islets (Fig. 2), compatible with the notion that glucose could enhance SYT8 and TNNI2 gene expression through increasing their interactions with INS. This study reveals a mechanism by which a strong tissue- or cell-type-specific regulatory element can coordinate cellular response to acute environmental and developmental signals through remote control of transcription of a second broadly expressed gene.

Our work indicates that CTCF is important for maintenance of interactions of INS with SYT8 and TNNI2 and for SYT8 and TNNI2 expression in islets. Since CTCF depletion had little effect on INS gene expression in human islets (Fig. 4a), CTCF probably does not act by directly activating transcription but more likely through regulating long-range chromosomal interactions as previously shown in the β-globin and H19 loci[20,22,24,25]. The observations that SYT8-INS interactions and SYT8 gene expression are markedly reduced upon CTCF depletion and that CTCF-binding sites are present in both INS and SYT8 loci suggests that CTCF-dependent SYT8-INS interactions positively regulate SYT8 and TNNI2 gene transcription in islets. CTCF may be involved in many cell types in organizing local chromatin architecture that could poise the SYT8 and TNNI2 genes for enhanced transcriptional activation once a cell type-specific activating signal is available (See model in Supplementary Fig. 9). Due to technical limitations in use of human tissues and potential indirect effects of CTCF knockdown, we can only demonstrate a correlation between CTCF mediated stabilization of long range contacts and effects on SYT8 and TNNI2 expression. Given the complex nature of networks stabilized by CTCF interactions, it is in any case difficult to isolate the effects of a single interacting pair of sites.

Interactions of the SYT8-TNNI2 locus with the INS gene may potentially promote SYT8 and TNNI2 gene transcription by recruiting to the SYT8 and TNNI2 promoters active histone marks present on the INS promoter[15,32,42], or by allowing transcription factors bound to the INS promoter to directly activate SYT8 expression in a manner analogous to transvection in Drosophila. We considered whether these effects could have arisen solely from independent recruitment of both INS and SYT8 promoters to a center of high RNA polymerase II concentration[34,35]. It is likely that these genes, when expressed, do share such a transcription factory. However, that is not sufficient to explain our observations that inactivation of the INS promoter reduces SYT8 and TNNI2 expression as well as INS gene transcription, and that INS expression is not affected by loss of CTCF binding and the SYT8-INS interaction, whereas SYT8 and TNNI2 expression are markedly decreased. Previously reported properties of transcription factories do not include the ability of some genes present within a factory to transactivate others in the same factory. In the case of the INS locus it appears that CTCF plays a role in facilitating the INS-SYT8/TNNI2 interaction, and that this interaction is coupled with an activation of SYT8/TNNI2 expression dependent on the level of INS expression. Finally, as impaired insulin secretion and diabetes have been observed in an individual with a balanced chromosomal translocation t(1;11)(p36.22;p15.5) that causes physical separation of the INS gene from SYT8 and other nearby genes[43], local genomic organization around INS might contribute to regulation of islet function.

An important role for SYT8 in insulin secretion, identified by this study, is unexpected because SYT8 lacks the critical aspartate residues for Ca2+ binding at both C2 domains and cannot bind phospholipids and target membrane SNARE (soluble N-ethylmaleimide-sensitive factor (NSF) attachment protein receptor)[44]. In contrast to Syt7 and Syt9, Syt8 fails to stimulate in vitro SNARE-mediated membrane fusion[44], an essential step for exocytosis of insulin. On the other hand, SYT8 has been shown to bind syntaxins with an affinity similar to or higher than Syt1[45], to regulate in vitro Syt1-stimulated membrane fusion[44] and to interact with Syt7 in a Ca2+-dependent manner although the interaction is weak and likely transient and hard to detect in cells[46]. These studies suggest that SYT8 may serve as a regulatory protein for other SYTs. The exact steps at which SYT8 acts during insulin secretion are unclear.

This work reveals that, through regulation of SYT8, the INS promoter may couple INS gene transcription with insulin secretion in islets. It seems advantageous that SYT8 gene transcription could be quickly adjusted in response to glucose through the INS promoter so that the cytosolic SYT8 pool in β cells could be refilled for the next round of insulin secretion. This novel mechanism for co-regulation is distinct from, and more direct than, the feedback regulatory effects of secreted insulin on β cells[38,47]. Together with the finding that secreted insulin stimulates INS gene transcription[38,47], our data reveal a positive feedback circuit that functions in islets to coordinate insulin transcription and secretion. This regulatory circuit might be important for robust and timely secretion of insulin in response to rapid changes of glucose concentrations. Furthermore, severe reduction in insulin transcription could potentially disrupt the regulatory circuit and lead to deregulation of insulin secretion. This may be relevant to pathogenesis of human diabetes: inactivating mutations in the INS promoter regulator genes PDX1, HNF-1α, HNF-1β, HNF-4α and NEUROD1[48-50] cause impaired insulin secretion in patients with monogenic forms of diabetes, maturity-onset diabetes of the young (MODY)[51], and loss-of-function mutations in the human INS promoter region cause neonatal diabetes[52].

Coordinated regulation of transcription by the INS promoter may not be limited to SYT8. Steady-state TNNI2 expression is similarly regulated in islets. This suggests that glucose-regulated INS promoter activity may coordinate multiple β cell functions. Since TNNI2 interacts with and activates the transcriptional activity of estrogen receptor-related receptor α (ERRα)[53], an important regulator of energy metabolism[54,55], regulation of TNNI2 gene transcription by the INS promoter might contribute to this function in human islets. Remote regulatory control by the INS promoter may have implications for pathology and treatment of type 2 diabetes. The UKPDS (United Kingdom Prospective Diabetes Study) clinical trials have shown the critical role of preserving β cell functions for the management of type 2 diabetes[56]. Our work suggests that recovering and sustaining active transcriptional activity on the INS promoter could constitute an important target to achieve this.

METHODS

Culture of human islets.

Isolated human pancreatic islets from independent cadaver donors were obtained from the Islet Cell Resource Centers (ICRs). Human islets were cultured in the shipping media for 2 h and placed on a 40 μm cell strainer (BD Falcon) to get rid of the dead or single islet cells. Islets on the cell strainer were washed twice with the fresh basal islet media, CMRL-1066 (99–663-cv, Cellgro) and then cultured in fresh media for at least 3 h or overnight in a 37°C cell incubator before further experiments.

Culture of human primary cells.

Human islet-derived precursor cells (hIPCs) were generated from human islets as previously described[32]. Normal human primary fibroblasts (NHPF) were obtained from ATCC and cultured as recommended.

3C analysis.

Dissociated single islet cells from 10,000 IEQs (islet equivalents) were suspended in ice-cold cross-linking buffer (10 mM Tris-HCl pH8.0, 10 mM MgCl2, 50 mM NaCl) containing 0.1% NP40 and proteinase inhibitors. Fresh formaldehyde was immediately added to 2% and the islet cells were incubated on ice for 15 min. After BglII digestion overnight, 3C-ligated DNA was prepared as previously described[57]. BglII digestion of cross-linked nuclei was efficient under this condition (Supplementary Fig. 2c). The duplex PCR reactions were run with PicoMaxx High Fidelity PCR Master Mix (Stratagene) and primers for β-actin (ACTB), INS and SYT8. PCR products were resolved on 2% ethidium bromide-containing agarose gel. The sequences of the primers used in this work are provided in Supplementary Fig. 12.

Quantitative 3C analysis.

60 ng of 3C ligated DNA was first preamplified for 25 cycles using primers of ACTB or SYT8-INS as above. After passing through a G25 column (GE Healthcare), the preamplified DNA was diluted 100-fold. TaqMan PCR reactions were then run with TaqMan Universal PCR Master Mix (Applied Biosystems). Relative interactions were determined by using a standard curve generated by a serial dilution of the religated BglII-digested templates, which were made by mixing equal molar amount of the corresponding purified PCR products[57]. The interactions in different 3C samples were normalized to that of the ACTB internal control. In control experiments, serial dilutions of 3C ligated DNA were preamplified and then quantitated by TaqMan qPCR. As shown in Supplementary Fig. 10, the SYT8-INS interaction or the ACTB control interaction in the preamplified DNA precisely reflects those in the non-amplified templates.

4C-Seq analysis.

4C analysis was done as previously described[24] except for the last few steps (see details in Supplementary Fig. 2). In brief, after reverse cross-linking, 3C ligated DNAs were digested with NlaIII and then re-ligated to make circular DNA. The first inverse PCR reaction was run for 23 cycles with Phusion DNA polymerase (Finnzymes) and one 5′ biotinylated primer. The PCR products were then digested with AvrII to cut the self-ligated products. The AvrII-digested DNA was incubated with streptavidin dynabeads and DNA-bound dynabeads were subject to nested PCR using two 5′ biotinylated primers. The final PCR products were digested with BglII and NlaIII. After incubation with streptavidin dynabeads, the known bait sequence located at both ends bound to dynabeads and was discarded while the unknown bait-interacting DNA sequences present in the supernatants were purified for preparation of the sequencing library. Solexa DNA sequencing analysis was performed as previously described[40]. The sequence tag profile obtained from 4C-Seq was uploaded into UCSC Genome Browser, and DNA sequences and the chromosome location of the 4C peaks were then downloaded from the UCSC Genome Browser. As shown in Supplementary Fig. 11a, 4C peaks surrounding the SYT8-TNNI2 locus with frequency larger than 150 all have BglII sites and NlaIII sites. 4C PCR products generated at the step prior to BglII and NlaIII digestion were also cloned and sequenced. Ten out of ten such clones had INS bait DNA sequence at both ends and interacting DNA sequence in the middle with BglII and NlaIII restriction sites connecting these two sequences (Supplementary Fig. 11b).

Inactivation of the INS promoter in human islets by promoter-targeting siRNA

Human islets were incubated at 1000 IEQs per ml of the basal islet media in non tissue-culture treated plates for four and a half days with 1 μM of one of two promoter-targeting Accell siRNA duplexes[58] (GGGAAAUGGUCCGGAAAUUUU and GGAUAGGGGUGUGGGGACAUU, Dharmacon) or the non-targeting control Accell siRNA pool (D-001910–10, Dharmacon). siRNA-treated islets were washed twice with fresh basal media and cultured in the fresh supplemented media containing insulin (99–603-CV, Cellgro) for at least 6 h before total cellular RNA was prepared.

Gene knockdown in intact human islets and normal human primary fibroblasts.

Human islets were incubated with 1 μM of Accell siRNA duplex(es) (a mixture of equal molar concentrations of GCCUUAUGAAUGUUAUAUU and GCUCUAAGAAAGAAGAUUC for CTCF; GGCUUAUUCCAGACCUUGU for SYT8) for four and a half days. siRNA-treated islets were washed twice with fresh media and cultured in the fresh media for 5 h before total cellular RNA was prepared. Normal human primary fibroblasts were incubated with 1 μM of CTCF-specific or control siRNA duplexes as above in Accell siRNA delivery media (Dharmacon) for 4 days.

Quantitative real-time RT-PCR analysis.

Total cellular RNA was prepared using RNeasy Mini reagents (Qiagen), and residual genomic DNA was removed by DNase I treatment using a DNA-free kit (Ambion). DNase-treated RNA was then converted to cDNA using the AffinityScript QPCR cDNA synthesis kit (Stratagene) and analyzed by real-time PCR using POWER SYBR Green PCR Master Mix (Applied Biosystems). Relative levels of cDNAs were determined by using a standard curve generated by serial dilutions of cDNAs. INS E2I2 preRNA levels and steady-state levels of SYT8, SYT7, SYT13, INS, CTCF, TNNI2, CTSD, and MRPL23 mRNAs were normalized to those of HPRT1. Each qRT-PCR assay was done in triplicate and independently repeated three times.

Islet insulin secretion experiments.

After siRNA treatment, islets were washed twice and suspended in fresh media and cultured for 3 h. 25 mM D-glucose or 20 mM L-arginine was added to the media for 30 min. Islets on 40 μm cell strainers were also placed into wells containing 5 ml of fresh stimulating media (containing 30.5 mM glucose) and were moved into a new well every 5 min. Finally, human islets were spun down and supernatants collected while total cellular RNA was prepared from the pellets of human islets. The insulin levels in the islet media were determined using a human insulin ELISA assay (IS130D, Calbiotech) and were normalized to the amount of total RNA prepared from the islets. Each ELISA assay was done in duplicate and repeated twice.