sampIc buffer, were then applied to an SDS-poly- acrylamide ge (10% to 15Wpolyacrylande gradi- ent) and subjected to electrophoresis as described (1). Electroblot transfer of the separated proteins onto Schleicher & Schull nitroceliulose paper was carried out with the use of a Bio-Rad apparatus. Blots were immersed in Denhardt's solution [0.3% Ficoll 400, 0.3% radioimmunoassay (RIA)-grade bovine serum albunmin (Sigma), and 0.3% polyvinyl- pyrollidone] for 2 hours and in 1% RIA-grade bovine serum albumin (BSA) (dissolved in Dulbec- co's phosphate-buffcred saline) for 1 hour. Rabbit antiserum to mouse cachectin was diluted 1: 200 in a solution containing 0.85% NaCI, 0.05% Tween 20, 1% RIA-grade BSA, and 0.01M tris-Cl buffer, pH 8.0 (TTB). The blot was incubated with the antiserum for 3 hours, and then rinsed three times with a solution of 0.05% Triton X-100 and 2% SDS in water. Affinity-purified goat antiserum to rabbit immunoglobulin (Miles-Yqda, Elkhart, IN) was then diluted 1:1000 in TTB, and applied to the blot fbr 3 hours. The blots were rinsed again with Triton/SDS solution and exposed to 5I-labeled protein A (New England Nuclear) diluted to a concentration of 0.15 ILCi/ml in TIMB. After 3 hours the blots were rinsed with Triton/SDS solution, then with distilled water, and were dried and used to produce autoradiograms. 24. BALB/c mice were injected intraperitoneally with 3 ml of sterile Brewer's thioglycoltate broth (Difco). After 5 days, macrophages were harvested by perito- neal lavage with sterile Hanks balanced salt solution (HBSS), washed once in the same, and plated at a confluent density in 3-cm tissue culture dishes (Bec- ton-Dickinson) with RPMI 1640 medium supple- mented with 5% fetal bovine serum (Gibco). After 1 hour, the adherent cells were washed twice with serum-free RPMI 1640 and stimulated by addition of Esberichia coi strain 0127:B8 lipopolysaccharide (LPS; Difco) (final concentration 1 I^g/ml). Control monolayers did not receive LPS. After 16 hours, the medium was aspirated for use in immunoblotting (Fig. 1B) and the cells were lysed with 5 ml of a solution containing 4M guanidinium isothiocya- nate, 5 mM sodium citrate, pH 7.0, 0.1M p- mercaptoethanol, and 0.5% Sarcosyl (Pharmacia). CsCl (2 g) was added to each sample, and total cellular RNA was isolated by centrifugation over a 5.7M CsCI cushion. RNA (approximately 2 .g per lane) was subjected to electrophoresis m a 1.2% agarose gel contamig 2.2M formaldehyde as a denaturant. Nitroccuose blots were allowed to hybridize with a nick-translated pUC-9 plasmid containing a mouse cachectin complemnentary DNA insert (specific activity = 10 dpm/,ug). Hybridiza- tion was carried out for 12 hours at 43°C in the presence of 50% formamide and 10% dextran sul- ate. Blots were then washed with two changes of 2x standard saline citrate (SSC) containing 0.1% SDS, and with two changes of 0.1 x SSC containing 0.1% SDS at 60°C, and used in autoradiography. 25. J. Weber, W. Jelinek, J. E. Darnell, Jr., Cell 10, 611 (1977). 26. We thank M. Salditt-Gcorgieff for help in perform- ing the nudear transcription assays. Supported by NIH grants AM01314 and AI21359 and RoZZ67- ler Foundation grant RF 84077. 24 October 1985; accepted 14 March 1986 Evolution of Human Influenza A Viruses over 50 Years: Rapid, Uniform Rate of Change in NS Gene DEBORAH A. BUONAGURIO, SUSUMU NAKADA, JEFFREY D. PARVIN, MARK KRYSTAL, PETER PALESE, WALTER M. FITCH Variation in influena A viruses was examined by comparison of nudeotide sequences of the NS gene (890 bases) of 15 human viruses isolated over 53 years (1933 to 1985). Changes in the genes accumulate with time, and an evolutionary tree based on the maximum parsimony method can be constructed. The evolutionary rate is approxi- mately 2 x 10-3 substitution per site per year in the NS genes, which is about 106 times the evolutionary rate of germline genes in mammals. This uniform and rapid rate of evolution in the NS gene is a good molecular clock and is compatible with the hypothesis that positive selection is operating on the hemagglutinin (or perhaps some other viral genes) to preserve random mutations in the NS gene. I NFLUENZA A VIRUSES HAVE A SINGLE- stranded RNA genome of eight seg- ments of negative polarity, with the shortest segment coding for the nonstruc- tural proteins (NS1 and NS2) (1). Figure 1 shows the nucleotide sequences of the NS genes of 15 human influenza A virus strains. The viruses were isolated over a 53-year period and represent all three human he- magglutinin serotypes (Hl, H2, and H3). Except for the three Houston isolates, the strains were obtained from diverse geo- graphical locations. The 15 sequences are easily aligned for analysis because of the size conservation of the NS gene segment of 890 bases. Nudeotide substitutions occur at 149 positions scattered throughout the gene and usually, once a base change is observed in a virus isolate, it is found in subsequent strains. The sequence information as presented in Fig. 1 was analyzed by maximum parsimony (2) to determine the phylogenetic tree of miniimum length. The best tree found con- tains a total of 186 substitutions and is illustrated in Fig. 2. The parsimony method also yielded four alternative trees containing 187 substitutions. These alternative trees contain only minor branch perturbations of the best tree. Figure 3 shows the number of nudeotide substitutions between the origin of the best tree and the tip of each branch (Fig. 2) plotted against the date of isolation of the viruses whose NS gene is represented by that tip. The major line, derived by linear regression analysis, shows that these se- quences are evolving at the steady rate of 1.73 ± 0.08 nucleotide substitutions per year, or 1.94 ± 0.09 x 10-3 substitution per nucleotide site per year. The WSN/33 and PR/34 strains appear to have more substitutions per year than expected and therefore were excluded from the evolution- ary rate calculation. Since these strains were isolated before refrigeration became avail- able in the laboratory, we believe that con- tinuous passaging in animal hosts and in embryonated eggs (particularly in the first 10 to 15 years after isolation of the strains) may have introduced additional mutations not present in the original isolates. Figure 3 also shows that the group of HlNl subtype strains, which reemerged in the human pop- ulation in 1977 and after a 27-year absence (3), is evolving at the same rate. These "new" HlNl viruses have been cocirculat- ing with the H3N2 viruses since 1977 and form a separate evolutionary branch (Fig. 2). In reality, the HlNl branch should be directly connected to the FW/50 branch of the main tree, since there are only five nudeotide differences between the FW/50 and USSR/77 virus NS genes. However, the viruses were isolated 27 years apart and, on the basis of the calculated evolutionary rate of 1.73 substitutions per year, we would predict approximately 46 additional substi- tutions in the NS gene of USSR/77 (repre- sented by the broken line in Fig. 2). The observed data thus suggest a unique epide- miology of the new HINM isolates. Several points can be made from the analysis of the data. First, calibration of the molecular clock is not affected by inaccurate paleontological dates, since the time of fos- silization (isolation) of these strains is re- corded. This may partly explain why the NS gene of influenza A viruses behaves as an accurate molecular clock (4). Thus, given only the NS gene sequence of a main line isolate, one can closely estimate the year of its isolation (Fig. 3). Although fewer points are available for measuring the rate in the new HlNl strains (1977 to 1985), the data (filled squares in Fig. 3) are compatible with a molecular clock ticking at the same evolu- tionary rate for these NS genes. The muta- tions seen in the NS genes of the new HlNl strains (1977 to 1985) are different from those seen in the 1950-1957 HlNl strains. The second point that can be made is that D. A. Buonagurio, S. Nakada, J. D. Parvin, M. Krystal, P. Palese, Department of Microbiology, Mount Sinai School of Medice, City University ofNew York, New York, NY 10029. W. M. Fitch, Departnent of Physiological Chemistry, University of Wisconsin, Center for Health Sciences, Madison, WI 53706. SCIENCE, VOL. 232 980 on February 1, 2015 www.sciencemag.org Downloaded from on February 1, 2015 www.sciencemag.org Downloaded from on February 1, 2015 www.sciencemag.org Downloaded from