Sanctum Sanctorum of the Coronavirus

Why is the coronavirus wrapping itself in so many vesicles?

One of the most notorious jokes in molecular biology lectures is, when starting the discussion of DNA replication, to place a fig leaf over the cartoon of the replication fork- the moment when one strand of DNA becomes two. It works on many levels, and signifies the importance of this most central event and structure. Coronaviruses never use DNA- they are RNA viruses exclusively. But RNA replication is processive just like DNA replication. Since RNA occurs only in one strand, rather than DNA's two strands, however, copying has to happen in two stages, first to a negative strand copy, then back to a positive strand copy, which goes into new virions. Amplification naturally can happen along the way as well.

A remarkable characteristic of coronaviruses and virtually all other positive strand RNA viruses is that they generate a complex and novel population of membrane structures and vesicles in cells. It is in or on these vesicles that their RNA replication takes place. There is a whole new vocabulary of membrane entities made, like convoluted membranes, zippered endoplasmic reticulum, reticulovesicular network, double membrane spherules, and double membrane vesicles. The last of these (called DMVs) is where viral genomic RNA replication takes place.

Vesicles and other membrane structures induced by coronaviruses. The large double-walled structures are called DMV, or double membrane vesicles. The tiny vesicles around them are called DMS or double membrane spherules. And the mess labeled CM (including a few clear spherules) stands for convoluted membranes. These are all induced by viral infection and viral RNA replication seems to happen in or on the DMVs.

These membranes are created by some of the earliest protein products the virus encodes- the nsp3, nsp4, and nsp6 proteins. The figure shows one possible structure of nsp6, which is thoroughly integrated in membranes, by virtue of the hydrophobicity of its transmembrane segments. But how this structure leads to membrane curviture and DMV formation is not at all clear. A paper from 2014 developed a drug that binds and disrupts the function of nsp6, which dramatically reduced the formation of DMVs, and viral replication.

Schematic prediction of the transmembrane structure of viral protein nsp6.

An even older paper from 2008 did the original work that identified the double membrane vesicles as the site of RNA replication. They took high resolution electron micrographs of infected cells, and stained for the presence of double-stranded RNA. This showed up very distinctly in the DMV structures. A more recent paper followed this up by short time course labeling with radioactive RNA precursors, so that newly made viral RNA would show up . These results were less clear, but labeling is roughly around the DMV structures. 

DMV structures are full of double-stranded viral RNA, indicating a core role in replication.

Brief labeling by radioactive nucleotides to find the location of replication gives ambiguous results, but points to DMVs. Note the beautiful electron micrograph of mitochondria (mm) and other cellular & viral structures. ER is endoplasmic reticulum, RO is replicative organelles (of the virus, which includes DMVs), and VCR stands for virion-containing region.

One major motive for the virus to hide its replication and replication products inside DMVs is that cells have various defenses that are triggered by double stranded RNA, and by RNAs that have non-cellular caps. Coronaviruses contain several enzymes to make cell-like caps on its RNA products, but still, hiding them entirely is probably a safer approach. Their replication requires double-stranded RNA at least transiently, and typically quite a bit builds up. On the other hand, DMVs also present significant problems. Are these vesicles completely closed to the cytoplasm? If so, nucleotides for replication would be difficult to come by. Also, what ultimately happens to the RNA hidden inside? None of this is clear yet. The recent paper makes a case that DMVs all have small openings to the other structures or the cytoplasm, but previous studies disagree.

Thus more work is needed, to validate this drug as something that could be used in humans, to figure out how and even where replication of the viral RNA really happens, why positive strand RNA viruses are so reliant on membrane-linked replication systems, how the viral proteins like nsp6 drive formation of these membrane structures, and what the life cycle of these structures is- what the fate of all this packaged RNA might be. The original model of coronavirus replication was that RNA replication takes place freely in the cytoplasm, and the N (nucleocapsid) protein joins it to make proto-viral genome packets. These bud into the endoplasmic reticulum, at sites where the M and S (spike) proteins have been concentrated in proto- envelope membrane regions. These enveloped buds then comprise complete virions that are exported by normal cellular pathways through the vesicular secretion system in large vesicles that bud with the plasma membrane to release the new viruses. But the extra complexity of the DMVs throws this simple schema into doubt. For such a tiny quasi-life form, coronaviruses do a huge amount of damage, and present a lot of mysteries.

• We are heading into a lawless state.
• Notes on satanic democracy.
• Oh, we took the money and fired the workers anyhow.
• Lindsey Graham, taking the high road.
• A gene for bigger brains.