The 'active life' of Hsp90 complexes

Abstract

Hsp90 forms a variety of complexes differing both in clientele and co-chaperones. Central to the role of co-chaperones in the formation of Hsp90 complexes is the delivery of client proteins and the regulation of the ATPase activity of Hsp90. Determining the mechanisms by which co-chaperones regulate Hsp90 is essential in understanding the assembly of these complexes and the activation and maturation of Hsp90's clientele. Mechanistically, co-chaperones alter the kinetics of the ATP-coupled conformational changes of Hsp90. The structural changes leading to the formation of a catalytically active unit involve all regions of the Hsp90 dimer. Their complexity has allowed different orthologues of Hsp90 to evolve kinetically in slightly different ways. The interaction of the cytosolic Hsp90 with a variety of co-chaperones lends itself to a complex set of different regulatory mechanisms that modulate Hsp90's conformation and ATPase activity. It also appears that the conformational switches of Hsp90 are not necessarily coupled under all circumstances. Here, I described different co-chaperone complexes and then discuss in detail the mechanisms and role that specific co-chaperones play in this. I will also discuss emerging evidence that post-translational modifications also affect the ATPase activity of Hsp90, and thus complex formation. Finally, I will present evidence showing how Hsp90's active site, although being highly conserved, can be altered to show resistance to drug binding, but still maintain ATP binding and ATPase activity. Such changes are therefore unlikely to significantly alter Hsp90's interactions with client proteins and co-chaperones. This article is part of a Special Issue entitled: Heat Shock Protein 90 (HSP90).

Fig. 1

The catalytically active unit of Hsp90. a), Pymol cartoon of the Hsp90 dimer (pdb, 2CG9; Sba1 not shown). N-terminal domains (N) are in green and magenta, middle domains (M) in gold and yellow and the C-terminal domains (C) in salmon and blue. AMPPNP is shown as spheres bound to the N-terminal domains. b), Stabilization of the N-terminal domains of Hsp90 in their dimerized state by Sba1 (pdb, 2CG9). Sba1 is shown in cyan bound between the dimerised N-terminal domains of Hsp90 (green and magenta). The ATP lids (red) are in their closed conformation. α-helix 1 and β-strand 1, which undergo an exchange of position are indicated. AMPPNP is shown as spheres. c), Stabilization of the catalytic loop (yellow with yellow residues) by interaction between Thr 22 of the neighboring N-terminal domain (pale green with green residues) and Leu 378 of the catalytic loop and between Ile 117 of the ATP lid (yellow with cyan residues) and Leu 374 of the catalytic loop of the same monomer (pdb, 2CG9). Hydrogen bonds between the catalytic Arg 380 and AMPPNP are shown with dotted blue lines. d), Model of the domain interface between the N-terminal- and middle-domain of Hsp90 showing that the catalytic loop may remain in a closed inactive state and is thus not coupled to other structural changes resulting from the closed dimerized state of Hsp90. The middle domain (pdb, 1HK7; yellow) with a closed catalytic loop (cyan, except for Arg 380 which is shown in gold) was superimposed on the full-length closed Hsp90 structure (pdb, 2CG9; N-domain, magenta; middle domain, green). No steric clashes are observed and a number of hydrogen bonds can be formed, which are shown as dotted blue lines. Water molecules are shown as cyan spheres.

Fig. 2

Kinetic cycles of the S. cerevisiae Hsp90 and E. coli Htpg. In the yeast cycle (blue and black arrows) the conformational changes leading to the catalytically active state (Closed-ATP-Active) involves transition via two intermediate conformations (I₁ and I₂). Aha1 is supposed to accelerate the cycle by bypassing the I₁ sate. In the E. coli cycle (red and black arrows) a two-phase transition via an intermediate state (I) leads to the closed active (T) state. The slowest step in the cycles, both representing conformational change, are indicated. The open, closed and active state as well as the nucleotide state of the chaperone throughout the cycle is indicated.

Fig. 3

Model structure of Hop. Evidence suggests that although Hop/Sti1 may be dimeric in solution, in Hsp90 complexes Hop/Sti1 monomers may interact with Hsp90 independently. The TPR1 and TPR2a domains are shown bound with their Hsp70 and Hsp90 target peptides, respectively. Co-ordinates obtained as a kind gift from J. Guenter Grossmann.

Fig. 4

Inhibition of the ATPase cycle by Cdc37/p50. Cdc37/p50 is shown in cyan and the N-terminal domain of Hsp90 in green with a red ATP lid (pdb, 1US7). Arg 167 of Cdc37/p50 interacts with the catalytic Glu 33 and prevents it from hydrolyzing ATP. The non-hydrolysable analog of ATP, AMPPNP is shown in magenta. Hydrogen bonds are shown as dotted blue lines.

Fig. 5

The stable Hsp90–Rar1 complex. a), Pymol cartoon showing the binding of the CHORD II domain of Rar1 (cyan) to the N-terminal domain of Hsp90 (pdb, 2XCM; open state, green; closed state, yellow ). The ATP lid in the open state is blue and in the closed state it is red. The closed state of the ATP lid is incompatible with the binding of the CHORD II domain. b), Pymol cartoon showing the interaction of the CHORD II domain of Rar1 with the ADP bound Hsp90 (pdb, 2XCM). The N-terminal domain of Hsp90 is shown in green with amino acids involved in binding ADP in salmon. CHORD II domain is shown in cyan with His 188 in yellow. ADP is shown in pale blue. Water molecules are shown as red spheres and the magnesium ion as a cyan sphere. Hydrogen bonds are shown as dotted blue lines.

Fig. 6

ATPase modulated states of Hsp90. a), The ‘inhibited’ state of Sba1 bound Hsp90 (pdb, 2CG9). Sba1 is shown in light blue with amino acid residues in cyan. The middle domain of Hsp90 is shown in yellow with amino acids residues in green. The N-terminal domain sequences from the neighboring Hsp90 monomer are shown in gold. ATP is shown bound to Arg 380 by hydrogen bounds (dotted blue lines). Trp 124 of Sba1 binds into a hydrophobic pocket formed by the Hsp90 residues, Leu 315, Ile 388 and Val 391, which signals the catalytic loop to be released into its active open conformation. This state represents an ‘inhibited’ form of Hsp90 that is committed to hydrolyzing ATP. b), Modulation of the catalytic loop of Hsp90. Hsp90 is shown in yellow (pdb2CG9; middle- and C-terminal-domain not shown). The catalytic loop is shown in magenta (closed state; pdb, 1HK7), green (partially open state of the N-domain Aha1 — middle domain of Hsp90 complex, pdb 1USU) and the open active state (blue) of the Sba1-bound structure (pdb 2CG9). Aha1 is in gold. The partially open state of the catalytic loop in the N-Aha1 — middle-Hsp90 complex results because in this structure Arg 380 cannot interact with ATP. Hydrogen bonds are shown as dotted blue lines.

Fig. 7

Model of the C-terminal small molecule-binding site of Hsp90. a), Pymol cartoon showing the proposed site of interaction of novobiocin (cyan spheres) with the Hsp90 dimer (green and gold; co-ordinates were a kind gift from Brian S. J. Blagg. b), Proposed interactions of novobiocin (cyan) with the C-terminal domains of Hsp90. Novobiocin is bound close to the C-terminal dimerization interface of Hsp90 and makes contact with both monomers of Hsp90 (green and gold). Hydrogen bonds are shown as dotted blue lines.

Fig. 8

Sites of post-translational modification of Hsp90. a), The closed active conformation of yeast Hsp90 showing amino acid residues that are phosphorylated (Thr 22 and Tyr 24), acetylated (K294), nitrosylated (Cys 598 In Hsp90α; equivalent position in yeast Hsp90 is Ala 577) and involved in client protein binding (Trp 300). ATP and selected amino acid residues are shown as gold and cyan spheres, respectively. b), Position of amino acid residues (Thr 22 and Tyr 24, yellow) in the open state of the N-terminal domain of yeast Hsp90. ATP is shown in spheres. Hydrogen bounds are shown as dotted blue lines.

Fig. 9

Radicicol resistance by modification of the ATP-binding site of Hsp90. A single mutation, L34I results in the opening of a pocket that alters the water structure of the active site. Three additional water molecules enter and repel the chlorine (Cl) of radicicol thus decreasing its affinity for Hsp90. Water molecules are shown as pale yellow (L34I) and cyan (wild type) spheres. Radicicol bound to the mutant protein is shown in yellow while that bound to wild type Hsp90 is in cyan. Amino acid residues of L34I are in yellow and residues of the wild type protein are in cyan. Mutations are shown as L34I and V35I in red text. Dotted red (L34I) and blue lines (wild type) represent hydrogen bonds.