Consolidation and reconsolidation Flashcards
What is memory consolidation? In terms of memory models
It’s the rehearsal part of the Atkinson and Shiffrin multi-store model of memory.
Its some process that results in the encoding of LTM.
Many people believe synaptic plasticity changes are responsible.
What four criteria need to be proved to say that synaptic plasticity underlies consolidation?
- Detectability - you should observe a change in synaptic weight when a memory is formed
- retrograde alteration - changing synaptic weights should erase or occlude memories
- mimicry - we should by able to create memories by manipulating synaptic weights
- anterograde alteration - blocking changes in synaptic plasticity should prevent the formation of new memories
Detectability - has it been shown? 3 strategies, 3 papers
Whitlock et al 2006 - used ‘inhibitory avoidance’ (dark and light cage sections, but the dark section has footshocks) to show changes in synaptic efficiency are present for up to 4 hours after a learning experience. But this is not long, maintenance of recording electrodes prevented proof over longer timescale
We can also label engrams using Immediate Early Genes such as Fos, Zif268, upregulated during learning, transcription factors that allow new protein synthesis. They return to baseline within hours, but transgenic mice produce fluorescent proteins when IEGs are active, permanently label engram. Strategies:
1. TetTag - Tetracycline controls labelling via a self-activating tTA-Tet system. Doxycyline can be given in food to prevent fluorescence until experimentally ready.
2. TRAP (Targeted Recombination in Activated Populations) - labelling is dependent on tamoxifen-inducible Cre recombinase system
Reijmers et al 2007 - Introduced TetTag, showed that engrams in BLA activated during auditory cued conditioning are also activated during retrieval.
It’s also been shown using contextual fear conditioning in the hippocampus and cortex
Tayler et al 2013 - changes can be detected 14 days after a learning experience
Problems - TetTag experiments have shown surprisingly low reactivation of the same neurons at recall - above chance but still low. Perhaps because of ‘over-tagging’ - neurons not in engram but active in the hours before and after training may be tagged. Also correlational but not causal evidence.
Retrograde alteration - evidence for (3)
Brun et al 2001 - applying HFS to hippocampus (to induce LTP) meant rats spent less time in the target area of a water maze
Denny et al 2014 - optogenetically inhibiting an hippocampal engram labelled in training reduces freezing to the cue.
Zhou et al 2009 - more excitable neurons are preferentially recruited to the engram. You can therefore allocate neurons to an engram, and then silence them. Silencing allocated engram neurons using an allatostatin-receptor-ligand system just before a memory test impaired memory. Unsilencing and testing again gave normal tone-fear memory results!
Mimicry
Tim Bliss said you should be able to induce the memory of dinner with Marilyn Monroe
Ramirez et al 2013 - labelled neurons recruited into an engram in context A with channelrhodopsin, then stimulated them (using blue light) in context B while receiving footshock. Stimulating the engram for context A whilst creating a new context-B-shock association meant rats put back into context A had a fear response, despite never having been fear conditioned in context A.
Ohkawa et al 2015 - labelled CA1 engram in neutral context memory. labelled BLA engram in foot-shock memory. activated both at once, induced fear of neutral context.
Alternatives to LTP=learning
Shors and Matzel 1997 - could be an attentional mechanism, or non-specifically increase salience of external stimuli to facilitate learning at a distant synapse.
We said LTP was a valid theory for memory because it’s ‘synapse-specific’. Then we found that LTP actually induces widespread responses, e.g. LTP in the dentate gyrus caused increased presynaptic glutamate receptor mRNA on the contralateral side after 5 hours. Smirnova et al 1993 said LTP in one part of the network can affect the downstream part of the network. Instead of seeing this as evidence against LTP=learning, we said it’s a valid theory because memory is distributed. We validated the theory by invoking the theory itself.
LTP is decremental - so it may be involved in short-term memory acquisition, but can’t be the final mechanism behind LTM storage.
If you let LTP decay back to baseline, the next induction of LTP is no easier than the first time. There is no facilitated reacquisition. This is not true of memories.
Drugs or genetic manipulations that impair NMDA-dependent LTP in the hippocampus improve performance on some tasks (passive avoidance learning) and worsen it on others (Morris water maze)
zif268 is upregulated after HFS-induced LTP in hippocampus, but not after contextual fear learning. So maybe the latter does not require the former.
Artificial induction of LTP had no impact on acquisition of water maze memory, but rats with higher capacity for LTP escaped the aversive environment faster and persevered for longer after platform removal. So maybe LTP is not about memory, but about perseverative behaviour.
An arousing experience (like restrained footshocks) causes changing e.g. in AMPA receptor binding that mimic changes during LTP. Maybe LTP is an arousal/attentional device.
Co-operativity in LTP (explanation)
A weak input alone will not cause LTP, but can contribute when it fires at the same time as a stronger input. The two inputs co-operate.
Associativity in LTP (explanation)
When two inputs co-operate, they both become strengthened - they are associated.
Input specificity in LTP (explanation)
If a strong input fires without the weak input firing, only the strong input is strengthened - the LTP is specific to this input.
NMDA receptors role in LTP
Require depolarisation to remove Magnesium ion block (usually the depolarisation is due to AMPA action), AND glutamate binding to open the pore.
These properties account for cooperativity and associativity (strong input needed to depolarise, but once depolarised any input producing glutamate will be strengthened) and input specificity (even if the cell’s depolarised, only inputs producing glutamate will strengthen).
Application of antagonist AP5 blocks LTP in hippocampal slices (Collingridge et al 1983) and intracerebroventricular AP5 impairs performance in Morris water maze (Morris et al 1986)
Calcium role in LTP
Malenka et al 1992 - used a photosensitive calcium chelator to determine that calcium is required for early induction of LTP, but not for later stages of LTP.
Miller et al 2002 - Mice with CaMKII mutations show impaired LTP, and poor performance on object recognition, spatial, and fear association tasks. CaMKII can autophosphorylate after calcium binding, acting as a molecular memory. Dependence on CaMKII therefore implies dependence on calcium.
Buard et al 2010 - Post-training block of CaMKII activity does not impair storage of contexual fear LTM, so it’s required for LTP induction but not maintenance.
Protein kinases role in LTP
In addition to CaMKs: PKA, PKC, MAPK are necessary for LTP induction. They act by:
- phosphorylating NMDA and AMPA receptors
- phosphorylating proteins involved in trafficking more AMPA receptors to surface
- initiating dendritic protein synthesis
- travelling back to soma and initiating somatic protein synthesis
Required for object recognition and for fear memory
IEGs role in LTP
zif268 is upregulated after HFS-induced LTP, but not contextual fear conditioning.
Jones et al 2001 - Transgenic mice that do not express zif268 show impaired L-LTP but not E-LTP, and poor spatial learning in water maze.
Lee et al 2004 - antisense knockdown of zif268 in hippocampus had no effect on cellular consolidation of contextual fear learning
Therefore is the mechanism of LTP different in different parts of the hippocampus? Or do some types of learning not require LTP?
Protein synthesis role in LTP
Frey et al 1988 - application of anisomycin 2-3 hours after training disrupts LTP, but earlier application leaves it intact.
HOWEVER, anisomycin also inhibits protein kinases, can induce apoptosis, and alters value of appetitive reinforcers.
Quevedo et al 1999 - anisomycin induces amnesia when administered prior to or 3 hours after inhibitory avoidance training, but not when administered immediately or 6 hours after. So waves of intracellular signalling.
Similar two-window effect found using dopamine receptor antagonists and waves of activation of kinase signalling pathways
Maintenance of LTP
Abraham et al 2002, 2003 - LTP can have very variable temporal stability, but can last up to a year.
PKMzeta inhibition impairs spatial memory and conditioned fear when ZIP (zeta inhibitory peptide) is injected into hippocampus and BLA respectively.
Shema et al 2007 - overexpression of PKMzeta enhances two consolidated memories
Tsokas et al 2012 - antisense PKMzeta shortly before aversive training blocks formation of LTM but not STM
BUT PKMzeta KO mice can still maintain spatial memories, although not so well when a weaker training protocol is used (Tsokas et al 2016)
ZIP impairs LTP maintenance and erases an established reward memory even in PKMzeta KO mice, so it must have other target/s.
Also ZIP injection into dorsal hippocampus does not impair contextual fear LTM, despite this area being necessary.
Tsokas et al 2016 - antisense PKMzeta after memory acquisition blocks spatial memory retention in WT but not PKMzeta KO mice. In KO mice, PKMgamma inhibition disrupts spatial memory maintenance. Perhaps this is a compensatory mechanism.
ZIP can block PKMgamma action too.
Proteins synthesised during cellular consolidation are those associated with cytoskeletal architecture and synaptic growth, so maybe maintenance requires structural change.
