Structural Features of Transcription Factors Associating with Nucleosome Binding

Abstract

Fate-changing transcription factors (TFs) scan chromatin to initiate new genetic programs during cell differentiation and reprogramming. Yet the protein structure domains that allow TFs to target nucleosomal DNA remain unexplored. We screened diverse TFs for binding to nucleosomes containing motif-enriched sequences targeted by pioneer factors in vivo. FOXA1, OCT4, ASCL1/E12α, PU1, CEBPα, and ZELDA display a range of nucleosome binding affinities that correlate with their cell reprogramming potential. We further screened 593 full-length human TFs on protein microarrays against different nucleosome sequences, followed by confirmation in solution, to distinguish among factors that bound nucleosomes, such as the neuronal AP-2α/β/γ, versus factors that only bound free DNA. Structural comparisons of DNA binding domains revealed that efficient nucleosome binders use short anchoring α helices to bind DNA, whereas weak nucleosome binders use unstructured regions and/or β sheets. Thus, specific modes of DNA interaction allow nucleosome scanning that confers pioneer activity to transcription factors.

Keywords: Ascl1; FoxA; NHLH2; Pu.1; RBPJ; TFAP2A; nucleosome binding; pioneer transcription factor; protein microarray.

Figure 1. Endogenous TF-Nucleosomal Targets Assemble into Stable Nucleosomes In Vitro

(A) Schematic diagram showing genomic data processing for the identification of TFs nucleosomal targets. (B-G) ChIP-seq profile for reprogramming TFs at identified nucleosomal targets and 3D representation of the DNA sequences used for nucleosome assembly containing TFs canonical motifs indicated (yellow). (B) FOXA1, GATA4 and HNF1α ChIP-seq (red) in liver and MNase-seq profile (green) in fibroblasts across the ALBN1 enhancer within the displayed genomic location. (C) 3D representation of the 160 bp- ALBN1-DNA. (D) ASCL1 and BRN2 ChIP-seq (red) at 48 hr induction in fibroblast and MNase-seq profile in fibroblasts near the NRCAM gene. (E) 162 bp NRCAM-DNA. (F) PU1, CEBPα, and CEBPβ ChIP-seq in macrophages and MNase-seq profile in fibroblasts near the CX3CR1 gene. (G) 162 bp- CX3CR1-DNA. (H-J) DNase-I footprinting showing the protection of (H) ALBN1-DNA, (I) NRCAM-DNA, (J) CX3CR1-DNA before and after nucleosome reconstitution in vitro. Electropherograms generated by digesting 5’−6 FAM end-labeled free DNA (top panel) and nucleosomes with low DNaseI (middle panel) and high DNaseI (bottom panel). Concentrations of DNase-I indicated. Dashed lines indicate central histone octamer protection within nucleosomes.

Figure 2. Reprogramming TFs bind Nucleosomes with Nanomolar Affinity

(A) Recombinant purified full-length TFs analyzed by SDS-PAGE and Coomassie staining. The factors are grouped by reprogramming to iHEP (induced hepatocytes), iN (induced neurons), and iMAC (induced macrophages). Recombinant single purification of ASCL1, E12α, and co-purification of ASCL1/E12α (right panel). (B-G) Representative EMSA showing the affinity of increasing amounts of TFs (B) FOXA1, (C) PU1, (D) ASCL1/E12α, (E) GATA4, (F) BRN2 and (G) CEBPα to Cy5-labelled DNA (lanes 1–6) and nucleosome (lanes 7–12). Black arrowheads indicate TF- DNA complexes. White arrowheads indicate TF-nucleosome complexes. (H) 2D plot of TFs dissociation constants for DNA (x-axis) and nucleosomes (y-axis). (I) Representative EMSA showing the affinity of CEBPβ WT, mutants T167D and S163D,T167D to CX3CR1-DNA and nucleosomes.

Figure 3. TFs Bind Nucleosomes with Specificity

(A-B) Representative EMSA of competition assays showing the affinity of recombinant (A) FOXA1, ASCL1/E12α and PU1 to ALBN1-NUC, NRCAM-NUC, and CX3CR1-NUC in the presence of 20-, 40- and 80-fold molar excess of specific competitor ( “s” lanes) or non-specific competitor (“ns” lanes) or absence of competitor (“-” lanes). (B) Same as (A) for BRN2 and CEBPα. (C) Representative WEMSA showing the binding of ASCL1/E12α to NRCAM-NUC. ASCL1/E12α :Nuc complex from EMSA were transferred onto a PVDF membrane (WEMSA) and blotted for H3, H2B, ASCL1 as indicated (the three panels on the right). White arrow heads indicate the observed TF-nucleosome complexes. (D-F) DNase-I footprinting electropherograms of 5’−6 FAM-labeled (D) NRCAM-NUC, (E) CX3CR1-NUC, and (F) ALBN1-NUC in absence (top) or presence of ASCL1/E12α (middle) or PU1 (bottom) end-labeled free DNA (top strand). 3D DNA representation (red) with each TF motif (yellow)..Filled circles, protections; open circles, enhancements. “deg. motif” = degenerate motif for PU1.

Figure 4. ZnF TF Zelda Bind to Nucleosomes

(A) Representative EMSA of ZLD showing affinity to short dsDNA probes a containing canonical ZLD-binding motif ( “s” lanes) or mutated motif ( “ns” lanes). (B) Graphical representation of ZLD motifs (yellow) identified on NRCAM-DNA sequence. (C) Representative EMSA showing the affinity of increasing amounts of recombinant ZLD to Cy5- NRCAM-DNA (lanes 1–5) and nucleosome (lanes 6–10). (D) Quantification competed fraction of ZLD:DNA (left panel) or ZLD:nucleosome (right panel) complexes by addition of molar excess of specific competitor ( “s” lanes), non- specific competitor (“ns” lanes) or absence of competitor (“-” lanes). Molar excess listed. (E) Representative WEMSA of ZLD :Nuc complex from EMSA transferred onto a PVDF membrane (WEMSA) and blotted for H3, H2B as indicated (the two panels on the right). White arrow heads indicate the observed TF-nucleosome complexes.

Figure 5. Systematic Assessment of Nucleosome Binding of Human TFs with Protein Microarrays

(A) Graphical scheme used to identify nucleosome interacting human TF using DNA and nucleosome probes binding to protein microarrays. (B) Cy5 fluorescence of TAP2-α on protein microarray printed spot in duplicate in absence (“-”) or hybridized with NRCAM-DNA or NRCAM-NUC. (C) Representative EMSA comparing the affinity of ASCL1/E12α with TFAP-2α to NRCAM-DNA or NRCAM-NUC. Black arrow heads indicate TF-DNA complexes. White arrow heads indicate the TF-nucleosome complexes. (D-E). Heatmap representations of TFs bound fractions to DNA (left) and nucleosome (right) by quantification from EMSA. (D) Clustered heatmap showing strong DNA and nucleosome binders (red-cluster 1), low DNA binders and (yellow-cluster 2), high DNA binders with low nucleosome affinity (blue-cluster 3) (E) Heatmap sorted on TF nucleosome bound fraction. TFs concentrations used on Figure S6A.

Figure 6. Strong Nucleosome Binding TFs Recognize DNA with Short Recognition α-Helixes

(A) Group I pioneer TFs DBDs crystal structures of FOXA3 (pdb 1VTN) (Clark, 1993), OCT4 (pdb 3L1P) (Esch et al., 2013), PU1 (pdb 1PUE) (Kodandapani et al., 1996), KLF4 (pdb 2WBS) (Schuetz et al., 2011) and, GATA3 (pdb 4HC9) (Chen et al., 2012). ASCL1 (SMR P50553). (B) Group IIA TFs with scissor-like DBDs crystal structures and extended recognition α- helixes of cMYC/MAX (pdb 1NKP) (Nair and Burley, 2003), CEBPα (pdb 1NWQ) (Miller et al., 2003), USF (pdb 1AN4) (Ferre-D’Amare et al., 1994), MYOG (SMR P15173), and CREM (SMR Q03060). (C) Group IIB TFs with immunoglobulin-like fold DBDs crystal structures of TBX1 (pdb 4A04) (El Omari et al., 2012), Brachyury (pdb 1XBR) (Muller and Herrmann, 1997), NF- kB p50 subunit (pdb 1SVC) (Muller et al., 1995) and, GAL4 (pbd 3COQ) (Hong et al., 2008). (D) TALE-PBC PBX1 (pdb 1PUF) (LaRonde-LeBlanc and Wolberger, 2003) and UBX (pbd 4UUS) (Foos et al., 2015) crystal structures showing scissor-like binding in dimer form with HOX TFs showing kink in recognition α-helixes (blue arrow).