1. LTP and LTD

2. Historical influences – what is a memory?
Santiago Ramón y Cajal - proposed memories are formed by strengthening connections between existing neurons to increase the effectiveness of their communication. (1894)
Hebb’s postulate
“When an axon of cell A is near enough to excite a cell B…
and repeatedly or persistently takes part in firing it…
some growth process or metabolic change takes place in one or both cells
such that A's efficiency, as one of the cells firing B, is increased.”
-Donald O. Hebb (1949)
To paraphrase:
“Inputs that fire together, wire together.”

3. LTP (long term potentiation) and LTD (long term depression)
Leading candidate molecular mechanisms underlying learning and memory.
Fundamental forms of synaptic plasticity.
Widespread in the brain, but may have different variations
-therefore important to specify which neurons are being studied.
LTP and LTD are models of what is occurring during learning and memory (proposed explanations)

4. Laminar Structure of Hippocampus
Hippocampus is critical for formation of spatial and explicit memories
Well characterized and highly organized, laminar cytoarchitecture and stereotyped wiring pattern
Amaral and Witter, Neuroscience 31: (1989)

5. Tri-synaptic Circuit of Excitatory Synapses in the Hippocampus
stim
Intracellular recording
fEPSP (“fields”)
CA1
s. oriens
s. pyramidale
s. radiatum
Schaffer Collaterals
to cortex 3
CA3
from cortex
Perforant Path 1
Mossy Fibers 2
Granule cell
s. moleculare
Dentate Gyrus
Modified from Blitzer et al., Biol Psychiat. 57:113 (2005)

6. Finding Hebb in the Hippocampus
“…repeatedly or persistently takes part in firing it…”
Induction
Stim. Schaffer Collateral at 100Hz, for 1 sec (tetanus)
CA1
100Hz tetanic stimulation
(High frequency stim (HFS)
tetanus
Baseline
Test pulse -
Stim. Schaffer Collateral at 0.2Hz
CA1
(Bliss and Lømo, 1973)

7. Producing LTP Baseline Induction Post-Induction
Stim. Schaffer Collateral at 0.2Hz
CA1
Induction
Stim. Schaffer Collateral at 100Hz, 1sec (tetanus)
CA1
tetanus
Post-Induction
Stim. Schaffer Collateral at 0.2Hz
CA1

8. Weak HFS Induces Transient Potentiation; Strong HFS Induces Stable LTP
cooperativity
Nature Rev. Neurosci. 4; (2003)
(2x 100 Hz, 1 sec, D1.5mV)
(1x, low intensity, D0.6mV)
Tsokas et al., J. Neurosci. 25:5833 (2005)

9. Repetitive stimulation induces stronger LTP
-One tetanus may cause LTP of several minutes to a couple of hours
-Repeated trains (4) leads to longer-lasting LTP (many hours, or until the slice dies)
-reminiscent of strengthening of memory with repetition

10. Associativity of LTP Weak and strong inputs are both potentiated
-but only if they are associated (occurring within short time window)
Three key properties of long-term potentiation (LTP) were elucidated in the 1970s and early 1980s. All of these can be explained mechanistically by the biophysical properties of the NMDA (N-methyl-D-aspartate) receptors that are required to trigger LTP.
Cooperativity
This term refers to the fact that when using high frequency stimulation to induce LTP, a crucial number of presynaptic fibres must be simultaneously activated — they must 'cooperate' to elicit LTP (that is, a threshold stimulation strength must be used). The frequency of stimulation interacts with the stimulus strength such that increasing one decreases the requirement for the other to trigger LTP. This property is explained by the fact that, to trigger LTP, the postsynaptic cell must be sufficiently depolarized to allow current (particularly Ca2+) to flow through the NMDA receptor channel.
Input (synapse) specificity
This property refers to the fact that, when LTP is elicited at one set of synapses on a postsynaptic cell, adjacent synapses that were not activated during the induction protocol do not show LTP. This property is explained by the requirement that, to elicit LTP, synaptic NMDA receptors must be activated, leading to a spatially restricted increase in intracellular Ca2+ in the relevant dendritic spine.
Associativity
This term refers to the fact that LTP can be elicited at synapses that are activated by low-frequency, sub-threshold stimuli if their activation is temporally concurrent with an LTP-inducing stimulus at another set of synapses on the same cell. This property is explained by the fact that the LTP-inducing stimulus provides the requisite depolarization, which is rapidly transmitted through the dendritic tree to those synapses in which the NMDA receptors were simultaneously activated by the sub-threshold stimulus. It is easy to envision how this property makes LTP an attractive mechanism for associating two pieces of information being conveyed by different sets of afferents that synapse on the same postsynaptic cell.
Please close this window to return to the main article.
Nature Rev. Neurosci. 4; (2003)

11. Input-specificity of LTPA B
specificity
associativity
Path A
Path B
A Only B Only A+B
Baseline
After HFS to Path A
Blitzer, Long-term potentiation: Mechanisms of induction and maintenance. Sci. STKE 2005

12. LTP Can Persist for Months / years In Vivo
(20-50T, dentate gyrus)
Abraham et al.,
J. Neurosci. 22:9626 (2002)

13. LTP has properties expected of a physiological basis for memory
1. RAPID – induced within seconds
2. GRADED – can be strengthened with repetition or intensity
3. ASSOCIATIVITY – can integrate inputs occurring at same time
4. SYNAPSE SPECIFIC - Induced only at appropriately stimulated synapses, and not neighboring synapses
5. PERSISTENT – can last up to months or longer in vivo
6. INDUCED BY PHYSIOLOGICAL STIMULI – “theta-burst stimulation” (TBS) mimics endogenous hippocampal rhythms

14. Temporal stages in LTP
(1) Induction (early phase, E-LTP)
Analogous to “learning?”
(2) Maintenance (late phase, L-LTP)
Analogous to “memory?”
Probably more stages exist
Different stages involve distinct mechanisms

15. Key Criteria for ‘Core’ events:
(1999)
Key Criteria for ‘Core’ events:
(1) Is the pathway activated upon LTP induction?
(2) Is pathway activation required for LTP?
(3) Does activation of the pathway mimic LTP?
(4) Does activation of the pathway occlude LTP?

16. Stimulation of NMDA-type Glutamate Receptors
LTP Induction : key steps
Stimulation of NMDA-type Glutamate Receptors

17. NMDARs are “coincidence detectors”
“Play The LTP Game”
Glu
Glu
Glu
Glu
-glutamate binds AMPA and NMDA receptors but NMDARs are blocked at resting membrane potential by extracellular Mg2+
-AMPA receptors mediate most fast excitatory synaptic transmission
-if glutamate is applied and membrane is depolarized at the same time, NMDA receptors are unblocked and allow calcium influx (Hebbian)
-biophysical properties of NMDARs well suited for association

18. NMDA receptors are critical for LTP and learning
I. Pharmacological inhibition
Inhibition of NMDAr with APV abolishes LTP in vivo
Nature 319, (27 February 1986)

19. What about behavior?
Morris water maze: test of spatial memory
Requires hippocampal function (lesioned animals perform poorly)
Nature 319, (27 February 1986)
AP5 infusion into brain impairs performance on water maze
II. Genetic inhibition
Knockout of obligate GluN1 subunit of NMDARs in CA1 also blocks LTP and impairs spatial learning
Tsien et al., Cell 87:1327 (1996)

20. LTP Induction Requires Postsynaptic Ca2+
Malenka et al. Neuron 9:121 (1992)
Requirement for Ca2+ is Brief
photoactivated Ca2+ chelator diazo-4
Stimulation of NMDA-type Glutamate Receptors
Increased Postsynaptic Ca2+

21. During LTP, CaMKII is Activated in Dendrites
Phospho-CaMKII
CONTROL LTP
Stimulation of NMDA-type Glutamate Receptors
Increased Postsynaptic Ca2+
Activation of CaMKII
+LTP
Giovannini et al. J. Neurosci. 21: 7053 (2001)

22. CaMKII: How It Gets Activated
Persistent activation can be a form of “molecular memory”
Binds directly to NMDA receptors
Lisman et al. Nature Rev. Neurosci. 3:175 (2002)

23. LTP Induction Regulates AMPAR channel properties and synaptic content
In 10 µM KN-62; 60 min after HFS:
Glu
Mostly GluA1/2 and GluA2/3 heteromers in adult hippocampus
Stimulation of NMDA-type Glutamate Receptors
Increased Postsynaptic Ca2+
Activation of CaMKII
Enhanced AMPA-type Glutamate Receptor Function / Content
Barria et al. Science 276:2042 (1997)

24. Phosphorylation of AMPARs During LTP
YES CaMKII Site  NO
PKA site 
Lee et al. Nature 405:955 (2000)

25. AMPAR delivery in LTP: Silent Synapses
functional
LTP
silent
functional
NMDAR
AMPAR added
LTP B1 B2
Isaac et al, 1997, Neuron, 18:269

26. Building a Smarter Mouse Overexpress NMDA Receptors
Two relevant key points:
1. GluN2B expression is high early in development and decreases with age (opposite to GluN2A)
2. GluN2B allows greater calcium influx than GluN2A
Nature 401: (2 September 1999)

27. Overexpression of GluN2B Improves LTP …
Mutant Lines
Controls
Day 18
GluN2B Tg
Tang et al., Nature 401:63 (1999) WT

28. … and Improves Recognition/Spatial Memory
Preference for Novel Object
Hidden Platform
Tang et al., Nature 401:63 (1999)
Been there, done that
Caveat: mice may feel more inflammatory pain, which is NMDAR-dependent in pain-related forebrain areas
Wei et al., Nat. Neurosci.  4, (2001)

29. Summary: Core Signaling Events in Induction of LTP at CA3-CA1 Synapse
Baseline Synaptic Transmission
(AMPAR-Mediated)
During LTP Induction
Blitzer, Long-term potentiation: Mechanisms of induction and maintenance. Sci. STKE 2005

30. The Consolidation of LTP: Induction vs. Maintenance
Theta pulse stim (5 Hz)
Stable reversal of LTP by theta pulse stimulation delivered 15 min after LTP induction in area CA1 of the freely moving rat. A, Group data from five animals showing that a 1 min train of TPS administered 15 min after TBS caused a lasting reversal of LTP. A second TBS episode administered 3 d after TPS produced a potentiation effect that was stable for the 4 d of recording thereafter. Each data point represents the group mean ± SEM (averages of 4 successive responses per animal and data point) of the initial slope of the dendritic field potential expressed as percentage of the baseline.
Failure to obtain LTP reversal when theta pulse stimulation is administered 30 min after induction. A, Group data from six animals illustrating that LTP was not affected by a 1 min train of TPS delivered 30 min after induction and continued to persist for the 4 d of recording thereafter. Each data point represents the group mean ± SEM (averages of 4 successive responses per pathway and data point) of the initial slope of the dendritic field potential expressed as percentage of the baseline
Similar to memory consolidation
Staubli and Scafidi J. Neurosci. 19:8712 (1999)

31. CaMKII Activity is Not Needed for LTP Maintenance
-Simiarly, NMDARs also are not required for LTP maintenance
(after LTP is induced, APV has no effect)
but induction still blocked
Chen et al. J. Neurophysiol. 85:1368 (2001)

32. LTP Maintenance Requires Protein Translation
(protein synthesis inhibitor)
Osten et al.
J. Neurosci. 16:2444 (1996)

33. Requirement for transcription in L-LTP
(E-LTP)
min
ACT = actinomycin D, a transcriptional inhibitor
Nguyen et al Science 265:

34. LONG TERM DEPRESSION (LTD)
-Dudek & Bear, 1992
-Widespread--possibly found in all excitatory synapses
-best studied in hippocampus, striatum, and cerebellum

35. LTD in the Hippocampus CA1
Induction
Low frequency stimulation (LFS)
(e.g. 1 Hz stimulation for 15 minutes)
J. Neurosci., July 1993, 13(7):

36. NMDARs are required for LTD at CA1 synapses
J. Neurosci., 13: 2910 (1993)
LTD is graded and saturable but does not occlude LTP
PNAS 89: 4363 (1992)
‘De-depression’
HFS
1Hz

37. How can NMDARs be required for both LTP and LTD?
Simple Model:
Ca2+
Presynaptic
Postsynaptic
NMDA receptor
HFS
LFS
Modifed from Lisman, 1989; Bear and Malenka, 1994.
The level and pattern of calcium influx is the critical difference

38. Bidirectional control of synaptic strength by kinases and phosphatases: Tri-state model
De-depression
Potentiation
PP2B
Depotentiation
Depression
Lee et al., Nature (2000) 405: 955

39. Clathrin-mediated endocytosis
LTD mechanisms II: AMPAR internalization
Clathrin-mediated endocytosis
Man et al., Neuron 25: 649–62, 2000
Insulin can also trigger endocytosis of ampa receptors for some bizarre reason – also clathrin coated pits.
GST-amphiSH3 (glutathione –S- fusion protein amphiphysin SH3 domain) specifically interacts with endogenous dynamin and prevents it from taking in receptors– the 3m form fails to bind to dynamin and it cannot prevent dynamin from endocytosingS

40. Bidirectional signaling in LTP and LTD
KINASES
PHOSPHATASES
Calcineurin/
PP1

41. What are spines?
Cajal, 1888, drawing of Purkinje cell, Golgi staining – “thorns or short spines”
1906 – Nobel awarded to Cajal + Golgi
Golgi-stained modern day micrograph of cortical pyramidal neuron

42. Morphology of spines 7 μm
Hippocampal CA1 neuron, calcein imaged using 2-photon laser scanning microscopy (2PLSM)
Nimchinsky et al., 2002 Annu. Rev. Physiol.

43. Large diversity exists in spine shape
Hippocampal CA1 neuron, GFP-filled: 3D reconstruction of a dendritic segment from confocal z-stacks
McKinney, 2005, Biochem Soc Trans

44. Spines and synapses
>90% of excitatory synapses in CNS terminate on spines
Neurotransmitter receptors are largely restricted to the surface of the spine
synapse
axon
axon
“en passant”
dendrite
dendrite
Menahem Segal
Nature Reviews Neuroscience 6,   (April 2005)

45. Dendritic spine development
“immature”
“mature”

46. Structural Synaptic Plasticity
Live confocal
imaging of hippocampal neuron transfected with GFP-actin
Dendritic spines are highly plastic and motile.
This motility depends on actin cytoskeleton dynamics.
Neuron. 1998 May;20(5): Rapid actin-based plasticity in dendritic spines. Fischer M, Kaech S, Knutti D, Matus A.

47. Functions of dendritic spines – insights from biophysical properties
Spines allow biochemical compartmentalization:
1. high calcium concentrations possible without causing excitotoxicity
2. input-specific signaling
3. neck controls diffusional coupling to dendrite and provides electrical insulation
4. electrical insulation prevents saturation of signaling by a few inputs (thus allows integration of inputs in a more linear fashion)
Spine head width
Spine length
Parent dendrite shaft
Spine neck width

48. Helical spine pattern maximizes connectivity
Purkinje cell dendrite
Filtered
(near face)
schematic
necks
heads
-Excitatory axons distribute to as many neurons as possible (minimize double hits)
-Excitatory spines receive as many different inputs as possible onto a neuron
-Creates a maximally distributed network with as many connections as possible to start with, which can later be modified by activity-dependent pruning and learning
Neuron Sep 8;71(5):772-81

49. Spine volume, postsynaptic area and vesicle number are correlated
What about synaptic strength?
Postsynapse area
Vesicle number
KM Harris, 1989

50. 2-photon glutamate uncaging
720 nm
Caged glutamate
glu
2P-EPSC mimics native mEPSCs

51. Spine volume correlates with synaptic efficacy
Matsuzaki et al., 2001, 2-photon glutamate uncaging, Hippocampal CA1 neuron

52. LTP increases spine number
EPSP
Engert & Bonhoeffer, 1999 Nature.
LTP (using pairing protocol) increases spine number in hpc slice culture

53. LTP increases spine size
1 μm
(uLTP)
(NMDAR unblocked)
(HFS)
Time-lapse images of dendritic spines on a hippocampal CA1 pyramidal neuron. Arrowhead indicates spot of two-photon uncaging of glutamate (stimulation for 1 min at 1 Hz). b, Time course of spine-head volume of stimulated (black circles) and neighbouring (green diamonds) spines shown in a.
Matsuzaki et al., 2004.

54. Larger spines are stronger
Spine-head volume (V, open symbols)
Maximal AMPAR currents (I, filled symbols)
All the small spines that showed enlargement immediately after pairing stimulation showed increase in synaptic strength as well.
Matsuzaki et al., 2004

55. Spine changes in long term plasticity
LTP
LTD
Smaller spines are weaker, but more easily potentiated
Larger spines are stronger, but harder to potentiate further
Small spines may be “plasticity” spines; large spines may be “memory” spines

56. Are spines stable over time in vivo?
Imaging results from Karel Svoboda’s lab say ‘no’
Imaging results from Wen-Biao Gan’s lab say ‘yes’
Conclusion: depends

57. In vivo imaging of dendritic spines
Images from somatosensory cortex (barrel cortex)

58. Are spines stable over time? (no, not really)
Pyramidal neuron in barrel cortex
17% of spines are transient
23% of spines are semi-transient
60% of spines are stable
Even “stable” spines are unstable
15% of “stable” spines disappear within 30 days
Trachtenberg et al., Nature 2002

59. Sensory deprivation enhances spine turnover
Barrel cortex
Trachtenberg et al., 2002
Deprived (black circles)  more unstable and fewer stable spines vs. control (white circle)

60. Are spines stable over time? (yeah, sorta)
Pyramidal neurons in visual cortex
Grutzendler et al., 2002

61. Spine turnover decreases with animal age
Pyramidal neurons in visual cortex
More stable with age…
…but fewer in number
Grutzendler et al., 2002

62. Persistence of spines depends on their volume
persistent spines
transient spines
Spine volume and spine brightness are tightly correlated
Larger spines are more stable
Holtmaat et al., 2005

63. Spines form excitatory synapses
Gray’s Type I, asymmetric synapse
Kennedy, 2000

64. The Postsynaptic Density (PSD)
An electron dense area on tip of spine head, occupying ~10-15% of total spine surface area.
Adjacent to the cytoplasmic face of postsynaptic membrane
Close apposition to the presynaptic active zone and docked synaptic vesicles
This organization focuses the synaptic signal, increasing its efficiency, accuracy and speed
Highly insoluble in nonionic detergents due to tight association with actin cytoskeleton – allows biochemical purification
PSD

65. PSDs have a large variety of shapes and sizes
PSDs can be continuous or “perforated”

66. PSD components Contains hundreds to thousands of different proteins:
1. Receptors and voltage gated ion channels (NMDAR, AMPAR, mGluR, calcium channels)
2. Scaffold proteins (PSD-95 family, GKAP, Shank, Homer)
3. Signaling and regulatory proteins (CamKIIa, PKC, PKA, MAP kinases, Ras/Rho/Rac family small GTPases)
4. Cytoskeletal proteins, crosslinkers, regulator (F-actin, cortactin, myosin, cofilin, ARP2/3, alpha-actinin, spectrin)
5. Cell adhesion molecules (cadherins, neurexin/neuroligins, Eph/ephrins)

67. Postsynaptic density (PSD) protein networks
Calabrese B et al. Physiology 2006;21:38-47

68. PSD-95 scaffold protein
PSD-95 is one of the major components of the PSD
Member of the MAGUK (membrane-associated guanylate kinase) family
Other family members:
PSD-93/Chapsyn-110, SAP97/hDlg, SAP102
Scaffold proteins contain several protein-protein interaction domains to tether multiple components
Domains of PSD-95
PDZ1-3

69. Signaling/cytoskeletal protein assembly by PSD-95
NMDAR
NMDAR
PSD-95
Fyn
nNOS
SynGAP
SPAR
PSD-95
CRIPT
GKAP
band 4.1
a-actinin
MAP1A
tubulin
F-actin

70. Postsynaptic glutamate receptors interact with distinct scaffolding / anchoring proteins
Heterogeneous
Dynamic
Less tightly associated with PSD
Consistent
Rel. Stable
NMDAR
AMPAR
PSD-95
Homer
GRIP
PDZ
domain
mGluR
peri-
synaptic
(D. Bredt, C. Garner, S. Grant, R. Huganir, M. Kennedy, P. Seeburg, P. Worley, E. Ziff, M. Sheng)

71. Many neurological diseases have spine defects
Fragile X
Mental retardation
Autism spectrum disorders
Alzheimer’s disease
Schizophrenia

72. Fragile X has supernumerary, thin spines
+Pak inhibitor
FRAX486 WT
FMR1 KO
FMR1 gene is silenced in Fragile X  long / thin spines, too many spines
FMR1 KO mice have too many spines, and they are long and thin (immature)
(also show repetitive movements, hyperactivity, reduced anxiety and LTP, excessive LTD)
P21-activated kinase (PAK) regulates actin dynamics and spine morphology
Dominant negative (dn) PAK kinase has too few spines, and they are short and stubby (opposite of FMR1 KO)
FMR1 KO x dnPAK rescues spines and behavior
(Proc Natl Acad Sci U S A Jul 3;104(27):
FMR1 KO given PAK kinase inhibitor FRAX486 has rescued spines and behavior
Proc Natl Acad Sci U S A Apr 2;110(14):5671-6

73. Spine abnormalities in many diseases
Nature Neuroscience 14,  285–293 (2011)

74. Learning objectives LTP and LTD
1. Understand what synaptic plasticity is, and why it is important for nervous
system function.
2. Understand the role of the hippocampus in memory, and why it is a good
model system for learning and memory studies.
3. Understand the concept of Hebbian plasticity and its relation to LTP
associativity and NMDA receptor coincidence detection.
4. Understand why LTP and LTD are attractive models of synaptic plasticity.
5. Understand how different stages of LTP are induced and the signaling proteins involved.
6. Understand the mechanisms that maintain LTP for different durations.
7. Understand how LTD is induced and expressed.
8. Have an idea of how synaptic plasticity (in models) relates to learning and
memory (in animals), and how this can be experimentally tested.
Different dendritic spine morphologies
Functions of dendritic spines
Morphological plasticity in response to stimuli
Rapid actin-based motility of spines
Spine turnover rates in vivo
What is the PSD and basic constituents
Structure and functions of scaffold proteins at the
Dendritic spine defects in neurological diseases