Disorder-mediated interactions target proteins to specific condensates

Abstract

Selective compartmentalization of cellular contents is fundamental to the regulation of biochemistry. Although membrane-bound organelles control composition by using a semi-permeable barrier, biomolecular condensates rely on interactions among constituents to determine composition. Condensates are formed by dynamic multivalent interactions, often involving intrinsically disordered regions (IDRs) of proteins, yet whether distinct compositions can arise from these dynamic interactions is not known. Here, by comparative analysis of proteins differentially partitioned by two different condensates, we find that distinct compositions arise through specific IDR-mediated interactions. The IDRs of differentially partitioned proteins are necessary and sufficient for selective partitioning. Distinct sequence features are required for IDRs to partition, and swapping these sequence features changes the specificity of partitioning. Swapping whole IDRs retargets proteins and their biochemical activity to different condensates. Our results demonstrate that IDR-mediated interactions can target proteins to specific condensates, enabling the spatial regulation of biochemistry within the cell.

Keywords: IDR; biomolecular condensates; condensate composition; condensate function; intrinsically disordered regions; nuclear organization; specificity.

Figure 1.. IDRs of FUS and MED1 both form condensates, but with distinct chemical features

(A) Consensus MobiDB disorder prediction plots for MED1 and FUS, IDRs are shaded orange or violet. 0 corresponds to lowest prediction of disorder and 1 corresponds to highest prediction of disorder. (B) Parametric heatmap for MED1 and FUS IDR regions representing a log2 of enrichment ratio relative to each other. (C) mEGFP-tagged recombinant protein. Droplets in 10% PEG-8000. Scale, 2 μm. (D) Heatmaps representing Q, FYW, KR, and DE amino acid content for IDR regions. (E) Representative Images of mCherry-MED1^IDR of mCherry-FUS^IDR transfected together with either CFP-LacI-MED1^IDR, CFP-LacI-FUS^IDR, or CFP-LacI (no IDR) (see Figure S1B). Scale bar, 5 μm. Inset scale bar, 1 μm. (F) Box plot (min-max) of partition coefficients for experiments in 1E. p-values represent one-way ANOVA vs. no IDR CFP-LacI control, n = 8. See also Figure S1 and Table S1.

Figure 2.. Condensates of MED1^IDR and FUS^IDR partition different proteins

(A) Schematic of in situ Lac-APEX method (cell-based method). (B) Schematic of in vitro extract method (cell-free method). (C) Representative images of cell lines expressing mEGFP-LacI-APEX2 noIDR, MED1^IDR, or FUS^IDR construct at the lac-array locus (top row), biotin signal visualized by Neutravidin-Rhodamine Red-X (bottom row) with and without biotin-phenol incubation (See Figure S2D). Scale bar, 5 μm. (D) Bar chart quantification of Neutravidin signal in (C) with and without biotin-phenol incubation for each cell line expressing mEGFP-LacI-APEX2 noIDR, MED1^IDR, or FUS^IDR. P-values represent Mann-Whitney test, n = 10. (E) Coomassie stained SDS-PAGE gel of Nuclear Extract (NE), pellet fractions, and recombinant purified MED1^IDR, or FUS^IDR. (F) Schematic of overlap analysis (G) Heatmap of pellet/supernatant (P/S) percentile rank (see methods) for uniquely partitioned proteins and overlapped proteins of MED1^IDR and FUS^IDR condensates. This heatmap is divided into three classes, differentially partitioned proteins for MED1^IDR condensates (72), differentially partitioned proteins for FUS^IDR condensates (61), and commonly partitioned proteins (non-DP). (H) Immunoblot of pellet fraction of indicated sample with indicated antibody. (I) Immunofluorescence using the indicated antibody. scale, 5 μm. (J) Quantification of (I). p-values represent one-way ANOVA vs. no IDR CFP-LacI control. See also Figure S2 and Table S2.

Figure 3.. IDRs are necessary and sufficient for partitioning

(A) Bar plots showing (left) enrichment of intrinsically disordered regions and Pfam domains in differentially partitioned proteins compared to proteome (P-value<0.0001) (right). (B) Same as (A) but for FUS^IDR partitioned proteins. (C) Schematic of experiment (left). Representative images of mCherry-IDR or delta-IDR fusions of candidates identified in MED1^IDR dataset tested for partitioning into CFP-lacI-MED1^IDR focus (right). Scale bar, 1 μm. (D) Box plot (min-max) of partition coefficients for experiments in (C), n = 5. p-values, two-tailed unpaired t test. (E) Same as (C) but for FUS^IDR. (F) Same as (D) but for FUS^IDR. See also Figure S3.

Figure 4.. IDRs are selective in their partitioning

(A) Schematic of experiment in (B). (B) Representative images of mCherry-IDR fusions (top row) from both datasets tested for partitioning in CFP-LacI-MED1^IDR focus (bottom row). Scale bar, 1μm. (C) Box plot (min-max) of partition coefficients for experiments in (B). p-values represent one-way ANOVA vs. no IDR control, n = 5. (D) Schematic of experiment in (E). (E) Representative images of mCherry-IDR fusions (top row) from both datasets tested for partitioning in CFP-LacI-FUS^IDR (bottom row) focus. Scale bar, 1μm. (F) Box plot (min-max) of partition coefficients for experiments in (E). p-values represent one-way ANOVA vs. no IDR control, n = 5. (G) Schematic of experiment in (H). (H) Representative images of cells transfected with mCherry-p300^IDR (left) and miRFP670-CDK11B^IDR (middle) to test for partitioning into CFP-LacI-MED1^IDR focus (right). Scale bar, 1μm. (I) Graph showing line profile of relative mCherry and miRFP670 signal centered at CFP foci from experiments in (H), n = 5. (J) Schematic of experiment in (K). (K) Representative images of cells transfected with mCherry-p300^IDR (left) and miRFP670-CDK11B^IDR (middle) to test for partitioning into CFP-LacI-FUS^IDR focus (right). Scale bar, 1μm. (L) Graph showing line profile of relative mCherry and miRFP670 signal centered at CFP foci from experiments in (K), n = 5. See also Figure S4.

Figure 5.. Amino acid features enriched in differentially partitioned proteins are required for partitioning

(A) Comparative analysis shows partitioned IDRs are enriched for Fraction of Charge Residues (FCR) and basic residues. (B) Comparative analysis shows MED1^IDR partitioned IDRs are enriched for acidic amino acid fraction and kappa. (C) Comparative analysis shows FUS^IDR partitioned IDRs are enriched for P, Q, K, R, and Y residues. (D) Representative images of mCherry-CDK11B^IDR WT or mutant E to A (top row) tested for partitioning in CFP-LacI-MED1^IDR focus (bottom row). Scale bar, 1μm. (E) Box plot (min-max) of partition coefficient for experiments in (D). p-values represent one-way ANOVA vs. no IDR control, n = 5. (F) Representative images of mCherry-EWSR1^IDR WT or mutant PQ to A (top row) tested for partitioning in CFP-LacI-FUS^IDR focus (bottom row). Scale bar, 1μm. (G) Box plot (min-max) of partition coefficient for experiments in (F). p-values represent one-way ANOVA vs. no IDR control, n = 5. See also Figure S5 and Table S3.

Figure 6.. Swapping amino acid features changes condensate specificity

(A) Schematic of expected results of changing sequence features of SPT6^IDR on partitioning. (B) Representative images of mCherry-SPT6^IDR WT or charge to PQ variant (top row) tested for partitioning into either CFP-LacI-MED1^IDR or CFP-LacI-FUS^IDR (bottom row). Scale bar, 1μm. (C) Box plot (min-max) of partition coefficient for experiments in (B). p-values represent one-way ANOVA vs. no IDR control, n = 7. (D) Schematic of expected results of changing sequence features of EWSR1^IDR on partitioning. (E) Representative images of mCherry-EWSR1^IDR WT or PQY to charge variant (top row) tested for partitioning into either CFP-LacI-MED1^IDR or CFP-LacI-FUS^IDR (bottom row). Scale bar, 1μm. (F) Box plot (min-max) of partition coefficient for experiments in (E). p-values represent one-way ANOVA vs. no IDR control, n = 7. See also Figure S6.

Figure 7.. Swapping IDRs redirects biochemistry

(A) Schematic of CDK11 partitioning and consequence on phosphorylation. (B) Representative images of immunofluorescence using an antibody for RNA polymerase II subunit RBP1 specific to CTD in CFP-LacI no IDR, -MED1^IDR, or -FUS^IDR foci in Lac array cells. (Right) Box plot (min-max), p-values represent one-way ANOVA vs. no IDR control, n = 5. (C) Representative images of immunofluorescence using phospho-specific antibodies for Serine 5 phosphorylation (S5ph) of the CTD of RNA Polymerase II (top row) within control CFP-LacI no IDR, -MED1^IDR, or -FUS^IDR foci in Lac array cells (bottom row). Scale bar, 5μm. (Left) Box plot (min-max) of S5ph signal enrichment within indicated LacI foci, p-values represent one-way ANOVA vs. no IDR control, n = 7. (Right) Box plot (min-max) of S5ph signal enrichment normalized to the average signal enrichment of total RNA Pol II from experiments presented in (B), p-values represent unpaired t-test, n = 7. (D) Schematic of predicted partitioning and consequence to phosphorylation of CDK11 kinase domain lacking its N-terminal IDR (CDK11^ΔIDR or CDK11^kinase) or of CDK11 chimera constructed by fusing CDK11^kinase and p300^IDR. (E) Representative images of immunofluorescence using phospho-specific antibodies for Serine 5 phosphorylation (S5ph) of the CTD of RNA Polymerase II (middle row), for Lac array cells expressing mCherry- no IDR, CDK11, CDK11 chimera, or kinase-dead (KD) chimera (top row) within CFP-LacI-FUS^IDR foci (bottom row). Whole-cell scale bar, 5μm. Inset scale bar, 1 μm. (F) Box plot (min-max) of mCherry constructs expressed in (E) (left) and S5ph signal enrichment (right) within CFP-LacI-FUS^IDR foci for experiments in (E). p-values represent one-way ANOVA vs. no IDR control, n = 5. (G) Schematic of predicted functional consequence of retargeted phosphorylation by CDK11 chimera. (H) Representative images of RNA FISH using a probe specific to the 2–6-3 Lac reporter to detect transcriptional activity (top row) for Lac array cells expressing mCherry- no IDR, CDK11, CDK11 chimera, or kinase-dead (KD) chimera (middle row) within CFP-LacI-FUS^IDR foci (bottom row). Whole-cell scale bar, 5μm. Inset scale bar, 1 μm. (I) Box plot (min-max) of RNA FISH probe signal for experiments in (H). p-values represent one-way ANOVA vs. no IDR control, n = 5.