to Memory -
The proposal concentrates on a central biomedical problem: the nature of nerve cell changes underlying learning and memory. The objectives are to elucidate the molecular changes occurring in the coupling between nerve cells taking part in a learning process. The objectives will be achieved by assessing the behavioural correlates of a cellular model of learning - long-term potentiation (LTP) - in the belief that this phenomenon represents a biological universal used by higher and lower animals. In spite of extensive effort, a clear relation between LTP and behaviourally induced learning is still lacking, largely due to the absence of convincing learning tests. In addition to behavioural assays, the project will employ state-of-the-art electrical and optical imaging techniques to determine the biophysical properties of the synapses involved. The molecular alterations of these nerve cells lie at the foundation for new ways to diagnose, prevent, and possibly treat learning and memory defects.
Attempts to establish the relationship between brain structure, activity, and function typically follow either of two generic approaches. The first is correlative: for example, this could seek to show how a particular function is correlated with patterns of neuronal activity in the brain. The limitations of these kinds of studies lie in their inability to demonstrate causal relations: to do this, manipulative studies are needed. These may show whether or not a structure or activity is necessary for a particular function. A classical manipulative approach is to make a lesion that destroys a particular target area in the brain. Lesion experiments have clearly shown selective memory impairments in rats, monkeys, or humans with hippocampal damage (Squire, 1991). Selective lesions confined to parts of the hippocampus also produce clear impairments, particularly in spatial memory (O'Keefe and Nadel, 1978). The key development in manipulative studies has been to increase the selectivity of the lesion, to damage the target area as completely as possible while causing the minimum damage to other, nearby areas. Nonetheless, however good the lesion technique, it is only one of several possible methods. More compelling evidence are likely to be derived from approaches that combine different methodologies while addressing the same issue.
The introduction of genetic engineering techniques to manipulate the genes that code for particular proteins has brought a new level of accuracy to manipulative approaches. By manipulating the right targets, experimenters can analyse the neural substrates of behaviour and cognition more precisely than ever before. The present project will use genetic engineering techniques to evaluate the neural substrate of learning and memory. In doing so the project will concentrate on the role of a specific neurotransmitter, namely the excitatory transmitter glutamate. At the majority of cortical synapses the glutamate binds to two ionotropic receptors with the acronyms AMPA and NMDA. While AMPA receptors mediate the standard impulse traffic between neuronees, the NMDA receptors have special roles during development and activity-dependent synaptic changes, effected through their ability to let calcium ions into specific postsynaptic sites. Both receptors are large, multi-component protein molecules which can be assembled in various ways to give variants with different properties. The AMPA receptor is a tetramer, a combination of four monomer molecules. Four different monomer types or GluR-A-D subunits, each coded for by a separate gene, are available for the assembly. A basic form of increased synaptic efficacy is due to an enhanced current through AMPA receptors.
Work carried out in Per Andersen's laboratory almost thirty years ago demonstrated that hippocampal neurones are sensitive to the history of previous activity (Bliss and Lømo, 1973). A brief high-frequency train of stimuli (a tetanus) increased the amplitude of the excitatory postsynaptic potential (EPSP) in target neurones. This facilitation is called long-term potentiation (LTP). LTP has the essential properties to explain learning - long duration, physiological induction, and associative properties. Such an electrophysiological phenomenon had long been sought as a potential neural substrate of memory formation. Subsequent pharmacological manipulations indicated that treatments that prevent the hippocampus from showing LTP also prevented the normal formation of spatial memories (Morris et al. 1986). Successful LTP induction requires synaptic activation of a sufficiently depolarised dendritic spine, leading to local calcium influx, which initiates LTP expression. However, the nature of the expression mechanism of LTP is not fully known. Recently, manipulation of the AMPA receptor at the the molecular level has efficiently interfered with the processes underlying changes in synaptic efficacy. Further analysis of these processes will, hopefully, enable us to understand how LTP, and in turn, how memory, is formed.
The relation between hippocampal LTP and hippocampus-dependent learning and memory functions is not unequivocal. There is a correlation between the time course of dentate LTP and ability to learn new tasks (Barnes, 1988), and the rate of forgetting a spatial task (Barnes, 1979). Further, blocking LTP in dentate perforant path synapses with 2-amino-5-phosphono-pentanoate (AP5) abolishes water maze learning (Morris et al., 1986). However, although spatial training may give LTP-like changes in dentate field potentials, the effect lasts less than 30 minutes (Moser et al., 1994). Moreover, attempts to interfere with spatial learning by saturation of LTP have given variable results (Bliss and Richter-Levin, 1993, Korol et al., 1993, Jeffery and Morris, 1993, Sutherland et al., 1993, Cain et al., 1993, Moser et al., 1998). Additionally, Bannerman et al.(1995) showed that prior spatial pretraining abolished the effect of AP5 on spatial learning tests in a new environment. Finally, the relative importance of CA1 and dentate LTP for learning is unknown. The present project is a fresh attack on these issues using the new and powerful technology of genetic engineering.
The central problem under investigation in this project is the nature and function of hippocampal synaptic plasticity. More specifically, the project intends to probe into the question of how the AMPA molecule is altered to execute critical changes at synapses that may support behavioural learning. Our main hypothesis is that a molecular alteration of glutamate AMPA receptors lies at the heart of activity-dependent synaptic changes as seen in LTP expression. The project takes as its starting point a new and fundamental finding made by a Heidelberg/Oslo collaboration: In mice lacking the gene for the GluR-A subunit (GluR-A-/-), AMPA receptors are probably composed of GluR-B-D subunits only. Although the excitatory dendritic synaptic transmission is normal in such mutants, they can not produce LTP in the CA1 field of the hippocampus (Zamanillo et al., 1999). Thus, we can conclude that the alterations point to an essential step in LTP expression. In spite of this failure of LTP, the GluR-A-/- mice show adequate learning in the basic reference memory version of the Morris water maze. It is not certain whether this failure was due to a real learning deficit, whether the test was too unspecific or whether sensory, motor, motivational, attentional or emotional factors were responsible. Moreover, recent experiments carried out by the applicant in the Oxford laboratory have found a pronounced spatial working memory (or temporary memory) deficit in the elevated T-Maze. While some types of memory may be spared in the deletion of LTP related AMPA activity, other types of memory may not be. In light of this, the data demand a re-investigation of the relation between LTP and different types of learning and memory.
objectives of the project are to use precise molecular manipulations
of LTP expression to study the relationship between LTP and memory formation.
In particular, we will re-assess the evidence that GluR-A-/- mice that
show no LTP in hippocampal area CA1 nonetheless show normal spatial
memory in the water maze (Zamanillo et al., 1999). There are three possibilities:
of GluR-A Mutants (Germany)
The first knockout type is already available. In this mouse, the GluR-A targeting vector pFC II contains 10.7 kb of the GluR-A gene (GRiA I) encoding parts of intron 10, intron 11 and exon 10. Exons are assigned according the GluR-B gene homology. A neo selection marker is inserted into a EcoNI site 700 bp downstream from exon 10. In addition, 225 nucleotides upstream of exon 10, a 28 bp PflMI fragment, is substituted by a 34 bp loxP site. Embryonic stem cells are electroporated with the targeting vector and linearised at the unique Hind III site in the polylinker. Positive clones are identified by PCR analysis with P1 sense (TCTCATTGTGATGGACCCATCC) and P2 antisense (CTGCCCATGAATAATAACTTCG) primers, and confirmed by Southern blotting of BamHI digested genomic DNA . As 5' outside probe a 162 bp Sac I fragment of the mouse GluR-A intron 10 is used. ES cell clones are transfected with an expression vector for CRE recombinase (pMC-Cre) and clones with one loxP site are identified by PCR of ES cell DNA with P1 sense and P3 antisense (CTGCCTGGGTAAAGTGACTTGG) primers. Subclones are injected into C57Bl6 blastocysts to generate germline transmitting chimaeric animals and backcrossed to C57Bl6 for one generation. The resulting F2 generation is intercrossed to produce GluR-A-/- mice at a Mendelian ratio of 25%. The identity of GluR-A genotypes is confirmed by Southern blot analysis and PCR analysis with primers P1 and P3. F2 cohorts of this mouse type have already been tested on various behavioural paradigms in Oxford during the last 9 months.
The second genetically modified mouse type is a GluR-AR/R mouse, which expresses a GluR-A subunit mutated at a single amino acid position in the pore region of the channel. The mutation changes a glutamine codon in position 602 to an arginine codon. It is expected to have a dominant effect on channel assembly and mediates impermeability for divalent ions through the AMPA receptor channel. In mice which carry this mutation the heteromeric AMPA receptor channel assemblies can be expected to be composed of GluR-B/GluR-C and GluR-A(R)/GluR-C subunits. GluR-B/GluR-A(R) assemblies are not expected since GluR-B carries an arginine in the channel segment and subunits containing arginine in position 602 co-assemble poorly. Thus, as in the GluR-A-/- mice the amount of the GluR-C subunit determines the amount of synaptically located AMPA receptors. In contrast to GluR-A-/- mice, however, in the GluR-AR/R mice the backbone of the GluR-A subunit participates in synaptically localized AMPA receptors. If the presence of GluR-A is of importance, then we may find LTP in these animals despite the fact that the number of AMPA receptors in CA1 cells is just as strongly reduced as in GluR-A-/- mice. These mice are made by gene targeting of embryonic ES cells. The targeting vector is like the targeting vector for the GluR-A gene described above. A loxP doubly flanked neo selection marker is inserted in intron 11. Together with diagnostic nucleotide exchanges in exon 10 and a change of the glutamine codon to an arginine codon, a third loxP site is positioned in intron 10. Manipulated ES cells are identified by PCR and confirmed by Southern blotting. Positive clones are amplified and transfected by Cre expression plasmid and subclones in which the neo gene has been removed by the Cre treatment are identified by PCR analysis and injected into blastocysts in order to obtain mice with the modified GluR-A(R) allele. Heterozygous GluR-A+/R mice are used to determine the expression level of the modified GluR-A(R) allele relative to the wild-type GluR-A+ allele and are used to obtain homozygous GluR-AR/R mutants.
The third genetically altered animal is a mouse mutant harbouring in its genome a bi-directional construct for a GFP-tagged GluR-A subunit and for the indicator gene lacz. Both genes are under the common control of minimal promoters with an intervening set of tet operators. The promoters are thus linked in their responsiveness to transcriptional activation by the tetracycline repressor-VP16 transactivation domain fusion (tTA), and such activation can be suppressed in presence of tetracycline and certain derivatives such as doxycycline (dox). These transgenic mice are crossed with transgenic mice expressing tTA in specific regions of the brain, such as the line described by Mayford et al., 1995, in which tTA is under the transcriptional control of the CaMKII promoter. In double-transgenic mutants, the GFP-tagged GluR-A subunit, and beta-galactosidase, will be expressed in a dox-regulated manner in forebrain principal excitatory neurones. By combining this double transgenic mutant with the GluR-A knockout genotype, we made available a mouse model in which we can regulate the amount of GluR-A in the CA3/CA1 neurones, and, thereby, influence LTP in their connecting synapses.
The primary aim of this grant proposal is to test the hypothesis that LTP-like forms of synaptic plasticity in the hippocampus may underlie hippocampus-dependent forms of learning. A clear prediction of this hypothesis is that animals which are deficient in hippocampal LTP should be impaired on behavioural tasks which require the hippocampus i.e. for our purposes that is tasks that are sensitive to hippocampal lesions. Clearly, the previously published observation that GluR-A-/- mice are capable of solving a spatial reference memory task in the water maze despite demonstrating a block of CA1 LTP (Zamanillo et al., 1999) is at odds with the hippocampal LTP/spatial learning hypothesis. However, spatial memory can be assessed in other ways than the water maze. In the past nine months, on a temporary student grant made available by Professor Per Andersen in Oslo, the applicant has already started to investigate other ways of testing the spatial working memory performance of the GluR-A-/- mice. We attempted to ascertain whether the lack of impairment found by Zamanillo et al., 1999, is specific to the aversively motivated water maze or whether it extends to the appetitively motivated Y-maze spatial reference memory task. First, we wanted to demonstrate that cytotoxic hippocampal lesions did in fact inhibit learning. The task proved highly sensitive to hippocampal damage, eliciting a statistically significant difference between the experimental and a control group (F(1,21)=69.65; p>0.0001). We then compared wild type and GluR-A-/- mice on the same task. By contrast, we found no significant difference between the groups (F<1; p>0.50). These results indicate that the initial conclusions regarding the behaviour of the LTP-deficient GluR-A-/- mice in the Morris water maze can be extended to the Y-maze. Similarly to the water maze, this task relies on the ability to learn and retain spatial information, yet it differs in terms of sensorimotor and motivational demands. The applicant will communicate these finding at the Annual Meeting of the Society for Neuroscience in San Diego, California, in November 2001.
In order to test the performance of the animals on a working memory task, we employed an appetitively motivated spatial working memory procedure (so-called rewarded alternation or spatial non-matching to position) using the elevated T-maze. Whereas the spatial reference water maze task requires the animal to retain spatial information over several days, performance on the T-maze is only dependent on temporary storage of information. In this task, mice are forced left or right down one arm of the T-shaped maze where they receive a small amount of sweetened condensed milk. Then, they are immediately given a free choice of either arm of the maze, and are rewarded for choosing the previously unvisited arm (i.e. for alternating). Whereas wild type mice performed well on this task (over 80% correct over 40 trials), the GluR-A-/- mice performed at chance (49% correct). At first glance, this finding contradicts the Zamanillo water maze study, which showed that the knockout mice were capable of normal spatial learning. Hippocampal lesions affect both these tasks. It is not immediately clear, therefore, why such a dissociation between water maze performance and T-maze performance has arisen. One possibility is that the difference may reflect differential task sensitivity. It is known that the T-maze is a more sensitive task than the water maze for picking up hippocampal dysfunction (Bannerman et al., 1999). Another possibility is that animals deficient in LTP are capable of showing some aspects of normal spatial learning, whereas they are impaired on others. The difference in performance on the two tests might then appear because the two tasks make different demands on the cognitive system.
We also plan to employ other procedures that measures working memory both spatially and non-spatially. One such task is the Differential Reinforcement of Low Rates Task, with an interval of 12 seconds (DRL-12). In this task, subjects are required to space their responses at least 12 seconds apart: responses that meet this criterion are rewarded; early responses are not. Early reports suggested that mice perform poorly on this task (Richelle, 1972). In fact they learn it well with a nose-poke response instead of lever press and liquid reward. Septal lesion-induced impairments in DRL efficiency can be seen after minimal training (Carlson et al., 1976). In rats, damage restricted to the hippocampus alone is sufficient to impair this performance; the extent of impairment increases with hippocampal lesion size, but does not depend on intra-hippocampal lesion site (e.g. Sinden et al., 1986; Bannerman et al., 1999). Performance is also impaired by administration of AP5 (Tonkiss et al., 1989). This procedure is relevant because it could answer the question whether the deficit found in the T-Maze extends to non-spatial, hippocampal-dependent working memory tasks.
The continuing development and validation of mouse equivalents of learning tasks that were developed for rats is strategically important to the project. The need for functional analyses of behaviour in mice, particularly genetically manipulated subjects, is expected to grow in the coming years. The range of potential tasks will therefore itself expand. We will continually refine our test battery in the light of experience, to exclude tasks giving poor performance in controls, or inconsistent results with low effect sizes for our manipulations of interest. A number of issues need to be resolved, and these issues form the basis of our immediate future experiments. The inclusion of the two other transgenic animal types will be of help in elucidating the issues. Especially interesting is testing of the experimenter-determined GFP-tagged knockout mice on the T-maze. Since the GluR-A-/- knockouts elicit such a marked deficit in temporary memory, it would be of great importance to be able to manipulate this freely by dox-withdrawal.
In particular, the work will include the testing for presence of LTP in the CA1 region and the dentate gyrus. The recordings will be made both intra- and extracellularly. Potentiation will be made by tetanic stimulation, theta burst stimulation or a pairing protocol. The ability of these protocols to distinguish between LTP properties in wild type and mutant mice will be determined. The efficacy of induction will be ensured by experiments with blockers of GABAA-mediated inhibition and by recording the voltage integral of the field and intracellular EPSPs during the tetanus. These techniques should allow us to achieve a major objective, namely to show how a specific molecular alteration of AMPA receptor molecules influences the degree and synaptic location of LTP expression. The applicant plans to continue and expand the analysis of the forebrain-restricted, GFP-tagged GluR-A-/- mice. With this reversible system, the Andersen laboratory has already been able to rescue GluR-A deficiency and restore cellular LTP in the CA3 à CA1 connection. The work will entail further investigation of the expression profile of the GFP-GluR-A fusion protein as well as analysis of cellular distributions at different stages in the animal's development. The experiments will, among other goals, probe the development of the LTP condition, and the possible presence of LTP in the dentate gyrus
The studies will relate hippocampal functional properties to the earlier behavioural tests of memory and cognition. Altogether, the work in Oslo is estimated to last for the duration of the second year. This will be an important step towards insuring the transferral of skills and techniques and the integration of the applicant into the Norwegian Neuroscience Community.
Optical Imaging of Synaptic Activity (Germany)
The work in Heidelberg is to a certain extent optional and will only occur if the time permits it. Because of the techniques available in this laboratory, it would certainly strengthen the project. We estimate that this part of the project will take approximately three months. Since it is likely to occur only in 2003, an invitation will be available for the NRF closer to that time.
Until recently, manipulative approaches have required the use of experimental interventions such as neurotoxic lesions, intracerebral infusion of specific drugs, or localised electrical stimulation of the brain, combined with sophisticated assessment of cognition to relate brain function to behaviour. Even the most selective of these techniques inevitably produces a variety of unwanted side effects, unrelated to the primary purpose of the intervention. Advances in genetic engineering techniques now mean that it is possible to make much more selective interventions, whose effects are confined to a single gene product. Selective knockouts of this kind thus represent a further refinement of methods for experimental intervention in brain function. The newly developed regulated types, employed in the present project, appear particularly useful. Such refinements may permit the neural substrate or mechanism under investigation to be studied while minimising the likelihood that the subject will experience potentially aversive or distressing side effects.
FOR MEDICINE AND HEALTH IN SOCIETY
The present project focuses on interactions between gene activity, synaptic plasticity, and the correlated behavioural learning and memory processes. The social impacts of an interdisciplinary research project clearly designed to increase the basic knowledge about the fundamental processes operating during learning and memory formation, may in the short term, be small. However, beyond its pure scientific aspects, the long-term impact on the public health may be considerable. Shedding light on the basic mechanisms underlying learning and memory will point to realistic models and experimental investigations useful for the study of pathological conditions present in diseases like for example dementia (prevalence 4000-8000/100.000 in European countries - 7 millions with Alzheimer's disease within Europe). The results of the project may also have beneficial effects within the educational systems, providing a framework where pedagogical theories and tools can be optimized. Professor Rawlins is already an invited contributor to OECD symposia that relate advances in neuroscience to the development of educational policy and practice.
By analysing the role played by the GluR-A-subunit of AMPA receptors for LTP appearance in hippocampal and bulbar synapses, the project will focus on a new, highly promising avenue in the search for molecular mechanisms underlying the LTP phenomenon. Hopefully, the new insight will form the basis for pharmacological approaches to facilitate LTP induction and expression. Such results might lead to new drugs for reducing learning difficulties and to slow or reduce memory failure and excitotoxic brain damage. Thus, the improved understanding of genetic, molecular and cellular mechanisms underlying learning and memory processes may, beyond its purely scientific aspect, help in the design of procedures to diagnose, prevent and treat learning and memory impairments in humans. In the light of past achievements, the research collaborators and supervisors in question should be well poised to contribute towards significant advances in this area. Finally, the results can be potentially transferable to the biotechnology, pharmaceutical, and information technology industries.