Sequences and features of VPS4A, CHMP1B, and MIT domains. (A) Schematic illustrations of human VPS4A (19) and CHMP1B. The predicted coiled-coil (CC) of CHMP1B was identified at a probability of 0.17 by using multicoil (41). (B) Sequence alignment of known MIT domains from eight human proteins (Upper) and two proteins lacking human homologs. The “consensus” sequence (Lower) was derived from all 116 MIT domains in the SMART database (23). Highly conserved residues (>50% identity) are shown in blue, and residue types conserved in >50% of the sequences are shown in lowercase (h, hydrophobic; +, positively charged; -, negatively charged; u, Ala or Gly; l, Ile, Val, or Leu). Numbering corresponds to VPS4A.

VPS4A_5–76 MIT domain structure. (A) Ribbon diagram of the VPS4A MIT domain (residues 5–76). (B) View down the three helix bundle of the MIT domain, emphasizing the asymmetry in the disposition of the three helices. (C) “Alanine zipper” connecting VPS4A_5–76 MIT helices 1 and 2. The five conserved alanine side chains within this motif are shown explicitly.

Side chain interactions at the different layers of the VPS4A MIT three-helix bundle. Sequential amino acid layers in the three helix bundle of the VPS4A MIT domain. Note that side chain interactions between helices 2 (H2) and 3 (H3) follow canonical alternating “knobs into holes” interactions between residues at helix positions +1 (A) and +4 (D), whereas side chain packing between helices 1 (H1) and 2 (H2) is different because alternating alanine residues (highlighted in green) project almost directly at the pairing helix and are sandwiched between hydrophobic side chains from the +4 and +5 positions (relative to the preceding alanine) in the pairing helix.

Interactions between VPS4A and CHMP1B proteins. (A) GST pull-downs demonstrating that both full-length VPS4A and its isolated MIT domain (VPS4A_1–84) bind GST-CHMP1B_65–196 (denoted GST-1B_65–196, lanes 5 and 7) but not GST alone (lanes 4 and 6). Molecular weight standards (MW, lane 1), pure GST (lane 2), and pure GST-CHMP1B_65–196 (lane 3) are shown for reference. (B) Biosensor binding isotherms showing wild-type (WT) and L64A mutant VPS4A_1–84 proteins binding GST-CHMP1B and GST-CHMP1B_65–196. Filled circles, WT VPS4A_1–84 binding full-length GST-CHMP1B captured from E. coli extracts; filled triangles, WT VPS4A_1–84 binding GST-CHMP1B_65–196 captured from E. coli extracts; filled squares, WT VPS4A_1–84 binding affinity purified GST-CHMP1B_65–196; open squares, VPS4A_1–84 L64A binding affinity purified GST-CHMP1B_65–196. Binding to the GST control surface was negligible (not shown). Dissociation constants (μM) for WT VPS4A_1–84 binding were: CHMP1B = 20 ± 13 and CHMP1B_65–196 = 13 ± 6 μM; and were >500 μM for VPS4A_1–84 L64A binding to both constructs (minimum three independent measurements). (C) VPS4A MIT surface renderings showing the locations and identities of conserved residues. Conserved residues are color-coded based on the reduction in CHMP_65–196 binding affinity for alanine substitution mutations: dark blue, <3-fold change; light blue, 3- to 30-fold reduction; green, >30-fold reduction. Dissociation constants were: WT, 13 ± 6 μM; I10A, 17 ± 1 μM; K15A, 14 ± 1 μM; E26A, 8 ± 1 μM; E37A, 16 ± 2 μM; K53A, 20 ± 2 μM; R57A, 34 ± 4 μM; K59A, 9 ± 1 μM; L64A, >500 μM; E68A, 55 ± 4 μM; K71A, 18 ± 2 μM (minimum of two independent measurements). (D) Overlays of the first four helices of the TPR domain of FKBP51 (gray) with the MIT domain of VPS4A. The overlay shows the similarity of the structures and the position of the second paired helix (fourth overall) in the FKBP51 TPR domain, which is missing in the VPS4A MIT domain. (E) Side view of the overlay in D, but with the VPS4A MIT domain shown in a space filling model and color coded as in C. The figure emphasizes that residues required for CHMP1B binding (green, light blue) map to the same surface as the binding site for the second paired helix of the FKBP51 TPR domain.