Turner Lectures (4-6)

Question 1

Engram/Memory Trace

Answer 1

The hypothetical physical manifestation of a memory, associated with the brain regions involved in memory systems.

Question 2

Aplysia Californica (basics, why it’s useful)

Answer 2

AKA Sea Hare
Mollusc, 15-30cm long, feeds on kelp.

Useful for studying because: withdrawal reflexes can be altered through experience, neural circuits underlying this are relatively well understood, neurons are easily accessible for intracellular recording

CNS: Relatively few (20,000) but large neurons. Organised into series of ganglia which communicate via pathways called CONNECTIVES. Made it possible to relate individual neurons to behaviours and create wiring diagrams.

Question 3

Gill and siphon reflex

Answer 3

Aplysia breathes through the gill on its back.
Gill is covered by overhanging mantle shelf.
Parapodia protect the gill by wrapping over dorsal surface of the animal.

Breathing occurs by water being drawn across the gill from the front (between the parapodia) then being ejected through the rear-facing siphon.

Withdrawal reflex of the gill and siphon is adaptive - protects the respiratory organs from harm.

Question 4

Habituation vs sensitization

Answer 4

Habituation: repeated weak tactile stimulation of siphon (non-noxious) –> gradual loss of reflex

Sensitization: strong stimulation (noxious) of tail i.e. electric shock –> enhancement of reflex

Both = NON-ASSOCIATIVE

Can learn to associate the two (i.e. electric shock = US, gill withdrawal = UR, weak siphon stim. = CS)

Question 5

Aplysia training (3 groups)

Answer 5

1) Paired (received CS followed by US)
2) Unpaired (same but with large interval)
3) US alone (sensitization control)

All received 30 training trials with 5 min intervals, then tested with CS alone.

(1) Responded more to CS alone than (2) or (3)
Learning lasted for 4 days.
(2) Had lowest response.

Closer temporal pairing –> larger response to tactile stim.
Duration of memories also increases as the number and spacing of trials increases (distributed learning)

Question 6

CS-US interval

Answer 6

Optimal learning occurs if CS precedes US by 0.5 secs (=forward pairing, positive CS-US interval).

Learning doesn’t occur with backward pairing (negative CS-US interval).

Question 7

Aplysia NS (+wiring diagram. How many MNs etc?)

Answer 7

Bilateral system of 10 ganglia (each ~2000 neurons)

E.g. buccal ganglion, cerebral ganglion etc (see diagram)

Gill withdrawal mediated via “abdominal ganglion”. Individual neurons have been mapped in this ganglion, with specific neurons related to specific functions.

6 MNs control the gill. Input from siphon to gill via these MNs. Also input to these MNs from 40 SNs.

Simplified:
(SN)
I
Siphon —-> MN —-> gill

(Think about it as population representation - see diagram for other inputs)

Question 8

Where do changes occur for habituation? Pathway?

Answer 8

3 possible places: the sensory nerve endings (at the siphon), the ganglion synapse (onto the MN) or the NMJ (onto the gill)

Recordings from cells tell us sensitivity to sensory stimulus is UNCHANGED (same no. of APs)
EPSP of MN is decreased - due to decreased sensitivity or decreased NT release?
Test activation of MNs with exogenous NTs e.g. glutamate –> sensitivity unchanged. Must be decreased release from presynaptic terminal onto MN (homosynaptic pathway)

Smaller EPSPs = less likely to generate APs in MN

Overall number of vesicles remains unchanged, they are just less likely to become docked/release L-glutamate. Originally: due to down-reg of Ca2+ channels (so less Ca2+ entering presynaptic). Now: caused by “silencing of release” - direct switching off of release machinery (mechanism unclear, involves GTP-binding protein Arf).

Question 9

Changes in circuit during (SHORT-TERM) sensitization

Answer 9

No direct modification of synapse used in habituation. Modulatory IN receiving sensory input from the tail –> indirect modification of the synapse output (heterosynaptic pathway)

Leads to increased NT release onto MN

Question 10

Modulatory INs

Answer 10

Modulatory INs that underlie sensitization = serotonergic. Increase in release due to modification of calcium and potassium channel function:

Activation of IN –> 5HT release –> activation of GPCRs (Gs and G0/q-linked)

Gs –> activation of adenylyl cyclase –> increase cAMP –> activation of PKA

Go –> activation of PLC (phospholipaseC) –> increase DAG –> activation of PKC

PKA phosphorylates K+ channels, decreasing K+ conductance and broadening AP
PKA and PKC phosphorylate Ca2+ channels, increasing Ca2+ conductance –> increased Ca2+ –> increase NT release

Question 11

Sensitization blocked by…

Answer 11

PKA inhibitors e.g. Rp-cAMPs

PKC inhibitors e.g. H7

Question 12

Mechanism underlying associative learning

Answer 12

Sensory neuron stim –> activates Ca2+ channels. Ca2+ –> activation of calmodulin –> increased activity of adenylyl cyclase –> increased cAMP level –> PKA –> more transmitter released

Difference between association and sensitization: increase transmitter release occurs DURING INDUCTION as well as after (in sensitization, this is only seen after a tail shock, once tactile stimulation of siphon is resumed). Therefore see larger EPSPs and more postsynaptic depol. during pairing-induction –> activation of L-glu receptors that are voltage dependent (NMDARs) during induction.

Forward pairing leads to calmodulin more effectively priming adenylyl cyclase (than backwards pairing)

Question 13

Effect of blocking NMDARs

Answer 13

Blocked with APV or AP5. Prevents conditioning. EPSP of MN will not increase.

NMDARs are blocked by Mg2+ at negative membrane potentials. Depolarisation –> unblocked, cations can flow through (Ca2+/Na+). Initial depolarisation provided by AMPA (another glutamate R which mediates the EPSPs)

Question 14

Cellular changes in conditioning

Answer 14

Presynaptic changes: same as sensitization (involved in early stages of conditioning)
Also get postsynaptic changes: increased EPSP is detected and fed back to presynaptic (tells it changes occuring should be maintained). Involves NMDA receptors –> increased calcium –> feedback

2x effects on presynaptic adenylate cyclase (CS–>increased calcium–>increased activated calmodulin–>activates AC) (US–>AC activation via 5HT)

Question 15

LTM (massed vs distributed learning paradigms for habituation and sensitization)

Answer 15

Prolonged biochemical up/down reg. = inefficient (could become saturated, meaning learning would cease - memory full)

Extent of LTM determined by whether learning occurs by “massed learning” (repeated exposure in single learning session) or “distributed learning” (repeated exposure over several sessions) - latter = higher retention.

E.g. habituation: tactile stim. over 40 trials. 10 trials on each of 4 days. Retention tested after 1, 7 and 21 days. Habituation still present after 21. If 40 trials given in 1 session, retention is far less.
Sensitization: 1s electric shock given as single shocks (30-120 mins apart) or as trains every 3 secs (30 mins apart). Distributed = 4 trains of 4 shocks (1 train per day for 4 days). Massed = 4 trains of 4 shocks in 1 day.
Or 4 single shocks in 1 day.
In order, where first paradigm has best retention, last has worst (tested 1, 4 and 7 days after training with tactile stim.)

LTM of sensitization requires repeated training. Applying 5ht to onto SN-MN synapse has same effect. (one application –> memory lasts 15 minutes, 5 applications spaced over 90 minutes –> lasts 24hrs)
Therefore 5ht involved in STM AND LTM of sensitizaton.

Question 16

LTM involves PROTEIN SYNTHESIS

Answer 16

5ht involved in sensitization for both STM and LTM. Difference is protein synthesis.

Blocking protein synthesis will block LTM ONLY IF DELIVERED DURING TRAINING and not after. No effect on STM.

Can use transcriptional blockers (e.g. actinomycin D) or translational blockers (e.g. anisomycin).

LTM of sensitization –> increased branching of sensory neurons. Habituation decreases.

Question 17

How is protein synthesis increased?

Answer 17

Thought that 5HT –> activation of PKA, which is then trans-located to the nucleus of sensory neurons to control protein synthesis.
PKA phosphorylates CREB (CRE-binding protein), which activates it.
CREB and cAMP-response-elements (CRE) bind, which –> transcription of IEGs (immediate early genes), which are usually transcription factors.

IEGs have fast but short activation period. Lead to transcription of LRGs (late-response genes), which have longer activation period. Leads to production of synaptic components (ion channels, receptors, cytoskeletal proteins etc) which are targeted to the terminal

Inhibiting CREB inhibits long term sensitization. Injected phosphorylated CREB induces facilitation of MN EPSP.

Question 18

Difference between LTM and STM

Answer 18

STM: supported by ACTIVE processes during working memory operations - persistant neuronal activity

LTM: involves the formation of ASSOCIATIONS between groups of neurons or memory traces - these may extend over large areas of the neural network, may remain latent for long periods without being recalled and are likely due to changes in synaptic strength

Question 19

Cell assemblies

Answer 19

Idea proposed by Hebb - that memories and other cognitive functions are represented in the form of cell assemblies. Any two cells or systems of cells that are repeatedly active at the same time will tend to become ‘associated’ so that activity in one facilitates activity in the other.

Initial experience gets held in STM (working memory). For use later, needs to be converted into latent memory (LTM). Once latent representation is stored in hippocampus, over time it will be reconsolidated and distributed across the neocortex.

Question 20

Delayed Match to Sample (DMTS) Task for working memory. Where does working memory occur?

Answer 20

Monkey fixates on point. Sample appears (item it learns to match), fixates on this. Delay. Multiple items appear, must fixate on correct one. Must have sample in working memory.

Working memory is associated with persistent activity in the prefrontal and higher cortical areas. During task, get neurons firing during delay phase in the prefrontal, motor and entorhinal cortices (PFC maintains info in working memory). Also get activity in hippocampus. Specific CA1 and CA3 neurons are active at different phases of the task e.g. as the trial is starting, during the sample phase, during the delay phase etc.

Question 21

How is persistent neuronal activity (persistent depolarisation) achieved for working memory? (X2 ways)
Where does it occur?

Answer 21

1) positive feedback loops e.g. basal ganglia-thalamocortical loops, reciprocal loops between cortical areas and local recurrent networks (within the cortex)
2) membrane potential bistability. Normally, remove depolarising stimulus –> firing rate of neuron drops back to original level. In some cortical neurons this doesn’t happen – get sustained activity from relatively small input. E.g. entorhinal cortical neurons have specific set of membrane proteins that allow for this. Includes calcium channels (depolarising influx of calcium) and Ican channels (calcium modulated mixed cation channels). Initial depolarisation causes influx of calcium, calcium causes Ican channels to open, causing more depolarisation. Calcium channels remain open.

mAChRs allow for initial depolarisation that activates Ican (ACh important in memory - binds to receptor, allows cations through)

Question 22

Delayed reaching task (receptors involved - 3 types)

Answer 22

Type of DMTS task. Key held down to initiate trial, one of two target switches is illuminated, delay, go signal means monkey can release key and reach for target. Recording from cortex shows NMDA-Rs have much greater contribution to persistent cortical activity than AMPA-Rs (shown using blockers of each - APV & CNQX)

AMPA-Rs tend to generate rapid responses but desensitize/inactivate quickly. NMDA-Rs have slower responses but show much less desensitization so can sustain activation.

Activation of D1-Rs also → up regulation of NMDA-R function (low doses of antagonists for both Rs, which are ineffective alone, –> reduced learning in rats. NMDA-R antagonists have also been shown to inhibit working memory in humans and short term D1-R activation improves working memory in aged primates)

Question 23

“Binding problem” - how are features linked in a representation?

Answer 23

EEG waves/oscillations - gamma and theta waves orchestrate timing of synchronous neural activity across networks.

E.g. implant electrodes in different regions of rodent’s brain. When it runs in a maze, subpopulations of cell assemblies repeatedly fire together. Timing of spikes occurs at specific time in theta oscillation. (cell assembly represented by single wave of gamma)

Also orchestrates timing of sequences (order of cell assemblies): Gamma oscillation - ensures the precise time of firing in a subset of pyramidal neurones. Theta oscillation - a carrier wave that organizes information about a moment in time

Question 24

Hippocampal gamma in freely moving rats

Answer 24

Two types of CA1 gamma:
input from CA3 Schaffer (slow, <60hz)
input from entorhinal cortex (fast, >60hz)

Each occur at different phases of theta cycle
Segregation of the two inputs critical for preventing interference between previous learned associations (via CA3) and new associations (via EC)

Question 25

Hippocampal theta in vivo (rodents)

Answer 25

Prominent in area CA1
Lesioning medial septal nucleus or diagonal band of Broca prevents theta - originally thought to be due to inputs from these areas

Actually generated from inputs from: slm (via entorhinal cortex) and sr (via CA3 Schaffer)

(slm = stratum lacunosum moleculare, sr = stratum radiatum)

Question 26

Hebb’s rule

Answer 26

Imagine 10 inputs to one cell. Locally generated EPSPs can all occur about the same time (within 1-2ms of one another) but input activity not sustained. No sustained pre and post synaptic activity –> no change

If cells fire more persistently, they all participate in the sustained firing of the target cell - summation of all inputs. Synaptic efficiency will be increased with time due metabolic or biochemical change.

Local EPSPs will now be LARGER so reactivation of inputs is more likely to generate output (closer to the threshold). Partial reactivation more likely to generate output.

This requires both pre- and postsynaptic neurons to generate APs. BUT not all synapses are Hebbian (some forms of LTP are non-associative and do not require coincident firing of pre and postsynaptic)

Question 27

Auto-associative neural networks

Answer 27

Bidirectional connections between nodes (a cell or group of cells) are automatically strengthened by conjoint activity (Hebb’s rule)
With partial pattern activation, the network retrieves the stored pattern (pattern completion)