Diversity in the structures and ligand-binding sites of nematode fatty acid and retinol-binding proteins revealed by Na-FAR-1 from Necator americanus

Necator americanus fatty acid and retinol-binding protein-1 (Na-FAR-1) is an abundantly expressed FAR from a parasitic hookworm. The present work describes its tissue distribution, structure and ligand-binding characteristics and shows that Na-FAR-1 expands to transport multiple FA molecules in its internal cavity.


INTRODUCTION
Lipids such as acids (FAs) and retinoids are relatively insoluble in water, can be susceptible to oxidation and are potentially damaging to membranes in their free form. Consequently, they are usually transported within proteins or protein-mediated lipid aggregates. Examples of vertebrate FA and retinoid transporter proteins include serum albumins (∼64 kDa), which bind a range of compounds, lipocalins (∼20 kDa), which are found in many secretions and the proteins of the cytoplasmic FA-binding/cellular retinol-binding/cellular retinoic acid-binding protein (FABP/CRBP/CRABP) family (∼14 kDa) that are confined to the cytoplasm, where they may bind FAs, retinol or retinoic acid, depending on the isoform. These proteins are variously involved in transport and storage of lipids and also in the delivery of small signalling lipids to their destinations [1].
The structures of Na-FAR-1 determined by NMR and X-ray have been deposited in the PDB under accession codes 4UET and 4XCP respectively.
isoforms of ∼20 kDa, eight having been found encoded within the genome of the free-living species Caenorhabditis elegans, each of which binds FAs and retinol to varying extents [10]. There are as yet uncharted numbers of FAR isoforms in parasitic nematodes, but FARs have drawn attention because they are secreted by both plant-and animal-parasitic species and their encoding transcripts are relatively abundant [11][12][13][14][15][16][17][18]. They have also proven useful for serodiagnosis, have shown promise in experimental vaccines [19,20] and have been proposed to facilitate infection by manipulating host lipid-mediated defences [21,22]. At least one FAR has been shown to bind an anthelmintic drug, these drugs being typically hydrophobic and so may require carrier proteins to conduct them to their site of action within parasites [23]. The definitive role of FARs in parasitism is not known, but their presence in secretions of the worms, coupled with their ligand-binding propensities, suggests roles for them either in acquiring lipids from the hosts, or in delivering or in sequestering signalling lipids and so modifying the tissues the parasites occupy or the immune responses against them. The structure of one FAR protein, C. elegans FAR-7 (Ce-FAR-7), has been solved by X-ray crystallography, revealing a helix-rich structure that is unlike any type of lipid-binding protein previously described [24]. The mRNA for Ce-FAR-7 does not encode a secretory signal peptide and its amino acid sequence indicates that it is in a different subfamily of FARs from those that are secreted from the synthesizing cell. We set out to confirm the expression pattern of a secreted FAR protein from a parasite and characterize its structure and ligandbinding characteristics. The protein Na-FAR-1 derives from the blood feeding intestinal hookworm of humans, N. americanus. This parasite and the other hookworm of humans, Ancylostoma duodenale, together infect over 300 million people worldwide [25,26], causing considerable morbidity, together with adverse social and economic consequences. We have determined the structure of Na-FAR-1 by both X-ray crystallography and NMR spectroscopy in solution. We find that it has a similar overall fold to Ce-FAR-7, indicating that FARs are structurally conserved despite considerable sequence diversity. But, in addition to the structural differences between Ce-FAR-7 and Na-FAR-1, we find that their ligand-binding sites differ significantly in position and form.

Protein expression and purification
Recombinant Na-FAR-1 (rNa-FAR-1) was expressed in BL21 (λDE3) Escherichia coli cells as described [27]. For native crystallographic studies, Na-FAR-1 was purified to homogeneity, as previously described [28], from cells grown in LB media. Selenomethionine-labelled protein was purified from B834 cells grown in M9 minimal medium supplemented with a cocktail of free amino acids (each 0.5 g·l − 1 ) and selenomethionine (50 mg·l − 1 ; Generon).
For NMR studies, samples of unlabelled, 15 N-labelled and 13 C 15 N-labelled protein were purified by nickel-affinity, size exclusion and reverse-phase chromatographies, as described [27], from cells grown in M9 minimal medium containing 15 NH 4 Cl, [ 13 C 6 ]-glucose or their unlabelled equivalents.

Western blotting and immunolocalization of Na-FAR-1
Antiserum prepared against recombinant Na-FAR-1 was raised in three rabbits by subcutaneous injection with 0.7 mg of purified recombinant Na-FAR-1 in Freund's complete adjuvant.
Antiserum was tested by ELISA and Western blot analysis against the recombinant protein. To analyse the expression of Na-FAR-1 in the worm, soluble extracts of adult N. americanus, as well as recombinant proteins (each 100 ng) of homologues from Ancylostoma caninum FAR-1 (Ac-FAR-1); Brugia malayi FARs (Bm-FAR-1 and Bm-FAR-2) and the unrelated protein, recombinant Ac-SPI (serine protease inhibitor from A. caninum) were separated on a 4 %-20 % gradient SDS precast polyacrylamide gel (Invitrogen) and subsequently electrotransferred on to a PVDF membrane (Millipore). Rabbit anti-Na-FAR-1 serum was diluted 1:5000 into PBS (phosphate buffered saline), pH 7.4, with 0.05 % Tween-20 and incubated with transferred membrane for 2 h. Horseradish peroxidaseconjugated goat anti rabbit-IgG was used as secondary antibody and ECL was used to develop the reaction (GE).
To determine the tissue-specific localization of Na-FAR-1, adult N. americanus worms were prepared as previously described [29]. Briefly, adult worms were collected from the intestines of hamsters infected with N. americanus L3 [third (infective) larval stage of a nematode] for 45 days and fixed with 10 % formalin. The fixed worms were sectioned and mounted on glass slides. The non-specific binding sites on worm sections were blocked with 5 % FBS in PBS for 1 h. The rabbit anti-Na-FAR-1 serum was applied (1:500 dilution) to each tissue section and incubated for 2 h at room temperature in a humidified chamber. Pre-immune rabbit serum at the same dilution was used as a negative control. Sections were washed six times for 5 min each in PBS and probed with anti-rabbit Cy3-conjugated IgG (Rockland). Sections were viewed under a Nikon TE-2000 Inverted fluorescence microscope using a 550 nm excitation filter block and emission at 565 nm.

Crystallization, data collection, processing and structure solution
We have shown previously that Na-FAR-1 crystallizes in two crystal forms, one of which (form 2) shows significant twinning [28]. Here, in order to obtain phasing information, selenomethionine-substituted protein was purified and crystallized, selecting only the cubic crystal form 1. Crystals were frozen in a stream of cool nitrogen gas (100 K) and brought to the Diamond Light Source, station I04 (DLS) for X-ray diffraction data collection.
Data were collected at 0.7 • increments per image, for a total of 200 images [wavelength 0.9793 Å (1 Å = 0.1 nm)] and processed by the automatic processing routines fast_dp, which utilized XDS [30], POINTLESS and SCALA [31]. The structure was solved using the SAS protocol of Auto-Rickshaw [32]. The input diffraction data were prepared and converted for use in Auto-Rickshaw, using programs of the CCP4 suite [33]. F A values were calculated using the program SHELXC [34]. Based on an initial analysis of the data, the maximum resolution for sub-structure determination and initial phase calculation was set to 2.14 Å based on the scaling statistics and the increase in R meas in the highest resolution bin. All the four heavy atoms requested were found using the program SHELXD [35]. The correct handedness for the substructure was determined using the programs ABS [36] and SHELXE [34]. Initial phases were calculated after density modification using the program SHELXE. 83.9 % of the model was built using the program ARP/wARP [37]. Despite a solvent content above 70 %, only one copy of the protein was observed in the asymmetric unit. The model was completed by hand using COOT [38] and iterative rounds of BUSTER [39]. Ligands and water were added using COOT. Based on its prevalence in the lipids co-purifying with recombinant Na-FAR-1, palmitate was fitted into unoccupied electron density within the binding pocket.
Further electron density that may represent additional ligand molecules was observed, but as the density was incomplete and not at full occupancy, it was left unmodelled. The geometry of the finished models was validated using Molprobity [40].

Preparation of NMR samples
Apo-Na-FAR-1 samples were delipidated by reverse phase-HPLC (high performance liquid chromatography) as previously described [27] using a C 8 -silica stationary phase and a water/acetonitrile gradient in the presence of 0.1 % trifluoroacetic acid as the mobile phase. Samples were lyophilized, reconstituted and concentrated to approximately 0.6 mM in 20 mM sodium phosphate, pH 7.20. 2 H 2 O was added to a final concentration of 5 % (v/v). The sample used for residual dipolar coupling (RDC) measurements was prepared by partial alignment of the uniformly 13 C 15 N-enriched Na-FAR-1 in a solution of magnetically aligned filamentous Pf1 bacteriophage [41] (ASLA biotech). The degree of alignment was evaluated by measuring the 2 H quadrapolar splitting in the HDO resonance. After testing several conditions, a NaCl concentration of 300 mM with 9 mg·ml − 1 of bacteriophage and 300 μM protein was selected.

NMR spectroscopy and analysis
All spectra were recorded at 311 K on a Bruker AVANCE 600 MHz spectrometer equipped with TCI cryoprobe. Resonances were assigned, as described previously [27]. All spectra were processed in AZARA (Wayne Boucher, http://www.bio.cam.ac.uk/azara). Maximum entropy reconstruction [42] was used to enhance resolution of the indirect dimensions of 3D experiments. Spectra were analysed with CCPNmr analysis software [43]. Frequency-based methods were employed to measure 1 D NH [44] and 1 D CαHα [45] couplings from in-phase/antiphase 15 N-HSQC (heteronuclear single-quantum correlation spectroscopy) spectra. Distance restraints for structure calculations were derived from 3D 15 N-NOESY-HSQC [46] and 3D 13 C-NOESY-HSQC spectra, each recorded with 100 ms mixing time.

Structure calculations from NMR-based restraints
Distance restraints were derived from NOESY (nuclear Overhauser effect spectroscopy) cross-peaks with the initial mapping from normalized intensity to distance and grouped in distance bins. NOE distance restraints were incorporated in restrained MD calculations using the ambiguous distance restraints formalism [47] using ARIA 2.3 [48] and CNS [49]. Loose backbone dihedral restraints for regions of regular secondary structure predicted based on secondary chemical shifts by DANGLE [50] were incorporated during the high temperature phases of the simulations but omitted during the final cooling phase. RDC and hydrogen bond restraints were then introduced. The average RDC alignment tensor was estimated from the ensemble calculated using only NOEs with PALES [51] and used to incorporate the RDC restraints via the SANI potential [52] in square-well mode. The 20 structures that best satisfy the experimental restraints were chosen from 100 structures generated in the final iteration and refined in explicit water [53]. The quality of these structures was analysed using PROCHECK_nmr [54] and their co-ordinates deposited in the Protein Data Bank under accession code 4UET. Structure figures were generated using PyMOL (http://www.PyMOL.org). 15 N-relaxation time constants, T 1 and T 2 were assessed using the method of Kay [55][56][57] at a field strength of 14.1 T. Relaxation delays for assessment of T 1 were 101, 601, 1001 and 1401 ms whereas those for T 2 were 17, 34, 68, 102 and 136 ms. Selected time points in each series were repeated in order to estimate the inherent error in calculation of cross-peak intensities. Relaxation times T 1 and T 2 were calculated using non-linear least squares fitting. Collection of 15 N-HSQC-heteronuclear NOE experiments with and without saturation allowed extraction of 1 H, 15 N NOE values. Both saturation and reference experiments were repeated for the purpose of error estimation.

Ligand-binding analysis by chemical shift perturbation
In order to examine Na-FAR-1's ligand binding properties by NMR, the chemical shift changes induced in 15 N HSQC spectra of the protein by the presence of increasing amounts of ligand were analysed. Double-labelled 13 C-15 N recombinant Na-FAR-1 (0.4 mM) was titrated by sequential addition of small volumes of unlabelled sodium oleate (Sigma-Aldrich) from a 125 mM stock solution in water (pH ∼9). Triple resonance experiments were recorded at selected points of the titration in order to assign the displaced cross-peaks. The titration started by addition of 0.5 molar equivalent of the ligand and was continued until turbidity due to the presence of insoluble oleic acid was observed.

Lipid extraction and analysis
Total lipids were extracted according to the methodology described by [58] and [59], with minor modifications. For unstripped Na-FAR-1 (no RP-HPLC purification), lipids were extracted from 15 mg of protein and compared with control apo-Na-FAR-1 (RP-HPLC purified protein). For comparison, approximately 3 ml of culture of E. coli BL21 (λDE3) cells were lysed by sonication. Each sample was mixed with 15 ml of CHCl 3 -CH 3 OH (2:1) and vigorously shaken for 15 min in an ice bath. The homogenate was washed with 250 μl of 2.9% (w/v) NaCl solution. After agitation, the phases were separated by centrifugation and the upper, aqueous phase discarded. The lower phase containing lipids was recovered and dried under a stream of N 2 gas, re-dissolved in CHCl 3 and stored at -20 • C under N 2 gas until analysis.
Non-esterified FA and phospholipid (PL) TLC spots were scraped and extracted from the silica with pure chloroform for FAs and chloroform-methanol-water for PLs respectively. FA composition was analysed by GC (gas chromatography)-MS of their methyl ester derivatives, prepared with BF 3 -methanol according to the method of Morrison and Smith [60] as described previously [61]. The individual FA methyl ester peaks were identified by comparison of their retention times with those of standards and by their mass spectra.

Fluorescence spectroscopy
Fluorescence experiments were performed with a Fluorolog-3 Spectrofluorometer (Horiba-Jobin Yvon). Buffer alone was used to correct for Raman and background scattering. RP-HPLC delipidated Na-FAR-1 was employed in all ligand-binding experiments.
The ligand-binding capacity of Na-FAR-1 was investigated with the fluorescent ligands 11-(dansylamino)undecanoic acid (DAUDA) and retinol. Stock solutions (10 mM) were prepared in ethanol and then diluted in PBS for use in the assays. Retinol solutions were diluted in ethanol and added directly to the cuvette to minimize degradation.
The FA chain length preference of rNa-FAR-1 was tested by displacement of the fluorescent ligand DAUDA as described [13]. Binding of non-fluorescent ligands was detected by a reversal of the wavelength shift and a decrease in fluorescence emission intensity on equal additions of test ligands to a DAUDA-rNa-FAR-1 complex recorded at the peak fluorescence emission wavelength of DAUDA in the protein (470 nm). The concentration of Na-FAR-1 in the cuvette was 1.5 μM. DAUDA ethanol stock solution was diluted 1:10 000 in PBS for use in the assays at 1 μM. Stock solutions of all the non-fluorescent competitors were made to approximately 10 mM in ethanol, then diluted in PBS for use in the assays.

Na-FAR-1
Na-FAR-1 had been identified in a gene survey of the N. americanus transcriptome under the name N. americanus LBP-20 (sequence ID NAC00128) [62]. LBP-20 proteins are currently named FAR due to their capacity for binding FAs and retinol and Na-FAR-1 was renamed accordingly [63]. More recently, the genome of N. americanus has been published and at least six FAR proteins have been recognized (Figure 1; Supplementary Figure S1), including Na-FAR-1, under the sequence name NECAME_14208 [64]. Na-FAR-1 cDNA predicts a 19364.57 Da protein with a 14 amino acid secretion signal peptide predicted by SignalP [65]. The post-translational removal of the leader peptide would yield a mature protein of 155 amino acids with a molecular mass of 17082.49 Da.

Na-FAR-1 localization within the parasite
Na-FAR-1-encoding mRNA had been detected in adult and larval L4 N. americanus life stages [62] and analysis of differences in gene expression between infective larvae (iL3) and adult parasitic stages shows that Na-FAR-1 is mainly expressed in the adult, blood-feeding stage [64]. Immunohistochemistry revealed that Na-FAR-1 protein occurs in the cloacal aperture and copulatory bursa of male worms (Figures 1c and 1d) and in the intestine (Figures 1f and 1g). The presence of Na-FAR-1 in the parasite's intestinal cells may indicate involvement in transportation and storage of lipids derived from host blood and its location in copulatory bursa of male worms suggests its function may additionally be related to reproduction. Na-FAR-1 was detected in the intestinal cells but not the reproductive structures of females ( Figure 1e).

3D structure of Na-FAR-1
The 3D structure of Na-FAR-1 was determined by both protein X-ray crystallography and solution state NMR spectroscopy. Na-FAR-1, which had not been subjected to reverse phase HPLC purification to strip out co-purifying ligands, crystallized as previously reported and diffracted to 2.14 Å [28]. Molecular replacement using the Ce-FAR-7 structure proved unsuccessful and we therefore obtained the crystal structure of Na-FAR-1 in complex with co-purifying ligands, hereafter referred to as holo-Na-FAR-1, by anomalous dispersion from crystals of Se-Met-labelled protein (   Figure S1 for the multiple sequence alignment from which the tree was constructed. (c-g) Localization of Na-FAR-1 within adult male (c and d) and female (e-g) worms.
Indirect immunofluorescence localization with rabbit anti-Na-FAR-1 serum stains the intestinal cells of adult N. americanus worms. Na-FAR-1 was also detected on the copulatory bursa and cloacal aperture of male worms. (h) Control carried out using pre-immune serum. Scale bars represent 100 μm. 76 1 D CaHa ) observed in a sample that had been partially aligned in Pf1 filamentous bacteriophage ( Table 2). The overall folds of the Na-FAR-1 structures in complex with ligands (holo) and without ligand (apo) are similar (Figure 2a-c). Superposition of the holo and apo structures gives a 1.814 Å co-ordinate RMSD (root-mean-square deviation) for all heavy atoms (1.579 Å for main chain heavy atoms only). Na-FAR-1 presents a wedge-shaped structure with two larger faces each of approximately 40 by 30 Å in area and of ∼17 Å in width at the wide end of the wedge. The fold is organized into 11 helices of various lengths, defining an internal cavity (Figure 3). The N-terminal helices, 3 10 (residues 3-5), α2 (8-11) and α3 (16)(17)(18)(19)(20)(21)(22)(23)(24) are co-planar with the C-terminal helices α9 (126-135), α10 (137-144) and α11 (147-152) and together with one of the longest helices, α6 (57-75), form one large face of the wedge. The other two long helices, α7 (79-99) and α8 (107-122), are co-planar with α5 (45)(46)(47)(48)(49)(50)(51)(52)(53)(54)(55) and form the other large face. These two faces enclose an internal cavity that is closed at the thick end of the wedge by α4 (29)(30)(31)(32)(33)(34)(35)(36)(37), which is almost perpendicular to the long axis of the molecule. Hydrophobic residues are located almost exclusively on the inward-facing sides of the helices with their side chains pointing towards the internal cavity. Most of the polar and charged residues are on the external surface of the protein generating a predominately hydrophilic surface. The notable exceptions are Ser 88 , Lys 96 and Tyr 100 whose side chains are located within the cavity, whereas His 67 and Arg 93 stand on either sides of the largest opening to the cavity. Despite their similarities, the structures of the apo and holo forms of Na-FAR-1 reveal significant differences that reflect structural alterations upon ligand binding. The internal cavity is much larger for the holo protein, achieved by a global outward displacement of the surrounding helices by a little over 1 Å compared with the apo form, with the biggest changes seen at the C-terminal end of helix α7 (Figure 2a). For the apo structure, the central cavity has a volume of approximately 1220 Å 3 accessible to a 1.4 Å radius probe (equivalent to a water molecule) or 940 Å 3 accessible to a 1.925 Å probe (equivalent to a CH 2 group). In the holo form, the cavity is more than twice as big, reaching 2570 and 2170 Å 3 for probes of 1.4 and 1.925 Å respectively.
The main part of the apo-Na-FAR-1 cavity is lined by the side chains of residues Leu 13 , Met 14  Val 117 and Phe 143 . The cavity has a single opening located on one side of the wedge between α6 and α7 and the α4-α5 loop that is surrounded by charged residues. In the holo structure the expanded cavity is lined by the same residues as in the apo structure, but with 21 additional, predominantly apolar, side chains contributing to the surface of the expanded cavity (Phe 1 , Ile 6 , Leu 25  ). In addition to the opening seen in the apo form, the cavity is also accessible to solvent via an opening between the α2-α3 and α9-α10 turns and α6. It is therefore possible that more than one region could be involved in ligand entrance or exit and the loop between α7 and α8, which exhibits a higher mobility than the average, could be a candidate region to participate in those processes. Within the cavity of the holo form, clear electron density was observed that is well fitted by a single FA molecule, containing 16 carbon atoms (Figure 2d). Since the predominant FA found in E. coli lipids is palmitate, we fitted this to the electron density. The palmitate appears to be curled into a C-shape around the side chain of Lys 96 and contacts 11 other residues (Phe 1 , Ile 6 , Met 14 , Phe 21 , Leu 22 , Leu 25 , Leu 33 , Leu 66 , Ile 95 , Tyr 99 and Tyr 100 ). Additional electron density, possibly indicative of several other bound ligand molecules, is seen in the cavity of Na-FAR-1, but in light of the heterogeneous mixture of lipids associated with the protein and the poor quality of the density, we have not modelled them.

Comparison of the structures of Na-FAR-1 and Ce-FAR-7
Ce-FAR-7 has previously been noted to lie in a distinct sub-group of FARs because it possesses two cysteines that may form a disulfide bond, has distinct lipid-binding characteristics, lacks a secretory leader peptide and because of the intron arrangement of its encoding gene [10,24]. The X-ray crystallographic structure of Ce-FAR-7 [24] displays similarities but several distinct differences to Na-FAR-1 that could relate to differences in their ligand binding and biological properties.
A structure-based sequence alignment (Figure 3a) reveals sequence identity between the two proteins at 32 positions (∼21 %). Among these are: hydrophobic residues, Leu 77 and Ala 81 (Leu 78 and Ala 82 in Ce-FAR-7) that help to determine the angle between α6 and α7; Leu 33 , Leu 59 (Leu 33 and Leu 60 in Ce-FAR-7) together with Leu 52 , which is replaced by another hydrophobic residue Val 53 , which are involved in maintaining the orientation between α6 and helices α4 and α5; a salt bridge between Glu 35 and Lys 55 (Glu 35 and Lys 56 in Ce-FAR-7) that helps to maintain the relative orientation of α4 and α5; and another salt bridge between Lys 73 (Arg 74 Ce-FAR-7) and Glu 17 that helps to maintain the relative orientation of α6 and α3. Na-FAR-1 has an insertion of four residues (107-110) that constitute an additional turn of helix at the N-terminal end of α8. Conserved prolines Pro 7 , Pro 15 and Pro 79 initiate helices α2, α3 and α7, corresponding to Pro 7 , Pro 15 and Pro 80 in Ce-FAR-7, as they seem likely to, in other FARs. Ce-FAR-7 has an additional residue in the loop between helices α4 and α5 and it is 14 residues shorter at the C-terminal region, thereby lacking helices α10 and α11. In Ce-FAR-7 α1 comprises a full turn of α-helix, whereas Na-FAR-1 has a 3 10 helix at this position. A comparison of the structures gives an RMSD of 1.711 Å (for a total of 636 atoms aligned) with the apo form and 1.646 Å (608 atoms aligned) with the holo form. The region comprising the α4-α5 joining loop, which might constitute part of the entrance to the cavity, was not modelled in the structure of Ce-FAR-7 due to insufficient electron density. It is likely that the lack of good definition is indicative of dynamic processes occurring in the α4-α5 loop, as we found by NMR for Na-FAR-1 (Supplementary Figure S2). The estimated cavity for Ce-FAR-7 calculated with CASTp has a much smaller size than for both forms of Na-FAR-1, with a volume of 687.7 Å 3 accessible to a probe equivalent to a water molecule and just to 390.0 Å 3 for a probe equivalent to a CH 2 group. In Ce-FAR-7, helices α2 and α3 approach α7 and α8 more closely, reducing the volume of the cavity and resulting in two distinct hydrophobic pockets. The majority of residues that line the Na-FAR-1 cavity and the Ce-FAR-7 P1 and P2 pockets are conserved or conservatively substituted. The exceptions are that the Na-FAR-1 cavity includes the side chains of more polar amino acids, such as Ser 88 , Arg 93 and Lys 96 , which are found in positions occupied by Leu 89 , Thr 94 and Leu 97 respectively, in Ce-FAR-7.

Ligand binding
In fluorescence-based ligand-binding assays, Na-FAR-1 bound the fluorescent FA analogue DAUDA and the naturally fluorescent lipid retinol (Figures 4d and 4e). The degree of blue shift in DAUDA fluorescence emission (from 543 nm in buffer to 480 nm) indicates Na-FAR-1 has a highly apolar binding site, as shown for other members of the FAR protein family [10]. It is also interesting to note that Ce-FAR-7 has shown no binding capacity for DAUDA [10] and binding to retinol showed no saturation [24]. The preference of Na-FAR-1 for FAs was investigated by the addition of FAs with different chain lengths to preformed protein-DAUDA complexes (Supplementary Figure S3), showing that DAUDA displacement was greatest with saturated FAs in the range C14:0-C19:0.
In order to determine the lipid classes bound to Na-FAR-1 in a cellular environment, a Folch extraction of the lipidic fraction of Na-FAR-1 purified from E. coli and subsequent TLC analyses were performed. The bacterial cytoplasm is not the natural environment of Na-FAR-1, but this approach could contribute to the understanding of its binding preferences. The lipid fraction bound to Na-FAR-1 exhibits heterogeneous lipid content (Figure 4a). Previous studies have shown that other members of the FARs bind retinol and FAs [10,13,15,19,21]. In addition, the present analysis reveals that the protein binds not only to FAs but also a broader range of lipid classes such as PLs.
The qualitative analysis conducted suggests the presence of polar lipids such as phosphatidylethanolamine (PE) and phosphatidylglycerol (PG). Esterification and analysis of the non-esterified FA and PL-derived FA methyl esters by GC-MS (Figures 4b and 4c) showed that in the lipids carried by Na-FAR-1, palmitate (16:0) was the most common FA. Examination of total E. coli lipids showed a similar FA composition.
Protein-ligand interactions were further investigated by titration of Na-FAR-1 with sodium oleate followed using NMR. The first step of the titration was made by adding of 0.5 molar equivalents of sodium oleate ( Figure 5), at which point two cross-peaks originating from each shifted amide group were detected, one remaining at the original shift that corresponds to the apo protein conformation and another cross-peak at a different shift resulting from the protein conformation in the presence of one ligand bound. At subsequent stages in the titration, further chemical shift changes were observed, in each case resulting in the appearance of a new cross-peak and disappearance of the original. The ligand binding process therefore exhibited slow exchange behaviour through the addition of 1, 2 and 3 molar equivalents of oleate, which would suggest that the protein binds three ligands with high affinity. Beyond this, additional chemical shift changes were observed in the fast exchange regime from a 1:4 protein-ligand ratio where only one cross peak was seen for each amide group. The titration was carried on to a maximum protein-ligand ratio of 1:10 where it was limited by the precipitation of excess, unbound ligand. These results suggest that the protein binds up to at least four oleate molecules, three of them with high affinity in the sub micromolar range and a fourth with lower affinity. Substantial chemical shift changes were observed for residues throughout the protein, consistent both with ligand binding affecting the local environment of residues within the helices surrounding the cavity and with the changes in conformation of residues in the loops required to allow the structure to expand.

CONCLUSION
The FAR proteins are unique to nematodes and we in the present study demonstrate the first structure of a member of this family from a parasitic species. FARs are present in the secretory products of parasites, which is apparent both for those that reside in the lumen of the gut and those that penetrate and reside in the host tissue of animals and plants [22,67]. This family of proteins comprises subfamilies that differ in their ligand-binding characteristics and in whether or not they are secreted from   the cells that synthesize them. Transcripts encoding Na-FAR-1 are relatively abundant in the parasite from which it derives and the encoded polypeptide bears a secretory signal sequence. It falls into a new subfamily of the FARs that has not been found so far in any other group of nematodes, although the inventory of FARs is only comprehensive for those species for which genomic information is essentially complete and annotated (e.g. C. elegans). It has been noted that all FAR protein sequences exhibit a consensus casein kinase II-type sequence that occurs in approximately the same place, despite the considerable degree of sequence diversity among FARs [19,24,67,68]. This argues that this site may have some significance in, for instance, altering the ligand-binding activity of the proteins or modifying interaction with other cellular factors such as cell surface receptors (host or parasite). There is currently no evidence that this site is phosphorylated naturally, though experiments with Ce-FAR-7 indicate that modification of this site does alter ligand binding. The previous study found that direct phosphorylation rendered the protein unstable in some unknown way, but when phosphorylation was mimicked by substitution of the presumed target threonine by an aspartic acid, the affinity of Ce-FAR-7 for retinol increased, number of oleate bound Figure 5 Titration of Na-FAR-1 with sodium oleate followed by NMR spectroscopy 15 N HSQC spectra of 0.4 mM Na-FAR-1 with successive additions of sodium oleate. The spectrum of the apo protein is shown in dark blue, with spectra coloured through purple to red at the final protein-ligand ratio of 1:10. A subset of the backbone and side chain (lettered) assignments are shown and the trajectories of selected residues indicated with arrows. The inset shows an expanded view of the cross-peaks from Gly 78 backbone amide with our interpretation of the protein-ligand stoichiometry responsible for each distinct cross-peak position.
although there was no change for FA. In both Ce-FAR-7 and Na-FAR-1, the amino acid side chain thought to be phosphorylated lies in a short unstructured region between two helices at one end of the molecule. In neither protein is the side chain of this residue oriented towards the internal cavity and nor in Ce-FAR-7 it is more distant from the internal surface than in Na-FAR-1. It is therefore not immediately clear how phosphorylation could affect the proteins' ligand-binding characteristics unless structural changes result.
The differences in the size and shape of the internal cavities of Ce-FAR-7 and Na-FAR-1 are probably indicative of differences in their ligand selectivity. Both proteins bind retinol and FAs, but whether these compounds are relevant in vivo and whether other classes of ligand (such as PLs here found to be bound by Na-FAR-1) are important, remains to be established. The fact that FARs are commonly found in the secretions of parasitic nematodes bolsters confidence that they play an important role in parasitism. Parasites need to acquire nutrients from their hosts, but also need to defend themselves against immune defence reactions and some parasites induce modifications to the tissues they occupy. If FARs do bind signalling lipids, then it is conceivable that they are released by parasites in order to subvert or modify the hosts' tissue and defence reactions. If so, then FARs would seem to be logical targets for vaccines, particularly against parasites that are embedded in host tissues. Development of drugs against them would need to take into account the possibility of interference with the hosts' proteins and enzymes that transport or modify FAs and retinoids. But, it is entirely possible that we have yet to discover the true range of the ligand-binding propensities of FARs in nematodes. An understanding of their structures and how they vary between those secreted by different species of parasites could illuminate their roles in parasitism and suggest possible targets for therapeutic interventions.