Almost autistic mice, part 2.

If we want to look for the autism mouse models, first we have to realize one basic caveat.

Mice can’t get autism.

A mouse doesn’t speak and doesn’t indulge in human-level sophisticated social interactions. Inter-mouse relations are mostly driven by the touch and smell, sight isn’t as important. Mice don’t have facial emotional expressions. Mouse focus and selection between varying stimuli is guided by a different set of rules. Mouse behavior can only deviate from a mousy norm.

What we can watch however are so called autistic-like behaviors. We can examine if one little rodent is more anxious than the other or how eager it is to explore novel surroundings. We can observe its communication with the other members of the species, including quality and quantity of the ultrasonic squeaks (yup, you can essentially do a whole PhD in mouse squeaks). We can see if there is repetitive behavior, for example excessive grooming. That’s about it if it comes to the symptoms.

Now we have to think which genes we’d like to turn off in order to trigger these symptoms.

I’ve been asked if it would be a good strategy to foul up their mirror neurons. After all, it has been described that mirror neurons aren’t working as they should in people with autism.

Actually this is not such a good idea, ant that’s why:
If you create an experiment like that, the question you ask isn’t “what can I do to get mouse models of autism”, only “what happens if you break mirror neurons in mice?”. In fact, these are at least three separate questions.

- Do mice have mirror neurons

- Which genes we have to turn off in order to break them (surgery wouldn’t be an option – at least in humans and monkeys we find mirror neurons all over the motor cortex)

- If we break them, do these mice exhibit “autism” or something entirely different (The mirror neuron hypothesis of autism is very much discussed. According to the latest research, autistic people don’t have a deficiency in this area.)

Getting back to our story, there are a few different strategies to develop mouse models for autism. Each has their pros and cons and probably they are all needed if we want to get a full picture. I’ll try to list them, bear in mind however that it’s not a perfect description.

1. Research on the syndromic autism

In some genetic syndromes autism is only one of the problems and not the worst one. Yes, it is possible to have that much bad luck. Those syndromes are caused by either a single mutation or loss/duplication of a chromosome fragment, and therefore a whole set of genes. Usually they are quite severe. Rett syndrome, Fragile X Syndrome (FXS), Apert syndrome or well-known Down syndrome (this is a special case, since there is a whole additional chromosome) don’t have much in common with the pop-image of the autistic savant genius. There is about 100 mutations associated with syndromic autism, some of them very rare.

We create a mouse model of one of these. We turn off or duplicate the same genes that are the cause of the syndrome in people.

Pros: We know the cause. We can do the same to mice. We are able to figure out which other biochemical pathways are broken because of the original mutation. We can look for the drugs that would alleviate the symptoms. For example the research on Fragile X Syndrome is very intensive at the moment. These syndromes probably aren’t ever to be completely cured, but if the patients cope better, it makes a great difference – for them and their parents.

Cons: these syndromes differ from the „vanilla” non-syndromic autism. It is possible that they don’t have anything in common in terms of biochemistry, just that the symptoms are alike. Additionally, in many of these syndromes autism isn’t always present. It depends on the genetic background. Which we know nothing about.

2. Research on the non-syndromic autism.

We take a group of people with autism – the more the better. We sequence their genomes (or parts of it). We check if there are some rare genetic variants, which are present in this group more often. This group would be ideally blood relatives or people of a very similar genetic background.
Then we turn off these genes in mice.
Actually there is already a lot of candidate genes for autism, so we don’t even have to sequence ourselves.

Pros: If it works, we have a more-or-less autism mouse model. Now we can culture the neurons, regulate the biochemical pathways with some drugs and figure out which would alleviate the symptoms. It might even give interesting results.

Cons: We mutate some mice… and nothing happens. Or something happens, only this isn’t autism at all. After all, non-syndromic autism is thought to be multigenic, caused by the accumulation of some genetic variants. Or simply it doesn’t work because a mouse is different from a human. D’oh!

3. Research on the inbred mouse with autistic-like symptoms

We take some mice. We inbreed them as long as we get the animals that sit in the corner of their cages and don’t squeak to each other. Then we look at whatever we have selected.

Pros: There will be papers out of it.

Cons: Seriously, does it have anything to do with the human autism? There is a hell lot of genes that influence the nervous system in different ways and the same symptoms might be caused by the different triggers. Fragile X Syndrome and tuberous sclerosis mutations, both known to cause autism, are - biochemically - opposites. We have autistic-like mice, derived by some warped kind of convergent evolution, but does it tell something about the universal mechanism?
Well, people think it might and do just that.

Phew. I have just provided a quick sketch on the topic, without all the technical details, for example what can get broken at the synapse. There are books and symposiums on it. We know a little bit about the biochemical pathways, receptors and other proteins involved in autism – or rather, autisms. Just like I wrote in the previous post, it isn’t a single disorder. If we ever start to treat it with drugs, it will be a personal composition for each patient. Once the prices of whole genome sequencing drop to a reasonable level, it will become possible.

And that alone will bring a new set of challenges. Even if we do not mention the insurance costs.