Importin α5 negatively regulates importin β1-mediated nuclear import of Newcastle disease virus matrix protein and viral replication and pathogenicity in chicken fibroblasts

ABSTRACT The matrix (M) protein of Newcastle disease virus (NDV) is demonstrated to localize in the nucleus via intrinsic nuclear localization signal (NLS), but cellular proteins involved in the nuclear import of NDV M protein and the role of M's nuclear localization in the replication and pathogenicity of NDV remain unclear. In this study, importin β1 was screened to interact with NDV M protein by yeast two-hybrid screening. This interaction was subsequently confirmed by co-immunoprecipitation and pull-down assays. In vitro binding studies indicated that the NLS region of M protein and the amino acids 336–433 of importin β1 that belonged to the RanGTP binding region were important for binding. Importantly, a recombinant virus with M/NLS mutation resulted in a pathotype change of NDV and attenuated viral replication and pathogenicity in chicken fibroblasts and SPF chickens. In agreement with the binding data, nuclear import of NDV M protein in digitonin-permeabilized HeLa cells required both importin β1 and RanGTP. Interestingly, importin α5 was verified to interact with M protein through binding importin β1. However, importin β1 or importin α5 depletion by siRNA resulted in different results, which showed the obviously cytoplasmic or nuclear accumulation of M protein and the remarkably decreased or increased replication ability and pathogenicity of NDV in chicken fibroblasts, respectively. Our findings therefore demonstrate for the first time the nuclear import mechanism of NDV M protein and the negative regulation role of importin α5 in importin β1-mediated nuclear import of M protein and the replication and pathogenicity of a paramyxovirus.


Introduction
Newcastle disease virus (NDV), a member of the genus Avulavirus in the family Paramyxoviridae, is an important avian pathogen that causes substantial economic losses to the poultry industry worldwide [1,2]. The genome of NDV is a non-segmented, single-stranded, negative-sense RNA encoding at least six proteins in the order 3'-NP-P-M-F-HN-L-5' [3]. Of all these viral structural proteins, the matrix (M) protein has the least molecular weight of around 40 kDa and forms an outer protein shell around the nucleocapsid, which constitutes the bridge between the viral envelope and the nucleocapsid [4]. Like most paramyxovirus M proteins, the NDV M protein is a multifunctional nucleocytoplasmic shuttling protein and plays crucial roles in NDV life cycle [5]. In addition to functioning for the assembly and budding of progeny virions in the cytoplasm and at the cell membrane later in infection [6], the NDV M protein is localized in the nucleus and nucleolus early in infection and remains in the nucleoli throughout infection [7][8][9]. The nuclear-nucleolar localization of NDV M protein is thought to inhibit host cell transcription and protein synthesis similar to the human respiratory syncytial virus (HRSV) [10] and vesicular stomatitis virus (VSV) M protein [11], and also ensure that viral replication and transcription in the cytoplasm proceed smoothly, which is by analogy with the measles virus (MeV) M protein [12]. Numerous studies have demonstrated that the nuclear and nucleolar localization of viral proteins depend on their own nuclear localization signal (NLS) and nucleolar localization signal (NoLS) as well as the cellular transport proteins [13][14][15]. Our recent study found that the nucleolar protein B23 targets NDV M protein to the nucleoli by interacting with the amino acids 30-60 (a putative NoLS) of M protein and enhances viral replication ability [16]. Previous study has shown that NDV M protein enters the nucleus via a bipartite NLS (KKGKKVIFDKIEEKIRR) independent of other viral proteins [17], but cellular proteins involved in the nuclear import of NDV M protein and the biological functions of this nuclear localization still remain unknown.
It has been demonstrated that nuclear transport of proteins carrying NLS through the nuclear envelopeembedded nuclear pore complexes (NPCs) is mediated by members of the importin superfamily including importin a and importin b [18][19][20]. The classical paradigm for nuclear import pathway is that importin a directly recognizes and binds to the NLS of cargo proteins, and then importin b1 directs binding of the binary complex to the cytoplasmic side of the NPC. The translocation of GDP-bound small GTPase Ran (RanGDP) conjunct ternary complex through the NPC is mediated by NTF2 via interaction with nucleoporins. Once inside the nucleus, binding of GTP-bound small GTPase Ran (RanGTP) to importin b1 causes the dissociation of the ternary complex [21,22]. Thus, cargo proteins are transported into the nucleus. In general, classical NLSs including monopartite and bipartite NLSs are imported by importin a/b1 heterodimer, while non-classical NLSs can be more complex in length, sequence and amino acid composition that are imported by importin b1 [21]. However, recent studies have found that classical NLSs can also be recognized and binded by importin b1 or homologs without the participation of importin a [23][24][25]. Moreover, some studies even confirmed that importin a can act as negative regulators for the nuclear import of some cargo proteins mediated by importin b1 alone [26,27]. Therefore, such diverse nuclear import pathways are receiving increasingly attention.
Many recent studies have proven that nuclear localization of viral proteins is crucial for viral replication and propagation [28][29][30][31]. For example, nuclear localization of the nucleocapsid protein of porcine respiratory and reproductive syndrome virus is essential for optimal virus replication and inhibition of cellular antiviral processes [32], and the successful production of infectious virions and efficient propagation of Japanese encephalitis virus require the nuclear localization of core protein [33,34]. For the members of paramyxoviruses, the M proteins of HRSV, Sendai virus (SeV), Nipah virus (NiV), and NDV all have the molecular mass less than 40 kDa and localize in the nucleus during the course of virus infection [5]. Although functional NLSs in the M proteins of these viruses have been characterized, so far, only the HRSV M protein is demonstrated to be recruited into the nucleus through direct recognition by importin b1 [35]. In addition, nuclear localization of M protein is important for the generation of HRSV progeny virions and is associated with the pathogenesis of viral infection [36]. Therefore, we speculated that the cellular importin members might also participate in the nuclear import of NDV M protein and regulate the replication and pathogenicity of NDV.
In this study, importin b1 was identified to be the nuclear transport receptor of NDV M protein and mediate the nuclear import of NDV M protein by binding its NLS region via the RanGTP-dependent pathway. Further studies showed that NLS mutation in the M protein disrupted its nuclear localization and reduced viral replication in chicken fibroblasts and attenuated viral replication and pathogenicity in SPF chickens. Interestingly, importin a5 was demonstrated to be a negative regulator in importin b1-mediated nuclear import of M protein and the replication and pathogenicity of NDV, as importin a5 deletion remarkably increased the nuclear accumulation of M protein and the replication ability and pathogenicity of NDV in chicken fibroblasts. Our studies will provide deep insights into understanding the more functions of M's nuclear localization in the life cycle and pathogenesis of NDV.

Yeast two-hybrid screening of NDV M-interacting cellular proteins
To define cellular proteins that interact with NDV M protein, a yeast two-hybrid screening strategy was employed. The NDV M protein was used as the bait for screening the cDNA library generated from DF-1 cells. The results showed that six colonies (No. 3 to 8) could grow on the auxotrophic medium SD/-Ade/-His/-Trp/-Leu and turn blue in the presence of X-a-gal, which had the same presentation as the positive control transformed with pGBKT7-53 and pGADT7-T (No. 2) ( Fig. 1A and B). However, the negative control could neither not grow nor turn blue on the medium SD/-Ade/-His/-Trp/-Leu/X-a-gal (No. 1). To further confirm the true interaction with M protein in yeast, b-galactosidase colony-lift filter assay was performed. We found that yeast colonies co-transformed with the pGADT7-derivative plasmids and pGBKT7-M plasmid all turned blue within 30 min, which was similar to the positive control (Fig. 1C). The nucleotide sequences of six colonies from the cDNA library were then sequenced and analyzed. The results showed that the sequences of six clones had the right open reading frame (ORF) in frame with the AD coding region, and five cellular proteins containing the interaction region were identified by bioinformatics analysis (Table 1). Fortunately, of all these proteins, importin b1 is demonstrated to be a karyopherin that transports multiple proteins owing NLS into the nucleus [23]. Therefore, we conclude that NDV M protein may bind importin b1 to enter the nucleus.

NDV M protein interacts with importin b1 in vivo and in vitro
To verify the interaction between NDV M protein and importin b1, we first performed co-immunoprecipitation assay with DF-1 cells transiently transfected with plasmid encoding Myc-importin b1 and infected with NDV. Expression of the fusion protein Myc-importin b1 and viral M protein was confirmed with anti-Myc and anti-M antibodies, respectively ( Fig. 2A, upper panel). In addition, the M protein and importin b1 in cell supernatants could be immunoprecipitated with each other when using anti-Myc or anti-M antibody ( Fig. 2A, middle and lower panels). Next, in vitro binding assay of the fusion protein GST-M to the purified His-importin b1 protein showed that His-importin b1 was pulled-down by GST-M protein but not by GST (Fig .2B). These results suggest that that NDV M protein physically interacts with importin b1 protein in vivo and in vitro.
Previous studies have shown that importin b1 mediates the nuclear import of cargo proteins by binding their NLS [23][24][25]. Here, the binding studies revealed that the NLS region in the M protein was important for importin b1 binding, since GST-M protein with the mutated NLS lost its binding activity to importin b1 (Fig. 2C). In addition, His-importin b1 deleting residues 336 to 433 (4336-433) could not be pulled down by GST-M (Fig. 2D), indicating that the amino acids 336-433 of importin b1 was essential for interaction with M. It is reported that human importin b1 contains importin-b N-terminal (IBN_N) domain at the N-terminus and several "HEAT repeat" motifs that mostly occupy the C-terminal portion [37]. After comparison with the amino acids of human importin b1, the residues 336 to 433 in chicken belonged to the 8-10 HEAT repeats, which were also the RanGTP binding region. Thus, we demonstrate that the NLS region of NDV M protein and the 8-10 HEAT repeats of importin b1 are important for interaction with each other.

NLS mutation in the M protein attenuates the replication and pathogenicity of NDV
Now that the NLS of M protein was essential for its interaction with importin b1, we investigated the effect of M/ NLS mutation on the virulence, replication ability and pathogenicity of NDV. The results of virus rescue showed that hemagglutination (HA)-positive allantoic   Figure  S1 in the supplemental material), and no additional mutations were observed in the M gene and other viral genes (data not shown). The immunofluorescence results showed that NLS mutation absolutely disrupted the nuclear localization of M protein in virus-infected cells (Fig. 3A)    revealed that the virus titers of rSS1GFP-M/NLSm were remarkably reduced in comparison to that of rSS1GFP from 12 to 72 hpi (P<0.01) (Fig. 3C). Meanwhile, the cytopathic effect (CPE) in rSS1GFP infected cells started at 12 hpi and cell monolayer was absolutely destroyed at 72 hpi, but the CPE in rSS1GFP-M/NLSm infected cells started at 24 hpi and cell monolayer was still existent at 72 hpi (Fig. 3D), which showed much slighter and slower CPE than that of rSS1GFP-infected cells at the same time points.
The in vivo pathogenesis assessment of M/NLS mutant virus rSS1GFP-M/NLSm in 4-week-old SPF chickens was then evaluated. The resulting survival curves were shown in Fig. 4A. Birds inoculated with the parental virus rSS1GFP exhibited slight depression at 3 days post-infection (dpi), severe depression (3/ 10), wing drop (2/10), hemiparesis/paralysis (3/10), and death (2/10) at 4 dpi, and 100% mortality by 5 dpi. At necropsy, all euthanized chickens presented severe gross lesions in multiple organs, such as conjunctivitis, hemorrhage of the throat, trachea, thymus, duodenum mucosa, and proventriculus, multifocal necrosis of the spleen, and marked atrophy of thymus and bursa of Fabricius at 4-5 dpi. In comparison, birds inoculated with the mutant virus rSS1GFP-M/ NLSm presented delayed and slight depression (4/10) at 5 dpi, severe depression (2/10) and one death (1/ 10) at 6 dpi, and two death (2/10) at 7 dpi, but there were no death in the subsequent days. Virus titration assays showed that the mutant virus rSS1GFP-M/ NLSm had little replication ability in spleen, thymus and bursa of Fabricius at 5 dpi; by contrast, the parental virus rSS1GFP replicated in multiple tissues and had relatively higher virus titers in the lymphoid tissues (spleen, thymus and bursa of Fabricius) (Fig. 4B). In addition, the results of histopathology observation of the lymphoid tissues showed that birds inoculated with rSS1GFP displayed multifocal confluent coagulative necrosis, severe lymphocyte depletion, and infiltration of macrophages, whereas no apparent histopathological changes were observed in the lymphoid organs of the rSS1GFP-M/NLSm group and the control group (Fig. 4C). Taken together, these results clearly demonstrate that M/NLS mutation can not only reduce viral replication ability in chicken fibroblasts but also attenuate viral replication and pathogenicity in SPF chickens.

Nuclear import of NDV M protein requires importin b1 and RanGTP
Numerous studies have shown that importin b1 together with RanGDP and/or RanGTP participate in the nuclear import of many cargo proteins [38][39][40][41][42]. To identify the cellular karyopherins responsible for M nuclear targeting and further verify the nuclear import pathway of NDV M protein, two dominantnegative (DN) mutants of importin a5 (DN-importin a5) [43] and importin b1 (DN-importin b1) [44], which lack the ability to bind importin b and Ran, respectively, and nuclear import inhibitors M9M [45] or Bimax2 46 that are specific for the transportin-1 pathway or the importin a1, a3, a6 and a7 pathways, respectively, RanGTP mutant (Ran/Q69L) [47], which is deficient in GTP hydrolysis, and NTF2 mutant (NTF2/E42K) [48,49], which fails to transport RanGDP into nucleus, were first introduced to determine whether they are required for the nuclear import of NDV M protein. As shown in Fig. 5A, DF-1 cells co-transfected with plasmid pEGFP-M and plasmids encoding DsRed-DN-importin a5 or DsRed-M9M or DsRed-Bimax2 or DsRed-NTF2/E42K did not impair the nuclear localization of EGFP-M, while co-expression of either DsRed-DN-importin b1 or DsRed-Ran/Q69L inhibited the nuclear accumulation of EGFP-M. In addition, detection of the intracellular localization of EGFP-M by Western blotting also verified that the EGFP-M had the same distribution as the fluorescence microscopy ( Fig. 5B).
In vitro nuclear import assay was then performed to further confirm that importin b1 and RanGTP are sufficient for the nuclear import pathway of NDV M protein. The results showed that the fusion protein GST-M-GFP was efficiently imported into the nucleus of digitonin-permeabilized HeLa cells in the presence of exogenous cytosol (Fig. 5C). However, GST-M-GFP showed no nuclear translocation with RanGTP or RanGDP alone, which is similar to the results observed in the absence of cytosol, whereas combination of purified importin b1 plus RanGTP exhibited a much stronger level of nuclear accumulation than that seen when the cytosol was added (Fig. 5C). Importantly, importin b1 plus RanGDP or importin b1 alone or replacement of RanGTP with RanQ69LGTP (importin b1 plus RanQ69LGTP) led to a small amount of nuclear accumulation of GST-M-GFP (Fig. 5C). Therefore, these results confirm that importin b1 together with RanGTP but not RanGDP are the components required to mediate the active nuclear import of NDV M protein, and this process depends on GTP hydrolysis by Ran.

Nuclear import of NDV M protein does not require importin a
Ivermectin is reported to be specific for importin a/b1-mediated nuclear import of cargoes, and has no effect on any of the other nuclear import pathways including that mediated by importin b1 alone [50]. Therefore, ivermectin was used to further verify whether NDV M protein can be imported into the nucleus without the participation of importin a. We found that there was no obvious difference in the subcellular localization of NDV M protein in rSS1GFP-infected DF-1 cells treated with ivermectin or DMSO when compared to the untreated cells at 12 hpi ( Fig. 6A and B). In addition, the ivermectin-or DMSO-treated DF-1 cells also had the same characteristics in the generation of CPE and the expression of GFP when infected with rSS1GFP ( Fig. 6C and D). Moreover, the replication capacity of rSS1GFP was not affected in either ivermectin or DMSO treated cells at 12 hpi and 24 hpi (Fig. 6E). These results demonstrate that nuclear import of M protein and the replication of NDV are not inhibited by ivermectin, and therefore do not require importin a.

Importin a5 interacts with NDV M protein by binding importin b1
Because some of the importin a members can act as negative regulators for importin b1-mediated nuclear import of cargo proteins, so we wanted to search for the potential importin a that may indirectly bind NDV M protein and study its role in the nuclear import process of NDV M protein. Up to now, there are seven importin a members including importin a1, a3, a4, a5, a6, a7 and a8 have been identified in humans and many animals [51][52][53]. The immunofluorescence assay was first performed to examine the subcellular localization of these fusion proteins expressed by the recombinant eukaryotic expression vectors. As shown in Fig. 7A, besides the cytoplasmic and nuclear envelope localization of Myc-importin a1 and Myc-importin b1, the other fusion proteins including Myc-importin a3 to Myc-importin a8 and HA-M exhibited the similar localization in the nucleus. Next, the co-immunoprecipitation and pulldown assays were used to identify the potential importin a proteins that interact with NDV M protein. The results showed that both Myc-importin a5 and Myc-importin b1 could be immunoprecipitated by HA-M (Fig. 7B). However, the pull-down assay indicated that GST-importin b1 but not GST-M could be pulled-down by His-importin a5 (Fig. 7C). Because endogenous importin a5 can interact with importin b1 in the process of transporting cargo proteins into the nucleus [21,22], we investigated if NDV M protein can directly bind importin a5. The results of subsequent protein-binding assays showed that when GST-M immobilized on Glutathione-Sepharose beads was incubated with His-importin a5 or His-importin b1, His-importin b1 but not Hisimportin a5 was pulled-down by GST-M (Fig. 7D,  lanes 2 and 3). However, when immobilized GST-M was incubated simultaneously with His-importin a5 and His-importin b1, both His-importin a5 and Hisimportin b1 were pulled-down by GST-M and detected by SDS-PAGE (Fig. 7D, lane 4). Therefore, these results demonstrate that importin a5 interacts with NDV M protein by binding importin b1.
Importin a5 reduces importin b1-mediated nuclear import of NDV M protein According to the above results and for the reason that importin a5 can act as negative regulator in importin b1-mediated nuclear import pathway, we hypothesized that importin a5 might play the inhibition function in the subcellular localization of NDV M protein. To test this hypothesis, the localization of M protein in importin b1or importin a5-depleted DF-1 cells was investigated. Three pairs of synthesized importin b1 or importin a5 siRNAs (see Table S1 in the supplemental material) were transfected into DF-1 cells respectively, and importin b1 RNAi#3 or importin a5 RNAi#2 could effectively lower the expression level of importin b1 or importin a5 without causing discernable changes in cell morphology ( Fig. 8A and B). In addition, the viability of those cells receiving siRNA using trypan blue exclusion and MTT assays were examined. Results showed that there was no difference between importin b1 or importin a5 RNAi and RNAi control in terms of viability of transfected cells (data not shown). We then infected DF-1 cells receiving the indicated siRNA or control siRNA with NDV strain rSS1GFP. As a result, at 6 or 12 hpi, importin b1 depletion caused M protein mainly in the cytoplasm, further confirming that importin b1 mediated the nuclear  import of M protein. However, importin a5 depletion markedly increased the nuclear accumulation of M protein when compared to that of control siRNA group (Fig. 8C). Similar results were obtained by examining the intracellular distribution of M protein in importin b1 siRNA-or importin a5 siRNA-or control siRNA-treated cells by immunoblot analysis (Fig. 8D). Together with the above results, these data reveal that the importin b1mediated nuclear import of NDV M protein is reduced in the presence of importin a5.

Importin a5 decreases importin b1-participted NDV replication and pathogenicity
Since importin a5 causes the reduction of nuclear localization of M protein, this might affect the replication and pathogenicity of NDV. To this end, siRNA-mediated knockdown of importin b1 or importin a5 in DF-1 cells infected with NDV strain rSS1GFP was investigated. As shown in Fig. 9A and B, the normal cells and control siRNA-treated cells showed the similar CPE and GFP expression level at 24 hpi, while siRNA-mediated knockdown of importin b1 markedly reduced NDV-induced CPE and GFP expression level, but knockdown of importin a5 greatly enhanced the CPE and GFP expression level when compared to the normal cell group and RNAi control group. In addition, significant reduction of viral loads in the cell culture supernatants and cell pellets of importin b1-RNAi cells was detected, but an increasing of viral loads in that of importin a5-RNAi cells were examined in comparison to that of normal cell and RNAi control (Fig. 9C and D). To further determine the effect of importin b1 or importin a5 knockdown on NDV replication, the mRNA expression levels of NDV M gene in NDV-infected importin b1-RNAi or importin a5-RNAi cells were examined at 24 hpi. We found that the mRNA expression level of NDV M gene in importin b1-RNAi or importin a5-RNAi cells significantly decreased or increased , respectively, in comparison to that of normal cells and RNAi control either in the culture supernatants or in the cell pellets (P < 0.001) ( Fig. 9E and F). Together, above results demonstrate that importin a5 acts as negative regulator in importin b1participated NDV replication and pathogenicity in cells.

Discussion
Paramyxoviruses are a diverse group of enveloped viruses with non-segmented negative-sense singlestranded RNA genomes that include a number of important human and animal pathogens [5]. To date, the pathogenic mechanism of the paramyxoviruses still attracts the global researchers' attentions. Two envelope glycoproteins, the attachment protein (termed HN for hemagglutinin-neuraminidase, H for hemagglutinin, or G for glycoprotein, depending on the virus) and the fusion (F) protein, have been always reported to be the major virulence and pathogenic factors for paramyxoviruses [54][55][56]. However, in recent years, increasing number of researches has focused on the M protein in the pathogenesis of paramyxoviruses due to its multifunction in inhibiting the host RNA and protein synthesis and facilitating the assembly and budding of progeny virions [57][58][59]. Among the members of paramyxoviruses, the M protein of HRSV, SeV, NiV and NDV is demonstrated to shuttle between the nucleus and cytoplasm through intrinsic NLS and NES [5]. Up to now, only the nucleocytoplasmic shuttling mechanism and the detailed function of HRSV M protein are elucidated [35,36]. We previously have demonstrated that the nuclear export of NDV M protein is mediated by three NESs via the CRM1-independent pathway [60], but the nuclear import mechanism of NDV M protein and the definite functions of M protein in the nucleus are still not known.
Numerous studies have revealed that NLS-mediated nuclear localization of viral proteins is crucial for the replication and pathogenicity of most viruses [28][29][30][31]. We previously found that a basic amino acid mutation, R42A, in the NDV M protein not only abrogates its nuclear localization but also attenuates NDV replication and pathogenicity [61]. However, R42A mutation in the M protein does not really reflect the effect of nuclear localization disruption of M protein on the replication and pathogenicity of NDV. Here, we successfully rescued the M/NLS mutant virus rSS1GFP-M/NLSm after three extra chicken egg passages and found that the disruption of M's nuclear localization caused by M/NLS mutation not only resulted in a pathotype change of NDV but also reduced viral replication in cells and attenuated the replication and pathogenicity of NDV in SPF chickens, thus demonstrating for the first time the importance of M's nuclear localization for NDV replication and pathogenicity. In addition, the rescued viruses rSS1GFP and rSS1GFP-M/NLSm could be used to further investigate the more detailed functions of M's nuclear localization in the pathogenesis of NDV utilizing transcriptomics and proteomics. However, it was strange that the MDT and ICPI values of rSS1GFP-M/NLSm did not correlate, which showed that the MDT value suggested the virus was lentogenic, but the ICPI value suggested the virus was velogenic. In our previous studies, we also found that there was no relevance in the MDT [61,62]. Another research group has studied the effect of some conserved amino acids mutation in the HN protein on the virulence of NDV. They similarly find that in comparison to the parental virus (MDT is 62 h, ICPI is 1.51), the MDT of Y526Q mutant NDV increases to 98 h, which belongs to the lentogen strain, whereas the ICPI value (1.33) of the virus indicates that it is still mesogenic strain [63]. We speculated that if the amino acid sequences at the F protein cleavage site that determine the virulence characteristics of NDV are not changed, mutating certain key amino acids in viral proteins will probably not make the MDT and ICPI values correlate, but the ICPI values of the mutant viruses still keep the original virulence characteristics.
In recent years, cellular nuclear transport receptor proteins-mediated nuclear import of viral proteins has been always the spotlight. In the present study, importin b1 was identified to mediate the nuclear import of NDV M protein by binding its NLS region via the RanGTPdependent pathway. Numerous studies have shown that Arg/Lys-rich NLSs within cargo proteins are the binding sites for the recognition and binding of importin a or importin b [20][21][22][23][24]. Generally, classical NLSs including monopartite and bipartite NLSs are transported into the nucleus by importin a/b heterodimer, whereas non-classical NLSs can be more complex in sequence, length and amino acid composition that are imported by importin b [22,24]. However, large numbers of studies have found that classical NLSs can also be recognized and bound by importin b1, such as the NLS of HTLV-1 Rex (RRRPRRSQRKR) [64], HIV-1 Tat (RKKRRQRRR) and Rev (RQARRNRRRR) [65], Smad3 (KKLKK) [66], TopBP1 (RKRK) [67], and BLM (RSKRRK) [68]. It is reported that NDV M protein localizes to the nucleus via a bipartite NLS (KKGKKVIFDKIEEKIRR) [17]. Interestingly, our results similarly demonstrated that the classical NLS of NDV M protein was critical for interaction with importin b1 without preferentially binded to importin a. It has been demonstrated that various types of importin as are expressed at widely divergent levels in different tissues and show very different affinities for distinct NLSs [69,70]. Therefore, we concluded that efficient nuclear import of NDV M protein in various kinds of tissues could be achieved by binding importin b1 directly, rather than relying on one or more patterns of importin a as an intermediary.
Previous study has indicated that importin b1 contains two major domains, including importin-b N-terminal (IBN_N) domain at the N-terminus and multiple "HEAT repeat" regions that mostly occupy the C-terminus [37]. Of which the HEAT repeats have the ability to form different conformations in different functional states that facilitate the accommodation of their binding partners by an induced fit type of mechanism [71,72]. Several studies have confirmed that the HEAT repeats of importin b1 can provide abundant binding regions for interaction with distinct cargo proteins. For example, the cellular proteins PTHrP [73], Snail [74], BLM [68], SREBP2 [75] and TopBP1 [67] interact with importin b1 by binding the 2-11, 5-14, 14-16, 7-17 and 18-19 HEAT repeats of importin b1, respectively. Although the HEAT repeats used to bind three cargo proteins (2-11 for PTHrP, 5-14 for Snail, and 7-17 for SREBP2) is overlapped, the binding mechanism for each protein is distinctly different [72]. But anyway, the binding regions of these cargo proteins all overlap with the RanGTP binding region of importin b1 (the 8-10 HEAT repeats), indicating that the formation of cargo protein-importin b1 heterodimer requires a large contact area and the ability for the binary complex to be disassembled by RanGTP binding upon entry to the nucleus. This is in agreement with the previous obtained results [73][74][75]. Similar to the above findings, we found that the HEAT repeats 8-10 of importin b1 was responsible for interaction with NDV M protein, and the nuclear import of M protein was dependent on RanGTP, further confirming that cargo proteins that interact with the RanGTP binding region of importin b1 require the RanGTP for nuclear targeting.
Although importin a/b1 heterodimer-mediated nuclear import is thought to be widely used in cells [22], but some studies find that importin a protein can negatively regulate the nuclear import of cargo proteins mediated by importin b1 [26,27]. The action mechanism shows that importin a interacts with the domain of cargo proteins to compete with the binding of importin b1 [26], or forms a ternary complex with the cargo protein-importin b1 complex to reduce the nuclear transport process [27]. In this study, we demonstrated that importin a5 interacted with the importin b1-M binary complex to decrease the nuclear import efficiency of NDV M protein, for the reason that importin a5 directly binded importin b1 but not M protein, and depletion of importin a5 remarkably increased the nuclear accumulation of M protein. In addition, we found that siRNA-mediated knockdown of importin a5 in DF-1 cells infected with NDV caused more serious CPE and more increased viral replication ability in comparison to that of normal cells and RNAi control cells. On the contrary, importin b1 depletion not only disrupted the nuclear localization of M protein but also greatly reduced the CPE induced by NDV infection and viral replication capacity. Moreover, we also found that knockdown of importin b1 decreased the expression levels of NDV M protein and GFP in cells. It has been shown that the nuclear localization of paramyxovirus M protein has two main functions: (i) inhibit host gene transcription and protein synthesis, and (ii) ensure that the replication and transcription of viral genome in the cytoplasm proceed until a certain level of viral protein and RNA expression is reached, at which point M is transported into the cytoplasm to participate in virus assembly [57]. Therefore, we speculated that the disruption of M's nuclear localization could not achieve the above functions and led to the reduction of the replication and transcription of viral genome during the course of virus infection. But more experiments were needed to verify the definite nuclear localization functions of NDV M protein. Together with the above results, our studies clearly demonstrated that importin b1 and RanGTP were responsible for the nuclear import process of NDV M protein, and importin b1 alone could increase the nuclear import efficiency of M protein and enhance NDV replication and pathogenicity (Fig. 10A), whereas such case could be decreased in the presence of importin a5 (Fig. 10B).
In summary, we demonstrated for the first time the nuclear import mechanism of NDV M protein and the negative regulation role of importin a5 in importin b1mediated nuclear import of M protein and the replication and pathogenicity of NDV. Our results will provide a better understanding of the exact role of M's nuclear localization in NDV life cycle and aid in understanding the poorly understood NDV pathogenesis.
To generate the recombinant infectious clones harboring alanine (A) substitution targeting basic amino acid residues in the NLS motif ( 247 AAGAAVIFDKIEEKIAA 263 ) of M protein, the fragment containing the restriction enzyme sties AgeⅠ and BstZ17Ⅰwas amplified by two pairs of specific primers, which were used to introduce multiple amino acid substitutions in the M protein, to generate two overlapping PCR fragments. The two PCR fragments were joined in a second PCR and the obtained fragment was digested with AgeⅠ and BstZ17Ⅰ to replace the corresponding region in the full-length cDNA clone pNDV/SS1GFP (see Figure S2 in the supplemental material) [77]. The resulting plasmid was named pNDV/SS1GFP-M/NLSm. All the recombinant plasmids were confirmed by PCR, restriction digestion and DNA sequencing. Primers used in this study are available upon request.
Yeast two-hybrid screening and colony-lift filter assay The yeast AH109 containing bait plasmid pGBKT7-M was grown on SD/-Trp/X-a-gal, SD/-Trp/-His/X-a-gal and SD/-Trp/-Ade/X-a-gal to exclude the autonomous transcriptional activity. Then the transformed AH109 mated with yeast Y187 containing pGADT7-Rec with the cDNA library of DF-1 cells for 24 h. The screening was performed according to the manufacturer's instructions (Matchmaker TM GAL4 Two-Hybrid System 3) and as previously described [78]. In b-galactosidase colonyfilter assay, the prey plasmids in the suspected positive clones were rescued and then co-transformed into AH109 with the plasmid pGBKT7-M. Positive clones grown on SD/-Ade/-His/-Trp/-Leu medium were tested for b-galactosidase activity. The pGBKT7-53 and pGADT7-T or pGBKT7-Lam and pGADT7-T co-transformed group was used as a positive and negative control, respectively. Yeast colonies co-transformed with pGBKT7-M and the pGADT7-derivative plasmids were checked periodically for the appearance of blue colonies. Yeast plasmid was isolated from blue colonies as previously described [79], and the inserted fragment was obtained by PCR amplification and then analyzed by bioinformatics methods.
Cell culture, transfection and fluorescence microscopy DF-1 cells were cultured in Dulbecco's modified Eagle's medium (DMEM, GIBCO) containing 10% fetal bovine serum (FBS, GIBCO) supplemented with 100 U/ml penicillin and 100 mg/ml streptomycin at 37 C under an atmosphere with 5% CO 2 . For the transfection experiments, 4 £ 10 5 DF-1 cells were grown to 80% confluence in 35-mmdiameter dishes and then double or single transfected with a total of 3 mg plasmid using the FuGENE HD Transfection Reagent (Roche) according to the manufacturer's recommendations. Twenty-four hours after transfection, cells expressing the fluorescence-fused proteins were rinsed with phosphate-buffered saline (PBS), fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.25% Triton X-100 in PBS for 5 min, and then counterstained with DAPI (Sigma) to detect the nuclei. Fluorescent images were obtained under a Nikon fluorescence microscope (Japan). Analysis and merging of the images were done with Adobe Photoshop 7.0 software.

Protein interaction assays
For co-immunoprecipitation assay, 4 £ 10 5 DF-1 cells grown in 35-mm-diameter dishes were transfected with plasmid pCMV-Myc-importin b1 for 24 h and then infected with NDV strain rSS1GFP at a multiplicity of infection (MOI) of 0.1. On the other hand, DF-1 cells cultured in 35 mm dishes were co-transfected with pCMV-HA-M and the indicated plasmids. At 24 h postinfection or post-transfection, cells were washed and lysed with immunoprecipitation buffer. After centrifugation, the supernatants were collected and incubated with the corresponding antibodies overnight at 4 C. The immune complexes were recovered by adsorption to protein A+G-Sepharose (Sigma) for 3 h at 4 C. After three washes in immunoprecipitation buffer, the immunoprecipitates were detected by Western blotting.
For pull-down assays, the His-importin b1 or Hisimportin b1(4336-433) or His-importin a5 fusion protein was expressed in E. coli BL21 (DE3) (4 h induction with 1.0 mM IPTG at 30 C), and the soluble Histagged proteins were purified on Ni-NTA His Bind Resin. GST-M and GST-M/NLSm (4 h induction with 0.5 mM IPTG at 28 C), and GST-importin b1 (4 h induction with 1.0 mM IPTG at 30 C) were expressed in E. coli BL21 (DE3), respectively, and the soluble fusion proteins were purified on Glutathione-Sepharose beads. In the GST pull-down experiments, the purified GST-M or GST-M/NLSm or GST-importin b1 protein was immobilized on Gluthatione-Sepharose beads (3 mg protein/10 mL beads). After washing with transport buffer, the immobilized proteins were incubated with the purified His-importin b1 or His-importin b1(4336-433) or His-importin a5 (3 mg for each protein, total volume 40 mL) for 2 h at 4 C. The beads were then washed three times with transport buffer and the bound proteins were eluted from the beads and used for SDS-PAGE followed by Coomassie blue staining or Western blot analysis. In the His pull-down experiments, His Ã Bind Resin-binded His-importin b1 or His-importin b1 (4336-433) (3 mg protein/10 mL resins) was incubated with purified GST-M (3 mg) for 2 h at 4 C. The resins were dealt with as described above, and the target protein GST-M was detected by Western blotting.

In vitro nuclear import assays
In vitro nuclear import assays in digitonin-permeabilized cells were performed as previously described [80]. Briefly, subconfluent HeLa cells were cultured on poly-L-lysinecoated glass coverslips for 1 day and then permeabilized with 70 mg of digitonin/ml for 5 min on ice. The digitoninpermeabilized HeLa cells were rinsed twice with transport buffer and then incubated for 15 min at room temperature with the import mixture. Import reactions contained an energy regenerating system (0.5 mM GTP, 5 mM phosphocreatine, and 0.4 U of creatine phosphokinase), plus various transport factors (0.5 mM importin b1; 3 mM RanGTP; 3 mM RanGDP; 3 mM RanQ69LGTP), plus the GST-M-GFP fusion protein (0.5 mM). The final reaction volume was adjusted to 20 mL with transport buffer. After incubation, the cells were washed with transport buffer and fixed with 3.7% formaldehyde on ice followed by methanol for 5 min at -20 C. After three washes with transport buffer, the nuclei were identified by DAPI staining. The results of GST-M-GFP nuclear import were analyzed with a Nikon fluorescence microscope.

Virus rescue and pathogenicity assay
For the virus rescue, BSR-T7/5 cells were grown in DMEM medium containing 10% FBS and 1 mg/ml geneticin G418 for five generation before transfection. Cells at 70% confluence in 35 mm dishes were transfected with the full-length cDNA clone (pNDV/SS1GFP or pNDV/SS1GFP-M/NLSm) together with three SS1derived helper plasmids at a total of 3 mg [77]. At 60 h post-transfection, the cell monolayers and culture supernatants were harvested and inoculated into the allantoic cavities of 10-day-old embryonated SPF chicken eggs. The HA test and DNA sequencing were performed to identify the rescued viruses rSS1GFP and rSS1GFP-M/ NLSm. Plaque formation assays and viral titers were performed using standard methods [81]. The pathogenicity assay of the rescued viruses was determined using the standard pathogenicity tests: the MDT test in 10-day-old SPF chicken eggs, and the ICPI test in 1-day-old SPF chicks [81].

Pathogenicity assessment of the rescued viruses in 4-week-old chickens
The pathogenicity assessment of the rescued viruses rSS1GFP and rSS1GFP-M/NLSm was determined in 4week-old SPF chickens. Forty-eight chickens were assigned randomly into three experimental groups, consisting of rSS1GFP-(n = 16), rSS1GFP-M/NLSm-(n = 16) and mock-(PBS, n = 16) infected groups. For each group, 6 birds and 10 birds were used for sampling and clinical observation, respectively. Chickens were inoculated via the eye drop/intranasal route with each virus at a dose of 10 5.0 EID 50 /100 mL per bird, or with 100 mL PBS as the negative control. The birds were monitored for clinical signs daily for 10 dpi. Three birds were euthanized daily from 4-5 dpi for gross lesion observation, and samples of the heart, liver, spleen, lung, kidney, brain, trachea, duodenum, bursa of Fabricius and thymus were collected to detect virus titration in DF-1 cells at 5 dpi. The virus titers were determined as the TCID 50 per gram (log 10 TCID 50 g ¡1 tissue) using the endpoint method of Reed and Muench. The part of collected lymphoid tissues (spleen, thymus and bursa of Fabricius) was fixed in 10% neutral formalin, routinely sectioned and stained with hematoxylin-eosin, and then examined for lesions using Nikon light microscopy.
Immunofluorescence antibody assay DF-1 cells grown in 12-well plates were treated with the drug ivermectin or DMSO as described previously [82]. Cells were then infected with NDV strain rSS1GFP at an MOI of 0.1 and prepared for immunofluorescence analysis at 12 hpi. At the stipulated time, cells were rinsed with PBS, fixed with 4% paraformaldehyde for 20 min, and then permeabilized with 0.2% Triton X-100 in PBS for 5 min. Cells were rinsed with PBS and blocked with 10% FBS in PBS for 30 min, and then incubated with anti-M polyclonal antibodies diluted in PBS containing 10% FBS for 1 h [16]. After three washes with PBS, the cells were incubated with Alexa Fluor 488 goat anti-rabbit immunoglobulin G antibody (Invitrogen) for 1 h. Cells were counterstained with DAPI to detect nuclei. For siRNA experiments, siRNA-transfected DF-1 cells were infected with rSS1GFP at an MOI of 1, and cells were collected at 6 hpi or 12 hpi to perform immunofluorescence antibody assay as described above. Images were captured with a fluorescence microscope and processed with Adobe Photoshop 7.0 software.

siRNA treatment and virus infection
The sequences of three pairs of siRNA designed to knockdown importin b1 or importin a5 in DF-1 cells were shown in supplemental Table S1. Negative siRNA control (Cat. No.12935-400) and siRNA transfection reagent were purchased from Invitrogen. For transfection with the siRNA against importin b1 or importin a5, low-passage DF-1 cells were transfected with the indicated siRNAs at a confluence of 80% on 35 mm dishes, and the knockdown efficiency was checked by immunoblot analysis at 48 h post-transfection. To study the effect of importin b1 or importin a5 knockdown on the subcellular localization of M protein and the replication of NDV, the NDV strain rSS1GFP was used to infect importin b1 or importin a5 siRNA-treated DF-1 cells at an MOI of 1. The subcellular localization of M protein was examined by immunofluorescence assay and immunoblotting analysis and at 6 hpi and 12 hpi, respectively. In addition, the cell culture supernatants and cell pellets were collected at the indicated time points (6,12,24,48, and 72 hpi), and the virus titers were determined as 50% tissue culture infective dose (TCID 50 ) in DF-1 cells [78]. Moreover, the CPE and green fluorescence in virusinfected cells were observed under fluorescence microscope and the GFP expression level was detected by Western blotting at 24 hpi.

RNA isolation and qRT-PCR analysis
Total RNA was prepared from siRNA-treated DF-1 cells using Qiagen RNeasy kit according to the manufacturer's protocol. One microgram of total RNA was used for cDNA synthesis by reverse transcription kit (TaKaRa). For quantification of the mRNA of NDV M gene, a SYBR green-based real-time PCR method (TaKaRa) was used, and GAPDH mRNA was quantified to normalize the total RNA concentration between different samples. The primers for detecting the M gene and GAPDH gene were designed with reference to previous publication [16]. The real-time PCR operation was carried out according to the previously described method [78]. The standard cure method was used to analyze the fold change of M gene mRNA expression level. For cell pellets analysis, the NDV M gene expression levels were calculated relatively to the expression of the GAPDH gene. The relative fold change of M gene expression in the cell pellets was calculated as follows: (mRNA expressions of M gene/GAPDH in Control RNAi or Normal cells)/ (mRNA expressions of M gene/GAPDH in importin b1 or importin a5 RNAi cells). The relative fold change of M gene expression in the cell supernatants was calculated as follows: (mRNA expressions of M gene in Control RNAi or Normal cell culture)/ (mRNA expressions of M gene in importin b1 or importin a5 RNAi cell culture).

Statistical analysis
Statistical analysis was performed using the GraphPad Prism 6.0 software (GraphPad Software Inc., La Jolla, CA, USA). The p-values between identified samples were generated using unpaired two-tailed Student's t-test. All experiments were repeated at least three times and the results were presented as the mean § standard deviation (SD). The significance levels were defined as P < 0.05.

Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed.