Han & Yiu: CREB and Memory Allocation -- Annie Bryant


Recalling a particular memory may cause us to think about memories of closely related events or topics. Is this due to an underlying structure governing memory organization, in which related memories are contained by overlapping Hebbian ensembles of neurons? Are neurons that become incorporated into a memory simply in the right place at the right time, or are there specific factors that predispose them for memory allocation? In 2007, Han et al. showed that lateral amygdala (LA) neurons with elevated CREB are selectively activated by auditory fear memories. They replicated this in 2009 using a combined model of cre-inducible diphtheria toxin receptors (iDTRs) and a CREB-cre vector to induce CREB overexpression. After fear memory retrieval, CREB-cre neurons showed increased activity (i.e. nucleus arc RNA) relative to non-infected neighboring cells. CREB-cre mice also showed enhanced fear memory formation. Systemic administration of DT after fear memory formation blocked memory expression for at least 12 days, suggesting permanent memory erasure. However, these mice were still able to acquire new fear memories thereafter, indicating that remaining noninfected neurons picked up the slack.

Han et al. had good controls to demonstrate that the CREB-cre virus did preferentially infect cells active during the fear conditioning and that DT didn’t induce any general toxin effects. However, I wasn’t totally sold on their conclusion that they “establish[ed] a causal link between a defined subpopulation of [LA] neurons and expression of a fear memory”. This paper is based entirely on neuronal apoptosis and changes to freezing behavior, yet they didn’t include an experiment to rule out potential confounding effects on baseline anxiety or general locomotion, both of which could affect freezing behavior. I was left with several questions: which of CREB’s myriad cellular roles is important to memory allocation? In what stage during the fear conditioning paradigm is CREB expression relevant? Was killing CREB-cre neurons with DT a better option than simply inhibiting them after memory acquisition?

Yiu et al. pick up where Han et al. left off, noting that CREB promotes neuronal excitability by (in part) decreasing voltage-gated K+ currents. They hypothesize that this CREB-mediated excitability enhances fear memory formation and biases engrams to recruit high-CREB neurons. They investigated this using HSV::CREB and HSV::dnKCNQ2 (a dominant-negative K+ channel), both of which increased excitability in cultured hippocampal neurons. Interestingly, Kir2.1 (an inwardly rectifying K+ channel) with CREB blocked CREB-induced excitability. These findings were paralleled in vivo following auditory fear conditioning: while CREB and dnKCNQ2 injected into principal LA neurons enhanced memory, CREB+Kir2.1 blocked memory enhancement. Yiu et al. did also account for potential anxiety and locomotion effects using the elevated plus maze test and open field test, respectively, and found that their manipulations didn’t affect either behavioral outcome.

Both Han et al. and Yiu et al. indicated that fear memories incorporate a large network of neural structures beyond the LA, and Yiu et al. explored two other amygdala nuclei: basal amygdala (BA) and central amygdala (CeA). When they injected the CREB vector into BA or CeA, they didn’t observe any fear memory enhancement associated with CREB overexpression, highlighting the regional specificity of CREB-enhanced memory. Yu et al. also address the specific time frame during which neuronal activity is relevant to memory allocation. Viral delivery of CREB or dnKCNQ2 two days before training (for optimal transgene expression during training) produced long-lasting memory enhancement beyond the end of transgene expression. However, viral delivery of CREB or dnKCNQ2 11 days before or 1 day after training had no effect on fear memory formation or recall. These series of experiments support the conclusion that neuronal excitability must be increased at the time of training in order to enhance memory. 

To delve deeper in temporal specificity, they also used DREADDs and optogenetics, with which they could transiently excite neurons with fine temporal and spatial control. They used the DREADD hM3Dq, a synthetic ACh receptor that increases excitability by binding the synthetic ligand CNO. When hM3Dq-injected mice were systemically administered CNO right before fear training, they exhibited memory enhancement 24 hours later – in the absence of CNO. Conversely, delivery of CNO after training failed to enhance memory, further indicating that excitability during training and memory encoding (rather than during consolidation) is key for memory enhancement. Furthermore, artificial reactivation of these neurons with CNO in a novel context test was sufficient to induce freezing. Lastly, Yiu et al. used ChR2 to show that activation of a small subset of LA neurons immediately before auditory fear was sufficient to enhance memory formation. These experiments seemed a bit overkill since they all drove home the same point: neurons must be excited at the time of memory acquisition to be incorporated into that memory and strengthen it.

Interestingly, memory traces were quite stable in size regardless of experimental group; however, neurons expressing CREB, dnKCNQ2, or hM3Dq were proportionally over-represented among arc+ neurons. Accordingly, noninfected neurons were under-represented among arc+ neurons, and were even less likely to be activated by fear training in CREB or dnKCNQ2 mice than in GFP mice. This suggests neurons compete to become part of a memory based on relative excitability, which echoes findings in Kim et al. 2013: the competition for fear memory allocation depends on both neuronal excitability and disynaptic inhibition – in which one excitatory neuron can inhibit another via an inhibitory interneuron.

Yiu et al. point out that although their manipulations were artificial, studies across species indicate that learning involves changes in intrinsic excitability via downregulating AHP K+ currents. Their artificial reactivation of hM3Dq-expressing neurons via CNO in a novel context reminded me of the Ramirez papers last week. In both studies, two stimuli (e.g. tone or light + shock) activated distinct populations of neurons that wouldn’t normally fire at the same time, and linked these neurons in a memory trace that can be reactivated with the CS. I wonder if this has implications for psychological disorders involving inappropriate sensory/memory associations like PTSD and schizophrenia; do individuals with such disorders have higher levels of CREB in fear-associated brain regions?


Comments

Post a Comment

Popular posts from this blog

Microbiotic effects on stress and social behavior in mice- Sierra Smith

For the final week of paper discussions, we dove into the fascinating and very new topic of the gut microbiome, and the influence of microbiotic composition on mice either under chronic subordinate colony housing situations or the offspring of dames fed a high-fat diet. In the first paper I read, Reber et al, there were fascinating findings that although supplementation with heat-killed Mycobacterium vaccae did not prevent all microbiotic changes induced by stressful housing conditions, it did have a significantly protective effect on the development of spontaneous colitis and stress-induced aggravation of pre-existing inflammatory bowel disease (colitis in mice). There was also a profound effect on proactive defensive behaviors in mice housed in CSC conditions, whereas they became defeated less easily and displayed significantly more proactive behaviors when faced with an aggressor. I was particularly excited to read about the increased Iba1 immunostaining in fear-regulato...
Read more

ACC-BLA Connections and Observational Learning- Annissa DeSilva

            Observational fear learning is an innate social behavior, Allsop et al. 2018 outlines the neural circuit underlying how the brain encodes emotionally salient (fear) behavior with neutral cues associated through observational fear learning. Allsop et al, found that projections from the Anterior cingulate cortex (ACC) to the Basolateral Amygdala (BLA) are essential in encoding observational learning. Specifically observing distress causes basal changes in the ACC neurons (encoding of the observed behavior through social cues) which enables the acquisition of the predictive nature of the aversive cue by BLA neurons. When ACC - BLA projections were optogenetically inhibited while mice were observing fear conditioning a reduced freezing response was seen when conditioned fear response was tested. Interestingly when the ACC - BLA projections were inhibited during testing of observational fear conditioning the...
Read more

Two sides of the same coin?: the bidirectional effects of VTA-NAc DA neurons - Emily Jones

Papers like the ones assigned this week reminded of the reasons I opted to take this class. Research into the biological nuances of complex clinical disorders really reinforces the idea that everything psychological is simultaneously biological. And in these papers, optogenetics amazingly gives us a fast and efficient way to explore these nuances. Interestingly, both papers demonstrated a bidirectional effect of dopamine neurons in the VTA-NAc pathway on depression phenotypes using phasic photoactivation. However, it seemed that these effects were flipped. In other words, when Chaudhury et al. inhibited the neurons, a resilient phenotype was induced in previously susceptible mice (indicated by a high preference for sugar water and decreased social avoidance compared to control mice); when Tye et. al. use the same opsin to inhibit the same kind of neurons, depression phenotypes were induced. Indeed the differences may be due in large part to the different stress conditions the mice we...
Read more