Plant non-specific lipid transfer proteins (ns-LTPs) are well known for their ability to bind and transfer lipids [1,2]. They exhibit a basic pI and a 9-kDa molecular mass ; eight cysteines all involved in disulphide bridges help in maintaining the structure of the protein. The three-dimensional structure of ns-LTP1 has been determined by 1H-NMR and crystallography and reveals a hydrophobic cavity within the protein [4–9]. Lersche et al.  were the first to report a binding constant, Kd, in the case of barley LTP1 with fatty acids, lysophosphatidylcholine and acyl-CoA. In contrast with wheat LTP1 , the affinity was very low with binding constants in the range 10−2 to 10−4 m, except for acyl-CoA where Kd = 10−6 m. In that study, the intrinsic tyrosine fluorescence was used to probe the binding of lipids. However, it will be shown herein that this method is not well suited in the case of barley because its LTP1 possesses three tyrosines, one of which is solvent exposed and is not sensitive to lipid binding. Kader and colleagues have reported binding of lipids by maize LTP1 by using displacement fluorescence methods that involve labelled lipids [11,12]. The binding of two lipid monomers was suspected from these experiments. However, it was not possible to determine any binding constants from these data [11,12]. More recently, we analysed the binding constant of wheat LTP1 complexed with several lipids as obtained by intrinsic tyrosine fluorescence . Our data confirmed the lack of specificity for fatty acids and phospholipids with various chain lengths, with Kd values of about 10−6 m. With the finding that wheat LTP1 is capable of binding two lipids [9,11,13], it becomes a new goal to determine whether it is a general feature of the LTP1 family. For this purpose, barley LTP1 is well adapted because the free protein possesses one of the smallest cavity volumes, as shown by comparing several structures of LTP1 . Moreover, several authors have proposed that these proteins could be involved in the formation of polyesters [15–17]. Then, binding of two ω-hydroxylated lipids, an abundant cutin monomer, represents an interesting task and should help in our understanding of the physiological role of this protein family.
The binding of two mono-acylated lipid monomers by plant lipid transfer proteins (LTP1s) presents an attractive field of research that could help our understanding of the functional role of this protein family. This task has been investigated in the case of barley LTP1 because it is known to exhibit a small cavity in its free state. The titration with lipids could not be followed by fluorescence with the native protein. Indeed, this LTP1 possesses a tyrosine residue on its C-terminus, Tyr91, which is not sensitive to lipid binding but mainly contributes to the fluorescence signal intensity. However, the binding of 1-myristoylglycerophosphatidylcholine (MyrGro-PCho) could be monitored by fluorescence after removal of Tyr91 by a carboxypeptidase. These experiments returned a dissociation constant of about 1 µm and showed that the protein can indeed bind two monomers. This result was corroborated by molecular modelling where the structure of the complex between barley LTP1 and MyrGro-PCho was derived from that determined in the case of wheat [Charvolin, D., Douliez, J.P., Marion, D., Cohen-addad, C. & Pebay-Peyroula, E. (1999) Eur. J. Biochem.264, 562–568.]. Results from isothermal titration calorimetry experiments indicated non-classic titration behaviour but also suggested that two lipids could be bound by the protein. Finally, barley LTP1 binds two ω-hydroxypalmitic acid, a compound found in the family of cutin monomers. The fact that the binding of two lipids could be related to the physiological role of this protein family is discussed.
isothermal titration calorimetry
lipid transfer protein
Materials and methods
Lipids and LTP1
The purification of barley LTP1 was performed according to Charvolin et al.. Lipids were obtained from Sigma and were used at concentrations varying between 4 and 6 mg·mL−1. ω-Hydroxypalmitic acid was solubilized in methanol and MyrGro-PCho in phosphate buffer (50 mm sodium phosphate buffer, pH 7). 10 mg protein was weighed exactly and solubilized in phosphate buffer (10 mL) to make the stock solution (104 µm). A twofold dilution of this stock solution was used for tyrosine intrinsic fluorescence experiments. Digestion of LTP1 with bovine pancreas carboxypeptidase A (Sigma) was performed as follows. 10 µL of the commercial enzyme solution was used with 10 mg·mL−1 LTP1 and the reaction was allowed to occur for 2 h at 40 °C. The mixture was re-purified by RP-HPLC (according to Charvolin et al.) and subjected to mass spectroscopy.
The molecular mass of the protein was obtained using a Perkin-Elmer APIII (Sciex, Thornill, Canada) triple quadrupole mass spectrometer equipped with an atmospheric pressure ionization source (Electro-spray mass spectrometer) . The sample analysis (1 mg·mL−1) was achieved by an on-line coupling between MS and RP-HPLC. Elution was carried out on an RP-HPLC column (Symmetry C18 Waters, Milford, MA, USA) at a flow rate of 0.25 mL·min−1 (40 °C) with a split to the MS ionization source , which was set at a flow rate of 0.3 mL·min−1. Ion detection was performed in positive mode and mass was calculated using biomultiview 1.2 (Software package Sciex) from scan mass to charge m/z.
Fluorescence intensity was measured at 25 °C with a Fluoromax-Spex (Jobin et Yvon, France). Excitation was set at 275 nm while emission spectra were recorded from 280 to 340 nm. 1 mL of a 50-µm ns-LTP1 was poured in the cuvette and titration was performed by adding the lipid solution in a stepwise manner. Titration curves reported maximum fluorescence intensity at 305 nm versus the molar ratio lipid/protein, Ri.
Isothermal titration calorimetry
Isothermal titration calorimetry (ITC) was performed using a MicroCal titration microcalorimeter (Northampton, MA, USA). Solutions were degassed under vacuum prior to use. Protein at a concentration of 1 mg·mL−1 was poured in the calorimeter cell and lipid (2.5 mg·mL−1) was added automatically in aliquots of 7 µL. In a blank experiment, the lipid was previously diluted three-fold and only buffer was placed in the cell. Thermogram data were integrated using the origin software supplied by MicroCal Inc.
Binding of MyrGro-PCho by native barley LTP1
In the case of wheat, tyrosine intrinsic fluorescence has been shown to be a powerful tool for studying the binding of LTP1 to various lipids . In that case, the interaction was followed by monitoring the increase of fluorescence emission upon addition of lipids. We applied this method for studying the interaction between barley LTP1 and LPC with a chain length of C14 (MyrGro-PCho). This lipid was chosen because it is known that two monomers can be loaded within the wheat LTP1 . In the case of barley LTP1, the fluorescence intensity of native protein was four times that of wheat LTP1 as shown in Fig. 1. This marked difference was due to an additional tyrosine residue (Tyr91) that barley LTP1 possesses on its C-terminus. Addition of various amounts of MyrGro-PCho did not yield any noticeable increase of fluorescence (data not shown), contrary to what was observed in the case of wheat . This meant either that binding did not occur or that the increase of fluorescence was hidden by the high contribution of Tyr91 to the total fluorescence signal. We then investigated the removal of this residue by a carboxypeptidase A. This enzyme was well suited for this purpose as it has a relatively high affinity for hydrophobic residues. Further to this enzymic treatment, MS returned a mass of 9412 Da compared with 9688 Da for the native protein, showing that the carboxypeptidase had also removed the non polar Ile90 residue of barley LTP1. For the sake of clarity, this truncated protein was named LTP189. The fluorescence spectrum of LTP189 is illustrated Fig. 1; it shows a signal intensity comparable with that of wheat LTP1.
Addition of MyrGro-PCho yielded an increase of fluorescence at saturation (1.3 times that of the original fluorescence) markedly lower than that obtained when wheat LTP1 was titrated with the same lipid . The use of a non-cooperative model  for fitting these data (Fig. 2) returned a dissociation constant, Kd, of 2 ± 0.3 µm and number of sites, n, of 1.7 ± 0.1. The affinity value (Kd) is in disagreement with the results from Lersche et al.  who reported a Kd higher than 10−2 m for various lysophosphatidylcholines. However, our result is analogous with the dissociation constant reported in the case of wheat LTP1 . As in the case of wheat , an attempt to fit our data with two independent sites did not provide accurate results. In order to validate our results, we used another technique for studying the binding of MyrGro-PCho to barley LTP1. ITC had already been used for probing the binding of lipids to proteins [20–23], and was performed with the native barley LTP1. In this case, the titration was followed by measuring the heat per second as a function of time after each injection. Upon addition of MyrGro-PCho, exothermic peaks, the intensity of which decreased with injection time, were recorded (Fig. 3). At the eighth injection (Ri ≈ 1.4), endothermic peaks appeared, forming a bell curve that vanished at the twelfth injection (Ri ≈ 2). An analogous study performed with wheat and maize LTP1s revealed the same exothermic/endothermic behaviour (unpublished results). A blank experiment in which the lipid was injected directly in buffer (data not shown) did not show any peaks. These data were integrated and an attempt to fit the curve with two independent sites failed. However, the titration behaviour supported the idea that two monomers of MyrGro-PCho could be loaded within the barley LTP1.
Molecular modelling of the MyrGro-PCho–barley LTP1 complex
Because the structures of the native barley LTP1 (Protein DataBank accession no. 1LIP.pdb)  and the wheat LTP1complexed with two MyrGro-PChos (Protein DataBank accession no. 1BWO.pdb)  are known, it appeared interesting to make use of molecular modelling to derive the structure of barley LTP1complexed with two MyrGro-PChos. Both structures were first superimposed on the basis of their backbone. When it was performed from the fifth to the 70th amino acid, that is, without including the C-terminus, the rmsd was 1.53 Å, while it was 2.38 Å from the fifth to 90th amino acid. This clearly revealed that the main difference arose from the structure of the C-terminal region which is more buried in the hydrophobic cavity in the case of barley than in wheat LTP1. This is well illustrated in Fig. 4, where both proteins are superimposed (displayed in the same orientation for the sake of clarity) and both lipids superimposed on the native barley LTP1. Close contacts between lipids and the protein were observed essentially in the region of the C-terminus.
A template procedure was then used; both lipids were re-introduced in the barley protein on the basis of the structure of wheat LTP1 (1BWO), and a minimization performed without any constraints. The result returned an interaction energy between lipids and protein similar to obtained with wheat and barley. This indicated that, on the basis of the structure of wheat LTP1, 1BWO, barley LTP1 can also bind two monomers. Tyr79 was also shown to make a hydrogen bond with the glycerol part of monomer A, and this has been suggested to generate the reported enhancement of fluorescence (see above, and Lullien-Pellerin et al.).
Binding of ω-hydroxypalmitic acid
As ω-hydroxypalmitic acid is well represented in the family of cutin monomers , the study of complex formation with LTPs is of considerable importance. Because this fatty acid is insoluble in buffer, the experiment could only be performed using fluorescence. The fatty acid was solubilized in methanol and LTP89 was used for the titration. As in the case of MyrGro-PCho, it generated an increase of fluorescence (not shown) and the fitting procedure returned a Kd of 1 ± 0.2 µm and n = 2.1 ± 0.1. Once again, these results are comparable with those obtained in the case of wheat LTP1 and the same lipid. Interestingly, the number of sites is closer to two for this fatty acid compared with MyrGro-PCho.
In the case of wheat LTP1, intrinsic tyrosine fluorescence is well suited for monitoring the binding of ligands. This protein possesses two tyrosine residues and only one, Tyr79, is suspected to be involved in the signal enhancement upon titration . Barley LTP1 has an additional tyrosine residue in its sequence at the C-terminal position, Tyr91. As revealed by the three-dimensional structure , Tyr91 is solvent exposed so that its fluorescence emission is not partially quenched as is the case for the other tyrosine residues (Tyr16 and Tyr79), which are embedded within the core of the protein. As a consequence, Tyr91 mainly contributes to the total fluorescence emission of barley LTP1, which is markedly higher than that of wheat LTP1 (Fig. 1) which lacks this terminal tyrosine residue. Moreover, Tyr91 hides the Tyr79 enhancement of fluorescence that is generated on lipid binding so that titration cannot be performed under these conditions. This could explain the reason for the poor affinity reported in the case of complex formation of barley LTP1 with various lipids . However, the removal of Tyr91 can be easily achieved by using a carboxypeptidase A even if it also hydrolyses Ile90. It should be noted that the structure of this truncated protein is not too affected by this procedure. Indeed, despite the carboxyl terminal has been shown to interact with Arg44 and contributes to the stability of the structure , this salt bridge does not always occur (see 1BE2.pdb ). In contrast with the native barley LTP1, this truncated protein can be used for monitoring the binding of lipids. Titration gives rise to an increase of fluorescence as obtained in the case of wheat LTP1. Our result shows an affinity comparable with that obtained for the complex formation between wheat LTP1 and various lipids, indicating that the interaction seems to be rather non-specific as long as the binding ligand possesses a hydrophobic tail. These similarities are consistent with the results of structural studies that have shown that the presence of the lipid polar head does not contribute significantly to the complex stability [7,9].
Of considerable interest is the finding that barley LTP1 is also capable of binding two lipids. This result is well illustrated by molecular modelling, which shows that both lipids can fit within the cavity. It is clearly shown that the loading of two monomers is almost only associated with a shift of the C-terminus. Moreover, it is known that this region is flexible as revealed by temperature factors from crystallography .
The fact that the value returned by fitting the fluorescence data (n = 1.7) remains lower than two indicates that both lipids do not contribute in the same way to the increase of fluorescence. In another way, as in the case of wheat LTP1, it is probable that both sites exhibit close but not identical affinity . The binding of two monomers appears rather surprising if considering the structure of the complex with palmitic acid or palmitoyl-CoA [7,26]. In that case, it was clearly shown that only one monomer of lipid binds within the cavity. Such a result was also reported in the case of maize with one palmitic acid  or wheat with one MyrGro-PCho (F. Vovelle, CBM, Orleans, France, personal communication), while it is only recently that a structure of LTP1 complexed with two lipids has been published . The reason why some studies have shown that LTP1 binds one monomer while others indicate that two can be bound remains obscure. It is tempting to speculate that binding of two monomers could strongly depend on the experimental conditions such as pH, buffers, temperature or lipid and protein concentrations.
As perturbation of the tyrosine fluorescence can be used to monitor the binding of ligands to LTP1, ITC also represents a promising technique for such studies. The binding experiment of LTP1 with MyrGro-PCho reveals non-classic behaviour with an endothermic event at the end of the titration. This apparently results from the superimposition of exothermic and endothermic behaviours. However, this cannot be fitted with the binding model so information is unavailable about the individual sites.
In the hypothesis of the participation of LTP1 in the biosynthesis of cutin [15,17], a binding study involving ω-hydroxypalmitic acid represents an attractive investigation. The affinity of this lipid for the barley LTP1 is analogous with that reported in the case of wheat . Moreover, in the same way, n is closer to two compared with values obtained with other lipids. This suggests that in this case, both sites exhibit similar affinity. This result is of strong importance if considering the head-to-tail orientation of both monomers within the protein cavity. An analogy with the structure derived from that of wheat suggests that the ω-hydroxyl group of monomer B would be located close to the carboxyl group of monomer A.
We have clearly shown that barley LTP1, which possesses a small hydrophobic cavity in its free state, is capable of binding two monomers of MyrGro-PCho. This behaviour probably depends on the biochemical conditions. This result, together with the binding of a hydroxylated fatty acid, strongly emphasizes the participation of LTP1 in cutin biosynthesis.