Tracking live-cell single-molecule dynamics enables measurements of heterochromatin-associated protein-protein interactions

Abstract

Visualizing and measuring molecular-scale interactions in living cells represents a major challenge, but recent advances in single-molecule super-resolution microscopy are bringing us closer to achieving this goal. Single-molecule super-resolution microscopy enables high-resolution and sensitive imaging of the positions and movement of molecules in living cells. HP1 proteins are important regulators of gene expression because they selectively bind and recognize H3K9 methylated (H3K9me) histones to form heterochromatin-associated protein complexes that silence gene expression, but several important mechanistic details of this process remain unexplored. Here, we extended live-cell single-molecule tracking studies in fission yeast to determine how HP1 proteins interact with their binding partners in the nucleus. We measured how genetic perturbations that affect H3K9me alter the diffusive properties of HP1 proteins and their binding partners, and we inferred their most likely interaction sites. Our results demonstrate that H3K9 methylation spatially restricts HP1 proteins and their interactors, thereby promoting ternary complex formation on chromatin while simultaneously suppressing off-chromatin binding. As opposed to being an inert platform to direct HP1 binding, our studies propose a novel function for H3K9me in promoting ternary complex formation by enhancing the specificity and stimulating the assembly of HP1-protein complexes in living cells.

Graphical Abstract

Figure 1.

Models of HP1-associated complex assembly. (A) Swi6 and Chp2 interact with different sets of heterochromatin-associated factors at sites of H3K9me. (B) Model 1: off-chromatin assembly model. HP1 proteins and their binding proteins bind off-chromatin and then survey the genome to find H3K9me sites. (C) Model 2: on-chromatin assembly model. HP1 proteins form complexes with their binding partners exclusively at sites of H3K9me.

Figure 2.

H3K9me and HP1 expression regulate their binding and epigenetic silencing. (A) PAmCherry is fused to the N-terminus of Chp2 and expressed ectopically using a series of inducible promoters: nmt1, nmt41, and nmt81 (high, medium, and low expression levels, respectively). Bottom: Schematic of the Chp2 domains. CD: chromodomain (H3K9me recognition); H: hinge (nucleic acid binding); CSD: chromoshadow domain (dimerization interface). (B) Silencing assay using an ura4+ reporter inserted at the mat locus (Kint2::ura4). 10-fold serial dilutions of cells expressing Chp2 from different nmt promoters were plated on EMMC, EMMC + FOA and EMMC − URA plates. All PAmCherry fusion proteins are inserted at the leu1 + locus unless otherwise specified. (C) ChIP-seq of PAmCherry-Chp2 showing reads mapped to the tel1L locus, cen1 locus, and mating-type (mat) locus for Chp2 expressed from the nmt1, nmt41, and nmt81 (high, medium, and low expression levels, respectively). Data shown are reads per million aggregated in 1 kb windows, scaled by 0.01 for ease of visualization (D, E) Concentration dependence curves of quantified electrophoretic mobility shift assays (EMSA) using H3K9me0 (D) and H3K9me3 (E) mononucleosomes in WT Chp2 (black), Chp2 W199A (pink), and Chp2 I370E (blue). Error bars indicate SD (replicates N = 4). (F) Table summarizing the apparent binding affinity (K_1/2) and specificity values observed for Chp2-WT, Chp2 W199A and Chp2 I370E.

Figure 3.

Single-molecule tracking reveals Chp2 dynamics and kinetics. (A) NOBIAS identifies two distinct mobility states for PAmCherry-Chp2-low (expressed from the nmt81 promoter) in WT cells. Each colored point is the average single-molecule diffusion coefficient of PAmCherry-Chp2 molecules in that state sampled from the posterior distribution of NOBIAS inference at a saved iteration after convergence. Grey points are the previously reported PAmCherry-Swi6 single-molecule dynamics (21). (B) NOBIAS identifies two distinct mobility states for PAmCherry-Chp2-low in clr4Δ cells (cross) and three distinct states for PAmCherry-Chp2-low in Clr4 F449Y cells (colored points). Grey points are the PAmCherry-Chp2-low data (Figure 3A). (C) Inferred transition probabilities between the two mobility states of PAmCherry-Chp2-low from single-molecule tracking (Figure 3A). Diffusion coefficients, D, in units of μm²/s and weight fractions, π, are indicated. (D) Fine-grained chemical kinetic simulations with Bayesian Synthetic Likelihood algorithm. The reaction on/off rate is proposed and simulated at a 0.4-ms time interval to calculate the likelihood based on transition probabilities from C at the 40-ms experimental imaging time interval. (E) Inferred rate constants for PAmCherry-Chp2-low. (F) Schematic of single-molecule time-lapse imaging. The time-lapse period, , is the sum of the 200-ms integration time and the time delay. Five different time delays, , were introduced. (G) Dwell time distributions for PAmCherry-Chp2-low. The distributions are shown with fits to an exponential decay. Inset: linear fit (red dashed line) of vs. , from which the dissociation rate constant, , and the photobleaching rate constant, , are obtained. Errors bars are the standard deviation of the exponential decay fitting.

Figure 4.

H3K9me regulates HP1-associated protein–protein interactions. (A) Two-color imaging of cells expressing mNeonGreen-Swi6 and Epe1-PAmCherry. Swi6 and Epe1 are expressed from their endogenous promoters. Green colorbar: Swi6-mNeonGreen intensities; Red colorbar: reconstructed Epe1-PAmCherry density map. Both color channels are normalized to the maximum pixel intensity. (B–D) NOBIAS identifies distinct mobility states for Epe1-PAmCherry. Each colored point is the average single-molecule diffusion coefficient of PAmCherry-Epe1 molecules in that state sampled from the posterior distribution of NOBIAS inference at a saved iteration after convergence in WT cells (B), clr4Δ cells (C) and highly overexpressed PAmCherry-Epe1-high (from the nmt1 promoter) (D). Grey points are the previously reported PAmCherry-Swi6 single-molecule dynamics in corresponding cells (21). (E) Dwell time distributions for Epe1-PAmCherry expressed under its endogenous promoter. The distributions are shown with fits to an exponential decay. Inset: linear fit (red dashed line) of vs. , from which the dissociation rate constant, , and the photobleaching rate constant are obtained. Error bars are from the standard deviation of exponential decay fitting. (F) Top: transition probabilities between the two mobility states of Epe1-PAmCherry from (B). Diffusion coefficients, D, in units of μm²/s and weight fractions, π, are indicated. Bottom: Inferred rate constants for Epe1-PAmCherry from the fine-grained chemical kinetic simulation.

Figure 5.

Mit1 and Clr3 preferentially form complexes at heterochromatin. (A, B) NOBIAS identifies three distinct mobility states for PAmCherry-Mit1 and PAmCherry-Clr3, both expressed from the nmt81 promoter. Each point is the average single-molecule diffusion coefficient of PAmCherry-Mit1 molecules in WT cells (A) or PAmCherry-Clr3 molecules in WT cells (B, colored points) and in mit1Δ cells (B, grey points) that state sampled from the posterior distribution of NOBIAS inference at a saved iteration after convergence. (C, D) Transition probabilities between the three mobility states of PAmCherry-Mit1 (C) and PAmCherry-Clr3 (D) from NOBIAS. The arrow widths are proportional to the transition probabilities; the transitions with probability below 0.8% are not drawn. Diffusion coefficients, D, in units of μm²/s and weight fractions, π, are indicated. (E, F) NOBIAS identifies three distinct mobility states for PAmCherry-Mit1 molecules from nmt81 promoter in chp2Δ cells (E) and in chp2Δswi6Δ cells (F). Each point is the average diffusion coefficient and weight fraction of that mobility state sampled from the posterior distribution of NOBIAS inference at a saved iteration after convergence. The cross colors and ranges show data of PAmCherry-Mit1 molecules in WT cells (A).

Figure 6.

H3K9me enables HP1-directed SHREC complex assembly. (A, B) NOBIAS identifies three distinct mobility states for PAmCherry-Mit1 molecules (A) and PAmCherry-Clr3 molecules (B) expressed from the nmt81 promoter in clr4Δ cells. Each point is the average diffusion coefficient and weight fraction of that mobility state sampled from the posterior distribution of NOBIAS inference at a saved iteration after convergence. The cross colors and ranges show data for PAmCherry-Mit1 molecules (Figure 5A) or PAmCherry-Clr3 molecules (Figure 5B) from the nmt81 promoter in WT cells. (C, D) Reconstructed single-molecule density map for PAmCherry-Mit1 (top) and PAmCherry-Clr3 (bottom) expressed from the nmt81 promoter in WT cells (C) and clr4Δ cells (D). Dashed lines: approximate S. pombe cell outlines; solid circles: approximate nucleus borders. (E) Ripley's H analysis for steps from all states for Mit1 and Clr3 expressed from the nmt81 promoter in WT cells and clr4Δ cells. Mit1 and Clr3 in clr4Δ cells have lower Ripley's H(r) values than Mit1 and Clr3 in WT cells.