DNA-binding landscape of IRF3, IRF5 and IRF7 dimers: implications for dimer-specific gene regulation

Abstract

Transcription factors IRF3, IRF5 and IRF7 (IRF3/5/7) have overlapping, yet distinct, roles in the mammalian response to pathogens. To examine the role that DNA-binding specificity plays in delineating IRF3/5/7-specific gene regulation we used protein-binding microarrays (PBMs) to characterize the DNA binding of IRF3/5/7 homodimers. We identified both common and dimer-specific DNA binding sites, and show that DNA-binding differences can translate into dimer-specific gene regulation. Central to the antiviral response, IRF3/5/7 regulate type I interferon (IFN) genes. We show that IRF3 and IRF7 bind to many interferon-stimulated response element (ISRE)-type sites in the virus-response elements (VREs) of IFN promoters. However, strikingly, IRF5 does not bind the VREs, suggesting evolutionary selection against IRF5 homodimer binding. Mutational analysis reveals a critical specificity-determining residue that inhibits IRF5 binding to the ISRE-variants present in the IFN gene promoters. Integrating PBM and reporter gene data we find that both DNA-binding affinity and affinity-independent mechanisms determine the function of DNA-bound IRF dimers, suggesting that DNA-based allostery plays a role in IRF binding site function. Our results provide new insights into the role and limitations of DNA-binding affinity in delineating IRF3/5/7-specific gene expression.

Biophysical Principles of Lineage Factor PU.1 Binding Revealed by NextPBMs

“…As all nuclear proteins are present in a nextPBM experiment, the method assays, in parallel, the DNA-binding of a TF in potentially multiple distinct protein complexes. We use a SNV approach (Andrilenas et al, 2018) to generate binding logos and thereby probe the biophysical configuration of the complexes bound to different DNA sequences. Applying this approach to sites bound by PU.1 cooperatively and non-cooperatively allowed us to recover the two known PU.1 binding logos from a single experiment, and to quantitatively characterize the sequence determinants of cooperative PU.1:IRF8 binding at nucleotide resolution.…”

“…To analyze the sequence determinants of PU.1 binding we queried the base preferences along select binding sites and generated DNA-binding logos using a single-nucleotide variant (SNV) approach (Materials and Methods) (Andrilenas et al, 2018). Briefly, we measure the binding of PU.1 to 20 bp-long seed sequences and all 60 SNV sequences; logos were then generated directly from binding scores to each SNV sequence ( Figure 3A-D, Methods and Materials).…”

IRF5 regulates unique subset of genes in dendritic cells during West Nile virus infection

“…These observations also corroborate results from other studies where IRF5 was found to bind "GAAA" half sites, rather than the "GAAANNGAAA" full site as other IRF factors prefer. 44 Another IRF5 ChIP study also identified composite (G)GAAA motifs. 45 We note that this GAAA motif was not identified in the replicate-combined samples using MEME-ChIP or the individual replicates, due to low peak numbers (data not shown).…”

Customizable high-throughput platform for profiling cofactor recruitment to DNA to characterize cis-regulatory elements and screen non-coding single-nucleotide polymorphisms

“…Each REF/SNP pair was screened for differential recruitment of p300, SMARCA4, TBL1XR1, RBBP5, and GCN5 as well as differential binding of representative ETS factor PU.1 nextPBM experimental results were preprocessed as above. Z-scores were obtained for each probe as previously described 57 Motif modeling using single variant (SV) probes was performed as previously described 17,57,61 for the SNP-QTL sites profiled in detail using CASCADE. For the multi-tile design used to model extended loci such as the LPS-responsive CXCL10 promoter segment, a weighted mean approach was applied as follows to overlapping positions in order to integrate 21 results across sequential tiles: all variant probes corresponding to a given nucleotide at a given position within the promoter segment were averaged using each probe's corresponding seed (reference genomic) z-score as a weight.…”

“…For CASCADE profiling of the SNP-QTLs, a minimal seed z-score of 1.5 was enforced for motif analysis. Recruitment energy matrices obtained from CASCADE cofactor profiling (fluorescence intensity z-scores) were converted to a probability-based matrix using the Boltzmann distribution as previously described 57 to be more directly comparable to previous TF binding models:…”