Site-directed spin labeling and electron paramagnetic resonance determination of vimentin head domain structure

Abstract

Intermediate filament (IF) proteins have been predicted to have a conserved tripartite domain structure consisting of a largely alpha-helical central rod domain, flanked by head and tail domains. However, crystal structures have not been reported for any IF or IF protein. Although progress has been made in determining central rod domain structure, no structural data have been reported for either the head or tail domains. We used site-directed spin labeling and electron paramagnetic resonance to analyze 45 different spin labeled mutants spanning the head domain of vimentin. The data, combined with results from a previous study, provide strong evidence that the polypeptide backbones of the head domains form a symmetric dimer of closely apposed backbones that fold back onto the rod domain, imparting an asymmetry to the dimer. By following the behavior of spin labels during the process of in vitro assembly, we show that head domain structure is dynamic, changing as a result of filament assembly. Finally, because the vimentin head domain is the major site of the phosphorylation that induces disassembly at mitosis, we studied the effects of phosphorylation on head domain structure and demonstrate that phosphorylation drives specific head domain regions apart. These data provide the first evidence-based model of IF head domain structure.

FIGURE 1.

Room temperature EPR spectra. EPR spectra are shown for each of the spin labeled head domain mutants described in this study. The number above each spectrum is the amino acid residue where the spin labeled cysteine residue is located. All of the spectra are from protofilaments harvested after dialysis against low salt buffer, as described under “Experimental Procedures.” All of the spectra are normalized to the same number of spins by double integration of the SDS-solubilized sample. Inset: the line width ratio, d₁/d, is shown on a model spectrum. This ratio, calculated from the spectra acquired at −100 °C, reflects the dipolar interaction strength and thus distance between spin labels.

FIGURE 2.

Intermediate filaments versus protofilaments. Room temperature EPR spectra were gathered from intact intermediate filaments (IF, red spectra) and from protofilament subunits (black spectra) generated by dialysis against a low salt (LS) buffer. The number at the left of each pair of spectra identifies the amino acid residue where the spin labeled cysteine residue is located. Spin labels showed a strong mobility shift as the sample transitioned from protofilamentous stage to intact IFs. The d₁/d values from spectra gathered at −100 °C are shown in parentheses.

FIGURE 3.

Changes in EPR spectra during in vitro assembly. Room temperature EPR spectra are shown from various stages during the in vitro assembly of intermediate filaments, starting with soluble monomers in 8 m urea (blue), 6 m urea (violet), 4 m urea (green), 3 m urea (red), and 2 m urea (black). The number at the left of each set of spectra identifies the amino acid residue where the spin labeled cysteine residue is located. The spectra are normalized for the protein concentration in each sample.

FIGURE 4.

The impact of phosphorylation on head domain structure. EPR spectra were gathered at room temperature and normalized as described under “Experimental Procedures.” The number at the left of each pair of spectra identifies the amino acid residue where the spin labeled cysteine residue is located. For each indicated spin labeled position, the red spectrum corresponds to the protein kinase A-treated sample (KA, phosphorylated), and the black spectrum to the buffer control sample (C, nonphosphorylated) sample. The change in the spectral line width reveals the degree of broadening and dipolar interaction and is calculated by d₁/d ratio as shown in parentheses.

FIGURE 5.

Refining the model of vimentin. The conventional model of the IF protein is shown at the top in a. The three major predicted domains are identified: head domain, central rod domain, and tail domain. The central rod domain is subdivided into coil regions that are predicted to form α-helical coiled-coils, and linker“ domains, which lack the characteristics of α-helical coiled-coil domains. The data presented here, in combination with a previous report, now establish that (a) the two head domains in a dimer reside in close proximity throughout essentially all of their length, (b) the closely aligned head domains folds back onto the rod, bringing residues 17 and 137 into close proximity, (c) because the closely aligned head domains fold back along one side of the rod domain, the vimentin dimer becomes a partially asymmetric structure, and (d) at least residues 12, 17, 32, and 83 move farther apart upon phosphorylation. These data suggest that the head domain may be largely constrained to one side of the dimer, as shown. This asymmetry in head domain placement could generate two classes of tetrameric interactions as shown in b (tail domains not shown); one class would have the head domains residing between the two dimers (bracket A), whereas the other would have the head domains facing away from the dimer-dimer interface (bracket B). This model could explain why phosphorylation results in tetramers; phosphorylation drives apart the dimers identified by bracket A, where the head domains are concentrated at the interface between two dimers, but not those defined by bracket B, where the head domains face away from the interface.