Neuroendocrine differentiation factor, IA-1, is a transcriptional repressor and contains a specific DNA-binding domain: identification of consensus IA-1 binding sequence

Abstract

A novel cDNA, insulinoma-associated antigen-1 (IA-1), containing five zinc-finger DNA-binding motifs, was isolated from a human insulinoma subtraction library. IA-1 expression is restricted to fetal but not adult pancreatic and brain tissues as well as tumors of neuroendocrine origin. Using various GAL4 DNA binding domain (DBD)/IA-1 fusion protein constructs, we demonstrated that IA-1 functions as a transcriptional repressor and that the region between amino acids 168 and 263 contains the majority of the repressor activity. Using a selected and amplified random oligonucleotide binding assay and bacterially expressed GST-IA-1DBD fusion protein (257-510 a.a.), we identified the consensus IA-1 binding sequence, TG/TC/TC/TT/AGGGGG/TCG/A. Further experiments showed that zinc-fingers 2 and 3 of IA-1 are sufficient to demonstrate transcriptional activity using an IA-1 consensus site containing a reporter construct. A database search with the consensus IA-1 binding sequence revealed target sites in a number of pancreas- and brain-specific genes consistent with its restricted expression pattern. The most significant matches were for the 5'-flanking regions of IA-1 and NeuroD/beta2 genes. Co-transfection of cells with either the full-length IA-1 or hEgr-1AD/IA-1DBD construct and IA-1 or NeuroD/beta2 promoter/CAT construct modulated CAT activity. These findings suggest that the IA-1 protein may be auto-regulated and play a role in pancreas and neuronal development, specifically in the regulation of the NeuroD/beta2 gene.

Figure 1

IA-1 functions as a transcriptional repressor in HeLa cells. Schematic representation of the various IA-1 fragments fused with the GAL4 DBD (amino acids 1–147) as an effector vector that was co-transfected with a reporter vector containing the GAL4 DNA-binding sequence, the thymidine kinase minimal promoter driven CAT reporter gene. The pSG424 vector contains the GAL4 DBD alone. The full-length IA-1 hybrid construct yielded 25% CAT activity of the pSG424 control vector. Amino acids 168–263 of the IA-1 protein possess repressor activity. A 263fs fragment represents a nucleotide deletion that causes a frame shift of the 1–263 amino acid sequence. Transient transfections were performed as described in the Materials and Methods in HeLa cells. The data are expressed as the percentage change as compared with the GAL4DBD construct alone (control). The CMV–βgal construct was used to normalize the transfection efficiency. The graph shows the average of four separate experiments and the SEM.

Figure 2

Competitive EMSA. IA-1 protein binds to a dominant C5 sequence. A competitive EMSA study revealed that the sequence located at the right-hand half of clone 5 accounts for the binding activity. The underlined sequence represents either half of the full-length C5 oligonucleotide (A). 500-fold molar excess cold competitor full-length C5 or the right hand half of the C5 sequence (C5R) competed away the specific protein/DNA complex. The left hand half sequence (C5L) had no effect on the protein/DNA complex (B). EMSA were performed as described in Materials and Methods.

Figure 3

Mutation analysis of the IA-1 consensus binding site. Two tandem repeats of the 12 bp C5 binding site (wild-type) as well as mutated sequences (Mut1, Mut2 and Mut3) were synthesized for the EMSA binding study. EMSA was performed as described in Materials and Methods. Mutated nucleotides are underlined. A stretch of G nucleotides is essential for IA-1 protein binding.

Figure 4

IA-1 promoter contains a consensus target sequence. The IA-1 promoter sequence (–103/–63 bp) binds to the IA-1 protein in EMSA analysis. The 500-fold cold IA-1p, C5 and C5R oligonucleotides compete efficiently but not the C5L probe showing that the interaction is specific (A). Different concentrations (10-, 50-, 100- and 500-fold) of cold excess IA-1 promoter sequence reveals dosage-dependent competition for GST–IA-1DBD protein binding (B).

Figure 5

Auto-regulation of the IA-1 gene expression. The –189/–16 bp IA-1 5′-upstream region, which contains the IA-1 target binding site, was cloned upstream of the E1bTATA/CAT reporter gene. Co-transfection of hEgr-1AD/IA-1DBD and the IA-1 –189/–16 bp-E1bTATA/CAT vector were performed as described in Materials and Methods. A stronger CAT activity was observed in insulinoma (β-TC-1) cells (A). The empty expression vector, the hEgr-1AD alone or the IA-1DBD construct had no effect on the CAT activity. Alternatively, we tested the IA-1 full-length suppressive activity using –426/+40 bp IA-1 pCAT and CMV–IA-1 cDNA expression vectors (B). Co-transfection of HeLa cells demonstrated 80% repression in IA-1 promoter activity. Transfections were performed in β-TC-1 or HeLa cells on three separate occasions. The data are expressed as fold induction or suppression over the control pcDNA3 (empty) expression vector. A CMV–βgal vector is used to normalize transfection efficiency. The graph represents the average of three separate experiments and SEM.

Figure 6

IA-1 zinc-fingers 2 and 3 are essential for transcriptional activity. Various zinc-finger constructs were fused in frame with hEgr-1 AD (a.a. 1–147). Co-transfection of zinc-finger mutants with the IA-1 –426/+40 bp promoter/pCAT3 reporter gene into β-TC-1 cells revealed that zinc-fingers 2 and 3 are the key motifs that contribute to the transcriptional activity. Zinc-finger 3 alone exhibits 50% of the control activity, whereas the combination of zinc-finger 3 and 4 only exhibits 25% of the control activity. The data are expressed as fold increase over the empty pcDNA3 expression vector. A CMV–βgal vector is used to normalize transfection efficiency. The graph represents the average of three separate experiments and SEM.

Figure 7

Target gene regulation by IA-1. EMSA analysis of potential target gene binding sites were performed using three consecutive repeats (12 bp)₃ of the putative IA-1 binding sites from the mouse Pax4, Pdx-1, Pax6 and NeuroD/β2 genes and incubated with GST–IA-1DBD protein. Both Pax4 and Pax6 sequences are in the reverse orientation with respect to the NeuroD/β2 site. The Pax6 and NeuroD/β2 sequences demonstrated binding activity to the GST–IA-1DBD protein (A). The 3X repeat sequences of the mouse Pax6 and NeuroD/β2 IA-1 binding sites were cloned upstream of the E1bTATA/CAT construct to assess the ability of IA-1 to modulate their activity (B). The hEgr-1AD/IA-1DBD fusion construct was used in the co-transfection experiments. A CMV–β-gal construct was used to normalize the transfection efficiency. The NeuroD/β2 binding site shows strong activity suggesting it is the target gene for the IA-1 transcription factor. Transient transfections were performed in β-TC-1 cells. The graph represents the average of three separate experiments and SEM.