1. Rationale for the claim that LTP is a substrate of memory
  2. Evidence
    1. Molecular approaches relating LTP to learning
    2. Electrophysiological approaches to relating LTP to learning


  1. Experience-dependent
  2. Long lasting
  3. Input specific
  4. Associative
  5. Cooperative


General difficulty for empirical testing:

Ultimate acceptance of LTP as the neural substrate of learning requires the demonstration of a causal link between LTP and behavioural learning


1.1. Is NMDAR-Dependent LTP in the Hippocampus Crucial for Spatial Learning in the Water Maze?

  1. Morris, Anderson, Lynch & Baudry (Nature, 1986)
    1. Chronic intraventricular infusion of AP5 causes a selective impairment of place learning, without affecting visual discrimination learning, see Fig 4-21.
    2. AP5 treatment also suppressed LTP in vivo, see Fig. 4-22.
    3. These results add support to the hypothesis that LTP is involved in some, but not all forms of learning.
  2. Saucier and Cain (Nature, 1995)
    1. NPC17742 (an NMDAR antagonist), completely blocked dentate gyrus LTP
    2. but did not prevent normal spatial learning in rats that had been made familiar with the general task requirements by non-spatial pretraining, see Fig. 4-23.
    3. These results indicate that this form of LTP is not required for normal spatial learning in the water maze.
  3. Bannerman, Good, Butcher, Ramsay, & Morris (Nature, 1995)
    1. A two pool technique
    2. AP5-induced learning deficit can be almost completely prevented if rats are pretrained in a different water maze before administration of the drug.
    3. Non-spatial pretraining can not prevent AP5-induced learning deficit, although it improved performance to some extent.
  4. Confounding side effects of NMDAR manipulation - NMDARs are involved in
    1. Sensorimotor mechanisms
      1. NMDARs are widely distributed throughout the central nervous system, and they function in all sensory modalities and in many aspects of motor output
      2. NMDARs are involved in motor initiation and control and receptor gene expression and release of neurotransmitters in the motor system.
      3. Normal coding and transmission of information in the visual, somatosensory, auditory, and pain modalities require NMDAR activity.
      4. The features that make NMDARs important for LTP are among the features that make NMDARs important both for mechanisms of sensory coding and transmission and for motor output.
    2. Fast synaptic transmission
      1. NMDARs can contribute to the EPSP at synapses the contain both NMDA and AMPA receptors
      2. NMDARs can by themselves induce LTP
  5. Alterations in behaviour caused by NMDAR antagonists could result from several factors
    1. Blockage of NMDAR-dependent LTP (or LTD)
    2. Disruption of NMDAR-mediated sensorimotor function
    3. Impairment of fast synaptic transmission
  6. Water maze task
    1. Experiments have shown that NMDAR antagonists cause severe acquisition impairment in naive rats tested on the water maze task.
  7. LTP and learning in the same test animals ?
    1. Majority of studies on LTP and learning have been correlational, in that the effects of NMDAR antagonists on LTP and learning have been evaluated separately in different groups of subjects.
    2. The fact that blockade of LTP has been confirmed in the same animals trained in the maze task (Saucier and Cain, 1995) avoids the limitation of the correlational approach used in much of the work in this field.
  8. Bottom line
    1. Water maze task is complex and requires animals to learn the general task requirement as well as the specific location of the hidden platform
    2. Non-spatial pretraining can separate the two kinds of learning
    3. Rats first made familiar with the general task requirements and subsequently trained after receiving NMDAR antagonists could learn the spatial location as quickly as controls (report from Cain's group, 1995) or showed (to some extent) improved performance (report from Morris's group, 1995)
    4. Robust spatial learning is possible without NMDAR-dependent LTP
  9. Limitation of the NMDAR approach
    1. Other pathways (incl. mossy-fiber pathway, the lateral perforant path to CA3 and dentate) in hippocampus display LTP that are NMDAR independent
    2. Alteration of any one of the LTP systems within the hippocampus may not be sufficient to produce a total or even a profound deficit in spatial learning

1.2. Knockout mutants

  1. The targeting of specific genes whose products are required for LTP has been used to evaluate the role of LTP in learning.
  2. Early studies by Tonegawa group (1992) and Kandel group (1992)
    1. Disrupted genes for CaMKII and kyrosine kinase
    2. found impairments in both hippocampal CA1 LTP and water maze acquisition.
  3. Sakimura et al (1995),
    1. targeted disruption of a mouse NMDAR subunit gene
    2. Found reduction of CA1 LTP and deficiency in spatial learning
  4. Although these findings suggested that CA1 LTP was required for spatial learning, there are limitation in these studies
    1. The gene disruptions were performed at embryonic stem cell stage.
    2. Thus, could alter both developmental processes and the expression of other genes.
    3. Animals could have anatomical physiological, and behavioural abnormalities that might play a role in the acquisition of specific tasks
  5. Tonegawa and Kandel groups (Cell, 1996) developed a new mutant with effects that are regionally and temporally restricted in the brain, see Fig. 4-24
    1. Lack NMDARs only on CA1 pyramidal cells and only beginning during the 3rd postnatal week, which avoids most of the potential developmental defects.
    2. Exhibit no LTP, impairment in the water maze task, and place cell deficiencies


2.1. Does Learning Produce LTP-like Changes?

  1. Sharp, McMaughton and Barnes (1989) demonstrated that exploration behaviour produced increases in synaptic responses -- field EPSP (at the site of perforant-path dentate gyrus)
    1. The increases persisted for a short periods of time (20-40 mins) after exploration
  2. Moser, Mathiesen, Andersen (1993)
    1. The increase in EPSP during exploration do not reflect learning-specific changes, but result from a concomitant rise in brain temperature that is caused by the associated muscular effort.
    2. Enhanced dentate field excitary potentials followed passive and active heating and were linearly related to the brain temperature.
  3. Rearing animals in complex environments produces
    1. In vitro hippocampal slices (at the site of perforant-path dentate gyrus), field EPSP was larger
    2. The effect was transient
  4. LTP might be involved in conditioning

2.2. Does Induction of LTP Influence Learning?

  1. LTP induced prior to learning might impair learning by saturating LTP processes that normally participate in the learning
  2. LTP induced after learning might obscure prior learning by occluding any distributed pattern of synaptic changes that were formed as a results of learning



Rationale For The Claim That LTP Is a Substrate of Memory (LTP looks like L&M)

  1. Experience-dependent
  2. Long lasting
  3. Input specific
  4. Associative
  5. Cooperative


  1. Molecular Approaches
    1. NMDAR-dependent LTP - spatial learning in the water maze?
      1. Water maze overall performance
      2. Non-spatial pretraining (general task requirement) and
      3. Spatial training (location of the platform)
    2. Gene knockout mutants (NMDAR and kinases)
      1. Gene disruption at embryotic stage
      2. mutants with effects that are regionally and temporally restricted
  2. Electrophysiological Approaches
    1. Does learning produce LTP-like changes?
      1. exploration behaviour
      2. enriched environments
    2. Does induction of LTP influence learning?
      1. prior to learning (to saturate normal LTP )
      2. Immediately after learning (to obscure prior learning)