A Liquid to Solid Phase Transition Underlying Pathological Huntingtin Exon1 Aggregation

Abstract

Huntington's disease is caused by an abnormally long polyglutamine tract in the huntingtin protein. This leads to the generation and deposition of N-terminal exon1 fragments of the protein in intracellular aggregates. We combined electron tomography and quantitative fluorescence microscopy to analyze the structural and material properties of huntingtin exon1 assemblies in mammalian cells, in yeast, and in vitro. We found that huntingtin exon1 proteins can form reversible liquid-like assemblies, a process driven by huntingtin's polyQ tract and proline-rich region. In cells and in vitro, the liquid-like assemblies converted to solid-like assemblies with a fibrillar structure. Intracellular phase transitions of polyglutamine proteins could play a role in initiating irreversible pathological aggregation.

Keywords: aggregation; electron tomography; fluorescence microscopy; huntingtin exon1; phase transition; polyQ.

Graphical abstract

Figure 1

Aggregation of HTT_ex1 Proteins Can Involve a Conversion between Distinct Macroscopic Assemblies (A) Domain organization of HTT_ex1 constructs in this study. (B) Representative confocal maximum intensity projections of bright and dim 43QP-GFP assemblies. Scale bar, 10 μm. (C and D) Time-lapse fluorescence microscopy of 43QP-GFP aggregation without (C), and with (D), a visible intermediate dim assembly. Orange arrows: bright assembly formation. Blue asterisk: coalescence of dim assemblies. Scale bar, 10 μm. (E) Quantification of aggregation events occurring without (orange) and with (orange/blue) visible intermediate dim assemblies. n > 92 aggregation events per construct from three independent experiments. p = 0.0003, chi-square. (F) FRAP experiment showing high HTT_ex1 mobility in dim assemblies but not in bright assemblies. Scale bar, 3 μm. (G) Averaged FRAP recovery curves. Shaded areas represent 95% confidence interval (CI). Dim assemblies estimated mobile fraction = 84%, 95% CI: 83%–85%, n = 20; bright assemblies estimated mobile fraction = 10%, 95% CI: 10%–11%, n = 20. See also Table S1. (H) EM projection image of a 43QP-GFP bright assembly (orange dashes). Higher-magnification image (inset) shows a network of fibrillar structures. (I) 43QP-GFP bright assembly (orange) and dim assemblies (blue). Higher-magnification image (inset) shows fibrillar structures emanating from the dim assemblies. Low-magnification scale bar, 5 μm; high-magnification scale bar, 500 nm.

Figure 2

Probing HTT_ex1 Assembly States in Yeast (A) Widefield microscopy of 43QP-GFP assemblies formed in [RNQ+] cells (green arrows), and [rnq−] cells (blue and orange arrows, showing dim and bright assemblies, respectively). Scale bar, 5 μm. (B) Averaged FRAP recovery curves. Shaded areas: 95% confidence intervals. See also Table S1. (C) Estimated mobile fractions of assemblies based on (B). Bright [rnq−] versus dim [rnq−] p = 6.3 × 10–13; bright [rnq−] versus [RNQ+] p = 0.003; dim [rnq−] versus [RNQ+] p = 3.2 × 10–12; Welch’s two-sample t test. (D) Photo bleaching of a dim assembly (red outline) in a yeast cell (gray outline) containing a second, coexisting dim assembly (purple outline). Scale bar, 3 μm. (E) Quantification of data in (D). Solid lines show mean fluorescence intensities in the corresponding colored regions in (D). Dotted lines highlight initial fluorescence levels in the corresponding colored regions. See also Table S2. (F) Intensity ratios of assemblies. Bright [rnq−] versus dim [rnq−] p = 0.0009; bright [rnq−] versus [RNQ+] p = 0.16; dim [rnq−] versus [RNQ+] p = 0.0019; Welch’s two sample t test. Black dots represent individual data points that fell outside the whiskers of the boxplots. (G) Circularity ratios of assemblies. Bright [rnq−] versus dim [rnq−] p = 0.0001; bright [rnq−] versus [RNQ+] p = 0.04; dim [rnq−] versus [RNQ+] p = 0.0006; Welch’s two-sample t test. (H) Plot of intensity ratios versus circularity ratios.

Figure 3

Dim Assemblies Display Liquid-like Properties in Cells (A) Addition of 1,6-hexanediol to digitonin-permeabilised cells can discriminate between liquid-like assemblies, which dissolve, and solid-like assemblies, which do not (Kroschwald et al., 2015). (B) Kymographs showing the effect of hexanediol addition on fluorescence intensities of HTT_ex1 assemblies in digitonin-permeabilized yeast cells. Hexanediol was added at time = 0 s. (C) Quantification of fluorescence intensities in kymographs in (B). (D) Snapshots and kymograph of hexanediol addition (first dotted white line) to a yeast cell containing dim assemblies, followed by hexanediol removal (second dotted white line). The red line indicates the slice through which the kymograph was plotted. Blue arrow highlights coalescence. Images were linearly scaled to increase the visibility of the merging assemblies, saturating some of the pixels in the early frames. Note that dim assemblies do not sequester all the cytoplasmic fluorescence, which is visible throughout the experiment. Scale bar, 3 μm. Related to Video S4.

Figure 4

Liquid-like and Solid-like HTT_ex1 Assemblies Have Different Nanostructures Electron tomography of yeast cells expressing 43Q-GFP and 43QP-GFP. Panels show slices through tomograms of the HTT_ex1 assemblies. Colors indicate the type of assembly and prion status of the cells: orange, [rnq−] SAs; blue, [rnq−] LAs; green, [RNQ+] SAs. (A) 43Q-GFP SA in a [rnq−] cell, showing a fibrillar nanostructure. (B) 43Q-GFP LA in a [rnq−] cell, with a characteristic “smooth” appearance. (C) 43Q-GFP SA in a [RNQ+] cell. (D) 43QP-GFP SA in a [rnq−] cell. (E) 43QP-GFP LA in a [rnq−] cell. (F) 43QP-GFP SA in a [RNQ+] cell. Scale bar, 200 nm.

Figure 5

HTT_ex1 Sequence Affects Assembly Formation (A) Yeast cells in the [rnq−] background expressing HTT_ex1 constructs with different polyQ lengths, with or without the P-rich region, were imaged by widefield fluorescence microscopy at specific time points after induction of HTT_ex1 expression. The percentage of cells containing an SA (orange), an LA (blue), or diffuse fluorescence (gray) was determined at each time point by manual counting. n = 207–259 cells per construct. (B) Quantitative dot blots calibrated with purified 25QP-GFP were used to determine HTT_ex1 concentration in cellular lysates (red). (C) Cellular concentrations of HTT_ex1 constructs at 4 and 24 hr. Plots show individual experiments (black) and mean ± SEM (gray).

Figure 6

HTT_ex1 Forms Liquid-like Assemblies by Liquid-Liquid Phase Separation (A) Liquid-liquid phase separation of 25QP-GFP induced by molecular crowding. Scale bar, 5 μm. (B) Fusion of 25QP-GFP droplets. Scale bar, 2 μm. See also Video S5. (C) Half-bleach (Brangwynne et al., 2009) of a 25QP-GFP droplet (white dashed outline). Kymograph shows redistribution of 25QP-GFP after the bleach. Scale bar, 1 μm. (D and E) Intermolecular interactions of 25QP-GFP in droplets. (D) Droplet formation at different salt and protein concentrations. The phase diagram indicates conditions where 25QP-GFP forms droplets (blue dots) and where it does not (gray dots). (E) Phase diagram showing the effect of 1,6-hexanediol concentrations on droplet formation. All experiments were carried out in the presence of 10% dextran as a crowding agent, and each condition was assessed at least twice.

Figure 7

Liquid-like HTT_ex1 Assemblies Convert into Solid-like Assemblies (A) Liquid-like droplets gradually convert into solid-like structures. Scale bar, 2 μm. (B) Quantification of the mean circularity of droplets during the conversion shown in (A). (C) FRAP of partially converted droplets comparing recovery of spike features (magenta) and droplet centers (green). Shaded region of graph represents 95% confidence interval. Scale bar, 2 μm. See also Table S1. (D) Droplets before (green) and after (magenta) hexanediol addition. Scale bar, 5 μm. (E) Quantification of early stages of droplet conversion (related to Figure 7D). Radial shift is the distance between intensity maxima in droplet linescans before (green) and after (magenta) hexanediol. n = 20 similarly sized droplets. Shaded regions: 95% confidence intervals. Scale bar, 0.5 μm. (F) Slices through a tomogram of a yeast cell containing a 43Q-GFP liquid-like assembly (blue dotted line) containing isolated fibrillar structures (orange arrows). Scale bar, 200 nm.