Skip to Content

Mutated Mutations (And Other Rodent Study Problems)

Here’s one for the “Dang it, now I have to think about that, too” file. A recent paper suggests that there are mutations in many genetically modified mouse models that could well be confounding their phenotypes. The problem is that so many of these are done from very similar embryonic stem cells and in very similar recipient mouse stains (C57 black!). “Passenger mutations” apparently show up flanking the targeted gene, and they’re not always silent, either.

For instance, it was thought that the gene Casp1 was the principal player that triggered an inflammatory response and cell death pathway in response to foreign organisms, a step involved in lethal shock. That’s because, according to a 1995 study and subsequent work, Casp1 knockout mice did not go into septic shock when challenged with molecules signaling foreign invaders. However, in 2011, researchers at Genentech showed that many Casp1 knockout mice also harbored a mutated Casp11 gene from 129 strain mice. The researchers showed that the passenger mutation to Casp11 was partly responsible for the animals’ resistance to shock.

This result triggered Vanden Berghe and his colleagues to look into their own work with Casp3 knockout mice. They found that their interpretation of Casp3’s role in septic shock had also been confounded by the Casp11 passenger mutation. “It affected two years of work,” Vanden Berghe said.

Here’s a web-based tool this multicenter team has developed to help others search for known mutations of this sort. Definitely worth a look if you’re altering mice for a living!

This gets back to something mentioned here the other day. It’s thought by many that a lot of animal assays are statistically underpowered, especially those from academic labs (where the budgets are tighter). This sort of variable (the new mutational problem) doesn’t help, but there are plenty of others to scatter your data already in a whole animal. That’s particularly true if you’re working in (say) neuroscience, where hard readouts are hard to come by. And the problem is, results in such assays are often the big final readout for a given research program, the test that shows whether the hypothesis was correct or not.

Something to think about next time you see an interesting paper that relies on rodent data. If you’re not a big statistics powerhouse, get someone who is to take a look before you get too wrapped up.

12 comments on “Mutated Mutations (And Other Rodent Study Problems)”

  1. anonymous says:

    We had a KO C57Bl6 line which was completely impervious to MOG-induced EAE (model of multiple sclerosis for those fortunate individuals who haven’t had to use it). A publication came out with the same gene KO in a C57Bl6 background & identical protocol of MOG-induced EAE- absolutely no effect in either the onset or extent of symptoms.
    On the bright side, KO very useful for testing the specificity of antibodies….

  2. Anon says:

    In unrelated news, I was wondering if you’ve seen the new “Big Pharma” simulation video game that just came out (linked in my handle). It sounds like it’s essentially one guy developing it, but it seems pretty well done. I haven’t played it, so I don’t know how well it really replicates the experience of managing a drug company.

    1. @Anon:
      I’ve already played a few rounds of Big Pharma. It’s completely not how pharmaceutical development works (in particular, there is no modelling of the grueling process of doing the kind of development that our friendly blog author does). But it’s still rather fun, nonetheless, if you like management simulator-style games.

  3. Mark Thorson says:

    It sounds like it’s essentially one guy developing it

    And that would be you, right?

  4. I was a postdoc at the time in the lab where the Casp11 mutation was discovered. I do think that this a truly revolutionary discovery and underscores the importance and influence of passenger mutations. Since then Caspase 11 has shown be one of the most critical effectors for facilitating non-canonical inflammasome activation

  5. LidiaKny619 says:

    Wow, that’s what I was looking for, what a information! existing here at this web site, thanks admin of this site.

  6. Kevin says:

    Any good lab backcrosses their KI/KO to a wild type or mixed background. The lab next door found lethal phenotypes in one background, perfectly viable in another. Additional, linked passenger mutations would make the situation worse because they wouldn’t be cleaned up by outcrossing.

  7. HT says:

    Hope that would raise enough awareness on this issue. With the advent of CRISPR, you can bet that the use of GM mice in experiments will explode. The scientific literature does not need more underpowered, irreproducible studies.

    On unrelated matters, this blog entry seems to have attracted many unrelated comments/ links …

  8. Paramus says:

    I totally agree that we do not want more underpowered, irreproducible results. One problem has always been a lack of understanding of the basic phenotype of the background strain. In a former life, we did work with alpha 7 KO mice imported from the US. I insisted on also importing the wt. When we compared the US wt to UK wt, same ‘background’, they were very different; even their physical appearance. Often studies are underpowered because many of these strains don’t breed as efficiently as out-bred animals; however, this is no excuse for bad science!

  9. Dr. Manhattan says:

    With the advent of whole genome sequencing, many of the underlying assumptions around genotypes and “inbred” strains (mice and other test lines) are being replaced by more informative (and initially more complicated) understanding of genes & their relationships. Add the epigenetic aspect and things only get more complex. One hopes that at some point all of this data will be sorted and can be applied to improve the models employed for testing.

  10. anon says:

    ” If you’re not a big statistics powerhouse, get someone who is to take a look before you get too wrapped up.”
    I don’t know how statistics will help here. Suppose you have cohort of mice with same breed and all of them have similar passenger mutation, you will get similar results even though it is well-powered experiment. Suppose you get different breeds of mice with different mutations. All statistics will tell you is that null hypothesis is right. It will not answer the question why a particular phenotype is observed. I think in this case whole genome sequencing of each mice used in experiment and understanding whether there are mutations is more helpful than statistics. Blind worship of statistics will not advance science.

Comments are closed.