Báo cáo khoa học: Thermodynamic stability and folding of proteins from hyperthermophilic organisms doc

Using a data set of 24 thermostable, five hyperthermo-stable, and 64 mesostable protein structures in 25 Keywords hyperthermostability; protein folding; stability profile; unfolding kinet

Thermodynamic stability and folding of proteins from

hyperthermophilic organisms

Kathryn A Luke1,2, Catherine L Higgins3and Pernilla Wittung-Stafshede1,2,4

1 Department of Biochemistry and Cell Biology, Rice University, Houston, TX, USA

2 Keck Center for Structural and Computational Biology, Rice University, Houston, TX, USA

3 Section of Atherosclerosis and Vascular Medicine, Department of Medicine, Baylor College of Medicine, Houston, TX, USA

4 Department of Chemistry, Rice University, Houston, TX, USA

Introduction

Proteins from thermophilic (growth temperature 45–

75C) and hyperthermophilic (growth

tempera-ture‡ 80 C) organisms exhibit remarkable thermal

stability and resistance to chemical denaturants [1–3]

Despite decades of research in this field, a general

con-cept for how this stability is achieved remains elusive

The necessary differences are subtle, because

homolo-gous proteins from thermophilic⁄ hyperthermophilic

and mesophilic organisms have nearly identical

sequences and overall structures [4] Thermostability

appears to be implemented by a variety of strategies,

using combinations of virtually all known structural parameters: increased number of ionic interactions, increased extent of hydrophobic-surface burial, increased number of prolines, decreased number of glutamines, improved core packing, greater rigidity, extended secondary structure, shorter surface loops, and higher states of oligomerization [4–11]

Some years ago, it was argued that proteins from extreme thermophiles (growth temperature around

100C) are stabilized in different ways compared to those from moderately thermophilic organisms [3] Using a data set of 24 thermostable, five hyperthermo-stable, and 64 mesostable protein structures in 25

Keywords

hyperthermostability; protein folding;

stability profile; unfolding kinetics

Correspondence

P Wittung-Stafshede, Department of

Biochemistry and Cell Biology, 6100 Main

Street, Rice University, Houston, TX 77251,

USA

Fax: +1 713 348 5154

Tel: +1 713 348 4076

E-mail: pernilla@rice.edu

(Received 28 February 2007, accepted

18 April 2007)

doi:10.1111/j.1742-4658.2007.05955.x

Life grows almost everywhere on earth, including in extreme environments and under harsh conditions Organisms adapted to high temperatures are called thermophiles (growth temperature 45–75C) and hyperthermophiles (growth temperature‡ 80 C) Proteins from such organisms usually show extreme thermal stability, despite having folded structures very similar to their mesostable counterparts Here, we summarize the current data on thermodynamic and kinetic folding⁄ unfolding behaviors of proteins from hyperthermophilic microorganisms In contrast to thermostable proteins, rather few (i.e less than 20) hyperthermostable proteins have been thor-oughly characterized in terms of their in vitro folding processes and their thermodynamic stability profiles Examples that will be discussed include co-chaperonin proteins, iron-sulfur-cluster proteins, and DNA-binding pro-teins from hyperthermophilic bacteria (i.e Aquifex and Theromotoga) and archea (e.g Pyrococcus, Thermococcus, Methanothermus and Sulfolobus) Despite the small set of studied systems, it is clear that super-slow protein unfolding is a dominant strategy to allow these proteins to function at extreme temperatures

Abbreviations

GuHCl, guanidine hydrochloride; TM, midpoint of thermally induced unfolding transition; DG U , change in free energy upon protein unfolding;

DC p , difference in heat capacity between folded and unfolded states; Fd, ferredoxin; [GuHCl] 1 ⁄ 2 , GuHCl concentration at midpoint of equilibrium unfolding transition.

structural families, Szilagyi and Zavodszky proposed

that hyperthermostable proteins have stronger ion

pairing, fewer cavities, and higher b-sheet contents as

compared to the thermostable proteins [3]

Hyper-thermophilic microbes are found in the most basal

positions in the universal tree of life in both bacteria

and Archea domains [1]; these organisms may thus

bear similarities to ancient life forms Whereas bacteria

only include two genera of hyperthermophilic

organ-isms (i.e Aquifex and Thermotoga), there is

consider-able phylogenic diversity among the hyperthermophilic

Archaea (e.g Pyrococcus, Thermococcus,

Methanother-musand Sulfolobus) [2] Notably, no hyperthermophilic

eukaryote has yet been discovered [1]

Comparisons of the thermodynamics and kinetics of

the folding of proteins from mesophilic and

thermo-philic⁄ hyperthermophilic organisms can provide an

insight into the mechanisms of stabilization that

can-not be obtained from static structural and sequence

investigations The thermodynamic stability of a

pro-tein is quantitatively defined by the Gibbs free-energy

change upon unfolding (DGU¼ –RTlnKU) deduced

from the equilibrium constant (KU) When postulated

as a simple reversible two-state transition [12], the

equilibrium constant (KU¼ kf⁄ ku) is characterized by

the rate constants of folding (kf) and unfolding (ku)

rates The stability of a protein therefore involves both

equilibrium and kinetic aspects; increased protein

sta-bility may be reflected either as slower unfolding (ku),

faster folding (kf), or a combination of the two

(Fig 1A) In vitro folding⁄ unfolding experiments in

solution often involve chemical (i.e urea or guanidine

hydrochloride, GuHCl) or thermal perturbations of

the protein; the progress of the reaction being

moni-tored by spectroscopic methods such as aromatic

fluorescence (tertiary interactions near fluorophores),

far-UV circular dichroism (secondary structure

con-tent), or visible absorption (cofactor environment) For

time-resolved folding investigations, stopped-flow

ing instruments are often necessary, which have a

mix-ing dead time of 1–2 ms Experimental analyses of the

kinetic and thermodynamic origin of protein

thermo-stability and hyperthermostbility, however, have often

been hampered by unfolding irreversibility of such

pro-teins in vitro [13–15]

Three thermodynamic models have been proposed

to explain the high stability of thermostable and

hy-perthermostable proteins [4,16] (Fig 1B) In the first

model (Model 1), compared to a protein from a

me-sophilic organism, the thermostable protein would be

more thermodynamically stable throughout the

tem-perature range (i.e have higher DGU at every

temper-ature, shifting the profile vertically upwards) A

second model (Model 2) implies that the free-energy profile of the thermostable protein would be horizon-tally displaced to higher temperatures In this model,

TS

A

B

U

Reaction Coordinate

F

Temperature

Model 1

Model 2

Model 3

ΔG‡ U

ΔGU

ΔG‡ F

Fig 1 (A) Scheme linking protein-thermodynamic stability (DG U ) to folding (k f ) and unfolding (k u ) rate constants U, unfolded; F, folded;

TS, transition state For a two-state folding process, the difference in equilibrium stability (i.e DG U ) is related to the difference in activation parameters (i.e DG 

F and DG 

U ) as: DG 

U ) DG 

F ¼ )RT*ln(k f ⁄ k u ) ¼ DG U (B) Thermodynamic profiles (i.e DG U versus temperature) illustrat-ing the three models by which thermostability can be achieved A protein (black solid line) can achieve higher thermal stability by increasing its free-energy at all temperatures (i.e Model 1, dotted line), by horizontally shifting its stability profile to higher temperatures (i.e Model 2, gray line), or by broadening the stability profile (i.e Model 3, dashed line) while keeping the temperature of maximum

DG U the same.

the maximum value for DGUwould be equal for both

proteins, but the maxima would occur at different

temperatures At high temperatures, the thermostable

protein would be more stable; at lower temperatures,

the protein from the mesophile would be more stable

Finally, a third model (Model 3) indicates that the

free-energy profile for the thermostable protein would

be a flattened version of that for the protein from the

mesophile Thus, the thermostable protein would have

a more shallow dependence of DGU on temperature,

corresponding to a lower specific heat capacity

change of unfolding (DCp) According to this model,

the maximal DGU value would again be equal for

both proteins and would occur at the same

tempera-ture Support for all three models, and combinations

thereof, has been reported for different thermostable

proteins [17,18]

In this minireview, we look at protein

hyperthermo-stability from an energetic point of view; specifically,

we describe existing data on equilibrium stability and

kinetic folding⁄ unfolding processes of proteins from

hyperthermophiles To collect as many examples as

possible, the literature has been searched

comprehen-sively In the following sections, we discuss biophysical

data for hyperthermostable: (a) co-chaperonin

pro-teins, (b) nonheme iron propro-teins, (c) DNA-binding

proteins, as well as (d) a few other proteins Although

the number of characterized hyperthermostable

pro-teins is rather small (Table 1, Fig 2), some common

themes are evident and will be discussed in the final

section

Co-chaperonin proteins Co-chaperonin protein 10 (cpn10) works in conjunc-tion with cpn60 to fold substrate proteins in most organisms in nature [19–21] The tertiary and quater-nary structures of cpn10 proteins appear conserved; seven irregular b-barrels assemble into a ring-shaped heptameric structure [22] Cpn10 from hyperthermo-philic Aquifex aeolicus (Aacpn10) is unique among cpn10 proteins in that each monomer contains a 25-residue C-terminal extension [23] The sequence of the C-terminal tail shows no significant similarity with any known protein domain; its orientation in the heptamer is yet unknown Comparative biophysi-cal studies using a truncated version of Aacpn10 where the tail has been removed, Aacpn10del-25, demonstrated that the tail protects against cpn10 aggregation at high temperatures and at high protein concentrations [24] The tail, however, is not neces-sary for protein folding, heptamer assembly, co-chap-eronin function, or protein hyperthermostability [24,25]

By contrast to many other oligomeric proteins, the unfolding and disassembly of Aacpn10 and Aacpn10-del-25 are fully reversible reactions in vitro [23] We have therefore been able to characterize, in detail, the equilibrium and kinetic unfolding⁄ dissociation and folding⁄ assembly behaviors of Aacpn10 and Aacpn10-del-25 [24,26] The results have been compared to the corresponding data for the mesostable human mito-chondrial cpn10 (hmcpn10) [27] and Escherichia coli

Table 1 List of hyperthermostable proteins for which chemical ⁄ thermal stability and ⁄ or folding ⁄ unfolding dynamic parameters (Table 2) have been reported in the literature For each protein, the source organism, its maximum growth temperature, the fold of the protein, the pres-ence of cofactors, the oligomeric state, and the protein databank accession code (PDB ID) (if known) are provided.

a Bacteria b Archaea.

cpn10 (GroES) [26] homologs Whereas Aacpn10 is

much more resistant to thermal perturbation (TM¼

119, 73, 72C for Aacpn10, GroES, and hmcpn10,

respectively; 50 lm protein, pH 7.5), the equilibrium

unfolding mechanism is similar for all three cpn10

proteins [24,26,27] In GuHCl, and upon heating,

Aacpn10, Aacpn10del-25, hmcpn10, and GroES exhibit

apparent two-state equilibrium transitions, in which

unfolding and dissociation steps are coupled

[22,24,26,27] Thermodynamic analysis revealed that

the increased stability of the Aacpn10 heptamer arises

due to more stable monomers and not to increased

subunit–subunit affinity Whereas the stability is

approximately 2–3 kJÆmol)1 for GroES and hmcpn10

monomers, it is greater than 5 kJÆmol)1 for the

Aacpn10 monomer (pH 7, 20C) [24,26,28]

Nonethe-less, over 85% of the overall heptamer stability comes

from the interface interactions in both the mesostable

and hyperthermostable variants of cpn10 [26–28] This

property may be a functional requirement to assure a

heptameric state of cpn10 when it cycles on and off of

the cpn60 complex in vivo

Cpn10 unfolds⁄ dissociates in a biphasic reaction in GuHCl that involves protein unfolding prior to hept-amer dissociation [29] When comparing the data for the two bacterial cpn10 variants, both unfolding and dissociation of GroES are much faster than for Aacpn10 [26,30] By contrast to unfolding⁄ dissociation, the time-resolved refolding⁄ reassembly pathways show notable variations among the three cpn10 homologs Refolding and reassembly of hmcpn10 follow along two, apparent two-state parallel pathways Most of the molecules (approximately 75%) fold before assembling into the heptamer, whereas the rest assemble prior to protein folding [29,30] GroES refolding⁄ reassembly,

by contrast, follows a single sequential pathway, with monomer folding preceding a much slower heptamer assembly step [26] The kinetic refolding⁄ reassembly path for Aacpn10 is similar to that of GroES but more complex [30] Upon triggering refolding⁄ reassembly, Aacpn10 molecules first populate a misfolded mono-meric species This unproductive intermediate then unwinds, and a productive intermediate species forms Finally, the productive intermediates assemble into the

A

L K

H

Fig 2 Structural models of the hyperthermostable proteins in Table 1 for which high-resolution structures have been reported (red, a-helix; yellow, b-sheet; green, loop) (A) AaFd (B) TmFd (C) PfRu (D) Sac7d from Sulfolobus acidocaldarius (E) TmCsp (F) MfrH (G HU from Ther-motoga maritima (H) TmDHFR (I) PfPCP (J) TkRNase (K) TmCheY HII (L) Aacpn10del-25 (model based on 1WE3).

heptamer, and final folding takes place [30] The high

thermodynamic stability of the folded Aacpn10

mono-mer [24] can explain why transient intermediates are

populated only for the hyperthermostable variant

Stability profiles for Aacpn10 and GroES have been

derived using equilibrium unfolding⁄ dissociation data

at a range of temperatures [26] Comparison reveals

that the hyperthermostable cpn10 uses a combination

of all three thermodynamic models described in the

Introduction to increase the heptamer stability at high

temperatures Careful inspection demonstrates that

Models 1 and 2 are most important for the stabilizing

effect [26]

Nonheme iron proteins

Iron-sulfur (Fe–S) clusters are common cofactors in

nature that facilitate electron transport in many

pro-teins (e.g ferredoxins; Fds) [31] Aquifex aeolicus is the

only hyperthermophile known to contain so-called

plant- and mammalian-type [2Fe)2S] Fds: AaFd1 and

AaFd5 [32,33] Fd unfolding in vitro is irreversible due

to cluster degradation and cysteine oxidation in the

unfolded state [34–37] Using linear extrapolations of

thermal midpoints in the presence of different GuHCl

concentrations, AaFd1 and AaFd5 were found to

exhi-bit midpoints well above 100C at pH 7 in buffer

(Table 2) At pH 2.5, both AaFd5 and AaFd1 are less stable than at neutral pH, indicating that electrostatic interactions are important for the high thermal stabil-ity at physiological pH [32,33] AaFd1 and AaFd5 unfold extremely slowly at pH 7 (20C), and polypep-tide unfolding and Fe–S cluster degradation processes appear kinetically coupled Extrapolation of kinetic data in the presence of denaturants suggests that unfolding of the hyperthermostable Fds at pH 7 in buffer (20C) requires hundreds of years [35] For the homologous [2Fe)2S] Fd from mesophilic Spinacea oleracea (SpFd), only a few hours are required for complete unfolding at the same experimental condi-tions [34]

The role of the disulfide bond in AaFd1 was assessed using the variant AaFd1-C87A (i.e Cys87Ala), in which one of the disulfide bond-forming cysteines is eliminated [33] We found AaFd1-C87A

to be less stable than the wild-type protein towards thermal [TM(wt) ) TM(C87A)  8 C] and chemical ([GuHCl]1 ⁄ 2(wt)) [GuHCl]1 ⁄ 2(C87A)  0.9 M) pertur-bations AaFd1 is therefore a rare case of a Fd that is stabilized by a disulfide bond [33] Disulfide bonds are not thought to be a method to achieve protein thermo-stability [5] In general, hyperthermostable proteins contain lower fractions of cysteines and are poorer in disulfide bonds than their thermostable and mesostable

Table 2 Thermal midpoints (TM), thermodynamic stability (DG U ), and kinetic folding ⁄ unfolding parameters (k f and ku) for hyperthermostable proteins If not otherwise stated, T M and DG U refer to pH 7, and k f ⁄ k u to pH 7 and 20–25 C, conditions In the last column, the thermo-dynamic models used to increase thermal stability are given (see Introduction for definitions).

Thermodynamic model used

0.0041d 5.5 · 10)5d 1, 2 (+ 3) Aacpn10del-25 [24] Aquifex aeolicus 111 a 279 (30 C) a,f 0.0033 d 2.7 · 10)4d 1, 2 (+ 3)

ORF56 [43] Sulfolobus islandicus 107.5 c 85 (25 C) 7 · 10 7 (M)1Æs)1) e 1.8 · 10)7 1

TkRNase HII [55] Thermococcus kodakaraensis 83 44 (50 C) 0.78 (50 C) e

5 · 10)8(50 C) 1, 2

a 50 l M monomer b 120 l M monomer c 5 l M monomer d Final folding ⁄ unfolding step (processes not two-state) e Two-state process.

f Coupled unfolding ⁄ dissociation.

counterparts [34] Because the variant is still much

more stable than SpFd, it was concluded that

electro-static interactions also contribute to the high stability

of AaFd1

Like the A aeolicus Fd proteins, the [4Fe)4S] Fd

from the hyperthermophile, Thermotoga maritima

(TmFd) and the di-cluster [3Fe)4S] ⁄ [4Fe)4S] Fd from

hyperthermophilic Acidianus ambivalens (AmFd),

dis-play irreversible unfolding reactions in vitro [15,38]

The time-resolved reactions appear to be two-state,

suggesting that unfolding and cluster degradation are

also coupled steps for these Fd proteins [36] The

ther-mal unfolding midpoints are 125C and 122 C

(pH 7) for TmFd and AmFd, respectively [38] At

pH 2.5, however, the unfolding midpoint for AmFd

decreased to 64C [15,36] Also, the apparent DGU

value for AmFd is strongly pH dependent; at 20C, it

decreases from 79 to 20 kJÆmol)1 when the pH drops

from 7 to 2.5 [15] Analysis of a structural model of

AmFd suggests that a combination of additional

sur-face ion pairs, the zinc cofactor, and an efficiently

packed core govern the high stability of this protein

[36] According to the crystal structure, TmFd also

contains an increased number of hydrogen bonds

between charged residues as compared to thermolabile

Fd proteins [38]

Rubredoxin from the hyperthermophile, Pyrococcus

furiosus (PfRu) is another hyperthermostable nonheme

iron protein (a single iron bound by four cysteines)

that has been well characterized with respect to its

unfolding features in vitro It was found that the

ther-mal unfolding midpoint of PfRu is 42C higher at

pH 7 than at pH 2 [39] In addition, the unfolding

rates for PfRu increase dramatically upon decreasing

the pH from 7 to 2 [40] Compared with rubredoxin

from mesophilic Clostridium pastureianum (CpRu),

PfRu unfolds much more slowly at all experimental

conditions Electrostatic-energy calculations suggest

that ion pairs placed at key surface positions play a

kinetic role by ‘clamping’ the hyperthermostable

vari-ant [13] Based on hydrogen-exchange experiments, a

thermodynamic stability profile was constructed for

PfRu, which displayed a maximum DGU of 63 kJÆ

mol)1 at 100C (pH 7) and an extrapolated TM (but

probably not realistic) close to 200C (pH 7) [41]

DNA-binding proteins

One of the first hyperthermostable proteins studied

with respect to folding was the Sac7d DNA-binding

protein from Sulfolobus acidocaldarius Sac7d is an

attractive model protein because it is a small,

66-resi-due monomeric protein that unfolds in a two-state

reversible process in vitro [42] Sac7d is highly resistant

to thermal (TM of 91C at pH 7 and 63 C at pH 0), chemical ([GuHCl]1⁄ 2¼ 2.8 m GuHCl, pH 7, 20 C) and acidic (remains folded in the pH range 0–10) perturbations The thermodynamic stability of Sac7d, however, is similar to that of many mesostable pro-teins; at pH 7 and 20 C, DGUis only 22 kJÆmol)1[42]

A comparison of the stability profile for Sac7d to those for mesostable proteins of similar sizes reveals that the curve for Sac7d is flattened compared to the others Thus, Sac7d employs Model 3 to increase its stability Accordingly, calorimetric experiments pro-vided a DCp value for Sac7d unfolding of 0.5 kcalÆ molK)1, which is significantly lower than DCp values for unfolding of mesostable proteins of similar sizes [42] It was hypothesized that Sac7d survives with a low free energy in vivo due to post-translational modi-fications as well as interactions with compatible osmo-lytes, and by binding to DNA [42]

Like Sac7d, ORF56 from Sulfolobus islandicus is a DNA-binding protein that appears to be stabilized by interactions with DNA [43] ORF56 is also a small protein (56 residues) It forms a tetramer when bound

to DNA and exists as a dimer in the absence of DNA Equilibrium unfolding of the ORF56 dimer in vitro is

an apparent two-state reversible reaction, in which unfolding and dissociation are coupled processes [43] The thermal unfolding midpoint for the ORF56 dimer

in the absence of DNA is 107.5C (pH 7) The stabil-ity profile constructed from GuHCl-induced unfold-ing⁄ dissociation data at different temperatures suggests that ORF56 uses the first thermodynamic model (Model 1) to increase dimer stability at high tempera-tures; the stability maximum remains at 30C and

DCp is equal to that for a mesostable protein of the same size [43] The kinetic unfolding⁄ dissociation and refolding⁄ reassembly reactions for ORF56 have been characterized; they are also two-state processes Because the rate constants of refolding⁄ reassembly are dependent on the protein concentration, association appears to be the rate-limiting step [43] The lack of

an initial monomer-folding phase suggests that the assembly takes place between unfolded monomers Several DNA-binding proteins act by protecting DNA from adopting unwanted secondary structures [44] The family of cold shock proteins has this func-tion and is a good model system for proteins with all b-sheet structures The folding reactions of the cold shock proteins from hyperthermophilic T maritima (TmCsp) and mesophilic Bacillus subtilis (BsCsp) have been extensively studied in vitro [44] Both equilibrium and time-resolved folding⁄ unfolding processes are two-state Interestingly, the rate constants of refolding are

similar for the two homologs and the processes occur

within milliseconds, although their native fold is all

b-sheet (pH 7, 20C) Comparing BsCsp and TmCsp,

all sequence variations map to the protein surface [44]

This agrees with the rate-limiting step in folding being

hydrophobic collapse of the protein core, which is

identical in both proteins TmCsp, however, has

signifi-cantly greater thermal and chemical stability (TM of

85C, pH 7; DGU of 26 kJÆmol)1, pH 7, 20C) than

BsCsp (TMof 50C, pH 7; DGUof 11 kJÆmol)1, pH 7,

20C) [44] This difference in thermodynamic stability

correlates with two orders of magnitude slower

unfold-ing of TmCsp as compared to unfoldunfold-ing of BsCsp [44]

Charged surface interactions unique to TmCsp appear

to increase the entropic barrier to unfolding and

thereby slow down the reaction [45]

In contrast to many other hyperthermostable

pro-teins, histone proteins do not use surface charges

to achieve thermostability The archaeal histones

from the hyperthermophilic Methanothermus fervidus

(MfrH) and Pyrococcus strain GB-3a (PyArH) were

found to have significant increases in bulky, aromatic

residues in their cores compared to mesostable histones

[46] As a result of more tightly packed protein

interi-ors, DGU is 65 (pH 7, 35C) and 72 kJÆmol)1 (pH 7,

44C) for MfrH and PyArH, respectively, compared

to 28 kJÆmol)1 (pH 7, 43C) for a mesostable histone

from Methanobacterium formicicum (ForH) The DCp

of unfolding for the hyperthermostable and mesostable

histone homologs is approximately the same Instead,

the stability profiles for MfrH and PyArH are shifted

vertically upwards, in line with the first

thermody-namic model [46] We note that the histone-like HU

protein from T maritima differs from MfrH and

PyArH in that it remains folded at high temperatures

using a combination of Models 1 and 3 [47]

More-over, this protein is thought to be stabilized by a high

percentage of charged residues scattered throughout

the structure [47]

One of the more complete studies of protein

hyper-thermostability focuses on the small, monomeric O6

-methyl-guanine-DNA methyltransferase from

hyper-thermophilic Thermococcus kodakaraensis (TkMGMT)

and the C-terminal domain of the Ada protein from

E coli (EcAdaC) [48–51] GuHCl-induced equilibrium

unfolding experiments show that both proteins display

two-state, reversible transitions, with TkMGMT being

significantly more stable than EcAdaC ([GuHCl]1 ⁄ 2¼

5.2 and 1.6 m GuHCl for TkMGMT and EcAdaC,

respectively, pH 8.0, 20C) [49] Inspection of their

stability profiles reveals that both proteins have the

same free energy of unfolding at their respective

organ-ism’s growth temperature It appears that TkMGMT

uses a combination of all three thermodynamic models

to generate its high stability [50,51] Time-resolved unfolding experiments in GuHCl indicated that EcAdaC will unfold in < 1 s, whereas the unfolding time for TkMGMT is 4.5· 106 s (approximately

2 months) when the data are extrapolated to buffer conditions (pH 8, 20C) [48] Disruption of internal ion pairs through residue-specific mutations was found

to increase the unfolding-rate constant of TkMGMT [50] This finding supports that charged interactions are of importance for governing TkMGMT hyperther-mostability

Other proteins

In addition to the described groups of proteins, only

a few other hyperthermostable proteins (i.e DHFR, PCP, RNase, CheY) have been characterized with respect to folding and stability in vitro Dihydro-folate reductase from hyperthermophilic T maritima (TmDHFR) is a very stable dimeric protein [52] Folded monomers have not been detected at any equi-librium solvent condition or during TmDHFR unfold-ing in vitro Denaturant-induced equilibrium unfoldunfold-ing

is an apparent two-state process, involving only folded dimers and unfolded monomers: DGU is 142 kJÆmol)1

at pH 7, 15C [52] The stability profile for TmDHFR

is shifted upwards and to the right compared to that for DHFR from E coli Like most other hyperthermo-stable proteins for which kinetics have been reported, the unfolding reaction for TmDHFR is several orders

of magnitude slower than for the mesostable homolog

at corresponding conditions [52]

Pyrrolidone carboxyl peptidase from P furiosus (PfPCP) and from Bacillus amyloliquefaciens (BaPCP)

is another set of hyperthermostable⁄ mesostable homo-logs for which equilibrium and kinetic folding data have been collected at different pH values [53] A vari-ant substituted with serines at Cys142 and Cys188 (PfPCP-142⁄ 188S) was prepared to eliminate complex-ity due to sulfur oxidation [53] GuHCl-induced unfolding reactions of PfPCP-142⁄ 188S and BaPCP are reversible for both proteins, but the DGU values differ dramatically: DGU is 57 kJÆmol)1 (pH 7, 60C) and 8 kJÆmol)1 (pH 7, 40C) for PfPCP-142 ⁄ 188S and BaPCP, respectively Unfolding-rate constants for PfPCP-142⁄ 188S and BaPCP are also drastically dif-ferent (1.6 · 10)15Æs)1 and 1.5· 10)8Æs)1, respectively;

pH 7, 25C), whereas the refolding rate constants are similar (9.3· 10)2Æs)1 and 3.6· 10)1Æs)1, respectively) [53] Also, at pH 2.3, where PCP exists in monomeric form, unfolding of PfPCP-142⁄ 188S is much slower than BaPCP unfolding [54]

Ribonuclease HII from hyperthermophilic

Thermo-coccus kodakaraensis(TkRNase HII) has also been the

subject of equilibrium and kinetic folding studies [55]

Both GuHCl- and heat-induced unfolding reactions

are reversible, albeit the very slow unfolding process

prohibited acquisition of equilibrium unfolding curves

at temperatures below 40C (pH 7) [55] At 50 C,

unfolding reactions attained their equilibrium values

after 2 weeks of incubation, and a DGU value of

approximately 44 kJÆmol)1(pH 7) could be calculated

The unfolding-rate constant for TkRNase HII is much

lower than those for RNase HI from E coli and

RN-ase HII from thermophilic Thermus thermophilus (Tt),

whereas the refolding speeds for all three proteins are

similar [55] The stability profiles of TkRNase HII and

TtRNase HII are similar, although TkRNase HII

exhibits a higher temperature of maximum stability

and is folded in a smaller range of temperatures The

DCp for TkRNase HII is higher than that for

TtRN-ase HII, explaining the more narrow range of

tempera-tures where the hyperthermostable protein remains

folded as compared to the thermostable homolog

Both TkRNase HII and TtRNase HII have higher

temperatures of maximum stability compared to the

mesostable EcRNase HI [55]

Finally, the thermodynamic parameters for two

CheY homologs, one from hyperthermophilic T

mari-tima(TmCheY) and the other from mesophilic B

sub-tilis (BsCheY) have been compared Based on

denaturant-induced unfolding studies TmCheY

dis-plays increased TM (98C versus 55 C, pH 7) and

DGU (40 kJÆmol)1 versus 13 kJÆmol)1; pH 7, 50C)

values as well as a decreased DCp for unfolding (1.2

versus 2.3 kcalÆmolK)1, pH 7) compared to BsCheY

[56]

Conclusions

We have summarized the in vitro data that exist on

thermodynamic stability and folding⁄ unfolding

reac-tions of proteins from hyperthermophilic organisms

The number of proteins that have been characterized

to date is low (i.e less than 20; Table 1) Clearly,

addi-tional studies are needed to make general conclusions

for how thermodynamic parameters correlate with

hyperthermostability Nonetheless, some common

themes are evident when analyzing the present data

First, most of the hyperthermostable proteins in

Table 2 have high TMandDGUvalues, at least around

neutral pH (Fig 3) To achieve high stability, the three

thermodynamic models (Fig 1B) are used in different

combinations by these proteins (Table 2, final column)

In our data set, Model 1 (vertical shift of DGU to

higher values) is clearly the most prevalent mechanism, and most often it is combined with Model 2 (horizon-tal shift of the profile to higher temperatures) This trend differs from previous reports, which have con-cluded that a decrease in DCp (i.e Model 3, either alone or in combination with Model 1) is the most common method for proteins to achieve high thermal stability [4,17,18,57] Notably, in the earlier com-parisons, no separation between thermostable and hyperthermostable proteins was made, and few hyper-thermostable proteins were included Perhaps proteins from hyperthermophilic organisms most often use Models 1 and 2, whereas thermostable proteins are more likely to use Models 1 and 3 It was recently pro-posed that the choice of structural strategy for thermal stabilization of hyperthermostable proteins depends on the evolutionary history of the organism [58]

Second, because stability and⁄ or TM is much reduced at low pH for most of the hyperthermostable proteins, electrostatic interactions and⁄ or specific ion pairing appear to be an important way for these pro-teins to govern high stability at neutral pH This is reasonable because charge–charge interactions become stronger, whereas the importance of the hydrophobic effect decreases, at higher temperatures [5]

Third, for all hyperthermostable proteins with reported unfolding kinetics (Table 2), the unfolding speed is always dramatically slower (up to eight orders

of magnitude!) for the hyperthermostable protein than for the mesostable homolog (at room temperature) Still, in the five cases tested (i.e Aacpn10, TmCsp, PfPCP, TkRNase HII and ORF56), protein refolding

0 20 40 60 80 100

40 60 80 100 120 140

TM (deg C)

Fig 3 T M versus DG U values for hyperthermostable proteins in Table 2 (filled circles, those for which both values are known; cpn10 proteins excluded) along with their mesophilic counterparts (open circles, data mentioned in the text) The plot shows that the two parameters are correlated (solid line) for both sets of proteins.

kinetics are similar for the hyperthermostable and

mes-ostable variants This suggests that protein

hyperther-mostability is linked directly to kinetic resistance to

unfolding There may have been evolutionary pressure

in hyperthermophiles to select proteins with reduced

unfolding rates, rather than with very high folding

rates, because the rates of irreversible modification

depend on the protein-unfolding speed [59] One may

speculate that an increase in favorable surface

interac-tions, such as extra ion pairs, creates an entropic

bar-rier towards unfolding of hyperthermostable proteins

Despite this apparent structural rigidity, some

hyper-thermostable proteins (i.e HU and PfRu) were found

to have unexpectedly high flexibility in their native

states [11,47,60] An important future task is to probe

folding⁄ unfolding kinetics as a function of

tempera-ture: most importantly, at temperatures closer to the

hyperthermophilic organisms’ growth temperatures In

the only study of this [44], it was found that TmCsp,

as compared to BsCsp, indeed had slower unfolding

rate constants in a wide temperature range

Despite the general theme of super-slow unfolding,

it appears that evolution can (and does) make use of

everything that works and therefore we will never find

an overarching chemical⁄ biophysical ⁄ energetic

expla-nation of protein hyperthermostability In other words,

‘There’s more than one way to skin a cat’ [61]

Acknowledgements

This work was funded by Grants from NIH

(GM059663) and the Robert A Welch Foundation

(C-1588) KAL is supported by the Houston Area

Molecular Biophysics Program (GM08280) CLH is

supported by NIH Training Grant (T32 HL007812;

TTGA)

References

1 DeLong EF (2001) A phylogenetic perspective on

hyper-thermophilic microorgansims Methods Enzymol 330,

3–11

2 Huber R & Stetter KO (2001) Discovery of

hypertherm-ophilic microorganisms Methods Enzymol 330, 11–24

3 Szilagyi A & Zavodszky P (2000) Structural differences

between mesophilic, moderately thermophilic and

extre-mely thermophilic protein subunits: results of a

compre-hensive survey Structure 8, 493–504

4 Jaenicke R & Bohm G (1998) The stability of proteins in

extreme environments Curr Opin Struct Biol 8, 738–748

5 Cambillau C & Claverie J (2000) Structural and

geno-mic correlates of hyperthermostability J Biol Chem 275,

32383–32386

6 Das F & Gerstein M (2000) The stability of thermo-philic proteins: a study based on comprehensive genome comparison Funct Integr Genomics 1, 76–88

7 Wallon G, Kryger G, Lovett S, Oshima T, Ringe D & Petsko G (1997) Crystal structures of Escherichia coli and Salmonella typhimurium 3-isopropylmalate dehydro-genase and comparison with their thermophilic

counterpart from Thermus thermophilus J Mol Biol 266, 1016–1031

8 Hendsch Z & Tidor B (1994) Do salt bridges stabilize proteins? A continuum electrostatic analysis Protein Sci

3, 211–226

9 Marqusee S & Sauer R (1994) Contributions of a hydrogen bond⁄ salt bridge network to the stability of secondary and tertiary structure in lamda repressor Protein Sci 3, 2217–2225

10 Macedo-Ribeiro S, Martins B, Pereira PJB, Huber GBR & Soulimane T (2001) New insights into the thermostability of bacterial ferredoxins: high-resolution crystal structure of the seven-iron ferredoxin

from Thermus thermophilus J Biol Inorg Chem 6, 663–674

11 Jaenicke R (2000) Do ultrastable proteins from hyper-thermophiles have high or low conformational rigidity? Proc Natl Acad Sci USA 97, 2962–2964

12 Fersht A (1999) Structure and Mechanism in Protein Sci-ence WH Freeman, New York, NY

13 Cavagnero S, Zhou Z, Adams M & Chan S (1998) Unfolding mechanism of rubredoxin from Pyrococcus furiosus Biochemistry 37, 3377–3385

14 Makhatadze GI & Privalov PL (1995) Energetics of pro-tein structure Adv Propro-tein Chem 47, 307–425

15 Moczygemba C, Guidry J, Jones K, Gomes C, Teixeira

M & Wittung-Stafshede P (2001) High stability of a fer-redoxin from the hyperthermophilic archaeon A ambi-valens: involvement of electrostatic interactions and cofactors Protein Sci 10, 1539–1548

16 Nojima HAI, Oshima T & Noda H (1977) Reversible thermal unfolding of thermostable phosphoglycerate kinase Thermostability associated with mean zero enthalpy change J Mol Biol 116, 429–442

17 Kumar S, Tsai C & Nussinov R (2001) Thermodynamic differences among homologous thermophilic and meso-philic protiens Biochemistry 40, 14152–14165

18 Razvi A & Scholtz JM (2006) Lessons in stability from thermophilic proteins Protein Sci 15, 1569–1578

19 Hartl FU (1996) Molecular chaperones in cellular pro-tein folding Nature 381, 571–580

20 Kawata Y, Nosaka K, Hongo K, Mizobata T & Nagai

J (1994) Chaperonin GroE and ADP facilitate the fold-ing of various proteins and protect against heat inacti-vation FEBS Lett 345, 229–232

21 Todd MJ, Boudkin O, Freire E & Lorimer GH (1995) GroES and the chaperonin-assisted protein folding

cycle: GroES has no affinity for nucleotides FEBS Lett

359, 123–125

22 Boudker O, Todd M & Freire E (1997) The structural

stability of the co-chaperonin GroES J Mol Biol 272,

770–779

23 Guidry JJ & Wittung-Stafshede P (2004) First

character-ization of co-chaperonin protein 10 from

hyper-thermo-philic Aquifex aeolicus Biochem Biophys Res Commun

317, 176–180

24 Luke K, Apiyo D & Wittung-Stafshede P (2005) Role

of the unique peptide tail in hyperthermostable Aquifex

aeolicusco-chaperonin protein 10 Biochemistry 44,

14385–14395

25 Luke K, Apiyo D & Wittung-Stafshede P (2005)

Dis-secting homo-heptamer thermodynamics by isothermal

titration calorimetry: entropy-driven assembly of

co-chaperonin protein 10 Biophys J 89, 3332–3336

26 Luke K & Wittung-Stafshede P (2006) Folding and

assembly pathways of co-chaperonin proteins 10: origin

of thermostability Arch Biochem Biophys 456, 8–18

27 Guidry JJ, Moczygemba C, Steede NK, Landry S &

Wittung-Stafshede P (2000) Reversible denaturation of

the oligomeric human chaperonin 10: denatured state

depends on chemical denaturant Protein Sci 9, 2109–

2117

28 Guidry JJ & Wittung-Stafshede P (2002) Low stability

for monomeric human chaperonin protein 10:

interpro-tein interactions contribute majority of oligomer

stabil-ity Arch Biochem Biophys 405, 280–282

29 Perham M, Chen M, Ma J & Wittung-Stafshede P

(2005) Unfolding of heptameric co-chaperonin protein

follows ‘fly casting’ mechanism: observation of transient

nonnative heptamer J Am Chem Soc 127, 16402–16403

30 Luke K, Perham M & Wittung-Stafshede P (2006)

Kinetic folding and assembly mechanisms differ for two

homologous heptamers J Mol Biol 363, 729–742

31 Beinert H, Holm R & Munck E (1997) Iron-sulfur

clus-ters: nature’s modular, multipurpose structures Science

277, 653–659

32 Mitou G, Higgins C, Wittung-Stafshede P, Conover

RC, Smith AD, Johnson MK, Gaillard J, Stubna A,

Munck E & Meyer J (2003) An Isc-type extremely

ther-mostable [2Fe-2S] ferredoxin from Aquifex aeolicus

Biochemical, spectroscopic, and unfolding studies

Biochemistry 42, 1354–1364

33 Meyer J, Clay MD, Johnson MK, Stubna A, Munck E,

Higgins C & Wittung-Stafshede P (2002) A

hypertherm-ophilic plant-type [2Fe-2S] ferredoxin from Aquifex

aeo-licusis stabilized by a disulfide bond Biochemistry 41,

3096–3108

34 Higgins C, Meyer J & Wittung-Stafshede P (2002)

Exceptional Stability of a [2Fe-2S] ferredoxin from

hyperthermophilic bacterium Aquifex aeolicus Biochim

Biophys Acta 1599, 82–89

35 Wittung-Stafshede P (2004) Slow unfolding explains high stability of thermostable ferredoxins: common mechanism governing thermostability? Biochim Biophys Acta 1700, 1–4

36 Wittung-Stafshede P, Gomez CM & Teixiera M (2000) Stability and folding of hyperthermophilic ferredoxin from the archaeon A ambivalens J Inorg Biochem 78, 35–41

37 Griffin S, Higgins C, Soulimane T & Wittung-Stafshede

P (2003) High thermal and chemical stability of Ther-mus thermophilusseven-iron ferredoxin Eur J Biochem

270, 4736–4743

38 Pfeil W, Gesierich U, Kleemann GR & Sterner R (1997) Ferredoxin from the hyperthermophile Thermotoga maritimais stable beyond the boiling point of water

J Mol Biol 272, 591–596

39 Cavagnero S, Zhou ZH, Adams MW & Chan SI (1995) Response of rubredoxin from Pyrococcus furiosus to environmental changes: implications for the origin of hyperthermostability Biochemistry 34, 9865–9873

40 Cavagnero S, Debe DA, Zhou ZH, Adams MW & Chan SI (1998) Kinetic role of electrostatic interactions

in the unfolding of hyperthermophilic and mesophilic rubredoxins Biochemistry 37, 3369–3376

41 Hiller R, Zhou ZH, Adams MW & Englander SW (1997) Stability and dynamics in a hyperthermophilic protein with melting temperature close to 200C Proc Natl Acad Sci USA 94, 11329–11332

42 McCrary B, Edmondson S, Shriver J (1996) Hyperther-mophile protein folding thermodynamics: differential scanning calorimetry and chemical denaturation of Sac7d J Mol Biol 264, 784–805

43 Zeeb M, Lipps G, Lilie H & Balbach J (2004) Folding and association of an extremely stable dimeric protein from Sulfolobus islandicus J Mol Biol 336, 227–240

44 Perl D, Welker C, Schindler T, Schroder K, Marahiel

M, Jaenicke R & Schmid F (1998) Conservation of rapid two-state folding in mesophilic, thermophilic, and hyperthermophilic cold shock proteins Nat Struct Biol

5, 229–235

45 Schuler B, Kremer W, Kalbitzer HR & Jaenicke R (2002) Role of entropy in protein thermostability: fold-ing kinetics of a hyperthermophilic cold shock protein

at high temperatures using19F NMR Biochemistry 41, 11670–11680

46 Li W-T, Grayling R, Sandman K, Edmondson S, Shriver J, Reeve J (1998) Thermodynamic Stability of Archael Histones Biochemistry 37, 10563–10572

47 Ruiz-Sanz J, Filimonov V, Christodoulou E, Vorgias C

& Mateo P (2004) Thermodynamic analysis of the unfolding and stability of the dimeric DNA-binding protein HU from the hyperthermophilic eubacterium Thermotoga maritimaand its E34D mutant Eur J Biochem 271, 1497–1507