How Memories Are Made: LTP and Engrams

Long-term potentiation, NMDA receptors, AMPA receptor insertion, protein synthesis and memory engrams: the cellular and molecular basis of learning — and how scientists have reactivated specific memories with light.

The idea that learning changes connections between neurons goes back to Cajal and was formalised by Donald Hebb in 1949: when one neuron repeatedly takes part in firing another, the connection between them strengthens. For decades that was a hypothesis. Then it was observed.

Long-term potentiation

In 1973 Timothy Bliss and Terje Lømo, working in Per Andersen's lab in Oslo, stimulated a pathway into the rabbit hippocampus with brief, high-frequency bursts. Afterwards, single test pulses produced much larger responses than before — and the increase lasted for hours, in some later experiments days or weeks. They had discovered long-term potentiation (LTP): a lasting, activity-dependent strengthening of synapses. Its opposite, long-term depression (LTD), weakens synapses.

The NMDA receptor

At many hippocampal synapses, LTP requires NMDA receptors. At rest, a magnesium ion blocks their channel. They open only when two things coincide: glutamate released by the sending neuron, and strong depolarisation of the receiving neuron that expels the magnesium. That's Hebb's rule built into a molecule. Calcium flowing through NMDA receptors activates enzymes such as CaMKII, which trigger the insertion of more AMPA receptors into the synapse — making it respond more strongly. Blocking NMDA receptors in rats impaired their learning of a water maze, as Richard Morris showed in 1986.

From hours to a lifetime

Early LTP uses existing proteins. Late LTP, lasting longer, requires gene expression and new protein synthesis via transcription factors such as CREB — paralleling Kandel's findings in Aplysia — and can involve growth of new synapses and enlargement of dendritic spines.

Engrams

In 1904 the German biologist Richard Semon coined the word engram for the physical trace of a memory. In 2012 Susumu Tonegawa's lab tagged the hippocampal neurons active while mice learned to fear a particular box, making them light-sensitive with optogenetics. Later, shining light on those cells made the mice freeze in a completely different box — reactivating the memory artificially. Engram research has since shown memories can be strengthened, weakened, linked and even "rescued" in mouse models of amnesia.