Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Alzheimer's disease

Lost memories found

Enhancing synaptic connections between neurons in the brain's hippocampus that are normally activated during memory formation rescues memory deficits in a mouse model of early Alzheimer's disease. See Letter p.508

Some 44 million people worldwide are affected by the neurodegenerative disorder Alzheimer's disease or related dementia1. Episodic memory decline in the early stages of disease often precedes biological hallmarks such as the abnormal accumulation of amyloid-β and tau proteins. It is also better correlated than these traits with defects that occur in the synaptic connections between neurons in patients' brains2. However, whether this memory impairment reflects a failure to encode new memories or an inability to retrieve stored ones is not known. In a fascinating paper on page 508 of this issue, Roy et al.3 shed light on the matter, rescuing synaptic and memory deficits in a mouse model of early Alzheimer's disease by reactivating an ensemble of neurons that had been previously activated by experience.

Memory is a biological function that allows animals to encode, retain and retrieve information. The brain's hippocampus is a key player in these processes. Hippocampal damage can be assessed in mice using a behavioural model called contextual fear conditioning, in which animals learn to show defensive behaviour in a certain environmental context after it becomes associated with an aversive stimulus — for example, learning to freeze when placed in a box in which they have previously received an electric shock to the foot. Roy and colleagues found that mouse models of the early stages of Alzheimer's disease can encode memories of an aversive experience, but that their recall of the experience when placed in the same environment a day later is severely impaired (Fig. 1a).

Figure 1: Light-induced rescue of episodic memory.

a, During contextual fear conditioning, a mouse is placed in a box where it undergoes an aversive experience, such as a footshock (yellow arrow), which causes it to freeze. Healthy mice recall this experience and freeze when placed back in the same box (not shown), but mice that model Alzheimer's disease do not, indicating that they have long-term memory defects. b, Roy et al.3 genetically tagged neurons that were activated during the aversive experience with a light-sensitive ion-channel protein (tagged neurons indicated in blue). They then removed the mice from the fear-conditioning environment and used pulses of blue light to repeatedly reactivate these neurons. In doing so, the authors rescued long-term memory in the model mice.

Neuronal ensembles that undergo enduring biochemical changes during an experience, and that are reactivated during recall of that experience, are referred to as engram cells4. Using mouse models of early-stage Alzheimer's disease, Roy et al. modified a previously described protocol5 so that engram cells in the dentate-gyrus subregion of the hippocampus that are activated by an aversive experience are tagged with a gene that encodes a light-sensitive ion channel. The authors then used blue light to selectively open the channel, activating the engram (the population of engram cells in the dentate gyrus) and so manipulating the memory trace at a later time.

Light-induced reactivation of the tagged engram cells led to recall of the aversive memory. However, memory rescue did not extend beyond the duration of the light treatment, so it could not reinstate memory recall when, a day later, the animals were returned to the context in which they experienced the footshock (known as a natural cue). This finding suggested that the standard optical reactivation regimen did not strengthen the engram enough for it to have a lasting effect. Indeed, it is known that associative memory traces need to be consolidated — strengthened and stabilized — to facilitate future retrieval in response to natural cues6.

Consolidation of contextual fear conditioning involves long-lasting strengthening of the synaptic inputs to neurons in the dentate gyrus from neurons in another brain region, the entorhinal cortex7. Such strengthening is known as long-term potentiation (LTP). Roy et al. reasoned that the memory-retrieval deficit in the model mice came from an inability to strengthen these 'perforant path' synapses. To test this theory, they repeatedly activated the engram with serial high-frequency light pulses8, optically inducing LTP in the perforant-path inputs.

With this clever strategy, the authors rescued long-term memory deficits in the mice (Fig. 1b). A reduction in the density of tiny neuronal structures called dendritic spines, which receive synaptic inputs, is associated with memory loss in Alzheimer's disease, and optically induced LTP returned spine density to normal. Moreover, this strategy rescued deficits in two other hippocampus-dependent memory tasks: active place avoidance, in which an animal learns to avoid the specific place where it previously received a footshock, and novel object recognition, which tests recognition memory. This indicates that the engram-based technique can be generalized to improve recall for various memory types.

Next, Roy et al. found that optically inducing LTP in perforant-path inputs to a broader range of dentate-gyrus neurons did not rescue long-term memory. This finding is intriguing, because it suggests that simultaneous reactivation of multiple neuronal ensembles in the dentate gyrus cancels out the effects of reactivating a specific engram. Consequently, treatments such as electrical stimulation of deep brain regions, which are used to treat human neurological disorders but cannot discriminate between engram and non-engram cells, may not improve memory in patients.

Notably, a previous study9 showed that electrical stimulation of the perforant path increases levels of amyloid-β in the interstitial fluid around hippocampal cells. Further work is needed to determine whether Roy and colleagues' engram intervention increases amyloid-β levels, and whether the strategy can ameliorate memory impairments in late-stage Alzheimer's disease if combined with techniques10 to reduce amyloid-β levels and aggregation of tau.

To both tag and manipulate engrams in mice, Roy et al. introduced genetic constructs in two viruses — a strategy that comes with caveats. One construct contained a short, 1-kilobase promoter region, which drives gene expression in active neurons. In its natural state in the genome, the promoter drives c-Fos expression, but in the viral construct it promotes expression of an 'activator' gene that, in turn, drives expression of a second construct that encodes the ion channel. However, this promoter naturally acts in concert with enhancer elements that span the 50 kilobases of DNA surrounding it11. Excluding these gene-regulatory elements from the viral construct results in an incomplete engram, because some neurons that are activated by the aversive experience less strongly than others will not be tagged. The engram could be labelled with greater specificity by incorporating the activator into the genomic position of c-Fos, such that all the gene-regulatory elements can act in concert to tag neuronal ensembles in response to aversive experience.

In addition, regardless of promoter expression, the construct containing the ion channel can be activated only when an antibiotic called doxycycline is removed from the animals' diet. Roy and colleagues tagged engram cells for 24 hours from the start of contextual fear conditioning. This design lacks precision, so some nonspecific neurons are probably included in the tagged ensemble. Engram labelling could be optimized by decreasing the time for which doxycycline is removed from the diet, or by using an alternative engram-tagging strategy that allows a shorter time window for labelling12.

Nonetheless, the potential to rescue long-term memory in dementia is exciting. In the future, Roy and colleagues' findings might help to guide engram-based strategies that rescue memory deficits in patients with early-stage Alzheimer's disease.

Footnote 1


  1. 1.

    See all news & views


  1. 1

    Van Cauwenberghe, C., Van Broeckhoven, C. & Sleegers, K. Genet. Med. (2015).

  2. 2

    Terry, R. D. et al. Ann. Neurol. 30, 572–580 (1991).

    CAS  PubMed  PubMed Central  Google Scholar 

  3. 3

    Roy, D. S. et al. Nature 531, 508–512 (2016).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Tonegawa, S., Liu, X., Ramirez, S. & Redondo, R. Neuron 87, 918–931 (2015).

    CAS  Article  Google Scholar 

  5. 5

    Reijmers, L. G., Perkins, B. L., Matsuo, N. & Mayford, M. Science 317, 1230–1233 (2007).

    ADS  CAS  Article  Google Scholar 

  6. 6

    Abel, T. & Lattal, K. M. Curr. Opin. Neurobiol. 11, 180–187 (2001).

    CAS  Article  Google Scholar 

  7. 7

    Nguyen, P. V. & Kandel, E. R. J. Neurosci. 16, 3189–3198 (1996).

    CAS  Article  Google Scholar 

  8. 8

    Nabavi, S. et al. Nature 511, 348–352 (2014).

    ADS  CAS  Article  Google Scholar 

  9. 9

    Cirrito, J. R. et al. Neuron 48, 913–922 (2005).

    CAS  Article  Google Scholar 

  10. 10

    Leinenga, G. & Götz, J. Sci. Transl. Med. 7, 278ra33 (2015).

    Article  Google Scholar 

  11. 11

    Joo, J.-Y., Schaukowitch, K., Farbiak, L., Kilaru, G. & Kim T.-K. Nature Neurosci. 19, 75–83 (2016).

    CAS  Article  Google Scholar 

  12. 12

    Cazzulino, A. S., Martinez, R., Tomm, N.K. & Denny, C. A. Hippocampus (2015).

Download references

Author information



Corresponding authors

Correspondence to Prerana Shrestha or Eric Klann.

Related links

Related links

Related links in Nature Research

Neuroscience: The power of positivity

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Shrestha, P., Klann, E. Lost memories found. Nature 531, 450–451 (2016).

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing