Ramirez et al. -- Annie Bryant

I liked how both papers this week were from the same lab, since this puts into perspective the tremendous progress they made in just two years in building upon their previous work. A friend once said that every time you think of a memory, it gets further distorted, since you’re actually recalling the last time you recalled that memory. As Ramirez et al. point out, humans are constantly unconsciously integrating misinformation into our memories from outside sources. While functional imaging studies have highlighted robust activity in the hippocampus during both genuine and false memory recall, research thus far hasn’t localized hippocampal subregions responsible for the genesis of false memories. They propose that by activating neurons with a fine spatial and temporal scope, they can delineate false versus genuine memories.

In 2013, Ramirez et al. hypothesized that combining fear conditioning with optogenetic light activation in dentate gyrus (DG) neurons would allow them to create a “false memory,” consisting of a fear response in a fear stimulus-free setting. They accomplished this using c-fos-tTa mice, injected with AAV-TRE-ChR2-mCherry into either the DG or the CA1 of the hippocampus. This inducible expression model is very powerful and precise: as a tet-off system, the absence of tetracycline (or its analogue doxycycline) activates transcription of a gene of interest – in this case, ChR2. By altering diet to either include or exclude doxycycline (dox), they were able to exclusively label DG or CA1 neurons with ChR2 during specific contexts; I thought of this like an external backup of the neuronal activation profile at that moment. They took mice off dox in context A to label cells that responded to exploring this new location, then put the mice back on dox before placing them in context B for foot shock-fear conditioning and simultaneous optic stimulation. This light selectively targeted and reactivated the neurons that were previously activated in context A.

This was a very complex set of experiments that required extensive controls, and I was really impressed with how well Ramirez et al. covered their bases. When DG-transduced mice returned to context A’, they froze significantly more, but didn’t freeze in the novel context C; this suggests freezing isn’t merely generic aftermath fear conditioning. Even if mice were also exposed to context C before conditioning (while still on dox), the mice still only froze in context A’ and not C’. Furthermore, ChR2-mCherry DG mice that underwent the same behavioral protocol minus light during fear conditioning didn’t freeze in either context, indicating that light stimulation is key and that false recall of fear memory is context A-specific. For yet another control, Ramirez et al. switched the repeated and novel contexts (A and C, respectively) in a new cohort to show that after light plus fear conditioning, the mice froze upon return to C’ but not A’. This was a smart design choice to show that the freezing response is specific memory recall and not a general artifact response to context A. The conditioned place avoidance (CPA) paradigm also revealed that when ChR2-transduced DG mice were optically stimulated in one chamber and fear conditioned elsewhere, they significantly preferred the opposite chamber the next day. Conversely, when ChR2-labeled CA1 cells were activated in context A and reactivated with light during fear conditioning in context B, there was no increase in freezing in any context thereafter. CA1-targeted mice also didn’t exhibit any chamber preference in the CPA paradigm following the same experimental protocol.

In 2015, Ramirez et al. optogenetically reactivated DG cells that were previously activated during a positive experience to acutely rescue stress-induced depression. Using the dox-mediated selective ChR2 labeling technique, DG neurons were labeled during either positive, neutral, or negative experiences. After 10 days of either no stress or chronic immobilization stress (CIS), mice underwent behavioral tests to measure anxiety, escape behavior, and anhedonia. In stressed mice, optical reactivation of only positive-associated DG cells rescued escape behavior and sucrose preference. However, no stressed animals responded to light in the open field test or elevated plus maze test, both of which measure anxiety. This is an interesting parallel with the papers we read in Week 1 (Santarelli and Bessa), which identified key differences in both the induction and resolution of depression versus anxiety. Ramirez et al. note that they focused on dorsal DG engrams in this paper but that they are keen to explore ventral DG pathways, as these may serve a role in regulating anxiety and stress responses.

Through optogenetic inhibition of circuits and pharmacological inhibition of neurotransmission respectively, Ramirez et al. identified the BLA à NAc circuit and glutamatergic signaling as key to behavior rescue via optic reactivation of positive experience-labeled DG cells. They also show that “chronic” (5 days) light stimulation to positive-labeled DG neurons in stressed mice reversed stress-induced tail suspension and sucrose preference test deficits.  Additionally, this group was the only one that maintained control levels of neurogenesis markers PSA-NCAM and DCX despite chronic stress. Interestingly, the increase in adult-born neuron counts positively correlated with the degree to which each group preferred sucrose. However, these results were obtained only one day after light stimulation ended. To conclude that these effects are long-lasting, I would have preferred that the experimenters measure these outcomes further out from stimulation (e.g. maybe a week, if not more). Nonetheless, it’s an exciting idea that persistent activation of a positive memory trace can promote and protect a positive emotional state after a traumatic event, as this has tremendous implications for PTSD research.

There are two issues that I’m still unclear on from these two papers. First, in the very last sentence of Ramirez et al. 2013, they mention the tri-synaptic circuit, in which each structure (CA1, CA3, and DG) presumably contributes heavily to memory acquisition and recall. Why did they focus only on DG and the CA1 and not the CA3 in their experiments? Second, I was confused by their interpretation of their 2013 comparison of neuronal activity overlap, both inter- and intra-context. They show that in DG-labeled mice, 1% of neurons were active in both A and C, which the authors say means the contexts recruit largely independent populations – that’s fine. However, 2% of neurons were activated upon repeat exposure to context A, which they describe as largely overlapping. The margin of change here is so small; how is 1% overlap independent, but 2% overlap is significantly overlapping? Also of note, 30% of CA1 neurons were active in both A and C, while 50% were active upon repeated exposure to A. I’m just surprised that so many CA1 neurons were active in both of these contexts, and yet DG neurons ended up being the key player in these papers.