Neurofilaments Function as Shock Absorbers: Compression Response Arising from Disordered Proteins

What can cells gain by using disordered, rather than folded, proteins in the architecture of their skeleton? Disordered proteins take multiple coexisting conformations, and often contain segments which act as random-walk-shaped polymers. Using x-ray scattering we measure the compression response of disordered protein hydrogels, which are the main stress-responsive component of neuron cells. We find that at high compression their mechanics are dominated by gaslike steric and ionic repulsions. At low compression, specific attractive interactions dominate. This is demonstrated by the considerable hydrogel expansion induced by the truncation of critical short protein segments. Accordingly, the floppy disordered proteins form a weakly cross-bridged hydrogel, and act as shock absorbers that sustain large deformations without failure.

Figure 1

(a) Neurofilament protein form bottlebrush filaments which interact via their protruding tails. Each filament is decorated with either one (NF-L), two (NF-L with NF-M or NF-L with NF-H), or three different types of tails. Tail charge distributions of native (b) NF-M and (d) NF-H are compared against the charge distributions of de-phosphorylated (c) NF-M, (e) NF-H, and (f) NF-L. Charges are averaged over a 5-amino acid window at $p\mathrm{H}$ 6.8 (see Supplemental Material [5]). (g) A closeup of the NF-L tip region shows the nonaveraged amino acid charges, with two vertical black lines denoting the 5 and 11 amino acid truncated versions. Hydrophobic residues are underlined.

Figure 2

(a) $\mathrm{\Pi }$ vs $D$ curves of NF-L and NF-L mutants at 150 mM show that truncation of the positive short tip segment results with a $\sim 15\mathrm{nm}$ expansion at lowest $\mathrm{\Pi }$. (b) A similar 20 nm expansion is observed in the NF-L:M network as a result of 11 amino acid tip truncation. (c) Sketch of the hexagonal model used for the theoretical calculations, with the equilateral triangular unit cell of volume $V$ marked with dashed line. (d) Dimensionless osmotic pressure $\tilde{\mathrm{\Pi }}$ vs ${\varphi }^{-1}$ of different brushes exhibits a power law decay above $\varphi \sim 0.01$. (e) A schematic illustration of opposite filaments interconnected by an effective contour length $S=Na$, shown in thick solid black line. The location of tail-bridging segments sets $S$, and, consequently, the attractive force prefactor $k$. (f) Modeled $\mathrm{\Pi }$ terms are plotted against experimental NF-L:H data (red squares). Fitted $\mathrm{\Pi }$ are plotted as solid lines in (a), (b).

Figure 3

(a) NF-L interfilament spacing vs monovalent salt concentration of NF-L at: 0% PEG (black triangles), 20% PEG (red circles), and EDTA assembly buffer with 0% PEG (blue triangle). Increasing concentrations beyond 100 mM do not significantly modify the stress response, in agreement with previous simulations [40]. The interfilament distance measured in NF-L networks formed with EDTA buffer show that the bridging mechanism does not require divalent salts. (b) $\mathrm{\Pi }$ vs $D$ results of NF-L at a wider range of monovalent concentration show significant deviations from the model at lower salt concentrations.