Crystal Structure of the Measles Virus Phosphoprotein Domain Responsible for the Induced Folding of the C-terminal Domain of the Nucleoprotein*

Measles virus is a negative-sense, single-stranded RNA virus belonging to the Mononegavirales order which comprises several human pathogens such as Ebola, Nipah, and Hendra viruses. The phosphoprotein of measles virus is a modular protein consisting of an intrinsically disordered N-terminal domain (Karlin, D., Longhi, S., Receveur, V., and Canard, B. (2002) Virology 296, 251–262) and of a C-terminal moiety (PCT) composed of alternating disordered and globular regions. We report the crystal structure of the extreme C-terminal domain (XD) of measles virus phosphoprotein (aa 459–507) at 1.8 Å resolution. We have previously reported that the C-terminal domain of measles virus nucleoprotein, NTAIL, is intrinsically unstructured and undergoes induced folding in the presence of PCT (Longhi, S., Receveur-Brechot, V., Karlin, D., Johansson, K., Darbon, H., Bhella, D., Yeo, R., Finet, S., and Canard, B. (2003) J. Biol. Chem. 278, 18638–18648). Using far-UV circular dichroism, we show that within PCT, XD is the region responsible for the induced folding of NTAIL. The crystal structure of XD consists of three helices, arranged in an anti-parallel triple-helix bundle. The surface of XD formed between helices α2 and α3 displays a long hydrophobic cleft that might provide a complementary hydrophobic surface to embed and promote folding of the predicted α-helix of NTAIL. We present a tentative model of the interaction between XD and NTAIL. These results, beyond presenting the first measles virus protein structure, shed light both on the function of the phosphoprotein at the molecular level and on the process of induced folding.

Measles virus (MV) 1 is a member of the Paramyxovirinae sub-family within the Mononegavirales order. Its non-segmented, negative-sense, single-stranded RNA genome is packaged by the viral nucleoprotein (N) within a helical nucleocapsid. Mononegavirales use this N-RNA complex as a template for both transcription and replication. These reactions are carried out by the RNA-dependent RNA polymerase polymerase (L) in conjuction with the phosphoprotein (P) (for a review see Ref. 1). During genome replication, synthesis of viral RNA and encapsidation by N are concomitant (1,2). N also has the capacity to self-assemble on cellular RNA to form nucleocapsid-like particles in the absence of viral RNA and viral proteins (3)(4)(5)(6)(7). Association of P with the soluble, monomeric form of N (N 0 ) prevents it from binding to cellular RNA (8 -10). The N 0 -P complex is the substrate used by the polymerase to initiate encapsidation of genomic RNA (10,11). N forms complexes with P and with the P-L complex during transcription and also, during replication, in its self-assembled form (N NUC ) (12)(13)(14)(15).
MV N consists of an N-terminal moiety, N CORE , and a Cterminal moiety, N TAIL , which is intrinsically unstructured (i.e. it lacks any constant secondary structure in physiological conditions) (16) and protrudes from the surface of viral nucleocapsids (17,18). The presence of flexible regions exposed at the surface of the viral nucleocapsid would ideally favor the interaction with different viral partners. In agreement, although N CORE contains all the regions necessary for self-assembly and RNA-binding (4,19,20), N TAIL contains the regions responsible for binding to P (16, 19 -21), to the polymerase complex P-L, and possibly to the matrix protein (22).
The P protein of Paramyxovirinae plays multiple roles in both transcription and replication, being an indispensable subunit of the viral polymerase complex and being responsible for binding and delivery of N 0 to the newly synthesized genomic RNA. P is a modular protein, organized into two moieties functionally and structurally distinct: a poorly conserved Nterminal moiety (PNT), which is intrinsically unstructured (23), and a well conserved C-terminal moiety (PCT). PCT contains all the regions required for transcription (14), whereas PNT provides several additional functions required for replication (10,14,24).
Paramyxovirinae PNTs are responsible for binding to N CORE within the N 0 -P complex (8,10,25,26). The main N 0 -binding site of MV PNT has been mapped to the extreme N terminus of P (21). Paramyxovirinae PCT share three common features: (i) oligomerization through a coiled-coil region, (ii) the presence at the extreme C terminus of a short predicted bundle of three ␣-helices, which is involved in binding to N (27), and (iii) the presence of an L binding site (28,29). The P protein of the related Sendai virus (SeV), a member of the Respirovirus genus, is tetrameric, whereas the oligomeric state of MV P is unknown (21,27). Experimental data available suggest that SeV P participates in both transcription and replication in its multimeric form, being tetrameric when complexed with both L (10) and N 0 . 2 Deletion of the extreme C-terminal domain of SeV PCT (aa 479 -568) abolishes binding to N NUC , as well as stable binding to N 0 (10,21,24).
The SeV P gene contains an in-frame internal ribosomal entry site which is used to produce the C-terminal part of the P protein (30,31). This internal translation product is named protein X and encompasses aa 474 -568 (12). Although SeV protein X encompasses an N binding site, stable binding of protein X on nucleocapsid determined by sedimentation through a caesium chlorure gradient could not be observed (12). Within the MV PCT gene region there are three in-frame AUG codons at amino acid positions 371, 422, and 446, but whether one of them is used to initiate translation of a protein X is unknown.
The only structural information available so far concerns MV PNT (23) and SeV PCT. The structure of a slightly shortened form (aa 320 -433) of the multimerization domain of SeV (P, aa 317-445) has been solved by x-ray crystallography (32). It forms a very stable coiled-coil buttressed by a bundle of three short ␣-helices. The structure of the protein X of SeV (aa 474 -568) has been solved by NMR (33). It is monomeric and consists of a flexible linker (aa 474 -517) fused to the X domain (XD) (aa 517-568). This linker is probably conserved also in MV PCT, as suggested by its high protease sensitivity (16).
We have recently reported that MV PCT is responsible for the induced folding of N TAIL (16). The induced folding of N TAIL in the presence of its physiological partner opens new perspectives in the structural study of disordered regions within the MV replicative complex. Indeed, the gain of structure arising from the interaction with PCT may lead to the formation of a complex rigid enough to be crystallized. Furthermore, although natively unfolded proteins have been receiving increased interest in the last few years, the molecular bases of the disorderedto-ordered transitions are still unknown. To identify precisely the PCT regions involved in the induced folding of N TAIL and to characterize structurally these complexes, we have analyzed the domain organization of MV P.
In this paper, we report the bacterial expression and purification of the extreme C-terminal domain, XD (aa 459 -507), of MV P. We show that MV N TAIL undergoes induced folding in the presence of XD. We report the crystal structure of XD at 1.8 Å resolution, which represents the first measles protein structure to be solved. Finally, we present a tentative model of the interaction between XD and N TAIL , which casts light on this widespread and still structurally undefined protein-protein mode of interaction.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Media-The Escherichia coli strain DH5␣ (Stratagene) was used for selection and amplification of DNA constructs. The E. coli strain C41 [DE3] (Avidis) was used for expression of recombinant XD. E. coli was grown in Luria-Bertani medium.
Cloning of the XD Coding Region-The XD gene construct, encoding residues 459 -507 of the MV P protein (strain Edmonston B) with a hexahistidine tag fused to its C terminus, was obtained by PCR using pET21a/P H6 (23) as template and Pfu polymerase (Promega). Forward primer (5Ј-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGAAGG-AGATAGAACCATGGCATCACGCAGTGTAATCCGCTCC-3Ј) was designed to introduce an AttB1 site (bold), whereas reverse primer (5Ј-GGGGACCACTTTGTACAAGAAAGCTGGGTCTTATTAGTGGT-GGTGGTGGTGGTGCTTCATTATTATCTTCACCAGCAT-3Ј) was designed to introduce an AttB2 site (bold) and a hexahistidine tag (underlined) at the C terminus of XD. Primers were purchased from Invitrogen. After purification (PCR purification kit, Qiagen), the PCR product was cloned into the pDest14 vector (Invitrogen) using the Gateway recombination system (Invitrogen). The final construct is referred to as pDest14/XD H6. The sequence of the coding region was checked by sequencing (MilleGen). Expression and Purification of XD-E. coli strain C41[DE3] (Avidis) was used for the expression of the pDest14/XD H6 construct. Because the MV P gene contains several rare codons that are used with a very low frequency in E. coli, co-expression of pDest14/XD H6 with the plasmid pLysS (Novagen) was carried out. This plasmid, which supplies six rare tRNAs, also carries the lysozyme gene, thus allowing a tight regulation of the expression of the recombinant gene, as well as a facilitated lysis. Cultures were grown overnight to saturation in LB medium containing 100 g/ml ampicillin and 17 g/ml chloramphenicol. An aliquot of the overnight culture was diluted 1/25 in LB medium and grown at 37°C. At OD 600 of 0.7, isopropyl ␤-D-thiogalactopyranoside was added to a final concentration of 0.2 mM, and the cells were grown at 37°C for 3 h. The induced cells were harvested, washed, and collected by centrifugation. The resulting pellets were frozen at Ϫ20°C.
Expression of seleno-methionine-substituted XD was performed using the method of methionine biosynthesis pathway inhibition (34). The pellet containing XD was resuspended in 5 volumes (v/w) buffer A (10 mM Tris-HCl, pH 8, 300 mM NaCl, 10 mM imidazole, 1 mM phenylmethylsulfonyl fluoride supplemented with lysozyme 0.1 mg/ml, DNase I 10 g/ml, protease inhibitor mixture (Sigma) (50 l/gr cells). After a 20min incubation with gentle agitation, the cells were disrupted by sonication (using a 750 W sonicator and 5 cycles of 30 s each at 60% power output). The lysate was clarified by centrifugation at 30,000 ϫ g for 30 min. Starting from a one-liter culture, the clarified supernatant was incubated for 1 h with gentle shaking with 4 ml Talon resin (Clontech) previously equilibrated in buffer A. The resin was washed with buffer A, and XD was eluted in buffer A containing 250 mM imidazole. Eluates were analyzed by SDS-PAGE for the presence of XD. The fractions containing XD were combined and concentrated using Centricon Plus-20 (molecular cutoff ϭ 5000 Da) (Millipore). The protein was then loaded onto a Superdex 75 HR 10/30 column (Amersham Biosciences) and eluted in either 50 mM sodium phosphate, pH 7, 50 mM NaCl or 10 mM Tris/HCl, pH 8, and 5 mM EDTA. The protein was stored at Ϫ20°C. Purification of the seleno-methionine protein was performed as described above.
Expression and purification of N TAIL and PNT was carried out as already described (16,23). All purification steps, except for gel filtrations, were carried out at 4°C. Apparent molecular mass of proteins eluted from the gel filtration column was deduced from a calibration carried out with LMW and HMW calibration kits (Amersham Biosciences).
Protein concentrations were calculated either using the theoretical absorption coefficient ⑀ (mg/ml/cm) at 280 nm (as obtained by using the program ProtParam at the EXPASY server) or the Bio-Rad protein assay reagent.
Dynamic Light Scattering-Dynamic light scattering experiments were performed with a Dynapro MSTC-200 (Protein Solutions) at 20°C. Samples of XD (7 mg/ml in 50 mM sodium phosphate, pH 7, 50 mM NaCl) were filtered prior to the measurements (Millex 0.22 m syringe filters, Millipore). The hydrodynamic radius was deduced from translational diffusion coefficients using the Stokes-Einstein equation. Diffusion coefficients were inferred from the analysis of the decay of the scattered intensity autocorrelation function. All calculations were performed using the software provided by the manufacturer.
Circular Dichroism-CD spectra were recorded on a Jasco 810 dichrograph using 1-mm thick quartz cells in 10 mM sodium phosphate, pH 7, at 20°C. Structural variations of N TAIL were measured as a function of changes in the initial CD spectrum upon addition of different amounts of either XD or PNT. CD spectra were measured between 185 and 260 nm at 0.2 nm/min and were averaged from three independent acquisitions. Mean ellipticity values per residue ([⌰]) were calculated as [⌰] ϭ 3300 m ⌬A/(l⅐c⅐n), where l (path length) ϭ 0.1 cm, n ϭ number of residues, m ϭ molecular mass in daltons, and c ϭ protein concentration 2 J. Curran, personal communication. expressed in mg/ml. Number of residues (n) were 56 for XD, 139 for N TAIL , and 236 for PNT, whereas m values were 8,600 Da for XD, 15,300 Da for N TAIL , and 24,800 Da for PNT. Protein concentrations of 0.1 mg/ml were used. When recording spectra of protein mixtures, mean ellipticity values per residue ([⌰]) were calculated as [⌰] ϭ 3300 ⌬A/ {[(C 1 ⅐n 1 )/m 1 ) ϩ (C 2 ⅐n 2 /m 2 )]⅐l}, where l (path length) ϭ 0.1 cm, n 1 or n 2 ϭ number of residues, m 1 or m 2 ϭ molecular mass in daltons, and c 1 or c 2 ϭ protein concentration expressed in mg/ml for each of the two proteins in the mixture. The theoretical average ellipticity values per residue ([⌰] Ave ), expected assuming that neither any unstructured-tostructured transition nor any secondary structure rearrangement occur, were calculated as follows: 2 correspond to the measured mean ellipticity values per residue, n 1 and n 2 to the number of residues for each of the two proteins, and R to the excess molar ratio of protein 2. The ␣-helical content was derived from the ellipticity at 220 nm as described in Ref. 35.
Crystallization and X-ray Data Collection-Crystallization experiments were performed immediately after protein purification. Screening experiments were performed using the nanodrop approach (36) with three commercial kits: Structure Screens 1 and 2, Stura Footprint Screen (both from Molecular Dimensions Ltd.), and Wizard Screens I and II (Emerald BioStructures). The screens were set up in 8 ϫ 12-well Greiner crystallization plates for sitting drops, with three shelves for each well. Reservoir solutions were 200 l in volume and crystallization drops were composed of 100, 200, or 300 nl of protein solution at 7.5 mg/ml and 100 nl of reservoir solution. The crystallization plates were sealed with transparent film after set-up of the drops and transferred to a storage cabinet at 20°C.
Crystallization of the SeMet protein was performed by doing a refinement of the conditions initially found for the native form, but this time using the hanging drop method. Crystallization drops were made by mixing 3 l of protein solution with 1 l of mother liquor, which were then left for equilibration at 20°C. Crystals typically grew within 48 h to a size of 0.1 ϫ 0.1 ϫ 0.5 mm 3 . Final crystallization conditions were 0.1 M 2-n-(cyclohexylamino)ethanesulfonic acid (CHES), pH 8.5, and 1.25 M sodium citrate.
The x-ray diffraction data on the SeMet XD were collected to a resolution of 1.8 Å on beamline ID14 -2 at the European Synchrotron Radiation Facility, Grenoble, France, using an ADSC Q4 charged-coupled device-based detector. Crystals were cryo-cooled to a temperature of 100°K by an Oxford Cryostream system. As the concentration of sodium citrate was sufficiently high to serve as cryo-protectant in itself, crystals were cryo-cooled without any additional cryo-protectant. The wavelength at the beamline was 0.933 Å, which was off by a few hundreds of an Å on the high-energy side of the selenium K-edge, which was at 0.979 Å. As the anomalous contribution is still significant (f Љϳ 3.5) at this "remote" wavelength, we could solve the structure by the single-wavelength anomalous dispersion (SAD) method.
The data were indexed, merged, and scaled using the programs Mosflm and Scala (Collaborative Computational Project, Number 4). The asymmetric unit contains one molecule, which corresponds to a solvent content of 49%. As the protein sequence contains four methionines (of 55 residues), including the N-terminal Met, the heavy atom search could directly be narrowed down to a maximum of four selenium sites.
Structure Determination and Refinement-For the initial heavy atom search, the program Solve (37) was used in the SAD mode with a resolution cut-off of 3 Å. This allowed us to locate three of the four possible selenium sites. The sites were then refined further using the analyze_solve option in Solve, and phasing was performed to the maximum resolution of 1.8 Å.
Subsequently, solvent flattening was performed by using the program Resolve (38), which produced an electron density map that could be interpreted as being a protein. The phases after density modification were input as initial phases to the program ARP/wARP (39,40), which then performed iterative automatic tracing of the density map, model building, and structure refinement in combination with Refmac5 (Collaborative Computational Project, Number 4). This produced an electron density map of outstanding quality as well as an almost complete model of the protein, with all but one of the side chains in the final model already in sequence (R/R free ϭ 21/26%). 5% of the reflections were used for R free calculations.
Maximum-likelihood refinement was performed with the program Refmac5. By inspection of 2F o Ϫ F c and F o Ϫ F c A electron density maps, the model was built using the program O (41). Building of the solvent structure was performed using ARP/wARP.
At the final stages of refinement, a strong difference density was still unaccounted for in a position between molecules in the crystal lattice. By inspecting its shape, it was interpreted to be a molecule of CHES that was present in the crystallization media, and was refined as such.
Model Minimization-Energy computations of the XD/N TAIL model were done with the GROMOS96 implementation of Swiss PDB Viewer (42,43). 200 steps of steepest descent followed by 20 steps of conjugate gradient energy minimization were performed.

RESULTS
Modular Organization of P-The P protein of MV consists of an N-terminal moiety, PNT (aa 1-230), which is responsible for binding to N CORE within the N 0 -P complex (21), and of a Cterminal moiety, PCT (aa 231-507), which is responsible for binding to N TAIL in both N 0 -P and N NUC -P complexes (9,29). We have recently reported that PCT interacts with the Cterminal moiety of N, N TAIL , and induces folding of the latter. As a first approach toward the identification of the PCT regions involved in the induced folding of N TAIL , and in view of their structural characterization, we have analyzed the domain organization of PCT. Using a combination of different computational approaches, we have shown that Paramyxovirinae PCT have a modular organization, consisting of alternating disordered and globular regions (56). In particular, MV PCT is composed of a disordered region (aa 231-303), a predicted coiled-coil (aa 304 -376), a disordered linker (aa 377-431), and a globular region (aa 432-507) (see Fig. 1). Based on the multiple sequence alignment reported by (27), this C-terminal globular region can be further subdivided into two regions: one encompassing residues aa 432-458, and one spanning residues 459 -507. This latter region, referred to as XD, is the structural counterpart of the extreme C-terminal globular domain (aa 517-568) of SeV P.
To investigate directly the structural properties of XD, we have expressed, purified, and characterized the XD domain alone.
Cloning, Expression, and Purification of XD-The P gene fragment encoding XD was cloned into the pDest14 plasmid (Invitrogen) to yield a C-terminal histidine-tagged recombinant product, the expression of which is under the control of the T7 promoter. Most XD was recovered from the soluble fraction of the bacterial lysate (not shown). XD was purified to homogeneity (Ͼ95%) in two steps: immobilized metal affinity chromatography, and size-exclusion chromatography (not shown). The elution profile from the gel filtration column (not shown) is consistent with a monomeric form. Moreover, dynamic light scattering studies of XD indicate that the protein sample is monodisperse (95%), consisting of a single, monomeric protein species with a Stokes radius of 13 Ϯ 2 Å, which corresponds to a molecular mass of 6 Ϯ 2 kDa.
N TAIL Undergoes Induced Folding in the Presence of XD-We have recently reported that MV N TAIL undergoes an induced folding in the presence of PCT (16). Biochemical data on SeV have shown that the extreme C terminus (aa 479 -568) of PCT contains the region responsible for binding to both N NUC and N 0 (10, 24, 27). Therefore, we have addressed the question of whether MV XD is involved in binding to and in the induced folding of N TAIL .
To monitor the possible structural transitions of N TAIL in the presence of XD, we have used far-UV CD spectroscopy. The far-UV CD spectrum of XD at neutral pH is typical of a structured protein, with a predominant ␣-helical content, as seen by the positive ellipticity between 185 and 200 nm and by two minima at 208 and 222 nm (see Fig. 2, gray line). The ␣-helical content (estimated at 32% based on the ellipticity at 220 nm) is in agreement with the secondary structure predictions as obtained using both PSI-PRED (44) and PHD (45) (not shown). Conversely, the CD spectrum of N TAIL is typical of an unstructured protein, as seen from its large negative ellipticity at 198 nm and low ellipticity at 185 nm (Fig. 2, black line).
After mixing N TAIL with different molar excesses of XD, the observed CD spectra of the mixtures differed from the corresponding theoretical average curves calculated from the individual spectra. Because the theoretical average curves correspond to the spectra that would be expected if no structural variations occur, deviations from these curves point out structural transitions. Most pronounced structural transitions are observed with an N TAIL :XD molar ratio of 1:2 (Fig. 2). When equimolar amounts was used, only slight deviations from the theoretical average curve were observed: the two spectra were almost perfectly superimposable except for the 185-195 nm region, in which a 70% mean increase of ellipticity was observed (data not shown). In the presence of a two-fold molar excess of XD, a random coil to ␣-helix transition can be observed, as indicated by the much more pronounced minima at 208 and 222 nm, and by the higher ellipticity at 190 nm of the experimentally observed spectrum, compared with the corresponding theoretical average curve (Fig. 2). Moreover, the ␣-helical content of the mixture (41%) is not only higher than that of the corresponding theoretical average curve but is also higher than the ␣-helical content of XD alone (32%). Gradually increasing the concentrations of XD to molar ratios as high as 8 does not result in more drastic structural variations than those observed with a two-fold molar excess of XD (data not shown).
As a control, we recorded CD spectra of N TAIL in the presence of either lysozyme or PNT (data not shown and (16)). In both cases, no significant structural variations are observed even with molar excesses as high as 5, thus confirming the significance of the deviation observed in the case of the N TAIL -XD mixtures (see Fig. 2). Therefore, these results indicate that N TAIL undergoes an induced folding upon binding to XD, with a random coil to ␣-helix transition taking place. The ␣-helixforming potential of N TAIL , already assessed by previous CD studies (16), is also in agreement with the secondary structure predictions provided by PSI-PRED (44) and PHD (45), which both predict an ␣-helix (residues 489 -504) as the sole secondary structure element within N TAIL . Therefore, it is conceivable that the induced folding of N TAIL could result in the formation of (at least) one ␣-helix encompassing residues 489 -504.
Overall Structure of XD-The development of the SAD technique over the last few years has shown that novel structures can be solved with single-wavelength data sets collected at wavelengths not necessarily optimized for the anomalous scatterer contained within the structure (46). The structure of MV XD was solved by SAD using synchrotron data collected at the high-energy side ( ϭ 0.933 Å) of the selenium K-edge ( ϭ 0.979 Å) on crystals of the seleno-methionine-substituted protein. The structure has been refined to R ϭ 0.196 and R free ϭ 0.246 (for refinement statistics, see Table I). The electron density is well defined throughout the structure (Fig. 3A), and the model includes residues 2-49 of the XD sequence, as well as four additional histidines from the hexahistidine tag, which adds an extra turn to the C-terminal helix. Most side-chain conformations are well ordered except for a few side chains on the surface.
The structure has an all-␣ architecture composed of three ␣-helices, forming an anti-parallel three-helix bundle (Fig. 3, B  and C). A hydrophobic core is formed by small hydrophobic residues and one aromatic residue (Phe-40), resulting in a tight packing between the three helices. The protein is monomeric in the crystal with no extended contacts between symmetry related molecules.
The final model of MV XD contains 53 amino acid residues, 64 water molecules, and one molecule of CHES. All non-glycin residues are in the most preferred regions of the Ramachandran plot (47) (not shown).
Comparison with Other Structures-The three-helix bundle is a common fold found in many other proteins either as a distinct domain or as part of a larger domain. When performing a search with the Dali server (48) for other proteins with the same fold, more than 200 hits were found.
In an attempt to find clues about how XD and the predicted ␣-helix of N TAIL (likely to be involved in induced folding) would interact in the complex formed by XD and N TAIL , two of the hits in the Dali search are noteworthy: (i) the N-terminal domain of glycerol-3-phosphate acyltransferase (GPAT) (49) (PDB ID code 1k30), and (ii) the Z-domain of protein A from Staphylococcus aureus (Z-domain) (PDB ID codes 1fc2 (50), 1lp1 (51), and 1h0t (52)).
The N-terminal domain of GPAT has a very similar structure to XD (root-mean-square deviation ϭ 1.6 Å for aa 12-75; sequence identity ϭ 18%). An interesting feature is that it has an extra ␣-helix at the N terminus that interacts with the threehelix bundle at the interface between helices ␣1 and ␣3. Thus, on GPAT, this surface, with its hydrophobic cleft running along the length of the ␣1/␣3 helices, is well suited for the binding of a one-helix partner.
It could be hypothesized that N TAIL , with its one predicted ␣-helix, could interact with XD in the same manner. Nonetheless, the ␣1/␣3 surface of XD (Fig. 4A) is chiefly hydrophobic but has no distinct binding cleft as does GPAT. It can be, however, as is the case in most protein interactions, that a conformational change of XD upon binding of N TAIL takes place, but this is impossible to predict by having just one partner alone, as is the case here.
The Z-domain of protein A is well superimposable to XD (root-mean-square deviation ϭ 1.7 Å for aa 126 -164). The sequence identity changes rather drastically, depending on whether a structural (17%) or a strictly sequence-based alignment (27%) is performed. A one-residue insertion in the loop between ␣1 and ␣2 in the Z-domain results in a one-residue "out-of-register" within the ␣2 helix of the two structures. By comparing the residue properties, however, the overall proper-ties are conserved, and the hydrophobic core is maintained in both structures.
The Z-domain has been shown to be able to induce the folding of another protein, namely an affibody targeted against the Z-domain (52). In the absence of the Z-domain, the affibody is a molten globule which has little ordered structure, although upon addition of the Z-domain, it folds into an ordered structure (52). The interface by which the Z-domain interacts with its binding partners, either the immunoglobulin Fc or the anti-Z affibody, is formed by helices ␣1 and ␣2. In both cases, this  3. Structure of XD. A, a representative part of the 2F o Ϫ F c A electron density map contoured at 1 . A molecule of CHES close to Tyr-23 (shown in stick representation) present in the crystallization buffer and located at the interface between symmetry-related molecules in the crystal lattice is also shown. B and C, ribbon representation of the XD model, colored blue at the N terminus and red at the C terminus. The structure of XD is an anti-parallel, three-helix bundle built up by ␣1, ␣2, and ␣3. C, same as in B but rotated by 90°.

FIG. 4. Surface of XD and model of the interaction between the predicted N TAIL helix and XD.
A, the surface created by ␣1 and ␣3 is mainly hydrophobic and rather flat, whereas the ␣1/␣2 face (B) is more hydrophilic, with only a small hydrophobic patch at the center. The ␣2/␣3 helices (C) form a long, mostly hydrophobic cleft running along the length of the helices, which seems well suited for being the interface of interaction between XD and N TAIL . The three different faces of XD are shown in Grasp surface representation with hydrophobic residues in yellow. Rotations of 120°along the indicated axis have been made between each image. D, surface representation of the cleft built up by ␣2/␣3. Residues conserved within Paramyxovirinae are colored in gold. Embedding of the modeled ␣-helix of N TAIL (in red, aa 489-506) with its hydrophobic side running along the hydrophobic cleft at the ␣2/␣3 face creates a pseudo-four-helix bundle. Hydrophobic residues from both proteins are interacting, and there is a nice shape-complementarity between the two proteins. region forms an extended, predominantly hydrophobic surface. In the complex with Fc, the Z-domain interacts with hydrophobic residues with three loops from the Fc. In the affibody complex, two clefts that are roughly perpendicular to the ␣1/␣2 helices of the Z-domain are formed, which are complementary in shape to the two interacting helices from the affibody.
One could imagine that if induced folding of N TAIL involves the formation of more than one ␣-helix (despite secondary structure predictions which predict only one ␣-helix), this latter mode of interaction could also be conserved in the XD-N TAIL complex. However, in the case of XD, the ␣1/␣2 face (Fig. 4B) is more hydrophilic, and no clear clefts are present. Rather, two long charged side chains (Arg-8 and Lys-12) are protruding from the surface, thus preventing binding of a partner in exactly the same manner as the Z-domain.
Model of the Interaction between N TAIL and XD-Based on the similarities in structure and sequence between XD and both GPAT and Z-domain, the XD region most promising for binding to N TAIL seems to be the surface created between ␣2 and ␣3 (Fig 4C). This face is largely hydrophobic and has a cleft running along the length of ␣2/␣3, which seems well suited to accept the binding of an additional helix. In addition, inspection of residues conserved within all Paramyxovirinae P (27) reveals that the ␣2/␣3 face contains about 2/3 of the conserved residues on the XD surface (see Fig. 4D).
N TAIL has only one predicted secondary structure element (an ␣-helix, aa 489 -504) which occurs within the region containing the P binding site (aa 456 -501) (53). A model of this predicted helix was made as an ideal helix with the side chains corresponding to the N TAIL sequence placed in their most common rotamer conformation. By comparing the surface properties of both XD and the helix of N TAIL , a model of how the interaction could take place has been built (Fig. 4D).
The N TAIL helix has one side that is mostly hydrophobic. By modeling this hydrophobic surface of the helix as running along the conserved hydrophobic cleft between ␣2 and ␣3 (Fig. 4D), an arrangement into a pseudo-four-helix bundle, another known stable arrangement of helices, is formed. After performing an energy minimization using GROMOS96 (42), there is a shape complementarity between the two proteins, with no steric clashes. The resulting model is not tightly packed, with a surface buried on XD by the ␣-helix of 800 Å 2 . The complex is stabilized by hydrophobic interactions involving side-chain carbon atoms from the interacting pairs N Leu-498/XD Leu-27 and N Arg-497/XD Met-43. Moreover, a methionine residue (Met-501) protruding from the N TAIL helix fits nicely into a hydrophobic cavity located close to the conserved Phe-40 of XD. Modeling of the pseudo-four-helix bundle does not result in any dramatic conformational changes in the XD structure (root-mean-square deviation ϭ 1.65 Å). The most significant structural rearrangements concern the loops connecting helices ␣1/␣2 and ␣2/␣3, with the maximal deviations occurring at positions 18 (C␣ distance of 2.2 Å) and 34 (C␣ distance of 1.75 Å). DISCUSSION Our previous report on the induced folding of N TAIL in the presence of PCT (16) opens the way to the structural study of this complex, because the gain of structure arising from the interaction with the partner may lead to the formation of a complex rigid enough to allow crystallization. However, the large heterogeneity of PCT (16) represents in principle a major inconvenience for crystallization purposes.
Therefore, we have carried out a bioinformatic analysis of MV P to identify putative globular domains within PCT. This led us to target XD for expression in E. coli. Indeed, this domain turned out to be expressed in the soluble fraction of the bacterial lysate. A slightly longer construct of the domain (aa 432-507) is not expressed in E. coli, supporting the fact that XD is indeed a distinct domain. Moreover, the final purified product is highly homogeneous, as judged by the sharpness of the gel filtration peak (data not shown) and by the high extent of monodispersion. These results, beyond validating the reliability of the prediction of the modular organization of PCT, point out the general interest of a domain approach for the biochemical and structural study of viral proteins.
Far-UV spectroscopy has been proven to be the method of choice to detect unstructured-to-structured transitions of N TAIL upon binding to PCT (16). Using the same approach, we show that N TAIL undergoes induced folding upon binding to XD. Binding of N TAIL to XD results in the same type of structural transitions as that observed with PCT (16) and is in agreement with the strong ␣-helical propensity of N TAIL (16).
The most pronounced structural transitions are observed in the presence of one equivalent of PCT, suggesting the formation of a 1:1 complex with a good affinity constant (16). The two-fold molar excess observed with XD could be ascribed to a lower affinity of N TAIL for XD, compared with PCT, rather than to the formation of a 1:2 stoichiometric complex. It is also conceivable that the induced folding of N TAIL occurs through the formation of a 1:1 stoichiometric complex, and that a molar excess of XD would titrate other low affinity sites, thus mimicking the possible contribution of other PCT regions to the binding to N TAIL . Indeed, in the case of SeV P, stable binding to the nucleocapsid requires the simultaneous presence of the XD region and of the region spanning aa 345-412, all within the PCT domain (12). Further studies, particularly biochemical studies of other PCT modules, are required to answer this question.
The crystal structure reveals that MV XD is a stable domain composed of three ␣-helices arranged in a three-helix bundle. A similar structural arrangement is likely to be also conserved in other Rubulavirus and Morbillivirus, as indicated by the fact that three ␣-helices are consistently predicted in this region (data not shown). Conversely, in the case of Respirovirus XD, secondary structure predictions predict one ␤-strand followed by two terminal ␣-helices (27), thus suggesting a different folding.
Only one of the three faces of the domain surface of MV XD seems to be suited for binding an additional helix within N TAIL . The hydrophobic surface that is conserved within Paramyxovirinae XD appears to be a potent candidate for being the surface of interaction with N TAIL , which is enriched in charged residues and is unstructured in solution (16). The contact with a hydrophobic patch could be the driving force in the induced folding of N TAIL by burying apolar residues at the proteinprotein interface. The proposed model is a tentative description of a possible mode of interaction. Precise structural information on the molecular mechanism of the induced folding of N TAIL upon binding to XD can only be provided by the availability of the crystal structure of the complex. Nevertheless, it is noteworthy that interactions similar to those occurring in our model take place in many other protein-protein interactions, such as those involving heat-shock proteins (54) and the Zdomain of protein A (52), as well as within individual proteins. Indeed, when a search was performed for homologous proteins using the Dali server (48), eight proteins were retrieved. These proteins, although exhibiting no significant sequence identity with our model, possess a similar four-helix bundle arrangement, with some differences in the angles between the helices, resulting in root-mean-square deviation values ranging from 2.8 to 3.6 Å.
The surface buried in XD by the ␣-helix of N TAIL (800 Å 2 ), although being in the range typically observed in complexes between antibodies and proteic antigens (700 -1200 Å 2 ) (55), is smaller than that observed in the complex between the protein A Z-domain and the affibody (900 Å 2 ), suggesting a less stable complex. Keeping in mind that the contact between P and N TAIL within the replicative complex is dynamically made/ broken to allow the polymerase to progress along the nucleocapsid template during both transcription and replication, as well as to deliver N monomers to the nascent RNA chain, the complex cannot be excessively stable for this transition to occur efficiently at a high rate.
The structure of XD supports the working model of P being a multimer "joined at the hip" (multimerization domain of P) from which arms and legs (the flexible linker) are protruding and which has hands and feet (XD) at the extremities that make contact with N TAILS while cartwheeling over the surface of the nucleocapsid during transcription and replication (2).