Báo cáo Y học: Binding of Thermomyces (Humicola) lanuginosa lipase to the mixed micelles of cis-parinaric acid/NaTDC Fluorescence resonance energy transfer and crystallographic study

Eur. J. Biochem. 269, 1613–1621 (2002) (cid:211) FEBS 2002

Binding of Thermomyces(Humicola) lanuginosalipase to the mixed micelles of cis-parinaric acid/NaTDC Fluorescence resonance energy transfer and crystallographic study

Ste´ phane Yapoudjian1, Margarita G. Ivanova1, A. Marek Brzozowski2, Shamkant A. Patkar3, Jesper Vind3, Allan Svendsen3 and Robert Verger1 1Laboratoire de Lipolyse Enzymatique CNRS-IFR1, Marseille, France; 2Structural Biology Laboratory, Chemistry Department, University of York, UK; 3Enzyme Research, Novozymes A/S, Bagsvaerd, Denmark

TLL(W89L),

[TLL(S146A),

fluorescence resonance energy transfer. In addition, a TLL(S146A)/oleic acid complex was crystallised and its three-dimensional structure was solved. Surprisingly, two possible binding modes (sn-1 and antisn1) were found to exist between oleic acid and the catalytic cleft of the open con- formation of TLL(S146A). Both binding modes involved an interaction with tryptophan 89 of the lipase lid, in agreement with fluorescence resonance energy transfer experiments.As a consequence, we concluded that TLL(S146A) mutant is not an appropriate substitute for the wild-type Thermomyces lanuginosa lipase for mimicking the interaction between the wild-type enzyme and lipids.

Keywords: lipase; X-ray crystallography; cis-parinaric acid; fluorescence resonance energy transfer.

The binding of Thermomyces lanuginosa lipase and its TLL(W117F, mutants W221H, W260H)] to the mixed micelles of cis-parinaric acid/ sodium taurodeoxycholate at pH 5.0 led to the quenching of the intrinsic tryptophan fluorescence emission (300–380 nm) and to a simultaneous increase in the cis-parinaric acid fluorescence emission (380–500 nm). These findings were used to characterize the Thermomyces lanuginosa lipase/cis- parinaric acid interactions occurring in the presence of resonance sodium taurodeoxycholate.The fluorescence energy transfer and Stern–Volmer quenching constant values obtained were correlated with the accessibility of the tryptophan residues to the cis-parinaric acid and with the lid opening ability of Thermomyces lanuginosa lipase (and its mutants). TLL(S146A) was found to have the highest

Lipases (EC 3.1.1.3) can be defined as enzymes that catalyze the hydrolysis of long-chain acyl-glycerols [1]. In addition to playing an important role in fat catabolism, they have numerous applications in the food, cosmetics, detergent and pharmaceutical industries [2–5].

In recent years, the three-dimensional structures of lipases and lipase–inhibitor complexes have been determined using X-ray crystallographic methods [6–11]. All lipases show a common a–b hydrolase fold [12] and a catalytic triad composed of a nucleophilic serine, which is activated via

hydrogen bonds as part of a charge relay system, along with the histidine and the aspartate or glutamate residues [6,7]. The crystal structures of some lipases have shown that the active site is covered by a helical surface loop or (cid:212)lid(cid:213) that renders the active site inaccessible to substrate. This is referred to as the closed conformation of the lipase. On the other hand, the three-dimensional structures of lipases complexed with inhibitors shows a rearrangement of the lid, allowing free access to the active site in the so-called open conformation, in which a large hydrophobic surface around the catalytic triad is exposed.

Thermomyces lanuginosa lipase (TLL) has four trypto- phan residues located in positions 89, 117, 221 and 260. The side chains of W117, W221 and W260 are buried into the protein core, whereas the W89 residue is located in the central part of the helical lid [13]. In the crystal structures of the open forms of TLL, W89 is in close van der Waals contact with the acyl moiety of an inhibitor mimicking the transition state [14,15].

The fluorescence technique was used previously to study the binding of TLL to small or large unilamellar vesicles of 1-palmitoyl-2-oleoylglycero-sn-3-phosphoglycerol (POPG) and to vesicles of zwitterionic phospholipids such as 1-palmitoyl-2-oleoylglycero-sn-3-phosphocholine [16]. The authors concluded that TLL may bind with a similar affinity to all types of phospholipid vesicles and may adopt a catalytically active conformation and be involved in inter- facial activation processes only when small unilamellar vesicles of POPG are used. Furthermore, molecular

Correspondence to R. Verger, Laboratoire de Lipolyse Enzymatique, CNRS-IFR1, 31 chemin Joseph Aiguier, 13402 Marseille cedex 20, France. Fax: + 33 91 71 58 57, Tel.: + 33 91 16 40 93, E-mail: verger@ibsm.cnrs-mrs.fr Abbreviations: cis-PnA, cis-parinaric acid; CMC, critical micellar concentration; FRET, fluorescence resonance energy transfer; KSV, Stern–Volmer quenching constant; NaTDC, sodium taurodeoxycho- late; OA, oleic acid; POPG, 1-palmitoyl-2-oleoylglycero-sn-3-phos- phoglycerol; RFI, relative fluorescence intensity; TLL, Thermomyces lanuginosa lipase; TLL(S146A), inactive mutant with S146 mutated to A; TLL(W89L), mutant with W89 mutated to L; TLL(W117F, W221H, W260H), mutant with only the W89; W117 mutated to F, W221 mutated to H and W260 mutated to H. Note: the atomic co-ordinates have been deposited in the Brookhaven Protein Data Bank with the accession code 1gt6. (Received 20 August 2001, revised 5 December 2001, accepted 14 January 2002)

1614 S. Yapoudjian et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

dynamics simulations [17] indicated that the replacement of a single amino acid at the active site (S146A) may lead to conformational alterations in TLL.

The aim of the present study was to investigate the TLL/ fatty acid interactions using the fluorescence resonance energy transfer (FRET) technique.

path length 1 cm) of 3.5 mL each: one for the assay and one for the control assay. The contents of each cuvette were mixed 5–10 times by gentle inversion of the cuvette capped with Teflon stopper, and were then left unstirred during the measurement procedure. Measurements were performed at room temperature. Two types of experiments were per- formed. (a) Titration of cis-PnA was carried out by the increasing amounts of NaTDC at pH 5.0, in the absence of TLL. The assay and control cuvettes were both filled with buffer and NaTDC at the various concentrations tested. cis- PnA was subsequently added to the assay cuvette from an ethanolic stock solution and differential absorption spectra were recorded between 200 and 450 mn. (b) Absorption spectra of cis-PnA in the presence of TLL at pH 5.0 or pH 8.0. The assay and control cuvettes were both filled with buffer, NaTDC and TLL. cis-PnA was added afterwards into the assay cuvette and the differential absorption spectra were recorded.

One of the prerequisites to be able to observe the FRET between a donor (TLL tryptophans) and an acceptor (fatty acid) is that there must exist a spectral overlap between the donor emission and the acceptor absorption spectra, and the donor and acceptor groups must be the right distance apart and properly oriented [18]. 9,11,13,15-cis,trans,trans,cis-octadecatetraenoic Therefore acid (cis-PnA), a naturally fluorescent fatty acid with thoroughly characterized spectroscopic properties [19], was chosen for use as a probe. It has been previously established that cis-PnA can act as an acceptor for the tryptophan fluorescence emission [20], and its spectroscop- ic properties have been used in studies on fatty acid binding to various proteins [20,21].

Fluorescence spectroscopy

Bile salts are the main detergent-like molecules respon- sible for the solubilization of lipolytic products (monoglyc- erides and free fatty acids) during the digestion of dietary fats. Sodium taurodeoxycholate (NaTDC) is a conjugated bile acid, which forms very small micelles in an aqueous solution [22]. The mixed micelles of cis-PnA/NaTDC turned out to be a convenient model system for studying the interactions between a water soluble protein (TLL) and a fatty acid in the form of mixed micelles.

Fluorescence measurements were carried out at 29 (cid:176)C under constant stirring using a SFM 25 spectrofluorimeter from Kontron Instruments and a 3.5-mL quartz cuvette (optical path length 1 cm). During all the fluorescence measurements, the optical density was < 0.1 in the spectral range between 280 nm and 500 nm to avoid inner filter effect. Two types of fluorescence experiments were performed.

Titration of cis-PnA at various NaTDC concentrations at pH 5.0. The cuvette was filled with buffer containing NaTDC at a given concentration. cis-PnA was then added to the cuvette and a fluorescence emission spectrum was recorded at an excitation wavelength of 320 nm by scanning at an emission wavelength ranging from 350 nm to 500 nm.

First, we studied the lipase free cis-PnA/NaTDC system in order to characterize the cis-PnA/NaTDC mixed micellar system. The binding behavior of TLL (and its mutants) to pure cis-PnA and to mixed micelles of cis-PnA/NaTDC was then studied using the FRET technique. In addition, X-ray crystallogaphy studies were performed on the S146A mutant in order to elucidate the particular properties of its complexes with fatty acids.

M A T E R I A L S A N D M E T H O D S

Materials

FRET experiments. TLL (wild-type or mutant) was titrated at pH 5.0 or pH 8.0 by adding increasing amounts of cis- PnA in the presence and absence of NaTDC. The excitation wavelength was set to 280 nm and the emission wavelength ranged from 300 nm to 500 nm.

The accessibility of tryptophan to cis-PnA was estimated by measuring the quenching of the TLL fluorescence effected by cis-PnA, according to the Stern–Volmer equa- tion [25]:

NaTDC was from Sigma and cis-PnA was from Molecu- lar Probes. A stock solution of 3.2 mM of cis-PnA in ethanol containing 0.001% (w/v) butylhydroxytoluene (BHT) as an antioxydant was stored in the dark at )20 (cid:176)C under an argon atmosphere. These precautions were taken to ensure that no polyene decomposition would occur [20].

(cid:133)1(cid:134)

(cid:136) 1 (cid:135) Ksv(cid:137)Q(cid:138)

F0 F

The TLL wild-type, its single mutants: TLL(S146A), TLL(W89L), and triple mutant TLL(W117F, W221H, W260H) were used. All enzymes were kindly provided by A. Svendsen and S. A. Patkar from Novo Nordisk, Denmark and prepared as described previously [23,24].

The buffers used were 10 mM Tris/HCl pH 8.0, 150 mM NaCl, 21 mM CaCl2, 1 mM EDTA and 10 mM acetate pH 5.0, 100 mM NaCl, 20 mM CaCl2.

where F0 and F are the fluorescence emission intensities in the absence and in the presence of a quencher, respectively, [Q] is the molar quencher concentration and KSV is the Stern–Volmer quenching constant. KSV is appropriate for collisional quenching in which binding is not involved. However, the Stern–Volmer equation fits well our experi- mental results, even though binding is clearly involved. Consequently, KSV will be replaced by (cid:212)KSV(cid:213).

UV absorption spectroscopy

Protein crystallization and crystallography

Differential absorption spectra were recorded on a Uvikon 860 spectrophotometer from Kontron Instruments. All assays were carried out using two quartz cuvettes (optical

TLL(S146A) solution was washed several times in 10K Centricon in 10 mM Tris/HCl pH 8.0 buffer and concen-

Lipase binding to lipid, FRET and structural study (Eur. J. Biochem. 269) 1615

(cid:211) FEBS 2002

trated up to 20 mgÆmL)1. Crystallization trials were performed using the hanging drop technique at 291 K. Screening for the crystallization conditions was performed simultaneously at pH 8.0 (0.1 M Tris/HCl buffer) and pH 5.0 (0.1 M acetate buffer). OA was used instead of cis- PnA for crystallization experiments to avoid oxidation during the crystallization. OA was dissolved in iso-propanol and mixed in this form with a protein sample at a 5 : 1 molar ratio (OA/lipase). After a 1-h incubation, the resulting precipitate was removed by centrifugation in a Sigma Eppendorf centrifuge (5 min, 18 000 g) and the remaining protein was used in the crystallization experi- ments. NaTDC was added to the crystallization trials separately at a concentration of 10 mM. Crystals were flash frozen in the liquid nitrogen and characterized in-house on a Rigaku RU200 rotating anode source (k (cid:136) 1.5418 A˚ ), MAR Research 345 imaging plate scanner, Osmic focusing mirrors and Oxford Cryosystem set at 120 K. The X-ray data were subsequently collected at the ESRF in Grenoble on the MAR Research CCD detector at 100 K, processed with DENZO and scaled and merged with SCALEPACK [26]. The structure was determined using the Molecular Replacement method. The lid was removed by molecular modelling in QUANTA to get a model for molecular (lipase minus lid). The AMORE software replacement program [27] was used and the wild-type TLL structure [14] (minus the lid) was used as a model. The structure was refined using maximum likelihood techniques with REFMAC [28]; other calculations were carried out using the CCP4 suite of programs (Collaborative Computational Project, Number 4, 1994).

Electron density map inspection, model building and analysis were carried out with the X-FIT options of the QUANTA software program (Molecular Simulations Inc.).

Fig. 1. Fluorescence Emission spectra of TLL (––) and UV absorption spectra of cis-PnA (- - -). TLL and cis-PnA concentrations were 0.8 lM and 10 lM, respectively. The buffer used was 10 mM acetate (pH 5.0) 100 mM NaCl, 20 mM CaCl2 or 10 mM Tris (pH 8.0) 150 mM NaCl, 21 mM CaCl2, 1 mM EDTA. NaTDC concentration was 1 mM. The excitation wavelength used to obtain the fluorescence emission spectra was 280 nm.

R E S U L T S

Absorption spectroscopy

The UV absorption spectrum of cis-PnA was determined in an ethanolic solution (95%) and found to be identical to that obtained by Sklar et al. [19]. As cis-PnA is prone to oxidation, the absorption spectrum of the stock solution was checked regularly and no changes in the cis-PnA absorption spectra were observed in the ethanolic solution upon storage.

increase in the attenuence of cis-PnA leveled off at NaTDC concentrations above 4 mM.

Fluorescence spectroscopy

The UV absorption spectrum of cis-PnA at pH 5.0 and pH 8.0 as well as the fluorescence emission spectrum of TLL (excited at 280 nm, at pH 5.0) in the presence of 1 mM NaTDC are shown in Fig. 1. At pH 5.0, the cis-PnA UV absorption spectrum overlapped the TLL emission spec- trum in the 290–380 nm wavelength range, whereas no overlap can be observed at pH 8.0. No significant changes in the TLL emission spectrum were detected between pH 5.0 and pH 8.0 (data not shown).

The effects of NaTDC on the UV absorption spectrum of cis-PnA at pH 5.0 are shown in Fig. 2. In the absence of NaTDC, the cis-PnA solution was slightly turbid. As soon as the NaTDC concentration reached at least 1 mM, the solution became optically clear changing simultaneously the absorption spectrum of cis-PnA. Three main absorption peaks appeared at 298 nm, 304 nm and 326 nm and increased in proportion to the NaTDC concentration. This

No significant NaTDC fluorescence was recorded under our experimental conditions. The excitation and emission spec- tra of cis-PnA were recorded at various NaTDC concen- trations at pH 5.0. The maximum of the excitation and the emission spectra were found to occur at 320 nm and 410 nm, respectively. In order to estimate the critical micellar concentration (CMC) of NaTDC, the relative fluorescence intensity (RFI) of cis-PnA at 410 nm (excita- tion wavelength at 320 nm) was measured as a function of the NaTDC concentration at pH 5.0. At NaTDC concen- trations lower than 1 mM, the fluorescence of cis-PnA was

Fig. 2. Effects of various NaTDC concentrations on the UV absorption spectra of a solution of cis-PnA. The cis-PnA concentration was kept constant at 10 lM. Buffer was 10 mM acetate (pH 5.0) 100 mM NaCl, 20 mM CaCl2. (- - -) 0 mM NaTDC, (– - –) 1 mM NaTDC, (– –) 2 mM NaTDC, (––) 4 mM NaTDC. The schematic diagram on the right illustrates the experimental protocol used.

1616 S. Yapoudjian et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

Fig. 4. RFI decrease (d) at kmax as well as RFI increase (s) at k410 nm as a function of cis-PnA concentration. Data from Fig. 3.

negligible. The RFI increased in parallel with the rise in the NaTDC concentration above 1 mM. This increase leveled off at NaTDC concentrations higher than 4 mM (data not shown).

The results of the FRET recordings obtained between TLL and cis-PnA, at wavelengths ranging from 300 to 500 nm in the presence of NaTDC at pH 5.0, are presented in Fig. 3. As the molar ratio (R) between cis-PnA and TLL increased, the RFI decreased at wavelengths ranging from 300 to 380 nm and increased simultaneously at wavelengths ranging from 380 to 500 nm.

From the data presented in Fig. 3, the decrease in RFI (%), measured at the maximal emission wavelength (kmax), as well as the increase of RFI (%), measured at 410 nm, as a function of cis-PnA concentration are presented in Fig. 4. A good quantitative correlation between increase and decrease of RFI as a function of increasing concentration of cis-PnA can be seen. Furthermore, a plateau value is reached when one molecule of TLL is added to one molecule of cis-PnA (R (cid:136) 1). Similar curves as those presented in Fig. 4 were also obtained for TLL(S146A), TLL(W117F, W221H, W260H) and TLL(W89L) (data not shown).

Similar FRET experiments were also performed with TLL, TLL(S146A), TLL(W117F, W221H, W260H), TLL(W89L) and cis-PnA in the presence and absence of NaTDC (Fig. 5). In the presence of NaTDC, the FRET was observed between TLL, TLL(S146A), TLL(W117F, W221H, W260H), TLL(W89L) and cis-PnA. In the absence of NaTDC, the FRET was negligible. Surprisingly, in the absence of NaTDC, a clear-cut quenching process was observed only with TLL(S146A) and TLL(W89L). Similar experiments were performed at pH 8.0, in the presence of

Fig. 3. FRET between TLL and cis-PnA. The numbers refer to the values of the molar ratio R of cis-PnA to TLL. In all the assays, the excitation wavelength was 280 nm. The dotted line corresponds to the fluorescence emission spectra of cis-PnA (1 lM) recorded in the absence of TLL under the same experimental conditions. The dashed line corresponds to the arithmetic sum of the TLL and cis-PnA spectra recorded separately. The correlation between quenched tryptophan RFI (325 nm) and sensitized RFI of cis-PnA (410 nm) is presented in Fig. 4. TLL concentration was 0.8 lM and cis-PnA concentration varied from 0 to 1 lM. For the sake of clarity, only spectrum corres- ponding to three cis-PnA concentrations (0, 0.4 and 0.8 lM) are shown (plain lines). The NaTDC concentration was 1 mM. Buffer was 10 mM acetate (pH 5.0) 100 mM NaCl, 20 mM CaCl2.

Fig. 5. FRET between TLL (and its mutants) and cis-PnA in the presence and absence of NaTDC. The protein concentration was 0.8 lM and the cis-PnA concentration was varied stepwise from 0 to 1 lM (0, 0.2, 0.4, 0.8, 1 lM). Excitation wavelength: 280 nm. Buffer pH 5.0 as in Fig. 2. The data in the graph at the uppermost left hand corner are identical to those shown in Fig. 3.

Lipase binding to lipid, FRET and structural study (Eur. J. Biochem. 269) 1617

(cid:211) FEBS 2002

Table 1. Effects of NaTDC (1 mM) and/or cis-PnA (1 lM) on kmax (nm) of the RFI of TLL and its mutants. Data from Fig. 5. Buffer was 10 mM acetate (pH 5.0) 100 mM NaCl, 20 mM CaCl2. The protein concentration was 0.8 lM and the excitation wavelength was 280 nm.

[NaTDC] (cid:136) 0 mM [NaTDC] (cid:136) 1 mM

Protein (0.8 lM) [PnA] (cid:136) 0 lM [PnA] (cid:136) 1 lM [PnA] (cid:136) 0 lM [PnA] (cid:136) 1 lM Dkmax Dkmax

NaTDC (data not shown). Quenching was observed only between TLL(S146A), TLL(W89L) and cis-PnA.

The maximum fluorescence emission wavelengths (kmax) of TLL, TLL(S146A), TLL(W117F, W221H, W260H), and TLL(W89L) (excitation at 280 nm, pH 5.0) with or without in the presence or absence of cis-PnA are NaTDC, summarized in Table 1. In the absence of cis-PnA, the addition of NaTDC led to a blue shift of the kmax of all the lipases tested, except for TLL(W89L). Furthermore, in contrast to what occurred with TLL(W89L), the addition of cis-PnA in the presence of 1 mM NaTDC also led to a significant blue shift in the case of TLL, TLL(S146A) and TLL(W117F, W221H, W260H). It is worth noting that TLL(W89L) displayed no significant blue shift under any of the experimental conditions tested.

Stern–Volmer plots for the fluorescence quenching of TLL (mutants) by cis-PnA were calculated from the data presented in Fig. 5, in the presence of 1 mM NaTDC (data not shown). The (cid:212)KSV(cid:213) constants calculated for TLL, and TLL(S146A), TLL(W117F, W221H, W260H) )1, TLL(W89L) were 3.2.106, 4.6.106, 3.4.106 and 1.5.106 M respectively.

X-ray crystallography

were set at zero during the refinement procedure and consequently in the final model as well. This was the case in particular with loop 24–44 in molecule B and few residues of this loop in molecule A. 262 water molecules were identified and refined. Both molecules have open ((cid:212)fully active(cid:213)) conformations, as discussed previously [15]. After satisfactory convergence of the refinement of the protein and water molecules, the remaining positive electron density in the surroundings of the active site cavities was analyzed. This made it possible to model and refine the full-length OA molecule in the active site of molecule B. Due to the residual electron density present in the active site of molecule A, the modeling of the ligand was restricted to its carboxylic group and the alkyl chain between C2 and C9. The remaining atoms of the OA in molecule A, C10–C18, were included in the final protein model for the sake of overall clarity but their occupancies were set to zero, as the electron density of these atoms was negligible. The OA compound was trapped in the active site of molecule B in an unexpected manner (Fig. 6A); it was rotated by (cid:25) 180(cid:176) with respect to the main sn-1 TLL alkyl chain binding site [14,15], which is referred to here as antisn1. Structural studies of fatty acids bound to human serum albumin also revealed this unexpected behaviour [29]. The OA carboxylic group of lipid molecule is thrust deeply into this alternative binding cleft of molecule B, where it is anchored via hydrogen bonds between its carboxylic group and the carbonyl oxygen of N92 (2.7 A˚ ) and NE2 of H110 (Fig. 6A). The C9–C10 cis double bond lies near the b carbon of A146 ((cid:25) 3.0 A˚ ), causing this alkyl chain to bend and become wrapped around the W89 residue of the lid, and the C18 carbon is finally wedged between CD1 of I255 and CH2 of W89.

The partially defined electron density map of the OA in molecule A binding site cavity was used to model the carboxylic group lying on the top of A146, which is stabilized by a hydrogen bond with the carbonyl oxygen of this residue (2.8 A˚ ) and the NE2 atom of H256 (2.8 A˚ ). The fact that C2-C9 carbon atoms of the lipid occupy the sn-1 position in the active site indicates that the lipid binds according to the (cid:212)conventional mode(cid:213) in this molecule. The location of the remaining OA atoms is not very clear, due to the very weak electron density but the position of the first atom of the cis C9–C10 bond was used to model the remaining part of thsis moiety in a similar manner to the model of one of the alkyl chains in the TLL complex with di-dodecyl phosphatidylcholine [15] (not presented in Fig. 6B for the sake of clarity), and with dodecyl phospho- nate inhibitor [14], which occupies the main sn-1 binding site as depicted in Fig. 6B.

Good X-ray quality crystals of TLL(S146A) were obtained in the presence of OA in 0.1 M Tris/HCl pH 8.0 buffer, 10 mM NaTDC, 25% w/v poly(ethylene glycol) 5K MME, 25 mM MgCl2. Crystallization at pH 5.0 and control setups were unsuccessful under similar conditions with wild-type TLL. Crystals of the TLL(S146A) mutant were found to belong to the P21212 space group and to have two molecules in the asymmetric unit, a packing density of 2.64 A˚ 3ÆDa)1 and a solvent content of 53%. The final X-ray data are 97.8% complete up to 2.20 A˚ resolution (96.2% in the 2.28–2.20 resolution shell) with an overall Rmerge of 0.075 (0.44), I/r(I) of 12.2, and a mean multiplicity of 3.2 observations per reflection. The final model has a crystal- lographic factor of 21.4 and a Rfree of 23.9 against all reflections in the resolution range of 20–2.20 A˚ . The overall root mean square deviations (rmsd) from geometrical ideality are 0.009 A˚ in bond lengths, 1.3(cid:176) in bond angles, and 1.2 A˚ 2 for the DB between bonded atoms. This model is composed of all the atoms of all the residues between E1 and L269 in the case of molecule A (chain A) and molecule B (chain B), with a rmsd of 0.17 A˚ between the corres- ponding Ca atoms of molecule A and B. However, due to the high mobility and the resulting lack of clarity of the electron density maps, occupancies of only a few residues

TLL TLL(S146A) TLL(W117F, W221H, W260H) TLL(W89L) 326 329 335 313 326 328 335 312 0 )1 0 )1 322 324 324 313 315 310 304 311 )7 )14 )20 )2

1618 S. Yapoudjian et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

Fig. 6. Oleic acid binding modes revealed by TLL-OA complex crystal structure. (A) Anti sn-1 position of the OA in the hydrophobic catalytic cleft of the open conformation of TLL(S146A). The REFMAC 2Fo-Fc electron density map of the active site surroundings in molecule B (contoured at 1 r level), with the ligand – oleic acid (OA) – omitted from this calculation; dashed lines indicate hydrogen bonds between OA and the protein. (B) Comparisons between several ligand binding modes in the catalytic cleft of TLL. The stereoview of OA binding modes in molecule A and B (OA_A and OA_B, respectively, thick lines); C9 and C10 correspond to the position of these atoms in the OA_A ligand, where the electron density was visible from its carboxylic group up to the C9-C10 carbon atoms: the alkyl chain of OA_A is included in the model beyond this double bond for the sake of clarity. The corresponding part of the covalent complex of the C12 posphonate inhibitor–TLL [14], in which the C12 alkyl moiety occupies sn-1 site of the active center, is also shown (protein shown by a thin line, ligand marked C12), for comparisons.

D I S C U S S I O N

continuous and specific lipase assay involving the use of the naturally fluorescent oil from Parinari glaberrimum [30].

Mixed micelles of cis-PnA/NaTDC

The CMC ((cid:25) 1 mM) of NaTDC measured using cis-PnA as reporter is in good agreement with the previously published values [22]. However, it is worth noting that cis- PnA cannot be used as a general fluorescent probe to evaluate the CMC of synthetic detergents such as Tween 20, Chaps, and Nansa (alkyl benzene sulfonate). In contrast to what occurred with NaTDC, no sharp increase was observed in the RFI of cis-PnA at the respective CMCs of the above mentioned detergents. Therefore, NaTDC appeared to be the most suitable detergent for the present studies, as FRET measurements can be performed after the incorporation of cis-PnA into mixed micelles.

We have observed by direct excitation at 280 nm an increase of the RFI of cis-PnA with increasing concentra- tions of NaTDC (data not shown). Consequently, we have selected a NaTDC concentration of 1 mM to perform the

Adding NaTDC to an aqueous solution of cis-PnA resulted in a drastic change in the UV absorption spectra of this fatty acid (Fig. 2). This spectroscopic property of cis-PnA is probably associated with the transformation of cis-PnA aggregates into mixed micelles of cis-PnA/NaTDC. The presence of mixed micelles of cis-PnA/NaTDC was also suggested by the increase in the fluorescence emission intensity of cis-PnA recorded with increasing amounts of NaTDC (data not shown). The results obtained using cis-PnA as a fluorescent reporter, indicate that the increase in the cis-PnA fluorescence intensity was due to its incorpor- ation into the NaTDC micelles, resulting in a drastic change in the microenvironment to which cis-PnA was exposed. This phenomenon has already been used as the basis of a sensitive,

Lipase binding to lipid, FRET and structural study (Eur. J. Biochem. 269) 1619

(cid:211) FEBS 2002

spectra obtained after mixing TLL and mixed micelles of cis-PnA/NaTDC. Moreover, as the molar ratio (R) between cis-PnA and TLL increased, the RFI recorded at wave- lengths ranging from 300 to 380 nm decreased, while increasing simultaneously at wavelength ranging from 380 to 500 nm (Fig. 4). This behavior indicates that a FRET occurred between TLL and cis-PnA. An isobestic point was observed at 380 nm. The distance between the donor and the acceptor can be estimated to be around 25 A˚ [20]. Similar experiments were performed with a TLL mutant in four tryptophan residues were mutated to which all nonfluorescent amino acids (data not shown). In this case, the RFI decreased by 10-fold and no significant FRET was observed. This indicates that any contributions of the tyrosine residues to the FRET were negligible under our experimental conditions.

FRET experiments in order to optimize the signal to noise ratio. Under these conditions (NaTDC concentration of 1 mM), some cis-PnA molecules are not incorporated into the mixed micelles of cis-PnA/NaTDC, as already indicated from the data presented in Fig. 2. It is known that the micelle formation process of bile salts is a complex process. Pre-micellar aggregates of various sizes have been described previously [22,31]. This may explain why at 1 mM NaTDC, we observed a clear and characteristic UV absorption spectrum of cis-PnA, different from the one recorded in the absence of NaTDC (see Fig. 2). Despite this limitation (a high RFI background resulting from a direct cis-PnA excitation at 280 nm), we observed a significant FRET at a NaTDC concentration of 4 mM (data not shown). Further- more, the addition of aliquots of pure ethanol (up to a concentration of 10%, v/v) to a mixed solution of cis-PnA/ NaTDC (1 lM/1 mM) does not significantly change the RFI of the cis-PnA (data not shown).

the kmax of

Binding of TLL (and mutants) to cis-PnA and NaTDC

In the presence of NaTDC, without cis-PnA. The presence of NaTDC at a concentration of 1 mM resulted in a blue shift of the kmax of the RFI of TLL (4 nm), TLL(S146A) (5 nm) and TLL(W117F, W221H, W260H) (11 nm) (see Table 1), which was probably due to the decreasing polarity of the local environment of W89 in these three molecules. The lack of any wavelength shift observed for TLL(W89L) and the highest blue shift observed for TLL(W117F, W221H, W260H) indicate that the lid’s W89 was involved in the interaction with NaTDC micelles.

These findings are in an agreement with the data obtained from other studies [16,32], showing that W89 is the only accessible tryptophan of the TLL and that the lid region is directly involved in the binding of TLL to micelles of the pentaoxyethylene octyl ether (C8E5) detergent [32].

When mixed micelles of cis-PnA/NaTDC were added, a blue shift of the RFI of TLL (7 nm), TLL(S146A) (14 nm) or TLL(W117F, W221H, W260H) (20 nm) was observed, which indicates that the environment of their tryptophan residues became less polar (see Table 1). The greatest blue shift which occurred with TLL(W117F, W221H, W260H) and the smallest one with TLL(W89L) (2 nm) suggest that the lid is involved in the binding of the mixed micelles of cis-PnA/NaTDC. On the other hand, these shifts might result from the quenching of the tryptophan fluorescence in the presence of the mixed micelles, which simultaneously reveals the fluorescence of the tyrosine residues. However, this explanation can be ruled out, as no significant shift was observed with the TLL(W89L) mutant in the presence of the mixed micelles. Similar experiments were performed with BSA in a control experiment and no blue shift was observed, although FRET showed a characteristic decrease in the RFI at 330 nm and a simultaneous increase at 410 nm (data not shown). If the quenching of the tryptophan emission had revealed the tyrosine emission, then we would also have observed a spectral shift in the experiments performed with BSA and TLL(W89L), which was not the case. We can therefore attribute the spectral blue shifts observed with mutants TLL(S146A) and TLL(W117F, W221H, W260H) to a change in the local surroundings of their W89 residues towards a less polar environment.

the fluorescence

The accessibility of the tryptophan residues to the cis- PnA quencher was used to estimate the changes taking place in the lid region of TLL and its mutants. As the values of (cid:212)Ksv(cid:213) calculated with TLL and TLL(W117F, W221H, W260H) were the same, we can assume that W89 is the only tryptophan side chain accessible to cis-PnA. This conclusion was also confirmed by the lowest (cid:212)Ksv(cid:213) obtained for TLL(W89L).

In the absence of NaTDC, with cis-PnA. The addition of cis-PnA to TLL (or mutants) did not lead to any significant changes in the kmax of the RFI of the proteins (see Table 1). Moreover, no FRET (Fig. 5) was observed for TLL or for its mutants upon addition of cis-PnA, probably due to the physico-chemical state of cis-PnA in the absence of NaTDC micelles. However, the presence of cis-PnA led to a quenching of emission only with TLL(S146A) and TLL(W89L). The S146A and W89L mutations are located either in the catalytic triad or in the lid, respectively, and favor the open structure of the lipase in solution, enhancing its interaction with cis-PnA. Hence, the mutant TLL(S146A) is not an appropriate substitute for the wild-type TLL for mimicking the interaction between lipase and lipids.

In the presence of NaTDC and cis-PnA. The absence of significant FRET at pH 8.0 with an excitation wavelength ranging from 250 to 290 nm (data not shown), may be attributed to the lack of spectral overlap between the UV absorption spectrum of cis-PnA and the fluorescence emission spectrum of TLL (Fig. 1).

The FRET between TLL and mixed micelles of cis-PnA/ NaTDC at pH 5.0 can be illustrated by comparing the separately taken spectra of TLL and cis-PnA (dashed curve from Fig. 3 corresponding to their arithmetic sum) with

The blue shift observed in the case of TLL(W117F, W221H, W260H) confirms that the lid is involved in the interaction between TLL and the mixed micelles of cis-PnA/ NaTDC. The highest accessibility of W89, assessed by (cid:212)KSV(cid:213) values, observed with TLL(S146A) indicates that this mutant has a higher binding affinity towards mixed micelles of cis- PnA/NaTDC than the wild-type TLL. The S146A mutation presumably destabilizes the closed conformation of the lipase, exposing a cluster of hydrophobic amino acids, including L206, F95, F113, F211, Y21, A146, L147 and A146, and consequently enhances the interaction between the lipase and the lipid aggregates, resulting in an efficient

1620 S. Yapoudjian et al. (Eur. J. Biochem. 269)

(cid:211) FEBS 2002

FRET. This is in agreement with the molecular dynamics simulations data obtained by Peters et al. [17], which indicated that the mobility of the lid was enhanced in the TLL(S146A) mutant. Furthermore, the results of independ- ent direct binding measurements carried out on TLL and TLL(S146A) with monomolecular films of substrate ana- logues have shown that TLL(S146A) has the highest affinity for these substrates (S. Yapoudjian, M. Ivanova, I. Douchet, A. Ze´ nath, M. Sentis, W. Marine, A. Svendsen and R. Verger, unpublished data). This confirms that TLL(S146A) is not an appropriate substitute for the wild-type TLL for mimicking the interaction between lipase and lipids.

Crystal structure of TLL(S146A) complexed with OA

contact of L147 with W89 of the lid that is present in all closed wild-type TLL structures, resulting in a higher mobility of the lid in the S146A mutant. Moreover, the removal of the potential structural strain of Ser146–His258 hydrogen bond may be propagated through freed His258 into the C-terminal (262–269) region of Tl lipase, which is thought to be one of the crucial structural element of this enzyme involved in the first steps of interfacial activation [15,16]. Any small changes in this C-terminus/Arg84 switch area may therefore contribute substantially to the destabil- ization of the closed (low activity or activated form [15]) forms of Tl lipase, facilitating faster opening of the lid and its transformation into the fully active form. It is also likely, that the lack of a hydroxyl group of Ser146 diminishes the role of this residue in the stabilization of the substrate– ligand molecule, leading to the (cid:212)confusion(cid:213) of the enzyme in the process of the ligand recognition. This may result in the two alternative OA binding modes observed here. The better defined electron density for OA in molecule B may than in molecule A suggests that the antisn1 binding mode of this lipid is favored by the mutant. This is probably due to the stabilizing effect of the hydrogen bonds of D92 and H110 with carboxyl group of the OA molecule, specific for this particular ligand conformation. Whether this form of OA binding is enhanced further by the more stable conforma- tion of the lid in molecule B is difficult to assess as generally very weak crystal contacts of the lid in molecule B are only marginally stronger than in molecule A, and the tempera- ture factors of these regions are similar in both molecules. The control crystallizations of the wild-type of TLL under the conditions used for the S146A mutant, have been unsuccessful. This deficiency of crystals of the wild-type TLL–OA complex may result from the lack of interactions between TLL and OA in the crystallization conditions that is in an agreement with the FRET data observed here. Hence the physiological relevance of the two binding modes of OA observed in S146A mutant should be interpreted with caution, as they may result from the small changes in the ligand binding cavity caused by the S146A mutation.

Micellar or molecular binding?

The crystal structure of TLL(S146A) in a complex with OA is virtually identical to other complexed structures of this enzyme [14,15], and can be classified as (cid:212)fully active(cid:213) conformation of TLL according to Brzozowski et al. [15]. The lids in both molecules (A and B) are well defined in the electron density maps and are in the same fully opened conformations. They are not involved in the interactions between molecule A and B, and are almost free from intermolecular contacts. For example, the W89 in molecule A is (cid:25) 8 A˚ from its nearest crystal lattice neighbour (F211) and (cid:25) 16 A˚ from W89 of symmetry related molecule. The nearest ((cid:25) 4 A˚ ) and only intermolecular contact of the lid in molecule A is between its E87 and symmetry linked D111. There is only one (nondirect) hydrogen bond via water molecules between E87 and N95–D96 of the symmetry related molecule. The lid in molecule B is also exposed to a large crystal cavity and is free from strong intermolecular interactions. However, the proximity ((cid:25) 4 A˚ ) of L90 from L90 and D94 of symmetry related molecule might have some stabilizing effect on the lid in this molecule. This would explain much better definition of the OA electron density in the molecule B in comparison with higher disorder of the ligand in the molecule A. However, despite of the lack of strong lattice contacts that may affect the conformation of the lid, its high mobility (described in other TLL crystal structure with unrestricted lid positioning [9]) is not observed in the reported structure.

It is worth noticing that the FRET technique used in the present study cannot, in principle, distinguish a micellar from a molecular binding mode. In other words, one could expect to observe comparable fluorescence signals whether a lipase molecule binds to the mixed micelle of cis-PnA/NaTDC or toa single molecule of cis-PnA incorporated into the mixed micelles. These two alternative models are reminiscent to a long standing discussion about the surface dilution pheno- menon concerning the interaction of phospholipase A2 with mixed micelles of Triton X-100/phospholipid [33].

the stability of

that

As seen in Fig. 4, the FRET signal reaches a plateau value when a molecule of TLL is added per molecule of cis- PnA. This stoechiometry is more in favour of a molecular recognition rather than a micellar binding mode. More- over, we determined the three-dimensional structure of the open TLL(S146A), co-crystallised with mixed OA/NaTDC micelles. We identified one OA molecule per TLL mono- mer lying in the catalytic site, in contact with W89, according to two binding modes (see Fig. 6). These structural considerations suggest the FRET was probably due to a molecular binding of TLL to cis-PnA.

The OA has been found in the exposed, catalytic cleft of the enzyme in two completely different orientations: (cid:212)clas- sical(cid:213) sn-1 position and, unexpectedly, in a antisn1 binding mode. These two different OA binding modes may shed some light on the unusual spectroscopic properties of the Ser146fiAla mutant described and discussed above. Firstly, the Ser146fiAla mutation abolishes OG Ser146–NE2 His258 hydrogen bond (2.7 A˚ ), which stabilizes 144–148 loop with the active serine at its apex [6,14,15]. As the Ala146 residue is released from this interaction it collapses slightly deeper ((cid:25) 0.62 A˚ S146Ca–A146Ca distance) into the protein core creating more space and flexibility for the putative ligand in the active site region. This relaxation, together with the lack of steric hindrance created usually by the hydroxyl group of Ser146 in the wild-type enzyme, results in a larger binding cavity, capable of accommodation of lipid analogues normally not acceptable by the wild-type active site. Secondly, the local perturbation caused by Ser146fiAla mutation may affect its neighbouring residue: L147. This can lead to a disruption of the weak ((cid:25) 3.8 A˚ ), but likely stabilizing, van der Waals

Lipase binding to lipid, FRET and structural study (Eur. J. Biochem. 269) 1621

(cid:211) FEBS 2002

Thus FRET can be taken as an index of the open conformation of TLL.

Humicola lanuginosa lipase through X-ray crystallography and intuitive modelling. Protein Eng. 7, 543–550.

A C K N O W L E D G E M E N T S

15. Brzozowski, A.M., Savage, H., Verma, C.S., Turkenburg, J.P., Lawson, D.M., Svendsen, A. & Patkar, S.A. (2000) Structural origins of the interfacial activation in Thermomyces (Humicola) lanuginosa lipase. Biochemistry 39, 15071–15082.

16. Cajal, Y., Svendsen, A., Girona, V., Patkar, S.A. & Alsina, M.A. (2000) Interfacial control of lid opening in Thermomyces lanugi- nosa lipase. Biochemistry 39, 413–423.

17. Peters, G.H., Svendsen, A., Langberg, H., Vind, J., Patkar, S.A., Toxavaerd, S. & Kinnunen, P.K.J. (1998) Active serine involved in the stabilization of the active site loop in the Humicola lanuginosa lipase. Biochemistry 37, 12375–12383.

18. Cantor, C.R. & Schimmel, P.R. (1980) Fluorescence Spectroscopy in Biophysical Chemistry, pp. 433–465. W.H. Freeman, San Francisco. This research was carried out with financial support of the BIOTECH program of the European Union under contract no. BIO4-CT97-2365. We would like to thank the staff and beamline managers at the European Synchrotron Radiation Facility (ESRF) (Grenoble) and SRS Daresbury for their assistance with the data collection. The infrastruc- ture of the Structural Biology Laboratory in York is supported by the Biology and Biotechnology Science Research Council (BBSRC). We would like to thank Dr Antonie J. VISSER (Wageningen Agricultural University, MicroSpectroscopy Centre, Laboratory of Biochemistry, the Netherlands) and Prof. J. Sturgis (LISM, CNRS, Marseille, France) for fruitful discussions. The assistance of Dr Jessica Blanc is acknowledged for revising the English manuscript.

19. Sklar, L.A., Hudson, B.S., Petersen, M. & Diamond, J. (1977) Conjugated polyene fatty acids as fluorescent probes: spectro- scopic characterization. Biochemistry 16, 813–818.

R E F E R E N C E S

1. Verger, R. (1997) Interfacial activation of lipases: facts and arti- 20. Sklar, L.A., Hudsen, B.S. & Simoni, R.D. (1977) Conjugated polyene fatty acids as fluorescent probes: binding to bovine serum albumin. Biochemistry 16, 5100–5108. facts. Trends Biotech. 15, 32–38.

2. Schmid, R.D. & Verger, R. (1998) Lipases: interfacial enzymes International applications. Angew. Chem. with attractive 1608–1633. 3. Faber, K. (1992) Bio-Transformations in Organic Chemistry. 21. Dansen, T.B., Westerman, J., Wouters, F.S., Wanders, R.J.A., van Hoek, A., Gadella, T.W.J. & Wirtz, K.W.A. (1999) High- a(cid:129)nity binding of very-long-chain fatty acyl-CoA esters to the peroxisomal non-specific lipid-transfer protein (sterol carrier protein-2). Biochem. J. 339, 193–199. Springer-Verlag, Berlin.

22. Small, D.M. (1971) The physical chemistry of cholanic acids. In: The Bile Acids (Nair, P.P. & Kritchevski, D., eds), pp. 249–356. Plenum Press, N.Y.

4. Alberghina, L. (2000) Protein Engineering in Industrial Biotech- nology, Harwood Academic Publishers, Australia, Canada, France. 5. Bornscheuer, U.T. & Kazlauskas, R.J. (1999) Hydrolases in Organic Synthesis: Regio- and Stereo-selective Biotransformations. Wiley-VCH, Weinheim, New York. 23. Svendsen, A., Clausen, I.G., Patkar, S.A., Borch, T. & Thellersen, M. (1997) Protein engineering of microbial lipases of industrial interest. Methods Enzymol. (Dennis, E. & Rubin, B., eds), pp. 317–340.

6. Brady, L., Brzozowski, A.M., Derewenda, Z.S., Dodson, E., Dodson, G., Tolley, S., Turkenburg, J.P., Christiansen, L., Huge-Jensen, B., Norskov, L., Thim, L. & Menge, U. (1990) A serine protease triad forms the catalytic centre of a triacylglycerol lipase. Nature 343, 767–770. 24. Holmquist, M., Martinelle, M., Clausen, I.G., Patkar, S., Svendsen, A. & Hult, K. (1994) Trp89 in the lid of Humicola lanuginosa lipase is important for e(cid:129)cient hydrolysis of tributyrin. Lipids 29, 599–603. 7. Winkler, F.K., d’Arcy, A. & Hunziker, W. (1990) Structure of human pancreatic lipase. Nature 343, 771–774.

8. van Tilbeurgh, H., Egloff, M.-P., Martinez, C., Rugani, N., Verger, R. & Cambillau, C. (1993) Interfacial activation of the lipase–procolipase complex by mixed micelles revealed by X-ray crystallography. Nature 362, 814–820. 25. Lehrer, S.S. (1971) Solute perturbation of protein fluorescence. The quenching of the tryptophyl fluorescence of model compounds and of lysozyme by iodide ion. Biochemistry 10, 3254–3263. 26. Otwinowski, Z. & Minor, W. (1997) Processing of X-ray diffrac- tion data collection in oscillation mode. Methods Enzymol. 276, 307–326. 27. Navaza, J. (1994) AMORE: an automated package for molecular replacement. Acta Crystallogr. A50, 157–163.

28. Murshudov, G.N., Vagin, A.A. & Dodson, E.J. (1997) Refine- ment of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D53, 240–255. 9. Derewenda, U., Swenson, L., Wei, Y.Y., Green, R., Kobos, P.M., Joerger, R., Haas, M.J. & Derewenda, Z.S. (1994) Conforma- tional lability of lipases observed in the absence of an oil–water interface. Crystallographic studies of enzymes from the fungi Humicola lanuginosa and Rhizopus delemar. J. Lipid Res. 35, 524–534.

29. Bhattacharya, A.A., Gru¨ ne, T. & Curry, S. (2000) Crystal- lographic analysis reveals common modes of binding of medium and long-chain fatty acids to human serum albumin. J. Mol. Biol. 303, 721–732.

10. Brzozowski, A.M., Derewenda, U., Derewenda, Z.S., Dodson, G.G., Lawson, D.M., Turkenburg, J.P., Bjorkling, F., Huge- Jensen, B., Patkar, S.A. & Thim, L. (1991) A model for interfacial activation in lipases from the structure of a fungal lipase–inhibitor complex. Nature 351, 491–494.

30. Beisson, F., Ferte´ , N., Nari, J., Noat, G., Arondel, V. & Verger, R. (1999) Use of naturally fluorescent triacylglycerols from Parinari glaberrimum to detect low lipase activities from Arabidopsis thaliana seedlings. J. Lipid Res. 40, 2313–2321. 11. Grochulski, P., Li, Y., Schrag, J.D. & Cygler, M. (1994) Two conformational states of Candida rugosa lipase. Protein Sci. 3, 82–91. 31. Small, D.M. (1968) Size and structure of bile salt micelles. Advan. Chem. Series 84, 31–52.

12. Ollis, D.L., Cheah, E., Cygler, M., Dijkstra, B., Frolow, F., Franken, S.M., Harel, M., Remington, S.J., Silman, I., Schrag, J., Sussman, J.L., Verschueren, K.H.G. & Goldman, A. (1992) The a/b hydrolase fold. Protein Eng. 5, 197–211. 32. Jutila, A., Zhu, K., Patkar, S.A., Vind, J., Svendsen, A. & Kinnunen, P.K.J. (2000) Detergent-induced conformational changes of Humicola lanuginosa lipase studied by fluorescence spectroscopy. Biophys. J. 78, 1634–1642.

13. Derewenda, U., Swenson, L., Green, R., Wei, Y., Yamaguchi, S., Joerger, R., Haas, M.J. & Derewenda, Z.S. (1994) Current pro- gress in crystallographic studies of new lipases from filamentous fungi. Protein Eng. 7, 551–557. 33. Deems, R.A., Eaton, B.R. & Dennis, E.A. (1975) Kinetic analysis of phospholipase A2 activity toward mixed micelles and its implications for the study of lipolytic enzymes. J. Biol. Chem. 250, 9013–9020. 14. Lawson, D.M., Brzozowski, A.M., Rety, S., Verma, C. & Dodson, G.G. (1994) Probing the nature of substrate binding in