The J.I.M. interview. Purnell W. Choppin, MD.

The Howard Hughes Medical Institute (HHMI) is known worldwide for its support of high calibre research in the basic biomedical sciences. At present, 332 men and women at 72 institutions are HHMI investigators, putting them in the ranks of the best and the brightest (and certainly the most generously supported) researchers in America. But the Howard Hughes Medical Institute's portfolio goes well beyond its core program as a medical research organization. Purnell W. Choppin, MD, discussed the breadth of Hughes' activities during a recent interview in his office at the Institute's main campus in Chevy Chase, Maryland, a few blocks away from the National Institutes of Health. Among other things, Choppin talked about US medical schools, clinical research, postgraduate education, and HHMI support of research abroad.

Control of virus-induced cell fusion by host cell lipid composition.

Virus-induced cell fusion has been examined in a series of stable cell lines which were originally selected for resistance to the fusogenic effects of polyethylene glycol (PEG). For a wide variety of viruses, including murine hepatitis virus (a coronavirus), vesicular stomatitis virus (a rhabdovirus), and two paramyxoviruses (Sendai virus and SV5), susceptibility to virus-induced fusion was found to be inversely correlated with susceptibility to PEG-induced fusion. This phenomenon was observed both for cell fusion occurring in the course of viral infection and for fusion induced "from without" by the addition of high titers of noninfectious or inactivated virus. The fusion-altered cell lines (fusible by virus but not by PEG) are characterized by their unusual lipid composition, including marked elevation of saturated fatty acids and the presence of an unusual ether-linked neutral lipid. To test the association between lipid composition and fusion, acyl chain saturation was manipulated by supplementing the culture medium with exogenous fatty acids. In such experiments, it was possible to control the responses of these cells to both viral and chemical fusogens. Increasing the cellular content of saturated fatty acyl chains increased the susceptibility of cells to viral fusion and decreased susceptibility to PEG-induced fusion, whereas lowering fatty acid saturation had the opposite effect. Thus, parallel cultures of cells can be either driven toward the PEG-fusible/virus-fusion-resistant phenotype of the parental cells or rendered susceptible to viral fusion but resistant to PEG-induced fusion, solely by the alteration of cellular lipids. The ability of cellular lipid composition to regulate virus-induced membrane fusion suggests a possible role for lipids in viral infection and pathogenesis.

Mice immunized with measles virus develop antibodies to a cell surface receptor for binding virus.

Mice were immunized with measles virus to determine whether an auto-anti-idiotypic antireceptor response could be generated as a probe for measles virus receptors. Mice initially responded to viral antigens (days 11 to 18) and subsequently developed antibodies to a putative measles virus receptor (peak at day 30 to 35) by three criteria: the sera (1) agglutinated erythrocytes which virus agglutinates, (2) reacted with Vero cells, and (3) inhibited virus attachment to Vero cells. Additionally, select sera inhibited virus infection of Vero cells. The cell-reactive activity was identified as immunoglobulin G antibody and was neutralized by sera reacting with virus (idiotype). The application of this anti-idiotypic antibody to identify measles virus-binding sites on Vero cells was revealed by the ability of sera to immunoprecipitate 20- and 30.5-kilodalton proteins from metabolically labeled ([35S]methionine) Vero cells.

Selective inhibition of WSN influenza virus haemolysis by pea lectin.

The capacity of lectins to inhibit viral haemolysis of chicken erythrocytes was tested, to evaluate the role of carbohydrate in the fusion reaction. Pretreatment of cells with pea lectin provided a 70% to 85% haemolysis inhibition with WSN influenza virus, but only 10% to 14% with PR8 influenza virus. Pea lectin did not detectably bind to virus, nor did it inhibit virus binding to cells, but it did inhibit WSN influenza virus elution. Additionally, pea lectin was active against Sendai virus and B/Lee influenza virus, but inactive against Newcastle disease virus. Haemolysis by WSN and PR8 influenza viruses was unaffected in cells pretreated with concanavalin A, peanut, wheatgerm or soybean lectins. A possible role of cellular carbohydrate in virus-cell fusion is discussed.

Resistance of a measles virus mutant to fusion inhibitory oligopeptides is not associated with mutations in the fusion peptide.

The nucleotide sequence and predicted amino acid sequence has been obtained for the fusion (F) protein gene of the R93 strain of measles virus and compared to that of the parental strain, Edmonston B. The R93 strain is a mutant measles virus which is able to grow and induce cell fusion in the presence of the fusion inhibiting oligopeptide, Z-D-Phe-L-Phe-L-(NO2)Arg (SV4814). Primer extension sequencing on isolated R93 mRNA demonstrated the presence of three nucleotide changes leading to three amino acid changes, none of which are in the hydrophobic NH2-terminal region of the F1 polypeptide.

Segment-specific and common nucleotide sequences in the noncoding regions of influenza B virus genome RNAs.

The nucleotide sequences of the 3' noncoding regions of all eight segments of influenza B virus RNA and the sequences of the 5' noncoding regions of segments 4-8 were determined in virus strains isolated over a period of 40 years. Nearly complete conservation of the noncoding sequences was found. Nine nucleotides at the 3' termini and 11 nucleotides at the 5' termini were common to all segments examined. In the region immediately adjacent to the common 3' terminal region, the nucleotides were specific for each segment and these segment-specific sequences were conserved in all strains examined. In each of the five segments in which both termini were examined, the segment-specific 3' sequences exhibited perfect inverted complementarity to a segment-specific sequence adjacent to the common 5' terminus. In addition, in the 3' noncoding region of RNA segments 1-3, which encode proteins involved in RNA synthesis, a single nucleotide substitution at position 10 was found that distinguishes these segments from segments 4-8. Comparison of these data with published reports has revealed that some of the features found in the noncoding regions of influenza B virus are also present in influenza A and C virus RNAs. In the RNAs of all three virus types, there is a segment-specific sequence of nucleotides near the 3' terminus that shows inverted complementarity to a sequence near the 5' terminus. This segment-specific sequence may play a role in the transcription of individual segments or in sorting of segments during virion assembly.

Protease activation mutants of Sendai virus: sequence analysis of the mRNA of the fusion protein (F) gene and direct identification of the cleavage-activation site.

Trypsin cleaves the fusion protein (F) of wild-type Sendai virus into two disulfide-linked polypeptides, F1 and F2, and thereby activates the membrane fusion activity of the virus. A. Scheid and P.W. Choppin [1976). Virology, 265-277) selected mutant viruses of which the F protein could be activated by different proteases, either elastase, chymotrypsin, or plasmin. Herein, we have further characterized five of these mutants. Sequencing of each mutant mRNA encoding the 60-70 amino acids surrounding the cleavage site revealed one or two amino acid changes near or at the cleavage sites. Virions cleaved in vitro by the appropriate proteases were assayed of their fusion activity by hemolysis, and the cleavage sites were determined by amino acid sequencing. In three cases, the change of protease specificity can be accounted for by changed amino acids right at the cleavage site, whereas several other mutations that potentiate cleavage at new sites by new proteases are somewhat removed from the actual cleavage site. We surmise that such mutations might alter local polypeptide conformation, thereby allowing the proteases access to existing sites. Cleavage at new sites produced fusion proteins with novel F1 NH-termini. We found that a mutant with a charged residue at the third position of this normally hydrophobic NH-terminal sequence retains activity in the hemolysis assay, whereas a mutant with a charged residue at the first position does not.

Biochemical studies on cell fusion. II. Control of fusion response by lipid alteration.

The preceding communication (Roos, D.S. and P.W. Choppin, 1985, J. Cell Biol. 101:1578-1590) described the lipid composition of a series of mouse fibroblast cell lines which vary in susceptibility to the fusogenic effects of polyethylene glycol (PEG). Two alterations in lipid content were found to be directly correlated with resistance to PEG-induced cell fusion: increases in fatty acyl chain saturation, and the elevation of neutral glycerides, including an unusual ether-linked compound. In this study, we have probed the association between lipid composition and cell fusion through the use of fatty acid supplements to the cellular growth medium, and show that the fusibility of cells can be controlled by altering their acyl chain composition. The parental Clone 1D cells contain moderately unsaturated fatty acids with a ratio of saturates to polyunsaturates (S/P) approximately 1 and fuse virtually to completion following a standard PEG treatment. By contrast, the lipids of a highly fusion-resistant mutant cell line, F40, are highly saturated (S/P approximately 4). When the S/P ratio of Clone 1D cells was increased to approximate that normally found in F40 cells by growth in the presence of high concentrations of saturated fatty acids, they became highly resistant to PEG. Reduction of the S/P ratio of F40 cells by growth in cis-polyunsaturated fatty acids rendered them susceptible to fusion. Cell lines F8, F16, etc., which are normally intermediate between Clone 1D and F40 in both lipid composition and fusion response, can be altered in either direction (towards either increased or decreased susceptibility to fusion) by the addition of appropriate fatty acids to the growth medium. Although trans-unsaturated fatty acids have phase-transition temperatures roughly similar to saturated compounds, and might therefore be expected to affect membrane fluidity in a similar manner, trans-unsaturated fatty acids exerted the same effect as cis-unsaturates on the control of PEG-induced cell fusion. This observation suggests that the control of cell fusion by alteration of fatty acid content is not due to changes in membrane fluidity, and thus that the fatty acids are involved in some other way in the modulation of cell fusion.

Biochemical studies on cell fusion. I. Lipid composition of fusion-resistant cells.

A series of stable cell mutants of mouse fibroblasts were previously isolated (Roos, D. S. and R. L. Davidson, 1980, Somatic Cell Genet., 6:381-390) that exhibit varying degrees of resistance to the fusion-inducing effect of polyethylene glycol (PEG), but are morphologically similar to the parental cells from which they were derived. Biochemical analysis of these mutant cell lines has revealed differences in whole cell lipid composition which are directly correlated with their susceptibility to fusion. Fusion-resistant cells contain elevated levels of neutral lipids, particularly triglycerides and an unusual ether-linked lipid, O-alkyl, diacylglycerol. This ether lipid is increased approximately 35-fold over parental cells in the most highly PEG-resistant cell line. Fusion-resistant cells also contain more highly saturated fatty acyl chains (ratio of saturated to polyunsaturated fatty acids [S/P ratio] approximately 4:1) than the parental line (S/P ratio approximately 1:1). Cells which are intermediate in their resistance to PEG have ether lipid and fatty acid composition which is intermediate between the parental cells and the most fusion-resistant mutants. In a related communication (Roos, D. S. and P. W. Choppin, 1985, J. Cell. Biol., 100:1591-1598) evidence is presented that alteration of lipid content can predictably control the fusion response of these cells.

Analysis of Sendai virus mRNAs with cDNA clones of viral genes and sequences of biologically important regions of the fusion protein.

cDNA clones representing five of the genes of Sendai virus (P, HN, NP, F, and M) were isolated and used to identify the viral mRNAs by hybridization. Five mRNAs that were monocistronic transcripts of these genes were identified. A sixth transcript, which was identified on the basis of size and of hybridization to viral RNA but not to the cDNA of the other five genes, is thought to represent the message for the L protein. In addition, polycistronic transcripts of the NP and P genes and of the M and F genes were also found. The latter establishes the position of the F gene adjacent to the M gene; these results confirm and extend the previously reported partial gene order of the virus. Nucleotide sequences and derived amino acid sequences of two biologically important regions of the F protein--approximately 25% of F proximal to its COOH terminus and the region spanning the site of the proteolytic cleavage that activates the fusion activity of the protein--are presented. The F protein has an unusually large "cytoplasmic domain" of 42 amino acids beyond the hydrophobic region by which it is anchored in the viral membrane. A single possible trypsin cleavage site was found at the junction of the F1 and F2 polypeptides, and 26 hydrophobic amino acids extend from this cleavage site at the NH2 terminus of the F1 polypeptide.

Tumorigenicity of cell lines with altered lipid composition.

A series of closely related mouse fibroblast cell lines that differ in their content of neutral ether-linked glycerolipid and fatty acids has been used to investigate the relationship between lipid composition and tumorigenicity. Although these cell lines, derived from the same parental culture, were selected without reference to transformation or tumorigenicity, their ability to form tumors in irradiated mice was found to be closely correlated with ether-lipid content. The cell line with the highest level of ether-lipid (designated F40) produces more tumors, the tumors appear more rapidly than when parental cells are injected, and the number of F40 cells required for tumor induction is less by a factor of approximately equal to 1000. F40 tumors are highly invasive, readily metastasize, and rarely regress, in contrast to the occasional benign tumors produced by the parental cell line. Cell lines that are intermediate in their lipid composition appear to be intermediate in tumorigenicity. This panel of graded cell lines provides a useful model system for both in vitro and in vivo studies on the acquisition of tumorigenicity and malignancy in cultured cells.

Studies on the synthesis of the influenza V virus NB glycoprotein.

The synthesis of the influenza B virus glycoprotein NB has been demonstrated in cells infected with four different virus strains isolated from 1940 to 1983. Time-course studies have shown that the neuraminidase protein (NA) and NB appear simultaneously in infected cells, in agreement with the previous observation that both proteins are translated from a single, bicistronic mRNA. Pulse-chase experiments indicated that the NB protein disappears from infected cells within 3 hr after synthesis. The influenza B virus nonstructural protein NS2 was also observed to disappear rapidly from infected cells, with none detected 1 hr after synthesis. Immunoprecipitation of viral proteins from infected cells using hyperimmune mouse serum has demonstrated antigenic cross-reactivity among the NB proteins of the four virus strains studied. The finding of polypeptides smaller than NB that exhibit a labeling pattern with different amino acids which is characteristic of NB and which are precipitated by the mouse antiviral serum suggests that specific cleavage products of NB may be formed in infected cells.

Oligopeptides that specifically inhibit membrane fusion by paramyxoviruses: studies on the site of action.

Previous studies from this laboratory showed that oligopeptides with amino acid sequences similar to the sequence of the N-terminal region of the F1 polypeptide of paramyxoviruses inhibited the membrane fusing activity of the F protein, and thereby inhibited virus infectivity at the level of penetration and virus-induced cell fusion and hemolysis. The site of action of these oligopeptide inhibitors has been investigated. Radioactively labeled oligopeptides were found to bind to cells, but not to virus. Pretreatment of cells, but not virus, at 4 degrees with oligopeptides inhibited the initiation of infection and hemolysis induced by measles virus. The binding of the oligopeptides to cells was reversible at 25 or 37 degrees. Oligopeptides were synthesized with a chloromethylketone group to enable them to bind irreversibly, or with an azido group to permit them to be cross-linked in situ by photoactivation. The results with these derivatized oligopeptides, which retained their inhibitory activity, confirmed that they bind to, and express their inhibitory activity on, cells and not virus. The results suggest that the oligopeptides react with receptor sites on the cell membrane and inhibit membrane-fusing activity by competing with the F1 polypeptide for such sites. A Scatchard analysis of the binding of an oligopeptide to CV-1 cells revealed that it bound with a dissociation constant of 1.2 X 10(-7) M and that there were approximately 3.0 X 10(6) binding sites per cell.

A previously unrecognized influenza B virus glycoprotein from a bicistronic mRNA that also encodes the viral neuraminidase.

RNA segment 6 of the influenza B virus genome codes for a previously unidentified polypeptide designated NB. The reading frame for this polypeptide begins with the first AUG codon on the mRNA and overlaps the reading frame for the viral neuraminidase by 292 nucleotides. The amino acid sequence of polypeptide NB deduced from the nucleotide sequence of the B/Lee/40 strain consists of 100 amino acids with a molecular weight of 11,242. The sequence contains four potential glycosylation sites, and the protein has been found to be glycosylated in infected cells. NB has not been found in virions. Sucrose gradient sedimentation and analysis of the structure of the mRNA by nuclease S1 mapping and sequence analysis by the primer extension method indicated that polypeptide NB and the neuraminidase are translated from a single bicistronic mRNA. A protein analogous to NB has not been found with influenza A virus, and this represents a major difference between the two virus types.