The memory molecules - interview with Todd Sacktor (and a feature in Eureka)

The latest issue of Eureka, the Times's monthly science supplement, is out today. I've been incredibly supportive of the venture and it's great to see that a major national newspaper is increasing its science coverage, rather than cutting back on it. For this issue (the fourth, I think), I've written a piece on fear and memory, including a lot of research that I've previously covered in this blog.

While writing the piece, I interviewed a scientist called Todd Sacktor who's doing some fascinating work in this field. Sacktor discovered that a protein called PKMzeta is vital for storing memory. Remove it, and memories are deleted, seemingly irreversibly. I'm printing the full transcript of the interview here, as a sort of companion piece to the Eureka feature. Think of it like one of the extras on a DVD.

These are the sorts of services that I think modern journalists can provide for their readers, to expand the boundaries of an article well beyond the first capital and the final fullstop. It won't work in every case and time is obviously a factor, but there are exceptions when a scientist will be so eloquent and enthusiastic that it would be a crime not to print all of their words. There's plenty of golden material here that didn't make it to the final piece because of word limits or because it didn't fit in the narrative. Here, you'll hopefully get a fuller picture of PKMzeta. And for non-journalists, it might be interesting to see where I've pulled out quotes for the actual piece.

When people think about memory, they often think of discrete things like files on a computer that can be stored or lost. How do such metaphors stand now?

We think memories are stored by the action of PKMzeta at specific synapses. So the commonplace notion of files on a computer hard disk isn't that far from the truth now. In a sense, it's actually closer than the old neuroscientific explanation - that you have the growth of new synapses that, once grown, simply connect networks of neurons more strongly. It's a bit like branches of a tree getting thicker or denser and that's the memory - they are are now stronger because of these new physical connections. 

But a computer hard disk, the structure is there. The hard disk has a certain size and certain places for the zeroes and ones, but you can store different information in the pattern of zeroes and ones. PKMzeta shows that it's kind of a mix between these two notions. PKMzeta turns up at specific synapses after you learn something. The unique properties of this enzyme allow it to be active all the time (which is really unusual) and active at specific synapses, doubling the strength of those connections rather than their number. A synapse with PKMzeta is twice as strong as it otherwise would be.

Does the amount of the protein matter, or is it a binary thing?

We don't really know but my guess is that it is binary. Whether you have the PKMzeta or not determines whether it's a one or a zero at that part of the brain. The way we proved that is by injecting an inhibitor of PKMzeta called ZIP into the brains of lab rats. That erased the memories that were stored in that part of the brain, even for learning that had happened months before and even for really strong memories.

It's also a universal mechanism. It applies to all parts of the brain that store different types of memory like the hippocampus that stores place information, or the amydgala that stores fear memories, or motor memories in the motor strip. They're all using PKMzeta. It's a real revolutionary change in how neuroscientists have thought about memory. Everything before was pretty much an assumption, a hypothesis. PKMzeta is pretty much the only protein we know about that's relevant for the actual persistent storage of memory. 

The far-reaching effects of ZIP are fascinating but also quite worrying. When I last wrote about this, many readers were concerned about potential applications. Do you ever feel the same?

I was initially worried too, if it somehow fell into the wrong hands. It wouldn't happen for a long time in the future, and you never know what's going to happen. But recently, lots of people who study addiction memory have taken up on the idea that you could erase addictions if you inject it into areas of the brain like the nucleus accumbens that are important for addictions. There's also a condition called central neuropathic pain syndrome, where people catch their finger in the car door and even after the injury heals, a memory for the pain is set up in the central nervous system. ZIP could erase that too.

So there's a lot of potential good. It's not just about the dystopian fantasies of making zombies or toying with people's memories. I think the actual good is going to far outweigh the potential for bad.

And basically, we didn't really understand how memory works! Now, we have the first inkling of how long-term memories and information are stored in the brain. That's an important thing for understanding how we're humans. We are our memories - our mental states are based upon everything we've learned. You can't do the technological side without doing the basic research first. You can hope to treat addiction or post-traumatic stress disorder in a fundamental way until you really know how these processes work.

 How does the work on PKMzeta tie in with the research on propanolol last year?

I say that PKMzeta is all we know about the storage of memory. But from work in the past, we know a lot about the consolidation of memory - the transition from short to long-term memory. That's not just one molecule, it's hundreds, with many parts of the brain working in concert. There's a transition period for an hour or two after you learn something when it's not consolidated and when it's easy to prevent it from doing so. Adding the modern view from our research onto this, we'd say that it's easy to prevent the synthesis of PKMzeta, but once you make it, the memories become consolidated at specific synapses.

For memories that are stored with big emotional components, it's been shown for decades that you can block the consolidation of the memory - the initial transformation from short into long - by dampening the response to neurotransmitters that are released in parts of the brain involved in encoding emotional memories.

So norepinephrine, which is related to epinephrine (adrenaline), is getting released and affects receptors in the amygdala, which is important for the consolidation or storage of emotional memories. It's been known for decades that if you give propanolol to a rat that has memorised something motivated by fear, you could block that memory.

Karim Nader rediscovered an idea that had been around since the 60s but had lost favour. The idea was that if you recall a memory, it undergoes a process called reconsolidation. This means that it's easier to block - almost magically, you can prevent it from going through a second consolidation period. Propanolol is one of those drugs that can block both consolidation and reconsolidation of emotional memories.

That process is basically kind of magical! People have no idea what it means; it's just sort of an abstraction. Reconsolidation means it's easier to disrupt but no one knows what's getting disrupted. It was more like phenomenology rather than a mechanistic understanding. So that's what's great about PKMzeta - we know what the mechanism is. It's likely that the reconsolidation is blocking the resynthesis of PKMzeta.

So the idea is that when you recall memories, PKMzeta at specific synapses is broken down and needs to reform before the memory can be reconsolidated?

Right, and that's a hypothesis that people are working on right now. But it's a lot easier to come up with hypotheses like that, than it is to come up with hypotheses based on the old branches-on-a-tree idea of memory storage. When we recall a memory, does reconsolidation mean that branches are pruned? Well, it's possible, but it's easier to explain with PKMzeta.

Now it's also possible that the new synapses are being made, and there's evidence that this happens. But what's radically different about what we're saying is that only synapses involved in storing memory, rather than those that were formed during the development of the brain, require PKMzeta all the time. As soon as you add the ZIP, those synapses disappear or weaken. 

Now, it's like the tree has two different branches - red ones and green ones - and when you add ZIP, the red branches disappear and the green ones stay. 

Does propanolol have any effect on PKMzeta?

We don't know that yet, but people are looking at it.  

Isn't it a bit peculiar to have something as important as our memories under the control of a single molecule?  

I would argue that information storage systems, no matter what you look at, tend to be quite simple. You could say that it's crazy that cells just use DNA for storing genetic information. Shouldn't there be 200 different molecules? That's what people used to think in the 30s, when they thought it was proteins that were storing genetic information. But for information storage, you want to keep the medium of the information simple, with just one or two types of storage.

In a computer, a hard disk is using just one type of storage - magnetic charge on a spinning disc. Of course, the information that's encoded within the pattern of zeroes and ones is incredibly complicated. By analogy, it's the structure of the entire brain that's incredibly complicated and that gives you the type of memory that's being stored by PKMzeta. For example, the amygdala is important for fear, the motor cortex is important for movement and the visual cortex is important for seeing. Each of those is going to have its own complex language, but ultimately, the long-term store is still going to be a simple pattern of zeroes and ones - PKMzeta or not. The analogy for DNA is that you've got four bases - it could have been hundreds, but it's a small number.

Making it simple also allows you to continuously update it. Imagine you have a computer that uses weird drives and hard disks and handwriting - it may be impossible to organise the thing in a way that can respond quickly to changes in the environment. You want to stick with one system.

What's next for you in terms of PKMzeta research?

We know that PKMzeta is storing memories for at least 3 months but we think that the half-life of the molecule is around 11 days. So there must be some sort of positive feedback that keeps levels of PKMzeta up. But more importantly, you have to maintain PKMzeta at specific synapses. It's the activity of the enzyme itself that does this because if you add ZIP, the PKMzeta exits the synapse and once you wash the drug out, the enzyme can't go back. The enzyme's continuous activity is keeping it in the right place. We want to really understand how that works. In other words... well, I want to know how it works! [laughs]

There's still a lot of magic here. PKMzeta is staying in the same synapse probably for decades. It's not the same molecule but the population is being maintained at a high level for maybe a hundred years. And you can erase it within an hour or two with ZIP! That's what people are responding too - you can have a memory that last decades and then it's just gone? It's the contrast between the long-term stability, and the fact that you can make it unstable by just giving a drug for an hour or so. Getting the magic out of it is the goal. 

More on memory and fear: 

• The guardians of fear - molecules that provide safety nets for scary memories
• Rewriting fearful memories by bringing them back to mind
• 9/11 memories reveal how flashbulb memories are made in the brain
• Drugs and stimulating environments reverse memory loss in brain-damaged mice
• Beta-blocker drug erases the emotion of fearful memories
• Erasing a memory reveals the neurons that encode it