Principles of Virology, Volume 1. Jane Flint
growth analysis can be used to study the single-cell reproductive cycle of viruses originated from the work on bacteriophages by Emory Ellis and Delbrück. In their classic experiment, they added virus particles to a culture of rapidly growing E. coli. These particles adsorbed quickly to the cells. The infected culture was then diluted, preventing further adsorption of unbound particles. This simple dilution step is the key to the experiment: it reduces further binding of virus to cells and effectively synchronizes the infection. Samples of the diluted culture were then taken every few minutes and analyzed for the number of infectious bacteriophages.
When the results of this experiment were plotted, several key observations emerged. The graphs were surprising in that they did not resemble the growth curves of bacteria or cultured cells. After a short lag, bacterial cell growth becomes exponential (i.e., each progeny cell is capable of dividing into two cells) and follows a straight line (Fig. 2.18A). Exponential growth continues until the nutrients in the medium are exhausted. In contrast, numbers of new viruses do not increase in a linear fashion from the start of the infection (Fig. 2.18B, left). There is an initial lag period in which no infectious viruses can be detected. This lag period is followed by a rapid increase in the number of infectious particles, which then plateaus. The single cycle of virus reproduction produces this “burst” of virus progeny. If the experiment is repeated, such that only a few cells are initially infected, the graph looks different (Fig. 2.18B, right). Instead of a single cycle, there is a stepwise increase in numbers of new viruses with time. Each step represents one cycle of virus infection.
Figure 2.17 Workflow for VS-Virome. Shown is the computational pipeline designed for defining the type and abundance of known and novel viral sequences in metagenomic data sets. VS-Virome first pre-processes the sequences (left) to remove adapter sequences (these are added to every DNA in the sample, and contain barcoding sequences, primer binding sites, and sequences for immobilizing the DNA), joins paired end reads if they overlap, performs quality control on sequences, and identifies low-complexity sequences and host sequences before subjecting all the sequences to BLAST (right) to detect viral sequences. Because integrated prophage are found in bacterial genomes, alignment to comprehensive databases could lead to removal of bona fide bacteriophage sequences. Bacteriophage hits are therefore placed into a separate output file. Candidate eukaryotic viral sequences are filtered to remove sequences that have high identity to bacterial genomes. Remaining reads are then aligned to the more comprehensive GenBank NT and NR databases to identify reads or contigs that have greater similarity to nonviral sequences than to viral sequences (i.e., increased likelihood of being a false positive). To have a high degree of confidence in viral classification, sequences that have significant hits to both viral and any nonviral reference sequence are placed in an “ambiguous” bin. Sequences in the viral bin only have significant alignment to viral sequences.
Once the nature of the viral propagation cycle was explored using the one-step growth curve, questions emerged about what was happening in the cell before the burst. What was the fate of the incoming virus? Did it disappear? How were more virus particles produced? These questions were answered by looking inside the infected cell. Instead of sampling the diluted culture for virus after various periods of infection, researchers prematurely lysed the infected cells as the infection proceeded and then assayed for infectious virus. The results were extremely informative. Immediately after dilution, there was a complete loss, or eclipse, of infectious virus for 10 to 15 min (Fig. 2.18B). In other words, input virions disappeared, and no new phage particles were produced during this period. It was shown later that the loss of infectivity is a consequence of the release of the genome from the virion, to allow for subsequent transcription of viral genes. Particle infectivity is lost during this phase because the released genome is not infectious under the conditions of the plaque assay. Later, newly assembled infectious particles could be detected inside the cell that had not yet been released by cell lysis.
METHODS
How to read a phylogenetic tree
Phylogenetic dendrograms, or trees, provide information about the inferred evolutionary relationships between viruses. The example shown in the figure is a phylogenetic tree for sequenced viral isolates from 10 different individuals. The horizontal dimension of the tree represents the degree of genetic change, and the scale (0.07) is the number of changes divided by the length of the sequence (in some trees this may be expressed as % change). The blue circles, called nodes, represent putative ancestors of the sampled viruses. Therefore, the branches represent chains of infections that have led to sampled viruses. The vertical distances have no significance.
The tree in the figure is rooted, which means that the root of the tree represents the common ancestor of all the sampled viruses. As we move from the root to the tips, we are moving forward in time, although the unit of time might not be known. The numbers next to each node represent the measure of support; these are computed by a variety of statistical approaches including “bootstrapping” and “Bayesian posterior probabilities.” A value close to 1 indicates strong evidence that sequences to the right of the node cluster together better than any other sequences. Often there is no known isolate corresponding to the root of the tree; in this case, an arbitrary root may be estimated, or the tree will be unrooted. In these cases, it can no longer be assumed that the order of ancestors proceeds from left to right.
Phylogenetic trees can also be constructed by grouping sampled viruses by host of isolation. Such an arrangement sometimes makes it possible to identify the animal source of a human virus. Circular forms, such as a radial format tree, are often displayed when the root is unknown.
Trees relating nucleic acid sequences depict the relationships as if sampled and intermediary sequences were on a trajectory to the present. This deduction is an oversimplification, because any intermediate that was lost during evolution will not be represented in the tree. In addition, any recombination or gene exchange by coinfection with similar viral genomes will scramble ordered lineages.
A fair question is whether we can predict the future trajectory or branches of the tree. We can never answer this question for two reasons: any given sample may not represent the diversity of any given virus population in an ecosystem, and we cannot predict the selective pressures that will be imposed.
Hall BG. 2011. Phylogenetic Trees Made Easy: A How- to Manual, 4th ed. Sinauer Associates, Sunderland, MA.
ViralZone. Phylogenetics of animal pathogens: basic principles and applications (a tutorial). http://viralzone.expasy.org/e_learning/phylogenetics/content.html
Figure 2.18 Comparison of bacterial and viral reproduction. (A) Growth curve for a bacterium. The number of bacteria is plotted as a function of time. One bacterium is added to the culture at time zero; after a brief lag, the bacterium begins to divide. The number of bacteria doubles every 20 min until nutrients in the medium are depleted and the growth rate decreases. The inset illustrates the propagation of bacteria by binary fission.