Skip to content Skip to navigation

What are the Physical Properties - Branching Characteristics and 3D sizes - of viral-length RNA molecules?

RNA as a Branched Polymer

The figure above is a cryo-electron microscopy image that we obtained [RNA 2012] from a purified solution of the RNA molecules (circled in red) corresponding to one of the genes of CCMV. Each of the molecules in the micrograph is chemically identical to the others – the same 3200 nucleotide (nt)-long sequence of RNA. But they appear different because they have different secondary and tertiary structures and because they have different orientations in the vitreous water in which they are trapped at low temperature; accordingly, they have different 2D projections in the transmission micrograph. Indeed, an RNA molecule this long must be regarded as a “statistical object” that must be represented by an ensemble of configurations, much like a long semi-flexible polymer. What all the molecules have in common, however, is a size that is comparable to that of the rigid protein shell in which they end up being confined, following self-assembly with the viral capsid protein. To illustrate this point we have added some purified virus to the sample, one particle of which (see arrow) appears in the lower right corner. Compare this image to that below of DNA viruses whose double-stranded genomes are huge compared to their capsids and whose packaging cannot be achieved by spontaneous self-assembly but instead requires a great deal of work to be performed.



We understand the relative compactness of a viral RNA genome in terms of the effective branching that results from its sequence self-complementarity, i.e., from the formation of successive base pairs (duplexes) involving complementary pairs of nucleotides separated by large distances along the single-stranded (ss) RNA. In particular, there are instances of three or more duplexes that emanate from one single-stranded loop, as illustrated below in a) for the lowest-free-energy secondary structure of a 120nt-long sequence. In b) we show this structure represented by its unique mapping onto a “tree graph”, a collection of connected points involving no closed path. For a sequence as long as 1000nt or more, however, there are many different secondary structures whose energies lie within kT of one another, and which are therefore present in comparable numbers at room temperature [Proc. Natl. Acad. Sci. 2008, 2012 RNA, PLoSONE 2014].
The corresponding ensemble of thermally accessible secondary and tertiary configurations cannot be crystallized, and hence their structure cannot be probed by the usual X-ray crystallography methods. Nor do solution NMR techniques for determining 3D structure work, because RNA molecules this long (and viral RNAs have to be this long in order to code for genes) rotate too slowly for motional narrowing to simplify their spectra. Finally, their secondary/tertiary structures would be disturbed by atomic force or negative-stain electron microscopies because of the interaction with solid substrates or with non-physiological solution (e.g., the uranyl acetate stain utilized in negative-stain imaging) conditions.

Accordingly, it is preferable to image them by cryo-electron microscopy in which the RNAs are probed as isolated molecules, without contrast agents, in vitrified aqueous films freely-suspended over holes in the electron-microscope grid [RNA 2012]. And theory performed on these molecules must explicitly deal with an ensemble (“forest”) of secondary structures (“trees”) associated with any single nucleotide sequence. This theoretical work is carried out in collaboration with Professor Avinoam Ben-Shaul of the Hebrew University of Jerusalem.


RNA as a ... Linear Polymer

In contrast to DNA, which is “just” the carrier of genetic information, RNA is often touted as the molecule that “does everything”. For example: it is the messenger between DNA sequence and protein product synthesis; it is the enzyme responsible for a host of nucleic acid cleavage reactions; it is an important structural molecule; and it is the genome of most viruses, etc. But there is still another way in which RNA is remarkable – RNA can behave either as a branched or as a linear polymer.

From the discussion immediately above (see (a) from the image above) it is clear that RNA becomes effectively branched because of the significant degree of self-complementarity that arises in any “reasonable-sequence” molecule, i.e., ones with comparable numbers of A, U, G and C bases. But what if there is no possibility of base pairing between nucleotides? In particular, how different is the behavior of an RNA molecule for which no secondary structure can arise?

To answer this question we use long strings of a single-letter RNAs, in particular, polyU RNA, which also shows the least tendency for base-stacking of all the single-letter RNAs. We work with samples whose lengths involve from 100s of nucleotides to ones with more than 10,000, and our goals are several-fold. The theme common to all of our studies is the contrast between polyU and “normal” RNA molecules as physical objects, and the consequences of these differences for their packaging by capsid protein. By “normal” RNA we mean the “reasonable-sequence” molecules defined just above, whether they be mathematically-random sequences, viral sequences, or non-viral biological sequences.

Because of the absence of duplexes in polyU, the usual intercalating dyes like ethidium bromide or SYBR Green do not work well in lighting it up. Accordingly, we use 30-nt fluorescently labeled polyA oligos to see the polyU. By running assembly mixes of fluorescent polyU (synthesized using the enzyme polynucleotide phosphorylase, PnPase) and CCMV capsid protein in electrophoretic gels, and staining the protein with Coomassie Blue, we determine the fraction of polyU packaged by capsid protein. Electron micrographs show size distributions for the virus-like-particles (VLPs) formed that differ significantly from assembly products formed from “normal” RNA and protein. In particular (see figure above) polyU lengths in the range 2500-9000nt are packaged into capsids that are 20% smaller than (the wildtype CCMV virions, and than) the VLPs formed from normal RNAs of this size. This is a first, qualitative, indication that the charge on the RNA is not the only important physical property for determining its packageability.