Unraveling Desmin's Head Domain Structure and Function

Abstract

Understanding the structure and function of intermediate filaments (IFs) is necessary in order to explain why more than 70 related IF genes have evolved in vertebrates while maintaining such dramatically tissue-specific expression. Desmin is a member of the large multigene family of IF proteins and is specifically expressed in myocytes. In an effort to elucidate its muscle-specific behavior, we have used a yeast two-hybrid system in order to identify desmin's head binding partners. We described a mitochondrial and a lysosomal protein, NADH ubiquinone oxidoreductase core subunit S2 (NDUFS2), and saposin D, respectively, as direct desmin binding partners. In silico analysis indicated that both interactions at the atomic level occur in a very similar way, by the formation of a three-helix bundle with hydrophobic interactions in the interdomain space and hydrogen bonds at R16 and S32 of the desmin head domain. The interactions, confirmed also by GST pull-down assays, indicating the necessity of the desmin head domain and, furthermore, point out its role in function of mitochondria and lysosomes, organelles which are disrupted in myopathies due to desmin head domain mutations.

Keywords: NDUFS2; desmin; homology modelling; protein interactions; saposin D.

Figure 1

NDUFS2 and Saposin D identified as desmin’s head domain binding partners. (A) Schematic representation of clone-111 (NDUFS2, GenBank accession number NM_004550; fragment 504–2059 bp) and clone-145 (pro-saposin D, GenBank accession number NM_002778; fragment 1337–2839 bp), identified using yeast two-hybrid system as desmin head domain binding proteins. (B) Specific binding of desmin to GST–saposin D. Desmin-enriched cardiac extracts (as a source of desmin, lane 1) and equivalent amounts of GST (lane 2), GST–saposin D (lane 3) and GST–saposin D fragments (lanes 4 and 5) expressed in bacteria were bound to glutathione Sepharose. Recombinant GST–saposin D fragment 412–524 aa and the GST–saposin D fragment 412–486 aa, but not GST or GST–saposin D fragment 488–524 aa, absorbed desmin, as indicated by Western blot analysis (aa numbering is from NCBI Reference Sequence: NP_002769.1). (C) Coomassie staining of the bound to glutathione Sepharose proteins from panel B, eluted by SDS. (D) Specific binding of NDUFS2 to GST–desmin. Equivalent amounts of GST (lane 1) and GST–desmin (lane 2) expressed in bacteria were bound to glutathione Sepharose. Recombinant GST–desmin, but not GST alone, absorbed NDUFS2 from total protein cardiac extracts preparation, as indicated by Western blot analysis. Total protein cardiac extracts from desmin null (Des−/−) mice (lane 3) used as a source of NDUFS2 were also analyzed.

Figure 2

(A) The docking pose between the homology models of desmin’s head domain (represented by the green ribbon) and NDUFS2 protein (represented by the red ribbon). The insert focuses on the three-helix bundle that is formed by the interaction of the two proteins. The interacting α-helix from desmin’s head is shown with a black arrow. The insert has been tilted 90° from the vertical axis. (B) The interaction between desmin’s head and NDUFS2. The interacting residues from desmin’s head domain are showing in space fill representation. Notably, Arg16 lies next to Arg15 and Ser32 is next to Ser31. This is a safety measure that ensures the viability of the hydrogen bonding interaction at the event of mutation or loss of either Arg16 or Ser32 residues. The core of desmin’s head domain interacting α-helix is lined by a series of three glycines (Gly20, Gly23 and Gly27), which contribute both sterically and electrochemically to the stabilization of the interaction. (C) The interaction between desmin’s head domain and saposin D protein. Desmin’s head 3D homology model used in this docking experiment is represented by the green ribbon, whereas the X-ray crystal structure of saposin D (RCSB entry: 3 BQP) is represented by the blue ribbon.

Figure 3

Two-dimensional Ligplot representation of the interaction established between desmin’s head domain and NDUFS2 protein. (A) This is the specific interactions that were established after the docking experiment, before the molecular dynamics simulation (MDs). As depicted, the only hydrogen bonding interactions are located on the outer sides of desmin’s interacting α-helix. (B) Same as (A), only this is the interaction after the molecular dynamics simulation. It is clear that the hydrogen bonds successfully retained their strength/position upon the molecular dynamics simulation. Note that this interaction map has been horizontally flipped by Ligplot.

Figure 4

A 2D Ligplot representation of the interaction established between the desmin’s head and saposin-D. (I) These are the specific interactions that were established after the docking experiment, before the molecular dynamics (MDs) simulation. The only hydrogen bonding interactions are between Arg16 and Glu33 and Ser32 and Lys45/Asp48 (renumbered from the 3BQP_B, NM_002778). (II) Same as (I), except these are the interactions after the molecular dynamics simulation.

Figure 5

Structural superposition of the two docked molecular complexes. The common structures of NDUFS2 (red ribbon) and saposin D (blue ribbon) interacting with desmin’s head domain (green ribbon) were superposed (right panel). It was found that the NDUFS2 3D homology model and the saposin D X-ray structure share identical 3D conformations that favor the specific interaction with the active modeled conformation of desmin’s head domain. Note that this interaction map has been horizontally flipped by Ligplot.

Figure 6

Prediction of desmin’s 3D structure by AlphaFold. The α-helical rod domain intervenes by the linker L12. Head and tail domains seem mainly unstructured, with the exception of a β-strand formation on the tail. Model confidence/color: very high (pLDDT > 90)/blue; high (90 > pLDDT > 70)/light blue; low (70 > pLDDT > 50)/yellow and very low (pLDDT < 50)/orange. (N: amino terminus).

Figure 7

(A) No binding of headless desmin to GST–NDUFS2 and GST–saposin D. Equivalent amounts of GST–NDUFS2 (lanes 1, 2 and 3), GST (lanes 4, 5 and 6) and GST–saposin D (lanes 7, 8 and 9) expressed in bacteria were bound to glutathione Sepharose. Only the normal desmin (ΝD, lanes 3 and 9), but not the headless desmin (DesminΔ1-48) (lanes 1 and 7), was absorbed by recombinant GST–NDUFS2 and GST–saposin D, as indicated by Western blot (WB) analysis. (B) Western blot analysis of desmin of total protein extracts of COS7 cells transfected with normal desmin, empty vector or desminΔ1-48 (lanes 1, 2 and 3, respectively) used for the GST pull-down assays. Abbreviations: DΔ1-48: desmin with 1-48 aa deletion, EV: empty vector and ND: normal desmin.

Figure 8

Immunofluorescence staining of cardiac tissue sections for desmin (green) and NDUFS2 (red) (A), indicating partial co-localization in areas close to sarcolemma and nucleus (arrows) and (B) for desmin and saposin D (red) at z-lines close to sarcolemma, presumably at costameres (arrowheads). (Blue: DAPI staining of nuclei, images are projections of confocal images; scale bar 10 μm).