Domain tethering impacts dimerization and DNA-mediated allostery in the human transcription factor FoxP1

Abstract

Transcription factors are multidomain proteins with specific DNA binding and regulatory domains. In the human FoxP subfamily (FoxP1, FoxP2, FoxP3, and FoxP4) of transcription factors, a 90 residue-long disordered region links a Leucine Zipper (ZIP)-known to form coiled-coil dimers-and a Forkhead (FKH) domain-known to form domain swapping dimers. We used replica exchange discrete molecular dynamics simulations, single-molecule fluorescence experiments, and other biophysical tools to understand how domain tethering in FoxP1 impacts dimerization at ZIP and FKH domains and how DNA binding allosterically regulates their dimerization. We found that domain tethering promotes FoxP1 dimerization but inhibits a FKH domain-swapped structure. Furthermore, our findings indicate that the linker mediates the mutual organization and dynamics of ZIP and FKH domains, forming closed and open states with and without interdomain contacts, thus highlighting the role of the linkers in multidomain proteins. Finally, we found that DNA allosterically promotes structural changes that decrease the dimerization propensity of FoxP1. We postulate that, upon DNA binding, the interdomain linker plays a crucial role in the gene regulatory function of FoxP1.

FIG. 1.

Domain topology and structure of human FoxP1. (a) A mostly disordered linker (lin) tethers ZIP (Leucine-Zipper) and FKH (Forkhead) domains in human FoxP proteins. Zooming into ZIP and FKH domains, we present the secondary structure topology (α-helix as round boxes, β-sheets as arrows, and coils as lines) and canonical dimerization mechanisms (antiparallel coiled-coil and three-dimensional domain swapping). (b) Tridimensional organization of the region ZIP-linker-FKH (ZIP-lin-FKH) in human FoxP1, as predicted by Alpha Fold.

FIG. 2.

Dimerization analysis of ZIP-lin and ZIP-lin-FKH. (a) Labeling strategy to determine the dimerization of the proteins and the three-dimensional domain swapping (3D-DS) of the ZIP-lin-FKH. We monitored the changes in anisotropy of a BODIPY-labeled single-cysteine mutant of both ZIP and ZIP-FKH proteins (in green) in the same position (C356) in the titration experiments, whereas the 3D-DS ability of the ZIP-lin-FKH was evaluated by single-molecule FRET by following the FRET efficiency between the donor (ATTO488) and the acceptor (ATTO647N) in C492 in each monomer. (b) Experimental molecular weight estimation of ZIP (molecular weight: 17 kDa) and LZ-FKH (25 kDa) at 10 µM of protein concentration. The value was extracted from a calibration curve using known proteins as reference (see the section titled “Material and Methods”). ZIP (olive) and LZ-FKH (blue) domains elute as a dimer. The arrow indicates the presence of a small population of high-order oligomers. (c) Dimer fraction comparison between ZIP and ZIP-FKH in each titration experiment. In both cases, 5 nM of each labeled protein was titrated with their respective unlabeled version. Each dimerization curve was measured in duplicate, showing the mean and its respective standard deviation in the text. The dissociation constant (Kd) value of the isolated FKH is shown in gray, as previously reported. (d) (left) Schematics for the identification of the double-labeled dimer by single-molecule FRET efficiency vs stoichiometry. Both donor (D, green) and acceptor (A, red) labeled the 3D-DS FKH* (middle) and ZIP-lin-FKH(C493) (right) were combined to adopt the double-labeled dimer. The stoichiometry parameter is used to identify the double-labeled dimer. An unlabeled protein concentration of 10 nM ZIP-lin-FKH was used to maintain the dimeric state. The FRET efficiency obtained was compared with the previous 3D-DS dimer of the isolated FKH* domain previously reported. Correction parameters and specific protocol for dimerization can be found in the section titled “Material and Methods” and in the supplementary material.

FIG. 3.

Stability and unfolding properties of ZIP-lin and ZIP-lin-FKH constructs. (a) Unfolding experiments using the isolated ZIP(C356)-lin (olive squares), ZIP(C356)-lin-FKH (blue squares), and ZIP-lin-FKH(C493) (blue circles). All the proteins were labeled with BODIPY-FL and used at 200 nM in each denaturant concentration. A two-state dimer ⇋ unfolded model was used to analyze the unfolding data for ZIP(C356) and ZIP(C356)-FKH, whereas a three-state dimer ⇋ intermediate ⇋ unfolded was used for the case of ZIP-FKH(C493). (b) Protein concentration effect on the unfolding transition of ZIP-FKH. The unfolding at 200 nM (a) was compared with the data using 10 µM of unlabeled ZIP-FKH and monitored by tryptophan fluorescence. (c) Chemical unfolding assay for ZIP(C356)-lin (olive) and ZIP(C356)-lin-FKH (blue). The labeled proteins were maintained at 5 nM of protein concentration to favor the monomer. The unfolding of the monomer ZIP-lin was fitted to a three-state monomer ⇋ intermediate ⇋ unfolded mechanism (see the section titled “Material and Methods”). In contrast, the monomer ZIP-lin-FKH was fitted to a two-state monomer ⇋ unfolded mechanism. All unfolding experiments were carried out in duplicates. Means and standard deviations are plotted.

FIG. 4.

rxDMD simulations for monomeric FoxP1 ZIP, ZIP-lin, and ZIP-lin-FKH. Energy landscape [Potential Mean Force (PMF)] against R_g and α-helical content of the different ZIP constructs: ZIP domain alone (a), ZIP-lin (b), and ZIP-lin-FKH (c), showing the energy minima for the predicted native (N), intermediates (I, 1, and 2), and the unfolded (U) states. A representative structure is shown for each energy minimum (basin) (see the section titled “Material and Methods”). For easy understanding, the ZIP-linker region is colored in olive, whereas the FKH domain is colored in gray. The inset in (a) corresponds to the zoom of the ZIP’s PMF plot to better show the energy minima using the same x axis scale.

FIG. 5.

Allostery effects in the ZIP and FKH domains upon binding to DNA. (a) DNA binding assays with the isolated FKH (gray circles) or the ZIP-FKH (blue circles). A labeled DNA ligand (see the section titled “Material and Methods”) was incubated with different concentrations of the specified wild-type protein (ZIP-lin-FKH and FKH). The DNA’s fluorescence anisotropy was determined at 525 nm upon excitation at 480 nm. Fluorescence anisotropy values are normalized to the value without protein. (b) Two-dimensional histogram of r_G vs ${⟨{\tau }_{\mathit{BODIPY}}⟩}_f$ for the ZIP(C356)-lin-FKH and ZIP-lin-FKH(C493) BODIPY labeled constructs under free or DNA-bound conditions. The dark and light lines represent the sample’s high and low anisotropy extracted by fitting the data using a time-resolved analysis with two rotational correlation components, giving high (r_High) and low (r_Low) anisotropies. (c) Rotational times and its respective fraction for each anisotropy component and respective sample (r_High and r_Low). Dark and light bars represent the high and low anisotropy components under free (green) and DNA-bound (yellow) conditions, as determined from the Perrin equation (see the section titled “Material and Methods”). (d) Dimerization analysis with DNA. 5 nM of the labeled ZIP-lin-FKH(C492) monomer was titrated with different concentrations of wild-type ZIP-lin-FKH without (free) or with 10 nM of unlabeled DNA. Both assays were normalized to the fluorescence anisotropy of the labeled protein without the unlabeled protein. The titration curves were carried out in duplicate. Means and standard deviations are plotted.

FIG. 6.

A proposed model of the impact of the tethering of ZIP-lin-FKH in protein properties. By our experimental and rxDMD, we predict that the equilibrium between two major conformations determines the ability of FoxP1 to dimerize. The closed conformation (high rotational correlation time in smFA) acts as an autoinhibited state for dimerization. The loss of interdomain contacts should be required to allow for coiled-coil dimerization. This open-closed equilibrium is maintained under dimeric conditions, and the presence of DNA stabilizes the dimer by increasing the closed conformation.