“We had a behavior that could be measured as a physical property”

Hopi Hoekstra: "For us, finding genes that contribute to naturally evolved behaviors represents really a first". (Photo Rosa Reis)

Did some genes evolve to “tell” a little desert mouse whether to dig a big or a small burrow, and whether to equip it with an emergency exit? Did other genes evolve to modulate a male’s capacity to be a good dad? Hopi Hoekstra’s work on the genetic basis of complex animal behaviors strongly suggests this is actually the case. Ultimately, this work can lead to a better understanding of how genetic variation and dysfunction underlying social behaviors impacts the wiring of the human brain.

Hopi Hoekstra, professor of Zoology at Harvard University, geneticist, evolutionary biologist, was born in California to Dutch parents. And she is not your garden-variety scientist. “I’m definitely not one of those people who as a young child says she wants to grow up and be a scientist”, this soft-voiced, motherly-looking woman told us during a visit last September to the Champalimaud Centre for the Unknown, where she had been invited to give a talk during the 2016 Neuroscience Symposium.
But there’s much more to her originality: a keen volleyball player when she started college, she initially planned on studying political science – and even admitted having hoped to become US ambassador to the Netherlands. Halfway through it, though, she switched to biology. It was only later, though, after joining a research lab, that she got “hooked by the context of passionate discovery” of a job that entailed finding things nobody yet knew – a privilege at which she still marvels today.
So first she went into biomechanics research (“running cockroaches on treadmills”, she says good-humoredly), then into fieldwork on grizzly bears at Yellowstone National Park. And finally, she ended up “somewhere in the middle” working on mice, which she has been doing ever since. Her main job in the last few years has been to try to unravel the genetic evolution of behavior.

You’ve been studying the genetics of the evolution of complex behaviors in mammals, including social behaviors, for about a decade. What drew you to this?

My interest came initially from the idea that we could link genes to phenotypes [the observable traits of an organism]. We started by looking at morphological traits, like pigmentation or the length of particular bones or appendages, that are much easier to measure than behaviors. Then, as we became more and more successful at understanding how genes work to change morphology, we got emboldened and moved into behavior – which, one could argue, is both more complicated and more interesting.

One of your latest works concerns burrow-digging behavior in different species of American desert mice.

Yes, and one of those reasons we started on this particular project was precisely because there were hints that it could be inheritable. In fact, natural historians had shown that you could take these mice and even if they had never seen dirt before – and maybe even if their parents and grand-parents hadn’t seen dirt either –, when you put them in boxes filled with dirt they would build burrows. So this suggested, at least, that they hadn’t learnt this trait from their parents or grand-parents, and thus hinted at a genetic component.

How can you know to what extent a complex behavior – and especially a social behavior – is inherited, innate?

One way we can measure heritability is to remove opportunities to learn, for example. In the case of burrowing, as I mentioned, you can take away the dirt and see if future generations can still burrow.
In the case of social behaviors, we can have the pups raised by a different type of parent and ask if that influences the way those mice then behave as adults. So, by manipulating their learning environment, we can start to disentangle nature from nurture, so to speak.
I think it’s very clear that, for many behaviors, you have both a genetic and a non-genetic component. I would say that there is a spectrum, and behaviors are on one side of the spectrum or the other. And because, as geneticists, we’re interested in finding the genes that contribute to behavioral variation, we tend to focus on the things that we think are more inheritable [she laughs].
We test that in the lab before we start doing our genetic experiments, and if there are hints that there is a genetic component, then we continue.

You studied architectural differences in burrow-digging between two species of desert mice. Why?

The three main reasons we focused on burrow-building behavior is not that I have a particular love of burrows or dirt [she laughs].
For one thing, there were consistent differences between the two species for this behavior: one species digs a little burrow, while the other digs a big burrow with an escape tunnel – which is much more complicated. This is really interesting.
The second reason is that you could bring the animals into the lab and they would reproduce those behaviors, enabling us to control the environmental effects in the lab better than in the field.
And the third reason that we looked at this particular trait was that we had a behavior that could be measured as a physical property.

How do burrow architectures differ?

The two species of mice that we focused on live in very different habitats and the burrow architecture is correlated with environmental differences. The one species that builds a little teeny burrow, that looks like a wool sock – with a little entrance tunnel and a nest chamber – lives in very heterogeneous environments, and to evade a predator they can run up a tree or under a rock, into a crevice or some other hiding place.
By contrast, the species that digs a burrow with a long entrance tunnel, a deep nest chamber and a secondary tunnel that radiates up towards the surface but doesn’t break it (and serves as an escape tunnel), lives in very open environments. They don’t have really any places to evade their predators, so it makes sense to invest a lot in digging a burrow that’s a really good hiding place, with a secret back tunnel escape hatch in case something comes into the entrance tunnel.

Behavior Caught in a cast

You said you can measure this behavior as if it were a physical property. How?

One of the nice things about burrows is that we can transform the behavior into a sort of morphological representation. This idea was first put forward by Richard Dawkins in a book that he published in 1982, entitled The Extended Phenotype.
The idea is that if you have a behavior that has a heritable component, and if that behavior leads to building an architecture – whether it’s a bird building a nest, a spider building a web, a termite building a mound – then you can genetically dissect that architecture just like you could the length of your leg or your hair, the color of your skin, etc.
In our case, if you put a mouse in a box and leave it alone for a day or two, you come back, take the mouse out and what you have left is precisely that: a physical representation of the behavior.

Measuring its properties still seems a bit awkward…

Yes. If you do it directly, it’s going to be sort of a sloppy measure. So to circumvent this, we made casts of those burrows and then we measured them like we would measure the length of a femur, for example. We can measure a mouse’s burrowing behavior as simply as with a ruler.
For that, we used a fantastic idea by a graduate student I had, Jesse Weber. Our high-tech approach [she laughs] was to use a polystyrene foam – an insulating foam you can buy in a store when you need to fill a hole in a wall. We injected it into the burrow, where it expanded and filled the space. Then, over the course of a few hours it hardened and we could dig the cast out, pull it out – and there we had the behavior.

Once you had a way to measure it, how did you study the genetics of this behavior?

One of the fun things we did was to take a little burrow digger and cross it, mate it, with a big burrow digger, to see what the first-generation hybrids, which got half their gene complement from the little-burrow-digging parent and the other half from the big-burrow-digging parent, would do.
We were taking bets in the lab, we had no idea what was going to happen. We put each of these mice into one of our burrowing chambers, let them do their thing for two nights, and came back eager to see what they had done. And, quite to our surprise, these mice had dug big burrows, just like one of their parents would. That meant that the alleles (or versions of genes) that control the behavior are largely dominant.
Next, we took those first-generation hybrids that dig perfectly digged burrows, with an escape tunnel and everything, and crossed them with little burrow diggers again. Here is where we got a mixing of the genomes (with each individual having a different combination of alleles from each of its parents), and started seeing all sorts of possible architectures in the burrows of those resulting second generation hybrids.

So this complex behavior is largely controlled by the genome.

Yes, I think that’s true, that it seems to be a largely heritable behavior.

Were you surprised?

To some extent. We had high hopes that this would be the case, but then actually seeing it in action was another thing.

You identified three DNA regions linked to digging long (versus short) burrows and one region linked to digging (or not) an escape hatch.

Yes. One of the things that we found when we looked at these second generation hybrids was that the genetic basis of the behavior is actually quite simple. That was a surprise, because the architecture of the big-burrow-digging species has an entrance tunnel and a very clever escape tunnel, so we sort of intuitively thought the genetics must be complex.
That we could narrow it down to just three regions of the genome associated with making a longer tunnel – and even maybe just one gene that says either they make an escape tunnel or they don’t – was quite surprising.
Also, the fact that these genes are on different chromosomes suggests that this behavior is in some sense modular, and that in those hybrids, you could get for example some individual mice that dug short entrance tunnels with an escape tunnel, while others dug long tunnels without an escape tunnel. So it does suggest that you could genetically dissociate the two.
Even more interesting, if we look at other species in this same group of mice, there are actually species that dig long burrows without an escape tunnel, for example. So it’s fun to imagine that perhaps the genes associated with entrance tunnel length evolved early and maybe have been showing up in different species. And then, as an add-on, you have the evolution of the escape tunnel gene.

You don’t know yet which one evolved first.

No, but once we identify those genes, I think there’s some hope we will be able to figure out the time in which those alleles arose – which would be fantastic. That’s because what we are really interested in is how things evolved in the past, but what we have is just the present to try to infer what happened in the past from the evolutionary point of view.

You haven’t identified any genes yet.

That’s right. So far we have identified three genetic regions, and those genetic regions clearly contain genetic changes that contribute to the behavior. But we don’t know, within those regions, if it’s one gene or several genes.

Good Dads, Bad Dads

You are now studying a complex social behavior in the same two species: the quality of parental care.

Yes. Our same two species that dig small and big burrows also happen to have different mating systems. One of our species of mice is promiscuous, which means that the litters from this species will have more often than not been sired by multiple males (the female will mate with males in rapid succession).
Promiscuousness is typical in most mammals. What’s less typical is that our other species of mice represents a case of monogamy. Not only do they act like they are monogamous, but if you check paternity, it’s really true that they are monogamous.
Parental care is correlated with these differences. So the first thing we asked ourselves was whether the different mating systems were actually associated with different parenting behaviors.
My post-doc Andrés Bendesky designed a very elegant behavioral assay and showed that monogamous mice are more parental in general, and that dads are just as good as moms at caring for their young. This is in stark contrast with the promiscuous species, which overall is less parental, and in which the moms are ok but the dads are very poor at providing care.
So we again showed that these differences are likely to have a strong genetic component.

And how are you going to “play” with this behavior?

For us, finding genes that contribute to these naturally evolved behaviors represents really a first. Once you have a gene in hand, you can ask how this works and understand how those behaviors evolved.
But what is potentially even more interesting is that we can ask how these genetic changes, in the case of behavior, act through neural circuitry to actually produce the behavior. And so the gene gives us a sort of a handle with which to start exploring the brain.
One simple question I think we don’t know the answer to is whether these genetic changes rewire the circuitry in the brain or just modulate that circuitry. That’s a fundamental question and with the genes in hand, we can start to see how they affect the brain quite easily. So I think the fun part is still to come.

Do you think it will be possible to untangle the genetic evolution from the cultural evolution of many behaviors?

Yes. But if you had, let’s say, a behavioral difference that was controlled by a hundred genes, then your power to find any one of those would be very small.
That’s why we do tend to favor traits where we think the genetic basis is going to be simple. We don’t yet have the power to find all the other “little” genes that matter. But we’re making great progress.
Perhaps even more importantly, we’re getting better at measuring behavior. Nowadays, we can do that in a completely automated, unbiased way, where we use computer learning and automated tracking systems so that measuring our behavior is not only more reliable and more unbiased, but we could imagine doing it on thousands of mice. I have high hopes for the field moving forward at a rapid pace.

If some human behaviors could actually be genetically untangled, would that have medical implications?

Absolutely. I think the more we understand the genetic component of behaviors the more enlightening it’s going to be when looking at human diseases. Social interactions are extremely important, and their dysregulation or dysfunction can certainly affect human health – in autism, for example.

Would you say complex human social behaviors are generally heritable?

I do think that it’s always interesting to take the step and ask whether similar things could be going on in humans. And although we have to tread carefully, I feel that there are certainly going to be genetic changes that contribute to behavioral differences even among human individuals. That seems pretty clear to me, but we don’t yet have a good handle on what those genetic changes are, on how much they matter – and on how they affect human behavior through neural circuitry. That’s going to be an exciting field.

Your passion for research dates back to when you first worked in a lab as a student. Is there something you particularly value about the way you do research at your own lab today?

The thing that I love most about our lab is the collaborative feel. The work in my lab tends to be very integrative. Even in a single paper, we can have data from wild populations and more ecological data all the way down to hard core developmental biology or neurobiology.
It’s very difficult for any one person to span that entire spectrum, and that’s why our lab tends to be very collaborative. It’s people with different expertise working together to try to formulate complete stories.
I think that if you get somebody who has been trained as a developmental biologist together with an ecologist, beautiful things can happen.



Ana Gerschenfeld works as a Science Writer at the Science Communication Office at the Champalimaud Neuroscience Programme



Edited by: Catarina Ramos (Science Communication office)
Photo Credit: Rosa Reis


Loading Likes...