major studies Flashcards
Wood (2000)
task 2 - relation spatial representation and memory
task/method: tmaze
experimental condition: experiment that would record cell activity when
animal was doing a “path” and would have to choose to turn left or right
results:
place cells can modulate their activity depending on whether an
animal will turn left or right in the future - In essence, the place cells have
a predictive code, signalling intended destination – this shows predictive
coding
klaus (2013)
task 2 - relation spatial representation and memory
task/method: t maze
experimental condition:
animal runs in a treadmill
and then has to decide to turn left or right – Place cells can also show a
time signal - If animals have to run on a treadmill for a fixed period of
time, different place cells will prefer to fire at different time points on
the treadmill
results:
cells fire at the beginning of the run, other cells in the
middle of the run and some other different ones at the end – this shows
temporal coding
aranov (2017)
task 2 - relation spatial representation and memory
task/method: conditioning paradigm experimental condition: animal needs to hold lever down until it hears a sound, then it needs to release
results: they found that place cells can also
develop a preference for different non-spatial
variables – such as tones! - Grid cells could
similarly adopt such auditory tuning but with
multiple firings as compared to place cells – this
shows auditory coding (cells respond to a tone of
a particular frequency)
kim and faselow (1992)
task 3 - consolidation
task/method:
experimental condition:
results:
wilson and mcnaughton (1994)
task 3 - hebbian plasticity
task/method: z maze
experimental condition:
results: they observed
place cells with overlapping fields in an environment tended to be active
together in subsequent sleep
- They proposed that through repeated reactivation of wakeful activity patterns,
new memories become consolidated (through Hebbian plasticity) and committed to long-term storage in the cortex.
olafsdottir (2016)
task 3 - replay for memory consolidation
task/method:
experimental condition:
results:
foster and wilson (2006)
task 3 - replay for memory consolidation
task/method:
experimental condition:
results: replay events occur more frequently during novel
experience.
- replay in memory consolidation is consistent with
SWRs occuring more frequently after rewarded runs
(more memory consolidation) vs unrewarded runs - increase in reverse replay is driven by rewarded
experiences – suggest reverse reply is involved in learning (ambrose, 2016)
girardeau (2009)
task 3 - replay for memory consolidation
task/method: radial arm maze
experimental condition: animal learns to locate food in 3
or 8 arms , then they sleep and they
either disrupt SWRs or not (some other
brain activity is disrupted).
results: swr ripples disruption impairs acquisition of a novel spatial memory task - provides casual evidence for the role
of replay in consolidation
maingret (2016)
task 3 - replay for memory consolidation
- provides casual evidence for the role of replay in consolidation
- replay is associated with higher communication btw hippo
and cortex – assessed here by measuring oscillatory coupling
during sleep that is associated with learning - trained animals on object rotation – free to explore objects in
environment then animals sleep and then they’re put in the
same environment but one of the objects moved, if the animal
notices it will go there to explore otherwise the animal does
not recall. Recall was successful for group that had 20 minutes
encoding period, but not for the 3 minutes encoding period.
Only for the 20 minutes group they found oscillatory coupling
btw hippo and cortex (= co-occurrence of ripples, spindles and
delta waves occurring in close temporal proximity during
sleep)
- Spatial memory learning is supported by coupling between hippocampal and cortical oscillations during sleep - Artificially inducing oscillatory coupling between the hippocampus and cortex during rest can create a long-lasting spatial memory
gridchyn (2020)
task 3 - replay for memory consolidation
memory targeted relay disruption leads to impaired recall of targeted memories ONLY
- one of the most important experiments in the field - animals navigate in either a control or target circular environment, which are very similar and they have to locate a reward, after 4 hours of sleep (controls get no optogenetic disruption and experimentals get opto disruption) they explore the environment again but no reward is located (do animals spend more time in the location where there was previously food or not?
singer (2013)
task 3 - replay for planning
on a WM task cells were more likely to be active in swr during swrs preceding correct decisions BUT only during initial learning of the task
- replay during working memory task
(double-u shape mazes) - early in training there were more SWRs
overall for correct trials
jadhav (2012)
task 3 - replay for planning
disrupting swrs only affects memory guided decision on a wm task
- if you disrupt ripples then their
performance is worse - WM is main support for planning function of
replay
olafsdottir (2017)
task 3 - replay for planning
Just before an animal initiates a goal-directed trajectory or the animal has just reached a
reward location, replay preferentially depicts task-focused information.
• However, if the animal disengages from the task, replay shows no biases
• Thus, replay may switch between planning and consolidation in response to task demands
During engaged periods, the replay also preferentially depicted local, congruent and forward replay trajectories • However, during engaged periods preceding incorrect decisions no biases in engaged replay were observed
- when they go on track 2 or 3 they have
different brain raresentations - as the animal arrived to the corner there was
a very strong bias for replay that was
depicting locations near the animal, this bias
disappeared if the animal linger that corner
(stays long in that corner) - replay seems to switch between planning
and consolidation mode depending how
engaged the animal was to the task - bias was looking at what was being replayed,
either remote (far) or local (close) locations - if they made a mistake they won’t be
rewarded (incorrect) – no bias (engaged
error = disengaged correct and error)
pfeiffer (2013)
task 3 - replay for planning
- animals navigate in rectangular environment where they had to locate food , every day food was in a different location - there were either goal-directed trials (food appears in same location) or random foraring trials (food appears in random location) - they predicted what the animal was
going to do before initiating a goal-
directed trajectory
During a goal-directed navigation task rat haves to alternate between foraging for food and running to a known goal location • Before animals initiate a goal- directed trajectory, replay seems to depict the animal’s upcoming trajectory • However, before random trajectories, replay events do not predict what path the animal will take
gupta (2010)
task 3 - replay for immagination
- animals follow a T maze (the dotted one
is imagination that is a shortcut to a goal) - they found replay in the task, when
animal was located in a reward location
but also depict never experienced paths
replay depict never experienced paths (i.e. shortcuts
olafsdottir (2015)
task 3 - replay for immagination
- animal is put in a T maze but can’t reach
its arms because there is a transparent
barrier, in one of the arms food is placed
but animal can’t reach it. Then animal
sleeps and then does T maze with no
barrier
Replay can also pre-play paths in unexplored environments • However, preplay only found for novel environments seen before and only for paths leading to a goal
green (1982)
task 4 - memory emergence in first years of life
Rat pups can successfully carry out a working
memory task from ~3weeks of age
• However, short-term reference memory (which does
not require the hippocampus) can be carried out by
2week of animals
Similar to humans, rodents show protracted development of episodic/spatial memory.
• As rodents age, the interval over which they can retain new learning increases
vargha (2001)
task 4 - memory emergence in first years of life
Although episodic declarative memory may develop gradually, the developmental timeline of semantic declarative memory may be different • Developmental amnesia patients can form new semantic memories despite extensive hippocampal lesions and severely impaired episodic memory function • Semantic memory may be mature at birth OR developmental amnesia is associated with alternate mechanisms of semantic memory formation
- It could be, we’re not sure yet
guskjolen (2018)
task 4 - memory emergence in first years of life
task 4 - memory emergence in first years of life
- used contextual fear conditioning task and labelled cells that were active during the encoding in infant animal, then artificially reactivate them when the animal is older doing the same task the memory is recovered – episodic recall hasn’t developed yet in infant
Memories formed during the infantile amnesia period
can be recovered if the original memory trace is
‘synthetically’ re-activated.
• Suggest, perhaps, infantile amnesia reflects a retrieval
deficit
donato (2017)
task 4 - stellate cells
Different sub-regions of the hippocampal formation
show distinct developmental timelines
• The superficial layers of the entorhinal cortex(EC)
develop first, followed by CA3/CA2 and CA1
• The deep layers of the EC and the dentate gyrus are
the last to develop
• These findings have been observed both in monkeys
and rodents
- CA2 is the first one to develop, then CA3 and
CA1 (which have numerous developing times,
depending on the area) and then the dentate
gyrus - Donato paper: entorhinal stellate cells
provide an activity-dependent instructive
signal that drives maturation sequentially and
uni-directionally through the intrinsic circuits
of the entorhinal-hippocampal network. The
findings raise the possibility that a small
number of autonomously developing neuronal
populations operate as intrinsic drivers of
maturation across widespread regions of the
cortex
wiltgen (2010)
task 6 - semantic memory in rodents
- semantic memories differ in quality as
they’re gist like – gist tested with fear
conditioning in context A , then there is
context B which is quite similar to A and then
C is very different from A and B – animals
generalize fear more to context B than C over
time which means they’re discriminating less
context A and B – if hippo is shut off then
generalization still happens – shows that
semantic memories are cortex dependent and
hippo independent - this model uses fear conditioning which is
not an adequate model as if animal is familiar
with context a and b before fear conditioning, then animal would be able to discriminate
yonelinas (2005)
task 6 - semantic memory in rodents
- Roc curves – means receiver operating characteristic curve – it is a graph that shows the performance of a
classification model at all classification thresholds - familiarity would be semantic (cortical) and recollection (perfectly remembering something) would be hippocampal
– this is easy in humans (they ask) but not in rodents - mind the definition of roc on the slide. – if familiarity is used as a strategy then there is a curved line , while if you
use recollection there would be a straight line in the plot
liang (1986)
task 8 - how stress influences memory
Role of epinephrine – How does it enhance memory?
If you block NE in amygdala, does it mean E effect on memory blocks too? Yup – this
study showed that by blocking E with propranolol retention latency decreases
quirarte (1997)
task 8 - how stress influences memory
Interaction between NE and GLUCORTICOIDS – how?
Dexamethasone: syntetic glucorticoid
Did same experiement as before but instead of NE they give glucorticoids – similar
results!
What does it mean?
If glu need NE and NE is realeased in arousal conditons then do glu modulate NE only
during arousal conditions? – do effects on memory of glu interact with emotional
arousal?
roozendaal (2006)
task 8 - how stress influences memory
Interaction between NE and GLUCORTICOIDS – how?
They want to know if NA is necessary for the effects of glu on memory
Object recognition task – first day gets used to two objects then in second day one is
replaced with a new object – if the animal spends more time on new object it means
it remembered the old one
Two conditions: one with stress and one without.
– to put a rat in a box is already stressful so they habituated them first (top right
figure) for one out of two conditions
Without habituation (NO STRESS)
- Memory is enhanced by glu and it has an interaction with NE as memory is
suppressed by propranolol (beta blocker)
With habituation (STRESS)
- Memory is not enhanced – meaning that glu (stress) enhances memory only in
stressful-arousing conditions
- if yohimbine (NE stimulant) is given after training then memory is enhanced
roozendaal (1999)
task 8 - amygdala and stress
Drug RU 29362 (GR agonist) injected in hippocampus after training
If NE is blocked in the amygdala then it does not modulate glu in hippocampus
If amygdala is not there, you can’t modulate memory – see next slide
cahill and mcnaugh (1998)
task 8 - amygdala and stress
Experiment:
Condition 1. arousing movie clips
Condition 2. static movie clips
They look amygdala activity and 3 weeks later they give them a memory test – the
higher amygdalar activation was when watching the arousing movie clip the more
movie clips they were able to remember :D , for neutral /static material there is NO
EFFECT!
van stegeren (2005)
task 8 - amygdala and stress
When humans are given a beta blocker (beta-adrenergic blocking agents) when
watching arousing material there is less activity in the amygdala (figure B) – showing
that NE plays a role in activating amygdala
tas (2017)
task 5 - sushi belt model
- microtubules: Microtubules in a neuron are used to
transport substances to different parts of the cell. For
example, neurotransmitters are made in the cell body
close to the nucleus, but need to travel long distances to
the end of axons where they will be used for synaptic
transmission. - motor-paint technique developed to see how are the
microtubules organized inside the dendrites, they’re
interested in that because the microtubules transport
the molecules travelling to the synapse . They look
specifically at tyrosinated or acetylated microtubules
(MT) - microtubules are either end positive or end negative
and the motors that travel in these microtubules they are
specific for one direction, in dendrides molecules in MT go
in both directions - on the left the whiter the more calcium there is
- on the right – black ampa receptors, when you have a line
that goes from left to right then it is going outward, from
top to bottom they are inward (going towards the soma),
pause/blue are not moving
doyle (2011)
task 5 - sushi belt model
- things always travel within the dendrite and when the
synapse needs something – takes it out of the pool what is
available
afner (2019)
task 5 - mRNA transmission
- where are proteins made in or around the synapse?
You can study this by using metabolic labelling, one of the
numerous ones is called puromycylation, puromycin is an
antibiotic very small molecule. t RNA is the intermediate btw
the ribosome and the amino acid which will at the end serve to
make a protein. Puromycin is a aminoacyl t RNA analogue so it
gets loaded at the place of the amino acid on all the t RNAs – in
other words puromycin is transported by t RNA – when all t
RNA is loaded with puromycin which is then transported to the
ribosome and released to substitute an amino acid – this
terminates the translation so you end up having truncated
proteins (proteins that are not fully ready) but it carries
puromycin that can be detected with an antibody – this way
you can see in what location the proteins were being made - figure: neuron expressing GFP in green and v GLUT1 in pink
(excitatory terminals)
Method used: expansion microscopy , when all the antibodies
are ready you put them into a gel that expands when it’s in
water so that cell becomes very big and this way we can see
synapses very well - red: detecting site of protein synthesis , there is a lot of
activity outside the neuron of interest as a lot of cells are on –
only a couple of neurons are expressing GFP (what interests us) - red shows that we have a lot of protein synthesis in the
dendrite, also inside the spine (in the middle and not in
postsynaptic density) and also, unexpectedly, a bit of protein
synthesis in the pre synaptic terminal – from these results we
can say that protein synthesis is basically happening
everywhere 65% in POST SYNAPSE of the spines produce
proteins after only 5 minutes, if labelling would have been
longer probably the percentage would have been higher, 37%
in PRE SYNAPSE of the spines produce proteins - so we induce plasticity on those neurons – protein
synthesis – in presynapse, postsynapses and also INHIBITORY
presynaptic TERMINALS (not mentioned above) - expain and be able to read the table!
afner (2015)
task 5 - surface diffusion
- how come the phosphorylation of this really remote RS
protein has any impact? To investigate this they use
FRET, you have two fluorophores, one that when is
directly excited is going to shine some green
fluorescence (GFP) – if you have another fluorophore in
proximity then it could also excite that and if this
happens the green disappears as it will excite another
fluorophore that will shine red – there is an overlapping
level of energy between the emission and excitation –
when an electrode (green) is excited the lifetime of it is
going to be reduced a lot as it is going to have multiple
ways to escape from that high level of energy as
electrodes when they’re in a high state they want to go
down so they look at the half-life of green protons – the
more green fluorophore interacts with red the more the
half-life will go down - in the dendrites it is more green and inside the spine
there is red – in the red (spine) there is strong
interaction between PDZ-95 and the receptor
REVIEW SLIDE MIDDLE AND LOW PART- CAN’T
UNDERSTAND FROM LECTURE - why if you modify domain WT (blue circles) then a
change in the receptor’s mobility happens? charge in
blue dots is positive , those positive charges bring the
domain close to the membrane. If you put a negatively
charged amino acid then it will bring it further from the
membrane – when charge balance is changed then
interaction PSD 95 - if there is a strong interaction between green and red
fluorophore then there is WHAT THE FUCK DOES SHE
SAY? - in her research she does a mutation that mimics
phosphorylation and it results ?
CaMKII – phosphorylation mediated by it, it also stops
the cargo, it recruits protein and cargos that enable
protein synthesis (including the protein synthesis itself),
it immobilizes directly the receptors within the
transmembrane
opazo (2011)
task 5 - surface diffusion
- AMPAR receptors are made of different
proteins, one (RS) that is very important is one
that binds to PDZ-95, which is the organizer of
the postsynaptic density (reason why
postsynaptic density looks so dark when looked
in microscopy). RS protein has a domain that
can be phosphorylated by CaMKII in 9 different
sites - if CaMKII gets activated and if you look at
receptors that are diffusing - in normal conditions you have a lot of mobile
receptors and little immobile if CaMKII is
always active then receptors do not move
anymore and they find a synapse thus caMKII
should play a role in recruiting the receptors …
but how does it do that?
goldman-rakic
task 7 - delay period
Correlates: persistent activities during
the delay period
- In PFC: neurons are cue-selective, some are
delay-selective and some are response-selective.
Oculomotor delayed response task in macaque monkeys ❖Some neurons in prefrontal cortex were active during the delay ❖These neurons were direction-tuned, i.e. a correlated for spatial ‘working’ memory ❖This example ‘delay neuron’ held on to this piece of information when monkey looked at a blank screen ❖She and colleagues suggested that similarly tuned frontal neurons formed a recurrently connected cluster— similar to how visual cortex works
inagaki (2019)
task 7 - delay period
Mechanisms: persistent firing: within a neuron or
from a network?
If persistent activity is within a neuron, photoinhibition should turn it off.
If persistent activity is within a neuron, photoinhibition should turn it off.
Otherwise: a network mechanism
Premotor cortex in mice (activity predictive of correct/incorrect response)
dotson (2018)
task 7 - persistent stimulus - selective activity
Networks? It could be local excitatory networks
(excitatory glu neurons giving excitation to each
other) or it could be mutual inhibition (excite local
network (group 1) but also interneuron which inhibits
more surrounding neurons (group 2) which are also
inhibiting, so inhibit inhibition leads to excitation
which then excite group 1. Plausible to think that
neurons are doing long-range synaptic interactions
across brain regions
- evidence of long-range synaptic interactions: stimulus selective activity during delay period. In vPFC + anterior and
ventral Intraparietal sulcus (AIP/VIP) there is a lot of persistent firing in delay period but not in visual cortex
masse (2020)
task 7 - attractor dynamic and sensorimotor transformation
- recurrent neural network – backward
feedback , last layer gives feedback back to
the input layer (A-B-A) - in recurrent nn- display increasing persistent
activity when the demand of the task
increases
The alternative: hybrid attractor dynamic and short- term synaptic plasticity model Masse et al 2019 Nat Neuro ❖ The spiking induces temporary (<1 s) changes in synaptic weights, perhaps via calcium dynamics ❖ Such short-term synaptic plasticity (STSP) could keep the information till the next burst of spikes! ❖ Orchestrated by beta and gamma rhythm
lundqvist (2018)
task 7 - gamma range oscillations
task/method:
experimental condition:
results:
olafsdottir (2016)
task 3 - replay for memory consolidation
task/method: experimental condition: - co-recorded MEC and CA1 grid and place cells to see whehter there was coordinated replay (measured by grid-place cell coherence) between the two brain areas which you would expect if replay is supporting memory consolidation –this was the case but coordination was stronger for forward replay than for reverse (graph gray and yellow)
results: hippo replay leads to coordinated grid cell replay in hippo’s principal cortical output region
- they also found that place cells
were initiating the replay and the
grid cells were behind 10 ms
Alonso et al (2020)
task 6 -
- hippocampus separates all experiences by time so it can differentiate them - complementary learning system theory consolidation: hippo is a “fast” learner and high plasticity and cortex is a “slow” learner with low plasticity – this allows extraction of overlapping principles to an abstract system after memory is encoded – this memory will lose details and only remember “gist” like info - advantage of slow learner (pfc) is that contrasting info (e.g penguin after learning wings=flying) is stored without erasing the other info
- Schema – great way to test semantic
memory – large cortical network that creates
long term memory only with one experience
and it comes along with rapid systems
consolidation (the longer you are in Nijmegen
the better schema you’d have) - in rodents there is a task to measure this,
rat needs to get the same cue as it ate before
(e.g. strawberry flavour)
tse (2011)
task 6
- gene expression in the cortex – new pair
associate has higher gene expression - if limbic cortex is turned off (ampa receptor
inhibitor) during encoding animals do not
remember – gene expression in limbic cortex
(= rat frontal cortex) is important otherwise
they wouldn’t remember – if PFC is not
active during updating of schema / semantic
memory, then there is no rapid systems
consolidation
buzsaki 2013
task 6
- semantic memory in rodents is a under
researched field - semantic memory is “facts”. Episodic and
semantic memories are both declarative. - there is no rodent test that tests semantic
memory - many contrasting ideas of what the hippo
does - Space: path integration and allocentric
system (not egocentric) and episodic memory
evolves from egocentric to allocentric view of
the world (semantic) - Episodic memory: hippo can be responsible
for that, hippo is good at bringing everything
together (H.M didn’t have episodic memory) - index theory: hippo (the librarian) is a point
where all the info are “put” together - circuit researchers: hippo functions only as
pattern completion (CA1) and pattern
(DG)separation - scene reconstruction: new theory – hippo is not very good for memory bc it has a fast turnover of synapses, instead
the hippo does is completing the story that the cortex is telling – need hippo for reconstructing a scene and this
would explain false memory formation (the hippo “fills” in missing info which are integrated from other memories)
genzel (2019)
task 6
- semantic memories could come from
episodic memories - semantic memories formation – you have
an experience (e.g. sitting in class) and some
neurons activate. Recall all the neurons
that were active during memory formation
are also active when recalling - some neurons are more important than
others (neuronal hubs) as if you inactivate
those you lose the whole memory - hippocampal hubs are especially important
for episodic memories - frontal cortex (human - pfc, mpfc, rodents -
prelimbic and acc) hubs are important for
storage - default mode network is the “memory”
network as it includes many memory hubs - object space task is developed – based on principle of
novelty, explore novel object inserted in space more – they
do differently bc they overlap (5 trials spaced with breaks of
45 minutes, they have different objects in each trial but
same location – some locations have been presented less
and they check whether the animal remembers by seeing if
it explores more the item at the least presented location
de quervain (1998)
SCHEMA FIG FROM THIS GUY - lecturer mentions a variety of experiments with him
Stress through glocorticoids impair retrival (see next slides) but enhances
consolidation
Train animal to be in platform,
Shortly before recall they stress animal – In control they recall training and stay more in platform, same for 2 minutes after
stressor or 4 hours after stressor. Significantly, we see that 30 min after stress they
can’t recall where the platform is as well as the other conditions – stress impairs
recall after 30 minutes as the cortisol response takes around 20-30 minutes to raise
plasma corticosterone levels
Glucorticoids impair memory recall!
Monyer 1994
Task 4
Molecular basis of memory development: NMDARs • GluN1 expression is already high at birth and does not change across development • GluN2A is only expressed post-natally and near-adult levels are reached by ~4weeks of age • GluN2B expression is high at birth and gradually goes down until ~4weeks of age • In the adult, GluN2A is the dominant NMDAR subunit. Monyer et al. (1994)
Tang 1999
Task 4 Molecular basis of memory development: NMDARs • Differences between GluN2A and GluN2B • GluN2Bs have higher affinity for glutamate • GluN2As deactivate more quickly • GluN2B synapses have a lower threshold for LTP! • Over-expression of GluN2B leads to enhanced memory
Travaglia 2016
Task 4
Molecular basis of memory development: NMDARs
• Training on a contextual fear conditioning task at
P17 (infantile amnesia period) accelerated the
switch in GluN2A/GluN2B expression
• Perhaps this molecular switch represents a critical
period for memory development?
Travaglia et al. (2016)
Lohmann 2014
Task 4
Molecular basis of memory development: NMDARs • Developmental changes in GluN2A and GluN2B expression levels show that there is a development ‘switch’ in which GluN2 subunit dominates NMDARs • This switch occurs coincidentally with the emergence of episodic memory • What are the differences between GluN2A and GluN2B?
Molecular basis of memory development
• Synaptic plasticity also involves altering
protein interactions in post-synaptic
cells via kinases
• A key kinase is CaMKII
• CaMKII is activated if calcium enters
the post-synaptic cell.
• Once activated it translocates to the
synapse and binds to GluN2Bs, inducing
spine growth.
• CaMKII-GluN2B interactions are vital
for LTP induction and activation of
CaMKII is sufficient to potentiate
synapses
• Interestingly, CaMKII levels reach near-
adult levels during 4th post-natal week
-The developmental period associated with the emergence of memory is marked by
dramatic changes in the molecular composition and processes within excitatory
neurons
• Based on the developmental timeline of various molecular candidates, memory
development may depend on the GluN2A/GluN2B switch and the expression of
CaMKII
Lohmann & Kessels (2014)
