In Vitro and In Vivo Interactions between the Hepatitis B Virus Protein P22 and the Cellular Protein gC1qR

Abstract

gC1qR, a mitochondrial matrix protein, was identified as the main cellular partner of the hepatitis B virus P22 protein. We demonstrated by immunofluorescence studies that some P22 molecules were colocalized with the endogenous gC1qR in both the cytoplasm and the nucleus but never in the mitochondria. We also showed that the last 34 amino acids of P22 were involved in the association with gC1qR.

The hepatitis B virus (HBV) is a DNA virus which can persistently infect the liver. Several studies have suggested an important role for the preC-C gene in this process (24). The product of the preC-C gene is a 25-kDa protein (precore protein) translated from the preC mRNA. During translocation of the nascent polypeptide into the lumen of the endoplasmic reticulum (ER), its signal peptide is cleaved, leading to P22 (24). The major part of the P22 molecule is further truncated at the C terminal by the furin protease (21) into the hepatitis B e antigen (HBeAg). However, it was observed in different experimental systems that some P22 molecules (25) are released back to the cytosolic side of the ER membrane (9, 25, 36). Later on, Guidotti et al. (11) detected cytosolic P22 molecules in the livers of transgenic mice expressing the precore protein.

HBeAg is usually found in the sera of infected patients when viral replication occurs. However, HBV variants which cannot express HBeAg molecules (HBeAg⁻ variants) have been identified in chronic carriers. Interestingly, it was observed that when infants are infected in the prenatal or postnatal periods, none become chronic carriers when they are born to HBeAg-negative mothers, whereas 90% become chronic carriers when their mothers are HBeAg positive (33). Similarly, it was reported that the preC-C gene of woodchuck hepatitis virus is important for chronic infection in the natural host (6). Later, Günther et al. (12) noticed that in chronic carriers, HBeAg⁻ variants are detectable only after several years of infection. Taken together, these observations suggest strongly that one of the preC-C gene products is necessary for the establishment of a persistent infection. It has been observed in transfected cells, as in transgenic mice, that an increase in the expression of the preC-C gene leads to inhibition of HBV replication (11, 17). On the other hand, mutations leading to the abolition or reduction of preC-C gene expression result in a significant increase in HBV replication (4, 29). The negative role of the preC-C gene in HBV replication was demonstrated to be due to cytosolic P22 molecules which form heterodimers with HBV capsid proteins (P21), leading to the formation of unstable nucleocapsids (29). Thus, in addition to its role as an HBeAg precursor, P22 may have an important function in the biology of HBV, leading us to research its cellular protein partners.

Identification of gC1qR as a cellular partner of P22.

Recently, we showed that a 32-kDa protein (P32) interacts strongly with P22 in human and simian cells (28). To obtain a sufficient quantity of P32 and identify it, we designed an affinity assay using a recombinant P22 protein (16). Magnetic beads coated with epoxy (Dynabeads M-270 Epoxy; Dynal) were coupled to 50 μg of the recombinant protein P22r at neutral pH. Then P22r-linked beads or magnetic beads alone were incubated for 1 h at 4°C with the proteins extracted (21) from 4.10⁷ simian COS-7 cells. The magnetic beads were collected at the tube wall, washed with phosphate-buffered saline (PBS), resuspended in 100 μl of Laemmli buffer, and run on a sodium dodecyl sulfate-polyacrylamide (SDS-PAGE) gel. As shown in Fig. 1, a 32-kDa protein was present in addition to P22r only when the magnetic beads were coupled with P22r. Other cellular proteins were associated with the P22r/magnetic bead complexes. Although they may be cellular partners of P22, their amounts were not sufficient to allow a microsequencing analysis.

FIG. 1.

An affinity assay based on P22r coupled to magnetic beads yielded the recovery of significant amounts of P32. Magnetic beads alone (lane 1) or coupled with P22r (lane 2) were mixed with cell extracts from proteins prepared from COS-7 cells. The proteins were immunoprecipitated with an anti-HBc antiserum, separated by 12.5% SDS-PAGE, and Coomassie blue stained. The positions of P32 and P22r are indicated on the right. On the left are indicated migrations of molecular mass standards.

A P32 Coomassie blue-stained spot from an SDS-PAGE gel was digested by trypsin according to the method of Rosenfeld et al. (27). Three internal peptides were sequenced by Edman degradation on a Procise sequencer (Applied Biosystems). Alignment of the sequences with sequences from a protein data bank by using the Smith-Waterman algorithm revealed a complete correlation with the human MA32 protein and important similarities with the MA32 protein from rat and mouse (Fig. 2). Since P32 was isolated from simian cells, this finding strongly suggests that P32 is the simian counterpart of the human MA32 protein also known as gC1qR (10).

FIG. 2.

Characterization of P32 as the counterpart of the human gC1qR protein. The alignment of three peptide sequences (in bold) with the entire sequence of the human gC1qR protein (Hs_gC1qR) and its mouse (Mm_MA32) and rat (Rn_MA32) counterparts (respective accession numbers, Q07021, O35658, and O35796) is shown. Residues which match perfectly with the three sequences are framed. Those matching with only the human sequence are in italic letters. X indicates the undetermined amino acids in the three peptides.

We next examined whether a strong interaction between P22 and human gC1qR can be found in cells supporting the replication of HBV. HepG2 cells were transfected by using Lipofectamin reagent (Invitrogen) with 3 μg of the plasmid pCMV-P25 per million cells. The pCMV-P25 plasmid, which encodes the precore protein, was obtained by insertion of the XhoI to BamHI fragment from pHPC (5) into a modified form of the pCEP4 plasmid (Invitrogen) (30). Cell extracts were prepared 48 h after transfection, and the proteins were immunoprecipitated with 5 μl of an antiserum directed against P22 (Dako). After an extensive washing under high-stringency conditions (28), the proteins were analyzed by Western blotting using a 1 to 5,000 dilution of a mouse monoclonal antibody specific for gC1qR (Covance) and were revealed with the Tropix Western-Star detection system (Applied Biosystems). Figure 3A shows that gC1qR was detected only in the cells in which P22 was expressed. A similar experiment was performed with human 293 cells, which led to the same observation (Fig. 3B). Then proteins from pCMV-p25-transfected cells were immunoprecipitated with the anti-gC1qR antibody and analyzed by Western blotting using the anti-HBc antibody. In this case, P22 was present only in the transfected cells (not shown). Taken together, these results allowed us to conclude that the gC1qR/P22 complex exists in transfected human cells.

FIG. 3.

gC1qR-P22 interaction in human cells. (A) Proteins from nontransfected HepG2 cells (lane 1) or from pCMV-P25-transfected cells (lane 2) were immunoprecipitated with anti-HBc antiserum and analyzed by Western blotting with an anti-gC1qR antibody. The bands detected in lanes 1 and 2 around 30 and 60 kDa most likely represent the heavy and light chains of immunoglobulins. (B) Proteins from nontransfected human 293 cell extract (lane 1) or from pCMV-P25-transfected cells (lane 2) were treated as described for panel A. Proteins from human 293 cells expressing GFP (lane 3) or GFP-C34 (lane 4) were immunoprecipitated with an anti-GFP antibody and analyzed by Western blotting as described for panel A. The migration of molecular mass standards, expressed in kilodaltons, is indicated in the middle.

We next determined which region of P22 is involved in this interaction. Previous results have shown that P32/gC1qR does not interact with P22 proteins truncated at the C terminus (28). This data might be explained either by an alteration of the spatial structure of P22 or by the fact that gC1qR interacts with the C-terminal region of P22. To differentiate between these two possibilities, the chimeric protein GFP-C34, corresponding to the green fluorescent protein (GFP) fused at its C terminus to the last 34 amino acids of P22, was constructed. The plasmid pGFP-C34 was obtained by insertion of the sequence coding for the P22 C-terminal fragment between the BamHI and BglII restriction sites of pEGFP-C1 (Clontech Laboratories Inc.). Proteins from cells transfected with pEGFP-C1 or pGFP-C34 were immunoprecipitated by an anti-GFP antiserum. Western blot analysis performed with an anti-gC1qR antibody showed that gC1qR was detected only in cells expressing GFP-C34 (Fig. 3B, lane 4). In parallel, a Western blot analysis using an anti-GFP revealed that identical amounts of GFP and GFP-C34 were immunoprecipitated in the cell extracts (not shown). Thus, our results demonstrated that the last 34 amino acids of P22 were involved in the interaction with gC1qR.

A fraction of P22 colocalizes with gC1qR.

We then determined in which cellular compartment the P22-gC1qR interaction takes place. Since gC1qR was found in different cellular compartments (mitochondria, cell membrane, and nucleus), we first examined by immunofluorescence its location in COS-7 cells. The cells were grown on a glass tissue culture chamber slide for 48 h and then fixated in PBS containing 4% paraformaldehyde. After this treatment, the cells were incubated in PBS-0.5% Triton for 15 min, incubated for 45 min with the anti-gC1qR monoclonal antibody or the anti-cytochrome c antibody (Oncogene Research Products), and diluted to 1 to 50 or 1 to 250, respectively, in PBS containing 1% bovine serum albumin. Primary antibodies were revealed with either tetramethyl rhodamine isothiocyanate-conjugated goat anti-mouse or CY2-conjugated goat anti-rabbit immunoglobulin G (Jackson Immunoresearch Laboratories). The cells were finally incubated in DAPI (4′,6′-diamidino-2-phenylindole) and diluted to 1 to 2,000 in PBS. The slides were examined with a Leica DMR microscope with a 63 × 1.32 oil immersion objective. Photographs were taken with a Leica CCD camera. This dual staining of COS-7 cells with the anti-gC1qR antibody together with the antimitochondrial cytochrome c antibody confirmed that gC1qR is present in the mitochondria (Fig. 4A). We then compared the cellular localization of P22 with that of the endogenous gC1qR. COS-7 cells were transfected with pHPC, which encodes the precore protein (5), and treated for immunohistochemistry with the relevant antibodies. Images were sequentially collected by confocal laser scanning on a Leica TCS-NT/SP with a 63 × 1.32 oil immersion objective. Z series were generated by collecting a stack consisting of 10 to 12 optical sections by using a step size of 0.284 μm in the Z direction. Staining of the cells with anti-gC1qR antibodies revealed a punctuate cytoplasmic pattern, as in nontransfected cells (not shown), and, in addition, one stained dot in the nucleus (Fig. 4B1). Expression of P22 showed predominant staining of a cytoplasmic area close to the nucleus (most likely the ER and the Golgi apparatus) together with some staining of the nucleus (Fig. 4B2). As shown by the confocal overlay of the single section scanned in the same optical plane, P22 and gC1qR appear to be colocalized in the nucleus of pHPC-transfected cells, as they are in the cytoplasm (Fig. 4B3). To avoid targeting P22 to the ER, we decided to express a P22 protein with its signal peptide deleted (P22Δsp), a strategy which was reported to increase the amount of P22 in the nuclear fraction of COS cells (25). Thus, the 57 nucleotides encoding the first 19 amino acids of P25 except the first Met were deleted from pHPC (21, 24, 25) by using the QuikChange site-directed mutagenesis kit (Stratagene). As shown in Fig. 4C2, P22Δsp was located exclusively in the nucleus, in which it appeared as dots. Moreover, when the cells were stained with the anti-gC1qR antibody (Fig. 4C1), we observed significant staining in the nucleus together with the classical cytoplasmic pattern. Overlaying of the two stainings confirmed that there is a significant colocalization of gC1qR with P22Δsp in the nucleus (Fig. 4C3). Taken together, our results suggested strongly that gC1qR interacted with P22 in both the cytosol and the nucleus.

FIG. 4.

Cellular colocalization of gC1qR and P22. (A) COS-7 cells were treated with anti-gC1qR (panel A1) or with anti-cytochrome c antibodies (panel A2). The images were visualized by direct microscopy, and the individual pictures are merged in panel A3. (B) COS-7 cells were transfected with pHPC plasmid and incubated with anti-gC1qR (panel B1) and with anti-HBc antibodies (panel B2). The individual images visualized by confocal microscopy are shown merged in panel B3. White arrows indicate gC1qR-P22 colocalization. (C) COS-7 cells were transfected with pHPCΔsp and then incubated with anti-gC1qR (panel C1) and anti-HBc (panel C2) antibodies. The individual images visualized by direct microscopy are shown merged in panel C3. The white arrow indicates the spot where gC1qR and P22 were colocalized.

In this report, we demonstrate that the HBV P22 protein interacted strongly with the endogenous gC1qR protein in hepatoma-derived (HepG2) transfected cells. Because these cells support HBV replication, this finding suggests that the association between these two proteins might have an important role, opening a new field of investigation on the function of P22 in the biology of HBV. The same observation was also made for human and simian transfected cells. Three kinds of data support this conclusion. First, we sequenced internal peptides of a 32-kDa cellular protein which interacts with P22 (28) and showed that their sequences match perfectly with that of the human gC1qR protein. Second, we showed that gC1qR and P22 were coimmunoprecipitated when proteins from human cells expressing P22 were treated with an antiserum directed against either P22 or gC1qR. Third, we observed colocalization of gC1qR with P22 in the cytoplasm and the nucleus of cells expressing P22.

gC1qR has been reported to be present in different cellular compartments. It was first described as a nuclear protein associated with the mRNA splicing factor SF2 (15) and later characterized as the receptor for C1q (10), located at the cell membrane. It has also been found to interact with nonstructural viral proteins such as human immunodeficiency virus type 1 Tat and Rev, herpes simplex virus type 1 IE63, Epstein-Barr virus EBNA-1, and structural viral proteins such as adenovirus core protein V, rubella virus core protein, and hepatitis C virus core protein (1, 3, 8, 14, 18-20, 34, 35). It was then established that gC1qR is translated as a full-length protein of 282 amino acids directed to the mitochondria by its 73-residue-long target peptide (7), which is localized mainly in the mitochondrial matrix (23).

Our immunofluorescence studies showed that the interaction between P22 and the endogenous gC1qR takes place both in the cytosol and the nucleus. Thus, we concluded that only the cytosolic P22 molecules, and not those present in the ER, are engaged in this interaction. As these molecules represent a small amount of P22 (estimated at 10 to 15%), this finding likely explains why few P22/gC1qR complexes were observed in cells expressing P22. To overcome this problem, we utilized a P22 protein with its signal sequence deleted (P22Δsp). In this case, several P22Δsp/gC1qR complexes were found in the nuclei, confirming that P22 and gC1qR were associated in the nucleus, as they were in the cytoplasm.

From our data, it is difficult to determine if P22 associates with the immature or the mature form of gC1qR, since it has not been demonstrated that the mature gC1qR can leave the mitochondria (2). The first possibility may be envisaged as the rubella virus core protein interacting with the immature gC1qR in the cytoplasm before being directed to the mitochondria (1). In this case, the core protein would interact with the C-terminal part of gC1qR (22), allowing recognition of the mitochondrial targeting sequence located in the N-terminal part of the protein. In our case, mitochondrial localization of the P22/gC1qR complex was not observed. Thus, it is tempting to speculate that P22 interacts with the immature gC1qR through a region containing its mitochondrial targeting sequence, thus impeding the targeting of the complex to the mitochondria. On the other hand, P22 might interact with the mature gC1qR after a retrograde transport of the mature gC1qR from mitochondria to cytoplasm. Such a possibility cannot be excluded, as several mitochondrial matrix proteins are found in other cellular compartments (31). We also showed that the presence of the last 34 amino acids of P22 is sufficient to allow the interaction of gC1qR with a recombinant GFP. This result is in good agreement with the fact that gC1qR does not associate with a C-terminally truncated P22 (28). Taken together, these results demonstrate the important role of this region in the association of gC1qR and P22.

What might be the biological significance of an interaction between gC1qR and P22 in the nucleus? Different data indicate a role for gC1qR in the mRNA splicing mechanism (15, 26). The potential effect of gC1qR on viral mRNA splicing might be relevant with the biology of HBV, as Soussan et al. (32) have shown that a novel HBV mRNA of 2.2 kb is produced after a splicing process. What is more, they have demonstrated that this mRNA directed the translation of a protein which was found in the liver of the patients. At the present time, the transcriptional control of this spliced mRNA is still unclear. Based on the possible role of gC1qR in the splicing mechanism, it is interesting to speculate that the P22-gC1qR association might play a role in this control. On the other hand, it has been suggested on the basis of structural studies that gC1qR might be a calcium binding protein which interacts with the mitochondrial permeability transition pore (13). One possibility is that the interaction between P22 and gC1qR may impede a role of this protein in apoptosis.