Local motion analysis reveals impact of the dynamic cytoskeleton on intracellular subdiffusion

Abstract

Intracellular transport is a complex interplay of ballistic transport along filaments and of diffusive motion, reliably delivering material and allowing for cell differentiation, migration, and proliferation. The diffusive regime, including subdiffusive, Brownian, and superdiffusive motion, is of particular interest for inferring information about the dynamics of the cytoskeleton morphology during intracellular transport. The influence of dynamic cytoskeletal states on intracellular transport are investigated in Dictyostelium discoideum cells by single particle tracking of fluorescent nanoparticles, to relate quantitative motion parameters and intracellular processes before and after cytoskeletal disruption. A local mean-square displacement (MSD) analysis separates ballistic motion phases, which we exclude here, from diffusive nanoparticle motion. In this study, we focus on intracellular subdiffusion and elucidate lag-time dependence, with particular focus on the impact of cytoskeleton compartments like microtubules and actin filaments. This method proves useful for binary motion state distributions. Experimental results are compared to simulations of a data-driven Langevin model with finite velocity correlations that captures essential statistical features of the local MSD algorithm. Specifically, the values of the mean MSD exponent and effective diffusion coefficients can be traced back to negative correlations of the motion's increments. We clearly identify both microtubules and actin filaments as the cause for intracellular subdiffusion and show that actin-microtubule cross talk exerts viscosifying effects at timescales larger than 0.2 s. Our findings might give insights into material transport and information exchange in living cells, which might facilitate gaining control over cell functions.

Figure 1

Tracking of single NPs, transported in living D. discoideum cells, yields this type of trajectories. Cytoskeleton depolymerization agents, namely Benomyl and Latrunculin A, are used to discern the influence of particular cytoskeleton constituents, MT and F-actin, on the intracellular transport. Trajectories are analyzed using a local MSD algorithm shifting a rolling window across the NP trajectories.

Figure 2

Comparison of NP motion in glycerol and WT D. discoideum cells. (a) The autocorrelation of frame-to-frame increments fluctuates around zero for the glycerol data (upper panel) but shows statistically significant negative correlations for the intracellular data (lower panel). Using the autocorrelation function and the increment's variance, an autoregressive process for the increments with the same correlation statistics can be simulated; increments for this simulated process are distributed according to a Gaussian. (b) The size distribution of frame-to-frame increments can be approximated by Gaussian distributions (lines show fits to the data), with slight deviations at large increment values. (c) Distributions of local mean-square exponent values α and (d) effective diffusion coefficients D (middle panels) for glycerol (upper panel) and the WT cell (lower panel); experimental data (histograms) and simulations (solid lines). (e) Joint distributions of mean-square exponent α and diffusion coefficient D for glycerol (upper panel) and WT cell (lower panel); experimental data (left) and simulations (right).

Figure 3

Increment correlation coefficient in x (a) and histograms of increments in x (b) in three different cytoskeleton states: WT cells (black lines without error bars) as reference, Benomyl-assisted MT depolymerization (first row), Latrunculin A-assisted F-actin depolymerization (second row), and simultaneous MT and F-actin depolymerization (third row). Correlations and histograms were similar for all cases when measured for the increments of y instead of x.

Figure 4

Distributions of α (a), of D (b), and their joint distribution (c) in various cytoskeleton states: WT cells (top row), Benomyl-assisted MT depolymerization (second row), Latrunculin A-assisted F-actin depolymerization (third row), and simultaneous MT and F-actin depolymerization (bottom row) for both experimental data (red bars) and simulations (black lines). In a mean values of α are indicated by vertical dashed lines.

Figure 5

Lag-time dependence of the local MSD exponent 〈α〉 (a) and the mean effective diffusion coefficient 〈D〉 (b) in various cytoskeleton states: WT cells (experiments as black circles, simulations as black line), Benomyl-assisted MT depolymerization (exp: blue triangles, sim: blue line), Latrunculin A-assisted F-actin depolymerization (green squares, green line), simultaneous MT and F-actin depolymerization (red diamonds, red line), and as reference NP diffusion in glycerol (yellow triangles, yellow line) for both experimental data (symbols) and simulations (solid lines). Data shown in Fig. 4a correspond to a lag time of 0.735 s (vertical dashed line).