Zinc finger domains bind low-complexity domain polymers

Abstract

Self-association of low-complexity protein sequences (LC domains) is important for polymer formation. Several molecular chaperones are involved in the regulation of LC domain polymer formation. However, the mechanisms underlying cell recognition of LC domain polymers remain unclear. Here we show that zinc finger domains (ZnFs) bind LC domains of RNA-binding proteins in a cross-β polymer-dependent manner. ZnFs bound to LC domain hydrogels and suppressed LC domain polymer formation. Moreover, ZnFs preferentially recognize LC domains in the polymeric state. These findings suggest that ZnFs act as physiological regulators of LC domain polymer formation.

Fig. 1. ZnFs interact with LC domains.

a Venn diagram showing gene overlap in two comparisons, RNA-seq data of motor neurons derived from FUS mutant hiPSCs (FUS mutant) and ChIP-seq data of CoCl₂-treated Human aortic endothelial cell (HAEC_CoCl₂). b Percentages of genes encoding zinc binding and ZnF domains in the three datasets of FUS mutant, HAEC_CoCl₂, and human proteome. c Binding assays of GFP fusion KLF4 fragments to hydrogels derived from the mCherry fusion hnRNPA2 LC domain (mCh:A2-LC). d Binding assays of various GFP fusion ZnF (GFP:ZEB1 ZnF5-7, ZEB2 ZnF6-8, and ZNF512B ZnF3-5) to mCh:A2-LC. e Binding assays of GFP fusion ZEB2 ZnF6-8 (GFP:ZEB2 ZnF6-8) to various mCherry fusion LC domains (mCh:A2-LC, mCh:FUS-LC, and mCh:TDP43-LC). f Binding assays of GFP fusion ZEB2 fragments to mCh:A2-LC hydrogels.

Fig. 2. The binding mode of ZnF to LC domains is different from that to DNA.

a ¹H-¹⁵N NMR spectra of ¹⁵N-labeled ZEB2-ZnF8 (¹⁵N-ZnF8) with mCherry (mCh; left), mCherry-fused hnRNPA2 LC domain (mCh:A2-LC; middle), ZnF-binding DNA (right). Samples were prepared in 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 20 mM β-ME, 0.1 mM PMSF, 0.12 mM ZnCl₂, and 5% ²H₂O. Signals labeled by amino acid type and residue number showed significant changes. b Change in signal intensity of ¹⁵N-ZnF8 upon LC-binding. The intensity ratio is the signal intensity of LC-bound ZnF8 divided by the signal intensity of LC-free ZnF8. Significantly changed residues (intensity ratios less than mean−0.5 SD and mean−SD) are indicated by orange and red bars, respectively. Residues for which no signal was detected in the absence of LC domains are indicated by asterisks. The cysteine (C) and histidine (H) residues forming the C2H2-type ZnF motif are indicated by outlines. Errors of intensity ratios were estimated based on the measured background noise level. c Changes in signal intensity (upper) and chemical shift (lower) of ¹⁵N-ZnF8 upon DNA-binding. In the graph of intensity ratios, significantly changed residues (intensity ratios less than mean−0.5 SD and mean−SD) are indicated by orange and red bars, respectively. Errors of intensity ratios were estimated based on the measured background noise level. In the graph of chemical shift perturbations (CSP), significantly changed residues (CSP greater than mean+0.5 SD and mean+SD) are indicated by orange and red bars, respectively. Residues for which no signal was detected in the absence of LC domains are indicated by asterisks. The cysteine (C) and histidine (H) residues forming the C2H2-type ZnF motif are indicated by outlines. d Surface model of ZEB2-ZnF8 predicted by AlphaFold2. In the upper panel, residues that were significantly changed by binding to mCh:A2-LC (intensity ratios less than mean−0.5 SD and mean−SD) are shown in orange and red, respectively. In the lower panel, residues that were significantly changed by DNA-binding (CSP greater than mean+0.5 SD and mean + SD) are indicated by orange and red bars, respectively.

Fig. 3. The binding of ZnF affects the phase separation of LC domains.

a Microscopic images of MBP:FUS droplets in the presence of GFP or GFP fusion ZEB2 ZnF8 (GFP:ZnF8). Three independent experiments were performed. Scale bars are 5 μm. b Refractive index (RI) images of MBP:FUS droplets in the absence (MBP:FUS) and presence of ZEB2 ZnF peptides (ZnF8, ZnF7-8, and ZnF6-8). Scale bars are 5 μm. c Quantitative analysis of Fig. 3b (n = 133 (MBP:FUS), 156 (ZnF8), 153 (ZnF7-8), and 152 (ZnF6-8), respectively). Statistical significance was examined by one-way analysis of variance with Tukey’s honest significant difference post hoc test. All statistical tests were two-sided. A significant difference was observed only for MBP:FUS droplets in the presence of ZnF6-8 compared with MBP:FUS droplets as the control (***p < 0.001). Data represents as mean ± SD. The unit of study was a single droplet, and each n represents an individual droplet measured from images acquired in separate fields of view. d Thioflavin T assays of hnRNPA2 LC domain (A2-LC) in the absence (black) and presence of ZEB2 ZnF6-8 peptide (ZnF6-8, blue), ZEB2 ZnF7-8 peptide (ZnF7-8, green) and ZEB2 ZnF8 (red). Data represent means ± SD (n = 4 technical replicates).

Fig. 4. ZnF preferentially recognizes LC polymers over monomeric LC domains.

a Hydrogel binding assay for mCherry fusion hnRNPA2 LC domain (mCh:A2-LC) with GFP fusion hnRNPA2 LC domain (GFP:A2-LC) and GFP:A2-LC mutants (GFP:A2-LC_F291S, GFP:A2-LC_MM, GFP:A2-LC_MM2 and GFP:A2-LC_MM3). b Electron microscopic images of negatively stained mCh:A2-LC and its mutants. Negatively stained samples were taken from ten different areas on the grid, and representative images are shown. Scale bars are 200 nm. c Gel filtration chromatography of mCh:A2-LC (red), mCh:A2-LC_F291S (green), mCh:A2-LC_MM (blue), and mCherry (black). Each NMR sample (0.1 mM) was loaded onto a Superdex 200 column. d ThT assays of mCh:A2-LC (red), mCh:A2-LC_F291S (green), and mCh:A2-LC_MM (blue). Data represent means ± SEM (n = 5 technical replicates). The lag time (t_lag) was estimated based on procedures described in “Methods”. e ¹H-¹⁵N NMR spectra of ¹⁵N-labeled ZEB2 ZnF8 (¹⁵N-ZnF8) with mCherry fusion hnRNPA2 LC domain mutants (mCh:A2-LC_F291S and mCh:A2-LC_MM). Samples were prepared in 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 20 mM β-ME, 0.1 mM PMSF, 0.12 mM ZnCl₂, and 5% ²H₂O. f Change in signal intensity of ¹⁵N-ZnF8 in the addition of mCh:A2-LC (red), mCh:A2-LC_F291S (green), and mCh:A2-LC_MM (blue). The intensity ratio is the signal intensity of A2-LC-bound ZnF8 divided by the signal intensity of A2-LC-free ZnF8. Residues for which no signal was detected in the absence of LC domains are indicated by asterisks. The cysteine (C) and histidine (H) residues forming the C2H2-type ZnF motif are indicated by outlines. Errors of intensity ratios were estimated based on the measured background noise level. In the upper panel, the NMR signals corresponding to the three residues K1059, G1061, and Y1070 are shown. g A model that ZnFs play as physiological regulators of LC domain polymer formation.