Specific interaction in vitro and in vivo of glyceraldehyde-3-phosphate dehydrogenase and LA protein with cis-acting RNAs of human parainfluenza virus type 3.

Human parainfluenza virus type 3 (HPIV3) genome RNA is transcribed and replicated by the virus-encoded RNA-dependent RNA polymerase, and specific cellular proteins play a regulatory role in these processes. To search for cellular proteins potentially interacting with HPIV3 cis-acting regulatory RNAs, a gel mobility shift assay was used. Two cellular proteins specifically interacted with the viral cis-acting RNAs containing the genomic 3′-noncoding region and the plus-sense leader sequence region. Surprisingly, by biochemical and immunological analyses, one of the cellular proteins was identified as the key glycolytic enzyme, glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The other protein was characterized as the autoantigen, LA protein. Both GAPDH and LA protein also interacted with the same cis-acting RNA sequences in vivo and were found to be associated with the HPIV3 ribonucleoprotein complex in the infected cells. By double immunofluorescent labeling, GAPDH was found to be co-localized with viral ribonucleoprotein in the perinuclear region. These observations strongly suggest that cellular GAPDH and LA Protein participate in the regulation of HPIV3 gene expression.

The human parainfluenza virus type 3 (HPIV3), 1 belonging to the paramyxovirus family, is one of the major causes of pneumonia and bronchiolitis in infants (1). HPIV3 contains a negative strand RNA genome that is encapsidated by a nucleocapsid protein NP (68 kDa) and tightly associated with two RNA polymerase subunits, a large protein L (251 kDa) and a phosphoprotein P (90 kDa), to form the viral ribonucleoprotein (RNP) core (2,3). The encapsidated genome RNA serves as a template for transcription to synthesize a leader RNA and six mRNAs as well as in replication to synthesize full-length genome RNA, both mediated by the viral RNA-dependent RNA polymerase. Recent studies demonstrate that participation of specific cellular proteins is critical for the regulation of gene expression of HPIV3 (4,5). Protein kinase Chas been impli-cated in the phosphorylation of the virion-associated RNA polymerase subunit, the phosphoprotein P (5). Introduction of protein kinase C--specific peptide inhibitor in cultured cells abrogated HPIV3 replication providing strong evidence that protein kinase Cis involved in the HPIV3 life cycle (5). Another cellular protein, actin, was found to be required in transcription of purified viral RNP in vitro and was found to be involved in maintaining a moderately coiled structure of the RNP that appeared to facilitate transcription of the genome RNA by the RNA polymerase (4). The productive infection of HPIV3, thus, appears to require a close encounter between the viral genome and several cellular proteins. A detailed search of such putative cellular proteins and their characterization would lead to better understanding of their roles in the regulation of the intricate steps in viral gene expression.
Sequence analysis of HPIV3 genome RNA reveals the presence of a sequence element at the 3Ј-end that serves as the binding site of the RNA polymerase to initiate synthesis of a 55-nucleotide plus-sense leader RNA followed by six monocistronic, capped, and polyadenylated mRNAs in vitro and in vivo (2,3). The leader RNA is involved in the initiation of assembly of viral nucleocapsid containing the plus-sense genome RNA that in turn serves as a template for the synthesis of minus strand genome RNA for packaging into progeny virions (6). Thus, the 3Ј-noncoding region of the genome RNA and the plus-sense leader RNA are the key cis-acting RNA sequence regions that presumably play important roles in the regulation of virus transcription and replication, respectively. A number of observations suggest that cellular proteins specifically interact with the viral cis-acting regulatory RNAs in several viral systems indicating their possible involvement in the regulation of viral gene expression (7)(8)(9)(10)(11)(12).
In this study, we searched for putative cellular proteins that might be involved in HPIV3 gene expression through interaction with the viral cis-acting regulatory RNAs. In a gel mobility shift assay, we used two regulatory RNAs described above, the 3Ј-genome sequence-containing RNA (3Ј-GS-RNA) and the leader sequence-containing RNA (LS-RNA) that are involved in binding of RNA polymerase and the viral nucleocapsid protein for transcription and replication, respectively. We have shown that the cellular glycolytic enzyme, GAPDH, and the nuclear antigen, LA protein, form specific complexes with these cisacting RNAs in vitro and in vivo. In addition, GAPDH is found to be co-localized with the viral RNP in HPIV3-infected cells. These results strongly suggest that both GAPDH and LA proteins are involved in the regulation of gene expression of HPIV3. the control of T7 promoter were constructed. To construct the plasmidcontaining 3Ј-GS, primer 1 (5Ј-AAG CTT TAA TAC GAC TCA CTA TAG  TCA ATG TCT TTA ATC C-3Ј) containing a HindIII site, T7 promoter  complementary sequence, and 17 nucleotides from the untranslated  region of the NP gene and primer 2 (5Ј-GGT ACC GAC GCT ATA TAC  CAA ACA AGA GAA GAA ACT TG-3Ј) containing KpnI-HgaI sites and a 22-nucleotide complementary sequence from the 3Ј-end of the leader region were synthesized (Operon Technologies, Inc.). These two primers were used in polymerase chain reaction containing pHPIV3-CAT plasmid DNA as template (13). Similarly, the plasmid containing LS-RNA was constructed using the pHPIV3-CAT DNA and the primer 3 (5Ј-AAG  CTT TAA TAC GAC TCA CTA TAG TCA ATG TCT TTA ATC C-3Ј)  containing a HindIII site, T7 promoter complementary sequence, and a  19-nucleotide complementary sequence from the leader region and  primer 4 (5Ј-GGT ACC GAC GCT ATA TGT CAA TGT CTT TAA  TCC-3Ј) containing KpnI-HgaI sites and 17 nucleotides from the untranslated region of the NP gene in a polymerase chain reaction. Sequences of the inserts in these constructs were confirmed by DNA sequencing. Radiolabeled RNAs were synthesized using these plasmid DNAs after linearization with HgaI in an in vitro transcription reaction containing [␣-32 P]UTP and T7 RNA polymerase according to the manufacturer's protocol (Boehringer Mannheim). The transcripts (73 nucleotides) would contain 55 nucleotides from the leader region and 18 nucleotides from the NP gene. The in vitro synthesized RNAs were analyzed in a 10% polyacrylamide-urea gel, and the radiolabeled RNA bands were excised. The RNAs were then eluted in a buffer containing 0.5 M ammonium acetate, 1 mM EDTA, and 0.1% SDS and purified by phenol extraction and ethanol precipitation.
Purification of Cellular Proteins-CV-1 cells were grown in monolayers in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The cells (5 ϫ 10 8 ) were harvested in phosphatebuffered saline (PBS) and pelleted by centrifugation at 1,000 ϫ g for 10 min. Lysate was prepared from these cells following the procedure described by Leopardi et al. (12). Briefly, the cell pellet was resuspended in 10 ml of buffer containing 10 mM Tris-HCl (pH 7.5) and 10 mM NaCl. The cells were lysed by 4 cycles of freezing and thawing, and the cell nuclei were removed by centrifugation at 1,000 ϫ g for 5 min. The lysate was further clarified by centrifugation through 30% glycerol containing 25 mM HEPES-KOH (pH 7.5) and 1 mM DTT at 150,000 ϫ g for 1 h. The soluble fraction from the top of the glycerol cushion was either directly used in gel-mobility shift assay or subjected to column chromatography. For column chromatography, the extract was dialyzed overnight against 1 liter of buffer A containing 25 mM Tris-HCl (pH 8.0), 5% glycerol, 0.1 mM EDTA, 50 mM NaCl, and 1 mM DTT and was loaded onto a DEAE-cellulose column (5 ml) equilibrated with the same buffer. The column was washed with 10 ml of buffer A, and the bound proteins were eluted with a linear 0 -0.5 M gradient of NaCl in buffer A (30 ml of total volume). The individual fractions (2-l aliquot) were used in gel mobility shift assay. Complex I-forming activity, present in the unbound fraction, was loaded onto a phosphocellulose column (3 ml) equilibrated with buffer A. The column was washed with the same buffer and eluted with a linear 0 -1 M gradient of NaCl in buffer A (20 ml). The complex I-forming activity, eluted from the column, was further purified by successive chromatography on DEAE and phosphocellulose columns. The fractions containing 3Ј-GS binding activity were pooled and concentrated, and the protein concentration was estimated as 1.5 mg/ ml. To purify the complex II forming activity, the pooled active fractions eluted from the first DEAE-cellulose column were loaded onto a Sephacryl S200 column (45 ϫ 1 cm) equilibrated with buffer A. The column was developed with buffer A, and individual fractions were monitored for binding radiolabeled LS RNA. The active fractions were pooled and stored at Ϫ20°C. The protein concentration in the pooled fraction was estimated as 0.5 mg/ml.
Gel Mobility Shift Assay-The binding of uniformly labeled 3Ј-GSand plus-sense LS-RNA with the cellular proteins was performed following the procedure described by Leopardi et al. (12). The reaction mixture (20 l) contained 15 mM HEPES (pH 8.0), 15 mM KCl, 0.25 mM EDTA, 0.25 mM DTT, 5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 200 g/ml yeast tRNA, 10% glycerol, 0.1 ng of radiolabeled RNA, and unless otherwise indicated, 0.5 g of purified cellular proteins. Incubation was done at room temperature for 30 min, and the samples were analyzed in a 6% native polyacrylamide gel in 0.5 ϫ TBE (0.045 M Tris-borate, 0.001 M EDTA) buffer. The gel was run at 150 V at room temperature and then dried and subjected to autoradiography.
UV Cross-linking-Binding of purified cellular protein to the radiolabeled RNA was performed at room temperature for 30 min in the gel mobility shift assay buffer in a 96-well plate. The reaction mixture (20 l) was then exposed to short wavelength UV light on ice at a 4-cm distance for 1 h. After UV cross-linking, the reaction mixture was incubated with RNase A (0.1 g) for 15 min at 37°C. The proteins were analyzed in a 10% SDS-polyacrylamide gel. The gel was stained, dried, and subjected to autoradiography.
Sequence Analysis-Proteins were resolved by electrophoresis in a 10% SDS-polyacrylamide gel and transferred onto polyvinylidene difluoride membrane according to the method of Matsudaira (14). The membrane was stained with Coomassie Blue, and the protein band was cut out. The protein was then sequenced on an Applied Biosystems model 470 sequenator equipped with on-line phenylthiohydantoin analysis using the regular program 03RPTH.
Immunoprecipitation of HPIV3 RNAs-Human lung carcinoma cells (A549) were grown in minimum Eagle's medium and were infected with HPIV3 at 10 PFU/cell. At 24 h postinfection, cell lysate was prepared according to Horikami and Moyer (15). The cell lysate was used for immunoprecipitation of HPIV3 RNA with anti-GAPDH and anti-LA protein according to Chang et al. (9). The RNA was purified from the precipitated complex by phenol extraction and ethanol precipitation and was analyzed by RNase protection assay using radiolabeled 3Ј-GSand LS-RNA probes according to Kurilla et al. (16).
Purification of Intracellular RNP-Intracellular RNP was isolated from HPIV3-infected CV-1 cells essentially as described by Toneguzzo and Ghosh (17), with slight modification. CV-1 cells in monolayer were infected with HPIV3 at 20 PFU/cell, and the cells were harvested at 20 h postinfection. The cells were washed with 10 mM phosphate buffer (pH 7.2) containing 0.15 M NaCl and disrupted in 10 mM Tris-HCl (pH 7.8) by sonication. The cell lysate was centrifuged at 10,000 ϫ g for 10 min, and finally the RNP was purified from this supernatant by centrifugation through a 30% glycerol cushion at 40,000 rpm in an SW 50.1 rotor. The pelleted RNP was suspended in 10 mM Tris-HCl (pH 7.8) containing 10% glycerol and 1 mM DTT and stored in liquid nitrogen.
Immunofluorescent Labeling-CV-1 cells were grown on coverslips and infected with HPIV3 at 1 PFU/cell. At 24 h postinfection, the cells were washed with phosphate-buffered saline followed by fixation with 3.6% paraformaldehyde and permeabilization with 1% Nonidet P-40. The fixed cells were treated with a mixture of rabbit anti-RNP and monoclonal anti-GAPDH or of rabbit anti-RNP and human anti-LA antibodies (18). For double labeling of RNP and GAPDH, the coverslips were washed and incubated with a mixture of fluorescein-conjugated anti-rabbit Ig and biotin-conjugated anti-mouse Ig secondary antibodies, followed by incubation with Texas Red-conjugated avidin. For double labeling of RNP and LA protein, the coverslips were washed and incubated with a mixture of fluorescein-conjugated anti-human Ig and biotin-conjugated anti-rabbit Ig secondary antibodies, followed by incubation with Texas Red-conjugated avidin. The coverslips were finally washed, mounted, and examined using a Leica CLSM confocal laser scanning microscope.

Interaction of Cellular Proteins with HPIV3 Cis-acting
RNAs-To identify cellular proteins that might be involved in HPIV3 gene expression, we inserted cDNA copies of the first 73 nucleotides from the 3Ј-end of the genome as well as its complementary sequence into transcription vectors under the control of T7 promoter referred to as pUC3ЈGS and pUCLS, respectively. Transcription of the plasmid pUC3ЈGS by T7 RNA polymerase after linearization with HgaI gives rise to 3Ј-GS-RNA, while transcription from the plasmid pUCLS yielded the plus-sense LS-RNA (Fig. 1). The regulatory elements present within the 3Ј-GS-RNA include sites for the binding of viral RNA polymerase and putative cellular factors and also the intergenic trinucleotide GAA and NP gene start sequence that are believed to be involved in termination of the leader RNA and the initiation and capping of nascent RNA transcripts. The LS RNA, on the other hand, contains sites for initiation of encapsidation by NP and for interaction of viral and cellular proteins near the first intergenic region at which termination of RNA transcripts must be suppressed during replication. Radiolabeled 3Ј-GS-and LS-RNAs were used in gel mobility shift assays with CV-1 cell cytoplasmic proteins. As shown in Fig. 1B, cellular proteins formed two distinct complexes with the LS RNA (complex I and II) and virtually one complex with the 3Ј-GS-RNA (complex I). The complexes with both 3Ј-GSand LS-RNA probes were abolished in the presence of 40-fold excess of corresponding unlabeled RNA, whereas 400-fold excess of unrelated competitor RNAs had no effect. In competition experiments with unlabeled heterologous RNA probes at 10fold excess, complex I was significantly inhibited by both RNA probes, whereas 50-fold excess 3Ј-GS-RNA was required to inhibit the formation of complex II with LS RNA (data not shown). These results indicate that the formation of complex I most likely involves both sequence and structure of the RNA and that the same proteins are involved in interaction with the two RNA probes. Similar complexes were also formed when extracts were prepared from other cell lines such as human lung carcinoma (A549) and baby hamster kidney (data not shown), indicating the ubiquitous nature of the cellular proteins that formed complexes with the 3Ј-GS-and LS-RNAs of HPIV3. Next, we fractionated the CV-1 cell cytoplasmic extract using a DEAE-cellulose column where the complex I-forming activity was present in the unbound fraction and the complex II-forming activity was eluted from the column at around 0.4 M NaCl concentration (Fig. 2). Because the complex I-and IIforming activities were separable, we reasoned that two separate proteins were involved and set out to characterize the putative RNA-binding proteins.

Glyceraldehyde-3-phosphate Dehydrogenase Is Involved in
Forming Complex I-To characterize the cellular protein that formed complex I, we first performed UV cross-linking analysis with the DEAE-cellulose unbound fraction. However, our attempt to identify the polypeptide directly by UV cross-linking failed, and accordingly we sought complete purification of the protein. The DEAE-cellulose unbound fraction was loaded onto a phosphocellulose column, and the bound proteins were eluted with a linear 0 -1 M NaCl gradient. The complex I-forming activity was eluted around 0.5 M NaCl concentration (data not shown). The active fractions were pooled and subjected to further purification by successive chromatography on DEAE-cellulose and phosphocellulose columns. The complex I-forming protein was purified to near homogeneity, and molecular mass was estimated as ϳ37 kDa by SDS-polyacrylamide gel electrophoresis (Fig. 3A). For further characterization, we performed microsequence analysis of the protein and compared it with the protein sequences available in the data base. As shown in Fig.  3, the partial sequence of the purified protein was virtually identical to the N terminus of bovine glyceraldehyde-3-phosphate dehydrogenase, a 37-kDa glycolytic enzyme (19). Consistent with these findings, the purified 37-kDa polypeptide reacted with a monoclonal anti-GAPDH antibody in Western blot analysis (Fig. 3B). Moreover, a commercial preparation of rabbit GAPDH (Boehringer Mannheim) also contained similar 3Ј-GS-RNA binding activity (Fig. 3C). Taken together, these data provide strong evidence that the complex I-forming protein is, in fact, GAPDH. Since GAPDH is not a bona fide RNA-binding protein but was shown to interact with poly(U) (20, 21), we investigated whether the stretches of U residues present in 3Ј-GS-RNA are involved in the interaction with GAPDH. As shown in Fig. 4A, the formation of complex I was inhibited by about 90% in the presence of 200-fold excess of poly(U), whereas a similar concentration of viral NP and P mRNAs had no effect, suggesting a role of U residues in this interaction. It is well documented that GAPDH contains an NAD ϩ binding site that is conserved among dehydrogenases (22) and has also been shown to be involved in the binding of GAPDH to AU-rich RNA sequences (23). Therefore we tested whether this site is involved in the binding of 3Ј-GS-RNA. As shown in Fig. 4A, a high concentration of NAD ϩ (Ͼ10 M) was required to abolish 3Ј-GS-RNA binding activity, and other dehydrogenases such as glucose-6-phosphate dehydrogenase and lactate dehydrogenase virtually failed to bind 3Ј-GS-RNA (Fig.  4B), suggesting that the NAD ϩ binding domain constitutes a part of the 3Ј-GS-RNA binding site.
Cellular Autoantigen, La, Is Involved in Forming Complex II-To characterize the complex II-forming protein, we pooled the fractions containing LS RNA binding activity that eluted at around 0.4 M NaCl concentration from the DEAE-cellulose column (Fig. 2) and used them for further purification. The pooled fraction was subjected to chromatography on a Sephacryl S-200 column where the activity was eluted in a single peak (data not shown), and the purified protein was referred to as LSBP. As shown in Fig. 5A, the purified fraction contained several protein bands in a silver-stained SDS-polyacrylamide gel. To identify the polypeptide that is directly involved in the interaction with LS RNA, UV cross-linking was performed. As shown in Fig. 5A, an ϳ50-kDa polypeptide was cross-linked to the radiolabeled LS RNA indicating its involvement in the formation of complex II. Involvement of the same polypeptide in the formation of complex II with 3Ј-GS-RNA, albeit at a low level, was also confirmed by UV cross-linking with radiolabeled 3Ј-GS-RNA (data not shown). We speculated that the 50-kDa protein identified in our study might be the cellular LS protein since in previous studies specific interaction of cellular LA protein, a bona fide RNA-binding protein, was shown to interact with several viral RNAs (7-9, 24). To examine this possibility, we performed Western blot analysis of the purified LSBP with anti-LA antibody. As shown in Fig. 5B, the anti-LS antibody   FIG. 2. Chromatographic separation of complex I-and II-forming proteins. The CV-1 cell cytoplasmic proteins (S100) were subjected to chromatography on a DEAE-cellulose column as described under "Experimental Procedures." Individual fractions (2 l) were used in gel mobility shift assay with radiolabeled LS RNA. The complexes were analyzed by electrophoresis in 6% polyacrylamide gel. The gel was dried and subjected to autoradiography. The numbers at the top indicate the fraction eluted from DEAE-cellulose column, and U represents the unbound fraction. The migration positions of complex I and II are indicated on the right.

FIG. 3. Characterization of complex I-forming protein.
The complex I Ϫforming protein referred to as GSBP was purified using DEAE-cellulose and phosphocellulose columns. A, polypeptide pattern of the purified GSBP. The purified protein was analyzed in a 10% SDS-polyacrylamide gel followed by silver staining. The sequence of the protein obtained by microsequence analysis is shown at the top. The numbers indicate the amino acid position of the corresponding protein. B, Western blot analysis of GSBP with anti-GAPDH. Purified GSBP (1.5 g) as well as commercial GAPDH (Boehringer Mannheim) (4 g) were analyzed in a 10% SDS-polyacrylamide gel and electroblotted onto a GeneScreen membrane. The blot was developed with monoclonal anti-GAPDH (Biodesign International) followed by peroxidase-conjugated goat anti-mouse Ig. Finally the antigen-antibody complex was detected by ECL reagent (Amersham Corp.). C, gel mobility shift assay with GSBP. The GSBP and commercial GAPDH at amounts as indicated were incubated with radiolabeled 3Ј-GS-RNA, and the complex was analyzed in 6% polyacrylamide gel. The gel was dried and subjected to autoradiography.
specifically recognized the 50-kDa protein and control recombinant LA protein, strongly suggesting that LA protein is involved in the formation of complex II. To further confirm its identity, the radiolabeled LS RNA was incubated with purified LSBP, and the complex was immunoprecipitated with anti-LA antibody using protein A-Sepharose. As the control, we used polyclonal anti-actin antibody in immunoprecipitation of radiolabeled LS RNA under identical conditions. As shown in Fig.  5C, only anti-LA antibody effectively precipitated the 73-nucleotide radiolabeled LS RNA. Finally, we used recombinant LA protein to study its ability to form complex II in a gel mobility shift assay using LS RNA. As shown in Fig. 5D, the recombinant LA protein efficiently interacted with the HPIV3 LS RNA forming complex II. This complex formation was solely mediated by the recombinant LA protein because identically processed bacterial proteins did not form any complex (Fig. 5D). These results clearly indicate that the cellular LA protein is directly involved in the formation of a specific complex, complex II, with the HPIV3 LS RNA.
GAPDH and LA Protein Associate with HPIV3 RNP in Vivo-The above results prompted us to investigate whether GAPDH also interacts with the viral cis-acting RNA in HPIV3infected cells. Monoclonal anti-GAPDH antibody was used to immunoprecipitate viral RNA, and the precipitated RNA was detected by RNase protection assay using radiolabeled LS RNA probe. As shown in Fig. 6A, radiolabeled LS RNA probe detected viral RNA precipitated by anti-GAPDH antibody from HPIV3-infected cells, indicating in vivo association of this cellular protein with HPIV3 RNA. Some precipitation of the genome-sense RNA by anti-actin antibody is observed that is possibly due to the association of actin with the RNP as a transcription factor (4). Next, we examined whether GAPDH remains associated with the viral RNP in HPIV3-infected cells. Intracellular RNP was isolated at different times postinfection, and the presence of GAPDH was determined by Western blot analysis. As shown in Fig. 6B, GAPDH was specifically associated with the viral RNP during infection. Similarly, in vivo association of LA protein with HPIV3 RNA was investigated by immunoprecipitation of viral RNA with anti-LA antibody and analysis of the precipitated RNA by RNase protection assay using radiolabeled 3Ј-GS-RNA probe. As shown in Fig. 7A, the radiolabeled 3Ј-GS-RNA probe detected the viral RNA precipitated by anti-LA antibody, thus confirming in vivo association of HPIV3 RNA with the cellular LA protein. We also examined whether the interaction of LA protein with the HPIV3 RNA leads to a specific association of this cellular protein with the viral RNP. The intracellular viral RNP was isolated, and the presence of LA protein was determined by Western blot analysis using anti-LA antibody. As shown in Fig. 7B, the LA protein was detected in the RNP as early as 8 h postinfection and continued to be present in the RNP during the virus life cycle.
Co-localization of GAPDH with HPIV3 RNP in the Infected Cells-Finally, to further confirm the interaction of viral RNP with these cellular proteins in the HPIV3-infected cells, indirect double immunofluorescence labeling and confocal microscopy were carried out. As illustrated in Fig. 8, GAPDH was   FIG. 5. Characterization of complex II-forming protein. The complex II-forming protein was purified by chromatography on DEAEcellulose, phosphocellulose, and Sephacryl S200 columns and referred to as LSBP. A, UV cross-linking of LSBP to the radiolabeled LS RNA. LSBP was incubated with radiolabeled LS RNA, and the resulting complex was subjected to UV irradiation as described under "Experimental Procedures." The cross-linked protein was identified in a 10% SDS-polyacrylamide gel analysis and autoradiography, and shown on the right is the polypeptide pattern of LSBP in a silver-stained gel. B, Western blot analysis of LSBP using anti-LA antibody. The LSBP and bacterially expressed LA protein (2 g each) were analyzed in 10% SDS-polyacrylamide gel and electroblotted onto GeneScreen membrane. The blot was developed by human anti-LA serum followed by rabbit anti-human Ig. The blot was finally treated with ECL reagent. C, immunoprecipitation of LSBP⅐LS-RNA complex with anti-LA antibody. Radiolabeled LS RNA was incubated with LSBP, and the complex was immunoprecipitated as described under "Experimental Procedures." The precipitated RNA was purified by phenol extraction and ethanol precipitation. The RNA was then analyzed in a 10% polyacrylamideurea gel. D, gel mobility shift assay with recombinant LA protein. The LA protein was expressed in bacteria (Bact) and purified, and as a control the proteins from bacteria harboring pET vector were processed in a similar manner. The purified proteins were used in gel mobility shift assay with radiolabeled LS RNA. Migration positions of molecular markers in kDa are shown. nt, nucleotide.

FIG. 6. In vivo interaction of GAPDH with HPIV3 components.
The interaction of GAPDH with HPIV3 RNA as well as the RNP in HPIV3-infected A549 cells were studied. A, detection of 3Ј-GS sequence containing RNA in GAPDH-bound form in HPIV3-infected A549 cells. Confluent monolayer of A549 cells was infected with HPIV3 at 10 PFU/cell, and at 24 h postinfection, the cell lysate was prepared. The cell extract was processed for immunoprecipitation with anti-GAPDH antibody as described under "Experimental Procedures." The precipitated RNA was purified, and the presence of 3Ј-GS sequence was examined by RNase protection assay with radiolabeled LS RNA as probe. The protected RNA was then analyzed by electrophoresis in 10% polyacrylamide-urea gel. B, in vivo association of GAPDH with HPIV3 RNP. CV-1 cells were infected with HPIV3 at 10 PFU/cell, and at 24 h postinfection (p.i.), intracellular RNP was isolated as described under "Experimental Procedures." The presence of GAPDH in the RNP was monitored by Western blot with anti-GAPDH antibody. nt, nucleotide. labeled predominantly in the cytoplasm of uninfected CV-1 cells with a perinuclear distribution (panel A). This distribution pattern of GAPDH remained unaltered in HPIV3-infected cells (panel B). An apparent increase in the level of GAPDH, following HPIV3 infection (compare panel A with B), is possibly due to formation of a multinucleated giant cell, which is typical of HPIV3 glycoprotein-mediated cell fusion. Interestingly, the viral RNP was also labeled in the cytoplasm with similar perinuclear distribution (panel C), suggesting specific interaction between the viral RNP and GAPDH. This notion was supported by the fact that, when confocal images were simultaneously acquired for both fluorochromes, the RNP and GAPDH were found to co-localize (panel D) in the HPIV3-infected cells. Immunolabeling of another cytoplasmic protein, tubulin, showed no co-localization with RNP, indicating that the interaction between the RNP and GAPDH was specific (data not shown). Similarly, we carried out double immunofluorescent labeling and confocal microscopy to examine specific interaction of viral RNP with the LA protein (data not shown). The LA protein was present primarily in the nucleus, and upon HPIV3 infection, a detectable amount was found to be redistributed in the cytoplasm in the perinuclear region where viral RNP was also present. However, co-localization as observed for GAPDH (panel D, yellow) could not be demonstrated due to low amounts of LA protein and the vast excess of RNP present in the same region. Nevertheless, these results strongly suggest that LA protein also becomes available for interaction with viral RNP in the cytoplasm.

DISCUSSION
In the present study, we have identified and characterized two cellular proteins, GAPDH, a cytoplasmic protein, and La, primarily a nuclear protein, that specifically interact with HPIV3 cis-acting regulatory RNAs, 3Ј-GS-and LS-RNA, in vitro as well as in vivo. We have also demonstrated that the same two cellular proteins interact with viral RNP in HPIV3infected cells. Moreover, by double immunoflourescent labeling and confocal microscopy, we have shown that GAPDH specifically co-localizes with viral RNP in the infected cells. The biological significance of these interactions, however, remain unknown at present. Nevertheless, both in vitro and in vivo specific interactions of GAPDH and LA protein with viral RNP strongly suggests that they must play a role in the life cycle of FIG. 7. In vivo association of LA protein with HPIV3 components. Interaction of LA protein with HPIV3 RNA and the intracellular viral RNP was studied. A, detection of LS RNA in La-bound form in HPIV3-infected A549 cells. The A549 cells were infected with HPIV3, and cell extract was prepared as described in Fig. 6A. RNase protection assay was done using radiolabeled 3Ј-GS-RNA as the probe. The protected RNA was analyzed in 10% polyacrylamide-urea gel. B, in vivo association of LA protein with HPIV3 RNP. The presence of LA protein in the viral RNP was determined following the procedure as described in Fig. 6B  HPIV3. The involvement of GAPDH is particularly interesting and unexpected because it is not a bona fide RNA-binding protein, and its specific association with a virus has not been demonstrated before. It is primarily involved in cellular metabolism as the key enzyme of the glycolytic pathway. There are earlier reports where GAPDH has been implicated in binding to single-stranded RNA in polyribosomes (20,25). Only recently a sequence-specific interaction of GAPDH with tRNAs (21) and AU-rich RNA sequences present in the 3Ј-untranslated region of several mRNAs has been reported (23). Since both 3Ј-GS-and LS-RNA contain AU-rich sequences, it is possible that GAPDH binds to these sequences. In addition, both cis-acting RNAs contain a similar stem-loop structure (Fig. 1) that could also be a part of the recognition site. The important question still remains with respect to the molecular basis of this selective interaction of GAPDH with HPIV3 RNA. Clearly, development of a reconstituted transcription or replication system in vitro using purified GAPDH would help delineate its role in these RNA synthetic processes.
It is important to note that GAPDH does not contain any consensus RNA binding motif similar to other well known RNA-binding proteins (26). In this regard, GAPDH appears to be similar to iron-responsive element binding protein (27), several small nuclear RNPs (28), and calreticulin (11), which bind to RNA but do not possess such RNA binding domains. The RNA binding properties of these proteins appear to be regulated by respective co-factors (21, 27) or by modification of the protein such as by phosphorylation (11). For GAPDH, the regulating co-factor is NAD ϩ , and its binding site within GAPDH is commonly referred to as the Rossmann fold, which is conserved among dehydrogenases (22). Our findings that the interaction between 3Ј-GS-RNA and GAPDH is inhibited in vitro by NAD ϩ , albeit at high concentration (10 M) (Fig. 4A), suggest that the Rossmann fold may be involved in this interaction. Since other dehydrogenases such as lactate dehydrogenase and glucose-6-phosphate dehydrogenase did not bind 3Ј-GS-RNA (Fig. 4B), it suggests that the Rossmann fold may constitute only a part of the 3Ј-GS-RNA binding site in GAPDH. HPIV3 infection may also lead to a significant decrease in the intracellular concentration of NAD ϩ , as observed recently in human immunodeficiency virus, type I-infected cells (29) leading to inhibition of GAPDH activity with impairment of cellular functions. Thus, it remains to be determined whether HPIV3 may utilize some other activity of this cellular enzyme for its own replication while inhibiting glycolytic function of GAPDH. In this regard, it is particularly interesting to note that GAPDH also interacts with cellular actin (30), which has been shown to be involved in the activation of HPIV3 transcription (4). A detailed study along these lines would lead to better understanding of this unique host-virus interaction process.
The other cellular protein identified is the autoantigen LA protein, which binds specifically to the plus-sense leader RNA in vitro (Fig. 1). This protein also is found to be associated with HPIV3 RNP during infection suggesting again its possible role in virus replication. Cellular LA protein is a ubiquitous phosphoprotein and a bona fide RNA-binding protein found predominantly in the nucleus of cells (24,31), and it was first identified as a target antigen of autoantibodies found in the sera of patients with systemic lupus erythematosus and Sjögren's syndrome (32). Interest in the LA protein was greatly stimulated by the finding that it binds to several RNA polymerase III transcripts (33) and facilitates their release from the template (34). Recently, cellular LA protein has also been shown to bind some viral RNAs such as adenovirus VA RNAs (35), Sindbis virus minus strand genome RNA (36), Epstein-Barr virus EBER RNAs (37), vesicular stomatitus virus and rabies virus leader RNAs (7,8,38), 5Ј-untranslated region of poliovirus RNA (24), and human immunodeficiency virus trans-activation response element RNA (9). Although the biological significance of the interaction between viral RNAs and cellular LA protein remained undefined, a specific role of LA protein in viral gene expression has more recently begun to emerge. For example, translation of poliovirus mRNA has been shown to require specific binding of LA protein to the 5Ј-untranslated region that relieves the structural constraint (39). Similarly, in the case of human immunodeficiency virus, the interaction of LA protein with the TAR element present at the 5Ј-end of the viral mRNAs was found to alleviate the translation repression by the TAR element (40). It is important to note that most of the viral RNAs reacting with LA are short, uncapped, and nonpolyadenylated, and the LA protein forms ribonucleoprotein complexes with these RNAs. Our studies also indicate that LA protein forms a ribonucleoprotein complex with the HPIV3 leader RNA because anti-LA antibody precipitated the La-bound leader RNA but not the free RNA. It is interesting to note that in the HPIV3 system the LA protein bound to leader RNA in vivo, which is elongated beyond the leader size (55 nucleotide) (Fig. 7). Leader length RNA (55 nucleotide), as found in VSV (7), was not detectable in HPIV3-infected cells raising the possibility that efficient elongation of RNA chains may occur once LA protein is bound to the nascent HPIV3 leader RNA. Thus, it would be interesting to determine whether LA protein binds at the intergenic region of the LS RNA. A consensus RNA motif for binding of LA protein has not been identified, and certain RNA sequences within the structural context, especially 3Јoligoribouridylate sequence, are believed to be involved in this interaction (41). The HPIV3 plus-sense LS RNA does not contain long stretches of U sequences, however, internal di-and triuridylate repeats are noticeable. Moreover, as stated above, the secondary structure of LS RNA (Fig. 1) may also be involved in LA protein recognition. It should be noted that the 3Ј-GS-RNA, although containing U-rich sequences, does not interact with the LA protein (Fig. 1). Thus, the selective interaction with the LS RNA underscores an important role of LA protein in HPIV3 replication. Perhaps LA protein acts as an anti-terminator during the replicative process. Again, an in vitro transcription/replication system for HPIV3 will be needed to study the function of LA protein in the HPIV3 life cycle.
Finally, both GAPDH and LA protein appear to play a role in the life cycle of the virus not only due to their ability to bind to the cis-acting viral RNA sequences but also to the fact that they specifically associate with the RNP in the infected cells (Figs. 6 and 7), which is confirmed by double immunofluorescent labeling studies (Fig. 8). The co-localization of GAPDH and viral RNP demonstrates for the first time the possible involvement of a key metabolic enzyme in the life cycle of the virus. It remains to be seen at which steps of the virus replicative pathway GAPDH acts. The association of LA protein with RNP is interesting since it is essentially a nuclear protein, whereas HPIV3 replicates in the cytoplasm. However, by immunofluorescent studies (data not shown), a detectable amount of LA protein seems to be released in the cytoplasm (relative to uninfected control) following HPIV3 infection, as observed for poliovirus infection (24). Experiments are in progress to address the significance of these interactions by delineating the role of GAPDH and LA protein in HPIV3 gene expression.