BBRC Biochemical and Biophysical Research Communications 323 (2004) 541–546 www.elsevier.com/locate/ybbrc
Expression of the antiviral protein MxA in cells transiently perturbs endocytosisq Shashidhar S. Jatiani, Rohit Mittal* Department of Biological Sciences, Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Mumbai 400 005, India Received 4 August 2004
Abstract MxA is an interferon-induced antiviral protein. Viral replication relies on the trafficking machinery of the host cell. Overexpression of MxA was found to perturb trafficking of internalized transferrin resulting in its accumulation in cells. Interestingly, this perturbation of endocytic trafficking was transient—with a maximal effect being seen 5–6 h after transfection. By 12 h after transfection the perturbation of endocytosis was seen to have subsided although MxA protein levels remained elevated even 24 h after transfection. The accumulation of transferrin is due to a block in transferrin recycling. It is further shown that MxA can physically associate with the endocytic protein dynamin, possibly accounting for the observed effect of MxA expression on transferrin endocytosis. These results uncover a hitherto unknown aspect of MxA action on trafficking processes within cells. 2004 Elsevier Inc. All rights reserved. Keywords: MxA; Antiviral; Endocytosis; Trafficking; Transferrin; Dynamin
In cells expression of the 76 kDa GTP-binding protein MxA is induced by the action of Type I interferons a/b, thereby rendering these cells resistant to viral infection [1]. The cytoplasmic protein MxA inhibits the growth of influenza, vesicular stomatitis, measles, thogoto, bunya, phlebo, hanta, and human parainfluenza viruses in cultured cells [2]. Such interferon-mediated effects are essential for the survival of higher vertebrates because they provide an early line of defense that sets in within hours of viral infection much before immune responses are mounted [3]. The exact mechanism of action of MxA is still unclear and seems to vary depending on the nature of the infecting virus [1]. In the case of Crimean-Congo hemorrhagic fever virus it is q Abbreviations: Tfn, transferrin; A-Tfn, Alexa568-conjugated diferric transferrin; Tfn-R, transferrin-receptor; REC, recycling endosomal compartment; SH3 domain, Src homology domain 3. * Corresponding author. Fax: +91 22 2280 4610. E-mail address:
[email protected] (R. Mittal).
0006-291X/$ - see front matter 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2004.08.134
suggested that MxA binds the nuclear capsid, prevents replication of viral RNA, and thereby inhibits the production of new infectious virus particles [4]. It is known that MxA blocks a poorly defined multiplication step of influenza A virus, after primary transcription but before genome replication [5]. This block may be effected during the transport of viral mRNAs to ribosomes, during translation or during transport of newly synthesized viral proteins to the site of genome replication. It is also known in the case of double-stranded RNA viruses that viral multiplication depends heavily on the transport machinery of the host cell. Therefore, it is conceivable that specific transport processes would provide excellent targets for interference with viral life cycles [6]. The C-terminus of MxA has been suggested to play an important part in mediating the biological function of MxA because a monoclonal antibody directed against the MxA C-terminus has been shown to neutralize the antiviral activity of MxA [7]. MxA could thus be involved in interactions with host cell factors or with viral
542
S.S. Jatiani, R. Mittal / Biochemical and Biophysical Research Communications 323 (2004) 541–546
components. The dynamin proteins are important regulators of membrane trafficking and transport processes in cells. Dynamin is known to assemble in the form of a helix about the neck of vesicles and to aid in their scission from parent membranes. The Mx proteins are structurally similar to dynamin and are considered part of the dynamin superfamily of GTP-binding proteins [8]. A characteristic feature of this family of proteins is their ability to form oligomers. We propose to test the possibility that MxA may carry out its physiological function by participating in or interfering with membrane trafficking events. We chose to study the biological activity of MxA by assaying the effect it would have on a dynamin-dependent cellular process, namely, receptor-mediated endocytosis of transferrin.
Materials and methods Materials. All chemicals and reagents were obtained from Sigma (USA) and fluorescent probes were from Molecular Probes (USA), unless otherwise specified. Anti-human transferrin-receptor antibody was from Roche (Germany). Cy5 labeling reagent was from Amersham (UK). Geneticin and other media formulations, except folatefree HamÕs F-12 media (f HF-12; Hi-Media, India), were from Life Technologies (USA). All dichroic mirrors, excitation and emission filters were from Chroma Technology (USA). Polyclonal antibodies against dynamin I/II (N-19), the His-probe (H-15), and alkaline phosphatase (AP) and horseradish peroxidase (HRP)-conjugated secondary antibodies were obtained from Santa Cruz Biotechnology (USA). Western blot detection reagents, BCIP and NBT, and protein G–Sepharose beads were obtained form Life Technologies (USA) while the enhanced chemiluminescence (ECL) kit and GSH–Sepharose beads were from Amersham (USA). Ni–NTA–agarose superflow beads were obtained from Qiagen (Germany). Preparation of cDNA constructs. Hexahistidine-tagged human MxA cDNA in pQE-9 vector was obtained from Prof. Otto Haller (Freiburg, Germany) for expression in Escherichia coli. After digesting pQE9-MxA with EcoRI and HindIII, the 1990 bp MxA cDNA including the hexahistidine tag was subcloned into pBS-SK(+) plasmid (intermediate construct) from where it was removed using EcoRI and XhoI and subcloned into a similarly digested pCDNA3 plasmid. For making the pIRES2-MxA and pCDNA3-myc-MxA constructs, untagged MxA cDNA was PCR amplified and ligated into the EcoRI– SalI and EcoRI–XhoI sites of the vectors, respectively. Cell culture and transfection procedure. The Chinese hamster ovary (CHO) cell line, TRVb-1 (having the hamster Tfn-receptor knocked out and stably expressing human Tfn-receptor [9]), was maintained in folate-free HamÕs F-12 medium supplemented with 5% fetal bovine serum, streptomycin (100 lg/ml), penicillin (100 U/ml), and geneticin (100 lg/ml). Cells were plated on poly-D -lysine coated 35 mm coverslip bottom dishes 24 h before transfection and transfected using Lipofectamine Reagent (Life Technologies, USA) with 1.0 lg MxA cDNA. One to twenty-four hours post-transfection, cells were assayed for internalization of transferrin. Cell labeling and transferrin internalization assay. Alexa568-conjugated diferric Tfn (A-Tfn) was prepared as described [10] except that Alexa568 was used in place of FITC. A-Tfn was made up to 10.5 g/ml in labeling medium (f HF-12, 0.3 mg/ml NaHCO3, 5% fetal bovine serum, and 15 mM Hepes, pH 7.2). MxA overexpressing and control TRVb-1 cells were incubated with A-Tfn for indicated times at 37 C. Cells were then cooled on ice and the excess label was washed in
medium 1 (150 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, and 20 mM Hepes, pH 7.4). Uninternalized Tfn was removed from its receptor at the cell surface by multiple washes with ice-cold acid rinse buffer (25.5 mM citric acid, 24.5 mM tri-sodium citrate, 280 mM sucrose, and 10 lM deferoxamine mesylate, pH 4.0). Cells were subsequently stained for surface expression of Tfn-receptor detected via Cy5-conjugated goat anti-mouse antibody against mAb for human Tfn-receptor. Cells were fixed on ice with 2% paraformaldehyde in medium 1 for 10 min, rinsed twice in medium 1, and taken for imaging. Fluorescence imaging, quantitation, and processing. Fluorescence imaging was carried out on a wide field imaging system (Nikon Eclipse TE300 inverted microscope equipped with a digital CCD camera with a Princeton Instruments (USA) TEK-512 · 512 D-series back-illuminated chip) and mercury arc illuminators were controlled using the Metamorph software. For all experiments 20· (0.75 NA) objective (Nikon, Japan) was used. Alexa568 was visualized with 560df30 excitation and 600EFLP emission filters while Cy5 was visualized using 654df25 excitation filter and 710df75 emission filter. The FITC-Cy3Cy5 dichroic was used. Fluorescence images were processed using routines in Metamorph software following previously described procedures [10]. For quantitation of the amount of endocytosed Tfn, the internal ATfn signal from individual cells was normalized against surface expression of Tfn-receptor as the I/S ratio. In each experiment, normalized fluorescence (I/S ratio) was calculated from 8 to 12 fields from duplicate dishes for each time point and the weighted mean value and the uncertainty in the mean was determined on a per field basis considering each field (consisting of 30–40 cells) to be an independent event as described elsewhere [10]. Interaction of MxA with dynamin. CHO TRVb-1 cells were transiently transfected with pCDNA3-myc-MxA and harvested 5 h post-transfection. The cells were lysed in lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 0.02% sodium azide, 1% Triton X-100,and protease inhibitors) and total soluble protein was obtained by centrifugation at 20,000g for 30 min at 4 C. GST-amphiphysinII-SH3 (expressed using pGEX-2T-amphiphysinII-SH3 obtained from Prof. Peter S. McPherson) and GST alone were immobilized on GSH– Sepharose beads and incubated with the cell lysate for 1 h with agitation at 4 C, to pull down endogenous dynamin from the cell lysate. The beads were washed extensively with lysis buffer, boiled in SDS–PAGE sample buffer, and resolved by 10% acrylamide SDS– PAGE and the presence of MxA in the pull-downs was tested by Western blotting with polyclonal anti-MxA antibody raised in rabbit. (His)6-MxA was expressed in E. coli BL21DE3 cells transformed with pQE9-MxA and pREP4 grown at 28 C and induced with 30 lM isopropyl-b-D -thiogalactopyranoside for 2 h at an optical density of 0.3 at 600 nm. Expressed protein was purified over Ni–NTA beads, concentrated, and repurified by gel filtration on Sephadex 75. Mouse brain extract was prepared as described elsewhere [11]. Twenty micrograms of (His)6-MxA was added to dilute mouse brain extract and incubated at 4 C for 4 h with agitation. Dynamin was immunoprecipitated using an anti-dynamin I/II antibody and protein G– Sepharose beads as per standard protocols. Coimmunoprecipitated (His)6-MxA was visualized by an anti-His blot.
Results and discussion We tested the hypothesis that MxA expression might perturb intracellular trafficking along the endocytic pathway. For this purpose we employed the well-characterized endocytosis of transferrin (Tfn) in mammalian cells as our experimental system.
S.S. Jatiani, R. Mittal / Biochemical and Biophysical Research Communications 323 (2004) 541–546
543
Transfection of MxA in cells leads to a transient increase in endocytosis of transferrin To examine the effect of MxA overexpression on receptor-mediated endocytosis, Chinese hamster ovary (CHO) TRVb-1 cells were transiently transfected with pCDNA3-MxA and pCDNA3 (vector control). At different times (1–24 h) post-transfection, cells were examined for their ability to endocytose fluorescently labeled transferrin (A-Tfn). Cells were assayed for transferrin uptake by incubation with labeled transferrin for 20 min at 37 C; subsequently; the cells were stained for surface expression of human Tfn-R, fixed, and imaged. The signal for internalized transferrin was normalized against surface expression level of the transferrin-receptor and is thus reported as the I/S ratio. The histograms depicted in Fig. 1 indicate that for early time points (up to 4 h after transfection) there is no difference in the endocytic activity of MxA-expressing cells compared to control cells. Interestingly, however, at intermediate time points (5 and 6 h) posttransfection, a significant increase in the number of cells exhibiting a higher I/S ratio is observed in the case of MxA transfected cells. When endocytosis was examined 12 h after transfection with the MxA cDNA, this difference in transferrin uptake was found to be abolished. We next examined the temporal profile of MxA expression levels in transfected cells. TRVb-1 cells were transiently transfected with MxA (or empty vector as control). At different times after transfection the expression of MxA was analyzed by Western blotting. MxA expression was seen to have already set in by 3 h and remained stable for 24 h post-transfection (Fig. 2). This is consistent with other reports where MxA was seen to be
Fig. 1. MxA overexpression in cells leads to a transient increase in endocytosed transferrin. CHO TRVb-1 cells were transfected with pCDNA3-MxA or pCDNA3 vector and were assayed for endocytosis of Alexa568-transferrin at different times after transfection. A frequency distribution, i.e., number of cells versus I/S ratio, is plotted for each time point. It is seen that at the 5 and 6 h time points there is an increase in transferrin signal in cells expressing MxA.
Fig. 2. Timecourse of MxA expression. CHO TRVb-1 cells were transiently transfected with pCDNA3-MxA or vector alone. At indicated times post-transfection, the cells were harvested and whole cell lysates were prepared. The lysates were resolved by 10% SDS– PAGE, subjected to Western transfer, and probed with an anti-His antibody. (His)6-MxA expression was seen as early as 3 h posttransfection and was found to remain elevated up to the 24 h time point. A set of two lanes represents each time point where the left lane is for cells transfected with pCDNA3-MxA and the right lane for vector alone.
stably present in cells up to 3 days post-transient transfection [12]. Hence, the transience of the perturbation of endocytic activity is not regulated by expression levels or stability of the MxA protein in transfected cells. The transience may be controlled by some other means— perhaps by cellular signaling events and/or post-translational modification of MxA. MxA-mediated perturbation of endocytic activity is due to retention of internalized cargo Diferric-transferrin is internalized into cells upon binding to its cell surface receptor. The receptor–ligand complex releases its bound iron in an early endosomal compartment. Apo-transferrin remains associated with its receptor and is trafficked into sorting endosomes. The transferrin-receptor complex then accumulates in the recycling endosomal compartment en route to the cell surface. It is on returning to the neutral extracellular environment that apo-transferrin is released from its receptor and is available for another round of iron-loading and receptor binding. The kinetics of the transferrin endocytic cycle are well worked out. Entry into and exit from sorting endosomes occur rapidly with t1/2 of 3 and 2.5 min, respectively. Exit of the transferrin-receptor complex from the recycling endocytic compartment has a t1/2 of 8.5 min. Thus, the transferrin internalization–externalization cycle reaches steady state (when internalization rate equals externalization rate) well within 20 min [10]. In MxA-expressing cells, we observe an increase in cell associated transferrin signal as compared to control cells upon a 20 min incubation with fluorescently labeled transferrin. This difference could be due to two possible reasons—an enhancement in the rate of internalization of transferrin from the cell surface, or a block or misrouting in the recycling of the Tfn–Tfn-R complex back to the cell surface. To investigate whether an increase in the rate of internalization was responsible for the above effect of MxA, we examined uptake upon short incubations (1–10 minutes)
544
S.S. Jatiani, R. Mittal / Biochemical and Biophysical Research Communications 323 (2004) 541–546
Fig. 3. MxA does not affect uptake but recycling of endocytosed transferrin. CHO TRVb-1 cells transfected with pcDNA3-MxA or vector alone were assayed for transferrin endocytosis 5 h posttransfection. Transferrin endocytosis was monitored following short (A) and long (B) periods of incubation of the cells with fluorescently labeled transferrin. The shorter incubations (1–10 min) assay the effect of MxA expression on internalization of transferrin, while the longer incubations (20–120 min) assay the effect of MxA on transferrinreceptor recycling. It is thus evident that MxA perturbs recycling of endocytosed transferrin but not its uptake.
with labeled transferrin. We found no difference in the cell associated transferrin signal between MxA transfected and control cells with these short incubations (Fig. 3A). This was a clear indication that the rate of internalization of transferrin from the cell surface is not affected by MxA expression. We next examined subsequent steps of the transferrin cycle by assaying endocytosis with longer times (20–120 min) of incubation with labeled transferrin (Fig. 3B). Here, we observed a progressive accumulation of A-Tfn in MxA-expressing cells as compared to control cells. These observations indicate retention of the endocytosed material within MxA-expressing cells. Internalization of transferrin is not affected by MxA expression in cells but subsequent trafficking of the cargo is perturbed leading to its retention in cells. Such an effect of MxA expression is consistent with the biological activity of MxA in retaining viral cargo in specific locales within infected cells [7]. A more direct readout of transferrin retention in MxAexpressing cells was developed by using the bicistronic expression vector pIRES2. Transfection of a pIRES2MxA construct in CHO TRVb-1 cells leads to concurrent expression of MxA and GFP in these cells. The robust
Fig. 4. Retention of transferrin in cells expressing MxA. CHO TRVb1 cells were transiently transfected with pIRES2-EGFP-MxA. Cells were assayed for endocytosis of A-Tfn 5 h after transfection and MxAexpressing cells (as judged by the concurrent expression of GFP provided by the pIRES2 vector, indicated by arrowhead) were compared with their untransfected neighbors. Endocytosis was scored after a 60 min incubation with A-Tfn at 37 C. Scale bar = 5 lm.
GFP fluorescence serves as a good marker for MxA-expressing cells. Cells transfected with this construct were assayed 5 h after transfection for endocytosis by incubation with A-Tfn for 60 min. It is seen (Fig. 4) that MxAexpressing cells (marked with arrowheads) clearly display an increased retention of transferrin in the cytoplasm compared to neighboring cells not expressing MxA. We next attempted to establish the site of MxA action by comparing the distributions of MxA and transferrin in cells. CHO cells were assayed for endocytosis 5 h after transfection with pCDNA3-MxA. Uptake of Alexa568-conjugated transferrin was imaged after a 20-min incubation with MxA-expressing cells (Fig. 5, left panel) while MxA distribution was visualized (Fig. 5, centre panel) using the 2C12 monoclonal anti-MxA antibody [13]. It is evident that significant overlap of transferrin and MxA occurs largely in the perinuclear recycling endosomal compartment (REC). This, taken together with our observations above, suggests that MxA may be affecting the trafficking of transferrin subsequent to its arrival at the REC. These observations are consistent with earlier reports wherein MxA is seen to colocalize with viral nucleocapsid proteins in perinuclear regions of infected cells [4,14]. MxA interacts with the endocytic protein dynamin Dynamins are involved in a variety of cellular membrane trafficking events. Having seen a perturbatory
S.S. Jatiani, R. Mittal / Biochemical and Biophysical Research Communications 323 (2004) 541–546
545
Fig. 5. Colocalization of transferrin and MxA in the recycling endosomal compartment. Five hours after transfection with pCDNA3-MxA CHO cells were examined for transferrin and MxA localizations. Scale bar = 2 lm.
effect of MxA expression on the endocytosis of transferrin, we asked whether MxA could be exerting its effect by interacting with dynamin. We addressed this question by coimmunoprecipitation analysis. Mouse brain extract, used as a natural source of dynamin, was allowed to incubate with exogenously added hexahistidine-tagged MxA (expressed and purified from bacteria). Dynamin was then immunoprecipitated using an anti-dynamin antibody and protein G beads. This was followed by immunoblotting with an antibody that recognizes the hexahistidine tag to identify any co-immunoprecipitated MxA protein. Fig. 6A shows that MxA was indeed co-immunoprecipitated by the anti-dynamin antibody. The mock immunoprecipitation reaction lacking the anti-dynamin
Fig. 6. Interaction of MxA and dynamin. (A) Mouse brain cytosol was incubated with recombinant His6-MxA. Immunoprecipitation was performed using an anti-dynamin antibody immobilized on protein G beads and resolved by 10% acrylamide SDS–PAGE. Immunoblotting was carried out with anti-His antibody to check for the presence of MxA in the immunoprecipitate (indicated by the arrow). The control immunoprecipitation reaction lacked the antidynamin antibody. (B) CHO TRVb-1 cells were transiently transfected with pCDNA3-myc-MxA. Endogenous dynamin was pulled down using GST-amphiphysinII-SH3 immobilized on GSH beads (GST-Amph) and resolved by 10% acrylamide SDS–PAGE. Immunoblotting was performed with anti-MxA antibody to detect the presence of MxA in the pull down (indicated by the arrow). In the control reaction immobilized GST was used.
antibody did not display the presence of a 76 kDa band corresponding to (His)6-MxA. Control experiments were performed to verify that the anti-dynamin antibody used above did not cross-react with recombinant (His)6-MxA. This interaction is an important and striking result. It was thus confirmed using an additional readout. We examined whether MxA overexpressed in cells would be complexed with dynamin in a pull-down assay. Dynamin has a very high affinity for the SH3 domain of amphiphysin and can be isolated from cells using a GST-amphiphysin-SH3 domain fusion protein immobilized on glutathione–Sepharose beads [15]. Lysates from cells overexpressing MxA were subjected to pull down using either immobilized GST-amphiphysin-SH3 or immobilized GST (as control). As expected, immobilized GST-amphiphysin-SH3 pulled down dynamin from the cell lysates whereas GST alone did not (data not shown); significantly, an anti-MxA Western blot (Fig. 6B) showed that MxA was also pulled down along with dynamin by GST-amphiphysin-SH3. No MxA immunoreactivity was seen in material pulled down from MxAexpressing cells by immobilized GST alone. This suggests that MxA may interact with dynamin in cells. Hence, we demonstrate an interaction between MxA and dynamin using two independent methods for the isolation of dynamin. This interaction could possibly interfere with and conceivably alter transport processes that are regulated by dynamin in the cellular context. Our results indicate that expression of MxA in cells perturbs trafficking along the endocytic pathway and that it does so by possibly interacting with the protein dynamin. These results are consistent with the earlier suggestion that Mx proteins might have important functions in protein and membrane trafficking and bind transiently to cellular proteins [16,17]. How this relates to the antiviral function of MxA is not presently clear. However, trafficking along the endocytic pathway is clearly an important area for further examination in the quest for understanding the basis of the antiviral activity of MxA.
546
S.S. Jatiani, R. Mittal / Biochemical and Biophysical Research Communications 323 (2004) 541–546
Acknowledgments We acknowledge the guidance and training provided by Prof. Satyajit Mayor. We thank Prof. Otto Haller for the MxA cDNA, Prof. Jonathan Howard for the 2C12 antibody, and Prof. Peter S. McPherson for the pGEX2T-amphiphysin-SH3 expression construct. This work has been supported by grants from the Department of Science and Technology, India, and from intramural funding of the TIFR, India. References [1] O. Haller, G. Kochs, Interferon-induced Mx proteins: dynaminlike GTPases with antiviral activity, Traffic 3 (2002) 710–717. [2] S.H. Lee, S.M. Vidal, Functional diversity of Mx proteins: variations on a theme of host resistance to infection, Genome Res. 12 (2002) 527–530. [3] G.R. Stark, I.M. Kerr, B.R. Williams, R.H. Silverman, R.D. Schreiber, How cells respond to interferons, Annu. Rev. Biochem. 67 (1998) 227–264. [4] I. Andersson, L. Bladh, M. Mousavi-Jazi, K.E. Magnusson, A. Lundkvist, O. Haller, A. Mirazimi, Human MxA protein inhibits the replication of Crimean-Congo hemorrhagic fever virus, J. Virol. 78 (2004) 4323–4329. [5] J. Pavlovic, O. Haller, P. Staeheli, Human and mouse Mx proteins inhibit different steps of the influenza virus multiplication cycle, J. Virol. 66 (1992) 2564–2569. [6] G. Kochs, O. Haller, GTP-bound human MxA protein interacts with the nucleocapsids of Thogoto virus (Orthomyxoviridae), J. Biol. Chem. 274 (1999) 4370–4376. [7] G. Kochs, O. Haller, Interferon-induced human MxA GTPase blocks nuclear import of Thogoto virus nucleocapsids, Proc. Natl. Acad. Sci. USA 96 (1999) 2082–2086.
[8] G.J.K. Praefcke, H.T. McMahon, The dynamin superfamily: universal membrane tubulation and fission molecules?, Nat. Rev. Mol. Cell Biol. 5 (2004) 133–147. [9] S. Mayor, S. Sabharanjak, F.R. Maxfield, Cholesterol-dependent retention of GPI-anchored proteins in endosomes, EMBO J. 17 (1998) 4626–4638. [10] S. Mayor, J.F. Presley, F.R. Maxfield, Sorting of membrane components from endosomes and subsequent recycling to the cell surface occurs by a bulk flow process, J. Cell Biol. 121 (1993) 1257–1269. [11] V.I. Slepnev, G.C. Ochoa, M.H. Butler, D. Grabs, P.D. Camilli, Role of phosphorylation in regulation of the assembly of endocytic coat complexes, Science 281 (1998) 821–824. [12] T. Ronni, K. Melen, A. Malygin, I. Julkunen, Control of IFNinducible MxA gene expression in human cells, J. Immunol. 150 (1993) 1715–1726. [13] P. Staeheli, P. Dreiding, O. Haller, J. Lindenmann, Polyclonal and monoclonal antibodies to the interferon-inducible protein Mx of influenza virus-resistant mice, J. Biol. Chem. 260 (1985) 1821– 1825. [14] G. Kochs, C. Janzen, H. Hohenberg, O. Haller, Antivirally active MxA protein sequesters La Crosse virus nucleocapsid protein into perinuclear complexes, Proc. Natl. Acad. Sci. USA 99 (2002) 3153–3158. [15] D.J. Owen, P. Wigge, Y. Vallis, J.D.A. Moore, P.R. Evans, H.T. McMahon, Crystal structure of the amphiphysin-2 SH3 domain and its role in the prevention of dynamin ring formation, EMBO J. 17 (1998) 5273–5285. [16] M.A. Horisberger, Interferon-induced human protein MxA is a GTPase which binds transiently to cellular proteins, J. Virol. 66 (1992) 4705–4709. [17] M.A. Accola, B. Huang, A.A. Masri, M.A. McNiven, The antiviral dynamin family member, MxA, tubulates lipids and localizes to the smooth endoplasmic reticulum, J. Biol. Chem. 277 (2002) 21829–21835.