Journal of General Virology (1997), 78, 1589���1596. Printed in Great Britain ........................................................................................................................................................................................................................................................................................... Influenza virus M1 protein binds to RNA through its nuclear localization signal Christine Elster,1��� Kjeld Larsen,1 Jean Gagnon,2 Rob W. H. Ruigrok1 and Florence Baudin1 1 EMBL Grenoble Outstation, c/o ILL, BP 156, 38042 Grenoble cedex 9, France 2 Institut de Biologie Structurale, CEA/CNRS, 41 Avenue des Martyrs, 38027 Grenoble cedex 1, France The RNA-binding activity of influenza A virus M1 protein was studied by cross-linking the protein to viral RNA followed by sequence analysis of the oligoribonucleotide bound to the protein as well as sequence analysis of the M1 peptide bound to the RNA. M1 was found to bind to RNA without any RNA sequence specificity, as verified in a series of filter- binding experiments using a large variety of nucleic acids including DNA. The peptide sequence that bound to the RNA was the RKLKR nuclear Introduction Influenza virus is an enveloped virus with a lipid membrane that separates the outside of the virus (membrane-embedded glycoprotein spikes which are responsible for attachment, penetration and virus release) from the inside of the virus, which contains the nucleocapsid this consists of eight distinct negative-strand RNA segments that are associated with nucleoprotein (NP) and an RNA polymerase complex (the complex of RNA, NP and polymerase is called RNP). In the presence of the four nucleotide triphosphates and a primer these RNPs are transcriptionally active in vitro. Separating the viral core from the membrane is a layer of M1 protein. M1 is thought to add to the structural integrity of the virus particle. In negatively stained virus, where stain has penetrated the lipid bilayer, M1 can be observed, aligned in rows resembling a finger-print (Ruigrok et al., 1989). M1 is generally believed to make contact with the lipid membrane (Bucher et al., 1980 Gregoriades & Frangione, 1981) and with the cytoplasmic tails of the two glycoproteins. However, experiments aiming to prove this interaction in vivo, using virus-infected cells or Author for correspondence: Florence Baudin. Fax ���33 4 76 20 71 99. e-mail Baudin!embl-grenoble.fr ��� Present address: Division of Experimental Cell Research and Oncology, Institute of Pathology, University of Graz, Auenbruggerplatz 25, 8036 Graz, Austria. localization signal of M1. Site-specific mutagenesis of recombinant M1 showed that most of the basic residues in that region had to be mutated in order to inhibit RNA-binding. We also constructed an M1 mutant that no longer bound to RNA but which was still able to inhibit the in vitro transcription activity of isolated viral ribonucleoprotein, albeit to a lower extent. Mutation of the zinc-binding sequence had no effect on RNA-binding or transcription-inhibition activity. expression of viral proteins in eukaryotic systems, have so far led to conflicting interpretations (Enami & Enami, 1996 Jin et al., 1994 Kretzchmar et al., 1996 Zhang & Lamb, 1996). Suggestions for the interaction of M1 with RNPs in vivo comes from two lines of research into two separate steps in the infection process. The first step where this interaction is revealed is during cell entry by the endosomal uptake process. In the acidic endosomes the viral membrane fuses with the endosomal membrane and during this process the interior of the virus is also supposed to be acidified through the action of the viral M2 membrane channel (Zebedee & Lamb, 1988 Hay, 1989 Wharton et al., 1990 Pinto et al., 1992). Blocking of the M2 ion channel with drugs like amantadine or rimantadine is believed to inhibit the disassembly process of M1 from RNP which inhibits subsequent virus replication (Bukrinskaya et al., 1982 Martin & Helenius, 1991b). Further, when newly expressed M1 is present in the cytoplasm, it interferes with the nuclear uptake of the RNPs, an activity that can be overcome by the temporary acidification of the cytoplasm (Bui et al., 1996). Later in infection, a second activity of M1 protein is needed in the nucleus of the infected cell in order to allow the export of newly synthesized RNPs from the nucleus to the cytoplasm for virion assembly (Martin & Helenius, 1991a Rey & Nayak, 1992 Whittaker et al., 1995), although it is not yet clear whether this export is mediated by a direct interaction of M1 with RNP. At this later stage of infection, newly expressed cytoplasmic M1 inhibits nuclear re-import of freshly made 0001-4664 # 1997 SGM BFIJ

C. Elster and others C. Elster and others RNPs, allowing these RNPs to be taken up into budding virions (Bui et al., 1996 Whittaker et al., 1996). Evidence for in vitro interaction of M1 with RNP comes from transcription-inhibition when purified M1 is added to transcribing RNPs (Zvonarjev & Ghendon, 1980 Ye et al., 1987, 1989 Hankins et al., 1990 Elster et al., 1994) and in vitro interaction of M1 with naked RNA has been shown by filter- binding assays and blotting techniques (Ye et al., 1989 Wakefield & Brownlee, 1989 Elster et al., 1994). It has been reported that M1 protein has two RNA-binding sites extending from amino acids 90 to 108 and from 135 to 164 (Wakefield & Brownlee, 1989 Ye et al., 1989). This first site contains an RKLKR sequence, residues 101 to 105, which was shown to be the nuclear localization signal (NLS) of M1 (Ye et al., 1995). Recently, using deletion mutants of recombinant M1, the residues extending from 91 to 111 were shown to be essential for RNA-binding activity and oligomerization of M1 on RNA (Watanabe et al., 1996). The experiments reported here were undertaken to identify the amino acids of M1 that bind to RNA. We used two cross- linkers, trans-diamminedichloroplatinum(II) (trans-DDP) and 4- azido-phenyl-glyoxal (APG), to induce RNA���protein cross- links, and we determined both the RNA sequence(s) and the peptide sequence involved in the M1���RNA cross-link. M1 was found to bind to vRNA without any sequence specificity. However, only one peptide of M1 protein was found cross- linked to RNA, amino acids residues 95 to 108. This peptide is rich in basic residues and contains the above mentioned NLS sequence "!"RKLKR"!& (Ye et al., 1995). Substitution of all four Arg and Lys to Ala led to a total loss of RNA-binding activity but only to a partial loss in transcription-inhibition activity. Methods + Viral M1 protein and vRNA. Egg-grown influenza A}PR}8}34 virus was obtained from Pasteur-Merieux, Marcy l���Etoile, France. The viral glycoproteins were removed by bromelain digestion which was stopped by addition of 100 mM iodoacetamide. Spikeless virus was purified by pelleting through 14% sucrose in PBS (150 mM NaCl 10 mM phosphate buffer pH 7���2 0���01% sodium azide). The virus was then disrupted with 1% Triton X-100 in PBS and centrifuged over a 10 to 30% continuous glycerol gradient in PBS (SW41 rotor, 36000 r.p.m., 4 C, 16 h). Pure M1 was collected from the upper fractions of the gradient as described in Elster et al. (1994). Segment 8 viral RNA (NS gene) was produced by in vitro transcription run-off synthesis as described in Baudin et al. (1994). + Recombinant M1 and mutagenesis. The M1 protein gene of influenza A}PR}8}34 virus cloned in a pAS1 vector was kindly donated by M. Krystal, Bristol-Myers Squibb, Wallingford, Conn., USA. The gene was subcloned into pET-16b and site-directed mutagenesis was per- formed on this construct using PCR (Ex-Site kit, Stratagene). The synthetic oligonucleotide primers containing the mutation were placed back to back on the template, with one of the oligonucleotides being 5�� phosphorylated. The whole plasmid was amplified leading to linear double-stranded DNA. The parental DNA was digested using DpnI, and the remaining PCR product was self-ligated and transformed into E. coli. Each construct was checked by sequencing. E. coli (BL21}DE3}pLysS) was transformed with each mutated plasmid. A 1 litre Luria broth culture was induced with 1 mM IPTG for M1 production. After 4 h induction, the cells were harvested by centrifugation and resuspended in ice-cold buffer (5 mM imidazole, 500 mM NaCl, 20 mM Tris���HCl pH 7���9). The cells were sonicated and M1 protein purified using an Ni-chelation resin according to the manufacturer (Novagen). After protein elution, each fraction was analysed by electrophoresis on 15% SDS���PAGE. The fractions con- taining M1 were slowly dialysed against 20 mM Tris���HCl pH 6���5, 200 mM NaCl and 1 mM DTT. + Cross-linking conditions with trans-DDP and vRNA frag- ment analysis. Viral M1 (2 ��M) was incubated with a 2-fold molar excess of viral RNA segment 8 and cross-linking was performed in 20 mM sodium phosphate buffer (pH 7���4), 1���5 mM magnesium acetate, 100 mM potassium acetate, 10% glycerol, 0���1% lubrol and 0���2 mM trans-DDP (Sigma) for 1 h at 20 C in the dark. About 30 to 40% of the RNA was cross-linked, as checked by a filter-binding assay using 5�� radioactively end-labelled RNA. The excess of cross-linking agent was removed using a speedy desalting column (Pierce). The cross-linked vRNA���M1 complexes were then adjusted to 0���3 M sodium acetate and precipitated with 3 vols of ethanol. The complexes were then subjected to RNase T1 digestion (1���5 U}��g RNA, 30 min at 37 C in 20 mM sodium phosphate pH 7���5, 1 mM EDTA and 2 M potassium acetate). The resulting covalent M1���oligoribonucleotide complexes were separated from non-cross-linked oligoribonucleotides on nitrocellulose filters (Millipore type HA, 45 ��m pore size, 25 mm diameter, previously soaked in the above buffer), which were then extensively washed with 20 ml of the same buffer. After filtration, the nitrocellulose filters were soaked in 500 ��l 2 M thiourea in order to reverse the cross-links. Supernatants were removed and the filters washed with 200 ��l 2 M thiourea. The liberated vRNA fragments were precipitated with ethanol and then labelled at their 5�� end with T4 polynucleotide kinase and 100 ��Ci [��-$#P]ATP according to Silberklang et al. (1977). The 5��-labelled fragments were fractionated by electrophoresis on a 20% polyacrylamide���8 M urea gel. After autoradiography, the fragments were excised, eluted according to Maxam & Gilbert (1977), and repurified on a 22% polyacrylamide���8 M urea gel. The fragments were then eluted, precipitated with ethanol in the presence of 10 ��g tRNA as carrier, dissolved in 10 ��l water, incubated for 5 min at 55 C and sequenced using several ribonucleases. For each sequence lane between 20000 and 50000 c.p.m. was used. Digestion was done with RNase T1 (0���25 U) for G RNase U2 (0���25 U) for A RNase PhyM (0���5 U) for A and U and B. cereus RNase (0���5 U) for C and U. Incubation was at 55 C for 15 min in 20 mM citrate buffer pH 4���5, 1 mM EDTA in the presence of 8 M urea for RNases T1, U2, PhyM and in the absence of urea for B. cereus RNase. The nucleotide ladder was made in 4 ��l deionized formamide at 90 C for 30 min. Analysis of the digests was carried out by electrophoresis on 15% polyacrylamide���8 M urea gel. + Cross-linking conditions with APG and M1 peptide analysis. M1 protein (1 mg) was incubated with a 2-fold excess of an RNA oligoribonucleotide (5�� AGUAGAAACAAGGGUG 3��) in PBS, 10% glycerol, 0���1% lubrol for 1 h at room temperature and cross-linked using a final concentration of 1 mg}ml APG (Aldrich) (Sgro et al., 1986). APG was solubilized in PBS at 65 C at a concentration of 10 mg}ml. After reaction of M1 with RNA and APG, excess APG was removed using a speedy desalting column. The second step of the reaction was then carried out by incubating the mixture for 90 min under UV light. The cross-linked fraction was irradiated using a 40% Co(NO $ ) # solution as a filter for cutting off the light below 300 nm. The cross-linked complex was separated from non-cross-linked material on an HPLC ��Bondapack column, using a gradient from 0 to 50% acetonitrile in 0���1% trifluoroacetic acid. Collected peaks were dried and submitted to endoprotease Asp-N BFJA