SNP CC in IFITM3 is a boon for H1N1, FWIW

The technology we have available to us today in the lab is both a boon and a bafflement. Example: The screens we have for RNA expression in cells is so sensitive we can see tiny changes in RNA expression levels in healthy/diseased/drug treated/etc cells. YAY! More information! More observations! More new ideas for research!… Except, the screens we have for RNA expression in cells is so sensitive, we can see tiny changes in RNA expression levels that dont really mean anything.

Example: The techniques we have for identifying viral RNA/DNA in cells is so sensitive… that we can pick up scant bits of contaminating DNA.

Another example: We can do all kinds of protocols for seeing differences in susceptible/non-susceptible cells for different viruses, to figure out which genes are stopping the virus, thus we learn more about the biochemistry of the virus and new angles to attack it in patients… but these screens ID genes. They dont ID what exactly is going on. So we have huge lists of genes that are up/down regulated in cells that successfully fend off, say, influenza, but we dont know what it means. We can see the trees, but not the forest.

An example of one of these IDed genes is Interferon-induced transmembrane protein 3 (:-/) We knew IFITM3 was part of the innate immune response (not antibodies or your T-cells, which learn and evolve in response to pathogens– your innate immune response sees patterns and reacts only to those patterns). And in in vitro studies, they found IFITM3 to be related to resistance to Influenza, West Nile, and Dengue infection… but we didnt really know what that meant for Influenza, West Nile, and Dengue victims. In vitro and in vivo are different worlds.

This recent paper in Nature tried to resolve that:

IFITM3 restricts the morbidity and mortality associated with influenza

First, mouse models. They generated mice that lacked the IFITM3 gene, then infected them with a low-pathogenicity murine-adapted H3N2 virus. This virus should not have given the mice any trouble.

And the regular mice did do fine. They lost a little weight for a bit, but they recovered. When the scientists looked at IFITM3 expression over time, it increased in the regular mice post infection.

The IFITM3(-) mice however, had real rough time with this weakened virus. They showed symptoms one would expect form a virulent virus, like high levels of influenza replication in the lungs, losing lots of weight, and generally having a rough time. From a virus that basically did nothing to the normal mice.

They saw the same pattern with several other forms of influenza in their mouse model, including Swine Flu.

Well, neat. Now we know how to kill IFITM3(-) mice.


The question remains as to whether/how IFITM3 plays a role in influenza infections in humans.

So, these folks looked at the IFITM3 gene in 53 people who got sick enough from Swine Flu that they had to be admitted to the hospital. Most of these people had a ‘normal’ IFITM3 gene with a TT at a splice acceptor site, or a variant commonly found in the general population, TC. BUT, 5.7% of the really sick influenza patients had a gene variant extremely rare in screenings of the general population (0.3%)– a CC.

Changing that TT to a CC, theoretically, leads to alternative RNA splicing, which leads to a alternative IFITM3 protein that is 21 amino acids shorter. If its shorter, it might not work well (or at all), leading to humans that get really sick from relatively benign influenza variants– like what we saw in the mice that lacked the IFITM3 gene. Maybe this is why some people got REALLY SICK from Swine Flu, but most people didnt even notice they were infected at all.

Well, you cant very well get a bunch of IFITM3-CC mutants in the lab to perform experiments on them to test this hypothesis. So these scientists generated a variant of IFITM3 that lacked the first 21 amino acids, like they think is happening in IFITM3-CC people. They then expressed it, IFITM3, or nothing in cells they infected with influenza. The shorter IFITM3 was only slightly better than nothing at restricting influenza replication, while the full length IFITM3 did just fine, implying that people with the IFITM3-CC mutation would, in fact, have a harder time controlling influenza infection.


Of course, this paper doesnt explain the other 94.3% of patients admitted to the hospital for Swine Flu infection with normal IFITM3, but its still pretty cool– It tells me that when genomic screens become cheap/widely available, we need to find the 0.3% of the population that is a IFITM3-CC mutant, and make damn sure they are first in line to get all of their immunizations.


  1. #1 Thanny
    March 27, 2012

    Maybe there are roughly 20 ways to suck at fighting Swine Flu, and that’s just one of them.

    Given how Rube Goldbergian life is, I’d be surprised if there were as few as 20 possible reasons for poor immune response to any given disease. In other words, there’s never going to be a panacea, so research like this is necessary, many more times over.

  2. #2 JKR
    March 27, 2012

    Thanny, you are right, but often that traditionally has taken years and sometimes luck to find. The questions is, whether or not US has the will to continue funding biological research at necessary levels for years to come.

  3. #3 Charl
    March 28, 2012

    They also found that IFITM3 is even better (or the CC is even worse) at dealing with Flu B than Flu A, so it’s a pretty cool anti-flu pathway. I also agree with Thanny – there are probably multiple host-virus interactions you can mutate that lead to sucky flu reactions.

  4. #4 cynical1
    March 28, 2012

    Interesting work. Although, I suspect that Thanny is correct, at least with HIV, to my knowledge, there is only one genetic mutation that imparts immunity to infection in humans: CCR5 delta 32 mutation. (Unfortunately, the drugs that tried to mimic that mechanism were less than stellar in the clinic.) So there are instances where a single genetic mutation explains the epidemiology.

  5. #5 hoary puccoon
    March 28, 2012

    How many of the 94.3% who didn’t have the mutant gene but were hospitalized had some other risk factor– being very old, very young, HIV positive or suffering from some other serious health problem?

  6. #6 davidp
    March 28, 2012

    Cool headline and really good writeup Abbie. Thanks.

  7. #7 windy
    March 29, 2012

    So there are instances where a single genetic mutation explains the epidemiology.

    The CCR5 resistance allele is only present in 10% of Europeans and in much smaller proportions in other populations, so it only explains a small part of the epidemiology… not that different from the IFITM3-CC situation.

    There is also at least one other genetic mutation that has been connected with HIV susceptibility, the absence of the Duffy antigen:

  8. #8 windy
    March 29, 2012

    #7 correction: the Duffy antigen receptor

  9. #9 cynical1
    March 29, 2012

    People who have inherited two copies of the delta 32 gene (from both parents) are essentially immune to HIV infection. People with one copy are more resistant to infection and show slower progression to full blown AIDS. The delta 32 mutant is very rare in Africans but occurs in about 20% of caucasians. It’s hard to argue that this mutation doesn’t explain all the epidemiology that I’ve heard of but maybe I’m missing something.

    The duffy antigen is contentious. The V.A. and also researchers at Case Western have published work that disputes the association, at least in African Americans. So I think the jury is still out on that one.

  10. #10 windy
    March 29, 2012

    It’s hard to argue that this mutation doesn’t explain all the epidemiology that I’ve heard of but maybe I’m missing something.

    Having a resistance/susceptibility allele can make all the difference for a particular person getting or not getting the disease, but I wouldn’t call that, in itself, the whole epidemiology. Both the CCR5 and IFITM3 variants only explain a few percent of the population level variation (which is what Thanny was referring to).

  11. #11 cynical1
    March 30, 2012

    @ windy: Ok I’m confused. What is the population level variation that you are referring to with regard to HIV infection? Are there groups out there, other than those with the delta 32 CCR5 mutation, that are repeatedly exposed to HIV but fail to become infected? I thought that there was one and only one protective gene for HIV infectivity but maybe I’m wrong. In the case of IFITM3, the rare genetic variant explains why SOME (5.7%) people get really sick but not why others (~94%) who don’t possess the variant also get really sick. That, at least to me, is a population level variation. Am I missing something? Once again, I am only talking about the epidemiology of infection, not progression rate to AIDS. Of course, the IFITM3 work is addressing disease severity.

  12. #12 hoary puccoon
    March 31, 2012


    I asked at #5, above, whether the 94.3% might have other risk factors, but got no answer. Before looking in the lab for lots of other genetic factors, it would be nice to know, just for starters, the ages of the hospitalized patients. Did the 5.7% of the total hospitalized make up a much more substantial % of those hospitalized who were, say, non-smokers, HIV-negative, between the ages of 18 and 45?

  13. #13 John weis
    March 31, 2012

    Very nice summary of that paper. I was excited to read it ’cause it underscored both the importance of IFITM3 as well as the resourcefullness of the virus that put the other 94% with a normal IFITM3 gene in the hospital. It also preceded our publication, by a day or two, that describes a potential mechanism of IFITM3 function by stabilizing the vATPase/Cathrin complex in endosomes of activated cells. There is still much to be learned about the interferon response.

  14. #14 ERV
    March 31, 2012

    Very cool, John! I will have to look up that paper too– The whole time I was reading this one, I was wondering what *exactly* was going on!