Plasmids and recombinant protein purification.

NS1 proteins used in this study are described in Table S1 in the supplemental material. Plasmids encoding the NS1 gene from an H6N6 avian influenza A virus isolate (A/blue-winged teal/MN/993/1980) and an H3N2 human influenza A virus isolate (H3N2 A/Memphis/14/1998) were provided by Clayton Naeve (40); the NS1 from an H5N1 human influenza A virus isolate (A/Viet Nam/1194/2004) and an H5N1 avian influenza A virus isolate (A/Hong Kong/156/97) were kindly provided by Yoshiro Kawaoka (University of Wisconsin) and Peter Palese (Mount Sinai School of Medicine), respectively. These plasmids were used as PCR templates to create Escherichia coli and mammalian expression plasmids. The NS1 effector domains (ED) were fused to glutathione S-transferase (GST) to create GST-NS1 expression plasmids (pGEX-2T). The H3N2, H6N6, and avian H5N1 ED contain residues 73 to 230, whereas the human H5N1 ED has a 5-amino-acid-residue truncation within the linker region, between the RNA binding domain (RBD) and ED, and contains residues 73 to 225 (6). GST-NS1 fusion proteins were expressed and purified as previously described (18). Mammalian constructs consisted of full-length NS1 fused to the FLAG epitope tag (pFLAG-CMV). To allow increased expression levels of NS1 proteins from the expression plasmids, the CPSF30 binding site at residues 184 to 188 was altered from GLEWN to RFLRY (38). Alterations to the PBM of the H6N6 and H5N1 NS1 proteins were generated with the QuikChange site-directed mutagenesis kit (Stratagene).

PDZ protein mammalian expression proteins were GW1/HA-MAGI-1B, GW1/HA-MAGI-1C (13), pcDNA3/HA-MAGI-2, pcDNA/MAGI-3-V5 (53), GW1/HA-MUPP1 (29), pRK5/Myc-PATJ (27), pcDNA/FLAG-Scribble (34), GW1/HA-Dlg1-I2, GW1/HA-Dlg1-I3, peGFP-C3-Dlg1-Δ PDZ1, peGFP- C3-Dlg1-Δ PDZ2, peGFP-C3-Dlg1-Δ PDZ3, peGFP-C3-Dlg1-Δ PDZ1-2, and peGFP-C3-Dlg1-Δ3PDZ (12). To construct a mammalian expression vector for Lin7C, PCR was used to insert the coding region of Lin7C (Origene) into a histidine-tagged expression vector (pET46; Novagen). Expression and purification of His-Lin7C were carried out according to the manufacturer's protocol (Qiagen).

GST-NS1 binding assays.

293T cells were transfected at 50% confluence with epitope-tagged PDZ protein expression plasmids with Lipofectamine and Plus reagent (Invitrogen). Cell extracts were prepared 24 h posttransfection, and binding assays with GST-NS1 proteins were performed as previously described (32).

siRNA, immunoprecipitations and immunoblots.

Small interfering RNA (siRNA) targeting human Scribble (51) with the target sequence 5′-CAGGATGAAGTCATTGGAACA-3′ was purchased from Ambion. siRNA targeting human Dlg1 was purchased from Santa Cruz. Canine Scribble and canine Dlg1 siRNA used the following target sequences: 5′-GGGAAGATGGCGAGAGCGA-3′ and 5′-GAGAAGAACCUGUCAGAAA-3′ (Dharmacon), respectively. 293T cells were reverse transfected with siRNA targeting human Dlg1 or Scribble with RNAiMAX according to the manufacturer's instructions. Protein depletion was confirmed at 48 h posttransfection by immunoblotting.

293T cells were transfected with FLAG-NS1 plasmids with Lipofectamine and Plus reagent according to the manufacturer's protocol, and extracts were prepared at 24 h posttransfection. Cell extracts were incubated with anti-FLAG M2 affinity gel (Sigma) for 1 h, and NS1 complexes were then washed with TBS (50 mM Tris [pH 7.4], 150 mM NaCl) and resuspended in SDS-PAGE loading buffer. Immunoblotting was performed as previously described (20). Antibodies used were to Myc (Cell Signaling), green fluorescent protein (GFP) (Clontech), β-actin, FLAG (M2, Sigma), hemagglutinin (HA), Scribble, Dlg1 (Santa Cruz), p85β (Serotech), V5, and Lin7C (Invitrogen).

Influenza A viruses.

Previously characterized influenza A viruses encoding the H6N6 NS segment (A/blue-winged teal/MN/993/1980) in a Udorn (A/Udorn/72) background were used for all infections (32). These viruses expressed H6N6 NS1 with a wild-type ESEV or mutant ESEA PBM. The identities of viral stocks were confirmed by DNA sequencing.

Immunofluorescence.

MDCK and A549 cells were grown on glass coverslips in 24-well plates until confluent. Cells were infected with ESEV or ESEA virus at a multiplicity of infection (MOI) of 3 in DMEM, and after 1 h the medium was replaced with DMEM containing l-(tosylamido-2-phenyl) ethyl chloromethyl ketone (TPCK) trypsin (2 μg/ml). At 10 h postinfection, the cells were washed twice with ice cold phosphate-buffered saline (PBS) and fixed with 4% formaldehyde in PBS for 30 min at 4°C. Coverslips were washed with PBS and permeabilized with 0.5% Triton X-100 for 20 min at room temperature. Coverslips were washed with PBS and blocked overnight in 5% nonfat dry milk (NFDM) in TBST (20 mM Tris-HCl [pH 7.4], 150 mM NaCl, 0.1% Tween) at 4°C. Coverslips were again incubated at 4°C overnight with primary antibody dilutions in 2.5% NFDM in TBST, followed by incubation with secondary antibodies conjugated to Alexa Fluor dyes (Invitrogen). The cells were fixed a second time and quenched with sodium borohydride (1 mg/ml) in PEM [80 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) (pH 6.8), 5 mM EGTA, 2 mM MgCl₂]. The cells were washed with PBS and then stained with 1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) in PBS to visualize the nuclei. Coverslips were mounted on glass slides using ProLong Gold antifade reagent and allowed to cure for 24 h. The cells were then imaged by deconvolution microscopy as described previously (19). A series of focal planes (Z-stacks) were captured for each coverslip with the Applied Precision DeltaVision microscope system and deconvolved with the softWoRx program (Applied Precision). A single focal plane from each Z-stack was further processed in Adobe Photoshop CS3. Alternatively, for transwell filter inserts, filters were imaged on a Carl Zeiss LSM 510 Meta confocal microscope (Carl Zeiss) and processed by LSM 510 image software (Carl Zeiss Inc.) and Adobe Photoshop CS3.

Primary antibodies used were to NP, Scribble, Dlg1 (Santa Cruz), Lin7C, ZO-1, Occludin (Invitrogen), NS1 (a generous gift from Robert Krug, University of Texas, Austin), ZO-3 (Cell Signaling), and NS1 (Immune Tech Corp.).

TER and MDCK siRNA.

MDCK cells were seeded on semipermeable transwell filter inserts (0.4-μm pore size; Costar), and TER was monitored daily. Approximately 5 days after seeding, monolayers were confluent and experienced a peak in TER, indicating polarization. Monolayers were infected with ESEV or ESEA virus at an MOI of 1 in DMEM, and after 1 h the medium was replaced with DMEM containing TPCK trypsin. TER was measured just before infection and at 3 h and 6 h postinfection. To normalize the readings, the TER of an empty well was subtracted from all infected and mock TER measurements. The ESEV- and ESEA-infected TER measurements were then compared to uninfected TER measurements and expressed as percentages of mock TER. Statistical differences between groups were determined by Student's t test, and P values of <0.05 were considered statistically significant.

To determine the percentage of infected cells at the conclusion of TER measurements, monolayers were fixed and processed for immunofluorescence. Monolayers were stained for NP and Scribble and then scored for the presence of NP.

MDCK cells were reverse transfected by RNAiMAX, following the manufacturer's protocol, with siRNA targeting canine Scribble and Dlg1 (Dharmacon). SiGLO (Dharmacon) and human Scribble siRNA served as a transfection marker and negative control, respectively. The siRNA-transfected cells were grown on transwell filter inserts, and TER was measured daily. Depletion of Scribble and Dlg1 were confirmed by immunofluorescence and immunoblotting.

Interaction of avian NS1 ESEV with Dlg1 acts in a cooperative manner to facilitate binding of Lin7C.

Dlg1 has key roles in regulating tight junctions, cellular polarity, and cell cycle progression. Lin7C forms a tripartite complex with Dlg1 and MPP7 within the Crumbs complex, and this protein complex is involved in cellular polarity (3, 4, 5, 31, 48). Dlg1 has been previously shown to be a target of the oncogenic adenovirus E4-ORF1, human papillomavirus (HPV) E6, and human T cell leukemia virus type 1 (HTLV-1) Tax proteins. The interaction between these viral proteins and Dlg1 during viral infection leads to the disruption of cellular polarity, perturbation of signaling pathways, or syncytium formation (12, 21, 33, 49, 54). Dlg1 contains three PDZ domains, a SRC homology domain 3 (SH3), an I3 or I2 domain, and a guanylate kinase (GK) homology domain (Fig. 1 A).

Open in a separate window

Fig. 1.

Requirements of Dlg1 and Lin7C for binding to the avian influenza NS1 protein in vitro. (A) Domain structure of Dlg1 is shown: three PDZ domains, a SRC homology domain 3 (SH3), an I3 domain, and an inactive guanylate kinase (GK) domain. Deletion mutants of the indicated PDZ domains are shown. (B) The indicated GFP-tagged Dlg1 deletion mutants were transfected into 293T cells, cell extracts were prepared at 24 h posttransfection, and binding assays performed with the GST-NS1 ED containing the ESEV PBM. Products of binding reactions were evaluated by immunoblots using GFP antisera; binding reaction products were also evaluated by Coomassie blue staining shown in the bottom panel. Lanes indicated by “−” represent input in binding reactions; lanes indicated by “+” are products of the binding reactions. (C) 293T cells were transfected with siRNAs targeting Scribble or Dlg1 for 48 h; cultures were then transfected with full-length FLAG-tagged NS1 protein (H6N6) with indicated PBM or were mock transfected. Lysates were prepared at 24 h posttransfection and used in coimmunoprecipitations. Products of immunoprecipitations were analyzed in immunoblots.

To determine which Dlg1 PDZ domains are required for an association with the avian NS1, we transfected GFP-Dlg1 deletion plasmids into 293T cells and used cell lysates in binding assays with the H6N6 GST-avNS1 ED protein containing the ESEV PBM (Fig. 1B). Deletion of PDZ domain 1 or 3 alone did not affect binding of H6N6 GST-avNS1-ESEV ED to Dlg1, and deletion of PDZ domain 2 substantially reduced but did not eliminate binding of H6N6 GST-avNS1-ESEV ED to Dlg1, whereas deletion of PDZ domain 1 and approximately half of PDZ domain 2 abolished the interaction. These data suggest that the interaction between the H6N6 GST-avNS1-ESEV ED and Dlg1 involves PDZ domains 1 and 2, which represent a single functional unit where truncated PDZ domain 2 may confer partial PDZ domain 1 binding to H6N6 ED (12, 28, 59). While it has been suggested that the Dlg1 PDZ domain 3 is critical for binding of ESEV containing NS1 to Dlg1, the association was not completely abolished until deletion of the second Dlg1 PDZ domain (52). Possible explanations for this difference may depend upon the Dlg1 constructs used, as the ones used here have an intact C terminus, containing a third of the total Dlg1 protein and the I3, SH3, and GK domains, which may influence PDZ domain availability and presentation for binding.

To characterize the association between NS1 and Lin7C, a His-tagged Lin7C protein was expressed and purified from E. coli. Purified wild-type GST-avNS1-ESEV ED and mutant GST-avNS1-ESEA ED proteins were used in in vitro binding assays. However, we were unable to demonstrate a specific direct interaction between the recombinant avNS1-ESEV ED and Lin7C proteins, suggesting that the interaction detected in coimmunoprecipitations is indirect (data not shown). Because Lin7C and Dlg1 are found in the Crumbs complex and associate through the L27 domain (5), we suspected that Dlg1 may allow Lin7C to associate with the avNS1-ESEV protein through an indirect interaction. To test this, 293T cells were first cotransfected with siRNAs that deplete Dlg1 and with a vector that expresses either FLAG-avNS1-ESEV or FLAG-avNS1-ESEA. As a specificity control, siRNAs that deplete Scribble were also included in the experiment. Cell extracts were then immunoprecipitated with an antibody against the FLAG epitope (Fig. 1C). In cells depleted of Dlg1, neither the avNS1-ESEV nor avNS1-ESEA coimmunoprecipitated with endogenous Lin7C as determined in immunoblots. It is likely that the association between Lin7C and Dlg1 occurs through L27 domains at the N terminus of each protein (5). However, in cells depleted of Scribble, wild-type avNS1-ESEV but not mutant avNS1-ESEA coimmunoprecipitated with endogenous Lin7C. These data indicate that the association between the avNS1-ESEV protein and Lin7C is dependent upon Dlg1, supporting the idea of an indirect interaction between NS1 and Lin7C.

Characterization of canonical and noncanonical NS1 PBM associations with PDZ proteins.

We have previously shown that an H6N6 NS1 protein with the ESEV PBM specifically associates with Scribble, Dlg1, MAGI-1, MAGI-2, and MAGI- 3 in vitro (32). To determine if other NS1 PBM sequences also confer binding to these PDZ proteins, we constructed GST-avNS1 fusion proteins with a variety of PBM sequences for use in in vitro binding assays. Our previously characterized H6N6 GST-avNS1-ESEV ED protein was used as the ED platform, and PBM sequences from a variety of influenza A virus isolates were appended to the ED C terminus. These PBM sequences were ESEV (H6N6 A/blue-winged teal/MN/993/1980), EPEV (H5N1 A/Hong Kong/156/1997), KSEV (H1N1 A/Brevig Mission/1/1918), and RSKV (H3N2 A/Memphis/14/1998). These fusion proteins were called GST-avNS1-ESEV, GST-avNS1-EPEV, GST-avNS1-KSEV, and GST-avNS1-RSKV, respectively. Additionally, a GST-avNS1-ESEA (modified H6N6 A/blue-winged teal/MN/993/1980) and a human NS1 (H3N2 A/Memphis/14/1998) with the RSKV PBM at its C terminus, termed GST-huNS1-RSKV, were used as specificity controls.

Epitope-tagged PDZ protein expression plasmids were transfected into 293T cells, and cell extracts were subjected to pulldown assays with GST-NS1 ED proteins immobilized on glutathione-Sepharose beads. The binding reaction products were examined in immunoblot assays (Fig. 2 A); Coomassie blue stains showed equivalent levels of GST-NS1 ED proteins were used in the binding assays (data not shown). As observed previously, neither MUPP1 nor PATJ associated with any of the GST-NS1 proteins. Surprisingly, only an NS1 ED with the ESEV PBM associated with Scribble, Dlg1, MAGI-1, MAGI-2, and MAGI-3. These data indicate that the EPEV, KSEV, and RSKV PBMs in the context of an H6N6 NS1 ED are unable to bind these PDZ proteins. As expected, the H3N2 NS1 ED with the RSKV PBM did not associate with any of these proteins, as reported previously (32).

Open in a separate window

Fig. 2.

In vitro binding assays with NS1 PBMs from a variety of viral isolates. (A) 293T cells were transfected with expression plasmids for the indicated epitope-tagged PDZ proteins, cell extracts were prepared at 48 h posttransfection, and binding assays were performed with GST-NS1 EDs containing the indicated PBM sequences. Products of binding reactions were examined in immunoblots. Coomassie blue stains of the products of binding reactions confirmed that equivalent amounts of GST-NS1 fusion proteins were used in binding assays (not shown). (B) H5N1 and H6N6 GST-NS1 ED fusion proteins were used in binding assays with extracts from 293T cells transfected with the PDZ protein expression plasmids indicated in panel A. The products of binding assays were resolved by SDS-PAGE and examined by immunoblotting. Gels were also stained with Coomassie blue to demonstrate that equivalent levels of GST-NS1 ED proteins were used in binding assays (not shown). (C) 293T cells were transfected with full-length H6N6 or H3N2 FLAG-tagged NS1 protein with the indicated PBMs. Cell extracts were prepared 24 h posttransfection and used in coimmunoprecipitations. Products of immunoprecipitations were examined by immunoblotting.

To further characterize the NS1 PBM, we examined two NS1 proteins from highly pathogenic H5N1 viruses with a canonical ESEV PBM and a noncanonical EPEV PBM. A previous mouse infection study found that recombinant influenza virus that expressed NS1 proteins with ESEV or EPEV PBMs exhibited increased morbidity and mortality compared to PBM-truncated NS1 proteins (23). We constructed GST-NS1 ED fusion proteins from the H5N1 NS1 proteins termed GST-H5N1NS1-ESEV (H5N1 A/Viet Nam/1194/2004) and GST-H5N1NS1-EPEV (H5N1 A/Hong Kong/156/97). Additionally, a GST-H5N1NS1-ESEA (H5N1 A/Viet Nam/1194/2004) and a GST-H5N1NS1-EPEA (H5N1 A/Hong Kong/156/97) were constructed to serve as specificity controls. Binding assays were performed as described above for the H6N6 GST-avNS1 ED proteins. As seen in Fig. 2B, Scribble, Dlg1, MAGI-1, MAGI-2, and MAGI-3 associated with the GST-H5N1NS1-ESEV ED protein but not with the GST-H5N1NS1-EPEA, -ESEA, or -EPEV ED protein (EPEA and ESEA not shown). These data suggest the noncanonical PBM EPEV may target a different subset of PDZ proteins or even non-PDZ proteins.

The GST-NS1 ED in vitro binding assays do not address possible contributions of the NS1 RBD on the PBM's ability to associate with cellular targets. To investigate the binding properties of PBMs in the context of a full-length NS1 protein, we constructed expression plasmids for the H6N6 avNS1 protein with EPEV, KSEV and RSKV PBM sequences. These H6N6 NS1 proteins were termed avNS1-EPEV, avNS1-KSEV, and avNS1-RSKV. FLAG-tagged expression plasmids for the full-length human NS1 with the RSKV PBM, termed huNS1-RSKV, and the avNS1-ESEA proteins served as specificity controls. Additionally, we constructed two full-length FLAG-tagged H5N1 NS1 proteins with the ESEV and EPEV PBM, termed H5N1NS1-ESEV (H5N1 A/Viet Nam/1194/2004) and H5N1NS1-EPEV (H5N1 A/Hong Kong/156/97). Following transfection of these expression plasmids into 293T cells, cell extracts were coimmunoprecipitated with anti-FLAG antibodies, and precipitated proteins were visualized by immunoblotting (Fig. 2C). The p85β subunit of phosphatidylinositol 3-kinase (PI3K) coimmunoprecipitated with all NS1 proteins as expected (data not shown), as its NS1 binding site lies within ED (44). Consistent with the GST-NS1 ED in vitro binding assays, only the avNS1-ESEV and H5N1NS1-ESEV coimmunoprecipitated with endogenous Scribble and Dlg1. The avNS1-ESEV and H5N1NS1-ESEV proteins also clearly associated with endogenous Lin7C, presumably in a complex with Dlg1 (see Fig. 1C).

PBM-dependent colocalization of NS1 with Dlg1 and Scribble and disruption of tight junction markers during a viral infection.

We previously constructed recombinant H3N2 influenza A viruses (Udorn strain) encoding an H6N6 NS1 protein with either a wild-type ESEV PBM or a mutant ESEA PBM, enabling the functional analyses of the NS1 PBM during influenza A virus infections in cultured mammalian cells. We found that infections with the ESEV but not mutant ESEA virus resulted in the partition of Scribble into perinuclear puncta that colocalized with the NS1-ESEV protein (32). We used these recombinant viruses to examine the localization of Dlg1 and Lin7C during infection. A549 cells were infected with the ESEV or ESEA virus and analyzed by immunofluorescence at 10 h postinfection. Infection with mutant ESEA virus does not alter the localization of Scribble or Dlg1 from the cell membrane, as the localization of these proteins was similar in mock-infected cells (Fig. 3 A). In contrast, infection with the ESEV virus resulted in the sequestration of Dlg1 to perinuclear puncta that colocalized with NS1, as seen previously with Scribble. Significant levels of all three proteins colocalized as white puncta in a merged enlargement (Fig. 3B). It should be noted that most NS1 is nuclear at this time point and the majority of NS1 proteins found within the cytoplasmic puncta appear to colocalize with at least one PDZ protein. Previous data have shown that Scribble, Dlg1, and NS1 partition into an insoluble cell fraction when extracts are prepared at 6 h postinfection, suggesting that these cytoplasmic puncta containing the three proteins represent an insoluble cell fraction (32). In the case of Scribble, the relocalization correlated with the loss of Scribble's proapoptotic function and enhanced viral replication (32). The data shown suggest that Dlg1 may also be functionally inactivated as it is sequestered within puncta containing Scribble and NS1 proteins.

Open in a separate window

Fig. 3.

Avian NS1 with ESEV PBM colocalizes with Scribble and Dlg1 in cytoplasmic puncta while disrupting Lin7C and tight junction markers. (A) A549 cells were infected at an MOI of 3 with the indicated influenza viruses. Cells were fixed at 10 h postinfection for immunofluorescence analysis with Dlg1, Scribble, and NS1 antisera. (B) Enlargement of indicated infected cell. Puncta with Scribble, Dlg1, and NS1 are displayed as white; puncta with only NS1 and Dlg1 are displayed as cyan; puncta with Scribble and Dlg1 are displayed as yellow; puncta with NS1 and Scribble are displayed as violet. (C) A549 cells were infected at an MOI of 3 with the indicated influenza viruses, and cells were fixed at 10 h postinfection for immunofluorescence analysis with Lin7C and nucleoprotein antisera. (D) A549 cells were infected at an MOI of 3 with the indicated influenza virus, and cells were fixed at 10 h postinfection for immunofluorescence analysis with Lin7C, Scribble, and Dlg1 antisera. (E) MDCK cells were infected at an MOI of 3 with the indicated viruses. Cells were fixed at 10 h postinfection for immunofluorescence analysis with nucleoprotein, ZO-1, or Occludin antisera.

We also examined Lin7C localization during infection with ESEA and ESEV virus (Fig. 3C). In mock-infected and ESEA virus-infected cells, Lin7C remained intact at the cell membrane, as seen with Scribble and Dlg1. However, in cells infected with the ESEV virus, Lin7C was partially mislocalized away from normal continuous staining at the cell border and relocalized to cytoplasmic puncta. Additionally, the relocalized cytoplasmic Lin7C colocalized with Scribble and Dlg1 in ESEV virus-infected cells, as indicated by the white puncta in the merged image (Fig. 3D).

We also examined TJ integrity following infection with the H3N2 recombinant viruses, as there is evidence to suggest downregulation of TJ proteins during influenza infection (60). Additionally, Dlg1 and Scribble have been shown to play important roles in cellular junction integrity (22, 42, 48). We used MDCK cells for these experiments, as this polarized epithelial cell line forms functional TJs. MDCK cells were infected with the ESEV or ESEA virus and examined by immunofluorescence at 10 h postinfection (Fig. 3E). We examined both AJ and TJ markers, as Lin7C and Dlg1 have been shown to associate in complexes at the AJ and to influence TJ proteins (5, 47). We found that infection with ESEV or ESEA virus does not discernibly affect the localization of AJ markers E-cadherin and β-catenin or the TJ markers claudin (data not shown) and ZO-3 (Fig. 3E). However, while uninfected cells and cells infected with the ESEA virus showed the expected TJ localization of ZO-1 and Occludin, cells infected with the ESEV virus instead exhibited increased cytoplasmic staining and discontinuous plasma membrane staining for ZO-1 and Occludin. It is notable that ZO-1 is reported to associate with Scribble and to be crucial for proper Scribble localization in cells (22). These findings suggest that binding of NS1 to Scribble and Dlg1 may structurally disrupt the TJ in cells.

This work was supported by grants 5U54AI057156-05; and R21AI083396; (to A.P.R.) and R01CA058541 (to R.T.J.).

We thank Rita Czako for recombinant Lin7C experiments and the Integrated Microscopy Core and Budi Utama for assistance with deconvolution and confocal microscopy. We also thank Sue Crawford for critical comments on the manuscript and technical assistance with TER experiments.