The negative regulator of Borna disease virus polymerase is a non-structural protein

Borna disease virus (BDV) is a non-segmented, negative-strand RNA virus that replicates in the central nervous system of a broad range of warm-blooded animals (Ludwig et al., 1988; Staeheli et al., 2000). Natural and experimental infections with BDV usually result in immune system-mediated neurological diseases including behavioural abnormalities (Hornig et al., 2003; Pletnikov et al., 2002). BDV infection is potentially linked to human psychiatric diseases, as BDV-specific antibodies were identified in sera of psychiatric patients with higher prevalence than in control cohorts (Jordan & Lipkin, 2001; Planz et al., 2002). Attempts to confirm human BDV infection by non-serological methods, such as detection of viral nucleic acid by nested RT-PCR or virus isolation, have given inconsistent results and therefore this issue remains controversial (Jordan & Lipkin, 2001; Planz et al., 2002).

In contrast to other members of the Mononegavirales, BDV replicates in the nucleus of infected cells and uses the splicing machinery for maturation of viral transcripts (Briese et al., 1992; Cubitt et al., 1994; Schneider et al., 1994). It encodes at least six viral proteins: the nucleoprotein (N), negative regulator (X), phosphoprotein (P), matrix protein (M), glycoprotein (G) and polymerase (L) (Briese et al., 1994). Whereas M and G are involved in particle formation, protein-binding studies suggest that P, N, L and X constitute the polymerase complex. P forms homo-oligomers and can act as a scaffold protein for polymerase complex formation (Schneider et al., 2004; Schwemmle et al., 1998; Wolff et al., 2000). The interaction domain for X is localized to the N-terminal half of P, whereas the oligomerization and L- and N-binding domains are found in the C-terminal half of P (Schneider et al., 2004; Schwemmle et al., 1998; Wolff et al., 2000). Data from a BDV-specific mini-replicon system has challenged the view that X is essential to reconstitute an active polymerase complex, since N, P and L alone were sufficient to achieve replication and transcription (Perez et al., 2003; Schneider et al., 2003). Furthermore, additional expression of X resulted in a pronounced inhibition of polymerase activity (Perez et al., 2003; Schneider et al., 2003) and thus identified this protein as a negative regulator of the polymerase.

The precise mechanism by which X regulates viral polymerase activity is unclear. It is believed that complex formation between X and P prevents the formation of an active polymerase complex in the nucleus. This hypothesis is based on the observation that co-expression of X and P partially results in cytoplasmic accumulation of both proteins (Kobayashi et al., 2003; Poenisch et al., 2004). A cytoplasmic localization of XP complexes is also frequently observed in MadinDarby canine kidney (MDCK) cells persistently infected with BDV (Kobayashi et al., 2003), supporting this hypothesis. Furthermore, P is only found in the nucleus of BDV-infected MDCK cells that lack detectable expression of X (Kobayashi et al., 2003). However, others found a co-localization of X and P in the nucleus of C6 cells persistently infected with BDV (Schwemmle et al., 1998), suggesting that the cytoplasmic retention of P by X is most likely not the only mechanism by which X regulates the polymerase activity. Recent observations that P multimers can bind simultaneously to X and L (Schneider et al., 2004) suggest that ribonucleoprotein (RNP)-bound X could modulate the viral polymerase activity in the nucleus as well.

Data from the BDV mini-replicon assay has revealed that BDV-X can inhibit polymerase activity by 30 % at X : P plasmid ratios of 1 : 6 and almost completely when equimolar amounts are used (Schneider et al., 2003). We therefore hypothesized that the X : P protein ratio is low in persistently infected cells to maintain virus replication and that X is not efficiently incorporated into viral particles, allowing efficient polymerase activity early after infection. To determine unequivocally the subcellular localization and expression levels of X in infected cells, we generated a monoclonal antibody (mAb) against this protein. mAb 10/1G3 specifically recognized a linear epitope of X (⁷⁰PLHDLRPRP⁷⁸) and revealed a co-localization of X with P and N in the nucleus and cytoplasm of BDV-infected cells. Based on Western blot analysis, the ratio of X : P : N in crude cell extract was found to be 1 : 6 : 40. Only traces of X were found in concentrated virus stocks, corresponding to an X : P ratio of 1 : 330. Thus, the X protein represents a non-structural protein and cannot interfere with virus replication steps early in infection, whereas the level of X in persistently infected cells is most likely sufficient to exert partial inhibition of the polymerase activity. We propose that X regulates the activity of P directly in the nucleus and not solely by retranslocation into the cytoplasm.

Generation of antibodies.
To generate a mAb directed against the X protein, female BALB/c mice were immunized intraperitoneally (i.p.) with 20 µg recombinant GSTX protein diluted in 200 µl 50 % complete Freund's adjuvant in PBS (Sigma). After 2 and 4 weeks, the mice received a second and third i.p. injection of 20 µg recombinant GSTX diluted in 200 µl 50 % incomplete Freund's adjuvant in PBS (Sigma). Six weeks after the initial i.p. immunization, mice were boosted intravenously with 20 µg maltose binding protein (MBP)X fusion protein diluted in 100 µl PBS. Four days later, the spleen was taken and fused with the mouse myeloma cell line X63-Ag8.653 (Kearney et al., 1979). Two weeks later, supernatants of hybridomas were tested in Western blot analyses for reactivity against recombinant His-tagged X protein and extract from BDV-infected human oligodendrocyte (Oligo) cells. Hybridomas producing reactive supernatant were expanded and cells were frozen at 80 °C for further analysis. Hybridoma 10/1G3 was subcloned three times to ensure a clonal origin. Antibody is available on request from O. Planz (Friedrich Loeffler-Institut, Tübingen, Germany).

To generate an M-specific antibody, New Zealand white rabbits were immunized subcutaneously (s.c.) with 200 µg peptideKLH (M35; N₂H-NQFLNIPFLSV-COOH) in complete Freund's adjuvant (Sigma). After 3, 6 and 9 weeks, the rabbits received a second, third and fourth s.c. injection of 200 µg peptideKLH in incomplete Freund's adjuvant (Sigma). Twelve weeks after the initial immunization, blood was taken from the rabbits. The specificity of this antibody was verified by Western blot analysis (data not shown).

Plasmids.
pET15b-based expression plasmids (Novagen) encoding His-tagged BDV N (pHis-N) and P (pHis-P) have been described previously (Schwemmle et al., 1997, 1998). To generate a His-tagged X expression plasmid (pHis-X), the complete ORF of X (BDV strain He/80) was amplified from plasmid pTRE-X (Schwemmle et al., 1998) and cloned into the NdeI/BamHI restriction sites of pET15b. In a second step, nucleotide exchanges (⁴⁹AAT⁵¹ to ⁴⁹GCG⁵¹) were introduced by PCR mutagenesis as described by the manufacturer (Invitrogen). The nucleotide exchanges resulted in a single amino acid substitution of the X protein (N17A). The bacterial expression plasmid pMAL-X encoding an MBPX fusion protein was constructed by inserting BDV X from pGEX-p10 (Schneider et al., 2004; Schwemmle et al., 1998; Wolff et al., 2000) cDNA into pMAL-c2 (New England Biolabs).

Protein preparation.
Purification of His-tagged proteins from Escherichia coli was carried out as described previously (Schwemmle et al., 1997). Briefly, cells were suspended in 50 mM Tris/HCl, pH 8·0, 5 mM MgCl₂, 500 mM NaCl, 10 % glycerol, 20 mM imidazole and 200 µM Pefabloc (Roth), disrupted by sonication, bound to a Ni-NTA agarose column (Qiagen) and stepwise eluted with 20 mM Tris/HCl, pH 8·0, 5 mM MgCl₂, 100 mM NaCl, 10 % glycerol and 500 mM imidazole. Fractions containing His-tagged viral proteins were pooled and stored at 80 °C. The purity of the His-tagged proteins was >95 % as judged by Coomassie blue staining. MBPX fusion protein was purified from bacterial lysates by affinity chromatography on an amylose resin (New England Biolabs) following the instructions of the manufacturer.

Western blot analysis and quantification of protein levels.
BDV-infected or uninfected cells were pooled and lysed in gel loading buffer (Laemmli & Favre, 1973), followed by ultrasonication. Protein extracts were size fractionated by 15 % SDS-PAGE and blotted on to a PVDF membrane (Millipore) for Western blot analysis. The membrane was blocked with milk powder (2 %, w/v, in PBS) for 2 h and then incubated with the indicated antibodies in PBS containing 0·2 % milk powder (w/v) overnight. After intense washing in PBS containing 0·1 % Tween 20 (Sigma), the blot was incubated with a 1 : 2000 dilution of a peroxidase-coupled donkey anti-mouse or anti-rabbit polyclonal antiserum (Dianova) for 1 h at room temperature. Finally, bound enzymic activity was detected using the enhanced chemiluminescence system (ECL+) from Amersham. Signal intensities of the virus-encoded proteins and the known amounts of the E. coli-purified His-tagged marker proteins were determined using ChemiDoc and the software package Quantity One (both from Bio-Rad). Based on these values and the molecular masses of the marker proteins (HisX, 11 421 Da; HisP, 24 479 Da; HisN, 42 985 Da), the amount of virus-encoded protein, as well as the X : P : N ratio, was determined.

Immunofluorescence analysis.
Immunofluorescence analysis was carried out as described previously (Geib et al., 2003) by applying the primary antibodies rabbit anti-N (Geib et al., 2003), rabbit anti-P (Geib et al., 2003) and mAb 10/1G3, using a laser scanning microscope (Zeiss).

Peptide array analysis.
Peptide arrays composed of overlapping 15mer fragments with an offset of 3 aa residues representing the protein sequences of X, P, N and M of strain He/80 (Pleschka et al., 2001) were chemically synthesized on cellulose sheets by the spot-synthesis technique as described previously (Frank, 1992). Probing the arrays with antibodies was essentially carried out as described by Frank (1992). Briefly, peptide arrays were blocked with membrane-blocking buffer overnight (Sigma-Genosys), followed by incubation with the corresponding antibody (1 : 1000 dilution) in blocking buffer for 3·5 h at room temperature. After three washes with Tris-buffered saline containing 0·05 % Tween 20, the peptide-bound antibodies were detected by species-specific alkaline phosphatase-conjugated IgG antibodies, which were visualized by the blue-coloured precipitate formed from the BCIP/MTT substrate as described previously (Frank, 1992). Signal patterns on the membranes were subsequently scanned for documentation.

Virus stock preparation and titration.
Virus stocks were prepared from Oligo cells persistently infected with BDV strain He/80 as described previously (Briese et al., 1992), titrated on Vero cells (Hallensleben et al., 1998) and concentrated by two steps of ultracentrifugation at 100 000 g for 1 h.

mAb 10/1G3 recognizes a linear epitope of BDV X
To detect X in BDV-infected cells we generated a mAb, designated 10/1G3, directed against this protein. In a peptide array analysis, the antibody recognized three neighbouring peptide spots corresponding to a linear epitope in the C terminus of X (⁷⁰PLHDLRPRP⁷⁸; Fig. 1a, left panel). mAb 10/1G3 did not react with N, P or M (Fig. 1a, left panel). A cocktail of mAb 10/1G3, monospecific antibodies directed against N and P (Geib et al., 2003) and a peptide-specific antibody directed against M verified the integrity of these peptide arrays (Fig. 1a, right panel).

(41K): Fig. 1. mAb 10/1G3 specifically recognizes a linear epitope of BDV X. (a) Arrays of overlapping membrane-bound 15mer peptides (with an offset of 3 aa residues) representing the entire amino acid sequence of BDV X, N, P and M were incubated with mAb 10/1G3 (left panel) or a mixture of mAb 10/1G3, monospecific sera directed against N (anti-N) and P (anti-P) and a peptide-specific antibody directed against M (M35) (right panel). Peptide-bound antibodies were identified with phosphatase-conjugated secondary antibodies and visualized with BCIP/MTT substrates leading to staining of positive spots. Bold letters indicate peptides recognized by mAb 10/1G3. (b) Cell extracts of BDV-infected (lanes 1, 3 and 5) or uninfected (lanes 2, 4 and 6) Oligo cells were analysed by Western blotting using mAb 10/1G3 (lanes 1 and 2), a rabbit anti-N polyclonal antiserum (lanes 3 and 4) or a rabbit anti-P polyclonal antiserum (lanes 5 and 6). The positions of X, P and the two isoforms of N are indicated.

In a Western blot analysis using total cell extract of Oligo cells persistently infected with BDV strain He/80 (Pleschka et al., 2001), mAb 10/1G3 recognized a major band with a molecular mass of ∼16 kDa, whereas no signal was observed in cell extract from uninfected Oligo cells (Fig. 1b, lanes 1 and 2). Although the calculated molecular mass of X is only 9433 Da (Pleschka et al., 2001), this band at 16 kDa most likely represents the X protein. BDV X is known to migrate slower in denaturing SDS-PAGE (Wehner et al., 1997; Kobayashi et al., 2003). Furthermore, mAb 10/1G3 recognized a 16 kDa protein after transient expression of X in 293T cells (data not shown). This atypical migration is independent of the phosphorylation status of X, since E. coli-purified His-tagged X, with a molecular mass of 11 421 Da, migrated at approximately 17 kDa (see Fig. 3a). The monospecific antibodies directed against N and P specifically recognized their corresponding antigens of the expected sizes (Fig. 1b). These results demonstrated that mAb 10/1G3 specifically recognized X in virus-infected cells by binding to a linear epitope of this protein.

(36K): Fig. 3. Levels of X in BDV-infected cells and virus stocks. (a) Determination of the detection limit of His-tagged X protein by mAb 10/1G3. Western blot analysis was performed using the indicated quantities of E. coli-purified His-tagged X (HisX) and X-specific mAb 10/1G3. Note that the detection limit was 600 pg (inset). M, Marker lane. (b) The amount of X was determined in cell extracts from BDV-infected Oligo cells (CE/BDV, 2x105 cells) or virus stock corresponding to 2·56x103 f.f.u. by Western blot analysis using mAb 10/1G3 and the indicated amounts of E. coli-purified His-tagged X. CE, Cell extract from uninfected Oligo cells (2x105 cells); M, marker lane. (c) The amount of P was determined as described in (b), using cell extract from 2x103 BDV-infected Oligo cells, virus stock corresponding to 320 f.f.u., the indicated amounts of E. coli-purified His-tagged P and a rabbit polyclonal antiserum recognizing P. (d) The amount of N was determined as described in (b) using cell extract from 2x103 BDV-infected Oligo cells, virus stock corresponding to 320 f.f.u., the indicated amounts of E. coli-purified His-tagged N (2-fold dilution steps) and a rabbit polyclonal antibody recognizing N.

X co-localizes with N and P in the nucleus of BDV-infected Oligo cells
We next analysed the subcellular localization of X, P and N in BDV-infected cells by confocal microscopy using mAb 10/1G3 and the above-mentioned monospecific antibodies directed against N and P. All three antibodies identified BDV antigen in the nucleus as well as in the cytoplasm (Fig. 2a, left panels). No specific signals were observed in uninfected cells (Fig. 2a, right panels). The nuclear punctuate staining of X was clearly visible in almost all infected cells and was similar to the signals observed with N and P, suggesting that these proteins co-localize in the nucleus. Double immunofluorescence analysis confirmed a co-localization of X with N (Fig. 2b, upper panel) and X with P (Fig. 2b, lower panel) in the nucleus, as well as in the cytoplasm. These findings are in line with previous observations that X co-localizes with P and N in C6 cells persistently infected with BDV (Schwemmle et al., 1998). Others, however, have observed a cytoplasmic accumulation of X and P in BDV-infected MDCK cells using a polyclonal X antiserum (Kobayashi et al., 2003). The reason for this discrepancy is unclear, but may reflect differences in the nature of the antibodies used in these studies.

(60K): Fig. 2. Subcellular localization of X, P and N in cells persistently infected with BDV. (a) The cellular localization of X, P and N in BDV-infected (Oligo/BDV) and uninfected Oligo cells (Oligo) was analysed by immunofluorescence using mAb 10/1G3 recognizing X (upper panels) and rabbit polyclonal antibodies recognizing either P (middle panels) or N (lower panels). (b) Double immunofluorescence analysis of the subcellular localization of X and N (upper panels) and X and P (lower panels). Note that X co-localizes with N and P in the nucleus as well as in the cytoplasm.

Levels of X protein in BDV-infected cells and virus stocks
To quantify the levels of X, P and N protein in persistently BDV-infected Oligo cells or virus stocks, we performed Western blot analysis using known quantities of E. coli-purified His-tagged X (HisX), P (HisP) and N (HisN) as protein standards. Detection limits were found to be approximately 600 pg HisX for mAb 10/1G3 (Fig. 3a), 40 pg HisP for antibody against P and 171 pg HisN for antibody against N (data not shown). Approximately 5 ng X could be detected in total extracts of infected cells, the equivalent of about 2x10⁵ cells (Fig. 3b). Due to the high abundance of N and P in this cell extract, 100-fold fewer cells (2x10³) were used to facilitate quantification. By comparison with protein standards, these analyses revealed 600 pg P (Fig. 3c) and 7 ng N (both isoforms together; Fig. 3d), resulting in a relative molecular ratio of X : P : N in infected cells of 1 : 6 : 40. Using BDV virus stock corresponding to 2·56x10³ focus-forming units (f.f.u.), we failed to detect X (Fig. 3b), whereas N and P could readily be detected (Fig. 3d). However, threshold levels of X were detectable in virus stock samples corresponding to 2·56x10⁵ f.f.u. (Fig. 4) and were estimated to represent 600 pg X protein. Thus, the calculated relative molecular ratio of X : P : N in BDV virus stock was 1 : 330 : 1130. Based on this ratio, the length of the BDV genome of 9810 nt (Pleschka et al., 2001) and the assumption that nucleoproteins of non-segmented, negative-strand RNA viruses cover on average 69 nt of the viral genome (Bhella et al., 2002; Calain & Roux, 1993; Thomas et al., 1985), we estimated that approximately 11·5 molecules of the X protein were present in one BDV virus particle. However, we could not exclude the possibility that the threshold levels of X detected in BDV virus stock reflected contamination with cell extract due to the preparation procedure.

(49K): Fig. 4. Detection of traces of X in concentrated BDV virus stock. Western blot analysis was performed with cell extract from BDV-infected (CE/BDV, 2x105 cells) or uninfected (CE, 2x105 cells) Oligo cells and virus stock corresponding to the indicated f.f.u. X was identified using mAb 10/1G3. Inset, longer exposure of lanes 13.

To study the expression of X during acute infection, we challenged Vero cells with BDV strain He/80 (0·1 f.f.u. per cell) and analysed the appearance and localization of this protein by immunofluorescence assay at 3, 4 and 5 days p.i. As judged by immunofluorescence assay, using the P-specific serum, approximately 10 % of the cells were infected with BDV at 3 days p.i. and the infection increased slightly to 12 % by 5 days p.i. (data not shown). By 3 days p.i., X was clearly detectable in the nucleus of the infected cells, co-localizing with P (Fig. 5a, upper panels). A similar co-localization of X and P was found in Vero cells persistently infected with BDV strain He/80 (Fig. 5a, lower panels). Despite the presence of X during acute infection, the level of this protein at 3, 4 and 5 days p.i. was too low to be detected by mAb 10/1G3 in Western blot analysis using a cell extract corresponding to 2x10⁵ cells (Fig. 5b, upper panel), whereas approximately 10 ng X could readily be detected using a cell extract from Vero cells persistently infected with BDV (Fig. 5b, upper panel). As expected from the low infection rate, the level of P in cell extract from acutely infected cells was significantly lower than in persistently infected cells (Fig. 5b, lower panel). These results indicated that both X and P are expressed during acute infection, albeit at lower levels.

(35K): Fig. 5. Subcellular localization and levels of X, P and N in acute and persistently infected Vero cells. (a) Double immunofluorescence analysis of the subcellular localization of X (left panels) and P (middle panels) and X and P images merged (right panels) in Vero cells at 3 days p.i. (upper panels) and persistently infected Vero cells (lower panels). (b) Western blot analysis was performed with cell extract from 2x105 BDV-infected Vero cells (CE/BDV), uninfected Vero cells (CE) and Vero cells infected for 3, 4 and 5 days with BDV [CE/BDV (acute)] using mAb 10/1G3 recognizing X (upper panel) and the P-specific rabbit polyclonal serum (lower panel).

The extremely low level of X found in virus stocks indicates that this protein is incorporated at very low levels, if at all, into virus particles and therefore represents a non-structural protein of BDV. Exclusion of X from virus particles might allow efficient replication/transcription of the viral genome early in infection and is compatible with previous results showing that X-expressing cells are resistant to infection with BDV (Geib et al., 2003). Later in the course of infection, the regulatory function of X might help to achieve and maintain viral persistence, e.g. by avoiding cell damage through extensive BDV replication.

The co-localization of X, P and N in the nucleus of infected cells does not support the hypothesis that X causes a cytoplasmic retention of P (Kobayashi et al., 2003; Poenisch et al., 2004). Although there is no experimental evidence, an association of X with NP complexes could explain co-localization of these proteins in the nucleus of BDV-infected Oligo cells. The non-overlapping X- and N-binding sites of P (Schneider et al., 2004; Schwemmle et al., 1998) may allow the simultaneous interaction with N and X. Alternatively, X may associate with RNP-bound P and thus block the polymerase activity. This is in line with the observation that P multimers can bind to L and X (Schneider et al., 2004). However, this interaction might only be transient and X may be actively stripped from RNPs prior to packaging. Recently its was shown that the polymerase activity was restored in the presence of X by co-expression of G and M (Perez & de la Torre, 2005). Thus, G and M might also prevent efficient packaging of X into viral particles.

We thank Christel Hässler for excellent technical support. We thank Peter Staeheli, Charles Samuel and Geoffrey Chase for critical reading of the manuscript. D. M. is supported by a grant from the Schweizerische Stiftung für medizinisch biologische Stipendien (SSMBS) through a donation by Novartis AG.