The role of Hsp70 (heat-shock protein 70) chaperones in assisting protein-folding processes relies on their ability to associate with short peptide stretches of protein substrates in a transient and ATP-controlled manner. In the present study, we review the molecular details of the mechanism behind substrate recognition by Hsp70 proteins.

Introduction

Members of the large family of Hsp70 (heat-shock protein 70) chaperones assist in an extraordinarily broad spectrum of protein-folding processes. They refold stress-denatured soluble proteins and, in co-operation with Hsp100 proteins, even aggregated proteins as part of their protein quality control function [13]. They also assist in the folding of newly synthesized proteins in the cytosol, the translocation into and folding within organelles and the assembly and disassembly of protein complexes; furthermore, they regulate signal transduction pathways by controlling the stability and activities of protein kinases and transcription factors [46]. All these functions rely on the ability of Hsp70 proteins to interact transiently with short peptide stretches of protein substrates. This interaction is controlled by ATP as well as by a plethora of different co-chaperones that modulate the ATPase cycle. Thus Hsp70 proteins, together with their co-chaperones, constitute a complex network of highly versatile and regulated folding machines [7].

All Hsp70 proteins share the same overall structure. They are composed of an actin-like N-terminal ATPase domain of 45 kDa [8,9], an SBD (substrate-binding domain) of approx. 15 kDa and a C-terminal domain of approx. 10 kDa that is involved in co-chaperone binding and probably has additional functions [10]. ATP binding to the ATPase domain of Hsp70 decreases the affinity of the SBD for substrates by 5–85-fold by increasing the association rates for substrates by approx. 50-fold and the dissociation rates by 2–3 orders of magnitude [11,12]. Thus Hsp70 chaperones alternate between an ATP state in which the substrate-binding pocket is open and an ADP state in which the binding pocket is preferentially closed and the associated substrates will be trapped (Figure 1). ATP hydrolysis leading to substrate enclosure is the rate-limiting slow step in the ATPase cycle of most Hsp70 proteins [13,14]. This step is strongly accelerated by DnaJ proteins (J proteins) in the presence of substrates [1519]. By their ability to couple ATP hydrolysis with substrate binding and also by their ability to associate with the substrates themselves, J proteins can target Hsp70 partner proteins to pre-selected substrates and, therefore, are critical regulators of Hsp70 activity [20].

Functional cycle of Hsp70 chaperones

Figure 1
Functional cycle of Hsp70 chaperones

Hsp70 is drawn in grey and the substrate in black. NXF, nucleotide-exchange factor (e.g. GrpE in E. coli and Bag in mammalian cells); JDP, J-domain protein (e.g. DnaJ in E. coli and Hdj1/Hdj2 in human cells).

Figure 1
Functional cycle of Hsp70 chaperones

Hsp70 is drawn in grey and the substrate in black. NXF, nucleotide-exchange factor (e.g. GrpE in E. coli and Bag in mammalian cells); JDP, J-domain protein (e.g. DnaJ in E. coli and Hdj1/Hdj2 in human cells).

We summarize below our current knowledge of the mechanism of substrate interaction by Hsp70 proteins.

Architecture of the SBD

The overall structure of the SBD is fairly conserved within the Hsp70 protein family, although some functional differences of potential importance exist [10,21,22]. The domain consists of a sandwich of two twisted β-sheets with four antiparallel strands each (Figure 2A). The substrate-binding cavity is formed by β-strands 1 and 2 and the connecting loops L1,2 and L3,4. These loops are stabilized by a second layer of loops (L4,5 and L5,6) and the α-helices A and B, which are packed on to the β-structure. The distal part of helix B is connected to the outer loops by hydrogen bonds and a salt bridge, thereby forming a latch, which contributes to the tight binding to substrates in the absence of ATP (Figure 2A). The substrate peptide that has been co-crystallized with the SBD fragment of the bacterial homologue DnaK is bound over a stretch of five residues through two types of interactions. First, hydrogen bonds are formed between the backbone of the cavity-forming loops and the backbone of the bound peptide. Secondly, van der Waals interactions are formed between hydrophobic side chains of the binding cavity and the substrate peptide. At a central position within the binding cavity, two residues of DnaK form an arch that reaches over the bound substrate, and a hydrophobic pocket facing downwards accommodates a single hydrophobic side chain of the substrate. These latter interactions make the most important contributions to substrate binding [12].

Structure of the substrate-binding domain

Figure 2
Structure of the substrate-binding domain

(a) A model illustrating possible conformational changes in the SBD. Movement is indicated by arrows and the shifted domains and loops are indicated transparent (PDB 1DKZ). (b) Overlay of 15 structures of the β domain of the SBD of DnaK in the peptide-bound form obtained by solution NMR (PDB 1Q5L). (c) Overlay of 20 structures of the β domain of the SBD of DnaK in the peptide-free form (PDB 1DGH). To illustrate the range of mobility, the three structures are coloured in red, green and blue. Black circles mark the loop regions having high mobility, where the range of mobility is wide in (c) but relatively restricted in (b). The green circles mark β-sheet 3, which is shifted between (b) and (c) but is more ordered in (b). All pictures were generated with the program WebLab Viewer Pro 4.0.

Figure 2
Structure of the substrate-binding domain

(a) A model illustrating possible conformational changes in the SBD. Movement is indicated by arrows and the shifted domains and loops are indicated transparent (PDB 1DKZ). (b) Overlay of 15 structures of the β domain of the SBD of DnaK in the peptide-bound form obtained by solution NMR (PDB 1Q5L). (c) Overlay of 20 structures of the β domain of the SBD of DnaK in the peptide-free form (PDB 1DGH). To illustrate the range of mobility, the three structures are coloured in red, green and blue. Black circles mark the loop regions having high mobility, where the range of mobility is wide in (c) but relatively restricted in (b). The green circles mark β-sheet 3, which is shifted between (b) and (c) but is more ordered in (b). All pictures were generated with the program WebLab Viewer Pro 4.0.

The most striking structural differences within the SBD exist for the Hsp110 and Hsp170 subfamilies of Hsp70 proteins [23,24]. Hsp110 and Hsp170 proteins are found only in eukaryotic cells and are poorly understood with respect to their biological functions and co-operating co-chaperones. They contain extensive insertions mainly in the β-strand connecting loops of the lower β-sheet and the helical part (Figure 2B). Deletion analysis suggests for Hsp110 proteins that, despite the relatively low sequence homology and the insertions in the C-terminal part relative to the classical Hsp70 proteins, their putative SBD has substrate-binding activity [24].

Another group of Hsp70 homologues contains a much-reduced C-terminal domain, including Ssz1 of yeast and the STCH protein of higher eukaryotes [2528]. Whereas Ssz1 may still have a β-sheet domain with some similarities to the classical Hsp70s, the sequence that extends beyond the ATPase domain in the STCH protein comprises approx. 60 residues, which is only one-half of the amino acids of the β-sheet domain. Nevertheless, STCH is more closely related to Hsp70 proteins than to actin, with which the Hsp70s share structural homology in the ATPase domain. Furthermore, more subtle differences exist throughout the Hsp70 protein family, e.g. those affecting the size and amino acid composition of the cavity-forming loops, which have an effect on the substrate specificity of Hsp70 (see below).

Conformational dynamics of the SBD

The structure of the isolated SBD of Hsp70, or truncated versions of it, has been investigated by a multitude of biophysical methods, including X-ray crystallography, solution NMR and neutron scattering [10,2933]. All the methods found the peptide substrate to be tightly packed into the binding cavity, raising the question as to how it can find its way into such a tight spot and how a large substrate protein can bind. It is clear that the substrate-binding site must undergo substantial conformational changes to allow substrate binding.

Emerging structural data from many different studies indicate that it is quite probable that movements of the helical lid as well as conformational changes in the β-sheet structure are both involved in the transition between these conformations. In the first crystal structure of the SBD of DnaK, two different conformations of the lid were found [10]. In one structure, the long α-helix lying over the substrate-binding site (α-helix B) was kinked in the middle and slightly bent upwards. Results of NMR studies, using a truncated form of the SBD lacking the smaller α-helices following helix B, showed that helix B in this mutant is capable of unwinding at almost the same region where the kink was found previously [30]. In addition, the truncated helix of this mutant does not lie directly on top of the loops that flank the binding pocket, as is the case with the crystal structure, but is bent sideways around them. Taken together, the results indicate that α-helix B has considerable flexibility at least in the truncated form. One can speculate that opening of the SBD could be achieved, at least in part, by bending the helix in the middle (see I in Figure 2a) or even by pivoting of the complete α-helical domain away from the substrate-binding site (see II in Figure 2a).

Interestingly, it has been shown that even a lidless variant of DnaK retains the capacity for allosteric stimulation of peptide release by ATP, albeit to a modest degree [12,18,33,34]. This indicates that at least a part of the conformational changes is due to the β-sheet domain and can be induced in the absence of the lid. NMR investigations of peptide-free [33] and peptide-bound forms [31] of the SBD lacking the complete α-helical domain have shown that the loops flanking the substrate-binding groove have a very high mobility (Figure 2b, black circles) and one of the β-strands (β 3; Figure 2b, green circle) is less ordered compared with the crystal structure. After peptide binding, the lidless β domain assumes a more rigid structure very similar to the crystal structure of the peptide-bound SBD possessing the lid. The dynamical range of the loops is then very restricted; β-sheet 3 is locked-in and the interaction between the peptide and β domain, including the arch formed by M404 and A429, are the same as in the lid-containing structures. This indicates that the β domain by itself can acquire both an open and a closed conformation. Figure 2(a) summarizes the different models for conformational changes that could lead to an opening of the SBD. Since these results were obtained using different truncated versions of the SBD or under extreme conditions such as crystallization, they can only hint at possible conformational changes. It is the challenge of future studies to test these models and to uncover the complex mechanism of the allosteric interactions ruling substrate interactions of the Hsp70 protein family.

What do Hsp70 proteins recognize within substrates?

Hsp70 chaperones interact promiscuously with almost all the unfolded proteins, but generally do not bind their native counterparts. However, they also recognize certain folded proteins with high specificity. Therefore an important question is, how does Hsp70 combine in its substrate specificity both seemingly contradictory properties?

The most extensive analysis of the substrate specificity of an Hsp70 protein used a library of cellulose-bound 13-mer peptides scanning the sequences of natural proteins with an overlap of ten amino acids [35]. Using this method on a library with more than 4000 peptides, the binding motif for DnaK was elucidated. It consists of a core of five amino acids enriched with hydrophobic residues, flanked on both sides by a region where positively charged residues are preferred. The binding motif of DnaK is abundant in protein sequences. In the native state, these sites are generally buried in the hydrophobic core of the protein, thereby explaining the promiscuous binding of DnaK to unfolded polypeptides (Figure 3).

DnaK–substrate interactions

Figure 3
DnaK–substrate interactions

(a) Hydrogen-bonding between the SBD of DnaK and the backbone of the co-crystallized substrate peptide (NRLLLTG). The substrate (green) is given in stick representation and the SBD without α-helices is presented as a ribbon model with the interacting residues in blue stick representation. Hydrogen bonds are indicated as black broken lines. (b) Hydrophobic interaction between the SBD of DnaK and the side chains of the substrate peptide, demonstrating the tight fitting of the peptide into the SBD. The substrate is given in green stick representation and the interacting residues of the SBD in blue CPK (Corey–Pauling–Koltun) representation. The pictures were generated with the program WebLab Viewer Pro 4.0 (PDB 1DKZ).

Figure 3
DnaK–substrate interactions

(a) Hydrogen-bonding between the SBD of DnaK and the backbone of the co-crystallized substrate peptide (NRLLLTG). The substrate (green) is given in stick representation and the SBD without α-helices is presented as a ribbon model with the interacting residues in blue stick representation. Hydrogen bonds are indicated as black broken lines. (b) Hydrophobic interaction between the SBD of DnaK and the side chains of the substrate peptide, demonstrating the tight fitting of the peptide into the SBD. The substrate is given in green stick representation and the interacting residues of the SBD in blue CPK (Corey–Pauling–Koltun) representation. The pictures were generated with the program WebLab Viewer Pro 4.0 (PDB 1DKZ).

It is necessary to determine how the Hsp70 proteins manage to bind selectively to certain folded substrates, such as transcription factors. The most advanced understanding of the nature of such binding sites comes from the bacterial sigma 32 protein, a heat-shock promoter-specific sigma subunit of RNA polymerase. This protein is inactivated through transient association with DnaK as part of a negative autoregulatory loop of the heat-shock regulation. Experimental results to identify the binding site of DnaK within sigma 32 will be presented elsewhere.

Comparison of the binding preference of DnaK with that of the Escherichia coli homologues HscA and HscC revealed that substrate specificity can vary substantially within the Hsp70 protein family, probably as a manifestation of adaptive specialization [21,36] (M.P. Mayer and B. Bukau, unpublished work). The structural reason for these differences lies probably in the substrate-binding cavity, in particular the arch-forming amino acids that, in contrast with other substrate-binding residues, show a significantly lower degree of evolutionary conservation [37]. In more general terms, these results indicate that we can expect to find considerable differences in substrate specificity within the Hsp70 protein family.

Heat Shock Proteins and Modulation of Cellular Function: Focused Meeting held at Guy's Hospital, London, U.K. Organized by C. Kelly (King's College London) and I. Dransfield (MRC Centre for Inflammatory Research, Edinburgh). Edited By C. Kelly.

Abbreviations

     
  • Hsp70

    heat-shock protein 70

  •  
  • SBD

    substrate-binding domain

References

References
1
Hartl
 
F.U.
 
Nature (London)
1996
, vol. 
381
 (pg. 
571
-
580
)
2
Mayer
 
M.P.
Rüdiger
 
S.
Bukau
 
B.
 
Biol. Chem.
2000
, vol. 
381
 (pg. 
877
-
885
)
3
Schlieker
 
C.
Bukau
 
B.
Mogk
 
A.
 
J. Biotechnol.
2002
, vol. 
96
 (pg. 
13
-
21
)
4
Frydman
 
J.
 
Annu. Rev. Biochem.
2001
, vol. 
70
 (pg. 
603
-
647
)
5
Hartl
 
F.U.
Hayer-Hartl
 
M.
 
Science
2002
, vol. 
295
 (pg. 
1852
-
1858
)
6
Bukau
 
B.
Deuerling
 
E.
Pfund
 
C.
Craig
 
E.A.
 
Cell (Cambridge, Mass.)
2000
, vol. 
101
 (pg. 
119
-
122
)
7
Mayer
 
M.P.
Brehmer
 
D.
Gassler
 
C.S.
Bukau
 
B.
 
Adv. Protein Chem.
2002
, vol. 
59
 (pg. 
1
-
44
)
8
Flaherty
 
K.M.
Deluca-Flaherty
 
C.
McKay
 
D.B.
 
Nature (London)
1990
, vol. 
346
 (pg. 
623
-
628
)
9
Flaherty
 
K.M.
McKay
 
D.B.
Kabsch
 
W.
Holmes
 
K.C.
 
Proc. Natl. Acad. Sci. U.S.A.
1991
, vol. 
88
 (pg. 
5041
-
5045
)
10
Zhu
 
X.
Zhao
 
X.
Burkholder
 
W.F.
Gragerov
 
A.
Ogata
 
C.M.
Gottesman
 
M.
Hendrickson
 
W.A.
 
Science
1996
, vol. 
272
 (pg. 
1606
-
1614
)
11
Schmid
 
D.
Baici
 
A.
Gehring
 
H.
Christen
 
P.
 
Science
1994
, vol. 
263
 (pg. 
971
-
973
)
12
Mayer
 
M.P.
Schröder
 
H.
Rüdiger
 
S.
Paal
 
K.
Laufen
 
T.
Bukau
 
B.
 
Nat. Struct. Biol.
2000
, vol. 
7
 (pg. 
586
-
593
)
13
McCarty
 
J.S.
Buchberger
 
A.
Reinstein
 
J.
Bukau
 
B.
 
J. Mol. Biol.
1995
, vol. 
249
 (pg. 
126
-
137
)
14
Theyssen
 
H.
Schuster
 
H.-P.
Bukau
 
B.
Reinstein
 
J.
 
J. Mol. Biol.
1996
, vol. 
263
 (pg. 
657
-
670
)
15
Liberek
 
K.
Marszalek
 
J.
Ang
 
D.
Georgopoulos
 
C.
Zylicz
 
M.
 
Proc. Natl. Acad. Sci. U.S.A.
1991
, vol. 
88
 (pg. 
2874
-
2878
)
16
Karzai
 
A.W.
McMacken
 
R.
 
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
11236
-
11246
)
17
Barouch
 
W.
Prasad
 
K.
Greene
 
L.
Eisenberg
 
E.
 
Biochemistry
1997
, vol. 
36
 (pg. 
4303
-
4308
)
18
Misselwitz
 
B.
Staeck
 
O.
Rapoport
 
T.A.
 
Mol. Cell
1998
, vol. 
2
 (pg. 
593
-
603
)
19
Laufen
 
T.
Mayer
 
M.P.
Beisel
 
C.
Klostermeier
 
D.
Reinstein
 
J.
Bukau
 
B.
 
Proc. Natl. Acad. Sci. U.S.A.
1999
, vol. 
96
 (pg. 
5452
-
5457
)
20
Mayer
 
M.
Bukau
 
B.
 
Biol. Chem.
1998
, vol. 
379
 (pg. 
261
-
268
)
21
Kluck
 
C.
Pazelt
 
H.
Genevaux
 
P.
Brehmer
 
D.
Rist
 
W.
Schneider-Mergener
 
J.
Bukau
 
B.
Mayer
 
M.P.
 
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
41060
-
41069
)
22
Rüdiger
 
S.
Buchberger
 
A.
Bukau
 
B.
 
Nat. Struct. Biol.
1997
, vol. 
4
 (pg. 
342
-
349
)
23
Easton
 
D.P.
Kaneko
 
Y.
Subjeck
 
J.R.
 
Cell Stress Chaperones
2000
, vol. 
5
 (pg. 
276
-
290
)
24
Oh
 
H.J.
Easton
 
D.
Murawski
 
M.
Kaneko
 
Y.
Subjeck
 
J.R.
 
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
15712
-
15718
)
25
Gautschi
 
M.
Lilie
 
H.
Fünfschilling
 
U.
Mun
 
A.
Ross
 
S.
Lithgow
 
T.
Rucknagel
 
P.
Rospert
 
S.
 
Proc. Natl. Acad. Sci. U.S.A.
2001
, vol. 
98
 (pg. 
3762
-
3767
)
26
Hundley
 
H.
Eisenman
 
H.
Walter
 
W.
Evans
 
T.
Hotokezaka
 
Y.
Wiedmann
 
M.
Craig
 
E.
 
Proc. Natl. Acad. Sci. U.S.A.
2002
, vol. 
99
 (pg. 
4203
-
4208
)
27
Otterson
 
G.A.
Flynn
 
G.C.
Kratzke
 
R.A.
Coxon
 
A.
Johnston
 
P.G.
Kaye
 
F.J.
 
EMBO J.
1994
, vol. 
13
 (pg. 
1216
-
1225
)
28
Otterson
 
G.A.
Kaye
 
F.J.
 
Gene
1997
, vol. 
199
 (pg. 
287
-
292
)
29
Morshauser
 
R.C.
Hu
 
W.
Wang
 
H.
Pang
 
Y.
Flynn
 
G.C.
Zuiderweg
 
E.R.P.
 
J. Mol. Biol.
1999
, vol. 
289
 (pg. 
1387
-
1403
)
30
Wang
 
H.
Kurochkin
 
A.V.
Pang
 
Y.
Hu
 
W.
Flynn
 
G.C.
Zuiderweg
 
E.R.P.
 
Biochemistry
1998
, vol. 
37
 (pg. 
7929
-
7940
)
31
Stevens
 
S.Y.
Cai
 
S.
Pellecchia
 
M.
Zuiderweg
 
E.R.
 
Protein Sci.
2003
, vol. 
12
 (pg. 
2588
-
2596
)
32
Cai
 
S.
Stevens
 
S.Y.
Budor
 
A.P.
Zuiderweg
 
E.R.
 
Biochemistry
2003
, vol. 
42
 (pg. 
11100
-
11108
)
33
Pellecchia
 
M.
Montgomery
 
D.L.
Stevens
 
S.Y.
Vander Kooi
 
C.W.
Feng
 
H.
Gierasch
 
L.M.
Zuiderweg
 
E.R.P.
 
Nat. Struct. Biol.
2000
, vol. 
7
 (pg. 
298
-
303
)
34
Buczynski
 
G.
Slepenkov
 
S.V.
Sehorn
 
M.G.
Witt
 
S.N.
 
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
27231
-
27236
)
35
Rüdiger
 
S.
Germeroth
 
L.
Schneider-Mergener
 
J.
Bukau
 
B.
 
EMBO J.
1997
, vol. 
16
 (pg. 
1501
-
1507
)
36
Hoff
 
K.G.
Ta
 
D.T.
Tapley
 
T.L.
Silberg
 
J.J.
Vickery
 
L.E.
 
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
27353
-
27359
)
37
Rüdiger
 
S.
Mayer
 
M.P.
Schneider-Mergener
 
J.
Bukau
 
B.
 
J. Mol. Biol.
2000
, vol. 
304
 (pg. 
245
-
251
)