Cellular Protein HAX1 Interacts with the Influenza A Virus PA Polymerase Subunit and Impedes Its Nuclear Translocation

Abstract

Transcription and replication of the influenza A virus RNA genome occur in the nucleus through the viral RNA-dependent RNA polymerase consisting of PB1, PB2, and PA. Cellular factors that associate with the viral polymerase complex play important roles in these processes. To look for cellular factors that could associate with influenza A virus PA protein, we have carried out a yeast two-hybrid screen using a HeLa cell cDNA library. We identified six cellular proteins that may interact with PA. We focused our study on one of the new PA-interacting proteins, HAX1, a protein with antiapoptotic function. By using glutathione S-transferase pulldown and coimmunoprecipitation assays, we demonstrate that HAX1 specifically interacts with PA in vitro and in vivo and that HAX1 interacts with the nuclear localization signal domain of PA. Nuclear accumulation of PA was increased in HAX1-knockdown cells, and this phenotype could be reversed by reexpression of HAX1, indicating that HAX1 can impede nuclear transport of PA. As a consequence, knockdown of HAX1 resulted in a significant increase in virus yield and polymerase activity in a minigenome assay, and this phenotype could be reversed by reexpression of HAX1, indicating that HAX1 can inhibit influenza A virus propagation. Together, these results not only provide insight into the mechanism underlying nuclear transport of PA but also identify an intrinsic host factor that restricts influenza A virus infection.

INTRODUCTION

Influenza A virus is an orthomyxovirus and an important pathogen of humans and animals. The viral genome consists of eight segments of negative-sense single-stranded RNA which are encapsidated as viral ribonucleoprotein complexes (vRNPs) by multiple copies of nucleoprotein (NP) and the trimeric RNA-dependent RNA polymerase consisting of PB1, PB2, and PA (1). Unusually among RNA viruses, influenza A virus transcribes and replicates its genome in the nucleus. The influenza A virus RNA genome (vRNA) is transcribed into mRNA and replicated through cRNA (complementary copy of vRNA) to produced a large number of progeny vRNA by the trimeric RNA polymerase. As a consequence, the polymerase subunits (PB1, PB2, and PA), which are produced in the cytoplasm, must be imported into the nucleus and assembled into a functional trimer (2). Indeed, nuclear localization signals (NLSs) have been identified on PB1 (3), PB2 (4–6), and PA (7), and it has been demonstrated that individually expressed PB1, PB2, and PA can enter the nucleus (3–5, 7–10). Various models have been proposed for the nuclear import and assembly of viral polymerase complex (11–15). However, on the basis of in vitro assembly observations and live-cell imaging studies, the following model is favored: PB1 and PA are imported into the nucleus as a dimer, and they then associate with separately imported PB2 to form the functional trimeric polymerase in the nucleus (11, 13, 14, 16). Cellular proteins play important roles in the nuclear transport of PB1, PB2, and PA. For instance, RanBP5 facilitates nuclear import of the PB1-PA dimer by interacting with PB1 (17–19). Hsp90 interacts with PB1 and PB2 and is involved in the nuclear transport and assembly of the viral RNA polymerase complex (15). PB2 is thought to enter the nucleus by interacting with importin isoforms (20, 21), and the interaction between PB2 and different importin isoforms governs the cell tropism of influenza A viruses (20–23). Although cellular proteins that promote nuclear transport of influenza A virus RNA polymerase subunits have been well characterized, cellular proteins that inhibit this process remain poorly defined. A recent study by Hudjetz and Gabriel showed that human importin-α3 acts as a negative regulator of human- and avian-like polymerase activity in vRNP reconstitution assays (23). However, the mechanism underlying importin-α3 inhibition of vRNP polymerase activity is not known.

During influenza A virus replication, the PB1 subunit plays a central role in the catalytic activities of the RNA polymerase and is directly involved in RNA synthesis (24–26). In the process of viral mRNA synthesis, both PB2 and PA are involved in the cap-snatching reaction to generate capped RNA primers for viral transcription. Whereas the PB2 subunit is responsible for recognition and binding of the cap structure of the host mRNAs (27–30), the PA subunit, which possesses endonuclease activity, cleaves the cap from host mRNA to generate capped RNA primers (31, 32). In addition to having endonuclease activity, PA has been shown to possess protease activity; however, the biological significance of this activity for viral replication remains obscure (33–35). Genetic analysis of PA suggests that PA is involved in several functions of the polymerase complex, including endonuclease activity, promoter binding, and serving as an elongation factor during RNA synthesis (36–39). Therefore, PA is involved in not only virus genome replication but also transcription.

Influenza A virus, as other viruses, is an intracellular parasite that exploits host cell machinery to multiply. For efficient viral genome replication and transcription, influenza A virus RNA polymerase needs to interact with cellular proteins. Knowledge of the interaction between viral RNA polymerase and cellular proteins not only informs us of the molecular mechanism underlying viral genome replication and transcription but also provides further targets for antiviral drug development. A number of cellular proteins that interact with influenza A virus RNA polymerase (or its subunits) have been identified (reviewed in references 40 and41). With regard to PA-interacting cellular proteins, PA has been shown to interact with hCLE, a putative transcriptional activator (42), and MCM, a putative DNA replication fork helicase (43). More recently, Bradel-Tretheway et al. have identified more than 100 cellular proteins that can interact with PA by liquid chromatography-tandem mass spectrometry (44). Notably, many of these PA-associated proteins are mitochondrial proteins (44).

To further search for cellular proteins that interact with PA, we performed a yeast two-hybrid screen of a HeLa cDNA library using PA as bait. Our screen data indicated that HAX1 is a new PA-interacting protein. HAX1 was originally identified as a protein associated with hematopoietic lineage cell-specific protein 1 (HS1) (45). HAX1 is a 35-kDa cytoplasmic protein with homology to Bcl-2 and is an antiapoptotic protein (45, 46). Previous studies have shown that HAX1 can interact with several viral proteins (46–50). One interesting study showed that HAX1 can interact with human immunodeficiency virus (HIV) Rev protein and inhibit Rev function by relocating Rev from the nucleus to the cytoplasm (49). In this study, we characterized the interaction between PA and HAX1. Our data demonstrate that (i) PA interacts with HAX1 through the NLS domain of PA, (ii) HAX1 inhibits nuclear accumulation of PA, and (iii) HAX1 inhibits influenza A virus replication. Together, these data suggest that HAX1 binding of PA is one of the defense mechanisms that the host uses to limit influenza A virus infection.

MATERIALS AND METHODS

Cell culture.

293FT (human embryonic kidney 293 cells transfected with simian virus 40 large T antigen), MDCK (the Madin-Darby canine kidney cell line), H1299 (a human non-small-cell lung carcinoma cell line), and H1299-derived cells were grown in Dulbecco's modified Eagle's medium (DMEM; HyClone, Logan, UT) supplemented with 10% fetal bovine serum (Gibco BRL, Gaithersburg, MD) in 5% CO₂. MDCK cells were used to amplify influenza A virus (A/WSN/33) and for plaque assays.

Plasmids.

Plasmids pPOL I-CAT-RT, pcDNA-PB1, pcDNA-PB2, pcDNA-PA, and pcDNA-NP were kindly provided by George G. Brownlee (51, 52). Plasmid pcDNA-PB1-GFP was kindly provided by Ervin Fodor (16). pEGFP-C1 was from Clontech Company (Mountain View, CA). pcDNA-BAX was constructed by inserting BAX cDNA into the pcDNA3.1 vector (Invitrogen, Grand Island, NY). Plasmids p3×Myc-PA (encoding PA tagged with 3 copies of Myc) and pGEX6P-1-PA (encoding PA fused with glutathione S-transferase [GST]) were constructed by inserting influenza A/WSN/33 virus PA cDNA, which was PCR amplified from pcDNA-PA, into the BamHI/EcoRI sites of M3-pcDNA3-hygromycin (53) and pGEX6P-1 (GE Healthcare Life Sciences, Piscataway, NJ), respectively. The plasmids encoding deletion mutants of PA were constructed by inserting PCR-amplified PA fragments into the BamHI/EcoRI sites of M3-pcDNA3-hygromycin. All constructs have an NH₂-terminal triple Myc tag. The primer pairs used for constructing various PA deletion mutants are as follows: for the Δ1 mutant, forward primer 5′-ATGGGATCCATATGGAAGATTTTGTGCGACAAT-3′ and reverse primer 5′-CTAGAATTCCTATCTCAATGCATGTGTGAG-3′; for the Δ2 mutant, the same forward primer as for the Δ1 mutant and reverse primer 5′-CTAGAATTCCTACACAAAGTTTACCACATCGGT-3′; for the Δ3 mutant, forward primer 5′-ATGGGATCCATGTAAATGCTAGAATTGAACCTTT-3′ and reverse primer 5′-CTAGAATCCCTATCTCAATGCATGTGTGAG-3′; for the Δ4 mutant, forward primer 5′-ATGGGATCCATACCGATGTGGTAAACTTTGTG-3′ and the same reverse primer as for the Δ3 mutant; and for the Δ5 mutant, forward primer 5′-ATGGGATCCATGTAAATGCTAGAATTGAACCTTT-3′ and reverse primer 5′-CTAGAATTCCTATCTCAATGCATGTGTGAG-3′. The internal deletion Δ6 mutant was constructed by a PCR-based overlap extension method (54) as described below. The first PCR generated two fragments using two pairs of primers: primers 5′-ATGGGATCCATATGGAAGATTTTGTGCGACAAT-3′ (forward) and 5′-TGGACATTTGTGTTACTCCAATTTCGATGAATC-3′ (reverse) and primers 5′-TGGAGTAACACAAATGTCCAAAGAAGTAAATGC-5′ (forward) and 5′-CTAGAATCCCTATCTCAATGCATGTGTGAG-3′ (reverse). The two fragments were then used as the template for a second PCR with primer pair 5′-ATGGGATCCATATGGAAGATTTTGTGCGACAAT-3′ (forward) and 5′-CTAGAATCCCTATCTCAATGCATGTGTGAG-3′ (reverse). The fragment obtained from the second PCR was inserted into the BamHI/EcoRI sites of M3-pcDNA3-hygromycin to generate the Δ6 mutant. The pPOL I-Luc-RT reporter plasmid was constructed by removing the chloramphenicol acetyltransferase (CAT) sequence (using BglII and SpeI restriction enzymes) from pPOL I-CAT-RT (52) and replacing it with the luciferase sequence (open reading frame in the minus sense). The pcDNA-HA-HAX1-R plasmid, which contains silent mutations in the target sequence of HAX1 short hairpin RNA (shRNA), was derived from pcDNA-HA-HAX1 (46) by site-directed mutagenesis using QuikChange II site-directed mutagenesis kits (Stratagene, Santa Clara, CA) with primers 5′-GAGGGTCAGACTCTTCGCGACTCTATGCTTAA-3′ (forward) and 5′-TTAAGCATAGAGTCGCGAAGAGTCTGACCCTC-3′ (reverse). The ferritin heavy polypeptide 1 (FTH1)-encoding plasmid pCMV-HA-FTH1 was constructed by inserting the FTH1 sequence, which was PCR amplified from the cDNA of 293T cells with primers 5′-CCGAATTCTTATGACGACCGCGTCCACCT-3′ (forward) and 5′-CTCAGATCTTTAGCTTTCATTATCACTGTCTC-3′ (reverse), into the EcoRI/BglII sites of pCMV-HA (Clontech).

Antibodies.

Anti-Myc (A-14), antihemagglutinin (anti-HA; F-7), anti-GST (1C9), and antitubulin (TU-02) antibodies were from Santa Cruz Biotechnology (Santa Cruz, Santa Cruz, CA). Anti-HAX1 (610825) antibody was from BD Bioscience Company (Franklin Lakes, NJ). Anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH; GTX100118), anti-green fluorescent protein (anti-GFP; GTX113617), and anti-PA (GTX118991) antibodies were from GeneTex Company (Irvine, CA). Anti-p84 (5E10) antibody was from Novus Biologicals Company (Littleton, CO). Goat anti-rabbit IgG–fluorescein isothiocyanate (FITC) (65-6111) and rabbit anti-mouse IgG–Alexa Fluor 594 (A-11062) antibodies were from Invitrogen Life Technologies Corporation. Anti-caspase-3 (3G2 clone) antibody was from Cell Signaling Technology (Danvers, MA).

Yeast two-hybrid screen.

The full-length coding sequence of PA was amplified by PCR from pcDNA-PA and cloned into pGBDU-C1 (55), which contains the DNA binding domain of the Saccharomyces cerevisiae GAL4 transcription factor and the uracil gene, to create a bait plasmid (pGBDU-C1-PA). The prey library used is composed of a pool of prey plasmids which contain HeLa cell cDNAs (Clontech) fused to the activation domain of the yeast GAL4 transcription factor and the leucine gene. The two-hybrid screen was conducted essentially as described previously (55). The prey library was transformed into the Saccharomyces cerevisiae PJ69-4A strain containing the PA bait plasmid pGBDU-C1-PA. Positive clones were selected on synthetic complete (SC) agar plates lacking uracil, leucine, histidine, and adenine (SC Ura⁻ Leu⁻ Ade⁻ His⁻). The positive clones were streaked on SC agar plates containing 5-fluoroorotic acid but lacking leucine to isolate clones that lost the bait plasmid. The prey-only-containing clones were then grown on SC Leu⁻ Ade⁻ plates, and those clones that could grow were eliminated for further analysis (because these preys can activate adenine gene expression by itself). To amplify prey plasmids, plasmids from prey-only-containing clones were purified and transformed into Escherichia coli DH5α, and the E. coli cells were plated out onto LB plates (with 50 μg/ml ampicillin). The prey plasmids were then purified from DH5α and retransformed into strain PJ69-4A containing the PA bait plasmid to confirm the interaction between prey and bait (grown on SC Ura⁻ Leu⁻ Ade⁻ His⁻ agar plates). Those prey plasmids that gave positive results were then subjected to DNA sequence analysis.

GST-pulldown assay.

pGEX6P-1-PA, which encodes the GST-PA fusion protein, and pGEX6P-1, which encodes the GST protein, were individually transformed into the Rosetta-gami strain (EMD Chemicals, Gibbstown, NJ), and the expression of GST or GST-PA protein was induced by isopropyl-1-thio-β-d-galactopyranoside. Bacterial cells were then lysed in TENT buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol [DTT], 0.5% Nonidet P-40 [NP-40] and complete protease inhibitor cocktail [Roche, Indianapolis, IN]) with mild sonication. Bacterial lysates containing 1 mg of GST or GST-PA proteins were incubated with glutathione Sepharose 4 Fast Flow (GE Healthcare Life Sciences) for 18 h at 4°C. The beads containing immobilized GST or GST-PA protein were washed, resuspended in radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 50 mM Tris [pH 7.4], 0.5 mM NaF, 0.5 mM Na₃VO₄, 1% NP-40, and complete protease inhibitor cocktail [Roche]), and then used to pull down the HA-HAX1 (HAX1 fused with an HA tag) or HA-FTH1 (FTH1 fused with an HA tag) protein as described below. 293FT cells were transiently transfected with the pcDNA-HA-HAX1, pCMV-HA-FTH1, or vector plasmid. Forty-eight hours later, cells were lysed in RIPA buffer and cell lysates containing equal amounts of extract protein (800 μg) were incubated with glutathione Sepharose 4 Fast Flow containing either GST or GST-PA at 4°C for 4 h. After an extensive wash with buffer I (150 mM NaCl, 50 mM Tris [pH 7.4], 0.5 mM NaF, 0.5 mM Na₃VO₄, 0.5% NP-40, and complete protease inhibitor cocktail [Roche]), the bound proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by Western blotting with anti-HA and anti-GST antibody.

Coimmunoprecipitation, Western blot analysis, and sub-G₁ apoptosis assay.

293FT cells were transfected with the indicated plasmids using the TurboFect transfection reagent (Fermentas, Glen Burnie, MD). Forty-eight hours later, cells were washed with cold phosphate-buffered saline (PBS) and lysed in RIPA buffer containing complete protease inhibitor cocktail (Roche). Cell lysates were precleared with 30 μl of protein G agarose (EMD Millipore, Bedford, MA). The lysates were incubated with the indicated antibody at 4°C for 18 h, and the protein G-agarose beads were then added and rotated at 4°C for 2 h. The beads were washed three times with lysis buffer, and the bound proteins were separated by SDS-PAGE followed by Western blotting with the indicated antibody. Western blot analyses were performed as described previously (56). Sub-G₁ apoptosis assays were performed as described previously (57).

Construction of lentivirus expressing HAX1 shRNA or Luc shRNA.

A lentivirus-based shRNA expression system (The RNAi Consortium [TRC]) was used to knock down gene expression of HAX1. A luciferase (Luc) shRNA construct was used as a nontargeting shRNA control. We obtained all pLKO.1-shRNA constructs from the National RNAi Core Facility, Academia Sinica, Taipei, Taiwan. The sequence targeting HAX1 is 5′-ACAGACACTTCGGGACTCAAT-3′ (TRCN0000061777), and the sequence targeting luciferase is 5′-CAAATCACAGAATCGTCGTAT-3′ (TRCN0000072246). To generate lentivirus expressing HAX1 shRNA or Luc shRNA, pLKO.1-HAX1 shRNA or pLKO.1-Luc shRNA was cotransfected into 293FT cells with pCMV-ΔR8.91 and pMD.G. Lentivirus production was carried out as described on the website http://rnai.genmed.sinica.edu.tw/protocols. It should be noted that the HAX1 shRNA used here can target mRNAs of both isoform A (∼35 kDa) and isoform B (∼32 kDa) of HAX1.

Establishment of HAX1-knockdown, nonspecific-knockdown, and HAX1-reexpressing H1299 cell clones.

H1299 cells were infected with lentivirus (which contains a puromycin resistance gene) expressing HAX1 shRNA or Luc shRNA. Twenty-four hours later, the cells were washed with PBS and incubated in medium containing 2 μg/ml puromycin (EMD Chemicals). After a 2-week selection using medium containing puromycin, puromycin-resistant colonies were cloned and pooled together to generate HAX1-knockdown H1299-shHAX1 cells or nonspecific-knockdown H1299-shLuc cells. To establish HAX1-reexpressed (H1299-shHAX1+HAX1) cells, H1299-shHAX1 cells were transfected with pcDNA-HA-HAX1-R, which contains silent mutations in the target sequence of HAX1 shRNA. Forty-eight hours later, cells were trypsinized and split at a 1:10 ratio into medium containing 800 μg/ml G418 (EMD Chemicals). After 4 to 6 weeks of selection using medium containing 800 μg/ml G418, G418-resistant colonies were cloned and tested for the expression of HAX1 by Western blot analysis. The clones that expressed a higher level of HAX1 than H1299-shHAX1 cells were pooled together to generate H1299-shHAX1+HAX1 cells.

Immunofluorescence.

Immunofluorescence assays were performed as described previously (58). H1299 and H1299-derived cells grown on 15-mm glass coverslips in 12-well plates were transfected with p3×Myc-PA by using TurboFect transfection reagent. Forty-eight hours later, cells were fixed, blocked with 5% bovine serum albumin, and incubated with anti-Myc and anti-HAX1 antibodies, followed by incubation with goat anti-rabbit IgG-FITC and rabbit anti-mouse IgG-Alexa Fluor 594 antibodies. To label nuclear DNA, cells were stained with 0.1 μg/ml Hoechst 33342 solution (Sigma-Aldrich, St. Louis, MO) for 30 min. Coverslips containing labeled cells were mounted on glass slides and visualized with a Leica TCS SP5 confocal microscope (Leica Microsystems, Mannheim, Germany).

Fractionation of cytoplasmic and nuclear proteins.

Cells were harvested by scraping them into ice-cold PBS and collected by centrifugation at 500 × g for 5 min. Cells were lysed with NP-40 lysis buffer (50 mM Tris-HCl [pH 8.0], 10 mM NaCl, 5 mM MgCl₂, and 0.5% NP-40) for 10 min on ice. Following centrifugation at 15,000 × g for 15 min, the supernatant was collected as the cytoplasmic fraction. The nuclear pellet was washed three times with the same lysis buffer and resuspended by vortexing in high-salt buffer (20 mM HEPES [pH 7.9], 0.5 M NaCl, 1 mM EDTA, and 1 mM DTT). Nuclear debris was removed by centrifugation at 15,000 × g for 10 min. The supernatant was collected as the nuclear fraction. The proteins in the cytoplasmic or nuclear fraction were analyzed by Western blotting using anti-PA, anti-p84, and anti-GAPDH antibodies.

RNA extraction and quantitative RT-PCR.

RNA extraction and quantitative reverse transcription-PCR (RT-PCR) were performed as described previously (59). cDNA was synthesized using 1 μg of total RNA as the template and oligo(dT)₁₈ (for NP mRNA), 5′-CTCATCCTTTATGACAAAGAAG-3′ (for NP vRNA), or 5′-AGTAGAAACAAGGGTATTTTTCTTT-3′ (for NP cRNA) as the primer according to the manufacturer's protocol for reverse transcription (Invitrogen).

In real-time PCR, 0.05 μg cDNA was used with SYBR green PCR master mix (Applied Biosystems, Foster City, CA) and sequence-specific primers (primers GAPDH-Forward [5′-GAAGATGGTGATGGGATTTC-3′] and GAPDH-Reverse [5′-GAAGGTGAAGGTCGGAGTC-3′]; primers NP gene-Forward [5′-CTCATCCTTTATGACAAAGAAG-3′] and NP gene-Reverse [5′-AGATCATCATGTGAGTCAGAC-3′]). Amplification of the target sequences was detected with an ABI 7500HT sequence detection system (Applied Biosystems) and analyzed with SDS (version 2.0) software (Applied Biosystems). The expression of GAPDH was used to normalize the abundance of the test RNAs.

Viral propagation and titration.

H1299 or H1299-derived cells (∼8 × 10⁵) were seeded in 6-well dishes. Twenty hours later, the cells were washed twice with PBS, replaced with 1× infection medium (serum-free DMEM, 0.3% bovine serum albumin, 1 mM sodium pyruvate, 1× nonessential amino acid solution, 10 mM HEPES [pH 7.0], and 2 μg/ml tosyl phenylalanyl chloromethyl ketone-trypsin), and then infected with influenza A/WSN/33 virus at a multiplicity of infection (MOI) of 0.01. At 12, 24, 36, or 48 h postinfection, supernatants were collected and the amount of virus in the supernatants was quantitated by plaque assay as described previously (60).

Minigenome replication and transcription assay.

Assays with the plasmid-based reverse genetics system were carried out as previously described to mimic influenza A virus transcription and replication (61). Cells in 24-well plates (5 × 10⁴ cells/well) were transfected with pPOL I-Luc-RT, pcDNA-PB1, pcDNA-PB2, pcDNA-PA, pcDNA-NP, Renilla luciferase expression plasmid pRL-CMV (which serves as a transfection efficiency control), and other indicated plasmids by TurboFect transfection reagent (Fermentas) according to the manufacturer's protocol. Forty-eight hours later, the luciferase activity was measured using a dual-luciferase reporter assay system (Promega, Madison, WI) according to the manufacturer's protocol. The luciferase activity was obtained by normalizing firefly luciferase light units against transfection efficiency (which was obtained by comparing the Renilla luciferase light units of different transfections).

Statistical analyses.

Data are presented as the mean ± standard deviation. The significance of the difference between groups was evaluated with the Student t test, and a P value of <0.05 was considered significant.

RESULTS

Identification of cellular HAX1 as a novel PA-interacting protein.

To identify cellular proteins that interact with influenza A virus PA protein, we performed a yeast two-hybrid screen using full-length PA as bait. Ten positive clones were obtained. Sequence analysis revealed that these clones represent 6 cellular proteins (Table 1). Among these 6 proteins, hCLE (represented by 3 clones) has been reported to interact with PA (42), HNRNPM (represented by 1 clone) has been shown to associate with influenza virus vRNP (62), and SIVA1 (represented by 1 clone) has been reported to interact with PB2 (63). The DNA sequences of 3 out of the 10 positive clones matched the DNA sequence of HAX1. Because HAX1 has been shown to interact with several viral proteins and play important roles in the life cycle of several viruses (46, 48–50), we focused our study on HAX1. To verify the interaction between PA and HAX1, we transformed the PJ69-4A yeast strain (55) with different combinations of the following plasmids: pGBDU-C1 (bait vector), pGAD-GH (prey vector), pGBDU-C1-PA (a plasmid encoding PA fused with the DNA binding domain of the yeast GAL4 transcription factor), and pGAD-GH-HAX1 (a plasmid encoding HAX1 fused with the activation domain of the yeast GAL4 transcription factor) (Fig. 1A). The transformants were then streaked on the SC Ura⁻ Leu⁻ plate or the SC Ura⁻ Leu⁻ Ade⁻ His⁻ plate. The growth of these transformants on the respective plates is shown in Fig. 1B. While all transformants could grow on the SC Ura⁻ Leu⁻ plate (indicating that these transformants contain the respective plasmids) (Fig. 1B, left), only the transformants transformed with pGBDU-C1-PA and pGAD-GH-HAX-1 could grow on the SC Ura⁻ Leu⁻ Ade⁻ His⁻ plate. Since the reporter genes ADE2 and HIS3 are turned on in yeast only upon bait-prey interaction, these data indicate that PA can interact with HAX1 by the two-hybrid criteria.

Table 1.

PA-interacting cellular proteins identified by yeast two-hybrid screen

Symbol	GenBank accession no.	Gene identifier	Gene description
hCLE/CGI99	NM_016039	51637	mRNA transcription modulator
HAX1	NM_006118	10456	Bcl-2-family-related protein, antiapoptotic protein
HNRNPM	NM_031203	4670	Member of the hnRNP subfamily, involved in RNA binding and pre-mRNA processing
SIVA1	NM_006427	10572	CD27-binding protein, apoptosis-inducing factor
PRDX1	NM_002574	5052	Member of the peroxiredoxin family, antioxidant enzyme
MMAB	NM_052845	362265	Vitamin B12-containing coenzyme for methylmalonyl coenzyme A mutase

Fig 1.

Yeast two-hybrid assay showing the interaction between PA and HAX1. (A) Template (for panel B) showing the transformants streaked on each section of the plate. (B) Growth of different transformants on SC Ura⁻ Leu⁻ and SC Ura⁻ Leu⁻ Ade⁻ His⁻ medium plates.

PA specifically interacts with HAX1 in vitro.

To confirm the interaction between PA and HAX1, GST-pulldown assays were performed. The GST-PA or GST protein was expressed in E. coli and immobilized on glutathione-conjugated Sepharose beads. Beads carrying GST-PA or GST were incubated with lysates from 293FT cells transfected with the pCMV-HA vector, pCMV-HA-FTH1, or pcDNA-HA-HAX1. After extensive rinsing, the protein complex captured on the beads was solubilized, subjected to electrophoresis in a denaturing gel, and immunoblotted with anti-HA or anti-GST antibody. As shown in Fig. 2A, top, the GST-PA protein could pull down HA-HAX1 but not HA-FTH1. In contrast, GST alone did not pull down HA-HAX1 or HA-FTH1. Figure 2A (bottom) shows the amount of GST-PA (or GST) bound to the beads used in different pulldown experiments. The beads used to pull down HA, HA-FTH1, and HA-HAX1 contained similar amounts of GST-PA (or GST) protein. Furthermore, more GST protein than GST-PA protein was used in pulldown experiments. Together these data indicate that PA can specifically interact with HAX1 in vitro.

Fig 2.

PA interacts with HAX1 in vitro and in vivo. (A) GST-pulldown analysis of interaction between PA and HAX1. Cell lysates from 293FT cells transfected with vector, pCMV-HA-FTH1, or pcDNA-HA-HAX1 were incubated with GST or GST-PA immobilized on glutathione-conjugated Sepharose beads, and the bound proteins were determined by Western blot analysis (WB) using anti-HA (top) or anti-GST (bottom) antibody. (B and C) Coimmunoprecipitation assays showing the interaction between Myc-PA and HA-HAX1. 293FT cells were cotransfected with the indicated plasmids, and the cell extracts were subjected to coimmunoprecipitation assays using either anti-HA (B) or anti-Myc (C) antibody to pull down the immune complex. IP, immunoprecipitation. (D) Coimmunoprecipitation assays showing that Myc-PA can specifically interact with HA-HAX1. 293FT cells were cotransfected with the indicated plasmids, and the cell extracts were immunoprecipitated with anti-Myc antibody, followed by Western blot analysis using anti-HA and anti-Myc antibodies. (E) Coimmunoprecipitation assays showing that PA can interact with endogenous HAX1. 293FT cells were transfected with pcDNA-PA, and the cell extracts were immunoprecipitated with anti-HAX1 antibody, followed by Western blot analysis using anti-PA and anti-HAX1 antibodies. V, vector. All the experiments whose results are shown in Fig. 2 were performed at least three times.

PA specifically interacts with HAX1 in mammalian cells.

To determine whether PA interacts with HAX1 in the human cells, we performed coimmunoprecipitation experiments. 293FT cells were cotransfected with p3×Myc-PA, pcDNA-HA-HAX1, or vector plasmid in different combinations, as shown in Fig. 2B. The cell extracts were immunoprecipitated with anti-HA antibody, and the immunoprecipitates were separated on denaturing gels and then immunoblotted with anti-Myc or anti-HA antibody. As shown in Fig. 2B, Myc-PA was detected in the anti-HA immune complex when Myc-PA and HA-HAX1 were coexpressed in the cells. In contrast, when cells were cotransfected with p3×Myc-PA and HA vector plasmid, Myc-PA was not detected in the anti-HA immune complex. Reciprocal coimmunoprecipitation experiments using anti-Myc antibody to bring down the immune complexes were performed to confirm the interaction between Myc-PA and HA-HAX1. As shown in Fig. 2C, HA-HAX1 was detected in the anti-Myc immune complex only when p3×Myc-PA and pcDNA-HA-HAX1 were cotransfected into the cells. Together, these data indicate that PA can interact with HAX1 in mammalian cells. To study whether the interaction between PA and HAX1 was specific, p3×Myc-PA, pcDNA-HA-HAX1, or pCMV-HA-FTH1 was cotransfected into 293FT cells in different combinations (Fig. 2D), and the cell lysates were immunoprecipitated with anti-Myc antibody. While HA-HAX1 could be coimmunoprecipitated with Myc-PA, HA-FTH1 could not (Fig. 2D), indicating that PA can interact with HAX1 specifically. To test whether untagged PA could interact with endogenous HAX1, 293FT cells were transfected with either pcDNA-PA or vector plasmid and the cell lysates were immunoprecipitated with anti-HAX1 antibody. As shown in Fig. 2E, untagged PA was coimmunoprecipitated with HAX1, further confirming the interaction between PA and HAX1.

HAX1 interacts with the NLS domain of PA.

To define the specific regions of PA required for interacting with HAX1. We constructed a series of Myc-tagged PA deletion mutants (Δ1, Δ2, Δ3, Δ4, Δ5, and Δ6) based on well-known functional domains of PA (40, 64–66) (Fig. 3A). These PA deletion mutants were individually cotransfected with pcDNA-HA-HAX1 into 293FT cells. Cell extracts were immunoprecipitated with anti-HA antibody, and the immunoprecipitates were Western blotted with anti-Myc antibody. As shown in Fig. 3B, whereas the Δ1 and Δ2 mutants of PA as well as the wild-type (WT) PA could be coimmunoprecipitated with HA-HAX1, other deletion mutants of PA (Δ3, Δ4, Δ5, and Δ6) failed to associate with HA-HAX1. It is noted that while both Δ1 and Δ2 mutants of PA contained the NLS domain, other deletion mutants of PA (Δ3, Δ4, Δ5, and Δ6) did not. That the Δ3, Δ4, Δ5, and Δ6 mutants did not contain NLS was demonstrated by the immunofluorescence assay, which showed the exclusively cytoplasmic localization of these mutant proteins (data not shown). Therefore, the NLS domain of PA is required for interaction with HAX1. Together, these data implicate that HAX1 interacts with PA through the NLS domain of PA.

Fig 3.

PA interacts with HAX1 through its NLS domain. (A) Schematic representation of full-length PA and PA deletion mutants. NLS, nuclear localization sequence; NBM, nucleotide binding motif (64). (B) Coimmunoprecipitation analysis of interaction between PA mutant proteins and HAX1. 293FT cells were cotransfected with pcDNA-HA-HAX1 and different Myc-tagged PA deletion mutants. Forty-eight hours later, cell lysates were immunoprecipitated with anti-HA antibody followed by Western blot analysis using anti-Myc and anti-HA antibodies. The experiment was performed three times.

HAX1 inhibits nuclear accumulation of PA.

Knowing that HAX1 interacts with the NLS domain of PA, we asked whether HAX1 could mask the NLS of PA and retain PA in the cytoplasm. To investigate this, we assessed the effect of HAX1 knockdown on the intracellular localization of PA. We constructed HAX1-knockdown H1299 cells (H1299-shHAX1) and nonspecific-knockdown H1299 cells (H1299-shLuc) by infecting human H1299 cells with lentivirus expressing HAX1 shRNA or Luc shRNA, respectively. We also constructed HAX1-reexpressed cells (H1299-shHAX1+HAX1) by transfecting HAX1, which contains silent mutations in the target sequence of HAX1 shRNA, into H1299-shHAX1 cells. The expression of HAX1 protein in these cell clones is shown in Fig. 4A. Compared to parental H1299 cells, whereas the expression of HAX1 in H1299-shHAX1 cells was significantly reduced (P < 0.01), that in H1299-shLuc and H1299-shHAX1+HAX1 cells was not altered. The growth rate of H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells was determined by 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. These cells grew at similar rate, indicating that knockdown of HAX1 does not affect the growth of the cells (data not shown). These H1299-derived cell clones and parental H1299 cells were transfected with p3×Myc-PA. Forty-eight hours later, cells were fixed and subjected to immunofluorescence assays. As shown in Fig. 4B, Myc-PA primarily localized in the cytoplasm of H1299-shLuc and parental H1299 cells. In contrast, in H1299-shHAX1 cells, Myc-PA localized both in the nucleus and in the cytoplasm. When HAX-1 was reexpressed in the H1299-shHAX1 cells (i.e., H1299-shHAX1+HAX1 cells), the cellular localization of Myc-PA was similar to that of the parental H1299 cells. The immunostained cells were counted and the intracellular localization pattern of Myc-PA was determined. Figure 4C shows a statistical summary of the intracellular localization patterns of Myc-PA in H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells. Knockdown of HAX1 decreased the proportion of cells with a predominantly cytoplasmic localization of Myc-PA and increased the proportion of cells in which Myc-PA was distributed evenly in the cytoplasm and nucleus. Together, these data suggest that HAX1 can inhibit the nuclear accumulation of PA.

Fig 4.

Nuclear accumulation of PA is inhibited by HAX1. (A) (Top) Western blot analysis showing the expression of HAX1 protein in H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells. Tubulin served as an internal control for the amounts of protein loaded on the gel. (Bottom) Relative levels of HAX1 protein in H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells. Values represent means ± standard deviations. **, P < 0.01; #, P = 0.06; ##, P = 0.1 (n = 3). (B) Immunofluorescence assays showing that HAX1 can inhibit the nuclear accumulation of PA. H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells were transfected with p3×Myc-PA. Forty-eight hours later, cells were fixed and stained with anti-Myc antibody (green), anti-HAX1 antibody (red), and Hoechst 33342 (blue). (C) Statistical summary of the intracellular localization patterns of Myc-PA in H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells. The immunostained cells were scored for PA location. The numbers of cells counted were 106 (for H1299), 111 (for H1299-shLuc), 112 (for H1299-shHAX1), and 127 (for H1299-shHAX1+HAX1). (D) Distribution of nuclear PA (N-PA) and cytoplasmic PA (C-PA) in the cells during influenza A virus infection. H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells were infected with influenza A/WSN/33 virus at an MOI of 5. Cells were harvested at 2, 4, 6, and 8 h postinfection (hpi), and the proteins in the nuclear fraction and the cytoplasmic fraction were analyzed by Western blotting using anti-PA, anti-p84 (p84 localizes exclusively in the nucleus), and anti-GAPDH (GAPDH localizes exclusively in the cytoplasm) antibodies. The relative N-PA/total PA ratio was calculated as follows. The level of N-PA was normalized against that of nuclear p84 (which served as a loading control). Similarly, the level of cytoplasmic PA was normalized against that of cytoplasmic GAPDH (which served as a loading control). The relative N-PA/total PA ratio was obtained by dividing the normalized level of N-PA with that of total PA. The total PA is the sum of N-PA and cytoplasmic PA. Mock, whole-cell protein extract from mock-infected H1299 cells was loaded onto the gel. The experiment was performed three times.

To further confirm this observation in viral infection, we infected H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells with influenza A/WSN/33 virus. Cells were harvested at 2, 4, 6, and 8 h after virus infection. The nuclei and cytoplasm of infected cells were separated as described in Materials and Methods by using a low concentration of nonionic detergent. Each fraction was subjected to electrophoresis in denaturing polyacrylamide gels and Western blotted with antibody against PA, p84 (a nuclear protein), or GAPDH (a cytoplasmic protein). As shown in Fig. 4D, at 2, 4, 6, and 8 h postinfection, the amount of nuclear PA (N-PA) was much higher in H1299-shHAX1 cells than in H1299 cells, suggesting that HAX1 may inhibit the nuclear accumulation of PA during virus infection. However, we also noticed that the total level of PA expression was higher in H1299-shHAX1 cells than in H1299 cells at different time points after virus infection. It is possible that the higher nuclear accumulation of PA in H1299-shHAX1 cells is due to higher PA expression rather than easier nuclear entry of PA. To demonstrate that PA can indeed enter the nucleus more efficiently in H1299-shHAX1 cells than in H1299 cells, the relative ratio of N-PA to total PA in the infected cells at different time points postinfection was calculated and is shown at the bottom of Fig. 4D. The relative N-PA/total PA ratio was higher in H1299-shHAX1 cells than in H1299 cells at all time points following virus infection, indicating that PA can enter the nucleus more efficiently in HAX1-knockdown cells. Moreover, the relative N-PA/total PA ratio was lower in H1299-shHAX1+HAX1 cells than in H1299-shHAX1 cells at all time points following virus infection, indicating that reexpression of HAX1 in H1299-shHAX1 cells can rescue the phenotype caused by HAX1 knockdown. Together, these data clearly demonstrate that HAX1 can inhibit the nuclear accumulation of PA during influenza A virus infection. This inhibition of nuclear accumulation of PA by HAX1 could cause lower activity of the viral RNA polymerase complex, leading to the lower expression of PA in parental H1299 cells. In contrast, in HAX1-knockdown cells, PA can accumulate in the nucleus efficiently, leading to higher activity of the polymerase complex and higher PA expression.

The binding of HAX1 to PA is reduced in the presence of PB1.

Previous studies have shown that PA alone is transported inefficiently into the nucleus (4, 9, 10, 16) and that PB1 can facilitate nuclear transport of PA (9, 16). Our results presented above indicate that HAX1 can interact with PA and inhibit the nuclear accumulation of PA. It is possible that PB1 can interfere with the interaction between HAX1 and PA and alleviate the inhibitory effect of HAX1, leading to efficient nuclear transport of PA. To test whether the binding of HAX1 to PA was disrupted in the presence of PB1, we cotransfected p3×Myc-PA and pcDNA-HA-HAX1 together with either the PB1-expressing plasmid or vector plasmid into 293FT cells. The cell lysates were immunoprecipitated with anti-Myc antibody, and the immunoprecipitates were then immunoblotted with anti-HA and anti-Myc antibodies. As shown in Fig. 5, the binding of HAX1 to PA was significantly reduced in the presence of PB1, indicating that PB1 can interfere with the interaction between PA and HAX1.

Fig 5.

Coimmunoprecipitation assays showing that the interaction between PA and HAX1 is reduced in the presence of PB1. (A) 293FT cells were transfected with p3×Myc-PA and pcDNA-HA-HAX1 together with either pcDNA-PB1-GFP (16) or vector plasmid pEGFP-C1. Forty-eight hours later, the cell extracts were immunoprecipitated with anti-Myc antibody, followed by Western blot analysis using anti-HA, anti-Myc, and anti-GFP antibodies. The experiment was performed three times, and the level of HAX1 and PA protein was determined by densitometry. (B) Relative HAX1/PA protein ratio. **, P < 0.01 (n = 3).

HAX1 inhibits the propagation of influenza A virus.

Knowing that HAX1 can inhibit the nuclear accumulation of PA, a protein involved in influenza A virus transcription and replication, we then asked whether HAX1 could inhibit the propagation of influenza A virus. To investigate this, we assessed the effect of HAX1 knockdown on the production of infectious influenza A viral particles. We infected parental H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells with influenza A/WSN/33 virus at an MOI of 0.01. At 12, 24, 36, and 48 h postinfection, the virus titer in the supernatant was determined by plaque assay. As shown in Fig. 6, the production of influenza A/WSN/33 viral particles in H1299-shHAX1 cells was significantly increased compared to that in H1299-shLuc and parental H1299 cells. Moreover, the production of influenza A/WSN/33 viral particles was significantly decreased in the H1299-shHAX1+HAX1 cells compared to that in the H1299-shHAX1 cells. Therefore, the increase in virus production observed in H1299-shHAX1 cells was indeed caused by HAX1 knockdown and not due to the off-target nonspecific effect of HAX1 shRNA. Taken together, the data presented above clearly demonstrate that HAX1 can negatively modulate the propagation of influenza A virus.

Fig 6.

HAX1 inhibits propagation of influenza A virus. H1299, H1299-shLuc, H1299-shHAX1, and H1299-shHAX1+HAX1 cells were infected with influenza A/WSN/33 virus at an MOI of 0.01. At 12, 24, 36, and 48 h postinfection, the virus titer in the supernatant was determined by plaque assay. Values represent means ± standard deviations. **, P < 0.01; *, P < 0.05 (n = 3).

Knockdown of HAX1 leads to increases of mRNA, vRNA, and cRNA synthesis during influenza A virus infection.

To study whether HAX1 had any impact on the transcription and replication of the influenza A virus genome, we infected H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells with influenza A/WSN/33 virus at an MOI of 2. At 4, 6, and 8 h postinfection, total RNAs were isolated and subjected to quantitative RT-PCR to measure the mRNA, vRNA, and cRNA levels of the viral NP segment. As shown in Fig. 7A to C, the mRNA, vRNA, and cRNA levels of the NP segment increased in H1299-shHAX1 cells compared to those in parental H1299 cells. Moreover, reexpression of HAX1 in H1299-shHAX1 cells (i.e., H1299-shHAX1+HAX1 cells) led to a significant reduction of the mRNA, vRNA, and cRNA levels of the NP segment. It was noted that the NP vRNA level was similar in H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells at 0 h postinfection (data not shown), indicating that these cells were infected with similar amounts of influenza A/WSN/33 viral particles at the beginning of infection. Together, these data demonstrate that knockdown of HAX1 leads to upregulation of transcription and replication of the influenza A virus genome.

Fig 7.

Knockdown of HAX1 leads to an increase in virus RNA synthesis. H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells were infected with influenza A/WSN/33 virus at an MOI of 2. At 4, 6, and 8 h postinfection, total RNAs were extracted and the levels of mRNA (A), vRNA (B), and cRNA (C) of the NP segment were measured by quantitative RT-PCR. The RNA level in virus-infected H1299 cells at 4 h postinfection was set equal to 1. Values represent means ± standard deviations. *, P < 0.05 (n = 3).

HAX1 downregulates the activity of the influenza A virus polymerase complex.

During influenza A virus infection, the NP-associated negative-sense vRNA is used as the template to synthesize mRNA and cRNA in the nucleus by the heterotrimeric polymerase complex consisting of PB1, PB2, and PA. To examine whether HAX1 could regulate the activity of the influenza A virus polymerase complex, we performed a minigenome transcription and replication assay using plasmid pPOL I-Luc-RT, which contains the luciferase open reading frame in negative polarity flanked by the noncoding regions of the NS gene of influenza A/WSN/33 virus, as a reporter (illustrated in Fig. 8A). We cotransfected pcDNA-PB1, pcDNA-PB2, pcDNA-PA, pcDNA-NP, pcDNA-HA-HAX1, and vector plasmids in different combinations together with the reporter plasmid pPOL I-Luc-RT into H1299 cells. Forty-eight hours later, the cell lysates were subjected to luciferase assay. The result of the luciferase assay is shown in Fig. 8B. When cells were transfected with pPOL I-Luc-RT together with PB1-, PB2-, and NP-expressing plasmid, no luciferase activity was detected (column 1), indicating that PA plays an important role in the transcription and replication of the virus-like genome. When cells were transfected with pPOL I-Luc-RT together with PB1-, PB2-, PA-, and NP-expressing plasmid, the luciferase activity increased dramatically (compare columns 1 and 2), indicating that PB1, PB2, PA, and NP together can stimulate the transcription and replication of the virus-like minigenome. When HAX1 was cotransfected with PB1, PB2, PA, NP, and the reporter into cells, the luciferase activity decreased significantly (compare columns 2 and 3), indicating that overexpression of HAX1 can inhibit the transcription and replication of the influenza A virus-like minigenome.

Fig 8.

Knockdown of HAX1 leads to an increase in RNA polymerase activity in the minigenome transcription and replication assay. (A) Schematic diagram showing the elements of the minigenome transcription and replication assay. NCR, noncoding region. (B) H1299 cells were transfected with pPOL I-Luc-RT reporter plasmid, Renilla luciferase-expressing plasmid pRL-CMV, and the indicated plasmids. Forty-eight hours later, the cell lysates were subjected to luciferase assay as described in Materials and Methods. (Bottom) Expression level of HAX1 protein. Values represent means ± standard deviations. **, P < 0.01 (n = 5). (C) H1299 cells were transfected with the pPOL I-Luc-RT reporter plasmid together with pcDNA-PB1, pcDNA-PB2, pcDNA-PA, pcDNA-NP, and pRL-CMV (column 1). H1299-shHAX1 cells were transfected with pPOL I-Luc-RT, pcDNA-PB1, pcDNA-PB2, pcDNA-PA, pcDNA-NP, and pRL-CMV plus 0 μg (column 2), 0.25 μg (column 3), or 1 μg (column 4) pcDNA-HA-HAX1-R (encoding the HA-HAX1 protein). Forty-eight hours later, the cell lysates were subjected to luciferase assay. Values represent means ± standard deviations. **, P < 0.01 (n = 5).

To further confirm that HAX1 can inhibit the transcription and replication of the influenza A virus-like minigenome, we transfected pPOL I-Luc-RT together with PB1-, PB2-, PA-, and NP-expressing plasmid into H1299 and H1299-shHAX1 cells. Forty-eight hours later, the cell lysates were subjected to luciferase assay. As shown in Fig. 8C, knockdown of HAX1 led to an increase in luciferase activity (compare columns 1 and 2). When increasing amounts of HAX1-expressing plasmid (pcDNA-HA-HAX1-R) were cotransfected with pPOL I-Luc-RT and PB1-, PB2-, PA-, and NP-expressing plasmid into H1299-shHAX1 cells, the luciferase activity decreased in a dose-dependent manner (compare columns 2, 3, and 4 in Fig. 8C). Together, these data confirm that HAX1 can inhibit the transcription and replication of the influenza A virus-like minigenome. Taken the data presented above together, we conclude that HAX1 can downregulate the activity of the influenza A virus polymerase complex by inhibiting the nuclear accumulation of PA.

PA does not affect the antiapoptotic function of HAX1.

Previous studies have shown that overexpression or activation of proapoptotic protein, such as BAX and caspase-3, leads to apoptosis, resulting in efficient influenza virus propagation (67, 68). In contrast, overexpression of antiapoptotic protein Bcl-2 leads to inhibition of apoptosis, which in turn results in downregulation of influenza A virus propagation (69, 70). HAX1 is an antiapoptotic protein with homology to Bcl-2 (45, 46, 71). It is possible that HAX1 inhibits influenza A virus propagation through inhibiting apoptosis and that PA could release this inhibition by interacting with HAX1 and inhibiting the antiapoptotic activity of the latter. We thus tested whether PA could affect the antiapoptotic function of HAX1. By using caspase-3 cleavage assays (Fig. 9A), we found that HAX1 could inhibit apoptosis induced by proapoptotic protein BAX (compare lanes 3 and 4), indicating that HAX1 indeed has an antiapoptotic function. When increasing amounts of PA were cotransfected into cells, the antiapoptotic activity of HAX1 remained intact (compare lanes 4, 5, and 6). Therefore, PA cannot affect the antiapoptotic function of HAX1. This conclusion was also confirmed in the sub-G₁ apoptosis assays (Fig. 9B).

Fig 9.

PA does not affect the antiapoptotic function of HAX1. (A) H1299 cells were transfected with pcDNA-BAX (0.5 μg), pcDNA-HA-HAX1 (2 μg), pcDNA-PA (2 or 4 μg), or vector plasmid in different combinations, as indicated. Forty-eight hours later, the cell extracts were subjected to Western blot analysis using anti-caspase-3 and anti-GAPDH antibodies. (B) H1299 cells were transfected with pcDNA-BAX (0.5 μg), pcDNA-HA-HAX1 (2 μg), pcDNA-PA (4 μg), or vector plasmid in different combinations, as indicated. Twenty-four hours later, both detached cells and adherent cells were collected and subjected to sub-G₁ DNA content analysis. Mock, mock-transfected cells. Values represent means ± standard deviations. **, P < 0.01 (n = 3).

HAX1 cannot inhibit apoptosis induced by influenza A virus infection.

Our data presented above indicate that HAX1 has an antiapoptotic activity that can inhibit apoptosis induced by BAX (Fig. 9). To test whether HAX1 could also inhibit apoptosis induced by influenza A virus infection, we infected H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells with influenza A/WSN/33 virus. The level of virus-induced apoptosis was then measured by caspase-3 cleavage assays. As shown in Fig. 10A, influenza A virus infection induced similar levels of apoptosis in H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells at 12 h and 24 h postinfection, indicating that HAX1 cannot inhibit apoptosis induced by influenza A virus infection. This conclusion was also confirmed in the sub-G₁ apoptosis assays (Fig. 10B).

Fig 10.

HAX1 cannot inhibit apoptosis induced by influenza A virus infection. (A) H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells were infected with influenza A/WSN/33 virus at an MOI of 3. Twelve or 24 h later, the cell extracts were subjected to Western blot analysis using anti-caspase-3 and anti-GAPDH antibodies. (B) H1299, H1299-shHAX1, and H1299-shHAX1+HAX1 cells were infected with influenza A/WSN/33 virus at an MOI of 3. Twelve or 24 h later, cells were collected and subjected to sub-G₁ DNA content analysis. Mock, mock-infected cells (n = 3).

DISCUSSION

The influenza A virus RNA polymerase, consisting of PB1, PB2, and PA, assembles in the nucleus, where influenza A virus replication occurs. To achieve efficient viral transcription and replication, the polymerase subunits must be transported into the nucleus. This nuclear transport of polymerase subunits depends on their interaction with host proteins and may be positively or negatively regulated by cellular factors (17, 72–75). Although cellular factors that facilitate nuclear transport of polymerase subunits have been well characterized, cellular factors that inhibit this process remain poorly defined. In this study, by using a yeast two-hybrid screening system, we identified cellular protein HAX1 to be a novel PA-interacting protein (Fig. 1) and demonstrated that HAX1 could inhibit nuclear accumulation of PA (Fig. 4) by interacting with the NLS domain of PA (Fig. 3). In accordance with this, we also found that HAX1 could inhibit the transcription and replication of the influenza A virus genome (Fig. 7 and 8) and inhibit the propagation of influenza A virus in human epithelial cells (Fig. 6). Together, these results suggest that HAX1 is an intrinsic host factor that restricts influenza A virus infection.

We demonstrated that PA could specifically interact with HAX1 by GST-pulldown and coimmunoprecipitation assays (Fig. 2). HAX1 is an antiapoptotic protein which is primarily located in the mitochondria, endoplasmic reticulum, and, to a lesser extent, the cytoplasm (76). It is possible that PA, through interacting with HAX1, affects the antiapoptotic function of HAX1. By using caspase-3 cleavage assays (Fig. 9A) and sub-G₁ apoptosis assays (Fig. 9B), we demonstrated that PA could not affect the antiapoptotic activity of HAX1.

To test whether HAX1 could affect the functions of PA, we first mapped the domain of PA that is targeted by HAX1. We found that HAX1 interacted with the NLS domain of PA (Fig. 3). This result led us to test whether HAX1 could mask the NLS of PA and retain PA in the cytoplasm. We found that when PA was transfected into parental and control knockdown cells, PA primarily localized in the cytoplasm. In contrast, when PA was transfected into HAX1-knockdown cells, PA localized both in the nucleus and in the cytoplasm (Fig. 4B). These data indicate that HAX1 can inhibit the nuclear accumulation of PA. This conclusion was confirmed by the observation that reexpression of HAX1 in the HAX1-knockdown cells resulted in PA being located in the cytoplasm (Fig. 4B). This conclusion was also supported by the observation that during influenza A virus infection, the relative nuclear PA/total PA ratio was higher in the HAX1-knockdown cells than in the parental cells and the HAX1-reexpressing HAX1-knockdown cells (Fig. 4D). Our data presented above are consistent with previous reports showing that when PA is expressed alone in the cells, PA is inefficiently transported into the nucleus (4, 9, 10, 16). For the efficient nuclear accumulation of PA, an interaction with PB1 is needed (9, 16). It is suggested that the NLS of PA is sterically hindered when expressed alone in the cells. However, in the presence of PB1, the binding of PB1 may induce a conformational change of PA, resulting in the exposure of NLS, leading to efficient nuclear transport of PA (16). We show here that HAX1 can interact with the NLS domain of PA (Fig. 3). This interaction may mask the NLS of PA, leading to inefficient nuclear accumulation of PA. Upon interaction with PB1, a conformational change in PA may cause the release of HAX1, resulting in the exposure of NLS, leading to efficient nuclear transport of PA. Indeed, by using coimmunoprecipitation assays, we have found that in the presence of PB1, the interaction between PA and HAX1 was reduced (Fig. 5). Thus, our data not only provide insight into the mechanism underlying the nuclear transport of PA but also provide an explanation for the previous observations that at an early time in the virus infection cycle, while PB1 and PB2 are already present in the nucleus, PA still localizes in the cytoplasm (8, 9). That HAX1 plays an important role in regulating the functions of viral nuclear proteins through interacting with them and trapping them in the cytoplasm has also been reported in the cases of HIV-1 Rev and Vpr proteins and Epstein-Barr virus EBNA-LP protein (46, 48, 49).

We showed that the propagation of influenza A virus was significantly increased in the HAX1-knockdown cells compared to that in the parental cells and control knockdown cells (Fig. 6). Moreover, reexpression of HAX1 in HAX1-knockdown cells could rescue the phenotype caused by HAX1 knockdown (Fig. 6). Together, these data indicate that HAX1 is a negative regulator of influenza A virus propagation. At least two models could explain why HAX1 plays an inhibitory role in influenza A virus propagation. First, HAX1, an antiapoptotic protein, may inhibit influenza A virus propagation through suppressing apoptosis. Second, HAX1 may inhibit influenza A virus propagation through inhibiting nuclear accumulation of PA. Previous studies have shown that activation of apoptosis results in efficient influenza virus propagation (67, 68), whereas suppression of apoptosis leads to inhibition of influenza A virus propagation (69, 70). It is possible that HAX1 inhibits influenza A virus propagation through suppressing apoptosis. However, this model, though not completely ruled out, is not favored on the basis of the following reasons. First, we found that HAX1 cannot inhibit apoptosis induced by influenza A virus infection (Fig. 10). Second, it has been shown that inhibition of apoptosis leads to suppression of influenza A virus assembly and release (68, 70) but does not affect viral genome replication and transcription (67). In this study, we found that HAX1 could inhibit the synthesis of influenza A virus mRNA, vRNA, and cRNA (Fig. 7) and inhibit the transcription and replication of the influenza A virus-like minigenome (Fig. 8). Together, these results suggest that the inhibitory effect of HAX1 on influenza A virus propagation may not be mediated through its antiapoptotic function. Rather, we prefer the model that HAX1 inhibits influenza A virus propagation through inhibiting the nuclear accumulation of PA.

Another possible mechanism that could account for HAX1 inhibition of influenza A virus propagation is that HAX1 inhibits the endonuclease activity of PA. Through HAX1-interacting domain mapping, we noticed that HAX1 may interact with the endonuclease domain of PA (Fig. 3). It is thus possible that HAX1 inhibits the endonuclease activity of PA, leading to inhibition of influenza A virus propagation. However, HAX1 is known to be primarily localized to the mitochondria and cytoplasm and rarely to the nucleus. The possibility that HAX1 inhibits the endonuclease activity of PA is minor.

The anti-influenza A virus drugs currently used primarily target neuraminidase and the M2 ion channel protein. Unfortunately, there is now widespread resistance to drugs in both of these classes (77). Therefore, it is important to develop new drugs for the treatment of influenza A virus infection. One potential therapeutic strategy to reduce the emergence of viral resistance is to target viral proteins involved in influenza A virus transcription and replication. Our finding that HAX1, through interacting with the NLS domain of PA, can inhibit the nuclear accumulation of PA and inhibit influenza A virus propagation could provide a potential strategy to treat influenza A virus infection. With regard to this point, we have found that truncated HAX1 containing only the N-terminal 100 amino acids could inhibit the transcription and replication of the influenza A virus-like minigenome by using the minigenome reporter assay (data not shown). It is conceivable that by narrowing down the HAX1 region that interacts with the NLS of PA, it will be possible to develop HAX1-derived peptides or peptidomimetics that have therapeutic potential.