W.G. Laver

Morphology of the isolated hemagglutinin and neuraminidase subunits of influenza virus

The BEL (A0) strain of influenza virus was disrupted with sodium dodecyl sulphate (SDS), and the hemagglutinin subunits were isolated by electrophoresis on cellulose acetate strips. Neuraminidase subunits were isolated from an A0-A2 recombinant influenza virus (X-7F1) in a similar way. Electron micrographs of the hemagglutinin subunits (in the presence of SDS) showed rods approximately 40 A wide and 140 A long. The neuraminidase subunits had a quite different appearance. In the presence of SDS, these were seen as oblong structures about 85 A long and 50 A wide with a centrally attached fibre 100 A long possessing what appeared to be a diffuse tail or small knob about 40 A in diameter at its end. When SDS was removed, the subunits aggregated. The hemagglutinin formed clusters of radiating rods; the neuraminidase subunits aggregated by the tips of their tails and formed structures similar in appearance to the heads of seeding dandelions. When SDS was removed from mixtures of the subunits, mixed clusters containing both kinds of subunits were formed. In the presence of SDS, the hemagglutinin subunits had a sedimentation coefficient of 7.5 S. They adsorbed to, but did not agglutinate red cells and thus appeared to be “monovalent.” Hemagglutinin activity was obtained only after removal of the SDS, and it is assumed that this activity was associated with the aggregates of rods seen in electron micrographs. The neuraminidase subunits in the presence of SDS sedimented faster (8.5 S) than the hemagglutinin subunits and possessed enzyme activity both before and after removal of the SDS. Both the hemagglutinin and neuraminidase, after the removal of SDS, sedimented more rapidly and as broader bands than in the presence of SDS, suggesting that they had formed aggregates—thus confirming the electron microscopic observations. These results suggest that the “spikes” seen on the surface of influenza virus particles are of two morphologically distinct kinds, one of which is associated with neuraminidase activity, and the other with the hemagglutinin.

Refined crystal structure of the influenza virus N9 neuraminidase-NC41 Fab complex

The crystal structure of the complex between neuraminidase from influenza virus (subtype N9 and isolated from an avian source) and the antigen-binding fragment (Fab) of monoclonal antibody NC41 has been refined by both least-squares and simulated annealing methods to an R-factor of 0.191 using 31,846 diffraction data in the resolution range 8.0 to 2.5 A. The resulting model has a root-mean-square deviation from ideal bond-length of 0.016 A. One fourth of the tetrameric complex comprises the crystallographic model, which has 6577 non-hydrogen atoms and consists of 389 protein residues and eight carbohydrate residues in the neuraminidase, 214 residues in the Fab light chain, and 221 residues in the heavy chain. One putative Ca ion buried in the neuraminidase, and 73 water molecules, are also included. A remarkable shape complementarity exists between the interacting surfaces of the antigen and the antibody, although the packing density of atoms at the interface is somewhat looser than in the interior of a protein. Similarly, there is a high degree of chemical complementarity between the antigen and antibody, mediated by one buried salt-link, two solvated salt-links and 12 hydrogen bonds. The antibody-binding site on neuraminidase is discontinuous and comprises five chain segments and 19 residues in contact, whilst 33 neuraminidase residues in eight segments have 899 A2 of surface area buried by the interaction (to a 1.7 A probe), including two hexose units. Seventeen residues in NC41 Fab lying in five of the six complementarity determining regions (CDRs) make contact with the neuraminidase and 36 antibody residues in seven segments have 916 A2 of buried surface area. The interface is more extensive than those of the three lysozyme-Fab complexes whose crystal structures have been determined, as judged by buried surface area and numbers of contact residues. There are only small differences (less than 1.5 A) between the complexed and uncomplexed neuraminidase structures and, at this resolution and accuracy, those differences are not unequivocal. The main-chain conformations of five of the CDRs follow the predicted canonical structures. The interface between the variable domains of the light and heavy chains is not as extensive as in other Fabs, due to less CDR-CDR interaction in NC41. The first CDR on the NC41 Fab light chain is positioned so that it could sterically hinder the approach of small as well as large substrates to the neuraminidase active-site pocket, suggesting a possible mechanism for the observed inhibition of enzyme activity by the antibody.(ABSTRACT TRUNCATED AT 400 WORDS)

Determination of the number of nonoverlapping antigenic areas on Hong Kong (H3N2) influenza virus hemagglutinin with monoclonal antibodies and the selection of variants with potential epidemiological significance

Abstract Monoclonal antibodies provided evidence for at least three nonoverlapping antigenic areas on the hemagglutinin molecule of A/Mem/V71 (H3N2) influenza virus. This was established by determining the reactivity patterns of 30 different monoclonal antibodies in hemagglutination-inhibition tests and by the failure to select antigenic variants of influenza virus when monoclonal antibodies from two nonoverlapping areas were used in combination. Antigenic analysis showed that most of the variants selected with monoclonal antibodies could not be distinguished from the parental virus with heterogeneous sera, suggesting that they are probably epidemiologically irrelevant. One variant, however, could be distinguished from the parental virus with heterogeneous sera and this variant had a change in sequence at residue 144 of the HAl polypeptide, from glycine in the parent to aspartic acid in the variant. A similar amino acid change has been found in naturally occurring variants at this residue. These studies suggest that some amino acid substitutions are more important than others for producing viruses with epidemiological potential. Antigenic analysis of naturally occurring H3N2 strains with monoclonal antibodies showed that antigenic variation occurs in each nonoverlapping antigenic area of the HA molecule and established that two distinct variants cocirculated in 1968, Hong Kong/1/68 being distinguishable from Aichi/2/68 in at least two antigenic areas. It appears that there may have been two separate lineages of H3N2 viruses, Hong Kong/1/68 giving rise to variants in England and Aichi/2/68 to variants in the USA and Australia.

Studies on the origin of pandemic influenza. III. Evidence implicating duck and equine influenza viruses as possible progenitors of the Hong Kong strain of human influenza.

Abstract Two strains of influenza virus isolated from horses and ducks in 1963, A/equine/Miami/1/63 (Heq2 Neq2) and A/duck/Ukraine/1/63 (Hav7 Neq2) were found to possess hemagglutinin subunits which cross-reacted in hemagglutination-inhibition and immunodiffusion tests with those of the Hong Kong strain of human influenza A/Hong Kong/1/68 (H3 N2). Peptide maps of the heavy polypeptide chains from the hemagglutinin subunits of these three strains showed a number of differences, but maps of the light chains were almost identical, indicating that the light polypeptide chains from the hemagglutinin subunits of these animal, avian and human viruses had practically the same amino acid sequence. One explanation of these results is that the three viruses arose, by genetic recombination, from a common ancestor.

Structural studies on the protein subunits from three strains of influenza virus

Particles of the MEL and BEL strains of influenza A virus were disrupted with SDS† and a clear-cut separation of three different protein fractions was obtained by electrophoresis. Fraction 1 from BEL virus contains a protein which agglutinates red cells. Fraction 1 from MEL virus has no hemagglutinin activity, but a protein in this fraction adsorbs to red cells and is assumed to be responsible for the haemagglu-tinin activity of the intact virus. Fraction 1 from each strain contains 37% of the viral protein and possesses a free N-terrninal aspartic acid residue, together with smaller amounts of N-terminal glycine. No free N-terminal amino acids were found in the other two protein fractions. The isolated haemagglutinin from BEL virus has a sedimentation coefficient of about 15 s, but chemical evidence suggests that this subunit is composed of several polypeptide chains. The fraction containing the haemagglutinin and one of the other proteins (fraction 3) from MEL and BEL viruses were digested with trypsin; the resulting peptides were mapped and some were analysed for arginine and lysine. With a single exception, the peptides from fraction 3 of MEL virus appear to be identical to the peptides from the corresponding protein of BEL, while peptides from the haemagglutinin fraction show many differences between the two strains. It is suggested that fraction 3 (which contains 38% of the viral protein) is the common internal antigen of these viruses, and that some of the differences in amino acid sequence observed in the haemagglutinin fraction are responsible for the immunological differences which exist between the external antigens of these two strains. Fraction 2 appears to contain a number of protein components, but these have not been characterized. Particles of influenza B virus (LEE strain), disrupted with SDS, were separated into three protein fractions by electrophoresis. None of these adsorbed to red cells, but one contained all the viral neuraminidase. N-terminal aspartic acid was found associated only with fraction 2, which was enzymically inactive.

Separation of two polypeptide chains from the hemagglutinin subunit of influenza virus

Abstract Particles of the BEL(Ao) strain of influenza virus (grown in the allantoic sac of embryonated chicken eggs) were disrupted with cold sodium dodecyl sulphate (SDS), and the hemagglutinin subunits were isolated by electrophoresis on cellulose acetate strips. The hemagglutinin subunits were found to contain two different polypeptide chains with molecular weights of approximately 60,000 and 21,000. In the intact subunits the heavy chain was joined by disulphide bond(s) to the light chain to form a dimer, and each hemagglutinin subunit contained two of these dimers. The two chains were separated by SDS-polyacrylamide gel electrophoresis in the presence of dithiothreitol or, on a preparative scale, by centrifugation on a guanidine hydrochloridedithiothreitol density gradient. The two chains were similar in amino acid composition, with the exception that the heavy chain contained about 9 times more proline than the light chain. The heavy chain also contained much more glucosamine. Maps of the tryptic peptides from the two chains were quite different, indicating that they differed in amino acid sequence. The other protein components of the BEL(Ao) influenza virus particle (neuraminidase, internal ribonucleoprotein antigen, and a small internal protein) were each associated with a different polypeptide. Thus particles of BEL(Ao) influenza virus contain at least 5 different polypeptide chains.

Antigenic drift in type A influenza virus: sequence differences in the hemagglutinin of Hong Kong (H3N2) variants selected with monoclonal hybridoma antibodies.

Abstract Three monoclonal hybridoma antibodies, designated H14/A2, H14/A20, and H14/A21, which appeared to bind to different sites on the hemagglutinin of A/Mem/1/71 (H3N2) virus, were used to select a total of 10 antigenic variants of this virus. The variants occurred with a frequency of about 1 per 100,000 infectious wild-type virus particles and were isolated after a single passage of the virus in the presence of the monoclonal antibody. Hyperimmune rabbit antisera did not distinguish wild-type and variant viruses, but the monoclonal antibodies, which reacted with the wild-type virus to titers of the order of 1 100,000, did not react at all (or to very low titer) with thevariants they selected. This dramatic change in antigenicity appeared to be associated with a single change in the amino acid sequence of the large hemagglutinin polypeptide, HA1. Four variants selected with H14/A2 monoclonal antibod showed the same antigenic properties and the same sequence change (asparagine to lysine) in the N-terminal half of HA1. Of three variants selected with H14/A20 monoclonal antibody, two showed a different change at a locus also in the N-terminal region of HA1 (a proline was replaced by serine in one variant and by leucine in the other). Of the other three variants (selected with H14/A21 monoclonal antibody) one showed a change in HA1 of serine to tyrosine. This change occurred in residue number 37 of cyanogen bromide fragment 2 (CN2). In the other two variants the change in HA1 has not been determined, but in these a tryptic peptide comprising residues 49–56 of CN2 was missing. No changes were found in the HA2 polypeptide from any of the variants.

The structure of influenza viruses: IV. chemical studies of the host antigen

Antiserum to uninfected chick chorioallantoic membranes inhibited the activity of the hemagglutinin isolated from influenza viruses grown in the allantoic sac of chick embryos. Antiserum to uninfected duck chorioallantoic membranes or uninfected cultures of calf kidney cells did not inhibit the activity of the hemagglutinin isolated from, influenza virus grown in these hosts. This inhibition of viral hemagglutinin by anti-host antibody was blocked by extracts of uninfected chick chorioallantoic membranes and by a heat-stable, low molecular weight substance isolated from these membranes and from purified virus. This substance is referred to as the host antigen. Analysis of the host antigen showed that it consisted predominantly of carbohydrate. It was bound covalently to the coat proteins of the virus, but was not found to be associated with the internal antigen. The host antigen was 5% or less of the dry weight of the virus. Peptide maps of the internal antigen and of the hemagglutinin of the BEL strain of influenza A virus, grown in chick embryos, were similar to the peptide maps of these proteins from virus grown in calf kidney cells. Peptide maps of one of the minor protein components of the virus (of unknown function) revealed some differences between the viruses from the two hosts. It was concluded that most of the protein in the virus was coded for by the viral genome, and that the preparations contained small amounts only of host proteins.

Purification and properties of chick embryo lethal orphan virus (an avian adenovirus)

Chick embryo lethal orphan (CELO) virus, an oncogenic avian adenovirus, and one of its antigens (the hexon) were purified from infected allantoic fluid and some of their properties were studied. Particles of CELO virus contained 80.7% protein and 17.3% DNA and had a buoyant density of 1.35 g cm−3 in CsCl. CELO virus DNA was a linear double-stranded molecule with a molecular weight of 30 × 106 daltons (greater than that of human adenovirus DNA) and a G + C content of 54%. Particles of CELO virus had typical adenovirus morphology and the hexons appeared to be similar to those of adenovirus type 2. CELO virus pentons, however, differed remarkably from those of other adenoviruses. Pentons isolated from the soluble antigen fraction of infected allantoic fluid, or dissociated by dialysis from particles of CELO virus possessed a base which carried two knobbed fibres, one long (425 A) and the other short (85 A). Besides the hexon and penton antigens, CELO virus particles possessed another antigen which was probably one of the internal proteins. The amino acid composition of CELO virus hexon was similar to that of adenovirus type 2 hexon but these two hexons were completely unrelated immunologically. SDS-polyacrylamide gel electrophoresis showed that CELO virus particles contained fewer internal proteins than particles of adenovirus type 2. CELO virus particles, disrupted in various ways, did not show hexons arranged in groups of nine characteristic of adenovirus types 2 and 5.

Refined atomic structures of N9 subtype influenza virus neuraminidase and escape mutants.

The crystal structure of the N9 subtype neuraminidase of influenza virus was refined by simulated annealing and conventional techniques to an R-factor of 0.172 for data in the resolution range 6.0 to 2.2 A. The r.m.s. deviation from ideal values of bond lengths is 0.014 A. The structure is similar to that of N2 subtype neuraminidase both in secondary structure elements and in their connections. The three-dimensional structures of several escape mutants of neuraminidase, selected with antineuraminidase monoclonal antibodies, are also reported. In every case, structural changes associated with the point mutation are confined to the mutation site or to residues that are spatially immediately adjacent to it. The failure of antisera to cross-react between N2 and N9 subtypes may be correlated with the absence of conserved, contiguous surface structures of area 700 A2 or more.