A short leucine-rich sequence in the Borna disease virus p10 protein mediates association with the viral phospho- and nucleoproteins

Borna disease virus (BDV) is characterized by a non-segmented negative-strand RNA genome that is replicated and transcribed in the nucleus of infected cells (Briese et al., 1992 ; Cubitt & de la Torre, 1994 ). BDV is the causative agent of Borna disease, a rare progressive meningoencephalitis of immunopathological origin. Borna disease primarily affects horses and sheep, although a broader host range for this virus is suggested by findings that other warm-blooded animal species, including rodents, ostriches and primates, can be experimentally infected with BDV. The detection of BDV-specific antibodies and viral nucleic acids as well as the isolation of infectious virus from psychiatric patients strongly suggest that humans are also susceptible to BDV infection (Bode et al., 1995 ; de La Torre et al., 1996 ; Kishi et al., 1995 ; Planz et al., 1999 ; Sauder et al., 1996 ). However, it is currently a matter of debate whether BDV can be considered to be a human pathogen (Bode et al., 1995 ; Richt et al., 1997 ).

The genome organization of BDV is similar to that of related members of the order Mononegavirales (de la Torre, 1994 ; Schneemann et al., 1995 ). Initially, five large open reading frames were identified and it was suggested that they encoded BDV-specific proteins that function as a nucleoprotein (p38/p39), phosphoprotein (p24), matrix protein (gp18), surface glycoprotein (gp94) and RNA-dependent RNA polymerase (p180/190). We have recently demonstrated that BDV expresses a sixth gene product of 10 kDa, the p10 protein, that is encoded by the predicted x1 open reading frame (Cubitt et al., 1994 ) which initiates upstream of and partially overlaps with the p24 coding region (Wehner et al., 1997 ). The p10 protein has no significant sequence homology to any known protein. However, p10 was found to co-localize with the viral nucleoprotein in the nucleus of BDV-infected cells suggesting an association with the viral ribonucleoprotein (Wehner et al., 1997 ). As an initial step to determine the roles of the p10 protein in the virus life-cycle, we analysed the interactions of the p10 protein with the BDV-specific phosphoprotein p24 and the nucleoprotein p38/p39 which are probably essential components of the virus replication machinery. Analysis of lysates from BDV-infected cells by anti-p10 immunoaffinity chromatography suggested that fractions of the p24 and p38 proteins form complexes with p10 that are stable at high salt concentrations. Moreover, the dissection of these proteinprotein interactions in glutathione agarose-based capture assays indicated that the BDV p10 protein interacts directly with the p24 phosphoprotein and indirectly with the viral nucleoprotein via bridging through p24. A detailed mutational analysis revealed that a short leucine-rich sequence in p10 at amino acids 815 is critical for binding to the viral phosphoprotein. Mutational inactivation of the p24 interacting domain resulted in failure of the p10 protein to associate with the BDV-specific intranuclear foci in infected cells. Due to the association with the predicted viral replicative phospho- and nucleoproteins, possible regulatory roles of the BDV p10 protein in viral RNA synthesis or transport are discussed.

Viruses and cells.
A human oligodendrocyte cell line persistently infected with the BDV strain He/80 (BDV-Oligo; Wehner et al., 1997 ) was maintained and passaged in Dulbeccos modified Eagles tissue culture medium (DMEM) containing 10% foetal calf serum.

Yeast strains, Escherichia coli strains and plasmids.
E. coli strains used for cloning and expression were DH5α and BL26. Saccharomyces cerevisiae EGY48 (Mata trp1 ura3 his3 LEU2::pLEX-Aop6-LEU2), plasmids pSH18-34, pRFHM1, pJG4-5, pcDNA-p10 and pGEX-p10 have been described previously (Gyuris et al., 1993 ; Wehner et al., 1997 ). Construction of plasmids followed standard cloning procedures (Ausubel et al., 1992 ). The plasmids pcDNA-p38 and pcDNA-p24 were prepared by subcloning BDV p38 cDNA (nucleotide positions 931166 according to Cubitt et al., 1994 ) and p24 cDNA (nucleotides 12721877) between the BamHI and EcoRI/BamHI sites of pcDNA3 (Invitrogen), respectively. Plasmids pJG-p10, pGILDA-p10, pGILDA-p24 and pGILDA-p39 were generated by subcloning cDNAs encoding the p10 (nucleotides 12201486), p24 and p39 (nucleotides 541166) proteins of BDV strain He/80 between the EcoRI/XhoI sites of pJG4-5 and pGILDA (Clontech), respectively. Plasmids of the pGILDA series express fusion proteins of the bacterial LexA DNA-binding domain whereas pJG constructs encode fusions with the B42 acidic transactivation domain attached to an HA-tag sequence. Derivatives of pGILDA-p10 expressing C-terminally truncated p10 proteins were constructed by introduction of translational stop codons through site-directed mutagenesis using a Quikchange kit (Stratagene). Plasmids encoding N-terminally truncated p10 proteins were made by PCR amplification of corresponding p10 cDNAs and subcloning the resulting products between the EcoRI/XhoI sites of pGILDA. pGILDA-p10 plasmids encoding p10 proteins with di-alanine substitutions were generated by site-directed mutagenesis. The bacterial expression plasmids pGEX-p24 and pGEX-p10 120 were constructed by inserting the corresponding p24 and p10 cDNAs between the EcoRI/BamHI sites of pGEX-2T (Pharmacia) and the EcoRI/XhoI sites of pGEX-5X-1 (Pharmacia), respectively. p10 cDNA was amplified and subcloned between the EcoRI/SfuI sites of pcDNA3.1/Myc-His C (Invitrogen) to obtain a vector expressing a p10Myc fusion protein in mammalian cells. Within this construct, the internal ATG codon in the p10 gene that initiates translation of the overlapping p24 reading frame was inactivated by site-directed mutagenesis without alteration of the derived p10 amino acid sequence resulting in pcDNA-p10-Myc. A further modification that resulted in the replacement of leucine residues 7, 9 and 10 by alanines (pcDNA-p10-Myc-mut) was introduced by site-directed mutagenesis. The plasmids were confirmed by DNA sequencing using a cycle sequencing protocol followed by analysis through an automated DNA sequencer ABI 377 (Perkin Elmer).

Generation of anti-GSTp10 and anti-GSTp24 sera.
Glutathione S-transferase (GST)p10 and GSTp24 fusion proteins were affinity-purified on glutathione Sepharose resin (Pharmacia) from extracts of E. coli XL-1 Blue transformed with pGEX-p10 or pGEX-p24 as recommended by the manufacturer. Rabbit monospecific anti-GSTp24 serum was raised by several injections with purified GSTp24 fusion protein suspended in Freunds adjuvant. Similarly, anti-p10 serum was generated by several immunizations of a rabbit with purified GSTp10 fusion protein. The specificity of the sera was tested by immunoblot detection of the respective BDV antigens in homogenate of BDV-infected rat brain.

Anti-p10 immunoaffinity chromatography.
Anti-p10 immunoaffinity resin was prepared by covalent immobilization of immunoglobulins from rabbit anti-GSTp10 serum (Wehner et al., 1997 ) on pre-activated Sepharose CL-6B matrix (Pharmacia) according to the protocol supplied by the manufacturer. The serum did not cross-react with any other BDV-specific proteins to detectable levels. Monolayers of BDV-oligo cells were collected in PBS and lysed by sonication. Extracts were clarified by centrifugation at 20000 g for 15 min and cycled several times over the anti-p10 affinity column. Subsequently, the column was washed with PBS and eluted with PBS containing 1 M NaCl, 3 M NaCl or 1 M NaClO₄. Eluate fractions of 2 ml were collected and assayed by anti-p10, anti-p24 and anti-p38/p39 immunoblotting as described previously (Wehner et al., 1997 ).

Yeast two-hybrid analysis of proteinprotein interactions by a β-galactosidase plate assay.
To assay for interactions between BDV p10, p24 and p39 proteins, EGY48 yeast cells harbouring the LexA-dependent lacZ reporter plasmid pSH18-34 were transformed pairwise with plasmids of the pGILDA and pJG series using the lithium acetate method (Ausubel et al., 1992 ). Yeast colonies were streaked onto plates containing X-Gal and incubated at 30 °C as described previously (Ausubel et al., 1992 ). The interactions between the vector-encoded proteins were assessed by blue colour development after 24 h. The expression of BDV fusion proteins was verified by immunoblot analysis of yeast cell extracts using MAbs recognizing the LexA DNA-binding domain and the vector-encoded HA-tag sequence fused to the activation domain, respectively.

Co-precipitation of BDV p24 and p38 proteins with GSTp10 by glutathione Sepharose.
The BDV p10 protein was expressed from pGEX-p10 as a GST fusion protein in E. coli BL26. Synthesis of GSTp10 was induced by addition of 1 mM IPTG. Bacterial cell lysate containing the GSTp10 fusion protein was adsorbed to glutathione Sepharose according to the protocol supplied by the manufacturer and contaminating proteins were removed by three washes with PBS. The BDV p24 and p38 proteins were synthesized and labelled with [³⁵S]methionine in coupled 50 µl transcription/translation reactions (TNT; Promega) programmed with pcDNA-p24 and/or pcDNA-p38. The translation reactions were mixed with 10 µl coated glutathione Sepharose beads in 750 µl NET-N buffer (10 mM TrisHCl, pH 8·0, 1 mM EDTA, 150 mM NaCl, 0·05% Nonidet P-40) for 2 h at 4 °C. The beads were washed three times with PBS/0·01% Nonidet P-40 and the precipitated proteins were separated by SDS gel electrophoresis and visualized by autoradiography.

Indirect immunofluorescence microscopy.
BDV-Oligo cells were transfected with the plasmids pcDNA-p10-Myc wild-type or pcDNA-p10-Myc-mut using the DOTAP reagent (Boehringer Mannheim) and seeded on glass cover slips in DMEM containing 10% foetal calf serum. After 36 h, cells were processed for immunofluorescence analysis by fixation in 2·5% methanol-free formaldehyde (Polysciences) and permeabilization of cells in 0·1% Triton X-100 was done as described previously (Wolff et al., 1998 ). Cells were stained with primary antibodies (rabbit anti-GSTp24 serum, 1:400; Myc-specific MAb 9E10, 1:50) diluted in PBS/3% BSA. The cells were washed and incubated with FITC-conjugated goat anti-mouse immunoglobulin G (IgG) and Texas red-conjugated goat anti-rabbit IgG. Subsequently, the cover slips were washed and mounted in MOWIOL 4-88 (Calbiochem). For immunofluorescence analysis, cells were viewed on a Zeiss Axiophot fluorescence microscope using a 63x objective and photographs were captured by a SPOT video camera (INTAS).

The p10 protein associates with the viral phospho- and nucleoproteins in salt-stable complexes
We have previously demonstrated nuclear co-localization of the BDV p10 protein with the viral nucleoprotein in BDV-infected cells. In order to assess the physical association of the BDV p10 protein with other viral antigens in BDV-infected cells, extracts from persistently infected human oligodendrocytes were analysed by anti-p10 immunoaffinity chromatography (Fig. 1). Proteins retained on the column were eluted with buffers of increasing stringency containing 1 M NaCl, 3 M NaCl and finally 1 M NaClO₄. Eluate fractions were collected and analysed by immunoblotting for the presence of the BDV phospho- (p24), nucleo- (p38/p39) and p10 proteins. The profile of the anti-p10 column demonstrated that a fraction of the viral nucleoprotein but only low amounts of the phosphoprotein eluted from the column in the presence of 1 M or 3 M NaCl. However, in the presence of chaotropic NaClO₄, the desorption of high amounts of BDV p24 and p38/p39 proteins was observed (Fig. 1). The p10 protein eluted from the column only in the presence of NaClO₄. This result suggests that the p10 protein is tightly associated with the viral phospho- and nucleoproteins in BDV-infected cells.

(37K): Fig. 1. Characterization of salt-stable complexes containing the BDV p10 protein and the viral phospho- (p24) and nucleoproteins (p38/p39) by anti-p10 immunoaffinity chromatography. Immunoglobulins from a monospecific rabbit anti-p10 serum were used to prepare an immunoaffinity column by immobilization on a Sepharose matrix. The column was loaded with cleared lysates of persistently BDV-infected human oligodendrocyte cells and washed with PBS containing 150 mM NaCl. Proteins were eluted by using buffers with increasing stringency containing 1 M NaCl, 3 M NaCl and 1 M NaClO4 (for a profile, see b; bar numbering, fractions eluted from the immunoaffinity chromatography column). Fractions were collected and analysed by immunoblotting using pooled phosphoprotein- and nucleoprotein-specific antibodies (a) or anti-p10 serum (c). Fraction numbers are given at the top of each lane. The positions of the p10, p24 and p38/p39 proteins are indicated on the right.

The N-terminal region of the p10 protein interacts with the p24 protein
We next wanted to determine if the p10 protein interacts directly with the viral nucleo- and/or phosphoproteins. To address this topic, yeast two-hybrid and biochemical co-precipitation analyses were used. In the two-hybrid system used, interactions between two proteins drive the expression of a lacZ reporter gene which can be visualized by conversion of a chromogenic substrate added to the growth medium. We observed that pairwise expression of the BDV p10 and p24 fusion proteins as well as that of the p24 and p39 fusion proteins resulted in intense blue colour formation indicating strong interactions between these proteins (Table 1). The interactions were specific since no β-galactosidase activity was detected in cells expressing only one of the two fusion proteins. There was no detectable interaction between p10 and p39 suggesting that the association between the two proteins as detected by immunoaffinity chromatography is not the result of a direct proteinprotein interaction. Similarly, the p10 protein did not interact with itself. To determine the p24-interacting region within the p10 protein, a panel of six N- and C-terminally truncated p10 deletion derivatives were constructed and tested by two-hybrid analysis. All p10 mutant proteins lacking the 20 N-terminal amino acids tested negative whereas up to 67 amino acids could be deleted from the C terminus without affecting the interaction with the p24 protein (Table 1). It is concluded that amino acids at positions 120 mediate the interaction of the p10 protein with p24.

Table 1. Interaction analysis of BDV p10 wild-type and mutant proteins with the BDV p24, p39 and p10 proteins in the yeast two-hybrid system

The viral p24 protein mediates the association of p10 with the viral nucleoprotein
The failure to detect interaction between the p10 protein and the viral nucleoprotein in the two-hybrid system suggested that a third factor(s) mediates assembly of the two proteins into the heteromeric complexes that eluted from the anti-p10 immunoaffinity column. Since the nucleo- and phosphoproteins of BDV have previously been shown to associate with each other (Berg et al., 1998 ; Hsu et al., 1994 ; Schwemmle et al., 1998 ; this study), we examined by GST capture analysis whether the presence of the p24 protein would mediate the interaction of p10 with the viral nucleoprotein. The p10 wild-type protein or p10 amino acids 120 were expressed as GST fusion proteins in E. coli and purified by adsorption to glutathione Sepharose beads. The coated beads were subsequently reacted with radiolabelled BDV phospho- (Fig. 2a) and/or nucleoproteins (Fig. 2b, c) and the precipitates were analysed by gel electrophoresis. In accordance with results obtained in the yeast two-hybrid system, we observed efficient precipitation of the p24 protein by the two GSTp10 fusion proteins but not by GST alone (Fig. 2). In contrast, no significant precipitation was observed when labelled nucleoprotein was reacted with the GSTp10 fusion proteins. However, the nucleoprotein was efficiently captured by the p10 wild-type or p10 120 protein in the presence of the p24 protein. This result strongly suggests that the BDV nucleoprotein associates indirectly with p10 via bridging through the p24 protein.

(48K): Fig. 2. The BDV p10 protein interacts with the viral phospho- (p24) and nucleoproteins (p38) in a GST co-precipitation assay. The p10 wild-type protein (lane WT) or p10 amino acids 120 (lane 120) were expressed as GST fusion proteins (GSTp10) in E. coli and immobilized on glutathione agarose. The coated beads were incubated with 35S-labelled p24 (a), p38 (b) or p24+p38 proteins (c) that were synthesized in coupled transcription/translation reactions in vitro. Precipitated proteins were analysed by SDS gel electrophoresis and autoradiography. As a control, GST alone was expressed and used to precipitate radiolabelled proteins (lane GST). The immunoreactivity of the translated p24 (a and c) and p38 proteins (b) was demonstrated in parallel immunoprecipitation reactions with rabbit anti-p24 serum (α-p24) or a nucleoprotein-specific monoclonal antibody (α-p38). The positions of the BDV phospho- and nucleoproteins are indicated on the right; the positions of molecular mass markers are indicated on the left. Lane T, 10% of total reaction used for precipitation; *, truncated translation products.

A leucine-rich stretch of amino acids at the p10 N-terminal region is essential for the interaction with the p24 protein
Our initial mutational analysis indicated that the binding site for the BDV phosphoprotein is located within the 20 N-terminal amino acids of the p10 protein. Interestingly, this domain contains a leucine-rich stretch of amino acids at positions 412 that is similar to short peptide motifs that were recently identified as nuclear export signals (NES) in cellular and viral factors like the human immunodeficiency virus (HIV) rev protein or the cellular cAMP-dependent protein kinase inhibitor (Fischer et al., 1995 ; Wen et al., 1995 ). NES sequences have been found to interact with specific nuclear receptor proteins like the CRM1 protein that can mediate transport of export-bound proteins and associated RNAs across the nuclear membrane (reviewed by Mattaj & Englmeier, 1998 ). Therefore, we analysed whether the leucine-rich region in the p10 protein would be available for interactions with third factors in the context of a p10p24 complex, or if the same amino acids would engage in p24 binding. Ten p10 mutant derivatives, in which pairs of adjoining amino acids at positions 221 were replaced by alanines, were constructed and tested for their interaction with the p24 protein (Fig. 3a). Changes at p10 positions 815, including several leucine residues, did not affect accumulation of the fusion proteins, but strongly interfered with p24 binding (Fig. 3). In addition, a triple mutant in which leucines at positions 7, 9 and 10 were altered did not interact with the p24 protein. In contrast, none of the di-alanine mutations outside positions 815 affected the p24 interaction. It is concluded that the leucine-rich sequence in the N-terminal domain of the p10 protein mediates binding to the BDV phosphoprotein.

Table 1. They are expressed as + (intense blue), +/- (light blue) and - (white) as indicated on the right. (b and c) To compare the accumulation levels of fusion proteins in the strains depicted in (a) and in Table 1, equivalent amounts of yeast extract were examined by immunoblotting using MAbs recognizing the LexA- (b) or HA-tag (c) moieties. The analysed constructs are indicated on the top of the lanes. The positions of molecular mass markers are indicated on the left.

Mutations in the p24-interacting domain prevent the association of the p10 protein with the nuclear BDV-specific foci in infected cells
A distinctive feature of BDV-infected cells is the accumulation of the viral nucleo- and phosphoproteins in nuclear foci that may also contain viral genomic RNA (Haas et al., 1986 ; Pyper & Gartner, 1997 ; Thierer et al., 1992 ). We have previously demonstrated that the p10 protein is associated with these nuclear foci in BDV-infected cells (Wehner et al., 1997 ). Our mutational analysis allowed us to address the question of whether the accumulation of the p10 protein in the BDV-specific foci would be mainly governed through interaction with the karyophilic viral phosphoprotein or if other factors determine this specific intranuclear localization. To this end, plasmids were prepared from which p10 proteins fused to a short C-terminal Myc-tag sequence were expressed in mammalian cells. Indirect immunofluorescence analysis of transfected BDV-infected oligodendrocyte cells with a Myc-tag specific antibody demonstrated that tagged p10 wild-type protein enters the nucleus and associates with the BDV-specific foci as judged by double-staining with the viral p24 phosphoprotein (Fig. 4a, b). In addition, a fraction of the tagged p10 protein was detected in the cytoplasm in the same distribution as the viral phosphoprotein. Thus, the plasmid-derived p10 protein was localized in the same manner in infected cells as the virus-expressed p10 (data not shown). Transient expression studies in several non-infected cell lines demonstrated a diffuse, non-focal appearance of the p10 wild-type protein in the cytoplasm and nucleus (data not shown). We then tested the intracellular localization of a p10 mutant protein (p10-Myc-mut) that, due to replacement of leucine residues 7, 9 and 10 by alanines, did not interact with p24 (Fig. 3). In contrast to the wild-type, the altered p10 protein was not found in association with the p24-reactive intranuclear foci, although comparable amounts were expressed and entered the nucleus of transfected BDV-infected cells (Fig. 4c). The p10 mutant protein rather had a diffuse nucleoplasmic distribution and tended to accumulate in nuclear clusters that were clearly different in size and position from the pattern of the virus-specific foci (Fig. 4c, d). In addition, a higher proportion of the p10 mutant protein accumulated in the nucleus in comparison to the wild-type, indicating a preference of the mutant for the nuclear compartment. These results suggest that it is primarily binding to the p24 phosphoprotein that governs the association of the p10 protein with BDV-specific foci in the nucleus.

(134K): Fig. 4. Disruption of the p24 interaction abrogates association of the p10 protein with the BDV-specific nuclear foci in virus-infected cells. The intracellular localization of expressed wild-type (b) and mutant p10Myc tag fusion protein (d) was analysed in transfected BDV-infected oligodendrocyte cells by immunofluorescence staining using a Myc-specific MAb. Within the mutant protein, leucine residues 7, 9 and 10 of p10 were replaced by alanines, thereby abrogating the interaction with the viral p24 protein. In the same cells, the viral phosphoprotein was detected by p24-specific rabbit antiserum (a and c).

BDV was recently assigned to the new family Bornaviridae within the order Mononegavirales (Pringle, 1999 ). Thus, the negative-strand RNA genome organization of BDV closely resembles the patterns found in other members of the Mononegavirales. Molecular analyses of other Mononegavirales such as vesicular stomatitis virus (VSV, a rhabdovirus), Sendai virus and measles virus (both members of the Paramyxoviridae) have firmly established that the minimal set of viral proteins that drives virus transcription and replication consists of a viral RNA-dependent RNA polymerase (L), a phosphoprotein (P) co-factor and a nucleoprotein (N) that assemble on the genomic viral RNAs to form a ribonucleocapsid (reviewed in Lamb & Kolakofsky, 1996 ; and in Wagner & Rose, 1996 ). In 3' to 5' order, these functionally homologous proteins are expressed from the first (N), the second (P) and the last (L) open reading frame found in the viral genomes. Analogous to the related Mononegavirales, it is expected that the BDV p38/p39 nucleoprotein, the p24 phosphoprotein and the L gene product constitute the basic apparatus for BDV replication and transcription, although formal proof of this hypothesis is lacking due to the current lack of a functional test system.

We have previously demonstrated that BDV expresses the p10 protein as a sixth viral gene product (Wehner et al., 1997 ). During the initial characterization, the p10 protein had been detected co-localized with the viral nucleoprotein in the nucleus of BDV-infected cells, suggesting an association with the virus replication complexes. Here, we have demonstrated that the p10 protein tightly associates with the viral phospho-and nucleoproteins in complexes that are stable under high salt conditions. Moreover, our two-hybrid and co-precipitation experiments strongly indicated that the p10 protein interacts directly with the viral phosphoprotein, as has been observed previously (Schwemmle et al., 1998 ). We did not detect significant interaction of the p38/p39 nucleoprotein with the p10 protein in the absence of p24, in contrast to the findings of Malik et al. (1999) . However, the viral nucleoprotein was efficiently co-precipitated with p10 in the presence of p24. Since it is known that p24 binds to p38/p39 (Berg et al., 1998 ; Hsu et al., 1994 ) and the interactive domains for the p10 and p38/p39 proteins have been located in different regions of the phosphoprotein (Schwemmle et al., 1998 ), it is likely that p24 bridges an indirect association between p10 and the nucleoprotein. Alternatively, the binding of p24 may induce a conformational change in the p10 protein allowing association with the viral nucleoprotein to occur.

We have explored the p10p24 interaction through extensive mutational analysis in genetic and biochemical assay systems. As a result, we identified p10 amino acid positions 815 to be critical for efficient p24 interaction. The failure of a mutant p10 protein, inactive in p24 binding, to associate with the BDV-specific intranuclear foci suggested that it is mainly the interaction with the viral phosphoprotein which mediates integration of p10 into viral nucleocapsids. Interestingly, this p10 region is rich in leucines and resembles NES that mediate interactions with the cellular export machinery (reviewed by Mattaj & Englmeier, 1998 ). The BDV genomic RNA is replicated in the nucleus of infected cells. Thus, there is a need for bornaviruses to regulate nuclear import and export of the viral genome. Other viruses with a nuclear replication strategy like HIV or influenza viruses express the small NES-containing rev and NEP/NS2 proteins, respectively, that associate with the viral genomes and mediate their nuclear export through interactions with cellular export receptor proteins (reviewed by Whittaker & Helenius, 1998 ). Based on the presence of a leucine-rich sequence at p10 positions 412, we and others have previously speculated that p10 may have a homologous function in nuclear export. Surprisingly, our mutational analysis suggested that the same region within the p10 protein engages in robust binding to the viral phosphoprotein. As a result, we expect p24 binding to prevent further interactions of p10 with third factors like export receptors. Preliminary quantification analyses suggested that there are considerably higher levels of p24 protein in persistently infected cells compared to p10 (Z. Mohammadi-Motahhari and T. Wolff, unpublished results). Therefore, disassembly of the p10p24 complex is probably required to make the N-terminal region of the p10 protein available for interactions with other factors. Clearly, further analysis will be needed in the future to assess potential functions of the p10 N-terminal domain in nuclear export.

We have previously reported that BDV expresses the p10 protein from a reading frame that initiates at an upstream start codon and overlaps with the coding region of the viral p24 phosphoprotein (Wehner et al., 1997 ). Thus, in terms of expression strategy, the BDV p10 protein shares common properties with the C proteins encoded by several paramyxo-and rhabdoviruses, although there is no noticeable sequence homology (Bellini et al., 1985 ; Giorgi et al., 1983 ; Spiropoulou & Nichol, 1993 ). For the C proteins of Sendai virus and VSV, modulatory effects on viral RNA synthesis have been demonstrated in vitro and in vivo (Cadd et al., 1996 ; Curran et al., 1992 ; Horikami et al., 1997 ; Latorre et al., 1998 ; Peluso et al., 1996 ; Tapparel et al., 1997 ). The Sendai virus C protein may exert this regulatory function through direct binding to the L polymerase subunit (Horikami et al., 1997 ). Once suitable assay systems become available for BDV, it will be interesting to address the question of whether the p10 protein, which associates with the predicted RNA polymerase co-factor p24, also has a regulatory function in viral RNA synthesis. We expect that the interaction analysis of the p10 protein will help us in exploring the architecture of the BDV replication complex in the future.

We like to thank Roger Brent (Massachusetts General Hospital) for the generous gift of two-hybrid reagents. This work was supported by an AIDS-Stipendienprogramm-fellowship of the German Ministry for Education, Science, Research and Technology to T.W. and a grant of the Deutsche Forschungsgemeinschaft to J.A.R. (SFB 535). T.W. would like to thank Hans-Dieter Klenk (Institute of Virology, Philipps-University Marburg) for his generous support.