Critiquing LaPlant et al, in Nature Neuroscience, Part 3: The spines and the depression

Oct 20 2010 Published by under Addiction, Behavioral Neuro, Neuroscience

Welcome to part 3 in the series of my coverage of LaPlant, et al. 2010. It's been a long day, and Sci is TIRED. She just ran a 13 hour experiment, and boy is she wiped. But she is also DEVOTED. And also has her teeth well into this paper, and refuses to let go just yet.


So, two times ago, we discussed DNA methylation (an activity which determined whether your DNA is available for use), and how it was regulated by cocaine. In part TWO, we discussed the further work they did on how DNA methylation affects aspects of cocaine REWARD.

And now...spine density. LaPlant et al, 2010. “Dnmt3a regulates emotional behavior and spine plasticity in the nucleus accumbens”. Nature Neuroscience, 2010.

To understand what spine density is, and what it has to do with, well, anything, we have to go back to a neuron.

What I want you to focus on is the part on the far left. There you have the main body of your neuron (containing the nucleus, the DNA, all the big important stuff), and the dendrites which stick out from the main body. The dendrites are little protrusions of cells, which are responsible for receiving the chemical transmissions from other cells, and transporting them electrically to the cell body (for an overview on neurotransmission, and another one on the action potential, check out those links. They are quite pretty if I do say so myself. And I do).

So it's very important for these dendrites to be connected. And they often need to maximize the number of connections they can make, which means maximizing the SURFACE area, over which another neuron can form a synapse. So these dendrites are even more hairlike than the ones pictured above, because off of each one there are little offshoots, called dendritic SPINES.

So how is this relevant to this paper? Well, one of the most long-lasting changes that is induced by cocaine in an increase in dendritic spines in the nucleus accumbens, an area of the brain associated with the rewarding properties of drugs. Unfortunately, we don't know WHY this increase in denditic spines occurs, or what it means, really. But the authors of this paper wanted to know whether this increase in spine density had anything to do with the changes in DNA methylation (and the increases in DNA methyltransferase 3a, or Dnmt3a, which is an enzyme which INCREASES DNA methylation).

So they used a virus to transfer Dnmt3a into the nucleus accumbens. I explained this last time, but basically, we can use the cell-invading properties of a virus to our advantage, by taking out the virus nasty and putting IN the DNA that we want to express. We then inject the virus into a very limited area, and the virus will infect the cells, but instead of spreading virus nasty, it instead spreads the DNA that we are interested in, and the infected cells will make proteins from that DNA.

So in this case, the authors used this virus as a vector to increase Dnmt3a in the nucleus accumbens, and thus to increase DNA meythylation. And then they looked at dendritic spines, comparing the Dnmt3a with just cocaine injections.


Isn't that pretty? All glowy and stuff. So you can see that cocaine AND Dnmt3a increase dendritic spines in the nucleus accumbens. They also found that by BLOCKING DNA methylation (using the drug they used in the last post), they could block cocaine-induced increases in dendritic spines.

So it looks like DNA methylation from Dnmt3a (and from cocaine VIA Dnmt3a) increases spine density. This is a very cool finding, but...what does it mean? What does the increase in spine density mean FUNCTIONALLY? Unfortunately, they didn't do any behavior to correlate with the injections of cocaine they gave (they had to give five injections of cocaine on a different schedule, to get the increases in spine density they observed). And this is a little disappointing to me, because we still don't KNOW what spine density has to do with addiction, and I feel like this part of the paper just brushed the edge of it...and then backed off. But that's not their fault, they had other fish to fry.

And the other fish they had to fry was...depression.

Depression? What does that have to do with THIS?!

Well, the laboratory who wrote this paper has been interested for a while in the concept of the nucleus accumbens and its involvement in "emotional behavior". The basic idea behind this hypothesis is that changes in the nucleus accumbens going one way affect reward, and going the other way, may affect NEGATIVE reward aspects, like depression. This is based on the symptom of depression known as anhedonia, which is a lack of reward experienced in previously pleasurable activities. So they reasoned that DNA methylation one way might affect cocaine reward, while DNA methylation the OTHER way might affect depression.

So they set out to make mice depressed using a method called chronic social defeat. Basically, you take a normal, unsuspecting mouse, and place it in the cage of a bigger, aggressive mouse. The bigger mouse calls that cage his home, and is going to defend his territory by whatever means necessary. So the first mouse (usually called the intruder mouse) gets his butt kicked. You do this over a series of days, and the mouse getting its butt kicked shows signs of depression. So they performed this protocol, and found INCREASES in Dnmt3a. Observe.


You can see the expression of Dnmt3a went up as the mice were exposed to social defeat. And over-expressing Dnmt3a (using the viral vector method I talked about up there), caused a sub-maximal dose of social defeat to look much bigger than normal:


The dark bar on the left half is the social interaction with a test mouse after only one day of social defeat. You can see that one day doesn't really affect how much a mouse interacts with a test mouse, he still feels ok. The right bar, on the other hand, is when the animals have received Dnmt3a increases in the nucleus accumbens, and they tend to not interact as much, which the authors describe as prodepressive.

The authors also increased Dnmt3a in rats, and did a forced swim test. A forced swim test is where you put a rat in a bath which it can't escape. It'll swim around for a while, and then give up and float (don't worry, they only stay in for about 6min). Antidepressant drugs are known to decrease the amount of time animals spend floating, and increase the amount of time they spend struggling, which is thought to indicate an aspect of depression (though which aspect, no one knows).

Here's another instance where I REALLY wonder why they used rats. Why not use mice here? This seems a little thrown in.

Finally, because they saw increases in DNA methylation in depressive paradigms, they tried DECREASING DNA methylation, and seeing how that affected social defeat.


This shows the interaction time in mice exposed to social defeat with nothing (far left), decreases in DNA methylation (middle), and fluoxetine, an antidepressant (right). You can see that decreases in DNA methylation increased social interaction time, just like the antidepressnat did, and that you could therefore say that decreases in DNA methylation have antidepressant properties.

The authors concluded from this and all the other data in the paper that DNA methylation, specifically Dnmt3a regulates "emotional behavior".

While I think there are a lot of really good things about this paper (I <3 the first and second figures in general, I think they really support the story), there are some issues, and this is what caused us to get heated in our discussion. These are not issues like I normally would talk about when I cover a paper. These are the deep nitty gritter annoying issues that reviewers smack scientists over the head with, and which, while often being annoying, and often adding nothing, also often make a point.

1) Mice vs rats: sure, they needed the rats for the self-administration, but the forced swim test? That could EASILY be done in mice. I don't really know what the point of that was.

2) What was the point of the spine density experiments? They started to go somewhere and then fizzled out.

3) The results from the cocaine half and the depression half of the study don't seem to agree with each other. Increases in Dnmt3a increased pro-depressive behavior, and ALSO attenuated cocaine conditioned place preference. This seems really odd to me, as there are many studies showing that social defeat (esp in monkeys) increases drug taking, and so presumably would INCREASE conditioned place preference.

4) I'm not sure why the depression stuff was in here. Oh, I know it was because of the "emotional behavior" angle. But, to be honest, that story doesn't feel very tight. The data doesn't correlate well. I feel like this paper should have focused EITHER on cocaine, OR on depression. They had a beautiful cocaine angle, why not go with that? Use the social defeat, look at DNA methylation, and then look at cocaine place preference in socially defeated animals? What about using ICSS as a measure of hedonia in animals that get coke or Dnmt3a, or get socially defeated? Why not look at drug self-administration in those animals? Do their progressive ratio numbers go up or down? How does Dnmt3a REALLY affect the motivation to take cocaine?

In the end, while this paper has a lot of good data (and a lot of GlamourMag, Nature Neuroscience is definitely a GlamourMag in my field), I really don't feel like this is a tight story. With all the rats vs mice vs cocaine vs depression, it feels like this wasn't a planned series of experiments focused on a single hypothesis or experimental goal. It more feels like a bunch of data thrown together in a gamish, and a story erected around it to make it fit together. That's not to say it's a bad paper, it's not, it's a good one, but I feel like it could be better. They could have stopped with the dendritic spines, and added some behavior on to that angle and called it a day. They could have gone for ICSS or more self-administration. I'm not sure why they went the direction they went in, and why they felt this was the strongest angle for their data.

But what should we take away from this? That Dnmt3a is a molecule to WATCH. I've got my bets on that one.

LaPlant Q, Vialou V, Covington HE 3rd, Dumitriu D, Feng J, Warren BL, Maze I, Dietz DM, Watts EL, Iñiguez SD, Koo JW, Mouzon E, Renthal W, Hollis F, Wang H, Noonan MA, Ren Y, Eisch AJ, Bolaños CA, Kabbaj M, Xiao G, Neve RL, Hurd YL, Oosting RS, Fan G, Morrison JH, & Nestler EJ (2010). Dnmt3a regulates emotional behavior and spine plasticity in the nucleus accumbens. Nature neuroscience, 13 (9), 1137-43 PMID: 20729844

2 responses so far

  • Janne says:

    My guess on rats versus mice: they were asked to do a follow-up experiment, and at that point they had access to rats but not to mice (the original mice experiments may well have been run much earlier).

    Guess on the depression angle: one of the main contributors to the paper (perhaps one of the people running the experiments themselves) had a thesis (over-)due. The thesis is on depression, and they really needed another depression-related publication to fill in a gap in it. On the other hand, they needed a paper on cocaine like they needed a hole in their head.

  • Liz says:

    Depression as a pet interest of one of the main contributors? Eh, I dunno.

    It does seem that they've left quite a few interesting angles unexplored. Your point (3) sounds particularly interesting. You *could* suggest that social defeat increases depressive behaviour, and increases drug taking, but that taking the drugs doesn't necessarily make the mice any happier. I guess the best way to test how the effects work would be to take depressive mice, check that they do show increased drug taking (or possibly use rats, as they self-administer), and then see how their place preference changes.

Leave a Reply