Up-to-date molecular mechanisms of short-term and long-term memory, retrieving, forgetting and memory management

1. In search of memory:
from molecules to mind
By Vasyl Mykytyuk

2. Acknowledgement:
Professor Jerry Rudy of the psychology department
at the University of Colorado at Boulder for his
lectures and textbook “Neurobiology of Learning and
Memory”

3. On average the human brain has 86
billion neurons
Glial cells are x10 to neurons
One neuron can have up to 10 000
synapses
During childhood and particularly
during adolescence a process
“synaptic pruning” occurs – up to 50%
of neurons and synapses are reduced
Our brain always uses 20% of all
energy

4. Brain functions
Arousal
Attention
Consciousness
Decision making
Executive functions
Natural language
Learning
Memory
Motor coordination
Perception
Planning
Problem solving
Thought
Mnemosyne (1881), mother
of all Muses, goddess of
memory by Dante Gabriel
Rossetti
Olin Levi Warner,Memory (1896).
Library of Congress Thomas
Jefferson Building, Washington, D.C.

7. The case of Henry Molaison
Henry Molaison
1926-2008

9. Forgetting curve by Herman Ebbinghaus (1913)

10. Two possible explanations of forgetting curve

15. Resting potential formation

17. ЕЛЕКТРИЧНЕ ПРОВЕДЕННЯ
ІНФОРМАЦІЇ
• Membrane potential
– Difference of charges across
the plasma membrane
• Resting potential
– Resting cells are (-) inside
and (+) outside
– Large amounts of Na+
outside the cell and K+ inside
• Action
potential/impulse
– Rapid reversals in charges
across the plasma membrane
– Caused by the exchange of
ions across the membrane of
the neuron
– Threshold level (-55mV)
needed to stimulate neurons
ALL-OR-NONE principle

18. Sequence of event
during synapse:
1. Arrival of action
potential to the
presynaptic axon
terminal
2. Neurotransmitter
release from synaptic
vesicles into the
synaptic cleft.
3. Molecules of
neurotransmitter
binding to specific
receptors.
4 Induction of a
chemical or
electrical signal in the
postsynaptic cell.

19. The synaptic plasticity hypothesis,
“the strength of synaptic connections—
the ease with which an action potential in
one cell excites (or inhibits) its target cell—
is not fixed but is plastic and modifiable”
(Squire and Kandel, 1999, p. 35).

20. Long-Term Potentiation
(LTP) is produced by
high-frequency
stimulation (100 Hz =
100 times in a second,
three times with 20
seconds between them)
Long-Term Depression
(LTD) is produced by
low-frequency
stimulation (900 pulses
of frequency 1–3 Hz ~
15 minutes)

23. Dendrites are extensively populated with structures called dendritic spines. A spine is a small (submicrometer)
membranous extrusion that protrudes from a dendrite. This figure illustrates a three dimensional
reconstruction of a section of a dendrite with spines of different shapes and sizes. Dendritic spines are of
special interest because (1) key receptors involved in the regulation of synaptic plasticity are
located in spines and (2) changes in the composition and architecture of the spine are
altered by neural activity.

25. (A) A schematic representation of an excitatory synapse. The PSD is opposed to the
presynaptic active zone, attached to F-actin. (B) A schematic of the PSD. Core scaffold
proteins of the PSD— PSD-95, GKAP, Shank, and Homer—interact with each other and
other proteins and are thought to form a lattice for the assembly of postsynaptic
membrane glutamate receptors and signaling molecules.

26. Mechanisms of ionotropic receptor action

27. Two types of ionotropic glutamate receptors:
The a-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA) receptor,
The N-methyl-D-aspartate (NMDA) receptor
They are located in dendritic spines, play a major role in the induction and
expression of LTP. When gluatamate binds to these receptors their channels open
and positively charged ions in the extracellular fluid (Na+ and Ca2+) enter the
neuron.

28. APV, an NMDA receptor antagonist, prevents the induction of LTP but has no effect on
its expression. CNQX, an AMPA receptor antagonist, prevents both the induction and
expression of LTP.

29. NMDA receptor has two
critical binding sites. One
site binds to glutamate and
the other site binds to
magnesium - Mg2+.
Magnesium binds to a site in
the pore, forming a
magnesium plug.
Calcium enter requires two
events:
1) glutamate has to bind to
the NMDA receptor;
2) the synapse has to
depolarize (by AMPA
receptor).

30. The induction of LTP occurs when extracellular calcium enters the spine via the NMDA calcium
channel. But entrance of calcium is not needed for LTP expression.

31. AMPA trafficking is regulated constitutively (double arrows) and by synaptic activity (single
arrow). The second one is a mechanism for LTP-formation.

33. Changes in synaptic strength that support LTP evolve in stages that can be
identified by the unique molecular processes that support each stage.

34. Generation and
stabilization of trace

35. Constitutive trafficking of AMPA receptors

37. AMPA receptors become trapped in the PSD when (1) the influx of calcium into the spine via NMDA
receptors (2) activates CaMKII. This kinase phosphorylates serine (P-serine) residues on the terminal
of the TARP Stargazin. This activity releases the terminal from the membrane and (3) facilitates its
binding to PSD-95 complexes, thereby trapping the receptor in the PSD. The stability of the trap
increases as CaMKII phosphorylates more of the Stargazin serine residues.

39. Filament actin in the spine head is crosslinked with spectrins. (B) The protease calpain is
activated when calcium enters the spine through the NMDA receptors. Calpain degrades
spectrins and facilitates the dissembling of actin. C. This activity contributes to the rapid
insertion of additional AMPA receptors in the PSD.

40. Stabilizing the Trace

41. (A) (B)

42. (A) When drugs such as latrunculin and cytochalasin that prevent actin polymerization are
applied before TBS, they do not prevent the induction of LTP, but it rapidly decays.
(B) LTP is maintained if these drugs are applied 15 minutes after LTP is induced.

44. N-cadherins are reorganized by an LTP-inducing stimulus. (A) Illustration of an unpotentiated
synapse with the presynaptic terminal and postsynaptic spine weakly bonded by cadherin monomers. (B)
A high-frequency stimulation is delivered to this synapse to induce LTP. (C) The high-frequency stimulus
promotes cadherin dimerization so the now-enlarged spine containing additional AMPA receptors is
tightly coupled to the presynaptic terminal and well positioned to receive glutamate released from the
presynaptic terminal.

45. An LTP-inducing stimulus enhances the function of integrins in two ways. (A) Prior to induction
integrins are below the plasma membrane. Integrin interacting epitopes are encrypted and the activity
of metaloproteinases (MMP9s) in the extracellular matrix (ECM) is low. (B) The induction stimulus
increases the expression of integrins in the spine head and the activity of MMPs in the ECM. The
proteolytic activity of MMPs directed at encrypted epitopes increases availability of integrin
interacting epitopes.

46. Consolidation of Synaptic Changes:
Translation and Transcription

47. Consolidation of Synaptic Changes: Translation and Transcription
New protein synthesis is essential for consolidation of synaptic changes

48. Synaptic activity
simultaneously
induces local
mRNA translation
and genomic
signaling cascade,
that trigger
transcription of new
mRNA

49. Messenger RNA and protein translation machinery, such as endoplasmic reticulum (ER) and ribosomes,
is present locally in the dendritic spine region. Synaptic activity can activate this machinery and translate
this mRNA.

50. Consolidation of Synaptic Changes: Specific Mechanisms

51. BDNF activates local protein synthesis through PI3K/mTOR pathway

52. (A) Release of caged glutamate, coupled with a spike-time protocol, produces
sustained growth and maintenance of a single spine. (B) Inhibiting the TrkB response
to BDNF prevents sustained growth and maintenance. This result indicates that this
protocol releases BDNF into the extracellular space.

53. General models
of how the
signals reach the
nucleus:
synapse-to-
nucleus
signaling and
soma-to-
nucleus
signaling

54. Synapse-to-Nucleus
Signaling
Different signaling cascades
lead to cAMP-responsive,
element-binding (CREB)
protein phosphorylation and
transcription initiation

55. Soma-to-Nucleus Signaling. Calcium enters through Voltage-Gated Calcium
Channels in soma of neuron and activates calmodulin and
Calcium/calmodulin-dependent protein kinase type IV which in its turn activate
CREB

56. “cAMP Response Element-Binding Protein Is a Primary Hub of Activity-Driven
Neuronal Gene Expression”
Eva Benito, Luis M. Valor, Maria Jimenez-Minchan, Wolfgang Huber and
Angel Barco
The Journal of Neuroscience, 14 December 2011
CREB was found to occupy ≈4,000 promoter sites in vivo, but many of them are
cell-specific
http://in.umh-csic.es/NADtranscriptomics/tablas/CREB_regulon.txt
http://in.umh-csic.es/NADtranscriptomics/tablas/CREB_timecourse.txt

58. Calcium entering the neuron through NMDA receptors is critical for the production of
LTP but this source of calcium alone may not be sufficient to produce any form of LTP

59. Ryanodine receptors (RyRs)
Calcium-induced calcium release
(CICR) - extracellular Ca2+ enters
through the NMDA receptor and binds to
ryanodine receptors (RyRs) to release
additional Ca2+ from the ER

60. Glutamate binds to a subtype of metabotropic receptor called
mGluR1, which is coupled which G-protein, which in its turn activate
an effector-enzyme (Phospholipase C) . When IP3 binds to IP3R
calcium is released from ER in the dendritic compartment.

61. A single TBS train produces a short-lasting form, LTP1; four trains produce a
moderately lasting form, LTP2; and eight trains produce a very long-lasting form, LTP3.
• LTP1 depends only on RyRs releasing ER calcium in the dendritic spine
compartment.
• LTP2 depends on IP3 receptors releasing calcium from ER in the dendritic
compartment.
• LTP 3 depends on calcium entering the soma through vdCCs.

62. How do mRNAs transcribed in response to synaptic activity find their way to the correct
synapses?
Synaptic tag and capture hypothesis(Frey, Morris, 1998)

64. What is the molecular basis of the synaptic tag?
Changes in functional and structural properties of spines that depend on actin
filaments are a major component of the tagging operation.
Potentiated spines contain more AMPA receptors and an enlarged actin cytoskeleton. Enlarged spines are better traps or tags
for capturing new mRNAs/proteins needed for consolidation. One possibility is that actin-associated proteins interact with
microtubules to cause them to invade the spine and permit motor proteins to cargo the relevant mRNAs/proteins to the spines
that have been potentiated. In addition, once consolidated, spines with an enlarged actin cytoskeleton might successfully
compete for the molecules that are needed to self-maintain and thus provide part of the solution to the molecular turnover
problem.

65. The degradation of existing proteins is important for LTP production.
Components of the UPS are
localized in dendrites,
synapses, and the nucleus.
Glutamate transmission
activates the ubiquitination
processes and the activation
of CaMKII plays a
critical role in recruiting
proteasome into dendritic
spines. Moreover,
stimulation that induces LTP
activates this system and
results in protein degradation
Protein Degradation Influences Three Phases of LTP: (a) an early stage that
involves only post-translation modifications, (b) a later stage that depends on local
protein synthesis in dendritic-spine regions, and (c) a second wave of proteins
synthesized from new mRNA that is a product of genomic signaling cascades

67. SUMMARY OF CONSOLIDATION PROCESS

68. Maintaining the Consolidated Trace

69. Maintaining the Consolidated Trace
the molecular turnover problem. The synaptic molecules that support memory traces
are shortlived in comparison to the duration of our memories.
Crick, however, suggested a
general answer: “… the molecules in synapses interact
in such a way that they can be replaced
with new material, one at a time, without altering the
overall state of the structure.”

70. (A) An unpotentiated synapse is depolarized. (B) As the synapse is potentiated,
AMPA receptors are added to the postsynaptic density. (C) Two processes—(1)
degradation of AMPA receptors and (2) receptor endocytosis — then pressure the
synapse to (D) return to its pre-potentiated state.

71. (A) Kinases like PKC have an
inhibitory–regulatory domain and a
catalytic domain. Such kinases
require a second messenger (SM)
to remove the inhibitory domain
and expose the catalytic domain.
(B) PKMζ has no inhibitory domain
so it does not require second
messenger to be catalytic. It is
perpetually active.
(C) PKMζ mRNA is present in
dendritic spine regions and its
translation requires synaptic
activity to activate several kinases
and actin filaments.
(D) Once translated, PKMζ self
perpetuates; it interacts with
other proteins to remove the
translator repressor and thereby
facilitates translation of PKMζ
mRNA in dendritic spines where it
was initially translated.

72. (A) As noted, PKMζ has no inhibitory domain; however, the peptide ZIP functions as an inhibitory
domain and inactivates it. (B) ZIP applied to a slice prior to the inducing stimulus does not affect the
early induction phase of LTP but does prevent the late phase. (C) ZIP applied 2 or 5 hours following
induction reverses late-phase LTP. (After Serrano et al., 2005.)

73. A pool of GluA2 AMPA receptors is trapped outside of the synapse by protein
interacting with C kinase 1 (PICK1). Release of this pool depends on a
trafficking protein enzyme, NSF. When PKMζ interacts with the NSF–PICK1
complex, the receptors are released for entry into the PSD.

74. (A) The activation of Trk (tyrosine
kinase) receptors
phosphorylates tyrosine sites on
GluA2 AMPA receptors. This
activity releases the receptors from
the PSD into the endocytotic zone
where they bond with the NSF–
PICK1 complex for recycling.
(B) The
presence of PKMζ disrupts the
normal endocytic
cycle—the receptors remain
trapped in the PSD and
the synapse remains potentiated.

75. PKMζ maintains potentiated synapses through two parallel and
complementary actions: (1) it facilitates the release of nonsynaptic
pools of GluA2 receptors to replace GluA1 receptors and (2) it
interferes with the endocytic cycle that normally removes GluA2s
from the synapse.

76. (A) Once synaptic activity induces
PKMζ synthesis in a spine
compartment (spine 2 above) in
response to synaptic activity,
continued synthesis could lead to
PKMζ spreading from the original
compartment to neighboring
spines (1 and 3), where it could
potentiate these spines by
releasing GluA2 receptors (not
shown).
(B) The spread of PKMζ is minimized
because once the GluA2 receptors
are released into the spine (1)
they are free to recapture the
PKMζ–PICK1 complex (2). In
principle this autotagging process
traps PKMζ in the spines where it
was synthesized. Because PKMζ
disrupts the endocytic cycle, an
additional outcome is that
GluA2s remain in the PSD region and
the spine remains potentiated.

77. Mice were genetically modified to lack PKMζ. However, established LTP was
maintained and, moreover, ZIP reversed the LTP maintenance.
These unexpected results support three conclusions: (1) enduring memories and
LTP can be established independent of the presence of PKMζ, that is, the
maintenance of LTP does not require PKMζ, (2) ZIP is not selective for PKMζ as it
also interacts with other molecules, and (3) there must be other kinases or other
molecules that can participate in the maintenance of LTP and memories. There
are at least two alternative explanations of these data.
• PKMζ is the primary memory maintenance molecule in normal rats, and the
genetic studies have uncovered redundant compensatory molecules whose
contribution is masked by the normal presence of PKMζ.
• PKMζ is not involved in the maintenance of LTP and maintenance depends on
other kinase targets of ZIP.
The field will now need to answer several questions:
• What are these new targets of ZIP that must be engaged to support
maintenance?
• Are the processes these new targets regulate redundant with PKMζ?
• What if any role does PKMζ play in maintenance?
Until these questions are answered it may be appropriate to conclude that PKMζ
and other kinase
targets of ZIP maintain the synaptic changes that support LTP.

78. Prion-like protein (cytoplasmic polyadenylation element binding
protein (CPEB)) is a candidate for maintenance molecule

79. (A and B) Schematic models of pathogenic (A) and functional (B) prions.
(C) Antibody that is specific for the aggregated (functional prionic) form of ApCPEB
selectively blocks the maintenance of long-term facilitation produced by 5HT.

81. Large spines become resistant to modifications by calcium efflux out of the neck

82. Some spines are large (A) – they remember, but there are always new filopodium-like
spines (C) grow to find synapses-contact, thus there can not be problem of dendritic
saturation with stable spines

83. Epigenetics:
our experience can affect
our DNA

84. Memory formation involves epigenetic mechanisms,

85. * 5-AZA – 5-azacytidine

90. Epigenetics: The sins of the father
The roots of inheritance may extend beyond the genome, but the mechanisms remain a puzzle.
Virginia Hughes

91. Every day we
change our brain,
that`s why we will
never be the
same tomorrow…

92. Molecular mechanisms
of forgetting

100. Figure 7. Model of Dopamine-Mediated Learning and Forgetting The left panel illustrates dopamine-mediated learning.
Upon electric shock, a US signal is delivered when DANs release large amounts of dopamine (red spheres) onto the MBs.
Binding of dopamine to the dDA1 receptor (green) leads to increased cAMP, while the CS odor stimuli results in increased
Ca2+ signaling. Coincident Ca2+ and cAMP signaling leads to strong memory formation within the MBs (Tomchik and
Davis, 2009). The right panel illustrates dopamine-mediated forgetting. After learning, a low level of ongoing dopamine
release is sensed by the DAMB receptor (red) and weakens the memory over time. Differences in affinity of the two
receptors for ligand, compartmentalization, coupling to downstream signaling pathways, or signaling kinetics may account
for the dominant action of dDA1 in acquisition and DAMB in forgetting.
2012

101. Spike-timing-dependen-plasticity (STDP)
those who fire together, wire together
NMDA receptor activation through depolarization after presynaptic activation
results in enhancement, NMDA receptor activation through depolarization
before presynaptic activation results in weakening of synaptic strength, both
thought to be mediated by different levels of calcium [34].

102. Memory modulation

103. NMDA receptor can include either GluN2A or GluN2AB
Receptors with GluN2AB take more time to close, so much
more Calcium is influxed
During our maturation our GluN2AB are replaced by GluN2A,
so it is more difficult to strenghten our synapses
Joe Tsien genetically engineered mice, overexpressing
GluN2AB in hippocampus and cortex.

104. Gary Lynch and his colleagues have developed a class of drugs
called ampakines that may enhance cognitive function, AMPAr are also
critical for retrieval
Sunifiram, Unifiram

105. Memory modulation framework

106. Epinephrine does not cross the blood brain barrier. However, when it is released
from the adrenal medulla it binds to adrenergic receptors on the vagal nerve. In response to activation, the
vagal nerve releases glutamate on neurons in the solitary tract nucleus (NTS). Activated NTS neurons
release glutamate onto neurons in the locus coeruleus, which in turn release norepinephrine that binds to
adrenergic receptors in the basolateral amygdala (BLA). Disrupting any component of this neuro-hormonal
circuit will prevent arousal from enhancing memory.

109. The Fate of Retrieved Memory

110. retrieving or reactivating the memory can return it to an active, labile state

111. Active trace theory

112. Reconsolidation theory
the synapses underlying the trace become unbound or weakened
retrieval also initiates another round of protein synthesis so that the trace is
“reconsolidated.”

115. Model representation of reconsolidation cascade(Tronson, Nature reviews, 2007)

118. Multiple memory systems
Hierarchy of Episodic memory system

119. The indexing theory of hippocampus role in Episodic memory

120. Hippocampus against entropy

121. Ribot`s Law explanation

124. Frederick Bartlett`s serial reproduction

126. Memory during sleep

127. 2004
Ribeiro, Nicolelis, 2004

128. Memory manipulations

132. Nature, 01 June 2014

133. 2014

134. Іноді достатньо згадати те,
що викликає неприємні емоцї
в іншому контексті. Адже
згадуючи, ми кожного разу
переписуємо свої спогади
Пам’ять – це не те, що
сталося з нами колись.
Пам’ять – це те, що ми є
зараз.
В нашому мозку не
зберігаються всі
“документи” нашого життя.
В ньому зберігається лише
остання копія.