Postsynaptic burst reactivation of hippocampal neurons enables associative plasticity of temporally discontiguous inputs
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
This manuscript, contains fundamental findings that substantially advance understanding of an important research question, mostly uses appropriate and validated methodology in line with the current state-of-the-art, with good support for the claims, and the message of the manuscript will have a profound and lasting influence on neuroscience. In essence, the manuscript reports that dopamine converts spike-timing-dependent synaptic depression into potentiation that requires cAMP/PKA second messenger cascade and protein synthesis. The mechanism enables a separate synaptic input to induce heterosynaptic potentiation in previously primed synapses, which is shown in a network model to have desirable computational properties. The significance of the findings is threefold: First, it is the longest-lasting synaptic eligibility trace identified so far; second, the mechanism enables memory linking between temporally separate events; and third, it indicates a novel function of postsynaptic reactivation events. In addition, the finding may inspire new reinforcement learning algorithms in machine learning.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
A fundamental unresolved problem in neuroscience is how the brain associates in memory events that are separated in time. Here, we propose that reactivation-induced synaptic plasticity can solve this problem. Previously, we reported that the reinforcement signal dopamine converts hippocampal spike timing-dependent depression into potentiation during continued synaptic activity (Brzosko et al., 2015). Here, we report that postsynaptic bursts in the presence of dopamine produce input-specific LTP in mouse hippocampal synapses 10 min after they were primed with coincident pre- and post-synaptic activity (post-before-pre pairing; Δt = –20 ms). This priming activity induces synaptic depression and sets an NMDA receptor-dependent silent eligibility trace which, through the cAMP-PKA cascade, is rapidly converted into protein synthesis-dependent synaptic potentiation, mediated by a signaling pathway distinct from that of conventional LTP. This synaptic learning rule was incorporated into a computational model, and we found that it adds specificity to reinforcement learning by controlling memory allocation and enabling both ‘instructive’ and ‘supervised’ reinforcement learning. We predicted that this mechanism would make reactivated neurons activate more strongly and carry more spatial information than non-reactivated cells, which was confirmed in freely moving mice performing a reward-based navigation task.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
This is a very interesting paper showing that postsynaptic bursts in the presence of dopamine produce input-specific LTP in hippocampal synapses 10 minutes after they were primed with negatively coincident pre- and postsynaptic activity. LTP requires NMDAR activation during priming and involves a cAMP-PKA cascade and protein synthesis. When this synaptic rule is incorporated into a computational model, reinforced learning is possible through selective reactivation of neurons. Experiments in behaving mice confirmed that neurons reactivated after an exploratory period display more activity than non-reactivated neurons.
We thank the Reviewer for their positive comments on our manuscript. We have incorporated the Reviewer‘s suggestions.
Reviewer #2 (Public Review):
Building on their …
Author Response
Reviewer #1 (Public Review):
This is a very interesting paper showing that postsynaptic bursts in the presence of dopamine produce input-specific LTP in hippocampal synapses 10 minutes after they were primed with negatively coincident pre- and postsynaptic activity. LTP requires NMDAR activation during priming and involves a cAMP-PKA cascade and protein synthesis. When this synaptic rule is incorporated into a computational model, reinforced learning is possible through selective reactivation of neurons. Experiments in behaving mice confirmed that neurons reactivated after an exploratory period display more activity than non-reactivated neurons.
We thank the Reviewer for their positive comments on our manuscript. We have incorporated the Reviewer‘s suggestions.
Reviewer #2 (Public Review):
Building on their previous 2015 study with Brzosko, Fuchsberger et al. propose a potential solution for how the brain associates with memory events that are separated in time. The authors find that in the presence of dopamine, postsynaptic bursts produce input-specific LTP at hippocampal CA3-CA1 synapses ten minutes after priming with a post-before-pre spiking-pairing protocol. They explore the signalling somewhat, for example showing a need for postsynaptic NMDARs as well as for protein synthesis. Using a computer model, they find that this form of plasticity enables reinforcement learning. A few key predictions were verified using an in-vivo spatial learning model.
This is a strong study that addresses a long-standing fundamental problem in modern neuroscience research, namely the temporal credit assignment problem of how temporally well-separated signals can be meaningfully associated and learned in the brain. The experiments are carefully executed, the rationale is clearly explained, and - excepting Fig 6-8 - the figures are for the most part easy to understand. The study ranges from in-vitro electrophysiology across computer modelling to awake-behaving in-vivo experiments to persuasively argue that their novel findings may provide a candidate solution to the temporal credit assignment problem. Taken at face value, this work is likely to be highly impactful, however, some control experiments were missing or are perhaps just not shown (e.g., stability, stability in the presence of anisomycin, the effect of anisomycin on firing, and similar), which makes the validity of the findings a bit hard to evaluate at times.
We thank the Reviewer for their positive evaluation of our study and address all the points raised below.
Reviewer #3 (Public Review):
Fuchsberger et al. demonstrate that an otherwise LTD-inducing STDP protocol can produce LTP if followed by burst reactivation of post-synaptic neurons in the presence of dopamine. Using computational modeling and single-photon imaging in the CA1 in mice, they propose these findings are relevant to spatial over-representation at a reward location.
This is a follow-up of the two previous studies from the same group (Brzosko et al., 2015 and Andrade-Talavera et al., 2016) where they showed a post-before-pre STDP protocol, which by default induces a (pre-synaptic) LTD, will induce synaptic potentiation in the presence of dopamine and continuous synaptic activity. The main conceptual difference between this manuscript and these previous studies is that continuous synaptic activity can be replaced by post-synaptic burst. This means that reactivation of post-synaptic neurons without any further pre-synaptic instruction is sufficient for successful LTP induction.
Mechanistically, the two protocols (continuous vs burst activation) appear to be similar (but not identical). For example, both require the activation of post-synaptic NMDAr during STDP pairing, and both depend on the AC/PKA pathways. Additionally, there are two new observations here: The activity of voltage-gated calcium channels during bursting is required for potentiation; the burst-induced potentiation also requires protein synthesis.
The evidence provided at this stage is strong.
Major point:
It is not clear to me how the STDP studied here relates to the next part of the study, the reward-based navigation task. My interpretation is that the authors consider the activity before reaching the reward location (approaching time) as resembling the STDP priming protocol, the activity at the reward location as equivalent to the bursting protocol, and consumption of the reward as similar to dopamine application. If so, what is the circumvential evidence that the activity during the approach induces any form of plasticity?
The link between the two is not obvious and I see the manuscript as two interesting but not naturally linked stories.
The Reviewer’s interpretation is correct. We considered the activity during navigation on the maze as the animal approaches the reward resembling the STDP priming protocol. Substantial evidence supports a role of NMDAR-dependent STDP in the formation of place fields during navigation (Mehta, Hippocampus 2015; Moore et al., 2021). It has been postulated that both LTP and LTD are involved in place field formation. This was based on the observation that place fields shift backwards with experience (Mehta & McNaughton PNAS 1997), and a computational model predicted that without LTD place field broadening would occur (Mehta et al. Neuron 2000). Thus LTP is required when entering the place field, and LTD when the animal exits the place field (Mehta et al. Neuron 2000). This is specific to navigation, as opposed to just walking on a linear track without task, and place field plasticity is predictive of navigational performance (Moore et al. Nature 2021).
We have added this to the Discussion section (page 13, line 344).
Mehta MR. 2015. From synaptic plasticity to spatial maps and sequence learning. Hippocampus 25:756-62.
Mehta MR, Quirk MC, Wilson MA. 2000. Experience-dependent asymmetric shape of hippocampal receptive fields. Neuron. 25: 707-15. Moore JJ, Cushman JD, Acharya L, Popeney B, Mehta MR. 2021. Linking hippocampal multiplexed tuning, Hebbian plasticity and navigation. Nature. 599: 442-448. -
Evaluation Summary:
This manuscript, contains fundamental findings that substantially advance understanding of an important research question, mostly uses appropriate and validated methodology in line with the current state-of-the-art, with good support for the claims, and the message of the manuscript will have a profound and lasting influence on neuroscience. In essence, the manuscript reports that dopamine converts spike-timing-dependent synaptic depression into potentiation that requires cAMP/PKA second messenger cascade and protein synthesis. The mechanism enables a separate synaptic input to induce heterosynaptic potentiation in previously primed synapses, which is shown in a network model to have desirable computational properties. The significance of the findings is threefold: First, it is the longest-lasting synaptic …
Evaluation Summary:
This manuscript, contains fundamental findings that substantially advance understanding of an important research question, mostly uses appropriate and validated methodology in line with the current state-of-the-art, with good support for the claims, and the message of the manuscript will have a profound and lasting influence on neuroscience. In essence, the manuscript reports that dopamine converts spike-timing-dependent synaptic depression into potentiation that requires cAMP/PKA second messenger cascade and protein synthesis. The mechanism enables a separate synaptic input to induce heterosynaptic potentiation in previously primed synapses, which is shown in a network model to have desirable computational properties. The significance of the findings is threefold: First, it is the longest-lasting synaptic eligibility trace identified so far; second, the mechanism enables memory linking between temporally separate events; and third, it indicates a novel function of postsynaptic reactivation events. In addition, the finding may inspire new reinforcement learning algorithms in machine learning.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)
-
Reviewer #1 (Public Review):
This is a very interesting paper showing that postsynaptic bursts in the presence of dopamine produce input-specific LTP in hippocampal synapses 10 minutes after they were primed with negatively coincident pre- and postsynaptic activity. LTP requires NMDAR activation during priming and involves a cAMP-PKA cascade and protein synthesis. When this synaptic rule is incorporated into a computational model, reinforced learning is possible through selective reactivation of neurons. Experiments in behaving mice confirmed that neurons reactivated after an exploratory period display more activity than non-reactivated neurons.
-
Reviewer #2 (Public Review):
Building on their previous 2015 study with Brzosko, Fuchsberger et al. propose a potential solution for how the brain associates with memory events that are separated in time. The authors find that in the presence of dopamine, postsynaptic bursts produce input-specific LTP at hippocampal CA3-CA1 synapses ten minutes after priming with a post-before-pre spiking-pairing protocol. They explore the signalling somewhat, for example showing a need for postsynaptic NMDARs as well as for protein synthesis. Using a computer model, they find that this form of plasticity enables reinforcement learning. A few key predictions were verified using an in-vivo spatial learning model.
This is a strong study that addresses a long-standing fundamental problem in modern neuroscience research, namely the temporal credit …
Reviewer #2 (Public Review):
Building on their previous 2015 study with Brzosko, Fuchsberger et al. propose a potential solution for how the brain associates with memory events that are separated in time. The authors find that in the presence of dopamine, postsynaptic bursts produce input-specific LTP at hippocampal CA3-CA1 synapses ten minutes after priming with a post-before-pre spiking-pairing protocol. They explore the signalling somewhat, for example showing a need for postsynaptic NMDARs as well as for protein synthesis. Using a computer model, they find that this form of plasticity enables reinforcement learning. A few key predictions were verified using an in-vivo spatial learning model.
This is a strong study that addresses a long-standing fundamental problem in modern neuroscience research, namely the temporal credit assignment problem of how temporally well-separated signals can be meaningfully associated and learned in the brain. The experiments are carefully executed, the rationale is clearly explained, and - excepting Fig 6-8 - the figures are for the most part easy to understand. The study ranges from in-vitro electrophysiology across computer modelling to awake-behaving in-vivo experiments to persuasively argue that their novel findings may provide a candidate solution to the temporal credit assignment problem. Taken at face value, this work is likely to be highly impactful, however, some control experiments were missing or are perhaps just not shown (e.g., stability, stability in the presence of anisomycin, the effect of anisomycin on firing, and similar), which makes the validity of the findings a bit hard to evaluate at times.
-
Reviewer #3 (Public Review):
Fuchsberger et al. demonstrate that an otherwise LTD-inducing STDP protocol can produce LTP if followed by burst reactivation of post-synaptic neurons in the presence of dopamine. Using computational modeling and single-photon imaging in the CA1 in mice, they propose these findings are relevant to spatial over-representation at a reward location.
This is a follow-up of the two previous studies from the same group (Brzosko et al., 2015 and Andrade-Talavera et al., 2016) where they showed a post-before-pre STDP protocol, which by default induces a (pre-synaptic) LTD, will induce synaptic potentiation in the presence of dopamine and continuous synaptic activity. The main conceptual difference between this manuscript and these previous studies is that continuous synaptic activity can be replaced by post-synaptic …
Reviewer #3 (Public Review):
Fuchsberger et al. demonstrate that an otherwise LTD-inducing STDP protocol can produce LTP if followed by burst reactivation of post-synaptic neurons in the presence of dopamine. Using computational modeling and single-photon imaging in the CA1 in mice, they propose these findings are relevant to spatial over-representation at a reward location.
This is a follow-up of the two previous studies from the same group (Brzosko et al., 2015 and Andrade-Talavera et al., 2016) where they showed a post-before-pre STDP protocol, which by default induces a (pre-synaptic) LTD, will induce synaptic potentiation in the presence of dopamine and continuous synaptic activity. The main conceptual difference between this manuscript and these previous studies is that continuous synaptic activity can be replaced by post-synaptic burst. This means that reactivation of post-synaptic neurons without any further pre-synaptic instruction is sufficient for successful LTP induction.
Mechanistically, the two protocols (continuous vs burst activation) appear to be similar (but not identical). For example, both require the activation of post-synaptic NMDAr during STDP pairing, and both depend on the AC/PKA pathways. Additionally, there are two new observations here: The activity of voltage-gated calcium channels during bursting is required for potentiation; the burst-induced potentiation also requires protein synthesis.
The evidence provided at this stage is strong.
Major point:
It is not clear to me how the STDP studied here relates to the next part of the study, the reward-based navigation task. My interpretation is that the authors consider the activity before reaching the reward location (approaching time) as resembling the STDP priming protocol, the activity at the reward location as equivalent to the bursting protocol, and consumption of the reward as similar to dopamine application. If so, what is the circumvential evidence that the activity during the approach induces any form of plasticity? The link between the two is not obvious and I see the manuscript as two interesting but not naturally linked stories.
-
-