Properties of intermediate filament networks assembled from keratin 8 and 18 in the presence of Mg²+

Abstract

The mechanical properties of epithelial cells are modulated by structural changes in keratin intermediate filament networks. To investigate the relationship between network architecture and viscoelasticity, we assembled keratin filaments from recombinant keratin proteins 8 (K8) and 18 (K18) in the presence of divalent ions (Mg(2+)). We probed the viscoelastic modulus of the network by tracking the movement of microspheres embedded in the network during assembly, and studied the network architecture using scanning electron microscopy. Addition of Mg(2+) at physiological concentrations (<1 mM) resulted in networks whose structure was similar to that of keratin networks in epithelial cells. Moreover, the elastic moduli of networks assembled in vitro were found to be within the same magnitude as those measured in keratin networks of detergent-extracted epithelial cells. These findings suggest that Mg(2+)-induced filament cross-linking represents a valid model for studying the cytoskeletal mechanics of keratin networks.

Figure 1

Keratin network morphology visualized by SEM. (A) The K8 and K18 filament network in epithelial cancer cells after removal of the cell membrane and other cytoplasmatic components by detergent extraction. Before extraction, 1 μm spheres were taken up by the cell through phagocytosis. (B–F) Filament networks assembled from 0.5 mg/ml purified K8 and K18 proteins at the indicated Mg²⁺ concentrations. The 1 μm spheres were embedded during filament assembly.

Figure 2

(Upper row) Morphometry of keratin filament networks imaged by SEM. The inset in the left panel illustrates the criteria for identifying physical cross-links, i.e., bending of filaments at intersections and threefold connections. (Lower row) Histograms depict the distribution of contour lengths between cross-links for the indicated Mg²⁺ concentrations.

Figure 3

Mean-squared displacement of microspheres from K8 and K18 networks assembled in the presence of the indicated concentrations of Mg²⁺.

Figure 4

Average storage (G′, panel A) and loss modulus (G″, panel B) for K8 and K18 networks assembled in the presence of different concentrations of Mg²⁺ as well as for keratin networks extracted from epithelial cancer cells. The insets show the mean and SD of G′₀ and G″₀ as determined at 1 Hz for the indicated Mg²⁺ concentration.

Figure 5

Dependence of the average elastic plateau modulus G′₀ of K8 and K18 networks on the Mg²⁺ concentration was investigated by fitting different models (thin black line: exponential model G′₀ ∼ (Mg²⁺)^x; thick black line: Flory (53); gray line: MacKintosh et al. (56)). The model by Flory (53) shows an excellent fit for Mg²⁺ concentrations of up to 1.0 mM, when the structural heterogeneity within the network is small (see inset in Fig. 4A).