Binding of activating transcription factor 6 to the A5/Core of the rat insulin II gene promoter does not mediate its transcriptional repression

Abstract

Pancreatic β-cells have a well-developed endoplasmic reticulum due to their highly specialized secretory function to produce insulin in response to glucose and nutrients. It has been previously reported that overexpression of activating transcription factor 6 (ATF6) reduces insulin gene expression in part via upregulation of small heterodimer partner. In this study, we investigated whether ATF6 directly binds to the insulin gene promoter, and whether its direct binding represses insulin gene promoter activity. A bioinformatics analysis identified a putative ATF6 binding site in the A5/Core region of the rat insulin II gene promoter. Direct binding of ATF6 was confirmed using several approaches. Electrophoretic mobility shift assays in nuclear extracts from MCF7 cells, isolated rat islets and insulin-secreting HIT-T15 cells showed ATF6 binding to the native A5/Core of the rat insulin II gene promoter. Antibody-mediated supershift analyses revealed the presence of both ATF6 isoforms, ATF6α and ATF6β, in the complex. Chromatin immunoprecipitation assays confirmed the binding of ATF6α and ATF6β to a region encompassing the A5/Core of the rat insulin II gene promoter in isolated rat islets. Overexpression of the active (cleaved) fragment of ATF6α, but not ATF6β, inhibited the activity of an insulin promoter-reporter by 50%. However, the inhibitory effect of ATF6α was insensitive to mutational inactivation or deletion of the A5/Core. Therefore, although ATF6 binds directly to the A5/Core of the rat insulin II gene promoter, this direct binding does not appear to contribute to its repressive activity.

Figure 1. Thapsigargin-induced ER-stress in isolated rat islets inhibits insulin pre-mRNA expression

(A) Expression of BIP, XBP-1s, ATF4 and ATF6 mRNA in isolated islets exposed to 2.8 and 16.7 mM glucose in the presence or absence of 1 μM thapsigargin for 6 h. (B) Expression of insulin pre-mRNA, PDX-1 and MafA mRNA in isolated islets exposed to 2.8 and 16.7 mM glucose in the presence or absence of 1 μM thapsigargin for 6 h. Pre-mRNA and mRNA levels were measured by RT-PCR and normalized by β-actin mRNA levels. Data are mean ± SEM of 4–6 independent experiments.*p<0.05

Figure 2. Identification of a putative ATF6 binding site on the A5/Core of the Insulin gene promoter region

Alignment of nucleotide sequences of the 5′-flanking region of the insulin I and II genes from mouse, rat and human. A box indicates a putative ATF6 binding site. The arrow indicates the previously described transcription start site (+ 1) (TSS). Asterisks indicate nucleotide homology across species. Flanking the A5/Core, sequences recognized by forward and reverse primers used for ChIP analysis are underlined (sequences shown in Supplementary Table 1). Bioinformatics analysis shows one putative conserved ATF6 binding site located within the A5/Core.

Figure 3. ATF6α and ATF6β bind to the A5/Core of the rat insulin II gene promoter

Nuclear extracts from MCF7 cells, isolated rat islets and HIT-T15 cells were tested by EMSA for their ability to bind to DNA probe containing the A5/Core. (A) EMSA of ³²P-labeled A5/Core probe. Increasing concentrations of nuclear extracts (2.5, 5 and 10 μg) isolated from MCF7 cells (Lanes 1 to 3). Two different anti-ATF6α antibodies were added to lanes 4 and 5, and anti-ATF6β was added to lane 6. Competition was done with 50-fold molar excess of unlabeled A5/Core probe (lane 7). (B) EMSA of ³²P-labeled A5/Core probe. Nuclear extracts were isolated from rat islets exposed for 6 h to 11.1 mM glucose in the absence or the presence of 1 μM thapsigargin (lanes 1 and 2). Anti-ATF6α and anti-ATF6β antibodies were added respectively to lanes 3 and 4. Competition was done with 100-fold molar excess of unlabeled A5/Core probe (lane 5). (C) EMSA of ³²P-labeled Intron 1 probe. Nuclear extracts were isolated from rat islets exposed for 6 h to 11.1 mM glucose in the presence of 1 μM thapsigargin. Anti-ATF6α and anti-ATF6β antibodies were added respectively to lanes 2 and 3. Competition was done with 100-fold molar excess of unlabeled Intron 1 probe (lane 4). (D) EMSA of ³²P-labeled A5/Core probe. Nuclear extracts isolated from immortalized pancreatic β-cells HIT-T15 transfected with increasing amount of ATF6p50 (0, 0.5 and 1.0 μg) (Lanes 1 to 3). Anti-ATF6α and anti-ATF6β antibodies were added respectively to lanes 4 and 5. Competition was done with 100-fold molar excess of unlabeled A5/Core probe (lane 6). EMSA probe sequences are indicated in Supplementary Table 1. Data shown are representative gels of at least three independent experiments.

Figure 4. Binding of ATF6 to the endogenous rat insulin II gene promoter, as assessed by ChIP analysis

Isolated rat islets were exposed to 2.8 or 16.7 mM glucose in the presence or absence of 1 μM thapsigargin for 6 h. Chromatin was immunoprecipitated with ATF6α antiserum (A), ATF6β antiserum (B), or normal rabbit serum. Data are expressed as the fold increase of the immunoprecipitated sample relative to the control and normalized to the amount of β-galactosidase recovered at the elution step. Data are mean ± SEM of 2–5 separate experiments. (C) Representative immunoblot from 3 independent experiments probed for antibodies against cleaved (ATF6α-p50) and uncleaved (ATF6α-p90) ATF6α and α-tubulin.

Figure 5. Overexpression of ATF6α-p50, but not ATF6β-p60, represses human insulin promoter activity

(A) HIT-T15 cells were cotransfected with the INS(-327)Luc with increasing amounts of the ATF6α-p50 or ATF6β-p60 expression vector or an empty vector (pcDNA3.1). Total DNA amount was identical amongst conditions. Firefly luciferase activity was corrected with Renilla or β-galactosidase activity. (B) Schematic representation of the different constructs used to assess the role of the A5/Core in INS(-327)Luc, INS(-230)Luc and mINS(-327)Luc containing a site-specific mutation of the A5/Core. (C) HIT-T15 cells were cotransfected with 0.5 μg ATF6α-p50 or the empty expression vector (pcDNA3.1), and INS(-327)Luc, mINS(-327)Luc or INS(-230)Luc. Transfection efficacy was corrected by normalizing Firefly luciferase activity to Renilla activity. Data are mean ± SEM of 3–4 separate experiments. *p< 0.05