Crystal structures of bovine odorant-binding protein in complexwith odorant molecules Florence Vincent1, Roberto Ramoni2, Silvia Spinelli1, Stefano Grolli2, Mariella Tegoni1 and Christia
Trang 1Crystal structures of bovine odorant-binding protein in complex
with odorant molecules
Florence Vincent1, Roberto Ramoni2, Silvia Spinelli1, Stefano Grolli2, Mariella Tegoni1
and Christian Cambillau1
1
Architecture et Fonction des Macromole´cules Biologiques, UMR 6098, CNRS, Marseille, France;2Dipartimento di Produzioni Animali, Biotecnologie Veterinarie, Qualita` e Sicurezza degli Alimenti, Universita` di Parma, Parma, Italy
The structure of bovine odorant-binding protein (bOBP)
revealed a striking feature of a dimer formed by domain
swapping
2 [Tegoni, M., Ramoni, R., Bignetti, E., Spinelli, S
& Cambillau, C (1996) Nat Struct Biol 3, 863–867;
Bian-chet, M.A., Bains, G., Pelosi, P., Pevsner, J., Snyder, S.H.,
Monaco, H.L & Amzel, L.M (1996) Nat Struct Biol 3,
934–939] and the presence of a naturally occuring ligand
[Ramoni, R., Vincent, F., Grolli, S., Conti, V., Malosse, C.,
Boyer, F.D., Nagnan-Le Meillour, P., Spinelli, S.,
Cambil-lau, C & Tegoni, M (2001) J Biol Chem 276, 7150–7155]
These features led us to investigate the binding of odorant
molecules with bOBP in solution and in the crystal The
behavior of odorant molecules in bOBP resembles that
observed with porcine OBP (pOBP), although the latter is
monomeric and devoid of ligand when purified The odorant
molecules presented Kdvalues with bOBP in the micromolar range Most of the X-ray structures revealed that odorant molecules interact with a common set of residues forming the cavity wall and do not exhibit specific interactions Depending on the ligand and on the monomer (A or B), a single residue – Phe89 – presents alternate conformations and might control cross-talking between the subunits Crystal data on both pOBP and bOBP, in contrast with binding and spectroscopic studies on rat OBP in solution, reveal an absence of significant conformational changes involving protein loops or backbone Thus, the role of OBP
in signal triggering remains unresolved
3
Keywords: crystal structure; domain swapping; odorant-binding protein; olfaction
Odorant-binding proteins (OBPs), first discovered in the
nasal mucus and epithelium of mammals at millimolar
concentrations [
4 1,2], were identified, by their sequence, as
lipocalins [3], a family of proteins generally involved in the
transport of hydrophobic ligands [4] The hypothesis of
their involvement in the olfactory process, perception and
transduction of the signal, was derived from their
localization and their ability to bind
2-iso-butyl-3-metoxy-pyrazine (IBMP), known as the odorant with the lowest
detection threshold in humans [5] Different roles for OBPs
have previously been proposed [6,7], such as (a) carriers of
odorants from the air to the olfactory receptors (ORs)
through the aqueous barrier of the mucus [11,12]; (b)
scavengers of odorants from the OR after transduction of
the olfactory signal [8,9] and/or of odorants present at a
high concentration in order to avoid saturation of the OR
[10,11]; (c) protectors of the nasal mucosa, which is
exposed to airflow and oxidative injuries, by binding
cytotoxic and genotoxic molecules, such as alkylic alde-hydes [9,12,13]; or (d) during transduction, to permit recognition of the OBP–odorant complex by the receptor,
as in bacterial chemotaxis [14,15]
spectrum of activity and a relatively weak affinity for odorants have been found in mammalian OBPs [9,12,16– 18], and clear experimental evidence of the role of OBPs in olfaction and odorant perception has not yet been produced Peculiar in the case of bovine OBP (bOBP) is the anti-cooperative binding resulting in the stoichiometry
of a single molecule of IBMP per dimer of OBP, repeatedly reported in the literature [1,8,9]
We have reported the first 3D structure of bOBP at 2.0 A˚ [19] In fact, the prototype of OBPs was demonstrated to be
a special case among lipocalins, as it is a dimer with a swapped helix, possibly providing interdependent properties
to the two subunits [19] As previously found in the case of retinol-binding protein [20] and Major Urinary Protein
a natural ligand that co-purified with the protein was observed in the b-barrel cavity of bOBP [19,22] The further unambiguous identification of this natural ligand as 1-octen-3-ol (OCT), an insect attractant produced by bovine rumination, suggested a role of bOBP in the ecological relationships between bovine and several insect species [23] Porcine OBP (pOBP) [24] was revealed to be a monomer and to have a cavity devoid of any ligand, which made it a good candidate for studying the interaction with a broad range of ligands of bOBP [9,12,16,18] or pOBP [16,18] The structure of different complexes of pOBP showed the presence of the odorants, in a stoichiometric molar ratio,
Correspondence to M Tegoni or C Cambillau, Architecture et
Fonction des Macromole´cules Biologiques, UMR 6098, CNRS, 31
Chemin Joseph Aiguier, 13402 Marseille Cedex 20, France.
Fax: +33 4 9116 4536, Tel.: +33 4 9116 4501,
E-mail: tegoni@afmb.cnrs-mrs.fr or cambillau@afmb.cnrs-mrs.fr
Abbreviations: AMA, 1-amino-anthracene; bOBP, bovine
odorant-binding protein; BZP, benzophenone; DHM, dihydromyrcenol;
IBMP or pyrazine, 2-isobutyl-3-metoxypyrazine; OCT, 1-octen-3-ol;
OR, odorant receptor; pOBP, porcine OBP; UND, undecanal.
(Received 23 June 2004, accepted 30 July 2004)
Trang 2inside the cavity formed by the b-barrel The orientation of
the ligands inside the cavity appeared to be opportunistic
with no specific target patches for aromatic or charged
groups and no correlation between the number of contacts
and the affinity measured in solution In particular, no
special characteristic could be ascribed to the good binders,
and upon interaction with all odorants, all residue
side-chains lining the cavity kept the conformation observed in
the native protein
While these results were not in favor of a specific carrier
role for OBPs [13], other studies performed in solution,
although sometimes contradictory, pointed to a contrasting
situation
7 Fluorescence studies in solution with three rat
OBPs and 49 different ligands revealed some preference of
each OBP to certain classes of chemical compounds,
suggesting a specific filter role of OBPs [17] Recently,
biophysical studies on a subclass of dimeric rat OBP
(OBP-1F) indicated that one ligand binds per monomer and that
subtle changes in the side-chains or backbone positions were
induced upon binding [25] Another group could express a
human OR in mammalian cells and demonstrate
high-affinity binding with empty porcine OBP, although binding
was further localized in several tissues besides nasal mucosa
Another class of lipocalins – tear lipocalin-1 – was shown
to bind with high affinity to a new class of receptor not
belonging to the GPCR family as ORs, and to be further
internalized [26]
With a view to collecting hard data based on crystal
structures, possibly displaying conformational changes,
we chose bOBP as a good candidate for ligand studies –
the native structure of bOBP showed some flexibility at
Phe89, and its dimer structure seemed favorable for
investigating putative cross-talk between both subunits
We report, in this study, the X-ray structures of
complexes with five ligands determined at a resolution
of 1.7–2.05 A˚ These X-ray studies were completed with
titration of bOBP by the same odorant molecules, in
competition with 1-amino-anthracene (AMA) The
odor-ant molecules were observed inside the two cavities
formed by the b-barrels, together with some residual
natural ligand (OCT) in three of the complexes The
interactions established with the protein were found to be
essentially hydrophobic, although few hydrogen bonds
were observed with ligands bearing polar groups The
internal cavity exhibits flexibility because Phe89 displays
two discrete alternate positions, as in the native protein
These results confirm previous studies on pOBP and
define the role of bOBP as able to recognize efficiently
molecules pertaining to different chemical classes The
absence of significant conformational change at the
protein surface is not in favor of an OR triggering role
for bOBP, thus leaving this question still open for new
experiments
Experimental procedures
Protein and odorants
Bovine and porcine OBP were purified from frozen
samples of nasal mucosa, as reported previously [23,24]
The purified proteins showed a single band in SDS/PAGE
AMA was purchased from Fluka The odorants
benzo-phenone (BZP), dihydromyrcenol or 2,6-dimethyl-7-octen-2-ol (DHM) and undecanal (UND) were purchased from Fluka, and IBMP and OCT were from Aldrich Stock solutions (2 mM) of the odorants were prepared in ethanol, then further diluted in 20 mMTris/HCl, pH 7.8, 0.5% (v/v) ethanol (TE buffer) In the binding studies, the odorant solutions in TE buffer were prepared just before each experiment and used only once
Fluorescence-binding assay The fluorescence-binding assay of AMA with bOBP, and the competition between AMA and odorants, were carried out according to a method published previously [27], with minor modifications [23] The influence of the concentration
of ethanol on the chasing process of AMA was tested and found to be negligible up to 1% (v/v), a result in contrast with the behavior of rat OBP reported by Briand et al [28]
In brief, the formation of AMA–OBP complexes was determined by following an increase of the fluorescence emission at 480 nm, upon excitation at 380 nm, using a PerkinElmer LS 50 luminescence spectrometer The disso-ciation constants of the AMA–OBP complexes were determined from the titration curves using the nonlinear fitting facility, SIGMA PLOT 5.0 (Cambridge Soft Corp., Cambridge, MA, USA) Saturation levels were determined from calibration curves obtained by incubating increasing concentrations of AMA (0.076–5 lM) with a fixed, satur-ating concentration of OBP (1 lM)
In the competition curves, the OBP samples (0.5 lMand
1 lM, respectively, for bOBP and pOBP) were incubated with a fixed amount of AMA (3 lM) and increasing concentrations of odorants (0.39–50 lM) The chasing of AMA bound to OBP was followed as a decay of the emission of the fluorescence intensity at 480 nm, and the curves, for each odorant, were analyzed using the SIGMA PLOT 5.0 software for the determination of the apparent dissociation constants (Kdiss app.) Kdiss true values were calculated from apparent Kdiss, using the following formula:
Kdisstrue¼ kdissapp: 1=½1 þ ð1=KdissAMA ½AMAÞ which takes into account the Kdiss for AMA and the concentration of AMA
Crystallization, data collection and refinement
of bovine OBP Except for the bOBP–UND complex, bOBP crystals were obtained by microdialysis of a 10 mgÆmL)1protein solu-tion against 28–32% (v/v) ethanol, in 50 mM citrate,
pH 5.4, at 4C Crystals of the bOBP–UND complex were obtained by vapour diffusion in the presence of 18–40% (v/v) ethanol, 20 mMcitrate, pH 4.2, 0.05% (v/v) poly(vinylpyrrolidone)
group P21, with cell dimensions a¼ 55.9 A˚, b ¼ 65.5 A˚,
c¼ 42.7 A˚ and b ¼ 98.8, and contain one homodimer in the asymmetric unit
The bOBP–odorant complexes were obtained by soaking the crystals, overnight at 4C, in synthetic crystallization solutions containing 2 mM odorant For the chasing of AMA by IBMP, the same procedure as in the solution fluorescence study was followed: a crystal of bOBP was first
Trang 3soaked for 1 h at 4C in a reservoir solution containing
2 mMAMA and then transferred to a reservoir containing
1 mMIBMP and soaked for 4 h
Data collection, scaling and reduction were performed as
with native bOBP (Table 1) Data on microdialysis crystals
were collected on a MAR Research 345 image plate (Mar
Research, Norderstedt, Germany) placed on a Rigaku
RU2000 rotating anode with Osmic mirrors (Rigaku Corp.,
Tokyo, Japan)
temperature, as cryocooling always yielded data sets that
could not be used in refinement Only the data set on the
crystal of the bOBP–UND complex was collected on
flash-frozen crystals [12.5% (v/v) methylpentanediol]
DW32 (LURE) These data, in contrast with the other data
sets, could be used and refined Indexation and integration
were performed usingDENZO[29], data scaling withSCALA
[30], and data reduction withTRUNCATE[30] (Table 1)
The refinement of bOBP complexes made use of the
previously determined bOBP structure as a starting model
(1OBP) The atomic structure of the odorants was built
using the programTURBO[31], while topology and force field
data were defined in the suitable files of CNS [30] by the
automated procedure XDICT(G L Kleywegt, Biomedical
Centre, Uppsala University, Sweden; http://alpha2.bmc
uu.se/hicup)
12 The models were then refined usingCNS[32]
for all bOBP–odorant complexes, except for UND which
was refined usingREFMAC[33] Cycles of refinement were
alternated with manual re-fitting into sigmaA-weighted
electron density maps with the graphic program [31] The
final models have Rworkand Rfreevalues of 18.8–22.0% and
22.3–23.9%, respectively (Table 1) The final models have
good geometries according toPROCHECK
coordi-nates have been deposited with the PDB at RCSB (Table 1)
Results
Binding of AMA and odorants in solution The dissociation constants of the AMA–OBP complexes for the bovine and porcine OBP isoforms were, respectively, 1.0 and 1.5 lM and the corresponding maximum saturation levels were 1.7 and 0.85 mol of AMA per mol of OBP These results were in agreement with values already reported for these proteins when assayed with fluores-cence-binding tests using AMA [23,27,35]
X-ray crystallography experiments and competitive bind-ing tests in solution indicated that AMA binds in the internal cavities of bOBP and could be completely displaced
by the natural ligand, OCT [3] Therefore, competitive displacement of AMA by different odorants should titrate the same internal binding site of bOBP
The procedure used with OCT for chasing AMA was also used with the five other odorant molecules (Fig 1) With all ligands but one (IBMP), the stoichiometry was found to be close to two molecules of odorant per bOBP dimer With IBMP, the stoichiometry was found to be close to one molecule of odorant per dimer, indicating that one of the two AMA molecules could not be chased by IBMP (see the structure section below) The calculated true Kdissvalues were found to range from 0.3 to 3.3 lM(Fig 1, Table 2) The values obtained with pOBP [13] are also reported for comparison (Table 2)
Table 1 Data collection and refinement statistics AMA, 1-amino-anthracene; BZP, benzophenone; DHM, dihydromyrcenol; IBMP, 2-isobutyl-3-metoxypyrazine; UND, undecanal.
Ligands (upper row) and PDB entry (lower row) UND
1GT4
DHM 1GT3
AMA 1HN2
AMA/IBMP 1GT1
BZP 1GT5 Data collection
Resolution limits (A˚) 13–2.06 35–1.7 17–1.8 18.6–1.7 20–2.05
Rsym (all last/shell) 7.6/33 5.2/11.6 5.2/30.1 6.5/26.8 6/24.2 Refinement
Resolution limits 9.5–2.1 10–1.8 0–1.8 10–1.71 18–2.08 Number of reflections 16 687 26 271 25 530 28 652 15 517
R factor /R free (%) 22.1/25.6 20.0/23.9 20.2/22.3 20.3/22.3 18.8/22.8
B factors (molecules A & B)
Main chain 27.06/18.86 32.69/27.65 35.11/28.78 36.58/31.52 36.26/29.37 Side-chains
Ligands 68.63/59.9 55.08/48.87 49.07/49.77 50.55/46.92 34.07/34.46
rmsd
Bonds (A˚)/angles 0.11/1.64 0.009/1.4 0.013/1.5 0.013/1.4 0.007/1.3
Trang 4Overall structures of the complexes
As described previously, bOBP is a 2· 159 residue dimer at
neutral or basic pH, and a monomer at pH values below 4.5
[36] The 2.0 A˚ resolution structure of bOBP has been
reported previously [19,22] and its natural ligand has been
identified [23] Briefly, each monomer is composed of a
lipocalin-type nine-stranded b-barrel comprising residues
15–121 (strands 1–8) and residues 145–149 (strand 9) from
the other monomer (Fig 2) From residue 123 onwards, the
topology diverges from the consensus lipocalin fold: the
b-barrel is connected by an extended stretch of residues
(123–126) to the a-helix protruding out of the b-barrel and
crossing the dimer interface (Fig 2) As a consequence, the a-helix of one monomer is placed close to where the a-helix
of the other monomer would be if bOBP had a classical lipocalin fold, in a peculiar arrangement named domain swapping (Fig 2)
In the present structures, the two bOBP polypeptidic chains are visible from residues 1–159 and 3–157 for molecules A and B, respectively The electron density map
of molecule A is generally better defined than that of molecule B, although its B factors are generally slightly higher (Table 1) When superimposing pairwise the Ca traces of the bOBP dimer for all complexes, the rmsd values range between 0.06 A˚ to 0.30 A˚, values within positional errors at this resolution (1.64–1.3 A˚, in the present study and 1.8 A˚ for the natural complex OCT–bOBP) [23] Superposition of monomers A and B for each complex confirms also that the lack of symmetry, owing to a different helix position relative to the b-barrel, is conserved
Complexes with the odorant compounds The internal cavities An electron density accounting for the presence of a bound ligand is found in each cavity of monomers A and B of bOBP (Fig 3) As for the polypeptide chain, the electron density map of the ligand
is generally better defined in monomer A than in monomer
B The interpretation of the map is more difficult than in pOBP [37], however, because residual binding of the natural ligand (OCT) occurred with some complexes and had to be estimated along the refinement (Table 3) This is the case for
Fig 1 1-Amino-anthracene (AMA) chasing by different odorant molecules The fluorescence of AMA decreases when chased by the odorant molecules; (A) 1-octen-3-ol, (B) undecanal, (C) benzophenone, (D) 2-iso-butyl-3-metoxypyrazine and (E) dihydromyrcenol Each fluorescence point on the y-axis shows the concentration of AMA still bound per monomer of bovine binding protein (bOBP) and porcine odorant-binding protein (pOBP), relative to the initial value, on a scale of 0–1 These values are plotted as a function of the micromolar concentration of total odorant competing for binding (x-axis) The concentrations of bOBP and pOBP were, respectively, 0.5 and 1 l M , while AMA was kept constant at
3 l M (m), bOBP; (d), pOBP Non-linear fit was calculated using the nonlinear facility of SIGMA PLOT 5.0 (Cambridge Soft Corp., Cambridge, MA, USA).
Table 2 Dissociation constants (Kdiss inlM ) of the fluorescent probe or
of odorant molecules for bovine odorant-binding protein (bOBP) AMA,
1-amino-anthracene; BZP, benzophenone; DHM, dihydromyrcenol;
IBMP, 2-isobutyl-3-metoxypyrazine; OCT, 1-octen-3-ol; bOBP,
bovine odorant-binding protein; pOBP, porcine odorant-binding
protein; UND, undecanal.
K diss bOBP K diss pOBP
Trang 5the complexes with AMA, AMA/IBMP and DHM The
complexes with BZP and UND do not include residual
OCT (Table 3) The positions of the complete sets of ligands
clusters, when superimposed, result in a global volume of
12 · 7 · 3 A˚3(Fig 4A) They establish contacts with 17
residues forming the cavity walls, among which 11 are
hydrophobic and six are polar noncharged (Table 4,
Fig 4B) No contacts are established with charged residues
as none point inside the cavity Some residues are
pre-eminently involved in interactions with the ligands
(thresh-old 10 A˚2), such as Phe36, Phe40, Phe54, Phe89, Phe119,
and, above all, Asn103
the various ligands is rather homogeneous, with an average
value of 156 ± 20 A˚2 (Table 4) While electron density
maps of the side-chains are well defined and unique, Phe89
presents two main positions, as reported previously
[19,22,35], i.e in a closed position (Fig 5A) or an open
(Fig 5B) position Phe89 may also be present in a mixture
of alternate open and closed positions, leading to different
cavity volumes
15 As seen in Table 5 and in Fig 5, the volume
of the larger cavity is smaller when Phe89 is closed
(Fig 5A), rather than open (Fig 5A), as the internal
volume is split into alternate large and small cavities
Indeed, the cavity volumes with Phe89 in both open and
closed positions are comparable to those with Phe89 in the
closed conformation
16
OCT, the natural ligand Although the structure of the complex of bOBP with OCT has been reported previously [23], we will briefly recall its binding mode here The nature
of OCT of bOBP was determined by GC/MS The two enantiomers of OCT (ratio 1 : 1) were fitted with the most appropriate conformations in the electron density maps within each barrel, and their occupancies were refined in the CNS The two isomers have very similar orientations, and are quasi superimposed (Fig 3M,N) The aliphatic chains of the two isomers are extended and remain close to each other The hydroxyl groups point to the same direction, although to an area of the cavity walls where no hydrogen bond donor is available
AMA The structure of the complex between bOBP and AMA has already been described [23] and we recall here these results and a more detailed analysis AMA in complex with bOBP occupies the same place as the natural ligand, i.e
in the internal cavity of the b-barrel of each monomer [19,35] In cavity A, the initial electron density map for AMA was very clear and a molecule of AMA was fitted readily (Fig 3A) The AMA molecule accounted for the total electron density, and its occupation was maintained at 100% (Table 3) All residues in the combining site are well defined in the electron density map and do not exhibit alternate conformations The side-chain of Phe89 presents
Fig 2 Bovine odorant-binding protein (bOBP) dimer (secondary structure representation) and superposition of all the ligands (sphere repre-sentation) in both cavities (A) Monomer A (green) and monomer B (blue-grey) (B) The structure is rotated 90 towards the reader with respect to (A); both monomers are rain-bow-colored The figure was produced using
PYMOL [48].
Trang 6only the open conformation among the two observed in the
native protein (Table 5) In contrast, in cavity B the electron
density map around AMA and Phe89 displays complex
features Some residual density appears at several places
when only the AMA molecule is taken into account for the
calculation of the electron density map These bulbs of
positive difference density disappear when 20% of the
natural ligand and 80% of AMA are introduced in the
calculations (Fig 3B, Table 3) Phe89 has two
conforma-tions: the predominant one is the open conformation, found
for Phe89 in cavity A; while the other conformation, with a slimmer density, is the closed one, similar to that found in the native structure with its natural ligand (Table 5) The position of AMA in cavity B is very similar to that observed
in cavity A
BZP The map of the complex with BZP exhibits a clear electron density for the ligand in site A associated with low B-factors (Table 1, Fig 3G) The ligand exhibits a unique conformation, and no trace of the natural ligand is detected
in the electron density map (Table 3) The two phenyl rings
of BZP form a dihedral angle of 40 (Fig 3) The carbonyl group of BZP establishes a canonic hydrogen bond (d¼ 2.8 A˚, angle ¼ 126) with the hydroxyl group of Thr38 The other contacts are exclusively hydrophobic (Table 4) In cavity B, the position of BZP is also well defined and unique (Fig 3H, Table 3) The hydrogen bond with Thr38 is longer than in subunit A (3.4 A˚) and its geometry less optimum Phe89 is observed in both the open and the closed positions in subunit A, and only in the closed position in subunit B (Table 5)
DHM The map of the DHM–bOBP complex exhibits intricate features First, Phe89 displays alternate conforma-tions in both monomers, as in the case of OCT (Table 5) This is not surprising as both OCT and DHM exhibit a
Fig 3 View of different ligands and the
elec-tron density observed in cavities A and B of
bovine odorant-binding protein (bOBP),
respectively (A,B) 1-Octen-3-ol (OCT) (both
enantiomers) co-purified with bOBP (C,D)
1-Amino-anthracene (AMA) and AMA and
OCT (E,F) Benzophenone (BZP) (G,H)
Di-hydromyrcenol (DHM) and OCT (I,J) AMA
plus 2-iso-butyl-3-metoxypyrazine (IBMP),
and AMA plus IBMP plus OCT (K,L)
Undecanal (UND) Electron density maps
were produced using TURBO [31] and
contoured at the 1 r level Ligands are
rep-resented in stick mode; color code: carbon ¼
yellow, oxygen ¼ red, nitrogen ¼ blue.
Table 3 Estimates in percentage occupancy of the ligands and of the
endogenous ligand [1-octen-3-ol (OCT)] in the cavity of bOBP monomers
A and B AMA, 1-amino-anthracene; BZP, benzophenone; DHM,
dihydromyrcenol; IBMP, 2-isobutyl-3-metoxypyrazine; UND,
unde-canal.
Molecule A Molecule B OCT A OCT B
AMA/IBMP 60 IBMP/
40 AMA
50 IBMP/
40 AMA
Trang 7similar elongated shape Furthermore, the electron density
map could not be accounted for using only DHM, but the
introduction of 70% DHM and 30% remaining OCT
almost suppresses residual electron density map (Fig 3I,J) The hydroxyl group of DHM exhibits a hydrogen bond with the Asn103 Nd2 atom (3.0 A˚)
Fig 4 Ribbon representation and molecular surface of cavity A of bovine odorant-binding protein (bOBP) All ligands are shown super-imposed (A) The ligands are represented (sticks, individual colors) inside the slabbed volume of the cavity Left and right views are rotated by 90 (B) Stereo view of the super-position of all ligands (sphere representation, individual colors) in cavity A of bOBP (transparent molecular surface) The side-chains (stick representation, color code as in Fig 3) of the residues in the cavity interacting with the ligands are shown The figure was produced using PYMOL [48].
Table 4 Interaction surfaces (A˚2) of the ligands with residues of cavities in monomers A and B These values were calculated by the surface value of each residue when the cavity is empty from the surface value when the cavity is filled by the ligand (using the program TURBO ) [31] AMA, 1-amino-anthracene; BZP, benzophenone; DHM, dihydromyrcenol; IBMP, 2-isobutyl-3-metoxypyrazine; OCT, 1-octen-3-ol; UND, undecanal.
Trang 8IBMP.A complex with IBMP has been submitted to X-ray
analysis, aiming to understand the peculiar behavior of
IBMP in solution (Introduction and Experimental
pro-cedures) With this aim, the procedure used to obtain the
complex reproduced exactly the fluorescence experiment,
i.e IBMP chasing AMA bound to bOBP (see the
Experimental procedures) In the structure, the electron
density map at both sites differed from that of AMA in the
bOBP–AMA complex, and the size of the density indicates
that some AMA is still bound (Fig 3E,F) IBMP has been
modeled in the electron density map of cavity A and B, and
AMA has been introduced After refinement with CNS,
including occupancy refinement for IBMP and AMA, the
electon density map could be modeled with 40% AMA and
60% IBMP, in cavity A (Fig 3E, Table 3) In cavity B,
some of the natural ligand had also to be introduced,
leading to occupation values of 42% AMA, 42% IBMP
and 16% OCT (Fig 3F, Table 3) Although the occupancy
values may differ in different experiments and should be
interpreted with caution, these results confirm that in the
crystal, as in solution, only one IBMP molecule binds per
bOBP dimer In subunit A, Phe89 is found in the open
conformation, while the closed one is found in subunit B
(Table 5)
UND The UND molecule is better defined in subunit A where all atoms are present in the electron density map (Fig 3K) In subunit B, the last atoms of the alkyl chain (C9-C11) are poorly defined (Fig 3L) In both subunits, the occupancy of UND is 100% (Table 3) UND adopts an extended conformation up to carbon 5 and then bends to form a U-shaped structure (Fig 3K,L) The aldehyde function does not establish any hydrogen bond with the residues of the cavity The complex with UND is the only one where Phe89 does not assume alternate positions in either subunits: Phe89 is in the open position in subunit A and in the closed position in subunit B (Table 5) In relation
to the position of Phe89, the position of the U-shaped part
of UND adopts opposite directions in subunits A and B (Fig 3K,L) Furthermore, the position adopted by UND in subunit A would not be compatible with the closed position
of Phe89 in subunit B
Communication of the internal cavity with the bulk solvent
The cavity of bOBP has no direct access to the solvent, which is also the case for other lipocalins, such as pOBP [24] or aphrodisin [3] A unique side-chain shields the cavity from the solvent, however Combined with the open position of Phe89, the rotation of Tyr83 by 120 (Fig 6Aa) opens a communication path to the internal cavity through which ligands might find access (Fig 6B) This rotation is not hindered by neighboring residues, and should therefore require a small amount of energy This suggests strongly that Tyr83 may be the door of the cavity, and might trigger the access of the natural substrate (Fig 6) Furthermore, no crystal packing contacts involve this part of the bOBP surface, which faces water crystal channels This is indeed the reason why soaking odorant molecules in the crystals was successful It is worthy of mention that in pOBP, Tyr83 is well situated to perform a similar role, and that the corresponding residue (Tyr76) in aphrodisin has already been proposed as being the cavity door [23]
Discussion
The average of the Kdiss values for bOBP is comparable (1.2 lM) to that found for pOBP (1.6 lM) (Table 2) The bOBP natural ligand, OCT, presents a slightly better affinity for bOBP than for pOBP Amazingly, IBMP, taken historically as the reference compound for binding studies with OBPs [1,2], is the poorest ligand of bOBP and an excellent ligand of pOBP; the affinity for bOBP is 10 times smaller than for pOBP There is no evident correlation between the bulkiness, the flexibility or the chemical functions borne by the different compounds and their affinity with either OBP It should be noted that the ratio of the Kdissvalues for bOBP are grouped within a factor of 10, meaning that the differences in energy involved are rather low This range of values is similar to that observed for rat OBP-F1 (Nespoulos) and for rat OBPs 1–3, for each of the proteins analyzed individually [17]
As mentioned above, the bOBP cavity is mostly hydro-phobic Besides BZP, the ligands make little use of the five semipolar residues in the cavity to establish hydrogen
Fig 5 Representation of the molecular surfaces of the cavities in
monomer A Left, bovine odorant-binding protein-1/amino-anthracene
(bOBP–AMA) complex The Phe89 side-chain presents only the open
position (yellow) and the cavity is unique (pink) Right, bovine
odor-ant-binding protein/benzyl-benzoate (bOBP–BZP) complex, with the
Phe89 side-chain in alternate positions: the open position is yellow, the
closed position is orange Note that the closed position of Phe89
sep-arates the cavity (as seen in the left view) into two (light and dark blue).
The figure was produced using GRASP [49] See also Table 5.
Table 5 Position of Phe89 in the cavity of monomer A and B, in each
complex The positions can be open, closed or alternate (alt) The
corresponding volumes (in A˚3) of the cavities are presented AMA,
1-amino-anthracene; BZP, benzophenone; DHM, dihydromyrcenol;
IBMP, 2-isobutyl-3-metoxypyrazine; OCT, 1-octen-3-ol; UND,
undecanal.
OCT AMA AMA/IBMP BZP DHM UND
Phe89
A Alt Open Open Alt Alt Open
B Alt Alt Alt Closed Alt Closed
Volumes of the cavities
A 396 491 504 396 400 443
B 377 378 401 375 377 377
Trang 9bonds Consequently, 80% of the contact surface involves
hydrophobic residues (Table 4) Bulky molecules, like
AMA and BZP, exhibit very close positions in both
subunits However, even in these two clear-cut cases, the
side-chain of Phe89 exhibits different orientations in
subunits A and B Slight differences have also been observed
in the position of the OCT molecules [35] and are observed
here for the molecules of DHM Only the UND molecule
exhibits a totally opposite position in subunits A and B,
clearly associated with the orientation of Phe89 The
position of Phe89 can be somewhat rationalized in terms
of subunit In subunit A, the open position is favored, with
three complexes exhibiting it, while three others present a
mixture of closed and open positions; no closed-only
position is observed In subunit B, two closed, four alternate
and no open-only positions are observed (Table 5) All the
conformational changes observed, however, are internal to
bOBP, and do not affect the protein surface It cannot be
excluded, however, that the different conformations of
Phe89 in each subunit result in a cross-talk of the
monomers, triggered by subtle differences in backbone/
side-chain dynamics and controlling the access of a second
molecule after the binding of the first one This mechanism
might be more general in dimeric OBPs, as rat OBP-F1 has
been shown to display an anti-cooperative mechanism [25]
Our results do not report a specificity of interaction between
bOBP and the limited number of odorant molecules studied
here However, considering the drastic differences between
the molecules chosen, a fine discriminating role of bOBP
seems to be excluded
The situation concerning a putative role of OBPs in OR
triggering, upon odorant binding, is far from clear The
high-affinity binding of empty pOBP to a human OR in
mammalian cells does not favor the role of OBP as ligand
carrier [38] Furthermore, the activity of the ORs in response
to odorant molecules alone has been reported [39] If a
model such as bacterial chemotaxis was applicable to
mammalian OBPs, it would be expected that they would
undergo a significant conformational change upon
com-plexation Such a behavior has been demonstrated for
serum retinol-binding protein, which, upon a conforma-tional change of a loop, can interact with a second protein, transthyretin, and prevent retinol excretion [40,41] With bOBP, the only conformational change observed in the crystal structure is internal to the molecule Similarly, no conformational changes have been detected by the struc-tural comparison between unliganded and liganded pOBP, nor were detected by IR spectroscopy [27] However, solution binding or spectroscopic studies on rat OBPs [17,25] have demonstrated a certain specificity of rOBPs for wide substrate classes as well as some conformational change upon receptor binding
In insects, antennal OBPs, pheromone-binding proteins (PBPs) and Chemosensory Proteins (CSPs)
perform considerable conformational changes [42] and could
be candidates for receptor triggering, as suggested recently from electro-sensillar recording [43] Fully sequenced insect genomes have shown the presence of multiple copies of putative OBP/PBP ORFs: 38 members in Drosophila mel-anogaster[44] and 29 members in Anopheles gambiae [45] Amazingly, the number of ORs expressed in Drosophila is
40 (on 62 candidates) [46,47], not disallowing the possible existence of functional OBP–OR couples In mammals, the number of ORs is in the order of several hundred, while the number of OBPs is at best a handful, not sufficient, by far, to suggest a specific functional interaction between a given OR with an OBP
Acknowledgements
We thank Dr Virna Conti for valuable technical assistance.
References
1 Bignetti, E., Cavaggioni, A., Pelosi, P., Persaud, K.C., Sorbi, R.T.
& Tirindelli, R (1985) Purification and characterisation of an odorant-binding protein from cow nasal tissue Eur J Biochem.
149, 227–231.
2 Pevsner, J., Trifiletti, R.R., Strittmatter, S.M & Snyder, S.H (1985) Isolation and characterization of an olfactory receptor
Fig 6 Solvent-accessible surface of bovine odorant-binding protein (bOBP) (A) The ribbon representation (purple) and molecular surface were calculated without taking into account Tyr83, leaving the cavity open The Tyr83 side-chain is presented in yellow in the closed and open positions, respectively; the arrow indicates the direction of the Tyr83 120 side-chain rotation for the opening of the cavity (B) The bOBP ribbon representation and molecular surface The surface has been slabbed, revealing the access channel to the cavity (double-headed arrow) when Tyr83 is rotated 120 outwards An undecanal molecule is represented (spheres) nested in the cavity The figure was produced using
GRASP [49] and PYMOL [48].
Trang 10protein for odorant pyrazines Proc Natl Acad Sci USA 82,
3050–3054.
3 Cavaggioni, A., Sorbi, R.T., Keen, J.N., Pappin, D.J & Findlay,
J.B (1987) Homology between the pyrazine-binding protein from
nasal mucosa and major urinary proteins FEBS Lett 212,
225–228.
4 Flower, D.R (1996) The lipocalin protein family: structure and
function Biochem J 318, 1–14.
5 Pelosi, P., Baldaccini, N.E & Pisanelli, A.M (1982) Identification
of a specific olfactory receptor for 2-isobutyl-3- methoxypyrazine.
Biochem J 201, 245–248.
6 Pelosi, P (1994) Odorant-binding proteins Crit Rev Biochem.
Mol Biol 29, 199–228.
7 Pelosi, P (1996) Perireceptor events in olfaction J Neurobiol 30,
3–19.
8 Bignetti, E., Cattaneo, P., Cavaggioni, A., Damiani, G &
Tir-indelli, R (1988) The pyrazine-binding protein and olfaction.
Comp Biochem Physiol B 90, 1–5.
9 Pevsner, J., Hou, V., Snowman, A.M & Snyder, S.H (1990)
Odorant-binding protein Characterization of ligand binding.
J Biol Chem 265, 6118–6125.
10 Burchell, B (1991) Turning on and turning off the sense of smell.
Nature 350, 16–17.
11 Schofield, P.R (1988) Carrier-bound odorant delivery to olfactory
receptors Trends Neurosci 11, 471–472.
12 Pelosi, P & Tirindelli, R.
19 (1989) Structure/Activity Studies and
Characterization of Odorant Binding Protein Marcel Dekker, New
York.
13 Vincent, F., Spinelli, S., Ramoni, R., Grolli, S., Pelosi, P.,
Cam-billau, C & Tegoni, M (2000) Complexes of porcine odorant
binding protein with odorant molecules belonging to different
chemical classes J Mol Biol 300, 127–139.
14 Koshland, D.E Jr (1981) Biochemistry of sensing and adaptation
in a simple bacterial system Annu Rev Biochem 50, 765–782.
15 Stewart, R.C & Dalquist, F.W.
20 (1987) Molecular components of
bacterial chemotaxis Chem Rev 87, 997–1025.
16 Herent, M.F., Collin, S & Pelosi, P (1995) Affinities of nutty
and green-smelling pyrazines and thiazoles to odorant-binding
proteins, in relation with their lipophilicity Chem Senses 20, 601–
608.
17 Lobel, D., Jacob, M., Volkner, M & Breer, H (2002) Odorants of
different chemical classes interact with distinct odorant binding
protein subtypes Chem Senses 27, 39–44.
18 Dal Monte, M.C.M., Anselmi, C & Pelosi, P (1993) Binding of
selected odorants to bovine and porcine odorant binding proteins.
Chem Senses 18, 713–721.
19 Tegoni, M., Ramoni, R., Bignetti, E., Spinelli, S & Cambillau, C.
(1996) Domain swapping creates a third putative combining site
in bovine odorant binding protein dimer Nat Struct Biol 3,
863–867.
20 Cowan, S.W., Newcomer, M.E & Jones, T.A (1990)
Crystal-lographic refinement of human serum retinol binding protein at
2A resolution Proteins 8, 44–61.
21 Bocskei, Z., Groom, C.R., Flower, D.R., Wright, C.E., Phillips,
S.E., Cavaggioni, A., Findlay, J.B & North, A.C (1992)
Phero-mone binding to two rodent urinary proteins revealed by X-ray
crystallography Nature 360, 186–188.
22 Bianchet, M.A., Bains, G., Pelosi, P., Pevsner, J., Snyder, S.H.,
Monaco, H.L & Amzel, L.M (1996) The three-dimensional
structure of bovine odorant binding protein and its mechanism of
odor recognition Nat Struct Biol 3, 934–939.
23 Ramoni, R., Vincent, F., Grolli, S., Conti, V., Malosse, C., Boyer,
F.D., Nagnan-Le Meillour, P., Spinelli, S., Cambillau, C &
Tegoni, M (2001) The insect attractant 1-octen-3-ol is the natural
ligand of bovine odorant-binding protein J Biol Chem 276,
7150–7155.
24 Spinelli, S., Ramoni, R., Grolli, S., Bonicel, J., Cambillau, C & Tegoni, M (1998) The structure of the monomeric porcine odorant binding protein sheds light on the domain swapping mechanism Biochemistry 37, 7913–7918.
25 Nespoulous, C., Briand, L., Delage, M.M., Tran, V & Pernollet, J.C (2004) Odorant binding and conformational changes of a rat odorant-binding protein Chem Senses 29, 189–198.
26 Wojnar, P., Lechner, M & Redl, B (2003) Antisense down-reg-ulation of lipocalin-interacting membrane receptor expression inhibits cellular internalization of lipocalin-1 in human NT2 cells.
J Biol Chem 278, 16209–16215.
27 Paolini, S., Scaloni, A., Amoresano, A., Marchese, S., Napolitano,
E & Pelosi, P (1998) Amino acid sequence, post-translational modifications, binding and labelling of porcine odorant-binding protein Chem Senses 23, 689–698.
28 Briand, L., Nespoulous, C., Perez, V., Remy, J.J., Huet, J.C & Pernollet, J.C (2000) Ligand-binding properties and structural characterization of a novel rat odorant-binding protein variant Eur J Biochem 267, 3079–3089.
29 Otwinovski, Z (1993)
21 DENZO: Oscillation Data and Reducing Program In Data Collection and Processing (Sawyer, L., Isaacs, N.W & Bailey, S., eds), pp.
22 56–63 DLSCI/R34 Daresbury Laboratory, Warrington, UK.
30 Collaborative Computational Project, Number 4
CCP4 suite: programs for crystallography Acta Crystallogr D 50, 760–766.
31 Roussel, A & Cambillau, C (1991) The TURBO-FRODO Graphics Package Silicon Graphics Geometry Partners Directory, Mountain View, CA.
32 Brunger, A.T., Adams, P.D., Clore, G.M., Delano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., Read, R.J., Rice, L.M., Simonson, T & Warren, G (1997) Crystallographic and NMR System (CNS), CNSsolve Yale University Press, New Heaven, CT.
33 Murshudov, G., Vagin, A.A & Dodson, E.J (1997) Refinement
of macromolecular structures by the maximum-likelihood method Acta Crystallogr D 53, 240–255.
34 Laskowski, R., MacArthur, M., Moss, D & Thornton, J (1993)
PROCHECK : a program to check the stereochemical quality of protein structures J Appl Crystallog 26, 91–97.
35 Ramoni, R., Vincent, F., Ashcroft, A.E., Accornero, P., Grolli, S., Valencia, C., Tegoni, M & Cambillau, C (2002) Control of domain swapping in bovine odorant-binding protein Biochem J.
365, 739–748.
36 Bussolati, L.R.R., Grolli, S., Donofrio, G & Bignetti, E (1993) Preparation of an affinity resin for odorants by coupling odorant binding protein from bovine nasal mucosa to Sepharose 4B.
J Biotechnol 30, 225–230.
37 Tegoni, M., Pelosi, P., Vincent, F., Spinelli, S., Campanacci, V., Grolli, S., Ramoni, R & Cambillau, C (2000) Mammalian odorant binding proteins Biochim Biophys Acta 1482, 229– 240.
38 Matarazzo, V., Zsurger, N., Guillemot, J.C., Clot-Faybesse, O., Botto, J.M., Dal Farra, C., Crowe, M., Demaille, J., Vincent, J.P., Mazella, J & Ronin, C (2002) Porcine odorant-binding protein selectively binds to a human olfactory receptor Chem Senses 27, 691–701.
39 Kiefer, H., Krieger, J., Olszewski, J.D., Von Heijne, G., Prestwich, G.D & Breer, H (1996) Expression of an olfactory receptor in Escherichia coli: purification, reconstitution, and ligand binding Biochemistry 35, 16077–16084.
40 Sivaprasadarao, A & Findlay, J.B (1993) Expression of func-tional human retinol-binding protein in Escherichia coli using a secretion vector Biochem J 296, 209–215.
41 Monaco, H.L (2000) The transthyretin–retinol-binding protein complex Biochim Biophys Acta 1482, 65–72.