Quantifying surface tension and viscosity in biomolecular condensates by FRAP-ID

Abstract

Many proteins with intrinsically disordered regions undergo liquid-liquid phase separation under specific conditions in vitro and in vivo. These complex biopolymers form a metastable phase with distinct mechanical properties defining the timescale of their biological functions. However, determining these properties is nontrivial, even in vitro, and often requires multiple techniques. Here we report the measurement of both viscosity and surface tension of biomolecular condensates via correlative fluorescence microscopy and atomic force microscopy (AFM) in a single experiment (fluorescence recovery after probe-induced dewetting, FRAP-ID). Upon surface tension evaluation via regular AFM-force spectroscopy, controlled AFM indentations induce dry spots in fluorescent condensates on a glass coverslip. The subsequent rewetting exhibits a contact line velocity that is used to quantify the condensed-phase viscosity. Therefore, in contrast with fluorescence recovery after photobleaching (FRAP), where molecular diffusion is observed, in FRAP-ID fluorescence recovery is obtained through fluid rewetting and the subsequent morphological relaxation. We show that the latter can be used to cross-validate viscosity values determined during the rewetting regime. Making use of fluid mechanics, FRAP-ID is a valuable tool to evaluate the mechanical properties that govern the dynamics of biomolecular condensates and determine how these properties impact the temporal aspects of condensate functionality.

Graphical abstract

Figure 1

(a) Correlative AFM-epifluorescence microscopy setup. (b) Pictorial representation of a liquid droplet confined between the AFM colloidal probe with radius R and the glass substrate, and exerting an attractive force on the retracting probe. The force is used to assess the condensate surface tension (γ). (c) Sketch reporting the different phases of the FRAP-ID method. The formation of a dry spot (top) within the condensed phase (green) is tracked until complete rewetting (regime I). The contact line speed during this process is monitored through epifluorescence to determine the viscosity (η). Subsequently, complete fluorescence recovery is obtained through a fluid relaxation, whose lubrication analysis leads once again to $\eta /\gamma$, cross-validating values obtained from regime I.

Figure 2

(a) AFM topography image of GFP-tagged ELF3 biomolecular condensates wetting a glass coverslip. Scale bar, 10 μm. The static contact angle (θ), shown in the inset (scale bars on both x and y axis = 1 μm), is determined along the profile (light blue) in the AFM image. (b) Two indentation cycles, performed with a colloidal probe (R = 2.5 μm) onto different droplets, are shown with both approach (blue) and retract (red). Indentations exhibit different hysteresis, reflecting the different size of the droplets. Retract curves are fitted (green) using Eq. 1, providing the droplet surface tension, whose distribution and the associated Gaussian fit (black) are shown in the inset.

Figure 3

(a) Probe-induced dry spot exhibiting complete rewetting, from formation (blue) to closure (yellow) over time, observed through epifluorescence. Scale bars, 2 μm. In each panel, the segment used to extract the geometric parameters is shown. (b) Hole geometrical profiles, color-coded as the segments in (a): the intercept with threshold values are used to determine the dry spot radius a (black horizontal line), the corresponding larger hole radius A (gray horizontal line), and the edge $\delta h/\delta r$. The intersection of such thresholds with the profiles defines the initial and final hole radii. Subscripts i and f, for A and a, denote the initial and final values, respectively. (c) Evolution of A, a (top), and $\delta h/\delta r$ (bottom) during the time of the experiment. Complete rewetting occurs at t = $t_{\mathrm{c}}$, when a$\approx$ 0 (end of regime I, highlighted in boldface $\delta h/\delta r$). Regime II ($t>t_{\mathrm{c}}$) is characterized by the recovery of the initial fluorescence through the relaxation of the fluid perturbation as tracked by A over time. (d) Evolution of hole radius a (black) as a function of the normalized time and associated best linear trend (red), providing an estimated viscosity. (e) Good agreement observed between numerical simulations (thick lines) and experimental profiles from regime II (thin lines) over a duration of 28 s.

Figure 4

(a) Evaluation of droplet viscosity using passive microrheology. 50-nm (left) and 180-nm (right) beads embedded within protein condensates (top). The MSD vs. Lag t (gray) for 180-nm beads is fitted with a linear trend (red) to determine slope and intercept, the latter providing the diffusion coefficient. The instrumental background noise (black) is characterized by observing immobile particles in a separate experiment. The inset shows a particle trajectory. Scales bars on both x and y axis = 100 nm. (b) Coalescence between two droplets (top). Characteristic fusion time (τ) plotted against the droplet length scale (blue circles), and estimated linear trend (black line). Error bars derived from the exponential fit are shown.