Synaptic plasticity-dependent competition rule influences memory formation

The formation of long-term memories involves the modification of synaptic connections between neurons, a process crucial for creating engrams – the physical representations of memories in the brain. While identifying engrams has advanced significantly, the precise mechanisms underlying engram formation during memory formation remain unclear. Existing research suggests that the neurons participating in memory encoding are not pre-determined but selected by specific mechanisms influenced by factors like CREB levels and cellular excitability at the time of learning. Synaptic plasticity, involving LTP and LTD, is widely accepted as a vital process in memory encoding and storage. However, the relationship between synaptic plasticity and the formation of memory-encoding neuronal ensembles remains poorly understood. This study hypothesized that the specific synapses undergoing plasticity-related activities during memory formation influence which cells and synapses are recruited for memory encoding. The challenge in studying this lies in the limitations of current engram labeling methods, which typically allow targeting only after memory encoding or consolidation. This research utilized an optogenetic priming approach to overcome this limitation. By targeting synapses that are usually not potentiated by learning and manipulating their plasticity using optogenetics, the researchers aimed to explore the role of synaptic plasticity in engram construction during memory formation.

The paper references numerous studies that have investigated memory engrams and the role of synaptic plasticity in memory formation. Studies are cited which identified ensembles of neurons and synapses where engrams are localized and where learning-dependent physical changes occur. The importance of CREB levels and cellular excitability in neuronal recruitment for memory encoding is discussed, citing research demonstrating that neurons with higher levels of CREB or greater excitability are preferentially selected. The role of activity-dependent synaptic plasticity, such as LTP and LTD, in mediating enduring changes in synaptic strength induced by learning is also highlighted. However, the study points out a gap in the literature: the unclear relationship between synaptic plasticity and the formation of memory-encoding neuronal ensembles.

The study used Pavlovian fear conditioning as a model of associative learning. Mice received injections of adeno-associated viruses (AAVs) into the auditory cortex and thalamus to express channelrhodopsin-2 (ChR2) in auditory neurons. Optogenetic stimulation was used to manipulate synaptic plasticity at auditory-to-lateral amygdala (LA) synapses. A priming paradigm was employed, where 10-Hz optogenetic stimulation was delivered to the ChR2-expressing auditory inputs immediately before fear conditioning (opto-FC). Electrophysiological recordings (field excitatory postsynaptic potentials or fEPSPs) were used to measure LTP induction at these synapses. Behavioral tests (freezing behavior) assessed fear memory formation and retrieval. Bidirectional optogenetic manipulation (using AAV encoding both ChR2 and halorhodopsin) allowed researchers to either activate or inhibit the manipulated synapses during memory retrieval. Anisomycin, a protein synthesis inhibitor, was used to test the lability of memories associated with the manipulated synapses. In addition to LTP manipulation, optical LTD was used to reverse the effects of priming and investigate its impact on memory encoding. Immunohistochemistry (using c-Fos as a marker for neuronal activation) was used to analyze the neuronal ensembles involved in memory encoding and retrieval in different experimental conditions. Various control groups were included (e.g., fear conditioning alone, optogenetic stimulation alone, anisomycin administration alone) to ensure the specificity of the findings. The researchers meticulously controlled various parameters, such as virus injection sites, light intensities, stimulation protocols, and behavioral testing conditions, using advanced techniques such as electroplating of silicon probes for improved multiunit recording and custom-made fluid-optic cannulas for simultaneous drug and light delivery.

The main findings are: 1. **Optogenetic priming before fear conditioning:** 10-Hz optogenetic priming of auditory inputs to the LA before fear conditioning induced LTP at these synapses, which were normally not potentiated by fear conditioning alone. This resulted in preferential encoding of fear memory in the manipulated neuronal ensembles. 2. **Optical LTD reverses preferential encoding:** Optical LTD delivered shortly after training reversed the effects of priming, abolishing the preferential memory encoding in the manipulated synapses. This suggests that a timely intervention during the encoding phase influences which neural ensembles are recruited for the memory. 3. **Optical LTP alone modifies memory encoding:** Optical LTP delivered shortly after fear conditioning alone, without prior priming, also resulted in preferential memory encoding in the stimulated neuronal ensembles. 4. **Late LTD disrupts memory retrieval:** Optical LTD delivered 24 hours after training disrupted fear memory retrieval specifically at the primed synapses. This effect was reversible by subsequently inducing LTP. 5. **Competitive memory encoding:** The consistent size of the c-Fos+ memory recall-activated cell population, irrespective of plasticity manipulation, points to a competitive, rather than cell-autonomous, selection process for memory encoding. 6. **Immunohistochemistry reveals ensemble changes:** Immunohistochemical analysis (c-Fos expression) showed that the overlap between the optogenetically manipulated neuronal population (tdTomato+) and the neurons activated during memory recall (c-Fos+) was significantly higher in the opto-FC group compared to the control group, demonstrating that the primed synapses were preferentially used for memory encoding. This preferential encoding was reversed by early LTD. 7. Similar results were observed with LTP-induced preferential memory encoding. The effects of LTP were also shown to be long-lasting in the experiments.

The findings strongly support the idea that activity-dependent synaptic plasticity plays a crucial role not only in consolidating but also in selecting neuronal ensembles for memory encoding. The results contradict the hypothesis that synaptic plasticity is solely involved in wiring selected neurons, but not in the selection process itself. Instead, the study demonstrates that the timing and type of synaptic plasticity manipulation (LTP vs. LTD) profoundly influence which neuronal ensembles participate in fear memory encoding. The competitive nature of memory allocation suggests a dynamic process where synaptic activities after an event contribute to the final engram composition. Potential mechanisms underlying this competitive process could involve synapse-specific retrograde signaling, mGluR signaling, AMPA receptor trafficking, and widespread heterosynaptic depression. The study highlights the plasticity of memory encoding even after the initial learning phase, suggesting a window of opportunity for modifying memory allocation by manipulating synaptic plasticity. The ability of post-learning LTD and LTP to modify the engram cells supports the idea that engram cells are formed by the strengthening of synaptic connections between neurons.

This study reveals a previously unknown synaptic plasticity-dependent competition rule that shapes memory formation. Optogenetic manipulation of synaptic plasticity demonstrated that the selection of neuronal ensembles for memory encoding isn't solely determined at the time of learning but continues through a dynamic competitive process involving LTP and LTD. Future research should focus on exploring the molecular and cellular mechanisms underlying this competitive process and investigate the broader implications of this finding for understanding various aspects of memory, including memory interference and reconsolidation.

One limitation of the study is that the optogenetic stimulation protocols used might not perfectly mimic physiological activity patterns. For instance, simultaneous stimulation of cortical and thalamic inputs, which might not accurately reflect natural temporal dynamics. Another limitation is that the study focused mainly on fear memory, and it remains unclear whether this synaptic plasticity-dependent competition rule is a generalizable phenomenon across various types of memory. Further research is necessary to address these limitations and extend these findings to other memory systems.