Unfolding and refolding of porcine odorant binding protein in guanidinium hydrochloride: equilibrium...

www.bba-direct.com

Biochimica et Biophysica Acta 1652 (2003) 115–125

Unfolding and refolding of porcine odorant binding protein in

guanidinium hydrochloride: equilibrium studies at neutral pH

Mariella Parisia,b, Alberto Mazzinia,b, Robert T. Sorbia,b,Roberto Ramonic, Stefano Grollic, Roberto Favillaa,d,*

a Istituto Nazionale di Fisica della Materia (INFM), Unita di Parma, Universita di Parma, Parma, ItalybSez. Biofisica, Dipartimento di Fisica, Universita di Parma, Parma, Italy

cDipartimento di Produzioni Animali, Biotecnologie Veterinarie, Qualita e Sicurezza degli Alimenti, Universita di Parma, Parma, ItalydDipartimento di Biochimica e Biologia Molecolare, Universita di Parma, Parma, Italy

Received 19 February 2003; received in revised form 8 August 2003; accepted 27 August 2003

Abstract

Unfolding and refolding studies on porcine odorant binding protein (pOBP) have been performed at pH 7 in the presence of guanidinium

hydrochloride (GdnHCl). Unfolding, monitored by following changes of protein fluorescence and circular dichroism (CD), was found to be a

reversible process, in terms of recovered structure and function. The equilibrium transition data were fitted by a simple two-state sigmoidal

function of denaturant concentration and the thermodynamic folding parameters, derived from the two techniques, were very similar (average

values: C1/2c 2.4 M, mc 2 kcal mol� 1 M� 1, DGunf,w0 c 4.7 kcal mol� 1). The transition was independent of protein concentration,

indicating that only monomeric species are involved. Only a minor protective effect by the fluorescent ligand 1-amino-anthracene (AMA)

against protein unfolding was detected, whereas dihydromyrcenol (DHM) stabilised the protein to a larger extent (DC1/2c 0.5 M). Refolding

was complete, when the protein, denatured with GdnHCl, was diluted with buffer. On the other hand, refolding by dialysis was largely

prevented by concomitant aggregation. The present results on pOBP are compared with those on bovine OBP (bOBP) [Biochim. Biophys.

Acta 1599 (2002) 90], where subunit folding is accompanied by domain swapping. We finally suggest that the generally observed two-state

folding of many lipocalins is probably favoured by their h-barrel topology.D 2003 Elsevier B.V. All rights reserved.

Keywords: Lipocalin; Odorant binding protein; Folding; Denaturant; Guanidinium chloride

1. Introduction have evolved chaperones to assist folding in vivo [2].

It is well documented that many single domain proteins

fold rapidly and spontaneously in vivo as well as in vitro [1],

in contrast to larger multidomain proteins, which are more

prone to aggregation. It is mainly for this reason that cells

1570-9639/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/j.bbapap.2003.08.009

Abbreviations: OBP(s), odorant binding protein(s); pOBP, porcine

odorant binding protein; bOBP, bovine odorant binding protein; GdnHCl,

guanidinium hydrochloride; DHM, dihydromyrcenol; AMA, 1-amino-

anthracene; NATA, N-acetyl-tryptophan-amide; FLU, fluorescence; FRET,

fluorescence resonance energy transfer; CD, circular dichroism; LS, light

scattering; SEC, size exclusion chromatography; PHOSPHATE, 20 mM

phosphate buffer pH 7; TRIS, 20 mM Tris–HCl buffer at pH 7.8

* Corresponding author. Dipartimento di Biochimica e Biologia

Molecolare, Parco Area delle Scienze 23, Universita degli Studi di Parma,

43100 Parma, Italy. Tel.: +39-0521-905-488; fax: +39-0521-905223.

E-mail address: [email protected] (R. Favilla).

However, at sufficiently high concentration, even small

proteins may also undergo misfolding and aggregation,

during renaturation [3]. For this reason, misfolding has now

been recognised as a highly important problem in biotech-

nology and medicine [4]. The new view about aggregation is

that it seems to be preferentially ruled by rather specific

interactions between partially unfolded intermediates with

hydrophobic surfaces exposed to solvent for sufficiently long

times [5].

Odorant binding proteins (OBPs) are small extracellular

proteins belonging to the superfamily of lipocalins [6],

among which are retinal binding protein, major urinary

protein and h-lactoglobulin (BLG). Though remarkably

different in their sequence, lipocalins present a highly

conserved tertiary structure, with a characteristic fold,

formed by an eight-stranded antiparallel h-barrel, where

M. Parisi et al. / Biochimica et Biophysica Acta 1652 (2003) 115–125116

several hydrophobic ligands can bind, and a short C-terminal

a-helix. The X-ray diffraction structures of both porcine and

bovine OBPs (pOBP and bOBP, respectively) have been

resolved [7,8], but only in the case of bOBP the C-terminal

a-helix is involved in domain swapping. Most OBPs are

monomers, but a few are stable as homodimers. For example,

pOBP is a monomer of 157 residue with a molecular mass of

18 kDa, while bOBP is composed of identical subunits held

together by domain swapping.

The biological role played by lipocalins seems to be

mainly involved in the transport of relatively small hydro-

phobic molecules, among them retinal, pheromones etc. [9].

In particular, OBPs are thought to be involved in the transport

of odorants to the olfactory receptors [10,11]. Recently,

however, other functional roles of OBPs have been proposed,

such as detoxification from lipid peroxidation [12] and/or

protection against insects carrying parasites [13].

With this background, we have investigated the unfold-

ing and refolding processes of pOBP extracted from nasal

mucosa, in the presence of guanidinium hydrochloride

(GdnHCl) at neutral pH. Because of the reversibility of the

unfolding process, we have been able to derive the thermo-

dynamic transition parameters. Aggregation has been ob-

served only when the protein has been refolded by dialysis.

The processes of unfolding/refolding have been monitored

by fluorescence (FLU), circular dichroism (CD), size exclu-

sion chromatography (SEC) and ligand binding experiments,

that is, following both intrinsic (protein) and extrinsic (bound

ligand) properties, whereas aggregation was monitored by

the increase of light scattering (LS) intensity. Since a similar

study on the orthologous dimeric bOBP has recently been

Fig. 1. Fluorescence spectra of 5 AM pOBP. Normalized uncorrected fluorescenc

PHOSPHATE with 4 M GdnHCl for 24 h; (- - -) renatured, 24 h after a 10 times

spectra of pOBP in the absence (—) and presence (. . .) of 4 M GdnHCl. kex 295

published [14], the results are compared to remark the pecu-

liar role played by domain swapping in bOBP dimerisation.

A comparison with the folding behaviour of other lip-

ocalins under similar conditions (i.e. at equilibrium and

neutral pH) is finally attempted to point out the probable

relevance of the common h-barrel topology.

2. Materials and methods

2.1. Protein extraction and purification

pOBP was purified from nasal porcine mucosa as previ-

ously described [14], with minor modifications. Briefly, pig

nasal mucosa (2 ml/g) was suspended in 20 mM Tris–HCl

buffer at pH 7.8 (TRIS) and homogenised in a Waring

blender. The suspension was centrifuged (48,000 � g, 30

min) and the supernatant fractionated with ammonium sulfate

in two steps: 60% and 100%. The precipitate was resus-

pended, dialysed against TRIS and purified by FPLC: first on

a Baker bond (Baker) column and then on a monoQ (Amer-

sham Pharmacia Biotech). Only fractions able to bind 1-

amino-anthracene (AMA) were stored for further investiga-

tions. Protein functionality was quantitatively determined by

measuring the ability of the protein to bind AMA: a ratio of

about 0.9 mol of AMAbound per mole of protein subunit was

usually found [13]. The degree of purity of these fractions

was evaluated by SDS-PAGE under reducing conditions.

For spectroscopic measurements, stock solutions of

pOBP in TRIS were transferred into 20 mM phosphate

buffer at pH 7 (PHOSPHATE) by ultrafiltration (Centricon

e spectra of 5 AM pOBP: (—) native, in PHOSPHATE; (. . .) denatured, indilution with PHOSPHATE from 4 M GdnHCl. Inset: Actual fluorescence

nm, 20 jC.

Fig. 2. CD spectra of pOBP. (A) Far-UV region; (B) near-UV region. In the

far-UV region, pOBP was 5 AM (cell pathlength = 2 mm) when native or

renatured, and 150 AM (cell pathlength = 0.1 mm) when denatured. In the

near-UV region, the protein was 50 AM in all cases (cell pathlength = 1 cm).

(—) native in PHOSPHATE; (. . .) denatured in PHOSPHATE with 4 M

GdnHCl; (- - -) renatured from 4 M GdnHCl by 10 times dilution with

PHOSPHATE. Each trace is an average of three accumulations. All

measurements at 20 jC.

M. Parisi et al. / Biochimica et Biophysica Acta 1652 (2003) 115–125 117

Plus-20, 10,000 MWCO, Millipore). Protein concentration

was evaluated using an extinction coefficient of 12,700

M� 1 cm� 1 at 278 nm, determined by the method of

Edelhoch, as described by Pace et al. [15].

2.2. Buffers and ligands

All solutions were prepared using deionised water, fur-

ther purified by means of a Milli-Q Millipore system. AMA

was purchased from Fluka, GdnHCl (z 99% pure) from

Sigma and dihydromyrcenol (DHM) from Aldrich. NaCl

and buffer salts were Carlo Erba analytical grade reagents.

2.3. Chemical denaturation and renaturation by GdnHCl

2.3.1. Denaturation

pOBP (usually c 5 AM) in PHOSPHATE was denatured

by addition of variable amounts of PHOSPHATE containing

6 M GdnHCl, to modulate the final concentration of GdnHCl

in the 0–5M range.Measurements were performed 24 h later,

leaving the solutions at room temperature. By that time, all

solutions were at equilibrium, as demonstrated by the absence

of any further change of their spectroscopic properties.

2.3.2. Renaturation

pOBP, previously denatured for 24 h in PHOSPHATE

containing 4 M GdnHCl, was refolded following one of the

three following methods: (a) dilution of denaturant by

addition of appropriate amounts of PHOSPHATE; (b) re-

moval of denaturant by centrifugal ultrafiltration (Centricon

Plus-20 filters); or (c) by dialysis. The concentration of

GdnHCl stock solutions was checked by refractometry.

2.4. Spectroscopic measurements

Absorbance measurements were performed on an 8453

model Hewlett-Packard diode array spectrophotomer. Fluo-

rescence measurements were carried out on an LS-50 Perkin-

Elmer spectrofluorimeter on samples with optical density

< 0.1 at the excitation wavelength, to avoid inner filter effects.

Fluorescence spectra were recorded using a 4� 10 mm

quartz cuvette, with 4 mm optical path and 5 nm slit in

excitation and 10 mm and 10 nm in emission. Maximum

emission wavelengths and areas under each spectrum were

calculated using a Gaussian fit function, after data conver-

sion from nanometers to wave numbers, according to the

following formula:

y ¼ A

wffiffiffiffip2

q e�2

ðx�x0Þ2

w2 ð1Þ

where A is the area, x0 the maximum position and w the full

width at half height. Protein aggregation, when present, was

detected as an apparent absorbance tail above 330 nm.

Quantitative data on LS intensity were recorded at 340 nm

for both excitation and emission, using a small volume

(4� 10 mm) cuvette.

CD measurements were performed on a J500-A Jasco

spectropolarimeter. Far-UV-CD and near-UV-CD spectra

were taken with 10 and 2 or 0.1 mm pathlength cells,

respectively. All samples were thermostated at 20 jC.

2.5. Fluorescence quenching and energy transfer

Solvent accessibility to the single Trp residue of pOBP

was estimated according to Stern–Volmer relationship:

F0=F ¼ 1þ KSV½Q� ð2Þ

where F0 and F are the fluorescence intensities in the

absence and presence of the quencher Q and KSV is the

Stern–Volmer quenching constant, also given by:

KSV ¼ kqs0

kq being the bimolecular quenching constant and s0 the

average lifetime of the fluorophor in the absence of


quencher. Lifetime measurements were performed with a

home assembled instrumentation, based on the time-corre-

lated single photon counting technique, previously de-

scribed [16]. Briefly, a thyratron-gated flash lamp (IBH

model 5000F) was filled with nitrogen (0.9 bar) and used at

50 kHz. The stop photomultiplier was a Philips XP2020 Q

model and the fast NIM electronics was from Ortec. Counts

were collected in 512 channels with a 107-ps time resolu-

tion and analysed by iterative convolution, using a non-

linear least-squares procedure based on the Marquard

algorithm.

Energy transfer efficiency from the single Trp residue of

pOBP to the bound ligand (AMA) was estimated using the

following relationship:

ET ¼ 1� FDA=FD ð3Þ

where FDA and FD are the protein fluorescence intensity in

the presence and absence of ligand, respectively. According

to the Foerster theory [17], a critical distance R0, at which

the energy transfer efficiency is 0.5, can be found for any

given donor–acceptor couple from the following formula:

R60 ¼

9000ðln10Þk2QD

128p5Nn4

Z l

0

FDðkÞeAðkÞk4dk ð4Þ

where QD is the quantum yield of the donor (Trp-16 of

pOBP) in the absence of acceptor (AMA); N is the Avoga-

dro’s number, n is the refractive index of the medium; FD(k)is the normalised fluorescence intensity of the donor; eA(k)the extinction coefficient wavelength function of the accep-

tor; k2 is a factor describing the relative spatial orientation of

the transition dipoles of the donor and the acceptor: in the

Fig. 3. Binding curves with AMA. Comparison between native (.) and renatured p

The fluorescence intensity was measured at 480 nm (kex = 380 nm) 24 h after ad

absence of any structural information, k2 is usually assumed

to be 2/3, corresponding to freely rotating dipoles. Once ET

and R0 have been estimated, an average donor–acceptor

distance R can be derived according to the following

relationship:

R ¼ R0ð1=ET � 1Þ1=6 ð5Þ

When X-ray structural data are available, R can be directly

estimated and its value used to check if energy transfer data

are consistent with it. If not, the donor and acceptor

transition dipoles are not randomly oriented or hindered in

their rotation, so that k2 differs from 2/3. In the limiting

case, in which they are mutually perpendicular, no reliable

Foerster distance can be derived from fluorescence reso-

nance energy transfer (FRET) measurements [18].

2.6. SEC measurements

SEC runs were performed by FPLC using a Superdex 75

16/60 gel filtration column (Amersham Pharmacia Bio-

tech), equilibrated with PHOSPHATE containing 0.1 M

NaCl and different amounts of GdnHCl from 0 to 2 M.

Apparent MW estimates of pOBP were obtained by interpo-

lation of calibration curves built up with standard proteins

(Pharmacia low MW calibration kit) eluted under similar

conditions.

2.7. Fitting of unfolding/refolding equilibrium data

Unfolding and refolding data were best fitted as function

of denaturant concentration by the following simple two-

rotein by dilution (5) and dialysis (E). pOBP was 1 AM in PHOSPHATE.

dition of the probe at 20 jC.


state sigmoidal function, where yF, yU, C1/2 and m are the

free parameters:

y ¼yF þ yU exp

½�mðC1=2 � CÞ�RT

� �

1þ exp½�mðC1=2 � CÞ�

RT

� � ð7Þ

With the best fit values of m and C1/2, the standard free

energy change of unfolding in buffer DGun,w0 can be esti-

mated according to: DGun,w0 =mC1/2.

All experiments have been performed in duplicate or

triplicate, so that the average values of the experimental

parameters are shown in the figures.

Fig. 4. Energy transfer from the protein to bound AMA. (A) Fluorescence

spectra of pOBP and AMA (kexc = 295 nm): (—) 5 AM pOBP; (- - -) 5 AMpOBP plus 30 AM AMA; (. . .) 30 AM AMA. All measurements were made

in PHOSPHATE using a 0.4-cm pathlength cell. (B) X-ray structure of

pOPB (pdb file 1A3Y) with AMA bound in its calyx. AMA is shown in the

same position it occupies in the homologous crystal of bOBP (pdb file

1HN2), to visualise the distance between AMA and Trp-16 in pOBP.

3. Results

3.1. Spectroscopic and functional properties of pOBP

Despite the fact that the protein contains only one Trp

and five Tyr residues, its fluorescence spectrum is largely

dominated by the Trp residue, when excited at 280 nm. This

is usually observed with most Trp-containing proteins,

because of energy transfer from Tyr to Trp residues [18].

However, to get rid of possible contributions from Tyr

residues, the protein was excited at 295 nm, where only

tryptophan absorbs. Under these conditions, the protein

shows a fluorescence spectrum with an emission maximum

at 346 nm, typical of a partially buried Trp residue (Fig. 1).

The spectra of the denatured and renatured protein are

also shown in the same figure. The former is red-shifted

(emission maximum at 358 nm) and more intense than that

of the native protein. Unfolding was preferentially moni-

tored following the shift of the emission maximum rather

than the change of the total fluorescence intensity, because

of the relatively larger increase of the former (Fig. 1, inset).

The spectrum of the renatured protein is almost indistin-

guishable from that of the native protein.

The far-UV-CD spectrum of native pOBP shows a trough

near 217 nm (Fig. 2A), typical of proteins with high h-structure content. The near-UV-CD spectrum shows a split

negative band around 280 nm (Fig. 2B), likely due to a rigid

environment around the aromatic residues, with possible

contributions from the unique disulfide bridge [19]. In the

presence of 4 M GdnHCl, these spectral characteristics are

largely lost, as a consequence of protein unfolding, as shown

in the same figures. Due to the strong absorbance of GdnHCl

in the far-UV, the CD spectrum of denatured pOBP could be

acquired only down to 208 nm, using a concentrated (150

AM) protein sample in a 0.1-mm pathlength cell.

Previous studies on the binding of AMA to pOBP

resulted in an almost 1:1 stoichiometry and a good affinity

(Kdc 1 AM) [20]. We exploited this property by following

the strong increase of fluorescence intensity and the con-

comitant blue shift of the emission peak from 550 to 480 nm

accompanying the binding of the ligand to the protein. The

binding curves, shown in Fig. 3, were obtained at pH 7

using the emission data recorded at 480 nm, after subtrac-

tion of the FLU intensity of free AMA. The curve for the

native protein is almost superimposable to that for the

protein renatured by dilution, in contrast to that for the

protein renatured by dialysis, which shows a largely reduced

binding capacity of AMA, though the affinity seems to be

almost unaffected.

Energy transfer from the single Trp residue of pOBP to

AMA was also observed. This conclusion is derived from

the observation that the fluorescence spectrum of the protein

without AMA (kex = 295 nm) is about twice more intense

than in the presence of AMA. At the same time, the

fluorescence spectrum of AMA, when excited at the same

wavelength, results much more intense in the presence than

in the absence of pOBP (Fig. 4). From the ratio of the

fluorescence spectra of pOBP in the absence and presence

Table 1

Transition parameters for GdnHCl-induced unfolding of pOBP in

PHOSPHATE at 20 jC

Transition parametersa DGw0

(kcal mol� 1)

m (kcal

mol� 1 M� 1)

C1/2 (M)

CD at 225 nm 4.6F 0.5 1.9F 0.2 2.40F 0.03

Protein fluorescence (kmax) 4.7F 0.3 2.0F 0.1 2.37F 0.02

Protein fluorescence (kmax)

with AMA

5.2F 0.3 2.0F 0.1 2.60F 0.02

Protein fluorescence (kmax)

with DHM

8.6F 0.4 2.9F 0.1 2.97F 0.01

AMA fluorescence

(intensity at kmax)

6.0F 0.8 2.4F 0.3 2.49F 0.04

Acrylamide quenching

rate constant

5.3F 0.6 1.8F 0.2 2.93F 0.03

Protein fluorescence kmax

(0.5 M acrylamide)

5.4F 1.5 1.7F 0.4 3.16F 0.11

a The values of m and C1/2 represent the best fit values, using Eq. (5) on the

average value of each point along the transition curve. The corresponding

absolute errors are on the second decimal digit for C1/2 and on the first

decimal digit for m, hence also for DGw0 =mC1/2.

Fig. 5. Unfolding equilibrium curve of pOBP as a function of GdnHCl. (A)

Unfolding transition curves of pOBP obtained by following fluorescence

kmax at three protein concentrations: (5) 1 AM, (.) 5 AM and (o) 50 AM,

all in PHOSPHATE. (B) Molar fraction of denatured protein (24 h in

PHOSPHATE with 4 M GdnHCl), calculated from changes of protein

fluorescence kmax (.) and molar ellipticity at 225 nm (E), as a function of

denaturant concentration. Five micromolar pOBP was incubated for 24 h in

PHOSPHATE containing different concentrations of GdnHCl. T= 20 jC.


of AMA, an energy transfer efficiency ET = 0.60 was de-

rived according to Eq. (3). Now, considering the overlap

between the donor emission and the acceptor absorbance

spectra, a critical Foerster distance R0 = 42 A as well as an

average donor–acceptor distance R = 39 A can also be

derived from Eqs. (4) and (5), respectively, on the assump-

tion that the two transition dipoles are free to rotate, that is,

k2 = 2/3.

3.2. Unfolding at equilibrium as a function of GdnHCl

Spectroscopic measurements on denatured pOBP were

performed after 24 h of incubation time with different

GdnHCl concentrations (from 0 to 5 M). Since absorbance

is not, in general, a very sensitive technique to detect protein

unfolding, unless the second derivative technique is used

[21], unfolding of pOBP induced by GdnHCl was studied

only by fluorescence and CD.

The maximum intensity wavelength of the protein intrin-

sic fluorescence, kmax, turned out to be a very sensitive

unfolding parameter, characterised by a large change (from

about 346 nm in the native state to about 358 nm in the

completely unfolded state). Changes of kmax as a function of

denaturant concentration are shown in Fig. 5A at three

different protein concentrations (1, 5 and 50 AM), together

with the best fit curves obtained using Eq. (7). These results

clearly show no protein concentration dependence in the

investigated range, thus ruling out the involvement of

oligomeric forms of pOBP during unfolding [22]. Change

of total fluorescence intensity was a less sensitive parameter

than kmax (data not shown).

A very similar unfolding behaviour was also monitored

by far-UV-CD. In fact, the normalised unfolding data,

obtained from ellipticity at 225 nm and fluorescence kmax

(Fig. 5B), were best fitted by very similar unfolding param-

eters (Table 1).

Protein unfolding was also monitored by SEC, to see if

any stable oligomeric intermediate formed in the denatur-

ation process. The apparent molecular weight Mapp of the

protein remained approximately constant near that of the

monomer (18 kDa) in the range 0.5–2 M GdnHCl. An

appreciable increase of Mapp was instead observed below

0.5 M GdnHCl, with a value close to 1.6 times that of the

monomer in PHOSPHATE alone (Fig. 6). A similar result

was already observed with this protein [23] and attributed,

as also stated by the resin’s manufacturer, to changes in the

retention time of acidic proteins, occurring at ionic strength

lower than 0.15 M.

No protein concentration effect was detected on the

elution volume of pOBP, when chromatographic runs were

performed with two different protein concentrations (5 and

50 AM, data not shown).

AMA and DHM, because of their good affinities, were

chosen to investigate the extent of protection that a bound

Fig. 6. Chromatographic behaviour of pOBP in the presence of GdnHCl. Samples of 10 AM pOBP were incubated for 24 h in PHOSPHATE at several

concentrations of GdnHCl (from 0 to 2 M) before chromatography, then eluted with the corresponding buffer. Inset: The apparent molecular weight of pOBP

was derived from calibration curves, obtained with standard proteins (Pharmacia, low MW calibration kit) under similar conditions, E: 0.2 M, o: 0.5 M, 5:

1 M and n: 2 M GdnHCl in PHOSPHATE.


ligand could give against denaturation. Fig. 7 shows the

unfolding curve of 5 AM pOBP in the presence of 30 AMAMA, as obtained by monitoring the change of the protein

fluorescence kmax (kex = 295 nm). Very similar data were

obtained by following the fluorescence intensity change of

the ligand itself at 480 nm (kex = 380 nm) (data not shown)

and the relative parameters are reported in Table 1. Only a

modest increase of C1/2 (0.2 M) was observed in the presence

Fig. 7. Effect of AMA and DHM on pOBP unfolding. Denaturation transition curve

AM AMA (E) or 20 AM DHM (5), as a function of GdnHCl concentration. The

of AMA. A more considerable protective effect against

denaturation by GdnHCl was instead observed in the pres-

ence of 20 AM DHM (C1/2c 3 M and mc 3), probably due

to a tighter binding of this ligand (Kd = 0.7 AM [24]), as

shown in Fig. 7 and reported in Table 1.

The change of the single Trp accessibility to solvent was

assessed at several denaturant concentrations using the

technique of fluorescence quenching by acrylamide. As

s of 5 AM pOBP in PHOSPHATE, alone (.) and in the presence of either 30plotted spectroscopic parameter is protein fluorescence kmax (kex = 295 nm).

M. Parisi et al. / Biochimica et Biophy122

expected, quenching of the protein fluorescence increases

with denaturant concentration, as deduced by the different

slopes of the Stern–Volmer plots (Fig. 8A). To rule out, this

was simply due to changes in solvent composition, the same

measurements were also performed on N-acetyl-tryptophan-

amide (NATA), a model compound with the Trp residue fully

exposed to solvent. The quenching rate constants of pOBP

were calculated from kq =KSV/s, with the Stern–Volmer

quenching constants KSV derived from the slopes of the

straight lines shown in Fig. 8A and the average lifetimes

calculated from s=(a1s12 + a2s2

2)/(a1s1 + a2s2) [18]. The ratio

of the quenching rate constants of pOBP to those of NATA,

as a function of denaturant concentration, is sigmoidal and

describes the unfolding process pretty well (Fig. 8B). The

protein fluorescence kmax in the presence of 0.5 M acrylam-

ide also shows a similar pattern as a function of GdnHCl

(Fig. 8B). However, both curves appear to be somewhat

shifted to higher denaturant concentration with respect to that

obtained in the absence of acrylamide, thus pointing out a

protective effect by the quencher.

Fig. 8. Quenching of pOBP fluorescence by acrylamide. (A) Stern–Volmer

plots of pOBP as a function of acrylamide, at several GdnHCl

concentrations from 0 to 4 M. The plot with the highest slope (*) refers

to NATA in PHOSPHATE and is almost invariant with GdnHCl. (B) Ratio

of the quenching rate constants of pOBP to NATA (o) and protein

fluorescence kmax at 0 M (.) and 0.5 M (n) acrylamide, as a function of

GdnHCl concentration. Small aliquots of 2 M acrylamide were added to

5 AM pOBP in PHOSPHATE at 20 jC.

3.3. Refolding at equilibrium from GdnHCl

The protein, previously denatured in 4 M GdnHCl for

24 h, was dialysed against PHOSPHATE for 2 days. The

spectroscopic and functional properties of the protein were

then measured. Though the recovered fluorescence and CD

spectra were very similar to those of the native protein, the

binding properties were indeed different to a large extent

(Fig. 3). This result was likely to be due to aggregation, as

deduced by a considerable LS intensity of the protein

solution (data not shown).

Actually, to understand if aggregation was prompted by

refolding or by the dialysis procedure itself, native pOBP

was also dialysed under a variety of experimental condi-

tions: (a) buffer was changed from PHOSPHATE to TRIS;

(b) 1 mM EDTAwas added to TRIS; (c) 200 mM NaCl was

added to TRIS; (d) 20 AM DHM was added to TRIS; (e)

dialysis membrane was changed, from 6–8 kDa (Spectrum)

to 14 kDa (Dolchimica Scientific Glassware) MWCO.

Protein concentration and volume were kept constant

(2 AM in 1 ml against 100 ml external buffer and dialysis

was prolonged for 24 h at 4 jC, with no change of buffer).

In all these cases, an appreciable change of LS intensity was

observed with respect to that under native conditions.

Since all these different refolding experiments by dialysis

gave invariably rise to aggregation, refolding was attempted

by two other different ways: (a) 1:10 dilution with buffer

and (b) ultrafiltration. In both cases, the recovered spectro-

scopic and functional properties of the renatured protein

were practically indistinguishable from those of the native

protein and no aggregation was detected by LS (Figs. 1–3).

sica Acta 1652 (2003) 115–125

4. Discussion

4.1. Analysis of pOBP properties

The fluorescence spectrum of native pOBP has its emis-

sion maximum near 346 nm, suggesting that its unique Trp-

16 residue is partially buried. This conclusion is confirmed

by the longer emission wavelength of NATA (360 nm, not

shown) and the crystallographic structure of the protein (Ref.

[8] and Fig. 4B). The longer emission wavelengths of pOBP

and NATA, observed by us, in comparison to those observed

by others [20], result from the fact that our spectra are

uncorrected for the instrumental response function.

Precious information on the secondary and tertiary struc-

ture of pOBP in solution was obtained from its far-UV and

near-UV-CD spectra, respectively. The secondary structure

content predicted by LINCOMB, a program based on the

method of Perczel et al. [25], is in good agreement with X-

ray data, where more than 60% is formed by h-structures,including h-turns, 18% by a-helices and the rest by random

coil or, more likely, irregular structure (Fig. 2A and Table

2). These results are not in complete agreement with those

published by others using CD and FT-IR [20,26], but are

Table 2

Secondary structure content of pOBP, as derived from far-UV-CDa and

X-rayb data

Secondary structure CD analysis (%) X-ray structure (%)

Alpha 18 15

Beta 70 72

Sheets 38 48

Turns 32 24

Random coil 12 13

a This work.b From Ref. [8].

Table 3

Dynamic quenching of pOBP by acrylamide

GdnHCl KSV (M)a s(ns)b kq (M ns� 1)c

pOBP NATA pOBP NATA pOBP NATA

0 0.89 11.5 2.10 2.62 0.42 4.39

1 1.12 15.3 2.00 2.46 0.56 6.24

2 1.12 12.76 2.11 2.45 0.53 5.21

2.5 2.28 12.52 2.64 2.46 0.86 5.11

3 4.13 12.5 2.78 2.45 1.49 5.10

4 5.73 11.5 2.75 2.44 2.08 4.69

a Stern–Volmer dynamic quenching constant.b Average fluorescence lifetime: s=(a1s1

2 + a2s22)/(a1s1 + a2s2) (from Ref.

[18]).c Dynamic quenching rate constant: kq =KSV/s.


similar to those in the crystal. Differences with FT-IR data

can be attributed to the different algorithms employed. The

peculiar shape of the near-UV-CD spectrum of pOBP (Fig.

2B) suggests that the aromatic residues, among which Trp-

16, are confined into an asymmetric rigid environment,

because of the complete loss of these structural features

upon protein denaturation.

By non-linear least-square fitting of binding data, a value

close to 1 AM was obtained for the dissociation constant of

AMA with native pOBP (Fig. 3), confirming previous

results [20]. pOBP renatured by dilution shows a very

similar binding curve, strongly suggesting a complete re-

covery of the functional properties. In contrast, renatured by

dialysis pOBP displays a much smaller binding capacity,

probably due to the presence of residual aggregates even

after filtration (Fig. 3).

When the protein fluorescence spectrum in the absence

of AMA is compared to that in the presence of this ligand,

energy transfer is detected with an efficiency of about 0.6

derived. Given a Foerster distance R0 of 42 A, calculated

assuming free rotation of the two transition dipoles, this ET

value should correspond to an average donor–acceptor

distance R of about 38 A. However, this is inconsistent

with X-ray data, if AMA is assumed to bind to pOBP as it

does into the bOBP calyx (Ref. [13] and Fig. 4B), since a

value of about 12 A for R can be easily measured.

Binding of AMA to pOBP was simulated using two pdb

files: one (1E00) relative to pOBP complexed with 2,6-

dimethyl-7-octen-2-ol and the other (1HN2) relative to

bOBP complexed with AMA. By overlapping the structure

of each protein monomer, AMA was positioned into pOBP

barrel as it is into bOBP barrel. This is shown in Fig. 4B,

where only AMA and Trp-16 residue are explicitly shown.

The very large discrepancy between expected and experi-

mental ET values can only be reconciled assuming that the

two transition dipoles are on average nearly perpendicular,

implying not only that AMA is rigidly bound inside the

cavity, but also that Trp-16 is hindered in its rotation, as also

suggested by the near-UV-CD spectrum of the protein.

4.2. Analysis of unfolding

The fluorescence spectrum of pOBP is progressively red-

shifted and more intense as denaturant concentration is

increased, showing that the single Trp protein residue is more

quenched in the native than in the fully denatured protein.

The fact that unfolding turned out to be protein concen-

tration independent clearly indicates that only monomers are

involved in the process. This result is consistent with what is

already known about the quaternary state of the protein in

solution in the absence of denaturant [8,23,27], as well as in

the crystal [8], but in disagreement with results by others,

who interpreted their SEC data at neutral pH as due to the

presence of pOBP dimers [26].

The ligand AMA exhibits only a small protection against

unfolding (C1/2 increases of about 0.2 M), though GdnHCl

induces a little loss of affinity of pOBP toward AMA (from

0.8 AM in PHOSPHATE to 1.4 AM at 1.5 M GdnHCl).

Similar results were also obtained from the unfolding curve

derived from AMA fluorescence data (Fig. 7 and Table 1).

The other ligand (DHM) was instead found to consider-

ably stabilise pOBP against denaturation by GdnHCl (C1/2

shifts from 2.5 to 3 M, Fig. 7), as expected from its higher

affinity to pOBP [24].

Quenching of protein fluorescence by acrylamide, or

other quenchers, reports on the degree of accessibility of

protein Trp residues to the solvent. The single Trp-16 residue

of pOBP results partially accessible to solvent, because it is

less quenched by acrylamide than NATA, a reference com-

pound with its indole moiety fully accessible to solvent. In

fact, the Stern–Volmer quenching rate constant KSV for

pOBP increases with denaturant concentration, as expected

for a protein undergoing denaturation, approaching that for

NATA, which is almost unaffected by denaturant, at high

GdnHCl (Fig. 8A). Since both pOBP and NATA fluores-

cence decays are best fitted by two exponentials, the average

lifetime was used to derive the quenching rate constant kq, as

described under Materials and methods and reported in Table

3. The quenching data can also be exploited to obtain another

unfolding plot (Fig. 8B); however, the C1/2 value thus

derived is significantly higher than that in the absence of

quencher (2.9 and 2.4 M, respectively, Table 1). A similar

value (C1/2 = 3.1 M) was also obtained when the protein

fluorescence kmax, measured in the presence of 0.5 M

acrylamide, was plotted versus denaturant concentration,


pointing out a significant protective effect by the quencher

against denaturation. It seems plausible to attribute this effect

to a weak binding of the quencher to pOBP, rather than an

indirect effect of the quencher on GdnHCl, since it saturates

near 0.5 M quencher (data not shown). Finally, it should be

mentioned that our results do not agree with those published

by others, who did not observe quenching by acrylamide

[20]. This may be due to differences in the purification

procedure, as suggested by Burova et al. [28], who found

different thermal denaturation profiles using different protein

preparations.

4.3. Analysis of refolding

pOBP, denatured in PHOSPHATE containing 4 M

GdnHCl, was found to completely renature, in terms of

regain of structural and functional properties, by simple

dilution with PHOSPHATE. However, when renaturation

was attempted by dialysis, aggregation was favoured under

a variety of conditions (different buffers, presence of EDTA,

high ionic strength, presence of the ligand DHM). Even

native pOBP itself was found to aggregate when dialysed

against PHOSPHATE. The most likely conclusion is that

aggregation was probably induced by hydrophobic contacts

between the protein and dialysis tube, whatever the chem-

ical composition of the membrane tube and the buffer.

In contrast, refolding by either dilution or centrifugal

ultrafiltration was rapid and complete and no aggregation

was observed, allowing us to obtain a refolding equilibrium

transition comparable to that for unfolding.

The unfolding/refolding equilibrium transition of pOBP

can be described as a simple two-state transition, because no

stable intermediate was found at any GdnHCl concentration.

This result can be compared to that obtained with bOBP

[14]. In both cases, unfolding occurs in a two-state fashion,

despite the fact that bOBP is a dimer stabilised by domain

swapping. Renaturation of bOBP presented hysteresis at

medium-long times, which allowed us to detect and char-

acterise a fully active monomeric intermediate by SEC

measurements as a function of time. The relative stability

of this intermediate was attributed to the specific but slow

dimerisation step due to domain swapping, presumably

involving an important conformational change of the mono-

mer from a closed to an open form present in bOBP dimer.

On the other hand, a comparison between the folding

behaviour of these two OBPs and other well-studied mono-

meric lipocalins, such as retinol binding protein (RBP) [29]

and BLG [30], is difficult, because many studies were

performed under a variety of conditions, such as different

pH, temperature and denaturant, and techniques. It is

therefore not surprising that different results were obtained

by different investigators even on the same protein. It is,

however, hazardous to compare the folding behaviour of

different proteins, even when they share the same overall

tertiary structure, such as lipocalins. If we limit ourselves to

consider the effect of GdnHCl on RBP and BLG at neutral

pH and equilibrium, folding appears in most cases to be

adequately described as a two-state process, without stable

intermediates [29,30]. This conclusion, if correct, probably

points out the importance of topology, in this case, the h-barrel topology, as a strong driving force in the folding of

these proteins.

Acknowledgements

The authors thank the Italian Ministry for University and

Research (MIUR) and the National Institute for the Physics

of Matter (INFM) for their financial support to this research,

and MIUR also for a PhD fellowship to M.P.

References

[1] S.E. Jackson, How do small, monomeric proteins fold? Fold. Des. 3

(1998) R81–R91.

[2] R.T. Ellis, F.U. Hartl, Principles of protein folding in the cellular

environment, Curr. Opin. Struct. Biol. 9 (1999) 102–110.

[3] E.I. Shakhnovich, Folding by association, Nat. Struct. Biol. 6 (1999)

99–102.

[4] C.M. Dobson, Protein misfolding, evolution and disease, Trends Bio-

chem. Sci. 24 (1999) 329–332.

[5] O.E. Ptitsyn, Molten globule and protein folding, Adv. Prot. Chem. 47

(1995) 83–229.

[6] D.R. Flower, The lipocalin protein family: structure and function,

Biochem. J. 318 (1996) 1–14.

[7] M. Tegoni, R. Ramoni, E. Bignetti, S. Spinelli, C. Cambillau, Domain

swapping creates a third putative combining site in bovine odorant

binding protein dimer, Nat. Struct. Biol. 13 (1996) 863–867.

[8] S. Spinelli, R. Ramoni, S. Grolli, J. Bonicel, C. Cambillau, M. Tego-

ni, The structure of the monomeric porcine odorant binding protein

sheds light on the domain swapping mechanism, Biochemistry 37

(1998) 7913–7918.

[9] A. Miyawaki, F. Matsushita, Y. Ryo, K. Mikoshiba, Possible phero-

mone-carrier function of two lipocalin proteins in the vomeronasal

organ, EMBO J. 13 (1994) 5835–5842.

[10] M. Tegoni, P. Pelosi, F. Vincent, S. Spinelli, V. Campanacci, S. Grolli,

R. Ramoni, C. Cambillau, Mammalian odorant binding proteins,

Biochim. Biophys. Acta 1482 (2000) 229–240.

[11] V. Matarazzo, N. Zsurger, J.C. Guillemot, O. Clot-Faybesse, J.M. Bot-

to, C. Dal Farra, M. Crowe, J. Demaille, J.P. Vincent, J. Macella, C.

Ronin, Porcine odorant-binding protein selectively binds to a human

olfactory receptor, Chem. Senses 8 (2002) 691–701.

[12] F. Vincent, S. Spinelli, R. Ramoni, S. Grolli, P. Pelosi, C. Cambillau,

M. Teloni, Complexes of porcine odorant binding protein with odor-

ant molecules belonging to different chemical classes, J. Mol. Biol.

300 (2000) 127–139.

[13] R. Ramoni, F. Vincent, S. Grolli, V. Conti, C. Malosse, F.D. Boyer, P.

Nagnan Le Meillour, S. Spinelli, C. Cambillau, M. Tegoni, The insect

attractant 1-octen-3-ol is the natural ligand of bovine odorant-binding

protein, J. Biol. Chem. 276 (2001) 7150–7155.

[14] A. Mazzini, A. Maia, M. Parisi, R.T. Sorbi, R. Ramoni, S. Grolli, R.

Favilla, Reversible unfolding of bovine odorant binding protein in-

duced by guanidinium hydrochloride at neutral pH, Biochim. Bio-

phys. Acta 1599 (2002) 90–101.

[15] C.N. Pace, F. Vajdos, L. Fee, G. Grimsley, T. Gray, How to measure

and predict the molar absorption coefficient of a protein, Protein Sci.

4 (1995) 2411–2423.

[16] R. Favilla, M. Goldoni, A. Mazzini, P. Di Muro, B. Salvato, M.

Beltramini, Guanidinium chloride induced unfolding of a hemocyanin


subunit from Carcinus aestuarii: I. Apo form, Biochim. Biophys.

Acta 1597 (2002) 42–50.

[17] L. Stryer, Fluorescence energy transfer as a spectroscopic ruler, Annu.

Rev. Biochem. 47 (1978) 819–846.

[18] J.R. Lakowitz, Principles of Fluorescence Spectroscopy, 2nd ed.,

Chap. 16, Kluwer Academic/Plenum Publishers, 1999.

[19] R.W. Woody, A.K. Bunker, in: G.D. Fasman (Ed.), Circular Dichro-

ism and the Conformational Analysis of Biomolecules, Plenum, NY,

1996, Chap. 4.

[20] S. Paolini, F. Tanfani, C. Fini, E. Bertoli, P. Pelosi, Porcine odorant-

binding protein: structural stability and ligand affinities measured by

Fourier-transform infrared spectroscopy and fluorescence spectro-

scopy, Biochim. Biophys. Acta 1431 (1999) 179–188.

[21] R. Ragone, G. Colonna, C. Balestrieri, R. Servillo, G. Irace, Deter-

mination of tyrosine exposure in proteins by second-derivative spec-

troscopy, Biochemistry 23 (1984) 1871–1875.

[22] R. Ragone, How the protein concentration affects unfolding curves of

oligomers, Biopolymers 53 (2000) 221–225.

[23] R. Ramoni, F. Vincent, A.E. Ashcroft, P. Accornero, S. Grolli, C.

Valencia, M. Tegoni, C. Cambillau, Control of domain swapping in

bovine odorant-binding protein, Biochem. J. 365 (2002) 739–748.

[24] J. Pevsner, V. Hou, A.M. Snowman, S.H. Snyder, Odorant-binding

protein. Characterization of ligand binding, J. Biol. Chem. 265 (1990)

6118–6125.

[25] A. Perczel, K. Park, G.D. Fasman, Analysis of the circular dichroism

spectrum of proteins using the convex constraint algorithm: a practi-

cal guide, Anal. Biochem. 203 (1992) 83–93.

[26] T.V. Burova, Y. Choiset, C.K. Jankowski, T. Haertle, Conformational

stability and binding properties of porcine odorant binding protein,

Biochemistry 38 (1999) 15043–15051.

[27] M. Dal Monte, I. Andreini, R. Revoltella, P. Pelosi, Purification and

characterization of two odorant-binding proteins from nasal tissue of

rabbit and pig, Comp. Biochem. Physiol., B 99 (1991) 445–451.

[28] T.V. Burova, H. Rabesona, Y. Choiset, C.K. Yankowski, L. Sawyer, T.

Haertle, Why has porcine VEG protein unusually high stability and

suppressed binding ability? Biochim. Biophys. Acta 1478 (2000)

267–279.

[29] L.L. Burns, P.M. D’Alessio, I.J. Ropson, Folding mechanism of three

structurally similar beta-sheet proteins, Proteins 33 (1998) 107–118.

[30] L. D’Alfonso, M. Collini, G. Baldini, Does h-lactoglobulin dena-

turation occur via an intermediate state? Biochemistry 41 (2002)

326–333.

Unfolding and refolding of porcine odorant binding protein in guanidinium hydrochloride: equilibrium...

Documents

Transcript of Unfolding and refolding of porcine odorant binding protein in guanidinium hydrochloride: equilibrium...