Structural analysis of the human respiratory syncytial virus phosphoprotein: characterization of an {alpha}-helical domain involved in oligomerization

Human respiratory syncytial virus (HRSV), a member of the genus Pneumovirus of the subfamily Pneumovirinae within the family Paramyxoviridae, possesses a single-stranded RNA genome of negative polarity (Huang & Wertz, 1982) of 15 222 nt for the A2 strain (Mink et al., 1991). Similarities in gene order and certain gene products are observed between HRSV and related viruses of the order Mononegavirales. However, the pneumovirus genome, including that of HRSV, encodes some unique gene products, such as the non-structural NS1 and NS2 proteins implicated in counteracting the interferon response (Schlender et al., 2000; Spann et al., 2004), the 22 kDa (22K or M2-1) protein, which acts as a transcription anti-terminator factor (Collins et al., 1996), and the M2-2 protein implicated in modulating the switch between transcription and replication (Bermingham & Collins, 1999).

The mechanism of HRSV genome replication and transcription has been inferred to a great extent from data obtained with other mononegaviruses (for a review, see Collins et al., 2001). Thus, it is thought that the template for RNA synthesis is the ribonucleoprotein (RNP) complex made of viral RNA (the genome) and the nucleoprotein (N), and that the large RNA-dependent RNA polymerase, encoded by the L gene, requires the phosphoprotein (P) as an essential cofactor. These three proteins, L, N and P, are required for genome replication but, in addition, the HRSV 22K protein is needed for efficient transcription.

HRSV P is composed of 241 aa (see Fig. 1), much shorter than its counterparts from other paramyxoviruses. It is phosphorylated mainly at Ser-232 (Barik et al., 1995; Sanchez-Seco et al., 1995), although other minor phosphorylation sites have been identified (Navarro et al., 1991; Asenjo et al., 2005). The precise role of phosphorylation for P activity remains unclear.

(36K): Fig. 1. Modular organization of HRSV P. (a) Consistently predicted disordered regions are shown as narrow boxes, while the central structured region is shown as a large box. Segments of P containing epitopes recognized by four different monoclonal antibodies (1P, 76P, 021/2P and 021/12P) are shown above the protein diagram. The two predicted α-helices within the N- and C-terminal disordered regions are indicated. The locations of the trypsin-resistant fragment described in this work (X) and the L polymerase-binding domain (LBD) are indicated below the protein diagram. (b) Multiple sequence alignment of aa 100200 containing the multimerization domain within pneumovirus Ps. The consensus sequence (identity cut-off of >70 %) is shown under the multiple sequence alignment. Positions marked by dots correspond to residues under the cut-off. Residues corresponding to homology of >70 % are boxed. The numbers correspond to the amino acid position in the HRSV P sequence. Secondary structure elements consistently predicted are shown above the alignment. The hydrophobic cluster plot of the aa 100200 region is shown above the multiple sequence alignment. The coiled-coil region is underlined. Conventions are explained in the key. Structured regions are characterized by a high number of hydrophobic clusters, while unstructured regions are low in or devoid of hydrophobic clusters. BRSV, Bovine RSV; ORSV, ovine RSV.

HRSV P interacts with N (Garcia-Barreno et al., 1996) and the RNA polymerase (Khattar et al., 2001), and possibly with the 22K protein (Mason et al., 2003), playing a central role in the process of RNA synthesis. The picture emerging from studies with HRSV and related paramyxoviruses is that P interacts with newly synthesized N (N⁰) to prevent illegitimate assembly of the latter and to deliver it to the nascent chain during genome replication (Curran et al., 1995; Castagne et al., 2004). In addition, it has been proposed that Sendai and measles virus P cartwheel on the RNP template via simultaneous breaking and reforming of contacts with N, opening the RNP structure so that the polymerase, tethered by P, can reach the bases in the viral RNA (Curran, 1998; Johansson et al., 2003; Blanchard et al., 2004; Kingston et al., 2004). The oligomeric nature of P is central to its interaction with the RNP template via simultaneous binding of multiple arms of the P oligomer with the exposed C-terminal tails of the assembled N monomers. Recent structural studies of Sendai and rinderpest virus Ps have firmly established that these molecules are homotetramers (Tarbouriech et al., 2000a; Rahaman et al., 2004). The C-terminal halves of Sendai and rinderpest virus Ps, which contain the oligomerization domain and the sites for interaction with L and RNP, are tetrameric and have very elongated shapes due to the coiled-coil structure of their oligomerization domains (Tarbouriech et al., 2000b; Rahaman et al., 2004).

It has been proposed that HRSV P is a homotetramer, based on cross-linking studies with bifunctional reagents and the behaviour of the protein in gel filtration chromatography (Asenjo & Villanueva, 2000; Castagne et al., 2004). Using a series of GST- and His-tagged deletion mutants of P expressed in bacteria, the oligomerization domain of P was located between residues 120 and 150 (Castagne et al., 2004). We have used a complementary approach to identify and characterize the P oligomerization domain. Native P was purified and subjected to limited proteolysis. A trypsin-resistant fragment spanning residues 104163 was identified that was oligomeric, had a high α-helix content and behaved abnormally in size-exclusion chromatography. These properties resemble those of the oligomerization domains of Sendai and rinderpest virus Ps, suggesting that the P from these three viruses may share structural motifs, despite their differences in length and the absence of significant sequence similarity.

Bioinformatic analysis.
P sequences used in this study were obtained using BLASTP (Altschul et al., 1997) against the TrEMBL databases (Bairoch & Apweiler, 2000) and the sequence of HRSV Long strain as the query. Sequences and accession numbers were as follows: HRSV group A, strains Long, P12579 (Lopez et al., 1988), and A2, P03421 (Satake et al., 1984); HRSV group B, CH18537 strain, P24567 (Johnson & Collins, 1990); bovine RSV, P33454 (Mallipeddi & Samal, 1992); ovine RSV, Q83956 (Alansari & Potgieter, 1994). Sequence alignment was done using CLUSTAL W (Thompson et al., 1994) and drawn using ESPript 2.2 (Gouet et al., 1999). Secondary-structure predictions were performed with PSIPRED (McGuffin et al., 2000) and the Predict Protein server (Rost, 1996). The result presented is the consensus of both methods. Prediction of coiled-coil regions was carried out using the coiled-coil program (Lupas et al., 1991). Hydrophobic cluster analysis was done by using the program DRAWHCA (Callebaut et al., 1997). Amino acid composition and calculation of the ratio between mean hydrophobicity and mean net charge were carried out as described by Karlin et al. (2003). Prediction of intrinsically disordered regions was done using DisEMBL () (Linding et al., 2003a), DISOPRED () (Ward et al., 2004), GlobPlot () (Linding et al., 2003b) and PONDR () (Li et al., 1999) with the default parameters.

Cells and viruses.
CV-1 and HEp-2 cells were routinely grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 % fetal calf serum and antibiotics. The Long strain of HRSV was grown in HEp-2 cells as previously described (Garcia-Barreno et al., 1996).

The recombinant vaccinia virus VV-P expressing HRSV P (Long strain) was obtained by the method of Blasco & Moss (1995). Briefly, the P insert of the previously described LP-5 plasmid (Garcia et al., 1993a) was subcloned into plasmid pRB21 (a gift of R. Blasco, INIA, Madrid), which carries the vaccinia VP37 gene. CV-1 cells were infected with the vaccinia virus vRB12, which lacks the VP37 gene and is unable to form plaques, and transfected with pRB21/P. Recombinant vaccinia viruses were recovered by plaque formation in CV-1 cells. After three rounds of plaque purification, VV-P stocks were grown and titrated in CV-1 cells.

Purification of HRSV P.
The previously described anti-P monoclonal antibody 1P (Garcia et al., 1993b) was purified from ascitic fluid by protein ASepharose chromatography. Eight to 10 milligrams of this antibody was bound covalently to 1 mg of CNBr-activated Sepharose, following the manufacturer's instructions (Amersham), to produce Sepharose1P.

HEp-2 cell monolayers were infected with VV-P (m.o.i. of 0·5) using DMEM containing 2 % fetal calf serum. Forty-eight hours later, the cells were scraped into the medium and sedimented at low speed. Cell pellets were resuspended in extraction buffer [10 mM Tris/HCl (pH 7·4), 1·5 M KCl, 5 mM EDTA], sonicated at maximum amplitude for 10 min (Braun Labsonic sonicator) and cell extracts were clarified by centrifugation at 500 g for 20 min. The supernatants were mixed with Sepharose1P and left rotating overnight at 4 °C. The slurry suspension was then packed into a column and, after extensive washing with extraction buffer, P was eluted with 20 vols 0·1 M glycine/HCl (pH 2·5). Fractions were collected and neutralized with saturated Tris, and the presence of P was visualized by SDS-PAGE and Coomassie blue staining. P-containing fractions were pooled, concentrated and buffer exchanged [buffer A: 10 mM Tris/HCl (pH 7·5), 150 mM NaCl] using Vivaspin (pore exclusion size 10 kDa; Sartorius). Protein concentration was determined by UV absorbance at 280 nm, with a calculated extinction coefficient of 0·2 at 1 mg ml¹.

P eluted from the affinity column was loaded onto a Superose 6 HR 10/30 gel filtration column previously equilibrated with buffer A and eluted with this buffer. UV absorbance was monitored at 280 nm and the presence of P in the column fractions was revealed by SDS-PAGE and Coomassie blue staining. Column calibration was carried out with the molecular markers indicated in the figure legends.

Although the majority of the current work was done with protein obtained from extracts of VV-P-infected cells (because manipulation of vaccinia recombinants is easier than manipulation of HRSV), P was also purified for comparative purposes from extracts of HRSV-infected cells (m.o.i. of 12) following the procedures outlined in the previous paragraphs.

Trypsin digestion of P and purification of resistant fragments.
Purified P was digested with increasing amounts of either TPCK-trypsin (Sigma) or recombinant trypsin (Roche) for 1 h at 37 °C. At the end of the incubation period, the proteolytic products were analysed by SDS-PAGE and Western blotting with the antibodies indicated in the figure legends. For large-scale production of trypsin-resistant fragments, P was incubated with trypsin and the digestion products were separated in a gel filtration column (Superose 6 HR 10/30) equilibrated in buffer A. To separate two fragments that co-eluted from the gel filtration column, aliquots of the fragment-containing fractions were loaded into several lanes of an SDS-polyacrylamide gel. After electrophoresis, the outer lanes were cut and stained with Coomassie blue to visualize the bands. From the unstained central part of the gel, two slices were cut out across the positions of the trypsin-resistant bands. The gel slices were macerated and mixed with 10 vols 0·05 M NH₄HCO₃. The fragments were eluted by shaking overnight at 4 °C and gel residues were eliminated by centrifugation. Both fragments were lyophilized and resuspended in a small volume of distilled water.

Cross-linking.
P or the trypsin-resistant fragment were incubated for 1 h at room temperature with increasing amounts of either glutaraldehyde (Sigma) diluted in PBS or suberic acid diluted in 20 mM sodium phosphate buffer (pH 7·4). Reactions were quenched by adding electrophoresis sample buffer and the cross-linked products were analysed by SDS-PAGE and Western blotting.

N-terminal sequencing.
Samples were adsorbed onto glass fibre paper, mounted in the reaction chamber of an Applied Biosystems Procise sequenator and processed for 10 rounds of automated Edman degradation.

Circular dichroism (CD).
Samples were diluted with 0·1 M phosphate buffer (pH 7·0) and analysed in a Jasco 810 spectropolarimeter at room temperature. Spectra were recorded between 190 and 260 nm. Spectra of buffer alone were recorded and subtracted from protein spectra. Molar residue ellipticity values were calculated by using the spectral analysis of the Jasco Spectra Management software. The α-helix content was calculated by using the method of Morris et al. (1999).

Mass spectrometry (MS).
Samples were mixed with an aliquot of α-cyano-4-hydroxycinnamic acid (Bruker Daltonics) in 50 % aqueous acetonitrile and 0·15 % trifluoroacetic acid. This mixture was deposited on to the matrix-assisted laser desorption ionization (MALDI) probe and allowed to dry at room temperature. MALDI MS and MS² (LIFT-TOF/TOF; time of flight) mass spectra were measured on a Bruker Ultraflex TOF/TOF MALDI mass spectrometer (Bruker Daltonics) (Suckau et al., 2003). Mass measurements were performed in both positive ion reflector and linear modes using delayed extraction and a nitrogen laser (337 nm). The laser repetition rate was 50 Hz and the ion acceleration voltage was approximately 25 kV.

To gain sequence information from MALDI metastable decay (Suckau et al., 2003), selected precursor ions submitted to LIFT-TOF/TOF analysis were initially accelerated at 8 kV and thereafter mass filtered by a precursor ion selector. Ion packets formed from a precursor ion and daughter ions thereof were then accelerated in an 18·5 kV electric field and allowed to decay in the field-free region to measure the corresponding MALDI MS² spectra. The measured fragment ion masses were transferred through the MS BioTools program (Bruker Daltonics) as inputs to search the NCBInr database using Mascot software (Matrix Science).

Bioinformatic analysis of the HRSV P protein
We retrieved five pneumovirus sequences homologous to that of HRSV P, with an overall identity of 76 % and an overall similarity of 92 % (see Methods), and performed a multiple sequence alignment. Examination of the HRSV P sequence using hydrophobic cluster analysis indicated the presence of one structured domain (aa 100200), flanked by two disordered domains (aa 199 and 201241) (Fig. 1a). Hydrophobic cluster analysis carried out on all of the pneumovirus P sequences revealed the same overall modular organization as for HRSV P (data not shown). The two intrinsically disordered regions were also predicted by PONDR, DISOPRED, DisEMBL, GlobPlot and the net charge/hydrophobicity method (Longhi et al., 2003). Analysis of the deviation in amino acid composition of the disordered N- and C-terminal regions confirmed that they were both depleted in order-promoting residues (W, C, Y, V, I, L and F) and enriched in disorder-promoting residues (P, K, E, D and S) (data not shown). Notably, PSIPRED and Predict Protein both predicted two α-helices encompassing residues 1425 and 220228 of HRSV P. These predicted α-helices may correspond to regions undergoing induced folding upon binding to a partner/ligand, as in the case of the N terminus of morbillivirus and respirovirus P (Karlin et al., 2003).

Within the structured domain of P, the 130156 region displayed a distribution of hydrophobic clusters (Fig. 1b) that typifies coiled coils (Ferron et al., 2005). This observation is in agreement with the results provided by the coiled-coil program, which indicated a high score for the region spanning residues 125146 (using a window of 21 residues). This region overlaps the L polymerase-binding domain (LBD, aa 121160; Khattar et al., 2001) (Fig. 1a). Both properties, i.e. the occurrence of a coiled-coil structure and an LBD, are features also present in the oligomerization domain of Sendai and rinderpest virus P (Tarbouriech et al., 2000a; Rahaman et al., 2004).

Purification and characterization of HRSV P
In order to assess the predicted structure, HRSV P was purified by immunoaffinity chromatography from extracts of HEp-2 cells infected with the recombinant vaccinia virus VV-P, followed by gel filtration fast protein liquid chromatography (FPLC). The protein eluted from the gel filtration column as a homogeneous peak of ∼500 kDa (compared with globular protein markers, Fig. 2a), included in the bed volume of the column. SDS-PAGE and Coomassie blue staining (Fig. 2b) revealed the presence of a single band in the fractions with the highest UV absorbance, with an apparent molecular mass of ∼35 kDa, in agreement with previous reports (Garcia-Barreno et al., 1996). This apparent molecular mass slightly exceeded the mass estimated from the P amino acid sequence (27 147 Da), but it has been reported that HRSV P exhibits an anomalous electrophoretic mobility that is very sensitive to single amino acid changes (Caravokyri & Pringle, 1992).

(21K): Fig. 2. Gel filtration of HRSV P. (a) P was purified by immunoaffinity chromatography and loaded onto a Superose 6 HR 10/30 column that was equilibrated and eluted with buffer A (flowrate 0·3 ml min1; fraction size 1·2 ml). The column was calibrated previously with the following molecular markers whose elution positions are indicated by arrows: thyroglobulin (670 kDa), IgG (150 kDa) and BSA (68 kDa). (b) Coomassie blue-stained SDS-PAGE of the gel filtration fractions indicated below each lane. Note that the P band (denoted by an arrow) was most prominent in fractions that were slightly delayed with respect to the absorbance peak of the chromatogram, due to the dead volume between the UV cell and the fraction collector. Lane C, P loaded in the gel filtration column. The positions of molecular mass markers in kDa are shown on the left.

Based on cross-linking results with glutaraldehyde, it has been proposed that the native P is a tetramer (Asenjo & Villanueva, 2000; Castagne et al., 2004). Although the estimated mass of P from the elution profile in Fig. 2(a) greatly exceeded the theoretical mass of a tetramer, glutaraldehyde cross-linking of purified P from VV-P-infected cells (Fig. 3a) reproduced the cross-linked profile reported for P expressed either in HRSV-infected cells (Asenjo & Villanueva, 2000) or in bacteria (Castagne et al., 2004). Increasing amounts of glutaraldehyde generated major products that were compatible with dimers (7580 kDa) and tetramers of P (160180 kDa). As published previously, no major band corresponding to trimers was observed among the cross-linked products (Fig. 3a). Cross-linking of purified P with increasing amounts of suberic acid (Fig. 3b) reproduced essentially the glutaraldehyde results. Differences in the intensity and sharpness of bands generated by the two cross-linkers might be explained by self-polymerization of the glutaraldehyde, as previously reported in the case of Sendai virus P (Tarbouriech et al., 2000a). The same elution profile from the gel filtration column and similar cross-linked products were obtained with P purified from HRSV-infected cells (not shown).

(28K): Fig. 3. P cross-linking. Two micrograms of P eluted from the gel filtration column was incubated with glutaraldehyde (a) or suberic acid (b) (amounts indicated below each lane) for 1 h at room temperature. The reaction products were analysed by Western blotting with antibody 021/2P. Bands that may tentatively correspond to P monomers, dimers and tetramers are indicated on the right. The positions of molecular mass markers are shown on the left.

Thus, the results of Fig. 3 confirmed the oligomeric nature of P used in this study. Whether or not P is a tetramer may require confirmation by other methods, given the propensity of cross-linkers to generate artefacts and the abnormal behaviour of HRSV P in both SDS-PAGE (Caravokyri & Pringle, 1992) and gel filtration chromatography.

Castagne et al. (2004) reported that HRSV P produced in bacteria eluted from gel filtration columns between aldolase (150 kDa) and albumin (68 kDa) markers. However, we found that P consistently eluted as a homogeneous peak of ∼500 kDa, when purified from either vaccinia- or HRSV-infected cells. Other authors have reported elution of P from HRSV-infected cells as an oligomer with a mass higher than 150 kDa (Asenjo & Villanueva, 2000). Whether these differences are linked to the protein source or the column resin used in each study is still not known. It should be noted that abnormal elution from gel filtration columns has been reported for other paramyxovirus Ps (Tarbouriech et al., 2000a; Rahaman et al., 2004) and may reflect their hydrodynamic properties, as demonstrated for Sendai virus P (Tarbouriech et al., 2000a).

Identification of trypsin-resistant fragments in the native P
To gain insight into the structure of the HRSV P molecule, the protein eluted from the gel filtration column was digested with increasing concentrations of different proteases. The proteolytic products were identified by Western blotting using four different monoclonal antibodies that recognized epitopes located at the N (1P) or C (021/12P) terminus or within two internal segments (76P and 021/2P) of the P polypeptide (see Fig. 1a; Garcia et al., 1993b; Garcia-Barreno et al., 1996). Fig. 4(a) shows the reactivity patterns of P trypsin fragments revealed with the four antibodies mentioned above. Partial digestion products were observed at low trypsin concentrations with the four antibodies, but a fragment (X) resistant to the highest trypsin concentration was detected only with antibody 021/2P. The same fragment was observed after trypsin digestion of P purified from HRSV-infected cells (Fig. 4b). Incubation of P with 10 times more trypsin than the amount needed to generate the X fragment reduced the intensity of this fragment only slightly (Fig. 4b). Other proteases (papain, thermolysin and V8) generated fragments of sizes similar to that of X, which reacted with antibody 021/2P but not with other antibodies (not shown), indicating that this fragment may represent a P domain with limited access to proteases.

(15K): Fig. 4. Digestion of P with trypsin. (a) Five micrograms of purified P eluted from the gel filtration column (see Fig. 2) was incubated with the indicated amounts of TPCK-trypsin for 1 h at 37 °C. The digestion products were split into four parts, which were resolved by SDS-PAGE and revealed by Western blotting with the monoclonal antibodies indicated above each panel. (b) The same amount of purified P, obtained from HRSV-infected cells, was treated with the indicated amounts of trypsin and analysed by Western blotting. A trypsin-resistant fragment reacting with antibody 021/2P is indicated (X).

The products of a large-scale trypsin digestion of P were subjected to gel filtration FPLC (Fig. 5a). The X fragment eluted in fractions that coincided with a homogeneous low-absorbance peak of the chromatogram, as revealed by Western blotting (Fig. 5b). This peak was ahead of a heterogeneous high-absorbance peak containing trypsin, as revealed by SDS-PAGE and Coomassie blue staining (Fig. 5c). The estimated molecular mass of the X fragment (∼120 kDa) from the gel filtration chromatogram greatly exceeded the mass estimated by SDS-PAGE (∼12 kDa). Although Western blotting detected only the X fragment, another smaller fragment (Y, ∼8 kDa), co-eluting with the former but not recognized by the 021/2P antibody, was detected by SDS-PAGE and Coomassie blue staining (Fig. 5c). The low absorbance at 280 nm of the 120 kDa peak in the chromatogram of Fig. 5(a), compared with the trypsin peak, is due to the lack of Trp and the low Tyr content of the X and Y fragments (see below).

(25K): Fig. 5. Purification of the trypsin-resistant fragment. (a) Purified P (1·6 mg) was digested with 200 µg TPCK-trypsin for 1 h at 37 °C. The digestion products were loaded onto a Superose 6 HR 10/30 column, calibrated as indicated in Fig. 2(a). UV absorbance was monitored at 280 nm. (b, c) Aliquots of the column fractions indicated below each lane were resolved by SDS-PAGE and revealed by either Western blotting with antibody 021/2P (b) or Coomassie blue staining (c). The locations of bands corresponding to fragments X and Y of P are indicated, as well as the location of the trypsin band visualized by Coomassie blue staining.

To identify the boundaries of fragments X and Y, N-terminal sequencing of the material eluted from the gel filtration column was performed. Two sequences were identified. The relative abundance of the two sequences correlated with the abundance of X and Y fragments in the fragment preparation (Fig. 6a). One of the sequences started after Lys-103 and the other after Tyr-118. To assign both sequences unambiguously, the two fragments were eluted individually from an unstained gel. After confirming the identity of both fragments by SDS-PAGE and Coomassie blue staining (Fig. 6b), 10 N-terminal sequencing cycles were done with the separated fragments. The X fragment yielded the sequence 104-ETIETFDNNE and the Y fragment the sequence 119-SYEEINDQTN. These two sequences are closely spaced in the P protein primary structure (see below), indicating that the two fragments overlapped extensively. The X fragment probably represents an authentic tryptic product since its N terminus started immediately after a Lys residue in the P protein sequence. Conversely, the Y fragment is probably a by-product of a chymotryptic activity that frequently contaminates trypsin preparations. This is consistent with the Y fragment starting after Tyr-118, which is a chymotrypsin target site.

(22K): Fig. 6. Characterization of the X and Y fragments of P by N-terminal sequencing and MS. P fragments eluted from the gel filtration column of Fig. 5(a) were separated by SDS-PAGE (a) and eluted from an unstained gel as indicated in Methods. The separated fragments were resolved by SDS-PAGE and stained with Coomassie blue (b). Aliquots of these fragments were used for 10 cycles of automatic Edman degradation. The sequences obtained are shown on the right. (c) Mass spectra of the material from fraction 12 of the gel filtration chromatogram of Fig. 5 (X+Y) and of the separated X and Y fragments shown in (b). (d) Amino acid sequences of the X and Y fragments. The partial sequences determined by N-terminal sequencing are shown in bold. The sequence of the Y fragment is shown in italics to denote that it was determined experimentally by MS, whereas the sequence of the X fragment was inferred from its mass (see text for explanation).

The material from the low-absorbance peak from the gel filtration column (Fig. 5) was then analysed by linear mode MALDI-TOF. Several molecular species were identified in the 47 kDa range, but not at higher masses, with two major peaks corresponding to protonated mean molecular masses of 4959 and 6748 Da (Fig. 6c). The mass spectrum of the X fragment eluted from the SDS-polyacrylamide gel (Fig. 6c) had a dominant peak corresponding to a protonated mean mass of 6750 Da and an adduct by-product of 6831 Da that was also visible in the spectrum of the unseparated X+Y fragments (6829 Da). The mass spectrum of the Y fragment eluted from the gel had a major peak corresponding to a protonated mean mass of 4961 Da (Fig. 6c). Thus, the masses of the peaks identified in the mass spectra of the separated X and Y fragments correlated with the masses of the major peaks in the spectrum of the P fragments eluted from the gel filtration column.

The MALDI spectrum of the low-absorbance peak from the gel filtration column was also recorded in reflectron mode to improve resolution in the lower-mass range. This spectrum showed two dominant peaks: a doubly charged ion at m/z=2478·75 and its corresponding singly charged ion at m/z=4955·49 (data not shown). A LIFT-TOF/TOF MS² spectrum could be successfully registered for the doubly charge species (not shown) that yielded the sequence indicated in Fig. 6(d), starting at Ser-119 and ending at Arg-163, after fragment ion database searching. The molecular mass of this fragment, calculated from its amino acid sequence (4959 Da), was in very good agreement with the experimental value determined by linear MALDI-TOF for fragment Y (49594961 Da). However, it was considerably lower than the molecular mass determined by SDS-PAGE (∼ 8 kDa) (Fig. 5c).

The large mass of the X fragment prevented MALDI MS² fragmentation analysis. However, the mass estimated for this fragment by linear MALDI-TOF (67486750 Da) was in very good agreement with the protonated mean mass calculated for a fragment starting at Glu-104 (coinciding with its N-terminal sequence) and ending at Arg-163 (6748 Da). Again, the molecular mass of the X fragment determined by MS was considerably lower than the mass estimated by SDS-PAGE (∼ 12 kDa). Thus, the two fragments resulting from the P trypsin digestion had overlapping sequences, except for an N-terminal 15 aa extension of fragment X (Fig. 6d). Since antibody 021/2P reacted with the X fragment but not with the Y fragment, its epitope must include some of the N-terminal residues of the larger fragment.

Characteristics of the P trypsin-resistant fragments
To avoid generation of the Y fragment, which might confound the results of structural studies (see later), a new preparation of the X fragment was made using a recombinant trypsin devoid of chymotrypsin contamination. Absence of the Y fragment in this new preparation was confirmed by SDS-PAGE (Fig. 7a), sequence analysis and MS (not shown). To test the oligomeric state of the X fragment, the material from the ∼ 120 kDa peak of the gel filtration chromatogram (Fig. 7a) was cross-linked with glutaraldehyde (Fig. 7b) or suberic acid (Fig. 7c) and the resulting products were analysed by Western blotting. Increasing amounts of the bifunctional reagents generated two major products that reacted with antibody 021/2P (indicated by asterisks in Fig. 7), indicative of oligomeric forms of fragment X. However, the abnormal electrophoretic behaviour of fragment X (see above) precluded the assignment of unambiguous stoichiometry for the cross-linked products.

(28K): Fig. 7. Cross-linking of the X fragment. Preparation of the trypsin-resistant fragment was as described in Fig. 5, except that recombinant trypsin devoid of contaminating chymotrypsin was used to digest P. One or five micrograms (as indicated) of the material eluted in the low-absorbance peak of the gel filtration column was analysed by SDS-PAGE and Coomassie blue staining (a). A 0·1 µg sample of this material was treated with the indicated amounts of either glutaraldehyde (b) or suberic acid (c) for 1 h at room temperature and the reaction products were resolved by SDS-PAGE and revealed by Western blotting with antibody 021/2P. Asterisks denote the major cross-linked products observed with the two reagents.

Analysis of the new X fragment preparation by CD indicated a high α-helical content (Fig. 8). When compared with the CD spectrum of P, the trypsin-resistant fragment exhibited more pronounced minima at 208 and 222 nm, characteristic of α-helices. The CD spectra of Fig. 8 were very similar to the spectra reported by Tarbouriech et al. (2000a) for Sendai virus P and for its oligomerization domain. Based on the ellipticity values at 222 nm of the CD spectra shown in Fig. 8, the α-helical content of HRSV P and of the trypsin-resistant fragment was estimated to be 27·4 and 43·3 %, respectively (values are representative of two independent determinations). Thus, the results of Figs 7 and 8 supported the notion that the trypsin-resistant fragment identified in this work represents the α-helical oligomerization domain of the native P molecule. This fragment encompasses the N-terminal two-thirds of the predicted structured domain of P, including the predicted coiled-coil region (Fig. 1a).

(13K): Fig. 8. Far-UV CD spectrum of the X fragment shown in Fig. 7(a) (broken line) and purified P (solid line).

Bioinformatic analysis predicted the existence of a central structured domain in HRSV P, flanked by two intrinsically disordered regions. In agreement with this modular organization, a large segment of the structured domain, recognized by antibody 021/2P, was shown to be resistant to trypsin. Similar fragments reacting with antibody 021/2P were generated by other proteases, such as papain, thermolysin and Staphylococcus aureus V8 protease. N-terminal sequencing and MS set the limits of the trypsin-resistant fragment (X) between aa 104 and 163. The fact that the X fragment could be cross-linked indicated that it represents the oligomeric domain of the native P molecule. A high α-helical content was calculated for the X fragment from its CD spectrum, in agreement with the prediction of a coiled-coil structure (see Fig. 1b), which was included in the X fragment sequence. Interestingly, the X fragment is preceded by a predicted disordered region of P (aa 199) and is followed by two predicted α-helices (α3 and α4, Fig. 1b). Thus, the X fragment may adopt a highly structured α-helical conformation in the native molecule bordered by regions that contain sites accessible to certain proteases. A second fragment (aa 119163, fragment Y), which co-eluted with the X fragment from the gel filtration column, was probably generated by a contaminating chymotryptic activity, indicating that there may be subregions within the X fragment with sites sensitive to cleavage by other proteases. MS analysis of fragment Y facilitated the unambiguous identification of the fragment borders in the P primary structure (Fig. 6).

The location of the X fragment in the P polypeptide is in agreement with the results reported by Castagne et al. (2004) for the location of the HRSV P oligomerization domain. These authors expressed a series of GST-fused deletion mutants of P in E. coli and tested their capacity to oligomerize with co-expressed HisP by pull-down assays. The boundaries of the P oligomerization domain that they identified (aa 120150) were shorter than those of the X fragment (aa 104163). This difference may reflect the different type of information provided by the complementary methodologies used in the two studies. Whereas the approach of Castagne et al. (2004) searched for sequences with minimal requirements for oligomerization with P, which may or may not define the domain limits, our method attempted a more classical approach of identifying domains in the folded P by resistance to protease digestion. These methodological differences may be important when attempting to crystallize the oligomerization domain of P for structure determination. In addition, our approach gave some structural information about the oligomerization domain in the native P by its accessibility to proteases.

Tarbouriech et al. (2000a) reported the isolation and characterization of a trypsin-resistant fragment from the C-terminal half of Sendai virus P that included its oligomerization domain. In the case of Sendai virus P, high-resolution X-ray diffraction analysis of the oligomerization domain revealed a homotetrameric coiled-coil structure (Tarbouriech et al., 2000b). The high α-helical content of the trypsin-resistant fragment of HRSV P (and the predicted coiled coil) suggests that it may adopt a conformation similar to that of the Sendai virus P oligomeric domain. It is worth mentioning that the calculated α-helical content of the trypsin-resistant fragment of Sendai virus P from CD analysis was 52 %, whereas the α-helices accounted for 83 % of the residues in the tetrameric crystal structure (Tarbouriech et al., 2000b). Thus, the α-helical content of HRSV P, calculated from the CD spectrum of Fig. 8 (43·3 %), may also have been underestimated. Although the trypsin-resistant fragment of Sendai virus P contains 126 aa, the C-terminal segment that forms the coiled-coil structure is only 65 aa (Tarbouriech et al., 2000b). This size is only slightly larger than that of the HRSV P trypsin-resistant fragment (60 aa). Interestingly, in both cases, the oligomerization domain overlaps the region involved in the interaction with the polymerase (Curran, 1998; Khattar et al., 2001).

The P polypeptide of viruses belonging to the subfamily Pneumovirinae (including HRSV) is almost half the size of its counterparts in viruses of the subfamily Paramyxovirinae and there is no sequence similarity between Ps of the two subfamilies (Karlin et al., 2003). Smaller size differences exist between Ps of the subfamily Paramyxovirinae and these differences are located mainly in the N-terminal half, which contains large disordered regions (Karlin et al., 2003). In contrast, the C-terminal half of Ps in the subfamily Paramixovirinae is more conserved in sequence and contains the oligomerization domain followed by a linker region and by the region of interaction with the RNP complex (PX domain) (Johansson et al., 2003; Karlin et al., 2003). It is noteworthy that the site of interaction with the RNP complex has also been mapped to the C-terminal end in the case of HRSV P (Garcia-Barreno et al., 1996). Thus, it seems that the P of members of the subfamily Pneumovirinae also has a modular structure, as proposed for the Paramyxovirinae, but with a shorter N-terminal half. These structural similarities may underscore conservation of similar mechanisms of action of P in the processes of transcription and replication of different paramyxovirus genomes.

We thank Bruno Canard, Jean-François Eléouët and Felix Rey for support and useful discussions. This work was funded in part by the European Commission (contract no. QLK2-CT2001-01225), under the specific programme Quality of Life and Management of Living Resources. It does not necessarily reflect its views and in no way anticipates the Commission's future policy in this area. Access to PONDR was provided by Molecular Kinetics (IUETC, 351 West 10th Street, Suite 318, Indianapolis, IN 46202 USA; main@molecularkinetics.com). PONDR is copyright ©1999 of the WSU Research Foundation, all rights reserved.