Andy Extance tells the astonishing story of the Arc protein and its capsid forms, and the questions it poses

Since its discovery in 1995, a protein called Arc has gradually revealed long-sought secrets about how memory works – but by 2013, Jason Shepherd was ‘a little bored’ of it. He’d worked on Arc since his undergraduate studies in 2001. By 2013, when he founded his current lab at the University of Utah, US, he thought ‘we’ve figured out most of what it does’. But within five years Shepherd’s team published what he called a ‘completely unexpected’ finding in the version of Arc that mammals like us have. At the same time, Vivian Budnik’s group at the University of Massachusetts, US, published a similar finding for the version of Arc found in flies. The findings reinforced an earlier suggestion that, millions of years ago, complex life forms found a way to use an ancestor of modern retroviruses like HIV to help us think.

In 2018, both teams showed that Arc can form itself into a capsid shell similar to a virus’s, and carry RNA genetic information between cells. This awesome idea surprised scientists in part because they didn’t have a full picture of Arc’s structure, and still don’t for the mammalian version. ‘The structure can give you a clue,’ Shepherd tells Chemistry World. ‘Which parts are sticking out of and into the capsid could determine the cargo.’ 

Since Budnik and Shepherd’s teams revealed the extent of Arc’s virus-like nature, researchers have revealed more about its structure and function. Some are trying to finally pin down a complete three-dimensional picture. Others meanwhile are working to further clarify how Arc helps us store memories when it’s not in capsid form. Shepherd’s team is preparing for experiments in animals that will show whether Arc’s capsid structure influences memory. Budnik’s group is hunting other capsid-forming proteins in complex life forms. In all these research directions, Arc’s discovery is poised to reveal more about how virus-like proteins make us what we are.

These are the latest steps in Arc’s progress into our main idea of how we learn, known as the synaptic plasticity and memory hypothesis. A synapse is where two of the brain’s approximately 85 billion neurons meet. An electrical signal can come in through a first neuron, and cause it to release chemical signals, such as glutamate neurotransmitters, into the synapse. The second neuron has receptors on the synapse’s other side to detect the signal. Depending on what’s happening to the second neuron, it may then send its own electrical signal. Whether the second neuron fires or not can change the synapse, and that changeability is called plasticity.

If the second neuron fires after the first neuron, it might gain glutamate receptors, making it more likely the second neuron will fire when the first one does in future. Scientists refer to this as strengthening, or long-term potentiation (LTP). If the second neuron doesn’t fire after the first one, then some glutamate receptors might be removed, making firing after the first neuron less likely next time. That’s referred to as weakening, or long-term depression (LTD). ‪Haruhiko Bito from the University of Tokyo in Japan calls the number of receptors sensitive to glutamate ‘a currency at the synapses’. When these changes persist, an organism’s neurons will fire the same way again when it has a similar thought process. Happening continually, on a vast scale, this process is a central part of how we form memories.

Survival of the fittest memories

In the 1990s, Bito helped show how a transcription factor called CREB turns on machinery used to make proteins from genetic instructions to help synaptic changes persist. Bito and other scientists wanted to find out which genes were being read by that transcription machinery. Arc ‘was one of the more interesting ones’, Bito recalls, because it was present in synapses and was being transcribed in large amounts.

For the strongest to survive, you need to remove the weakest

To form long-term memories, scientists believe that somehow synapses are tagged in a way that that makes them collect and use nearby proteins made during LTP and LTD. Researchers call this synaptic tagging and capture (STC). Unexpectedly, what Arc does is nearly the opposite, Bito’s team’s work shows. Instead of synapses involving at least one neuron that has been active getting tagged, in this case completely inactive synapses get tagged. Arc then helps weaken their connection, so that when a neuron fires, its neighbour is less likely to. ‘This is survival of the fittest,’ says Bito. ‘For the strongest to survive, you need to remove the weakest.’

Arc’s importance stretches far and wide, including into the critical period at the start of many organisms’ lives. Working with Shepherd’s team and others in 2017, Bito helped show that Arc genes are expressed to make protein at higher levels in young mice than older ones. Suturing one of their eyes shut during the critical period permanently worsens sight in that eye through LTD. Increasing Arc levels extended the mice’s critical period, and actually increased plasticity in mature mice, reopening their critical periods. Yet to date, researchers do not completely understand the precise biochemical details involved.

This is a problem that Clive Bramham’s team at the University of Bergen in Norway is now trying to resolve. Like Bito, the Bergen researchers came to Arc when investigating the brain’s protein transcription machinery. In 2002, the team found Arc synthesis was being turned on by BDNF, which stimulates CREB. Since then, they have studied Arc’s molecular function, using microscopy and x-ray scattering techniques. They have found that inside cells, Arc binds to glutamate receptors themselves. When it does so, it’s competing with other proteins that anchor the receptors to the synapse. ‘When Arc is made in the synapse, it would be in a position to dislodge receptors,’ Bramham explains. The receptors would be free to move around in the cell membrane or to be pulled into the cell through endocytosis.

Building up to capsids

This is just one of various roles for Arc. Many labs have shown it is a flexible protein that can act as an interactive hub, binding many different molecules. ‘It works by engaging specific effector proteins to mediate plasticity,’ Bramham says. Scientists knew even before the 2018 capsid finding that single Arc strands have roles on their own, or can grab other strands, forming pairs and groups of four. But otherwise it’s not clear exactly how most such functions operate at a molecular level.

An image showing the model of Arc function

Source: © 2017 Oleksii Nikolaienko et al

The Arc protein can change between three states, which alters its ability to act in helping synapses form memories

Arc has two parts that help it take various forms that play roles in synapse strengthening and weakening. The C-terminal domain is reasonably well-studied, and looks like the capsid protein of retroviruses like HIV. The mammalian form has a ‘nice, juicy, potentially druggable, hydrophobic binding pocket’, not present in its fly or virus form according to Bramham. ‘Discovery of this pocket by Paul Worley’s lab at Johns Hopkins University provides a big molecular clue as to how Arc regulates intracellular signalling and plasticity,’ Bramham says. The pocket’s structure helps show how it can bind different partners and act as a hub. It might one day be possible to design drugs to enhance or block interactions, he suggests. However, it is still important to understand even more about how it can bind so many different molecules.

The other part, the N-terminal domain (NTD), is the key to how Arc strands link together to build bigger structures. Using total internal reflectance microscopy, Bramham’s team can see how many Arc molecules assemble together at any one time. That showed that RNA triggers Arc’s assembly into capsid-sized structures, via a short segment in the Arc NTD, and is required for Arc binding to itself and forming capsids. ‘Although we don’t have crystal structure for the entire NTD, we obtained the crystal structure for the piece that mediates self-binding,’ Bramham says.

Budnik’s team came upon Arc capsids after studying synaptic plasticity in Drosophila fruit flies, and how the cells on either side of the synapse coordinate. They had found that tiny spheres of mostly fat and protein called extracellular vesicles were important in helping synapses develop. Extracellular vesicles can transport RNA across the synapse, Budnik explains. ‘They can change genetic function of the other cell,’ she says. The two versions of Arc RNA and Arc protein that Drosophila have were among the most common cargoes in these vesicles, her team found.

At the time, Budnik’s colleague Travis Thompson had just joined her group. He is an expert in biological pathways involving retrotransposons, genes that copy and re-insert themselves into their owner’s genetic code. They do this using reverse transcription, converting their RNA to DNA and integrating it with their host’s genome, and can form capsids. This is the same trick that retroviruses like HIV use. Some people think that retroviruses evolved from retrotransposons, Budnik notes. ‘They evolved, acquired these envelope proteins that allow them then to come out of the cell and into other cells to infect them.’ Such genes make up more than half of the human genome, and so are often overlooked as junk DNA, she adds.

Charting a detailed map

Thompson noticed that Arc’s amino acids looked like those in a retrotransposon capsid region of a retrovirus. Scientists had noticed this before, but never worked out why this was the case, says Budnik. ‘So we asked “Can we form a capsid?”’ she recalls. Her team made the protein by encoding its genetic sequence in bacteria, and made capsids that they could see by electron microscopy. The capsids also contained RNA, some of which encoded Arc. They then went on to isolate Arc capsids from extracellular vesicles. Before publishing these findings, Budnik saw Shepherd presenting his similar results at a meeting. ‘It was very interesting, because we had started from one point of view, and without any communication whatsoever, he, from the biochemical point of view, found the same thing,’ she says. ‘So we decided at this point to send our papers in parallel.’

Currently the Massachusetts team’s attention has turned to another retrotransposon gene that has an important job, called Copia. ‘This opens up this idea that maybe these things got domesticated for use by the cell in many different ways,’ she says. ‘They have evolved over billions of years and are so perfect that you use them in different contexts.’

Images detailing the virus structure

Source: © Simon Erlendsson et al/Springer Nature Limited 2020

Over 200 Arc proteins can assemble into a virus-like capsid

Simon Erlendsson and John Briggs at the Laboratory of Molecular Biology in Cambridge, UK, have recently shed more light onto that evolutionary history. As it became clear that Arc assembled into capsids, Erlendsson and Briggs teamed up with Shepherd to figure out the structure and function of Arc and its viral-like capsids. ‘In flies, we know both the structure of the full-length proteins on their own and the structure of the protein when it forms viral-like capsids,’ Erlendsson says. ‘The capsids contain 240 individual copies of the Arc proteins.’

The LMB researchers have also got 3D structures of the mammalian C-terminal domain from both crystallography and NMR. It comprises about half of Arc’s roughly 400 amino acid residues. ‘We are still, however, missing a detailed molecular map of the mammalian Arc capsids,’ Erlendsson admits. ‘Arc can bind RNA, DNA, lipids, other proteins and itself. It makes different complexes when it binds one or the other or all. While this makes perfect sense from a biological point of view, it makes them difficult to work with.’

Capsid challenges

The Arc gene was probably introduced into our genome more than 250 million years ago, Erlendsson explains, which makes it hard to know how exactly it got there. ‘It is, however, evident from our structures that the Arc protein and capsids are highly similar to long terminal repeat retrotransposons, such as Ty3, and retroviruses, such as HIV,’ he says. ‘One hypothesis is that failure in removing or silencing the Arc gene has led to a domestication of the gene.’ That means we have repurposed Arc into  performing other functions in our brains related to memory and learning. The possibility that retroviral-like signalling could be involved in one of the most precious functions of the brain is ‘just astonishing’, Erlendsson says.

Erlendsson is wary about the idea that this knowledge might lead to drugs, however. ‘Currently, it is difficult to predict how exactly drugs neutralising the effects of Arc would affect the physiology – the same binding pocket binds many different ligands,’ he says. ‘The pharmacological interest in Arc is mostly due to is ability to form viral-like capsids. The Arc capsid may be used to transport cargo in the brain in the same way that they transport and protect their own genetic material.’

We are still missing a detailed molecular map of the mammalian Arc capsids

Shepherd thinks that Arc may have therapeutic potential in re-opening the critical plasticity period. ‘You could use it for recovery after traumatic brain injury or strokes,’ he explains. ‘And we’re trying to figure out if this capsid cell-to-cell signalling is involved in Alzheimer’s.’

As well as fully understanding mammalian Arc’s 3D structure, Shepherd wants to know when and where capsids are made. ‘That’s hard because there’s probably not a lot of them that are being made at any given time, because you need a lot of protein.’ Part of the challenge is that each capsid might only contain four RNA molecules, which is far from enough to make the proteins needed for another capsid.

Shepherd’s next step is doing animal experiments to fully explore how the capsid’s structure determines its function. ‘We’ve done experiments in cultured neurons; the real question is going to be in brains,’ he says. Shepherd and colleagues have nearly completed developing the experimental systems needed. However he warns that they may find that it’s only simpler Arc systems involved in memory, not the capsid form. If the capsid is involved, ‘it opens up a lot of new avenues for thinking about how information is stored’, Shepherd says. ‘But until we show that, all bets are off.’

Whether or not our synapses often directly transfer information in virus-like packages, Arc has revealed new ways that biochemistry makes us what we are. We now know its ancient virus origins are not unique, as Budnik, Thompson and colleagues will soon publish research showing how creatures have harnessed other capsid-forming retrotransposons. Every new example will give researchers structural and functional mysteries to decipher. Perhaps they too will reveal more wonders like how Arc molecules help conjure the thoughts running through our heads.

Andy Extance is a science writer based in Exeter, UK