Molecular Characterization of the Human La Protein·Hepatitis B Virus RNA.B Interaction in Vitro *

The La protein was recently identified as a host factor potentially involved in the cytokine-induced post-transcriptional down-regulation of hepatitis B virus (HBV) RNA. The La binding site was mapped to a predicted stem-loop structure within a region shared by all HBV RNAs, and it was concluded that the La protein might be an HBV RNA-stabilizing factor. To characterize the RNA binding mediated by the different RNA recognition motifs (RRMs) of the human La protein, several La deletion mutants were produced and analyzed for HBV RNA binding ability. The data demonstrate that the first RRM is not required for binding, whereas the RNP-1 and RNP-2 consensus sequences of the RRM-2 and RRM-3 are separately required for binding, indicating a cooperative function of these two RRMs. Furthermore, the results suggest that multimeric La disassembles into monomeric La upon binding of HBV RNA.B. By gel retardation assay the affinity of the wild type human La·HBV RNA.B interaction was determined in the nanomolar range, comparable to the affinity determined for the mouse La·HBV RNA.B interaction. This study identified small regions within the human La protein mediating the binding of HBV RNA. Hence, these binding sites might represent targets for novel antiviral strategies based on the disruption of the human La·HBV RNA interaction, thereby leading to HBV RNA degradation.

The human La protein is a 47-kDa phosphoprotein predominantly localized in the nucleus. It was first discovered as an autoantigen recognized by antibodies present in sera of patients suffering from systemic lupus erythematosus and Sjögren's syndrome (1,2). The La protein is a member of a large group of RNA-binding proteins containing RNA recognition motifs (RRM) 1 (3)(4)(5)(6)(7)(8) and is implicated in several steps of RNA metabolism. Among the different La proteins identified in a variety of organisms, the N-terminal part is highly conserved (9). La was shown to co-immunoprecipitate with a number of small RNA molecules (10). A role for La in the termination of RNA polymerase III transcription has been described. It was shown that La interacts with RNA polymerase III transcripts such as pre-tRNA by binding to a small stretch of uridines at the 3Ј-end common to these transcripts and might be necessary for proper processing of these precursors (11)(12)(13)(14)(15)(16)(17). In addition, La is known to interact with a variety of viral and other cellular RNAs (18 -26). La is also suggested to be involved in the capindependent translation initiation of several viruses, including polio virus and hepatitis C virus (19,(27)(28)(29), and more recently evidence is growing that La stabilizes various RNAs, such as histone and hepatitis C and B virus RNA (22,23,25,30,31). At this time point it is not clear yet how La fulfills all of these different functions, however, assuming that this protein acts as a RNA chaperone, thereby stabilizing RNA structures, a function in these varied processes might be envisaged.
The human La protein contains three RNA recognition motifs (RRM) involved in the binding of RNAs (9), although the RRMs do not match very well to the RRM core structure identified by comparison of 70 known RRMs (7). The RNA recognition motif is one of the best characterized RNA binding motifs present as single or as multiple copies in a multitude of RNAbinding proteins. The three-dimensional structure was resolved for some RRMs, and the secondary structural elements composing the three-dimensional structure are ␤1␣1␤2␤3␣2␤4 folding (32,33). Two conserved amino acid signatures referred to as RNP-1 and RNP-2 (RNP consensus sequence) are described as essential motifs for RNA binding. Maraia and coworkers established a model for the interaction of La with pre-tRNA (9). In this model the N-terminal RRM interacts with the 3Ј-end of pre-tRNA by binding to the poly-U stretch, the second and third RRM interacts with the pre-tRNA molecule, and the C-terminal part of La binds the 5Ј-end of the pre-tRNA (9). This C-terminal region contains a basic amino acid stretch and a Walker-A motif (34 -36). After complete processing of tRNA, La is unable to interact with mature tRNA, signifying that the 5Ј-and 3Ј-ends of the pre-tRNA are essential for the recognition by La.
Studying the immune response against the hepatitis B virus using the HBV transgenic mouse model (37), Chisari and coworkers (38 -42) have shown that injection of hepatitis B virus surface antigen-specific cytotoxic T lymphocytes into HBV transgenic mice led to suppression of all viral products by an non-cytotoxic pathway. This as of yet unresolved process was mediated by the cytokines interferon-␥ and tumor necrosis factor-␣. Furthermore, it was shown that these cytokines lead to post-transcriptional degradation of the viral RNA (43). In an attempt to identify host factors involved in degradation of the viral RNA, the mouse La protein (mLa) was identified as an HBV RNA-specific binding protein. A strong correlation between the cytokine-mediated disappearance of HBV RNA and the cytokine-induced processing of full-length mLa was ob-served, indicating that full-length mLa is involved in stabilizing HBV RNA (22,23). The binding site of mLa was mapped to a predicted stem-loop structure within an element 91 nucleotides long (referred to as HBV RNA.B) located at the 5Ј-end of the post-transcriptional regulatory element of HBV, and the specificity of the interaction was confirmed by competition experiments using a variety of competitors (22,23). In addition, the affinity of the interaction was determined in the nanomolar range, further indicating a specific mLa⅐HBV RNA.B interaction (23). The idea that this viral RNA element might represent a stabilizing element was further supported by a recent publication showing that the viral RNA was accessible to endoribonucleolytic cleavage near the mLa binding site (31). The endoribonucleolytic activity present in nuclear extracts of HBV transgenic mice was characterized, and it was shown that the HBV RNA substrates were more efficiently cleaved after induction of HBV RNA degradation and mLa processing. These data imply that the mLa protein is involved in the HBV RNA metabolism.
To gain better insight into the role of human La protein (hLa) in the viral RNA metabolism, we characterized the hLa⅐HBV RNA interaction in more detail. We evaluated the optimal conditions for the interaction between recombinant hLa protein and in vitro transcribed HBV RNA.B and determined the binding affinity. Our results indicate that multimeric hLa disassembles into monomers upon binding of HBV RNA.B. Furthermore, we investigated the requirement of each of the three RRMs for binding by deleting the whole RRM-1 and the RNP-2 and RNP-1 motifs located in the second and third RRM. Deletion of short amino acid stretches gives us the ability to discover a cooperative binding mechanism between RRM-2 and RRM-3 and reduces the possibility of major structural changes triggered by this kind of manipulation. Our study shows that HBV RNA.B is bound by the RNP-2 motifs of RRM-2 and RRM-3, suggesting an interplay goes on between those two RRMs. We identified short amino acid stretches that might be useful as targets for specific disruption of the hLa⅐HBV RNA interaction to destabilize the viral RNA.
PCR products were purified using the QIAquick PCR purification kit (Qiagen, Germany), phosphorylated at the 5Ј-end with 10 units of T4 polynucleotide kinase (Roche Molecular Biochemicals, Germany), subsequently ethanol-precipitated, and resuspended in an appropriate volume of H 2 O. The PCR products were ligated with the Rapid DNA Ligation kit (Roche Molecular Biochemicals, Germany) following the manufacturer's instructions. Finally, the ligated DNA was digested with 10 units of DpnI enzyme (New England BioLabs, Beverly, MA) for 1 h at 37°C with appropriate buffers and transformed into Escherichia coli strain DH5␣. The introduced mutations were checked by sequencing using the Licor 4000L system (MWG Biotech, Germany).
Expression and Purification of Recombinant Proteins-Recombinant proteins were expressed in 100 ml of E. coli BL21 cultures containing the appropriate antibiotic. 100 ml of bacteria cultures was grown at 37°C to a density of A 600 ϭ 0.5, treated with isopropyl-1-thio-␤-Dgalactopyranoside to a final concentration of 1 mM, and shaken for additional 4 -5 h at 37°C. The cells were subsequently precipitated by centrifugation. All purification steps were carried out with chilled buffers at 4°C, and all solutions contained 1% protease inhibitor mix Complete (Roche Molecular Biochemicals, Germany). Cells were lysed in 2 ml of lysis buffer containing 50 mM NaH 2 PO 4 , 10 mM imidazole, 300 mM NaCl and sonicated three times for 10 s. His-tagged hLa protein was purified with Ni-NTA Spin columns (Qiagen, Germany) as follows. The cell lysate was recovered by centrifugation at 20,000 ϫ g for 15 min at 4°C. Before loading, the Ni-NTA spin columns were equilibrated with 600 l of lysis buffer by centrifugation at 700 ϫ g for 2 min at 4°C. The supernatant containing the soluble proteins was loaded onto the spin columns and centrifuged for 2 min at 700 ϫ g. Nonspecifically bound proteins were removed by washing the column four times each with 600 l of wash buffer (50 mM NaH 2 PO 4 , 42.5 mM imidazole, 1 M NaCl, 0.11% Triton X-100). The His-tagged hLa protein was eluted with 3 ϫ 200 l elution buffer (50 mM NaH 2 PO 4 , 300 mM imidazole, 300 mM NaCl).
For large scale hLa preparations, 1 liter of bacteria solution was treated as above, and the pellet was resuspended in 20 ml of lysis buffer, sonicated, and centrifuged for 15 min at 4°C with 20,000 ϫ g. The supernatant was then incubated with 5-8 ml of equilibrated Ni-NTA-agarose (Qiagen, Germany) for 30 -60 min with gentle stirring at 4°C and loaded to a 1.5-ϫ 15-cm column. Nonspecifically bound proteins were removed with 50 ml of wash buffer and His-tagged hLa was eluted with elution buffer in 1-ml fractions. Eluted hLa was dialyzed three times against 500 ml of buffer A (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 3 mM MgCl 2 , 0.5 mM EDTA, 1 mM DTT, and 5% glycerol). When higher protein concentrations were desired, the protein solution was concentrated with Amicon Centricon YM-10 concentrators (Millipore, Germany) following the manufacturer's instructions. Protein concentrations were determined using the Bio-Rad Protein Assay (Bio-Rad, Germany). Correct size and purity of the expressed and purified recombinant hLa were analyzed by 12.5% SDS-PAGE minigels using the MiniProtean-III System (Bio-Rad, Germany) followed by Coomassie Blue staining and Western blot analysis (see Fig. 5 below). The latter was achieved according to standard methods, briefly, by electrotransfer (90 min, 4°C, 65 V) to a nitrocellulose membrane (Protran, Schleicher & Schü ll, Germany) in a wet-blot system using the Bio-Rad Trans-Blot Cell (Bio-Rad, Germany) and immunoblotting with primary mouse monoclonal ␣hLa antibodies 3B9, 4B6, SW5 (M. Bachmann, Oklahoma Medical Research Foundation, Oklahoma City, OK) or rabbit polyclonal ␣-hLa antibody 1982 and horseradish peroxidase-conjugated secondary goat-anti-mouse IgG antibody (HϩL) or goat-anti-rabbit IgG antibody (HϩL) (Dianova, Germany), respectively. Visualization of immunodetected proteins was achieved with the SuperSignal chemiluminescent peroxidase solution system (Pierce, Rockford, IL), followed by x-ray film development. To verify the results, at least two to three different protein preparations were tested that gave similar results.
Gel Filtration-As a further step of protein purification, we performed preparative gel filtration analysis using a HiPrep 26/60 Sephacryl S100-HR column (Amersham Bioscience, Germany), which had been equilibrated with buffer A (see above). To maintain constant flow rates of 0.5 ml/min and for sensitive detection of eluted protein by monitoring the absorbance at 280 nm, the column was connected to a high-performance liquid chromatography BioCad Sprint System (Per-Septive Biosystems Inc./Applied Biosystems, Germany) using the BioCad Perfusion Chromatography Workstation software (version 3.00), including an external fraction collector. To establish a calibration curve for the column, the void volume was determined using 1 mg of dextran (M r 2,000,000, Sigma, Germany). Furthermore, 1 mg each of alcohol dehydrogenase (M r 150,000), albumin (M r 66,000), ovalbumin (M r 42,700), and carbonic anhydrase (M r 29,000) were loaded separately, and the elution volume for each protein was detected. A standard curve was obtained for all calibration proteins with the y-axis showing the log of molecular weight and the x- where V t is bed volume of the column, V o is void volume, and V e is elution volume. For protein purification at least 1 mg of Ni-NTA-purified His-tagged hLa protein in a volume of 1-2 ml was loaded to the equilibrated column, the elution volume of the different peaks was determined and compared with those of the standard proteins thereby calculating the K av and molecular weight of each peak.
Characterization of the oligomerization of WT hLa protein and of the synthesized mutants was performed by gel filtration analysis using an analytical Superdex 200 HR 10/30 column (Amersham Bioscience, Germany). This column was connected to the same high-performance liquid chromatography-system as mentioned above and calibrated with the same proteins (0.2 mg each, loaded separately) with a flow rate of 0.5 ml/min. The buffer used corresponded to buffer A but contained 10 mM EDTA to prevent cation-dependent oligomerization of His-tagged recombinant hLa. For analysis of hLa proteins 0.05 mg of Ni-NTA-purified His-tagged hLA protein in a volume of 0.5 ml was loaded to the equilibrated column. The elution volume was determined, K av was calculated, and the molecular weight was then derived from the standard curve of this column.
In Vitro Transcription-The in vitro transcript HBV RNA.B was generated by in vitro transcription as previously described (22,23). For competition experiments, actin and glyceraldehyde-3-phosphate dehydrogenase RNA were produced as described elsewhere (23). An FspIlinearized plasmid (phTR1) containing the human telomerase RNA coding sequence (kindly provided by W. Filipowicz) was used for the generation of human telomerase competitor RNAs by in vitro transcription.
Gel Retardation Assay-The standard binding reaction was carried out in a final volume of 40 l with 200 ng or, as indicated, Ni-NTA spin columns or, if indicated, preparative gel filtration-purified (Peak-2, 85-90 kDa) recombinant hLa protein and about 200,000 cpm or molar concentrations, as indicated, of 32 P-radiolabeled HBV RNA.B in binding buffer containing 10 mM Tris-HCl, pH 7.4, 3 mM MgCl 2 , 100 mM NaCl, 0.5 mM EDTA, and 0.5% Nonidet P-40 (unless otherwise stated). The in vitro transcribed 32 P-labeled RNA.B was denatured at 75°C for 10 min and renatured by cooling slowly to RT prior to addition to the reaction mixture. Samples were incubated for 10 min at RT. After addition of 5 l of electrophoresis buffer containing 10% glycerol and 0.01% bromphenol blue, reaction mixtures were separated on an 8% native polyacrylamide gel (18 ϫ 18 cm) for 3-4 h at 200 V at room temperature. The gels were prerun at 240 V for 1 h in 1 ϫ TBE containing 45 mM Tris, 45 mM boric acid, 1 mM EDTA. Gels were dried for 1.5 h at 80°C on Whatman paper (Whatman, UK) using a Bio-Rad Slab Dryer Model 483 (Bio-Rad, Germany), and signals were evaluated using a FUJIX BAS 2000 phosphorimaging system (Fuji, Germany) or documented via exposure to x-ray films. All EMSAs with wild type and deletion mutants of hLa protein were performed at least three to five times with at least two different protein preparations, and similar results were obtained in all cases.
For saturation experiments and calculation of dissociation constants (K D ), increasing amounts of protein were added to constant amounts of RNA as indicated in the figures. The signal intensity of the hLa⅐RNA complexes was quantified using TINA2.09d software provided by the supplier (Raytest, Germany). The band intensities of monomeric hLa⅐RNA complexes were measured, background intensities were subtracted, and the values were transformed to relative ratios calculated as a percentage of maximum signal, and data were fitted to non-linear regression curves. Data were expressed as the mean out of five independent experiments.
For supershift analysis, 200 ng of Ni-NTA-purified hLa was preincubated with 5 l of antibody (as indicated in the figures) and 20 units of RNasin (Promega, Madison, WI) in a 20-l volume of 10 mM Tris-HCl, pH 7.4, at 4°C for 1 h, before the other components of the standard reaction and the labeled RNA were added. Competition experiments were carried out by addition of excess cold competitor to the binding reaction 3 min before the addition of the labeled transcript.

RESULTS
Characterization of the hLa⅐HBV RNA.B Interaction-To study the interaction between hLa and HBV RNA.B, an RNA gel retardation assay was established using recombinant hLa and in vitro transcribed HBV RNA.B as interaction partners. In the first experiments, we tested whether the specificity and the binding conditions for the complex formation between recombinant hLa and HBV in vitro transcribed RNA.B were similar to the previously shown specific interaction and optimized binding conditions between HBV RNA.B and endoge-nous mLa prepared from HBV transgenic mice livers and HBV RNA.B.
The recombinant hLa protein used in this study was fused to a hexa-histidine tag, expressed in E. coli, and purified by Ni-NTA affinity chromatography or gel filtration as indicated as described under "Experimental Procedures." HBV RNA.B was synthesized by in vitro transcription using [ 32 P]UTP for uniform labeling of the transcript. Incubation of recombinant hLa with labeled HBV RNA.B leads to the formation of hLa⅐RNA.B complexes with different electrophoretic mobility as shown by RNA gel retardation assays (Fig. 1A, lane 2). The signal intensities of these hLa⅐RNA.B complexes varied between different experiments, but the complex with the highest mobility (Fig.  1A, hLa monomer) was consistently predominant, suggesting a preferential formation of this complex. To confirm that these complexes were formed between hLa and HBV RNA.B, a supershift analysis was performed using two La-specific antibodies. The monoclonal antibody SW5 and the polyclonal rabbit anti-hLa serum shifted all of the hLa⅐HBV RNA.B complexes (Fig. 1A, lanes 2 versus 3 and 4), whereas the rabbit preimmune serum had no effect on complex formation or on complex mobility (Fig. 1A, lanes 2 versus 5). These results show that hLa forms RNA-binding competent multimers, although the hLa⅐HBV RNA.B complex with the highest mobility, referred to as monomer, was the most prominent one, and the formation of higher molecular weight complexes varied from experiment to experiment as seen in this study.
Next we asked whether recombinant hLa binds HBV RNA.B as specific as it was shown recently for the interaction between endogenous mLa present in nuclear protein extracts of HBV transgenic mice and HBV RNA.B (22,23). Competition exper- iments were performed using several unlabeled in vitro transcribed RNAs, including glyceraldehyde-3-phosphate dehydrogenase, telomerase RNA (both not shown), HBV RNA.B and actin transcripts (Fig. 1, B and C). As exemplified in Fig. 1C it is shown that unlabeled actin RNA competes almost as efficiently as unlabeled HBV RNA.B (Fig. 1B) for the binding of recombinant hLa to labeled HBV RNA.B. Similar results were obtained with glyceraldehyde-3-phosphate dehydrogenase and telomerase competitor RNAs (not shown). Because a highly specific interaction between endogenous mLa and HBV RNA.B was shown recently (22,23), we believe that auxiliary cofactors present in nuclear extract are required to specify the interaction between La and HBV RNA.B. In a first attempt to address this assumption we combined recombinant hLa with nuclear extracts prepared from human hepatoma cell line Huh-7. Under this experimental setting, it was observed that factors in nuclear extracts were indeed able to modulate the recombinant hLa⅐HBV RNA.B complex formation (not shown).
To gain further insight into the binding reaction and the nature of the different hLa⅐RNA.B complexes, the influence of increasing salt and DTT concentrations were tested as well as the influence of different pH, temperature, and MgCl 2 , and EDTA were monitored. In these experiments the binding conditions were as described for standard conditions under "Experimental Procedures," except for the adjusted compound, which is as indicated in the figures. As shown for the interaction between endogenous mLa protein and HBV RNA.B (22), the efficiency of RNA binding was only slightly reduced in the presence of 1 M NaCl, indicating non-electrostatic interactions may contribute to this interaction (Fig. 2, compare lane 2 and  lanes 7-10). The hLa⅐HBV RNA.B complex was stable at pH 4.5, 7.4, and 9.5 (Fig. 2, lanes 12-14) and was formed to a similar extent at 4°C, room temperature, and 37°C (not shown). Furthermore RNA⅐protein interaction can be expected to depend on MgCl 2 and EDTA (44). Therefore, different MgCl 2 and EDTA concentrations both alone and in combination were tested. The formation of hLa⅐HBV RNA.B complexes were observed with or without 3 mM MgCl 2 (not shown), indicating that MgCl 2 was not necessary for complex formation, as shown previously for the interaction between mLa and RNA.B (22). Addition of EDTA (5 mM) strongly increased the formation of hLa⅐HBV RNA.B complexes independently of MgCl 2 (not shown). Increasing concentrations of the reducing agent DTT were without effect on complex formation (Fig. 2, compare lane   2 and lanes 3-5), showing that disulfide bridges are not involved in formation of hLa multimers. Therefore, standard binding reactions were performed in 10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 3 mM MgCl 2 , 0.5 mM EDTA, and 0.1% Nonidet P-40. Taken together, these data show similar binding characteristics of recombinant hLa to HBV RNA.B compared with the binding conditions reported earlier for the interaction between endogenous mLa protein and HBV RNA.B (22), signifying a similar method of RNA binding for recombinant hLa and endogenous mLa.
To determine the molecular weight of the hLa protein and potential hLa multimers in our preparation, purified recombinant hLa was submitted to preparative gel filtration chromatography. Gel filtration analysis of the wild type hLa protein revealed two major peaks with estimated molecular masses of Ͼ150 kDa (peak 1) and 85-90 kDa (peak 2). The presence of full-length hLa (47 kDa) in each of these protein peaks was confirmed by Western blot analysis using hLa-specific antibody 3B9 (Fig. 3B). We conclude from this analysis that peak 1 contains hLa multimers and peak 2 hLa dimers, however, we did not observe a peak corresponding to the molecular mass of hLa monomers (47 kDa), indicating that hLa prepared under our conditions exists preferentially as dimers and multimers. These observations confirmed the previous assumption that the different hLa⅐HBV RNA.B complexes detectable during gel retardation assay analysis consist of hLa multimers and monomers. At this point it was still not possible to decide whether the complex with the highest mobility represents hLa monomers or multimers bound to HBV RNA.B. Previously, it was shown that hLa contains a dimerization domain located in the C-terminal part between amino acids 298 and 348 of La (45). To test to which extent this domain contributes to the formation of the different complexes, we deleted the dimerization domain. This deletion mutant (hLa-⌬6, deletion of amino acids 274 -354) and wild type hLa were purified by nickel affinity chromatography (see Fig. 5, C and D) and subsequently submitted to analytical gel filtration analysis. To exclude the formation of recombinant hLa multimers via the interaction of divalent cations and the His-tags, 10 mM EDTA was included in the buffer for the analytical gel filtration. Under this condition separation of wild type hLa revealed a major peak of molecular mass of ϳ100 kDa and one smaller peak of ϳ50 kDa representing hLa dimers and monomers, respectively. Analysis of hLa-⌬6 revealed one major peak of ϳ45 kDa, indicating that the deleted region was required for hLa dimerization. The presence of full-length hLa-⌬6 (approximately 40 kDa) in these protein peaks was confirmed by Western blot analysis using hLa-specific antibody 3B9 (Fig. 3C).
Next, we submitted wild type and ⌬6 hLa proteins to gel retardation analysis. Although wild type hLa forms several hLa⅐RNA.B complexes of different mobilities (Fig. 3A, lane 2), hLa-⌬6 leads to the formation of a single complex (Fig. 3A, lane  3). The electrophoretic mobility of this complex was similar to the wild type hLa⅐RNA.B complex with the highest mobility, but the signal intensity appears to be less intense. Because in analytical gel filtration analysis hLa-⌬6 was only eluted as a monomeric protein, the results of the gel retardation assay indicate that the complex with the highest mobility consists of hLa monomers bound to HBV RNA.B. Additionally, these results indicate that, upon binding of HBV RNA.B, the state of hLa oligomerization changes: gel filtration analysis revealed a multimeric form of the protein, whereas gel retardation assays predominantly revealed monomers bound to RNA. This signifies that hLa monomers preferentially bind to HBV RNA.B, although the original fraction contains hLa multimers, sug- gesting a shift from hLa multimers to monomers following RNA binding.
In this context it is also very important to calculate how much HBV RNA.B binding competent WT hLa protein exists in our gel filtration fraction. We determined that 100 ng (50 nM) gel filtration-purified recombinant hLa (Fig. 3B, GF peak-2, 85-90 kDa) was able to bind ϳ2 nM HBV RNA.B (see below). Assuming that the hLa WT purified by preparative gel filtration (peak-2, 85-90 kDa) was not contaminated with monomeric hLa, then approximately 1 of 12 hLa dimers was able to bind one HBV RNA.B transcript. On the other hand, we cannot ultimately exclude a minor contamination of monomers in gel filtration peak 2 (hLa dimers, 85-90 kDa), although rechromatography of the hLa dimer fraction showed no monomeric hLa peak (data not shown).
To test the influence of additional regions in the C-terminal part of hLa on RNA binding, amino acids 353-393 were deleted, referred to as mutant hLa-⌬7 (Fig. 5, A, C, and D). This mutation reduced the efficiency of the interaction with HBV RNA.B relative to WT hLa (Fig. 3A, compare lanes 2 and 4), indicating that amino acids in this region were also partially required for binding or that structural changes account for the reduced binding activity. This may also be true for the reduced RNA binding activity of hLa-⌬6 (Fig. 3A, compare lanes 2 and  3). Taken together, it was shown that hLa monomers preferentially interact with HBV RNA.B and that amino acids and/or structural features of the C-terminal part might be involved in binding.
Affinity of the Wild Type hLa⅐HBV RNA.B Interaction-Ex-periments were performed to determine the binding affinity of recombinant wild type hLa to HBV RNA.B. Protein purified by Ni-NTA affinity chromatography and subsequent preparative gel filtration chromatography was used for the calculation of the binding affinity. First, increasing amounts of preparative gel filtration-purified (peak 2, 85-90 kDa) wild type hLa proteins were added to constant concentrations of RNA.B to determine the optimal protein concentration required for titration of RNA.B (not shown). Gel retardation assay results were analyzed by phosphorimaging with arbitrary units to quantitate the ribonucleoprotein complexes formed at varying hLa protein concentrations. We observed a linear increase in RNA.B binding at hLa protein concentrations between 25 (50 ng) and 100 nM (200 ng) at 0.1 nM RNA.B per binding reaction (data not shown). Based on these results, 50 nM (100 ng) of preparative gel filtration-purified WT hLa (peak 2, 85-90 kDa) was used to determine the binding affinity at increasing RNA.B concentrations. Gel retardation results were analyzed by phosphorimaging to quantitate RNA⅐protein complex formation at increasing RNA.B concentrations (Fig. 4A). The apparent affinity (K D ) was calculated according to the mass action equation. The maximal monomeric hLa⅐HBV RNA.B complex formation as determined by phosphorimaging was set as 100%, and the percentage of complex formation was plotted against the corresponding RNA.B concentration. By using this method, we determined the apparent affinity for the formation of the monomeric wild type hLa⅐HBV RNA.B complex as K D ϳ 0.8 nM (Fig. 4B), which represents the mean of five independent experiments. Note the intense formation of monomeric hLa⅐HBV RNA.B complexes, although dimeric hLa protein was applied. The affinity for the interaction between recombinant WT hLa is in accordance to the previously determined affinity (K D ϳ 1.4 nM) for the interaction between endogenous mLa protein and HBV RNA.B (23).
Contribution of the Different RNA Recognition Motifs to the Binding of HBV RNA.B-The hLa protein contains three RRMs potentially mediating the recognition of HBV RNA.B. To gain better insight into the contribution of the individual domains to HBV RNA.B binding, we introduced several deletions within the three different RRMs. The RRM-1 was completely deleted, whereas the RNP-2 and RNP-1 motifs present in each of the RRM-2 and RRM-3 domain were separately deleted (Fig.  5, A and B). The mutant hLa proteins were expressed in E. coli, purified, and analyzed by Coomassie Blue staining (not shown) and Western blotting using the two different monoclonal hLa antibodies 3B9 and 4B6 (Fig. 5, C and D). The Western blot analysis was performed to confirm that equal amounts (200 ng) of each of the recombinant hLa mutants were applied to the gel retardation assay analyses. Fig. 5 (C and D) shows that equal amounts of the different hLa mutants were applied in this study. Furthermore, at least two different protein purifications were performed for each mutant to verify the results, and representative gel retardation assays are shown in this study. Analysis of the HBV RNA.B binding activity of hLa-⌬1 re-vealed that the RRM-1 was not required for binding HBV RNA.B (Fig. 6, compare lanes 2 and 3). However, it seems that the complex formation was even stronger. This observation is of particular interest, because the RRM-1 was implicated in the interaction of pre-tRNA by mediating the binding of the poly-U-stretch located at the 3Ј-end. The RNA used in this study contains two internal stretches of uridines, but not at the 3Ј-or 5Ј-end, indicating that the RRM-1 mainly mediates the binding of a specific subset of RNAs.
Analysis of hLa-⌬2, in which the RNP-2 motif of RRM-2 was deleted, revealed that the amino acid sequence 113 VYIKGF 118 was absolutely essential for binding to HBV RNA.B (Fig. 7,  compare lanes 2 and 3). Valine (V) and Isoleucine (I) residues fit to the RNP-2 core consensus (UXUXXL, Fig. 5B) identified by Birney and co-worker (7) but not the phenylalanine. Single amino acid substitutions at positions 114 (Y114P) and 117 (G117E) revealed that these amino acids are not required for binding (not shown). The RNP-1 motif 151 KGSIFVVF 158 of RRM-2 shows strong homology to the RNP-1 core sequence XXXUXVXF (Fig. 5B). Deletion of the RNP-1 motif diminished but not completely abolished the RNA binding activity of mutant hLa-⌬3, as was observed with the mutant hLa-⌬2 (Fig. 7,  compare lanes 2, 3, and 4). These data demonstrate that the RNP-2 is essential and the RNP-1 partially required for the hLa⅐HBV RNA.B interaction. Therefore, in contrast to the first RRM, the second RRM is obligatory for the interaction with HBV RNA.B.
Taken together, these data show that the RRM-1 is redundant for binding HBV RNA.B. The RNP-2 motifs of the second and third RRM are autonomously involved in the recognition of RNA.B, whereas the RNP-1 motifs are partially required. We conclude that a complex mechanism of HBV RNA.B recognition is mediated by RRM-2 and RRM-3 but not by RRM-1.

DISCUSSION
During a search for hepatocellular proteins that mediate the cytokine-induced degradation of HBV RNA, we recently demonstrated a correlation between the disappearance of the viral RNA and the appearance of a La protein fragment in the mouse liver following cytotoxic T lymphocytes injection or viral infection (22,23). In the present study, we extended the knowledge about the interaction between recombinant hLa and HBV RNA.B by characterization of the binding conditions, determination of the affinity, and identification of hLa motifs mediating the interaction with HBV RNA.B.
It is shown that the binding conditions for the interaction between endogenous mLa and recombinant hLa were very similar, in particular it was observed that the mouse and hLa⅐HBV RNA.B complexes were stable under high salt conditions (Fig.  2) and that binding is independent of MgCl 2 . Other protein⅐RNA complexes are destabilized at increasing ionic strength (44,46,47), presumably due to competition between salt anions and nucleic acid anions for protein interaction sites. Therefore, the stability in presence of high ionic strength might FIG. 4. A, apparent affinity of the hLa⅐HBV RNA.B interaction determined in gel retardation assays. Standard gel retardation assay was performed under conditions described under "Experimental Procedures." In all binding reactions 100 ng of wild type recombinant hLa (WT) purified by Ni-NTA spin columns and subsequent preparative gel filtration chromatography (GF, peak 2, 85-90 kDa) was analyzed for HBV RNA.B binding (lanes 2-11). The concentration of labeled HBV RNA.B was increased as indicated in lanes 2-11. Lane 1, reaction without hLa. Binding reactions were analyzed as described under "Experimental Procedures." B, plot of the percentage of complex formation versus increasing HBV RNA.B concentrations. The band intensities of monomeric hLa⅐RNA complexes were measured, background intensities were subtracted, the resulting values were transformed to relative ratios calculated as a percentage of maximum signal, and data were fitted to non-linear regression curves. Data are expressed as the mean out of five independent experiments.
indicate that non-electrostatic interaction (i.e. hydrophobic forces) contributes to the hLa⅐HBV RNA.B interaction. To verify this assumption structural information about the hLa protein and hLa⅐RNA complexes will be needed.
The binding affinity of monomeric wild type hLa was determined to be ϳ0.8 nM (Fig. 4), which is similar to the affinity calculated by UV cross-linking experiments for the interaction between mLa and HBV RNA.B (ϳ1.4 nM (23)), signifying a high affinity interaction between recombinant human or endogenous mLa protein with the HBV RNA.B. The monitored value for the affinity is in the range of other ribonucleoprotein complexes described, such as has been calculated for the interaction between the HIV Tat protein and the TAR element (ϳ0.14 nM (48)), hnRNP proteins (10 pM to 10 nM (49)), and the binding of the hLa protein and the TAR element (17 nM (50)). Previously the specific interaction between endogenous mLa protein and HBV RNA.B was confirmed by competition experiments using different RNA substrates (23). In contrast we were not able to show a specific interaction between HBV RNA.B and recombinant hLa. Therefore, we assume that auxiliary factors are a prerequisite for a specific interaction between hLa and HBV RNA.B, as contained in nuclear mouse liver protein extracts, where the interaction was shown to be specific (22,23). The first experiments addressing this point suggest that the complex formation between recombinant hLa and HBV RNA.B could indeed be modulated by addition of nuclear extracts prepared from Huh-7 cells (not shown).
The formation of several hLa⅐HBV RNA.B complexes was revealed by gel retardation assays (Fig. 1A). As shown in this study these complexes were composed of monomeric and multimeric hLa proteins bound to HBV RNA.B, and evidence is provided that these complexes were not stabilized by disulfide bridges, because the reducing agent DTT did not change the ribonucleoprotein complex pattern (Fig. 2, lanes 2-5). It will be of general interest to evaluate how the equilibrium between hLa monomers and oligomers is regulated and which functions are accomplished by oligomeric or monomeric hLa. The importance of this issue was supported by studies addressing the function of La in poliovirus translation initiation (45). The authors identified the dimerization domain located between amino acids 298 and 348 and showed that deletion of this sequence leads to RNA-binding competent La monomers, but, interestingly, the stimulating effect of La on poliovirus translation initiation was abolished by this deletion (45). Consequently, it seems that La might fulfill certain functions as oligomer and others as monomer, leading us to question in which conformation hLa preferentially binds HBV RNA. Because the monomeric hLa⅐RNA.B complex was the most dominant, we assume that hLa preferentially binds HBV RNA as a monomer. This assumption is supported by our observation that the 85-to 90-kDa gel filtration fraction of wild type hLa represents La dimers but forms predominantly monomeric hLa⅐HBV RNA complexes (Fig. 4A). This could be explained by a minor contamination of monomeric hLa in our dimeric gel filtration peak or by a regulative mechanism following binding of RNA, suggesting an equilibrium between free oligomeric hLa and RNA-bound monomeric hLa. The latter assumption appears more likely because a contamination with hLa monomers was not detectable and would suggest that traces of monomeric hLa bind HBV RNA.B with very high affinity. Because the La protein is associated with, and implicated in, the processing of a variety of small nuclear RNPs (51), small nucleolar RNPs (52), and pre-tRNA (14,16), it is tempting to speculate that the function of La as an RNA chaperone, thereby stabilizing RNA molecules during their processing, might depend on the state of oligomerization of hLa. However, the proposed regulation of the equilibrium between hLa oligomers and monomers following RNA binding must be tested by other methods.
It was shown in this study that deletion of amino acids 274 -348, covering the dimerization domain, was involved in the oligomerization of hLa, however, in addition it was observed that complex formation between hLa-⌬6 and HBV RNA.B was diminished (Fig. 3A). Therefore, additional amino acids in the deleted regions were required for high affinity binding. On one hand it might be due to the deletion of the C-terminal part of the RRM-3, including the last two amino acids of RNP-1. This RNP-1 motif contributes partially to the binding as discussed below. On the other hand the Walker-A motif (amino acids 333-340 (34)) was completely removed as well as a partial stretch of basic amino acids (amino acids 328 -363) located in this region. These elements might also contribute to the binding, because the 5Ј processing of pre-tRNA was dependent on these elements (9). However, La⅐RNA binding studies have shown that deletion mutants missing the C-terminal part of La are still able to bind to HIV TAR and hY1 RNA (50) as well as different single-and double-stranded RNA substrates (35,53). In contrast, the binding of La to the HCV internal ribosome entry site was diminished after substituting amino acids in the basic region between aa 328 and 344 (27), thus indicating that different La domains are involved in the recognition of various subsets of RNAs described in the literature. Further work will be needed to discriminate between the specific amino acids in the C-terminal part of hLa necessary for direct binding or stabilization of a structure allowing interaction with HBV RNA.B. Structural changes within the hLa protein may be induced by the deletion in hLa-⌬7 (amino acids 353-393, Fig. 3A), which might artificially reduce the binding capability of this mutant. However, we cannot rule out the possibility that the deleted acidic amino acid stretch (amino acids 367-375) and/or the casein kinase II phosphorylation site at position 366 contribute to a certain extent to the binding of HBV RNA.B. The casein kinase II phosphorylation was shown to control the maturation of pre-tRNA by releasing the 5Ј-end of the RNA after phosphorylation (15) without rendering the RNA binding activity to hY1 RNA (13). Recently, it was shown that phosphorylation of recombinant hLa by casein kinase II reduced the interaction with synthetic RNA oligomers carrying a poly-U stretch at their 3Ј-end (35). This suggests that serine 366 phosphorylation did not regulate the general RNA binding activity of hLa but does effect the interaction with certain RNA substrates and controls other functions mediated by hLa. It is likely that the interaction between hLa and HBV RNA.B is regulated by phosphorylation, because dephosphorylation of mouse nuclear extracts prior to UV-cross-linking abolished binding of mLa protein to HBV RNA.B (23). Therefore, future work will address the question as to what extent potential phosphorylation sites are involved in the regulation of the hLa⅐HBV RNA.B interaction.
Another deletion mutant hLa-⌬1, in which the RRM-1 was deleted (amino acids 11-99, Fig. 6), bound HBV RNA.B very intensively, indicating that deletion of the RRM-1 did not contribute to the binding at all and did not induce unfavorable structural changes. This mutant is of special interest, because the RRM-1 is clearly implicated in the interaction with pre-tRNA (15). Recently, a model for the interaction between La and pre-tRNA was established (9). In this model the RRM-1 mediates the binding of the typical 3Ј-UUU signature of RNA polymerase III transcripts, whereas the RRM-2 and RRM-3 motifs are required for general pre-tRNA binding. In addition, the 5Ј-flank of the immature tRNA is bound by the Walker A motif located in the C-terminal part of hLa. Therefore, the RRM-1 and the Walker-A motifs are essential determinants for a stable interaction between La and pre-tRNA, because mature tRNA is not bound by La. In comparison to this method of binding, the recognition of HBV RNA.B differs in the way that the RRM-1 is not involved and that the high affinity binding might be established by a cooperative binding mode of RRM-2 and RRM-3 as discussed below.
RRMs are well described RNA binding motifs found in a variety of RNA-binding proteins. These motifs are between 70 and 80 amino acids long and contain the RNP-2 and RNP-1 amino acid signatures required for RNA binding. The structure of RRMs was determined e.g. for the human hnRNP C protein, the small nuclear ribonucleoprotein A and the poly(A)-binding protein (32,33,54) and revealed the secondary structure of the RRMs composed of ␤1␣1␤2␤3␣2␤4 folding structure. The RNP-2 is located in the ␤1 and the RNP-1 is located in the ␤3 structural element. In the three-dimensional organization, the RNP-2 and RNP-1 are placed between helices 1 and 2, assembling a RNA binding surface. RNPs are required for general RNA binding activity, but they do not necessarily determine the binding specificity. In some cases the amino acids leading to the specificity for the interaction are located in the loop 3 between ␤2␤3 or in the C-terminal region of the RRM (49). The comparison of 70 different RRM-containing proteins leads to the formulation of a RRM core sequence (7). In that study only the RRM-3 of the hLa protein was used for comparison and interpreted as an atypical RRM. Although the RNPs are not very typical, deletion of the RNP-2 and RNP-1 sequences clearly show that each of these motifs were required for RNA.B binding. Interestingly, separate deletion of the RNP-2 motifs of RRM-2 and RRM-3 identified these elements as the most important ones, because the RNA-binding activity was almost completely abolished (Figs. 7 and 8). These results strongly indicate that the binding of HBV RNA.B is not only mediated by a single RRM but that RRM-2 and RRM-3 are required. Deletion of the RNP-1 signatures located in the RRM-2 and RRM-3 partially reduces binding (Figs. 7 and 8). Hence, all RNPs of RRM-2 and RRM-3 contribute to binding, suggesting that the RRM-2 and RRM-3 domains are functionally linked. Many RRM-containing proteins carry not only one RRM but multiple RRMs. For the HuD protein and the Sex-lethal protein, it was shown that two RRMs are forming a cleft in which the RNA was bound (55,56). This study provides evidence that the RRM-2 and RRM-3 motifs of hLa are the central requirements for HBV RNA.B binding, most likely by forming an RNA binding domain composed of RRM-2 and RRM-3, as is shown for the binding of AU-rich elements by the HuD protein (55).
In conclusion, the La protein interacts with a variety of RNA molecules with a wide diversity of structures and sequences. It is likely that La interacts with these subsets of RNAs by establishing varying binding modes accomplished by different La domains. For binding HBV RNA.B the RRM-1 is redundant, whereas the RRM-2 and RRM-3 domains function presumably in a cooperative manner. Whether the C-terminal part directly interacts with HBV RNA.B or indirectly stabilizes a structure necessary for a stable interaction remains to be clarified. Furthermore, we assume that the equilibrium between oligomeric and monomeric hLa is regulated by RNA binding. This is of general interest to understand the variety of hLa functions. The RNP-2 motifs of RRM-2 and RRM-3 were identified as the most important regions for the hLa⅐HBV RNA.B interaction. It is tempting to speculate that the selection of molecules specifically interfering with the hLa⅐HBV RNA interaction might induce HBV RNA degradation.