DEAF1 Binds Unmethylated and Variably Spaced CpG Dinucleotide Motifs

Editor: Osman El-Maarri, University of Bonn, Institut of experimental hematology and transfusion medicine, Germany Received: September DEAF1 Binds Unmethylated and Variably Spaced CpG Dinucleotide Motifs Philip J. Jensik 0 Jesse D. Vargas. 0 Sara N. Reardon 0 Shivakumar Rajamanickam 0 Jodi I. Huggenvik 0 Michael W. Collard 0 0 Department of Physiology, Southern Illinois University School of Medicine , Carbondale, Illinois , United States of America DEAF1 is a transcriptional regulator associated with autoimmune and neurological disorders and is known to bind TTCG motifs. To further ascertain preferred DEAF1 DNA ligands, we screened a random oligonucleotide library containing an ''anchored'' CpG motif. We identified a binding consensus that generally conformed to a repeated TTCGGG motif, with the two invariant CpG dinucleotides separated by 6-11 nucleotides. Alteration of the consensus surrounding the dual CpG dinucleotides, or cytosine methylation of a single CpG half-site, eliminated DEAF1 binding. A sequence within the Htr1a promoter that resembles the binding consensus but contains a single CpG motif was confirmed to have low affinity binding with DEAF1. A DEAF1 binding consensus was identified in the EIF4G3 promoter and ChIP assay showed endogenous DEAF1 was bound to the region. We conclude that DEAF1 preferentially binds variably spaced and unmethylated CpG-containing half-sites when they occur within an appropriate consensus. - Introduction Deformed Epidermal Autoregulatory Factor 1 (DEAF1) is a transcription factor that binds to TTCG half-sites through a centralized DNA binding SAND (Sp-100, AIRE, NucP41/75 and DEAF1) domain [13]. The SAND domain contains a positively charged region encompassing a conserved KDWK motif [3]. An adjacent zinc finger domain and nuclear localization signal are necessary for DEAF1-DNA interactions [4]. Transcriptionally, DEAF1 displays dual activity, repressing its own promoter activity while activating other promoters such as Eif4g3 [3, 5, 6]. DEAF1-DEAF1 and DEAF1-Ku70 protein interactions also occur through the SAND domain [4, 7]. DEAF1 contains a nuclear export signal that acts as part of a second DEAF1-DEAF1 and DEAF1-LMO4 protein interaction domain [4, 810]. A C-terminal cysteine rich MYND (Myeloid translocation protein 8, Nervy, and DEAF1) domain likely mediates other protein-protein interactions [11]. Specific mutations in the SAND domain of the DEAF1 gene result in moderate to severe non-syndromic intellectual disability in humans [6, 12]. These mutations eliminate or greatly reduce both DEAF1 interactions with TTCG-containing DNA sequences and DEAF1 transcriptional repression of its own promoter [6]. DEAF1 is also linked to human mood disorders [1316], cancer [17, 18], autoimmune disorders [5, 19] and interferon-b production [20]. DEAF1 deficiency leads to neural tube closure defects in mice [21] and early embryonic arrest in Drosophila [22]. Deletion of Deaf1 in mouse brain results in an anxiety-like phenotype and causes severe deficits in 24-hour contextual memory [6]. In our previous study, a degenerate random oligonucleotide library was used to identify TTCG motifs in DEAF1-binding sequences [2]. Subsequently, Burnett et al. [23] demonstrated that introduction of an anchored CpG half-site core into a degenerate oligonucleotide library allowed identification of the optimal spacing and preferred sequences surrounding the CpG-containing half-sites for the SAND domain-containing glucocorticoid modulatory element binding 1/2 (GMEB1/2) protein. The objectives of this study were to: 1) further delineate the DNA consensus sequence required for DEAF1 binding using affinity selection of a CpG-anchored oligonucleotide library, 2) assess the effects of CpG methylation on DEAF1-DNA interactions, and 3) characterize the binding of DEAF1 to a sequence within the EIF4G3 promoter. Increased understanding of DNA sequences that DEAF1 can or cannot bind should aid in identifying potential DEAF1 target genes and provide insight into their regulation in normal biology and DEAF1-related disease. Materials and Methods Plasmids GST-DEAF1 and DEAF1-FLAG constructs have been previously described [4] and were derived from human DEAF1 cDNA (accession number AF049459). Purification of DEAF1 proteins Full-length recombinant bacterial expressed GST-DEAF1 and HEK293T expressed DEAF1-FLAG proteins were purified as previously described [4, 7]. Relative purities of the proteins are shown in S1 Figure. DEAF1 DNA Consensus Selection DEAF1 affinity selection of DNA sequences was similar to that previously described [2] using GST-DEAF1 and DEAF1-FLAG proteins, but was modified as in [23] to include an anchored CpG dinucleotide in degenerate oligonucleotides and to also include an electrophoretic mobility shift assay (EMSA) for affinity purification of DEAF1-DNA complexes. The degenerate oligonucleotide library was made with the following three oligonucleotides: 63-mer-59-CTGCTGGATCCTGCAGCTCTGAGN3CGN13GTCTGACAAGCTTCTAGAGTCA-39 Selection Forward Primer- 59-CTGCTGGATCCTGCAGCTCTGAG-39 Selection Reverse Primer- 59-TGACTCTAGAAGCTTGTCAGAC-39 The 63-mer oligonucleotide consists of an 18-mer of random nucleotides with an internal anchored CpG dinucleotide flanked by a 59 23-mer with a BamHI site and a 39 22-mer with a HindIII site (sites are underlined) to facilitate subcloning into pBluescript II KS+ vector. Briefly, GST-DEAF1 fusion protein immobilized on glutathione-agarose beads was incubated with the CpG anchored degenerate oligonucleotide library. Bound oligonucleotides were eluted and amplified by PCR using Selection Forward and Reverse primers and one-tenth of the PCR product was used in the next round of selection. A total of 6 rounds of selection were performed. Oligonucleotides in the final round of selection were amplified by PCR (10 cycles) with 32P-ATP to generate radiolabeled oligonucleotides that were used in a single round of EMSA selection with mammalian expressed DEAF1FLAG protein. DNA in the shifted bands were excised, amplified by PCR and digested with HindIII and BamHI prior to subcloning. DNA from individual colonies was sequenced on the CEQ8000 DNA sequencer (Beckman Coulter) using T7 and T3 primers. Consensus Analysis Sequences were compared and aligned using MEME (Multiple Em for Motif Elicitation) [24]. Resultant half-site sequences were further analyzed using DMatrix [25] and pictogram (http://genes.mit.edu/pictogram.html). Genomic scans were performed using RSA-Tools Genomic Scale PatternSearch [26] from the RSAT server, Brussels, Belgium. EMSA Binding Analysis The indicated 32P-Labeled dsDNA probes were synthesized by PCR and incubated with 200 ng of DEAF1-FLAG protein for 30 min at room temperature in 1x EMSA binding buffer with 1 mg of dA:dT. Complexes were separated on 5% native polyacrylamide gels and migration of the DNA probes were visualized by PhosphorImager. EMSA analysis using fluorescent IR700 and IR800 DNA probe (...truncated)