Solution Structure of Human Pex5·Pex14·PTS1 Protein Complexes Obtained by Small Angle X-ray Scattering

The Pex5p receptor recognizes newly synthesized peroxisomal matrix proteins which have a C-terminal peroxisomal targeting signal to the peroxisome. After docking to protein complexes on the membrane, these proteins are translocated across the membrane. The docking mechanism remains unclear, as no structural data on the multicomponent docking complex are available. As the interaction of the cargo-loaded Pex5p receptor and the peroxisomal membrane protein Pex14p is the essential primary docking step, we have investigated the solution structure of these complexes by small angle x-ray scattering and static light scattering. Titration studies yielded a 1:6 stoichiometry for the Pex5p·Pex14p complex, and low resolution structural models were reconstructed from the x-ray scattering data. The free full-length human Pex5p is monomeric in solution, with an elongated, partially unfolded N-terminal domain. The model of the complex reveals that the N terminus of Pex5p remains extended in the presence of cargo and Pex14p, the latter proteins being significantly intermingled with the Pex5p moiety. These results suggest that the extended structure of Pex5p may play a role in interactions with other substrates such as lipids and membrane proteins during the formation of functional multiprotein complexes.

of the Pex5p/cargo interaction may also apply to the PTS2 cargo recognition of the Pex7p receptor (5).
The next step of PTS-protein import, docking of the cargo loaded receptor to the translocon, involves the peroxisomal protein Pex14p (6). Multiple Pex14p binding sites with diaromatic pentapeptide motifs (WXXX(F/Y)) were shown to be present in the N terminus of Pex5p (7)(8)(9). The number of these motifs, however, varies among species. The human Pex5p receptor, which has been investigated in this contribution, has a total of seven motifs. A recent NMR structure of the N-terminal domain of Pex14p and the first WXXX(F/Y) motif of Pex5p reveals an ␣-helical conformation of the motif (10). Interactions between Pex5p and other proteins and by their association with the peroxisomal membrane possibly lead to dissociation of the PTS-protein from Pex5p (11)(12)(13). The exact sequence of events in the import mechanism remains, however, unknown. It is in particular unclear how, in contrast with other organelles, peroxisomes can import folded oligomeric, functional proteins (14).
Previous biophysical work indicated that the N terminus half of Pex5p is unfolded in vitro (15,16). Recent protease sensitivity assays showed that the proteolytic profiles of the full-length receptor Pex5p(F) change in the presence of PTS1 peptide and the Pex13p Src homology 3 domain, which is another docking factor (16,17), indicating conformational changes of Pex5p upon binding these receptor ligands. Furthermore, it was found that Pex5p may even traverse the peroxisomal membrane, leaving only a small N-terminal fragment in the cytosol while exposing the C-terminal TPR domain to the luminal side of the membrane (11).
Although recognition of many PTS cargos seems to be confined to the C-terminal TPR domains of Pex5p, it has become clear that the N-terminal part of Pex5p is primarily involved in docking of the receptor onto the peroxisomal membrane and other docking factors. Because only poorly diffracting crystals have been purified to date, we investigated its solution structure by small angle x-ray scattering (SAXS) and static light scattering (SLS). Complexes with the PTS1 cargo sterol carrier protein 2 (SCP2), which functions as lipid transfer protein, were also studied as the crystal structure of Pex5p(C)/SCP2 is already known (4). Our results indicate that human Pex5p(F) is a monomer with an extended N terminus. The stoichiometry of Pex5p(F)⅐ Pex14p(N)⅐PTS1 complex has been assessed by titration with SAXS, SLS, and gel filtration, and a low resolution structural model of the complex has been reconstructed in which Pex5p(F) remains extended upon Pex14p(N) binding.
SCP2 was purified by successive passage over Ni 2ϩ -NTAagarose in binding buffer and glutathione-Sepharose 4B (Amersham Biosciences) columns in buffer II, containing 50 mM HEPES-KOH (pH 7.5), 100 mM KCl, and 1 mM TCEP. The glutathione S-transferase tag of SCP2 was removed with Histobacco etch virus protease (30 g/ml) for 2 h at room temperature. The SCP2 released from the resin was passed through Ni 2ϩ -NTA-agarose and finally purified over Superdex 75 in buffer II.
Pex14p(N) was purified by successive passage over Ni 2ϩ -NTA-agarose in binding buffer and Superdex 75 columns, ending with buffer II. Fractions containing Pex5p(N) and Pex5p(C) were purified by successive passage over Ni 2ϩ -NTA-agarose in binding buffer and Superdex 75 columns, ending with buffer II.
To form the binary Pex5p(F)⅐Pex14p(N) complex, after harvesting the two sets of E. coli cells containing (Pex5p(F) and Pex14p(N)) were mixed on ice for 30 min and resuspended in lysis buffer (see above) with protease inhibitors. The Pex5p(F)⅐ Pex14p(N) complex was purified by successive passage over Ni 2ϩ -NTA-agarose in binding buffer and Superdex 200 size exclusion columns, ending with buffer III containing 50 mM HEPES-KOH (pH 7.5), 100 mM KCl, and 2 mM TCEP.
The binary Pex5p(F)⅐SCP2 complex was formed by incubating a (1:1) mixture of purified Pex5p(F) and SCP2 on ice for 30 min and purified on a Superdex 200 size exclusion column, ending with buffer III. The ternary Pex5p(F)⅐Pex14p(N)⅐ SCP2 complex was formed by incubating a (1:1) mixture of purified (Pex5p(F)⅐Pex14p(N)) and SCP2 on ice for 30 min and purified on a Superdex 200 size exclusion column, ending with buffer III.
To form the ternary Pex5p(C)⅐Pex14p(N)⅐SCP2 complex, after harvesting, three sets of E. coli cells containing Pex5p(C), Pex14p(N), and SCP2, were mixed on ice for 30 min and resuspended in lysis buffer (see above) with protease inhibitors. The complex was purified by successive passage over Ni 2ϩ -NTAagarose in binding buffer and glutathione-Sepharose columns in buffer II. The glutathione S-transferase tag of SCP2 was removed with His-tobacco etch virus protease (30 g/ml) for 2 h at room temperature. The complex released from the resin was passed through Ni 2ϩ -NTA and finally purified over Superdex 200 in a buffer II. All constructs were pooled and concentrated for the SAXS measurements using a Vivaspin 20 spin concentrator (Vivascience) with a 3-or 5-kDa molar mass (MM) cutoff.
The purity and homogeneity of all proteins were confirmed by SDS-PAGE, native PAGE, and dynamic light scattering (DLS). The extinction coefficients of His 6 -Pex5p(F), His 6 -Pex14p(N), and SCP2, used in determining concentrations and/or mixing ratios, were calculated from their sequences to be 100,500, 5,690, and 5,500 M Ϫ1 ⅐cm Ϫ1 , respectively.
SLS-Pex5p(F), Pex5p(F)⅐Pex14p(N), Pex5p(F)⅐SCP2, Pex5p(F)⅐ Pex14p(N)⅐SCP2, and Pex5p(C)⅐Pex14p(N)⅐SCP2 were injected at concentrations of 0.3, 0.7, 0.5, 0.7, and 0.1 mg/ml, respectively, into a Superdex 200 10/300 GL tricorn column (Amersham Biosciences), equilibrated with buffer II at room temperature, and connected with an SLS-detector (mini DAWN Tristar; Wyatt Technology, Santa Barbara, CA). The instrument measures the protein concentration by UV absorbance, the refractive index, and the scattering intensity of the eluted peaks. The ASTRA software (Version 4.90.08; Wyatt Technology, Santa Barbara, CA) was used to compute the MM via the basic light-scattering equation (20). A set of standard proteins (Bio-Rad) was used to calibrate the column. The fractions containing the complexes of interest were analyzed on SDS-PAGE and native gels (Invitrogen).
SLS Titration Experiment-Pex5p(F) and Pex14p(N), at 3.5 and 6.3 mg/ml, respectively, were dialyzed in buffer III. Solutions containing appropriate molar ratios of Pex5p(F) and Pex14p(N) between 1:1 and 1:8 were incubated on ice for 1 h to form complexes and processed as indicated above. The fractions containing the complexes of interest were analyzed on a native gel.
Small Angle X-ray Scattering Measurements-The synchrotron radiation x-ray scattering data were collected on the X33 beamline of the EMBL on the storage ring DORIS III (DESY, Hamburg, Germany) (21). Solutions of Pex5p and its binary and ternary complexes with Pex14p(N) and SCP2 proteins were measured at 15°C at solute concentrations of 0.5, 1.0, and 2.0 mg/ml. Before the measurements, 20 mM TCEP was added to samples to reduce the radiation damage. The data were recorded using a MAR345 image plate detector at a sampledetector distance of 2.7 m and a wavelength of ϭ 0.15 nm, covering the range of momentum transfer 0.12 Ͻ s Ͻ 4.5 nm Ϫ1 (s ϭ 4 sin/, where 2 is the scattering angle). No measurable radiation damage was detected by comparison of successive time frames with 2-min exposures. The data were averaged after normalization to the intensity of the transmitted beam, and the scattering of the buffer was subtracted. The difference data were extrapolated to zero solute concentration following standard procedures. All data manipulations were performed using the program package PRIMUS (22).
The forward scattering I(0) and the radius of gyration R g were evaluated using the Guinier approximation (23) assuming that at very small angles (s Ͻ 1.3/R g ) the intensity is represented as I(s) ϭ I(0)exp(Ϫ(sR g ) 2 /3). These parameters were also computed from the entire scattering patterns using the program GNOM (24), which provides the maximum particle dimensions D max and the distance distribution functions p(r). The molecular masses of the solutes were estimated from the forward scattering by normalization against reference solutions of bovine serum albumin. The excluded (Porod) volumes of hydrated particles were computed as (25) Before the calculation, an appropriate constant was subtracted from each data point to force the s Ϫ4 decay of the intensity at higher angles following Porod's law (25) for homogeneous particles. This procedure yields a "shape scattering" curve corrected for the unwanted scattering contribution from the internal structure. For globular proteins the computed Porod (i.e. hydrated) volumes in nm 3 are expected to be about twice the MMs in kDa.
Ab Initio Shape Determination-The low resolution shapes of Pex5p(F) and the ternary Pex5p(F)⅐Pex14p(N)⅐SCP2 complex were reconstructed by the program DAMMIN (26) and its multiphase version MONSA (27). These programs represent the particle as a collection of MϾ Ͼ1 densely packed beads inside a sphere with diameter D max . In DAMMIN, each bead is assigned either to the solvent or to the particle, and the latter is represented by a simple "phase" (non-solvent beads). In MONSA each bead is assigned either to the solvent (index ϭ 0) or to one of the parts of the complex (index ϭ 1 corresponding to Pex5p(N) (residues 1-316), index ϭ 2 for Pex5p(C) (residues 317-639), index ϭ 3 for the first Pex14p(N) protein binding to Pex5p(F) together with SCP2 protein, and index ϭ 4 for all other binding Pex14p(N) proteins). The particle is, therefore, represented at low resolution by four phases (portions of the complex), and the overall model is described by a string of length M containing the phase index for each bead (0 ϭ solvent or 1, 2, 3, 4 for any of the parts of the complex). Starting from a random string, simulated annealing (SA) is employed in both DAMMIN and MONSA to search for a model composed of interconnected compact phases, fitting single (DAMMIN), or multiple shape (MONSA) curves to minimize the overall discrepancy, where I k (s) is the scattering intensity from a construct k, the index k runs over the scattering curves, N k are the numbers of experimental points in each curve, c k are scaling factors, and I calc (s) and (s j ) are the intensities calculated from the subsets of the beads belonging to the appropriate phases and the experimental errors at the momentum transfer s j , respectively.
In MONSA, the ab initio procedure was applied to simultaneously fit five scattering curves (one curve from Pex5p(F), two curves from binary complexes of Pex5p(F)⅐Pex14p(N) and Pex5p(F)⅐SCP2, and two from ternary constructs of Pex5p(F)⅐ Pex14p(N)⅐SCP2 and Pex5p(C)⅐Pex14p(N)⅐SCP2). Given the assumption that each phase has a uniform density, the shapescattering curves after appropriate constant subtraction (see above) were fitted up to the resolution of about 3 nm (i.e. in the range of the scattering vectors up to s ϭ 2 nm Ϫ1 ). The expected volume fractions of the four phases, corresponding to their theoretical molecular masses, were used to further restrain the SA procedure. In independent DAMMIN calculations, the relevant curves were fitted separately to yield the low resolution models of the corresponding subcomplexes.
Molecular Modeling-Rigid body modeling was done using the atomic models of the C-terminal domain of Pex5p (residues; 280 -602, Protein Data Bank (PDB) code 1FCH) (2), Pex5p(C)/SCP2 (residues; 335-639/22-143, PDB code 2C0L) (4), and Pex14p(N) (Personnel communication) as input for the programs BUNCH and SASREF (28). Note that the residue numbers of 1FCH correspond to those of the Protein Data Bank and differ from 2C0L. The scattering amplitudes of the individual subunit structures from the ternary Pex5p(F)⅐Pex14p(N)⅐ SCP2 complex were calculated from their atomic coordinates by the program CRYSOL (29).
To model the structure of Pex5p(F) in solution, the unknown N-terminal segment was first represented by interconnected chains composed of dummy residues. An SA protocol implemented in the program BUNCH (28) was employed to find a native-like configuration of the N-terminal fragment without steric clashes, fitting the experimental scattering from the fulllength protein. The results of multiple BUNCH runs were averaged to determine common structural features using the programs DAMAVER (30) and SUPCOMB (31). The latter program aligns two arbitrary low or high resolution models represented by ensembles of points by minimizing a dissimilarity measure called normalized spatial discrepancy (NSD). For every point (bead or atom) in the first model, the minimum value among the distances between this point and all points in the second model is found, and the same is done for the points in the second model. These distances are added and normalized against the average distances between the neighboring points for the two models. Generally, NSD values close to one indicate that the two models are similar. DAMAVER generates the average model of the set of superimposed structures and also specifies the most typical model (i.e. that having the lowest average normalized spatial discrepancy with all other models in the set).
The resulting low resolution structure of the N-terminal part of Pex5p(F) was further used to model the ternary Pex5p(F)⅐ Pex14p(N)⅐SCP2 complex. The position of SCP2 with respect to Pex5p(C) was fixed as in the crystallographic model of truncated Pex5p(C)⅐SCP2 complex (as in 2C0L), and the portion of Pex5p(C) unresolved in the latter model was added (as in 1FCH). This model was treated as a single rigid body in all subsequent calculations. The six copies of Pex14p(N) were added, and their positions were refined by rigid body modeling using the program SASREF (28). This program employs a SA protocol to generate an interconnected assembly of subunits without steric clashes fitting the scattering data. Intersubunit contacts were loosely defined for the Pex14p(N) interaction with the WXXX(F/Y) motifs (residues 28 -32, 118 -122, 140 -144, 159 -163, 243-247, and 308 -312) of Pex5p(F), based on NMR data (10) (Fig. 1A).
To select the most typical ab initio model of the complex and estimate its possible conformational space, 10 MONSA reconstructions were pairwise aligned using SUPCOMB (31). For each model, the values of NSD against all other models were computed, and an average value ͗NSD͘ was calculated. The most typical reconstruction was the one with the lowest average NSD with respect to the other reconstructions. The same selection was made for the SASREF reconstruction to select the most typical rigid body model. Circular dichroism (CD)-CD spectra of Pex5p(F), Pex5p(N), and Pex5p(C) were measured on a Jasco J-810 spectropolarimeter equipped with a Peltier temperature control unit using a cylindrical 1-mm Hellma Quartz Suprasil cuvette in the wavelength range 190 -240 nm. The protein concentrations were 0.6 -1.5 M in 10 mM sodium phosphate buffer (pH 7.4) at 10°C. The spectra were averaged over three scans, corrected for buffer background, and scaled against the camphor-10sulfonic acid ellipticity calibration.

Determination of the Molar Mass and Stoichiometry of the Pex5p(F)⅐Pex14p(N) Complex by SLS and Native Gel Analysis-
All Pex5p-containing constructs were first analyzed to assess their oligomeric states in solution. For this, recombinant human Pex5p(F), Pex5p(F)⅐Pex14p(N), Pex5p(F)⅐SCP2, Pex5p(F)⅐ Pex14p(N)⅐SCP2, and Pex5p(C)⅐Pex14p(N)⅐SCP2 were studied by native gel analysis (Fig. 1C) and by SLS (Fig. 1D). The average MMs from SLS presented in Table 1 indicate that Pex5p(F) alone is monomeric and forms complexes with multiple Pex14p(N)s at the concentrations used for gel filtration. The complex formation was verified by SDS and native gels, and the purified constructs were found to be homogeneous in native gels ( Fig. 1C) but displayed the expected bands of the individual components in SDS gels (Fig. 1B). The interaction between Pex5p(F) and Pex14p(N) was further assessed by SLS titration. The apparent MM of Pex5p(F)⅐Pex14p(N) increases as illustrated in Table 2 and Fig. 2 up to a molar ratio of 1:6. Above a 1:7 molar ratio, the value saturates, suggesting that Pex5p(F) binds six or seven Pex14p(N)s in solution. This result was further verified by a SAXS titration experiment (see below).
Molecular Parameters from SAXS-The structural parameters of Pex5p(F) determined from the experimental scattering pattern (Fig. 3A) are given in Table 1. The estimated MM and hydrated particle volume agree well with those predicted from the primary structure, indicating that Pex5p(F) is monomeric in solution. The R g and D max values are 5.2 Ϯ 0.1 and 18 Ϯ 1 nm, respectively, confirming recent observations indicating that Pex5p(F) is elongated (15,16). The distance distributions p(r) computed from the experimental data of all complexes investigated are given in Fig. 3B. The bell-shaped p(r) of the truncated Pex5p(C)⅐Pex14p(N)⅐SCP2 complex (curve 5) is typical for globular particles (32), pointing to a compact shape of the construct. The other p(r) functions have similar skewed profiles characteristic of elongated particles with a maximum at around 5 nm, suggesting that little or no compaction occurs when complexes with Pex14p(N) and SCP2 are formed. The p(r) functions for complexes in the presence of the scaffold ligand Pex14p, Pex5p(F)⅐Pex14p(N) (curve 2), and Pex5p(F)⅐Pex14p(N)⅐SCP2 (curve 1) go above that for the binary cargo Pex5p(F)⅐SCP2 complex (curve 3) in the distance range between 6 and 12 nm, suggesting that this range mostly corresponds to the distances between Pex14p molecules in the complex. Further evidence about the conformation of the N terminus of Pex5p(F) is provided by the comparison of the sizes of different constructs determined by DLS and SAXS. For the truncated Pex5p(C)⅐Pex14p(N)⅐SCP2 complex, the hydrodynamic radius R h from DLS (Table 1, last row, first data column) coincides with the radius of gyration R g from SAXS (Table 1, last row, third data column), as expected for globular particles. In contrast, for all remaining constructs involving the full-length Pex5p, R g significantly exceeds R h . Such an effect is observed for flexible macromolecules. For unfolded proteins, the ratio R g /R h is about 1.5 (33). Therefore, given that this type of effect has not been observed for complexes involving the C-terminal cargo binding part of the receptor only, the data demonstrate that the observed molecular flexibility must originate from the N-terminal part of the receptor regardless of whether is of the ligands investigated (SCP2, Pex14p(N)) are bound.  Fig. 2A (Table 2), the apparent R g and MM exp values do increase up to a 1:6 Pex5p(F)⅐Pex14p(N) molar ratio and start to decrease above a 1:7 molar ratio. This result is consistent with those of the SLS titration experiment (see above) and is further corroborated by the native gel analysis (Fig. 2B), also displaying saturation above a 1:6 molar ratio.

Determination of the Stoichiometry of the Complex between Human Pex5p(F) and Pex14p(N) by SAXS-For
Ab Initio Modeling of Pex5p(F)-The structure of Pex5p(F) protein in the absence of ligands, restored ab initio from the scattering pattern (Fig. 3A), is shown in Fig. 4A. It was obtained by averaging 10 independent reconstructions yielding a good fit to the experimental data in the entire scattering range (the discrepancy is given in Table 1). The model depicts a compact globular domain, which may correspond to the C-terminal part of Pex5p and an elongated tail. Several independent runs yielded reproducible ab initio models of the Pex5p(F)⅐Pex14p(N)⅐SCP2 complex, which neatly fitted the experimental scattering profiles (Table 1). A typical model presented in Fig. 4B demonstrates that the Pex5p(F)⅐Pex14p(N)⅐ SCP2 complex adopts an elongated conformation in solution. The overall appearance of the model is similar to the shape of Pex5p(F) alone, implying that there are no large structural rearrangements upon complex formation with Pex14p(N) and SCP2. The overlap of this model with the shape of apoPex5p(F) (Fig. 4A) indicates that both models have similar overall shapes. The model of the complex displays additional mass, suggesting possible binding sites of Pex14p(N) molecules. The above results of the ab initio modeling suggest that the overall structure of Pex5p(F) remains extended in complex with Pex14p(N) and SCP2. For a more detailed analysis, the program MONSA was employed to build a multiphase bead model of the complex by simultaneous fitting of the available data. Several independent runs yielded reproducible solutions, and the most typical one is shown in Fig. 5. The overall shape of the MONSA model is similar to the one from DAMMIN (Fig. 4B), but the former model highlights the shapes of individual parts of the ternary complex. The shape of N terminus of Pex5p(F) in the complex remains elongated when the Pex14p(N) molecules are bound to TABLE 1 Overall parameters of Pex5p constructs from DLS, SLS, and SAXS R g , D max , V p , and R h are, respectively, the experimental radius of gyration, maximum size, excluded volume, and hydrodynamic radius. MM exp , MM sls , and MM th and molar masses were determined by SAXS and SLS and calculated from the appropriate primary sequences, respectively. C h and C sls are the concentration of samples used for DLS and SLS, respectively. M is the overall discrepancy between the experimental data and computed scattering curves from the ab initio MONSA model, RB denotes the discrepancy of a typical rigid body model by SASREF assuming 1:6:1 ternary complex formation of Pex5p(F)⅐Pex14p(N)⅐SCP2.  the N-terminal region of Pex5p(F). The Pex14p(N) moiety depicted within the overall shape of the complex (red beads in Fig. 5) appears essentially intermingled with that of the N terminus of Pex5p(F). At the same time the C-terminal domain of Pex5p(F) displays a globular form, matching previous structural data (5,34). Molecular Modeling-In an independent approach, a model of the ternary complex was constructed making use of the high resolution models of the individual domains. These domains were treated as rigid bodies, whereas the fragments with unknown structure were represented as flexible chains of dummy residues. First, the model of Pex5p(F) was generated consisting of the compact C-terminal domain and the N terminus containing 322 dummy residues. Multiple runs of BUNCH (28) starting from random initial configurations yielded an extended configuration of N-terminal region of Pex5p(F), providing a good fit to the data (Fig. 4A). Twenty reconstructions were pairwise compared, and the most typical model, which had the smallest average normalized spatial discrepancy with respect to the other models, was selected. This model overlaps well with the ab initio model of Pex5p(F) (Fig. 4A).

Sample
The program SASREF (28) was then used for rigid body modeling of the ternary Pex5p(F)⅐Pex14p(N)⅐SCP2 complex by simultaneously fitting all available scattering profiles. The N-terminal region of Pex5p(F) was taken as rigid body from the BUNCH reconstruction, and SAS-REF was run for two stoichiometries of the Pex5p(F)⅐Pex14p(N)⅐SCP2 (1:6:1 and 1:5:1). The individual Pex14p(N) molecules were required to have contacts with the appropriate binding sites of Pex5p(F) (see "Experimental Procedures"). Multiple runs of SASREF yielded reproducible models whereby calculations with a 1:6:1 stoichiometry gave somewhat better overall fits (see Table 1). The most typical model of the ternary complex (with the lowest average NSD value) displays the overall shape similar to that provided by DAMMIN (Fig. 4B). A comparison with the most typical multiphase MONSA model in Fig. 5 reveals that the two independent modeling approaches predict a similar organization of the complex. The Pex14p(N) molecules (red traces in Fig. 4B), similar to the MONSA model in Fig. 5, do not build a compact cluster but, instead, are spread over the surface of the entire N terminus.
The variability and accessible conformational space of the Pex5p(F)⅐ Pex14p(N)⅐SCP2 is illustrated in supplemental Fig. S1. It is evident that the ternary complex displays rather flexible organization of the individual components, but the overall shape of all the models remains similar. The average NSD values for the most typical ab initio and most typical rigid body models are 1.31 and 1.45, respectively, indicating a good reproducibility of the results.
CD-To independently verify the SAXS results suggesting a disordered conformation of the N-terminal part of Pex5p(F), the full-length Pex5p, Pex5p(N), and Pex5p(C) were analyzed by CD. The results (Fig. 6) indicate that Pex5p(C) is largely ␣-helical, whereas Pex5p(N) shows little secondary structure. The absence of secondary structure further corroborates the SAXS-based model displaying an extended Pex5p(F).

DISCUSSION
The SAXS model of the solution structure of human Pex5p(F) and the SLS results indicate that in solution human Pex5p(F) is an elongated monomer at the concentrations used (0.14 -2.5 mg/ml), in agreement with recent studies (15,16). Furthermore, the shape obtained from SAXS reveals that the unfolded N-terminal region and the globular C-terminal TPR domain are arranged in line. The Pex5p(F) tetramer was also observed by electron microscopy (EM) (35) together with filamentous aggregates. We performed titration experiments by SAXS and SLS, which independently pointed to a Pex5p(F)⅐Pex14p(N) stoichiometry of about (1:6). These results are in good agreement with previous surface plasmon resonance experiments and fluorescence titration analysis of the interaction of Pex14p (1-78) with synthetic peptides (WXXX(F/Y) motifs), which proposed 1:6 -7 complexes (8,9). Previous data indicated, however, that not all Pex14p binding motifs in Pex5p are required for its function as an PTS1-import receptor (9). Electrophoretic experiments of the related Pex5p(F)⅐Pex14p(F) complex, isolated from rat liver, also indicated that more than one Pex14p ligand is bound to the Pex5p receptor, with a probable stoichiometry of about 1:5 (17). Of course, the methods employed for titration in the present study (SAXS and SLS) are not site-specific, i.e. they give the overall stoichiometry only and do not provide evidence about the occupancy of the specific Pex14p(N) binding sites on Pex5p(F).
No significant structural changes are detected by SAXS upon complex formation of human Pex5p(F), i.e. Pex5p(F)⅐Pex14p(N)⅐ SCP2 retains an elongated shape, and if any, only minor conformational changes occur. Dynamics or instability in the protein fold would be expected for a region involved in protein-protein   (2)) is displayed as green C traces. The N-terminal part of Pex5p restored from SASREF is shown as blue C traces. The five Pex14p(N) proteins bound to the Pex5p(F) N terminus are drawn as red C traces, the sixth Pex14p(N) and SCP2 proteins are indicated by yellow C traces. Right and bottom views are rotated counterclockwise by 90°around the y and x axes, respectively.
interaction (36). The natively disordered region and extended Stokes radius can lead to "Fly casting" (37), where proteins can "fish" for binding partners. This is a particularly effective means of recruiting protein assemblies at the membrane surfaces. For example, in mammalian endocytosis, the native disordered region of adapter proteins recruits clathrin at the membrane surface (38). Pex5p(F) containing a significant unfolded N-terminal domain could connect with other proteins or the membrane across a relatively large distance. It is, therefore, possible for Pex5p(F) to present several binding sites to a relatively wide area, as the SAXS model showed, and to improve the efficiency with which proteins are assembled or become recruited to the membrane. In addition, the long and unfolded N-terminal domain of Pex5p(F) on the membrane could allow the binding partners to remain close together, which facilitates the efficient exchange of binding partners after each binding interaction has fulfilled its purpose. By analogy with these observations, the unstructured N-terminal part of Pex5p(F) and its elongated shape might be necessary for the Pex5p(F)⅐Pex14p(N)⅐PTS1 complex to interact with other substrates, such as lipids and membrane proteins during the formation of functional multiprotein complexes (36,39,40).
The binding properties of human Pex5p differ significantly from those of its equivalent in lower organisms. In higher eukaryotes (plants, mammals), Pex5p and Pex7p interact with Pex13p and Pex14p at the peroxisomal membrane during the docking step of peroxisomal protein import, whereas in lower eukaryotes (yeast, fungi), they bind to Pex13p, Pex14p, and Pex17p (41). EM studies have revealed that Hansenula polymorpha Pex5p(F) (Hp Pex5p) is a globular tetramer near pH 7 which exists in two forms (open/closed) and undergoes a conformational change upon interaction with Hp Pex20p (42). A large conformational change was also observed by fluorescence spectroscopy in Hp Pex5p upon interaction with Hp Pex8p, an intraperoxisomal protein for which there is no equivalent in humans (18). In contrast, our data on the human Pex5p receptor do not indicate either homo-oligomerization of any of the Pex5p fragments or major structural changes upon Pex14p(N)/SCP2 binding. It remains to be determined whether the different findings could be attributed to variations in experimental conditions or they reflect possible taxonomic differences in terms of architecture and function of the Pex5p receptor.