Sunday, 22 May 2016

Zika in the mouse...

Just over a week ago there were a few published Zika virus (ZIKV) studies in mice and brain balls (neurospheres and organoids). These were big deals. One was a letter to Nature by Cugola and colleagues, describing defects in the brains of mouse pups after their mothers were infected with virus.[1] This was one big letter and well outside my expertise to review in any sort of depth. But it raised a few questions for me.

Briefly, there were two parts to the letter - in part 1, two lines of pregnant mice (SJL and C58BL/6) were injected with virus and the pups evaluated immediately after birth. In Part 2, cell culture created human progenitor stem cells (hPSCs) and brain balls (neurospheres and cerebral organoids) were incubated with ZIKV.

This study included a ZIKV variant currently circulating in Brazil (ZIKVBR) rather than using the virus from six decades ago, but disappointingly, it did not include dengue virus or Chikungunya virus as control viruses against which to compare the activities described for ZIKV. There was some use of a slow-growing, attenuated (low risk of nerve pathology) vaccine strain of yellow fever virus as control virus - but exactly how wasn't clear to me. There was no virus control in the mouse work though - just the organoid experiments - as far as I can tell.
  1. One big takeaway message was that the C56BL/6 pups did not have any notable differences compared to healthy pups. Virus did not seem to cross the placenta. All the problems found were in the other line of mice - the immunologically defective SJL.[8,9]
    The authors note the C56BL/6 line had a robust antiviral immune response.
    If ZIKV is indeed the cause of congenital disease in humans, there is something to that result that should be of much interest. Whether that is at the level of a genetic immune deficiency, a microbiome-level issue or the occurence/history, absence or order of past infections, among other things, is unclear
  2. The SJL pups were smaller, had lots more ZIKV RNA in the brain than in kidney, liver or spleen, tissues, had eye abnormalities and had elevated markers suggesting that brain cell death was linked to apoptosis and autophagy. The authors remarked how these shared similarities with foetal human disease
  3. I have my usual question about how relevant to a human is a model that directly introduces a dumpload of virus into the blood (40,000,000,000 plaque forming units) by injection when the natural route of infection of a human is considered to be mostly due to a (presumably) much lower load from mosquito bite. Adding the same dose of inactivated ZIKV would have been very interesting to test for non-viral effects from the inoculum (h/t KatA)
  4. Human pluripotent stem cell (hPSC)-derived neural progenitor cells (NPCs) were also exposed to virus and found to die via apoptosis. But where did these cells originate? I can't tell from the methods and references cited. Was it from human skin cells as other experiments have used? Skin cells are known to host ZIKV [3] Could this be a confounding factor? 
  5. Mock infected NPCs - "pretend" infection in the absence of any actual virus to check whether the method or material carrying the virus caused the observed damage - actually upregulated expression of the likely receptor for ZIKV, AXL, but ZIKV infection didn't.  Does that mean ZIKV down-regulated expression caused by something in the mock inoculum?
  6. High doses (10 MOI) of ZIKV killed NPCs, but a lower dose (1 MOI) did not - what does that mean for the heavily dosed mouse model; and for a human bitten by a mosquito? ZIKVBR and an African lineage ZIKV (ZIKVAF) acted similarly. 
  7. In three dimensional cultures, neurospheres growing in the presence of 10 MOI of ZIKVBR were smaller and cell death was apparent - the effect was not as strong with ZIKVAF suggesting lineage differences. Effects were dose dependent - stronger with higher doses (10 MOI) than lower (1 MOI). Differences between ZIKV lineages is something yet to be fully explored by virology-for some inexplicable reason(s).
You'll possibly have noticed two units used by the authors - plaque forming units and MOI (multiplicity of infection). These are different because of different methods used to determine the endpoints. Briefly (but still technically):

From Sloutskin et al.[6]
  • For plaque-forming units (PFU) - we titrate infectious virus that can damage cells (make plaques) and look at where the effect finishes. By titrate I mean serially dilute and then add each diluted solution to a separate well of the same cells, usually grown in a multiwell plate. The effect is the formation of plaques - areas of clearing due to virus-induced death of cells that had first been grown into a single layer in each plate well before being infected by a virus preparation. The dilution before the effect finishes is the PFU.
    There are also issues around how well PFU value determined using one cell culture model holds up when infecting a cell/tissue/organ/animal that is different from the one you determined your PFU on - you may get different results.
    For example you may find your virus preparation contains 10,000 PFU on the original cells, but if you did that same titration with the same preparation but using a different cell type, it may contain 100,000 PFU or 1,000 PFU.
    In the study above, ZIKV was grown up using a C6/36 mosquito cell line, then titrated using porcine kidney epithelial cells to determine the PFU then that C6/36 preparation was used for the mouse, NPC, neurosphere and organoid incubations. This is normal approach but can raise questions.
  • For multiplicity of infection (MOI) we are talking about the average number of virus particles in a preparation that infect a target cell. 1 MOI means 1 virus particle per cell. An MOI of 10 means 10 virus particles per cell.[4] This works well in cell culture where we often use a single cell type grown in a single layer - conditions, viral density and culture volumes are optimized and we use the same cells as for the PFU.[5] When you take that MOI of 1, calculated on your cell line, and add it to a different and complex tissue, or animal, with lots of different cell types, perhaps spread over a larger or smaller surface, in the presence or absence of an antiviral response, with more or fewer receptors etc...the ratio of 1 virus : 1 cell will probably not hold up. So using a higher MOI makes it more likley that each cell in a different tissue/organ/animal will get infected. Thus an MOI of 1 and 1 PFU are not always the same thing. A virus preparation determined to have an MOI of 1 using one system might require 0.1 or 273.64 PFU of that preparation in another cell/tissue/organ/animal system, because it takes that amount to finally show the desired effect (cell death, PCR positivity, virus protein detection etc) in that target cell/tissue/organ/animal.

    We should also think about how relevant an MOI of 10 is to the thing we originally sought to study. In this case it is the early brain structure of a foetus whose mother was infected from a mosquito bite.
    The bite itself would not have delivered anything like an MOI of 1 to any tissue in that mother's body.
    But how much ZIKV crosses the human placenta in the rare cases that it does at all? Well, we assume placental crossing is a rare event for ZIKV.
    I'd presume it's also not an MOI of 1 for any tissue in the foetus - unless the placenta is where the virus is being amplified. But it does seem like there is a lot of ZIKV in the brain of studied infected foetuses.  So this mouse model might reflect what happens in a forming human brain that amplifies virus it acquired after placental crossing, or that was present from conception (here for more on that theory[7]).  

One of my biggest questions about this study is - so what? 

This is an incredibly dense and impressive piece of work - don't get me wrong. It may well be the definitive result that so many have commented on it being. But I wonder if the same degree of very detailed investigation gone into the study of other mosquito-borne viruses in mouse brains? Have we sought out the congenital infection potential of other arboviruses in mouse pups and investigated arbovirus impact on neurospheres and organoids to this extent? Are these results unique to ZIKV or do the same or similar effects from other viruses, that - to our knowledge - are not considered threats for congenital disease in humans? We should probably establish that, by using more experimental controls, before we continue on our journey along this limb.

I have trouble saying that this experiment equals that disease as much as I am troubled by saying that the (possible) surge of microcephaly diagnoses in north east Brazil is caused by that particular parallel viral epidemic.

References...
  1. The Brazilian Zika virus strain causes birth defects in experimental models
    http://www.nature.com/nature/journal/vnfv/ncurrent/full/nature18296.html
  2. http://jvi.asm.org/content/82/12/6024.full
  3. http://www.ncbi.nlm.nih.gov/pubmed/26085147
  4. https://en.wikipedia.org/wiki/Multiplicity_of_infection
  5. http://www.virology.ws/2014/05/06/virology-question-of-the-week-what-matters-more-multiplicity-of-infection-or-virus-concentration/
  6. http://www.bio-protocol.org/e1295
  7. http://virologydownunder.blogspot.com.au/2016/04/the-three-parent-hypothesis-mum-dad-and.html
  8. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1453197/
  9. http://www.informatics.jax.org/external/festing/mouse/docs/SJL.shtml