Skip to main content

Animal Testing

Muting Mutations

A solid, believable animal model of a human disease (preferably in a small animal like a mouse!) is a very important thing to have in a drug discovery project, but they are hard to come by. A mouse is not a human, and neither are the other small organisms that we’d like to use. But there’s no other way to go. We still share huge amounts of biochemistry with basically any living creature on Earth, and the complexity of living cells and organs cannot (as yet) be recapitulated by anything else – no in vitro system, no digital simulation.

I’m especially struck by the suggestions that come up for the latter, mostly from people outside of research who don’t realize what they’re asking for. How do you simulate something that has a bunch of important parts that you haven’t even discovered yet? We keep finding big, crucial systems and mechanisms in living cells that we didn’t even know were there, and I’m sure that there are plenty more head-slappers on the way. Even when we find them, it’s not like we suddenly arrive at a complete understanding of them, either, that’s for sure. Look (just to pick one example) at the bewildering variety of RNA types that we now know are present in cells. Just realizing that they exist has been quite an eye-opening adventure, but have we found them all? Why would anyone think so? And do we understand all their functions? Not at all.

So we have to carry on with cellular and animal models, even though we know that they’re imperfect at best, and even though we know that we don’t know (in many cases) why and how they’re so imperfect. But here’s a new paper that might shed some light on some of them. One of the well-known complications shows up when you try to create mutant disease models, which should (in theory) be very powerful and useful. And they can be, but they can also be weird and frustrating. The number of those that have generated less of an effect than foreseen is just uncountable.

Why is that? The standard explanations, which have nothing wrong with them as far as they go, are that other cellular pathways can often compensate for a defective gene’s product, and that even if there’s not an immediate backup protein ready to step in, that during development things might get rewired on the fly (after all, there’s a lot of incentive). The end result is a much more normal-looking animal than you ever expected. And it’s often difficult or plain impossible to tell what exactly compensated for the gene deletion and how it managed the trick. There really is a “genetic robustness” effect (which makes excellent evolutionary sense), something that makes the whole machinery less fragile and more able to compensate for trouble. Mutations certainly happen with functional consequences, of course, but their effects do seem to be damped as much as possible by some system we haven’t been able to work out.

And that’s where this new paper comes in. The authors have studied both mouse and zebrafish models of defined mutations (either a premature stop codon or deletion of their last exon), and they find that there are changes in protein expression that show up. It doesn’t seem to be a broad stress response, and indeed, when you look at the upregulated genes in general, a disproportionate number of them have sequence similarity to the one that was mutated. Heterozygous mutants show the same effects (but less) and it doesn’t seem to be induced by exposure to the mutant protein products themselves or by loss of the wild-type protein functions. That makes you think that the effect is upstream.

This transcriptional adaptation doesn’t happen with all mutations, it should be noted. And when the group looked at those non-adapting mutations versus the adapting ones, the difference seems to be that the adaptation-sensitive ones have mRNA that breaks down more quickly. Something about the decay of mutant RNA sets off the response to upregulate genes that are related to the mutated one and to open up the chromatin structure to get this done. Messing around with the known RNA surveillance machinery affected this process, as it should, as did introducing RNA sequences that were deliberately made even less stable.

Now, this also makes you wonder about mutations that occur in forms where no RNA gets generated at all, such as ones where the promoter region (or the entire gene locus) has been deleted. And indeed, these don’t seem to set off the adaptation response, which means that they can produce phenotypes that are attenuated (sometimes completely) in mutations that can actually produce unstable mutant RNA. This point, I’d say, has not really been appreciated to date – you have to keep that mutant RNA from being produced at all if you want to see the full effects of your change.

This makes sense, both from an evolutionary perspective and in light of the evidence that’s been accumulating about how RNA stability affects gene transcription in general. There’s an awful lot of mRNA-monitoring activity going on in the cell, and that was thought to be there to keep suspicious sequences from being turned into protein. And while that’s true, it now looks like another function is to go further and to call up related genes that might be able to compensate for the apparent trouble. We might, in fact, already have been looking at these effects in the pattens of human genetic disorders, without quite realizing what we’ve been seeing:

The current dogma is that pathogenic missense mutations tend to be more common in affected individuals because they might lead to constitutively active or dominant negative proteins. However, we propose that nonsense mutations are less common as they might result in mRNA decay-triggered upregulation of related genes and therefore not cause noticeable symptoms. Detailed transcriptomic analyses of relevant individuals will help to test this hypothesis. 

We’ve been talking about “penetrance” of genetic disorders without really knowing what we’re referring to in detail, but that’s changing. The transcriptional machinery is startling stuff, and it gets more impressive and complicated the further we dig into it. Evolutionary pressure could hardly be higher than on this part of the cell’s workings, and we living creatures have had a long time to build up redundant multifunctional backup systems that do several things simultaneously. It’s something to see, and there are no doubt even more complications to come.

7 comments on “Muting Mutations”

  1. John Wayne says:

    “How do you simulate something that has a bunch of important parts that you haven’t even discovered yet?”

    Nicely summarized. A computer can whip our butts at Go, but the rules of that game at 100.0% understood. Biology? Maybe we can round up to 5% understanding, but I think that is generous.

  2. mr mcknuckles says:

    The part about missense mutations being less common than thought reminds me of the possibly apocryphal story about UK bombers during WW2. Someone got the bright idea to armor the undamaged parts of bombers that barely made it back to base (because planes that suffered hits there were presumably over-represented in those that were shot down).

  3. Ed says:

    Very interesting. Does this only happen with paralogs, or are there broader mechanisms to upregulate non-homologous genes that are in the same biological pathway? Of course with biology, if a mechanism exists, it will be co-opted for other processes…

    I don’t know the details of the pathways studied in this paper, but most of the compensentory pairs seem to be paralogs (e.g. mutate hbegfa and hbegfb is upregulated). I wonder if that could be a transcript-specific feedback mechanism that spills over to “recently” duplicated genes, which have the same putative mRNA regulatory sequences. This paper starts to address that, finding multiple genes with compensatory upregulation for a given mutant, with “sequence similarity” in the 3′ UTR. However I don’t see that it’s demonstrated that all such compensatory genes are regulated by a mechanism upstream of translation.

  4. Barry says:

    Its reported that elephants:
    1-are relatively resistant to cancer
    2-have many copies of p53 in their genomes
    but over-expression of p53 has been shown to have deleterious effects in the mutants. This current paper suggests a mechanism whereby elephants can have those multiple copies but only express them one at a time at need.

  5. luysii says:

    I’m going to have a post fleshing out the letters shown below presently. The Nature paper above provides a new and rather generic way to treat genetic disease. The idea was not mentioned in the paper, nor Derek’s note, nor the comments so far. It appears to be original.

    It started after I got back from a great talk by Dr. Amy Wagers of Harvard on April 13th about Duchenne muscular dystrophy, something I dealt with for 15 years running a muscular dystrophy clinic in Montana ’72 – ’87.

    I wrote her twice but have heard nothing back.

    Here are the letters. Stay tuned for the post fleshing it out. Feel free to try the idea. I am retired with no academic affilitation.

    Sent 21 April 2019 @ 9;58 PM

    Dr. Wagers:

    I’ll have a post on this in the future. Consistent with the idea below is the mdx mouse has a premature termination codon in exon #23 resulting in a dystrophin containing only 27% of the amino acids of normal dystrophin. I spent a lot of time over the years reading papers on the mdx and wondering why the animal was NOT weak.

    I haven’t heard back from you, and I know you must get tons of Emails. Hopefully this isn’t winding up in your junk folder


    Begin forwarded message:

    From: Lewis Robinson
    Subject: A suggestion
    Date: April 17, 2019 at 12:17:09 PM EDT

    Dr. Wagers:

    I greatly enjoyed your talk last Saturday. I’m the guy who ran a muscular dystrophy clinic in Montana, and suggested that you look at satellite cells in patients with a variety of genetic dystrophies in atrophied and relatively spared muscles to determined why some functioned well and others did not.

    However, an even better suggestion and one involving far less work with CRISPR and AAV comes out of 3 papers in last week’s Nature (which I’ve just read since getting back home) — vol. 568 pp. 179 – 180, 193 – 197, 259 – 263. They concern Nonsense Induced Transcriptional Compensation (NITC), something I’d never heard of.

    The claim is that mutations with premature termination codons (PTCs) are less severe (and less common) than missense mutations, because of NITC, as PTCs result in nonsense mediated decay (NMD — don’t molecular biologists just love acronyms !) releasing fragments of mRNA which then get into the nucleus in some way with the help of the COMPASS complex. The fragments bind to similar complementary sequences in the genome for proteins with similar function, recruiting histone methyl transferases turning on a histone modification which make these genes more likely to be transcribed. Almost too incredible to be believed, but the evidence the papers present is pretty good.

    So rather than try to repair dystrophin with a large chunk of DNA inside an AAV, just use CRISPR to put in a PTC somewhere and let NMD do the trick. It should be easier to accomplish. Treating a genetic disease by making the gene worse certainly sounds paradoxic, but that’s what this work implies

  6. Ken says:

    I suppose it could be called ironic if it turns out we never really left the RNA world, and DNA and proteins are just adjuncts to the real biochemical framework of life.

Comments are closed.