Tweaking the epigenome may erase a haunting fear

TLDR: Old fear memories are hard to erase because of a protein called HDAC2 that shuts down the expression of neuroplasticity genes required for memory updating. Inhibiting HDAC2 before recall and extinction training persistently erased a remote fear memory in rats. HDAC2 inhibitors are in clinical trials for cancer, but currently lack the specificity needed to treat PTSD.

Quite a few science bloggers have expertly written above this study, but mostly from a "new drug to treat PTSD" angle. If you're interested in the molecular bonanza that allows you to form (and erase) long-term memories, read on! Image: A Fearful Eye by KarotheFox on Deviant Art

Let’s zoom in on the inner workings of a memory.

Within minutes after experiencing something terrifying, dozens of proteins begin shuffling around in your synapses, frantically transcribing the experience into a new fear memory. Nascent memories are fragile and easily disrupted; to stabilize (“consolidate”) them, the neuron constructs many new proteins involved in synaptic signaling, either in the cell body or directly at the synapse. These reinforcement plasticity components unite at the synapse, and together with other “memory molecules”, change the strength between synaptic connections. A new long-term fear memory is thus steadily stored.

However, focusing on synaptic remodeling only gives you half of the picture. Far away from the synapse, deep inside the nucleus, chemical modifications to the DNA (and its packaging material) are changing how certain genes are expressed. In order to fit into the nucleus, long strands of DNA wrap around a group of proteins called histones, like thread around a spool. When DNA is wound tightly, it makes it difficult for RNA-making machinery to access the gene – and so the cell can’t make the protein.

Certain stimuli from the environment (such as a traumatic experience) trigger the addition or removal of small chemical groups from histones, briefly changing its structure so that the DNA strand can loosen enough to allow gene transcription. Past studies have shown that epigenetic changes in the hippocampus after learning help lay down long-term fear memories.

What if similar tweaks to the epigenome could also help remove traumatic memories?

Who’s in charge of memory updating anyway?

Say you train some rats to fear a cue by pairing it with a small electric shock – soon the rat freezes in fear at the cue alone. To get rid of the memory, scientist would use something akin to exposure therapy in humans: they repeatedly present the cue (say, a tone) without the shock. The reason extinction training works is that when a memory is recalled, it temporarily transforms into a labile state vulnerable to change. During this period, the rat learns the tone is safe – and loses the fear.

Unfortunately, extinction training only works on recent (day-old) memories; wait 30 days and no amount of relearning can permanently ease the rats’ fears. Since recall opens up a period of neural plasticity for memory updating, researchers in this study reasoned, perhaps old memories simply can’t induce the same degree of plasticity required for modification.

A bit of molecular sleuthing confirmed that they were on the right track. Recall of a recent memory increased the expression of cFos (left graph below, "1d"), a gene regulated by neuronal activity and critical for neuronal plasticity. However, cFos upregulation didn’t occur after retrieving old memories ("30d"). The culprit seemed to be HDAC2 (histone deacetylase 2), an enzyme that “parks” near parts of DNA that promote gene expression (aptly called “promotors”). HDAC2 removes acetyl groups from histones, which causes DNA to bind more tightly to the histone core and limits transcription.

S-NO of HDAC2 dissociates it from its gene targets (for example, cFos), functionally "inactivating" it in terms of modifying gene expression. The enzyme itself still retains its activity.

When the rats recalled a recent memory, neuronal activity triggered chemical modification of HDAC2 and transiently dissociated it from its gene target (right graph, S-NO causes dissociation), which allowed transcription of plasticity genes to stay on. However, when recall was delayed, HDAC2 stubbornly stuck to its job – it continued to remove acetyl groups, effectively shutting down the production of neuroplasticity proteins.

But is HDAC2 the reason why old fear memories are so persistent, or is it only an unlucky correlate?

Rejuvenating an old memory (…sort of)

Researchers first artificially activated HDAC2 (and thus limited gene translation) in rats with a non-specific drug* before prompting the rats to recall a recent memory; this somewhat mimics the molecular state of retrieving an old memory. After extinction learning, while these rats did forget the fear, relief was brief: soon the day-old memory returned, as if it were “old”. Conversely, inactivating HDAC2 – mimicking recall of a recent memory - before remote memory retrieval erased it for good. Thus HDAC2 seems to act like a spring-wound timer: set a time, and the memory becomes vulnerable to change in the interim. As the memory ages, the timer freezes at 0, no longer able to open a window in time for memory updating. (*They used a drug called L-NAME that inhibits NO synthase, which is an enzyme required to dissociate HDAC2 from its targets. NO synthase nitrosylates HDAC2 roughly an hour after recall, which correlates with the memory reconsolidation window. For inactivating HDAC2 they used MOL. Both drugs have many targets other than HDAC2.)

Here's where it gets sexy. Researchers repeated this experiment with a more specific drug: CI-994, an HDAC inhibitor that selectively inhibits HDAC2 (and a few other HDAC2-like proteins). This hot class of drugs is currently in clinical trials for many types of cancer and neurodegenerative diseases. Once again, old memories were transformed into a recent memory-like labile state: recall and extinction amped up brain metabolism and neural activity, boosted the expression of several major neuroplasticity genes, strengthened connections between memory-related hippocampal neurons and wiped out the old, haunting memory. The learned fear memory did not return, even 30 days later (see graph below).

The remote fear memory is strong after recall (white), diminishes after extinction training (side dash) but pops back 30 days later without any intervention (black, left). CI-933 prevents the memory from returning (black bar, middle graph), but only if the memory was recalled before extinction training (black bar, right graph - without recall, CI-933 doesn't work, presumably because the memory wasn't made labile)

Amazingly, HDAC2 inhibitors worked their magic only on one fear memory at a time. A group of rats learned to fear both a tone and a box. Roughly a month later, they received the drug right before recalling one of the fear memories. CI-944 only erased the recalled memory, but spared the other.

HDAC inhibitors for PTSD? 

One obvious application of the drug would be in post-traumatic stress disorder (PTSD), but making the leap from mice to men begs caution. HDACs are expressed all over the body and regulate the expression of many genes at a time, all of which would be disrupted by an HDAC inhibitor. Perhaps a more sophisticated PTSD therapy would be to specifically block the interaction between HDAC2 and key neuroplasticity genes. That is, if we can find the crucial few.

Another curious thing is the massive S-NO of HDAC2 (helping it dissociate) in the hippocampus after recall of a recent memory, so much it can be measured by a rather crude technique. Generally memory storage is considered to be quite "elitist", involving only a selected few synapses. Can activity of only a few synapses at recall truly be responsible for such as big change?

Gräff J, Joseph NF, Horn ME, Samiei A, Meng J, Seo J, Rei D, Bero AW, Phan TX, Wagner F, Holson E, Xu J, Sun J, Neve RL, Mach RH, Haggarty SJ, & Tsai LH (2014). Epigenetic Priming of Memory Updating during Reconsolidation to Attenuate Remote Fear Memories. Cell, 156 (1-2), 261-76 PMID: 24439381