Báo cáo khoa học: Physico-chemical properties of molten dimer ascorbate oxidase doc

Keywords conformational change; dimeric intermediate; high pressure; protein compressibility; protein folding Correspondence G.. These findings provide new aspects of the protein folding

Physico-chemical properties of molten dimer ascorbate oxidase

Eleonora Nicolai1,2, Almerinda Di Venere1,2, Nicola Rosato1,2, Antonello Rossi2,

Alessandro Finazzi Agro’2and Giampiero Mei1,2

1 INFM, University of Rome, ‘Tor Vergata’, Italy

2 Department of Experimental Medicine and Biochemical Sciences, University of Rome, ‘Tor Vergata’, Italy

Dimeric enzymes are largely diffused in living

organ-isms, being present in prokaryotes and eukaryotes,

and in plants as well as in animals [1,2] Their

func-tionality is strictly dependent on quaternary

interac-tions that regulate the stability of subunit interface,

as demonstrated by the loss of biological activity

often occurring when the two monomers fall apart

Studying the folding process of dimeric proteins is

important for several reasons [3] In fact, their

qua-ternary structure represents the simplest case of

pro-tein–protein interaction, i.e., the basic mechanism that drives and controls most metabolic pathways in cells [1] Furthermore, it is the common opinion that oligomers evolved from simpler ancestral structures, namely monomeric proteins whose internal domains might have undergone a swapping process, transform-ing intrachain contacts in a new dimeric interchain surface [4] Thus, learning the ‘story’ of folding of dimeric (and oligomeric) enzymes is also important to understand part of life evolution

Keywords

conformational change; dimeric

intermediate; high pressure; protein

compressibility; protein folding

Correspondence

G Mei, Department of Experimental

Medicine and Biochemical Sciences,

University of Rome ‘Tor Vergata’, Via

Montpellier 1, 00133 Rome, Italy

Fax: +39 06 72596468

Tel: +39 06 72596460

E-mail: mei@med.uniroma2.it

(Received 9 August 2006, revised 26

Sep-tember 2006, accepted 27 SepSep-tember 2006)

doi:10.1111/j.1742-4658.2006.05515.x

The possible presence of dimeric unfolding intermediates might offer a clue

to understanding the relationship between tertiary and quaternary structure formation in dimers Ascorbate oxidase is a large dimeric enzyme that dis-plays such an intermediate along its unfolding pathway In this study the combined effect of high pressure and denaturing agents gave new insight

on this intermediate and on the mechanism of its formation The transition from native dimer to the dimeric intermediate is characterized by the release of copper ions forming the tri-nuclear copper center located at the interface between domain 2 and 3 of each subunit This transition, which is pH-dependent, is accompanied by a decrease in volume, probably associ-ated to electrostriction due to the loosening of intra-subunit electrostatic interactions The dimeric species is present even at 3· 108Pa, providing evidence that mechanically or chemically induced unfolding lead to a simi-lar intermediate state Instead, dissociation occurs with an extremely simi-large and negative volume change (DV)200 mLÆmol)1) by pressurization in the presence of moderate amounts of denaturant This volume change can

be ascribed to the elimination of voids at the subunit interface Further-more, the combination of guanidine and high pressure uncovers the pres-ence of a marginally stable (DG 2 kcalÆmol)1) monomeric species (which was not observed in previous equilibrium unfolding measurements) that might be populated in the early folding steps of ascorbate oxidase These findings provide new aspects of the protein folding pathway, further sup-porting the important role of quaternary interactions in the folding strategy

of large dimeric enzymes

Abbreviations

AAO, ascorbate oxidase; ANS, 1-anilino-8-naphthalene-sulfonic acid; GdmHCl, guanidinium hydrochloride; MD, molten dimer.

About 10 years ago, the extensive study of small

globular protein denaturation allowed a description of

the folding process in terms of sound schemes [5–7],

from the hydrophobic collapse (typical of molten

glob-ule states) to the so called framework model (based on

early secondary structure interactions) The discovery

of multiple intermediate states in several protein

fold-ing experiments, the presence of nearly independently

structured domains and the existence of parallel

fold-ing pathways led to the hypothesis that these simplified

models could actually reflect different aspects of a

much more complex process, whose features may be

explored by a combination of different experimental

tests [8] In particular, they might be viewed as

extremes of a more general nucleation-condensation

mechanism, dependent on the balance between early

secondary and tertiary interactions [9] This hypothesis,

that has been proposed to hold also in the case of

lar-ger enzymes [9], was successfully tested so far only for

small proteins where a molecular dynamics simulation

is possible [9–11] In parallel to these experimental

approaches, the development of a statistical mechanics

theory of the protein folding problem is opening new

perspectives in protein folding research [12] According

to this viewpoint, the search for the most stable

poly-peptide chain conformation occurs through a rough

funnel-shaped energy surface, whose local minima

rep-resent possible intermediate species [13]

All these new concepts can be in principle extended

also to dimeric proteins However, the characterization

of their folding kinetics and the determination of

inter-mediate states is often quite a difficult task, because

oligomerization can compete and thus interfere with

the assembling of monomers The analysis of the

avail-able crystallographic structures has shown that in the

majority of dimeric proteins the intersubunits contact

region is quite large with respect to the overall protein

surface [2] Generally the interface is mostly

character-ized by hydrophobic side chains that shield each other

from water upon dimerization [2] This feature is very

common, but exceptions are found especially in larger

dimers, like ascorbate oxidase (AAO; EC 1.10.3.3)

This enzyme is a blue copper protein that catalyzes

the oxidation of ascorbate with concomitant oxygen

reduction to water It is a homodimer, each subunit

containing 14 tryptophans [14] distributed in three

dis-tinct domains Only domains 2 and 3 are involved in

the monomer–monomer interaction, through the side

chains of residues located in b-turns or random coil

segments of the polypeptide chain Thus, none of the

amino acids forming a and b structures are present at

the interface The contact area between the subunits

shows a relatively low hydrophobic character, because

about half of the amino acids involved are polar Despite these unusual features the AAO dimer with-stands partial denaturation by  1.5 m guanidinium hydrochloride (GdmHCl) or  3 m urea In this condi-tion no larger aggregates are formed, as demonstrated

by analytical ultracentrifugation [15] In fact, a dimeric folding intermediate is rather observed which lacks catalytic activity and shows a partial loss of tertiary structure These characteristics recall those of molten globule state of small, globular proteins [16], therefore this dimeric intermediate has been defined as a molten dimer (MD) state [15]

The structural properties of this partially unfolded dimer are obviously correlated to the properties of the intersubunit contacts Thus, in order to get further information on the monomers association and on the role played by quaternary structure in the stability of large sized enzymes such as AAO, we carried out new measurements using both chemical and physical dena-turing agents In particular, we took advantage of the possibilities offered by the combination of high pres-sure techniques and low concentrations of GdmHCl [17], to perform a detailed analysis of the intermediate states The results of steady state and dynamic fluores-cence measurements allowed us to evaluate the volume and free energy changes from the native to the MD state Further transition to the fully unfolded state has been also shown to be a complex process, involving at least one weak, unstructured monomeric intermediate These findings provide new insights on the native structure in solution as well as further details on the unfolding pathway that should reflect the mechanism

of the protein assembly

Results

From native to MD state) denaturant-induced conformational changes

The structural changes of AAO upon partial unfolding

by 1.4 m GdmHCl have been investigated by fluores-cence measurements The intrinsic steady state emis-sion spectra of partially unfolded AAO by 1.4 m GdmHCl or 2.8 m urea are indicative of a greater hydration with respect to the native state (Fig 1A) In fact a larger full width at half maximum, due to the appearance of a red-shifted shoulder at  344 nm typ-ical of partially solvent-exposed tryptophan residues, was observed Lifetime measurements confirm this observation In fact, in a previous study [14] we repor-ted the rather complex fluorescence decay of AAO, which requires at least two continuous distributions

of lifetimes to fit the data The faster and slower

distributions were attributed to buried and

solvent-exposed tryptophan residues, respectively [14] As

shown in Fig 1B, there is an increase of the

compo-nent at  2.5 ns in the presence of 1.4 m GdmHCl or

2.8 m urea, while the shorter lifetime component is

decreased Unfortunately these fluorescence decays are

unsuitable to monitor rotational correlation times of

dozens of nanoseconds, as theoretically expected for a

large sized enzyme (i.e.,‡ 58 ns, see below) Therefore,

AAO has been labeled with an external probe, namely

dansylchloride, which allows anisotropy measurements

on longer time scales Figure 2A shows that the steady

state anisotropy of the dansylated enzyme increased

upon addition of GdmHCl, up to 1.5 m, indicating

a slower tumbling of the protein The steady state anisotropy of AAO-bound dansylchloride has there-fore been used to get preliminary estimations of the protein volume under native and partially unfolded states These values, reported in Table 1, have been obtained from the slopes of the linear Perrin plots [18] shown in Fig 2B The volume calculated for native AAO is reasonably close to that estimated using the protein molecular mass and an average hydrated

speci-fic volume of  1 cm3Æg)1 [19] Instead the volume of the protein molecule at 1.4 m GdmHCl, i.e., the MD species is  1.8 times larger (Table 1), indicating a rele-vant loss of the original stiffness These results have been quantitatively confirmed by dynamic anisotropy measurements, in which both the dynamics of the probe and of the whole protein can be determined through their rotational correlation times [18] As shown in Table 2, a double exponential fit was needed

to adequately fit the data In particular, the longer

cor-Fig 1 (A) Normalized steady state emission spectra, at 10 5 Pa, of

AAO at pH 6 (solid line), AAO + 1.4 M GdmHCl (short-dashed line),

AAO + 2.8 M urea (dotted line) and AAO + 3.5 M GdmHCl (long

dashes) The spectrum at 3 · 10 8 Pa has been reported for

com-parison as d (pH 6.0) and h (pH 8) Inset: relative fluorescence

total intensity of AAO versus pressure at two different protein

con-centration, namely  10)7M (n) and  10)5M (m) (B) Fluorescence

lifetime distribution profiles of native AAO, AAO + 1.4 M GdmHCl,

AAO + 2.8 M urea, AAO + 3.5 M GdmHCl and AAO at 3 · 10 8 Pa

[symbols as in (A)].

Fig 2 (A) Steady state anisotropy of dansylated AAO as a function

of GdmHCl concentration (B) Perrin plots of the native dansylated enzyme in the absence (s) or in presence of 1.4 M GdmHCl (d) The solid and dashed lines correspond to the best linear fit (yielding

Y1¼ 309.9 X + 5.12 and Y 2 ¼ 152.5 X + 4.47, respectively).

relation time obtained for native AAO (/2 67 ns)

matches closely the theoretical value calculated for a

prolate ellipsoidal molecule (/2  68 ns, see

Experi-mental procedures) having the 3D dimensions of AAO

[20,21] The MD, instead, showed a longer correlation

time (/2 124 ns), diagnostic of a much slower

tumb-ling Similar results were also obtained by light

scatter-ing measurements, which indeed yielded molecular

volumes in good agreement with those estimated from

dynamic anisotropy data (Table 1)

The MD state is more sensitive to trypsin

digestion

A further evidence of a less compact tridimensional

structure of the MD state has been obtained by trypsin

digestion The results obtained after 30 and 60 min of

trypsin digestion are illustrated in Fig 3, for both

native and MD AAO It appears that already after

30 min the partially folded intermediate is cleaved to

smaller fragments (Fig 3, column E), at variance with

native AAO (Fig 3, column C) After a longer

incuba-tion (60 min, columns F, G, H) the proteolysis of MD

was more pronounced but a partial digestion of the

native sample was observed (column F)

Pressure effects on the tertiary structure

of native AAO

The effect of pressure on AAO intrinsic fluorescence is

reported in Fig 1A At 3 · 108 Pa a red-shift of the

spectrum is accompanied by a 15–20% quenching of

tryptophan emission (Fig 1A, inset) The transition

was reversible because the native spectrum was obtained by bringing back the sample to room pres-sure (data not shown) No significant difference was obtained at higher protein concentration (Fig 1A, inset) ruling out the occurrence of subunit dissociation [22] Both steady state (Fig 1A) and dynamic fluores-cence data (Fig 1B) indicate that high pressure pro-motes a partial, pH-dependent loss of tertiary structure (Fig 1A)

These structural changes are generally characterized

by the exposure of internal hydrophobic patches as revealed by 1-anilino-8-naphthalene-sulfonic acid (ANS) binding [23,24] In a previous study [15], we have in fact shown that the ANS fluorescence is much higher in the partially denatured than in native AAO

A similar effect is achieved under pressure in the range

105–3· 108Pa (Fig 4A), further supporting the idea that physical denaturation might parallel the results obtained with chemical denaturants [15] When nor-malized (Fig 4A, inset), the data display a sigmoidal shape, suggesting that a two-state transition model (N2<- -> MD) may be used to get a fit A similar trend is shown by the tryptophan fluorescence change

Table 1 Theoretical and experimental volumes of native AAO and AAO + 1.4 M GdmHCl Theoretical dry volume estimated from crystallo-graphic data; theoretical hydrated volume estimated using a hydrated specific volume of 1 cm3Æg)1[19].

Sample

Theoretical dry volume [A˚3 ]

Theoretical hydrated volume [A˚3 ]

Perrin volume [A˚3 ]

Dynamic anisotropy volume [A˚3 ]

Light scattering volume [A˚3 ]

Table 2 Rotational correlation times of native AAO and AAO +

1.4 M GdmHCl.

AAO (2 exponentials) 1.3 0.74 ± 0.05 67 ± 6 0.39 ± 0.02

Molten dimer

(1 exponential)

Molten dimer

(2 exponentials)

0.9 0.71 ± 0.06 124 ± 9 0.37 ± 0.03

Fig 3 SDS ⁄ PAGE of AAO incubated with trypsin Columns A and

B show the markers and the native AAO, respectively Patterns C,

D and E correspond to AAO, AAO + 1.4 M urea and AAO + 2.8 M urea, incubated with trypsin at 37 C for 30 min Patterns F, G and

H represent the same runs after 60 min incubation with trypsin.

as a function of hydrostatic pressure (Fig 4B) The

best fit parameters, representing the free energy and

volume changes of the transition, are reported in

Table 3 The average DG value ( 2.9 kcalÆmol)1) is closer than that previously obtained for GdmHCl- and urea-induced denaturation ( 3.4 kcalÆmol)1) in the transition from the native to the MD state [15] On the other hand, the small, negative volume change, DV1¼ )60 ± 4 mLÆmol)1 (Table 3), is consistent with that observed for many proteins undergoing partial dena-turation under pressure [25]

Denaturant- and pressure-induced effects on the bound copper

Copper ions are fundamental for the AAO redox activ-ity The enzyme contains four metal ions per subunit classified as mononuclear (type I Cu) and tri-nuclear (type II and type III Cu) copper centers, respectively [20] They are characterized by distinct optical proper-ties with absorption peaks at 610 (type I) and 330 nm (type II–III) [26] The absence of activity in the MD state [15] indicates that even a partial loss of tertiary structure has strong effects on copper We have there-fore investigated the copper sites by absorption spec-troscopy (Fig 5) Figure 5A demonstrates that the tri-nuclear copper center structure is already lost under quite mild unfolding denaturant concentrations or pressure values On the contrary, the type I Cu binding site is only partially affected by the addition of 1.4 m GdmHCl and, to a lesser extent, by high pressure, but only at pH 8 (Fig 5B)

Pressure-induced dissociation and unfolding of the MD species

A next set of experiments concerned the combined effect of GdmHCl (1.4 m) and pressure on AAO The fluorescence of AAO-bound ANS decreased at increasing pressure indicating a gradual dissociation

of the probe from the protein, due to the collapse of the enzyme tridimensional scaffolding (Fig 6A) This result was confirmed by intrinsic fluorescence meas-urements as shown in Fig 6B The center of mass of the steady state fluorescence spectrum was progres-sively shifted to longer wavelengths upon increasing

Fig 4 (A) Pressure-induced ANS binding to AAO The arrow

indi-cates the dependence of the ANS fluorescence intensity upon

increasing the hydrostatic pressure up to 3 · 10 8

Pa (ANS ⁄ protein

 10 : 1) The inset shows the ANS fluorescence intensity as a

function of pressure The best fitting parameters obtained by a

two-state fit (N 2 <–> MD) are reported in Table 3 (B) Effect of high

pressure the fluorescence center of mass of AAO at pH 6.0 (d)

and pH 8.0 (h) Solid lines represent a fit for two-state transition

(N 2 <–> MD) and the corresponding parameters are reported in

Table 3.

Table 3 Results of native to MD (N2 <–> MD) and MD to unfolded (MD <–> 2M* <–> 2U) pressure-induced transition fits.

Transition and fitting model v 2 DG1 [kcalÆmol)1] DV1[mLÆmol)1] DG2 [kcalÆmol)1] DV2[mLÆmol)1]

MD <–> 2M* <–> 2U TRP center of mass 1.1 10.6 ± 0.3 )192 ± 20 2.0 ± 0.1 )54 ± 4

the hydrostatic pressure In particular, at 3· 108Pa

the spectrum is almost completely red-shifted, as

expected for fully solvated tryptophan residues

(Fig 6B, inset) Both the intrinsic fluorescence

inten-sity and spectrum position are much more affected

than ANS fluorescence at low pressure values (105–

4· 107Pa), suggesting that the two fluorophores are

probing nonsimultaneous structural transitions

Indeed, at variance with the transition from the

native to the MD state, it was impossible to fit the

data reported of Fig 6B as a simple two-state process

More complex fits were therefore attempted, taking

into account a further intermediate state Because the

transition is protein-concentration dependent (Fig 6B,

inset) both ANS (Fig 6A) and tryptophan fluorescence

data (Fig 6B) have been interpreted according to the

following scheme:

MD < > 2M< > 2U

where M* and U represent partially and fully

unfol-ded monomers, respectively The overall free energy

change (DG1+ DG2) calculated from these

experi-ments ranges between 12.2 and 13.3 kcalÆmol)1

(Table 3), a value remarkably close to that previously

obtained for the transition of MD to the fully

unfol-ded state, upon chemical denaturation (urea or

GdmHCl [17]) The dissociation is accompanied by a

very large volume change (Table 3) and is inhibited by

glycerol (Fig 6B), a known stabilizing agent [27]

Figure 6 demonstrates that at 30% glycerol the

trans-ition curve was shifted by 5 · 107 Pa to higher pres-sure, while at 70% glycerol the displacement of the spectral center of mass was hardly more than 1.5 nm,

up to 2.5· 108 Pa The addition of glycerol did not produce detectable changes to the native protein (data not shown)

Discussion

Early events in the denaturation of AAO) preferential hydration at the domain interface The characterization of predissociated states plays a crucial role in the study of oligomeric proteins, yield-ing important information on the relationships between tertiary and quaternary interactions In fact, besides the stabilization of the tridimensional structure, these interactions regulate volume fluctuation and con-formational changes, i.e., those dynamic properties that actually keep the enzymes working Independent measurements on AAO partially unfolded by GdmHCl, such as dynamic fluorescence, anisotropy and light scattering provide quantitative structural infor-mation along the denaturation pathway (Tables 1–3)

In particular, the slower rotational dynamics (/2

125 ns) of the dimeric intermediate suggests the occur-rence of a swelling effect, probably associated with the progressive hydration of the protein tertiary structure The pressurization of native AAO in the range 105–

3· 108 Pa also produces a predissociated intermediate

0.006

0.008

0.010

0.012

wavelength (nm)

A

0.000 0.004 0.008 0.012

wavelength (nm)

B

Fig 5 Dependence of the AAO absorption bands on pressure (10 5 Pa, d; 6 · 10 7 Pa pH 6, ––; 6 · 10 7 Pa pH 8, —) The spectrum at 10 5 Pa

in the presence of 1.4 M GdmHCl or 2.8 M urea has been reported for comparison (h) (A) represents the near UV absorption (type II–III tri-nuclear copper center), while (B) corresponds to visible absorption (type I, blue copper center).

where the content of residual structure is

pH-depend-ent (Figs 1A and 4B) The data indicate that pressure

generates an intermediate much alike the

denaturant-induced MD state, showing a larger heterogeneity of

the emission decay (Fig 1B) and an enhanced

propen-sity to bind ANS (Fig 4A)

The effect of high pressure on proteins’ stability has been extensively studied in the last decade, in order to gain new insight into its peculiar denaturation mechan-ism The protein’s interior is known to behave as a solid particle, its compressibility being generally very low [28–30] In the case of monomeric globular pro-teins, pressure unfolding has therefore been associated with a partial hydration of the peripheral protein regions, in the presence of a still and stiff hydrophobic core [30,31] On the other hand, the behavior of large proteins is more complex For instance, dimeric enzymes display both elastic and anelastic changes prior to subunit dissociation [32] Moreover, large pro-teins gain a greater flexibility through the reciprocal movements of their domains [33], suggesting that their weak points with respect to both physical and chemical stress are mainly located in their interdomain regions rather than inside each individual domain The X-ray structure of AAO revealed the existence of three dis-tinct domains per subunit [20,21], sharing a common b-barrel topology This particular scaffolding can indeed account for the protein resistance to high pres-sure in the range 105–3· 108 Pa Beta structures show

a lower compressibility than loops and a-helical regions [34] Furthermore, they are also the most appropriate shelters for hydrophobic residues from water, according to a recent report on the accessible surface of about 600 protein structures [35]

Even though volume changes upon protein unfold-ing may arise from different sources [36], the so called electrostriction effect seems to play a major role in the pressure-induced MD state of AAO [25,29,30,37] Such

a hypothesis is not only supported by the strong pH-dependence observed in the first pressure-induced transition, but also by the presence of an uncommonly large number [46] of ion pair interactions [21], 13 of which occur between different domains of the same subunit This finding and the relapse of the tri-nuclear copper site (located between domains 2 and 3) at only

6· 107 Pa (Fig 5A) are further evidence that the swelling effect from the native to the MD state arises from the preferential hydration of interdomain sites, rather than of the domains’ core In this model, the hydration of the interdomain regions would recall the mechanism that characterizes the early denaturation steps of the peripheral shell of smaller monomeric pro-teins This is in keeping with the overall similarity of the three domains in AAO each of which mimics the small blue copper proteins [20,21] The reversibility principle would require that the formation of interdo-main contacts within the same subunit was a very late event of the folding process It is important to recall that in all copper proteins studied the metal is not

A

B

Fig 6 (A) Pressure-induced ANS dissociation from the AAO MD

intermediate (at 1.4 M GdmHCl) The arrow indicates the

depend-ence of the ANS fluorescdepend-ence intensity upon increasing the

exter-nal hydrostatic pressure from 10 5 –3 · 10 8 Pa The inset shows the

ANS fluorescence intensity as a function of pressure The best

fit-ting parameters (Table 3) were obtained by a three-state fit (MD

<–> 2 M * <–> 2U) (B) Effect of high pressure and glycerol on the

intrinsic fluorescence of the MD intermediate Filled symbols

repre-sent the spectral center of mass in the absence (circles), or

pres-ence of 30% (diamonds) and 70% (triangles) glycerol Empty

circles correspond to the change of the total fluorescence intensity

in the absence of glycerol The three state (MD <–> 2M* <–> 2U)

fits of the tryptophan center of mass and total intensity as a

func-tion of pressure are reported as solid lines The results of the two

fits are shown in Table 3 The inset shows the spectra of the MD

state at 1.8 · 10 8

Pa at two different AAO concentrations, namely

6 l M (solid line) and 0.2 l M (short dashes) The spectrum at

 3 · 10 8 Pa (0.2 l M , dotted line) and the spectrum of the fully

unfolded AAO (105Pa in the presence of 3.5 M GdmHCl, long

dashes) are also reported for comparison.

required for folding Furthermore it is possible to

obtain partially or fully copper-depleted AAO without

significantly affecting the structure

The role of dimerization in the folding strategy of

a large protein

The combination of low denaturant concentration and

pressure opens new possibilities to the study of protein

folding [17], allowing us to trap intermediates

charac-terized by a smaller volume than that of the native

state, undetectable with traditional denaturation

tech-niques The ‘softer’ action of hydrostatic pressure has

also been proposed to increase the local roughness of

the folding energy landscape [38], depending on the

protein’s structural features Our results are in keeping

with these ideas In fact, no stable monomeric species

has been previously observed upon urea or GdmHCl

denaturation of AAO [15] However, the analysis of

the GdmHCl + pressure unfolding curves (Fig 6A,B

and Table 3) demonstrates that monomeric species

might be formed when the MD state at 1.4 m GdmHCl

is pressurized The addition of glycerol, which is

known to reduce the size of internal cavities [27],

coun-teracts the loss of the protein quaternary structure

(Fig 6B) Therefore, at variance with the

electrostric-tion mechanism, which might explain the volume

chan-ges from the native to the MD states, the collapse of

the voids at the dimeric interface might be the main

reason for the unusually large negative DV value

()200 mLÆmol)1) observed upon pressure-induced

dissociation Indeed, the visual inspection of the AAO

structure demonstrates that a large gap exists between

the two subunits (Fig 7A) A more quantitative

calcu-lation of the AAO gap volume index, i.e., the ratio

between the gap volume and the interface area [39,40],

shows that it is by far the largest value obtained

among the available crystallographic structures of

dimeric proteins (Fig 7B) This is due to both the gap

volume and the very small size of the contact interface,

which in fact involves only 25 amino acid side chains

per subunit Interestingly, these residues are not

ran-domly distributed along the two polypeptide chains,

but rather form 3–4 stretches of neighbor, if not

adja-cent, residues This arrangement would suggest for

AAO a very early quaternary interaction between a

few patches of neighbor residues, probably in order to

stabilize only partially folded monomeric species A

recent study of a small homodimeric protein, namely

the factor for inversion stimulation, has demonstrated

that the formation of a dimeric intermediate can

indeed be a strategy to accelerate the folding process

[41] Unfortunately detailed kinetic folding studies on

larger oligomeric proteins are still at the early stages, but there are examples in which a fast subunit associ-ation occurs before the final folding state is achieved [42] On the other hand, if the folding of the AAO mo-nomers is strictly intertwined with the formation of the quaternary interactions, one might expect to find clues

of this process in the same folded protein For instance, an early dimerization followed by a collapse toward a more compact structure might lead to larger inclusion of solvent molecules within the protein mat-rix The crystallographic model of AAO reported in Fig 7A supports this hypothesis In fact it appears that several water molecules are entrapped in cavities not accessible from the outside More interestingly, their distribution is uneven, with the area near the dimeric interface being crowded by water molecules

B A

Fig 7 (Top) Crystallographic model of dimeric AAO (PDB file 1AOZ) obtained by removing all water molecules at the protein sur-face The AAO structure has been reported in the transparence mode, in order to show the water molecules fully entrapped within the protein matrix (purple) (Bottom) Gap volume index (i.e., the ratio between the total empty gap volume at the dimeric interface and the dimeric interface area) of  30 dimeric proteins as a func-tion of their size (the black bar corresponds to AAO).

Water plays a major, dual role in proteins’ life [43,44]

The final amount of water molecules entrapped within

a protein structure arises as the best compromise

among several requirements, namely folding, stability

and functionality In the case of enzymes characterized

by a quaternary structure these features are particularly

critical to understand the reason for oligomerization,

when no simple explanation (e.g., allosteric effects) hold

[2,3] The analysis of specific cases, such as the folding

process of AAO, can be therefore paradigmatic for

other large dimers The stabilization of the partially

folded monomers through nonlocal interactions within

each subunit’s tertiary structure is entropically costly,

as it would drastically reduce the degrees of freedom of

the polypeptide chain [45] We propose that quaternary

interactions are fundamental to stabilize each subunit

in AAO and to drive the final folding from a hydrated

MD state to the native conformation Although a

gen-eralization is premature, it will be interesting to see

whether, and to what extent, the stabilization through

similar intermediates is a potential, more general trick

to drive (and probably accelerate) the overall assembly

process, thereby influencing the balance between

enthal-pic and entroenthal-pic contributions After all, such a

mech-anism would also avoid exceedingly high stabilization

energy barriers for the folding of large dimers at the

expense of water inclusion in the final structure The

relatively low stabilization energy of large molecules

could be a natural, built-in advantage that might also

allow a more rapid turnover of these proteins in vivo

Experimental procedures

Materials

Ultrapure GdmHCl and urea were purchased from USB

(United States Biochemicals; Cleveland, OH, USA);

dansyl-chloride and ANS were purchased from Sigma (St Louis,

MO, USA) Ascorbate oxidase from green zucchini was

purchased from Boehringer Mannheim (Mannheim,

Germany) and dissolved in 80 mm potassium phosphate

buffer, pH 6, at 20C, unless otherwise specified

Protein dansylation

AAO in potassium-phosphate buffer (pH 8) was

incuba-ted, at 20C, in the presence of dansylchloride (ratio

 1 : 30) After 24 h, the dansylchloride in excess was

removed, by filtration of the solution through a D-salt

ex-cellulose plastic desalting column (exclusion limit Mr5000,

equilibrated at pH 6), thus isolating the dansylated protein

fractions

Spectroscopic assays Steady state fluorescence spectra and anisotropy were recor-ded on a photon counting spectrofluorometer equipped with Glan Thompson polarizers (ISS, model K2, Champaign, IL) Dynamic fluorescence measurements were carried out using the phase-shift and demodulation technique [18] The light sources of the dynamic fluorometer (ISS, model Koala) were either a K180 laser diode, with emission wavelength at

370 ± 8 nm or an arc-xenon lamp modulated in the range 2–

200 MHz The data analysis was performed using the glo-bals unlimited software (LFD, Urbana, IL, USA) [46] Steady state and dynamic fluorescence at high pressure were measured with the same instruments, using the high pressure ISS cell, equipped with an external bath circulator Light scattering measurements were performed on a Horiba (Kyoto, Japan) LB-500 dynamic light scattering nanoparticle size analyzer, equipped with a 650 nm, 5 mW laser diode Data analysis was performed using the accompanying soft-ware based on a Fourier-transform deconvolution procedure Absorption spectra in the range  300–700 nm were recorded connecting the pressure cell to a PerkinElmer (Wellesley, MA, USA) Lambda 18 spectrometer, through a couple of Hellma (041.002-UVS) fiber optic cables (Hellma, Milan, Italy)

Theoretical estimation of the average rotational correlation time

A preliminary, rough estimation of the overall rotational correlation time for a spherical molecule with the same size

of AAO has been obtained using the following approxima-ted relationship [19]:

Usph¼ 1=6Drot ðMr=2:4Þ1012s 58 ns where Drotand Mrrepresent the rotational diffusion coeffi-cient and the protein molecular mass (140 000 Da), respect-ively A more reliable value has been instead obtained considering that the AAO dimer is a prolate ellipsoid with semiaxes a 110 A˚ and b  55 A˚ [20], yielding the follow-ing relative rotational correlation values [19]:

Ua=Usph¼ 1:51

Ub=Usph¼ 1:05 i.e., Fa  88 ns and Fb  61 ns The harmonic mean of these values (i.e., the parameter obtained by anisotropy dynamic measurements) is therefore:

<U> 68 ns

Molecular graphics The model of dimeric AAO was obtained from the PDB file 1AOZ [20,21] and elaborated with the visual molecular dynamics(vmd) software (Urbana, IL, USA) The analysis

of the dimeric interface of AAO was carried out using the

‘Protein–Protein’ Interaction Server (http://www.biochem

ucl.ac.uk/bsm/PP/server/ [39,40])

References

1 Marianayagam NJ, Sunde M & Matthews JM (2004)

The power of two: protein dimerization in biology

Trends Biochem Sci 29, 618–625

2 Mei G, Di Venere A, Rosato N & Finazzi Agro’ A

(2005) The importance of being dimeric Febs J 272,

16–27

3 Ali MH & Imperiali B (2005) Protein oligomerization:

how and why Bioorg Med Chem 13, 5013–5020

4 Jaenicke R & Lilie H (2000) Folding and association of

oligomeric and multimeric proteins Adv Prot Chem 53,

329–401

5 Brockwell DJ, Smith DA & Radford SE (2000) Protein

folding mechanisms: new methods and emerging ideas

Curr Opin Struct Biol 10, 16–25

6 Radford SE (2000) Protein folding: progress made and

promises ahead Trends Biochem Sci 25, 611–618

7 Fersht AR & Daggett V (2002) Protein folding and

unfolding at atomic resolution Cell 108, 573–582

8 Wallace LA & Matthews CR (2002) Sequential vs

par-allel protein-folding mechanism: experimental tests for

complex folding reactions Biophys Chem 101–102, 113–

131

9 Daggett V & Fersht AR (2003) Is there a unifying

mechanism for protein folding? Trends Biochem Sci 28,

18–25

10 Gianni S, Guydosh NR, Khan F, Caldas TD, Mayor U,

White GWN, De Marco ML, Daggett V & Fersht AR

(2003) Unifying features in protein-folding mechanisms

Proc Natl Acad Sci 100, 13286–13291

11 White GWN, Gianni S, Grossmann JG, Jemth P,

Fersht AR & Daggett V (2005) Simulation and

experi-ment conspire to reveal cryptic intermediates and a slide

from the nucleation-condensation to framework

mechanism of folding J Mol Biol 350, 757–775

12 Brooks CL III, Onuchic JN & Wales DJ (2001) Taking

a walk on a landscape Science 293, 612–613

13 Onuchic JN, Schulten ZL & Wolynes PG (1997) The

energy landscape perspective Annu Rev Phys Chem 48,

545–600

14 Di Venere A, Mei G, Gilardi G, Rosato N, De Matteis F,

McKay R, Gratton E & Finazzi Agro’ A (1998)

Resolution of the heterogeneous fluorescence in

multi-tryptophan proteins: ascorbate oxidase Eur J Biochem

257, 337–343

15 Mei G, Di Venere A, Buganza M, Vecchini P, Rosato N

& Finazzi Agro’ A (1997) The role of the quaternary

structure in the stability of dimeric proteins: the case of

ascorbate oxidase Biochemistry 36, 10917–10922

16 Kuwajima K (1989) The molten globule state as a clue for understanding the folding and cooperativity of glob-ular-protein structure Proteins 6, 87–103

17 Perrett S & Zhou JM (2002) Expanding the pressure technique: insights into protein folding from combined use of pressure and chemical denaturants Biochim Biophys Acta 1595, 210–223

18 Lakowicz JR (1999) Principles of Fluorescence Spectros-copy, 2nd edn Kluwer Academic⁄ Plenum Press, New York, NY

19 Cantor CR & Schimmel PS (1980) Biophysical Chemis-try, Vol II Freeman, San Francisco, CA

20 Messerschmidt A, Rossi A, Ladenstein R, Huber R, Bolognesi M, Gatti G, Marchesini A, Petruzzelli R & Finazzi Agro’ A (1989) X-ray crystal structure of the blue oxidase ascorbate oxidase from zucchini J Mol Biol 206, 513–529

21 Messerschmidt A, Ladenstein R, Huber R, Bolognesi M, Avigliano L, Petruzzelli R, Rossi A & Finazzi Agro’ A (1992) Refined crystal structure of ascorbate oxidase at 1.9 A˚ resolution J Mol Biol 224, 179–205

22 Silva JL & Weber G (1993) Pressure stability of pro-teins Annu Rev Phys Chem 50, 245–252

23 Lakowicz JR (1983) Principles of Fluorescence Spectros-copy Plenum Press, New York, NY

24 Semisotnov GV, Rodionova NA, Razgulyaev OI, Uversky VN, Gripas’ AF & Gilmanshin RI (1991) Study of the ‘Molten Globule’ intermediate state in protein folding by a hydrophobic fluorescent probe Biopolymers 31, 119–128

25 Royer CA (2002) Revisiting volume changes in pres-sure-induced protein unfolding Biochim Biophys Acta

1595, 201–209

26 Avigliano L & Finazzi Agro’ A (1997) Biological Function and Enzyme Kinetics of Ascorbate Oxidase

In Multi-Copper Oxidases (Messerschmidt A, ed), pp 265–278 World Scientific Publishing Co Pte Ltd, Singapore

27 Priev A, Almagor A, Yedgar S & Gavish B (1996) Gly-cerol decreases the volume and compressibility of pro-tein interior Biochemistry 35, 2061–2066

28 Gavish B, Gratton E & Hardy CJ (1983) Adiabatic compressibility of globular proteins Proc Natl Acad Sci

80, 750–754

29 Kharakoz DP (2000) Protein compressibility, dynamics and pressure Biophys J 79, 511–525

30 Chalikian TV (2003) Volume tric properties of proteins Annu Rev Biophys Biomol Struct 32, 207–235

31 Taulier N & Chalikian TV (2002) Compressibility

of protein transitions Biochim Biophys Acta 1595, 48–70

32 Cioni P & Strambini GB (1996) Pressure effects on the structure of oligomeric proteins prior to subunit disso-ciation J Mol Biol 263, 789–799