Dna Microarrays And Animal Models Of Learning And Memory

Sebastiano Cavallaro

Institute of Neurological Sciences Italian National Research Council, 95123 Catania, Italy

I. Introduction

II. DNA Microarray Technology

A. Basic Principles

B. Microarray Data Analysis

III. Use of DNA Microarrays for Studying Learning and Memory

A. Physiology of Learning and Memory

B. Pathology of Learning and Memory

IV. Conclusion References

I. Introduction

Identifying the mechanisms responsible for learning and memory consolidation remains a critical goal of behavioral neuroscience. Many experiments over the past few decades have demonstrated that inhibitors of transcription or translation interfere with long-term memory (LTM) formation, indicating the requirement of de novo gene expression (Davis and Squire, 1984; Stork and Welzl, 1999). Proteins newly synthesized during memory consolidation may contribute to restructuring processes at the synapse and thereby alter the efficiency of synaptic transmission beyond the duration of short-term memory. Revealing the dependence of LTM on protein synthesis, however, provides no information about the identity and specificity of the required proteins. Because the quantity of a particular protein is often reflected by the abundance of its messenger RNA (mRNA), a variety of methods have been used to describe a limited number of differentially expressed mRNAs during LTM. Increased or, less often, decreased expression of genes has been demonstrated during specific time windows following learning (Stork and Welzl, 1999).

In the past we have used RNA fingerprinting to identify genes that were upregulated in the hippocampus of water maze—trained rats (Cavallaro et al., 1997). Spatial learning—induced changes in expression of some of these genes occur at selective times and in specific hippocampal subfields (Cavallaro et al., 1997; Zhao et al., 2000), indicating distinct contributions to learning and


Copyright 2004, Elsevier Inc.

All rights reserved. 0074-7742/04 $35.00

memory. Increased expression of one of these genes, the ryanodine receptor type-2, could result in increased mobilization of [Ca2+] that may participate in the synaptic changes underlying associative memory storage (Alkon et al., 1998). In these past studies, however, we screened only a small fraction of the genes that may have been differentially expressed during LTM. Thus, the questions remain how many genes are involved in memory and how do they interact functionally to effect memory storage. In addition, each of the identified genes may not act in a linear sequence but in complex networks. Successive screenings at different times, therefore, were needed to uncover the networks of genes involved in distinct steps of memory storage.

Knowledge of the structure and organization of the human genome and high-throughput gene expression technologies are now opening the door to a new dynamic and functional dimension to the exploration of learning and memory. In this chapter, we highlight the use of one of these technologies, DNA microarray, and illustrate how this can be applied to dissect and analyze the pathophysiology of learning and memory in the mammalian brain. For a more general description of microarray technology, see other reviews (Heller, 2002; Hess et al., 2001; Noordewier and Warren, 2001).

II. DNA Microarray Technology

A. Basic Principles

A DNA microarray is a grid of DNA spots, called probes, each containing a unique DNA sequence (Fig. 1). Spots contain either DNA oligomers or a longer DNA sequence designed to be complementary to a particular mRNA of interest. When a microarray is hybridized to fluorescence-tagged complementary DNAs (cDNAs) or RNAs (cRNAs) derived from mRNA or total RNA, each spot is a target for the mRNA encoded by a gene. A laser can then excite the bound cDNAs or cRNAs, and fluorescence intensities from each spot on the slides are collected by a scanner. The intensity of the fluorescence at each array element is proportional to the expression level of that gene in the sample. The choice of having oligomers or longer cDNA sequences yields two different microarray technologies: oligonucleotide and cDNA microarrays, respectively. With cDNA microarrays, two fluor-labeled cDNA samples can be combined and simultaneously hybridized to the same microarray, where they competitively react with thousands of arrayed cDNA molecules. Oligonucleotide microarrays, instead, require that each sample be hybridized onto separate arrays. The thing that makes microarrays the most promising technology for genome-wide expression analysis is the number of DNA probes that it is possible to place on a microarray.

Already there are microarrays with probes for every gene in yeast, and others with more than 30,000 human genes. This allows researchers to observe the response of whole genomes to various stimuli instead of one gene at a time.

B. Mlcroarray Data Analysis

Microarray analysis results in large amounts of data that are difficult to interpret without computational methods. The simplest analysis involves two samples, representing a test condition and a control condition, and yields a list of paired expression values, one pair for each gene. As illustrated in Fig. 1, these pairs can be represented graphically by a scatter plot, with the values of sample one plotted on the x-axis and the values of sample two plotted on the y-axis. The resulting correlation plot provides a visual image of the relationship between the two expression profiles. In this plot, genes with similar expression levels in the two samples should have points on the identity line (y = x) and genes that are expressed differentially lie at some distance from this line. However, the problem is that microarrays do not measure expression levels directly, but intensity levels, as represented by the amount of phosphorescent dye that was recorded by a scanner. Many other factors, such as the overall mRNA concentration of the two samples, the saturation effects in the hybridization, or the quenching effect of the phosphorescent dyes, can affect these intensity values. To correct these differences in intensity levels, the raw data can be ''normalized,'' for example, by using a normalization constant derived from ''housekeeping'' or ''spiked'' control genes. Once normalized, a series of restrictions (or filters) can be applied to the data obtained. These restrictions include factors such as quality control, expression-level constraints, sample-to-sample fold comparison, and statistical group comparisons. The simplest way to identify interesting genes in DNA microarray experiments is to search for those that are consistently either upregulated or downregulated. To this end, fold-difference thresholds and/or statistical analysis of gene expression levels can be applied. Relative differences in expression levels (fold changes) are typically employed in group comparisons of gene expression and have much intuitive appeal for biologists. The choice of thresholds, however, is somewhat arbitrary and inherently subject to high error rates because information on sample variance is not exploited. If array experiments are replicated to an extent that permits direct estimates of the variance of each transcript, parametric or non-parametric statistics can be applied. In these cases, however, many false-positive results are expected by chance when one relies on the nominal p value. For instance, when testing 10,000 transcripts we would expect to misidentify about 500 genes as significant (p < .05), even when there is no real difference in gene expression. Multiple testing corrections, therefore, are needed to adjust the individual p value to account for this effect.

Fig. 1. Schematic representation of DNA microarray methodology. Total or messenger RNA is extracted, reverse transcribed, labeled, and hybridized to oligonucleotide microarrays. In complementary DNA microarray platforms, two RNA samples can be reverse transcribed, labeled with different fluorochromes (e.g., Cy3 and Cy5), and simultaneously cohybridized to arrays. At the end of the hybridization, the image produced by the dye is collected by a laser scanner. Intensity values from each

Fig. 1. Schematic representation of DNA microarray methodology. Total or messenger RNA is extracted, reverse transcribed, labeled, and hybridized to oligonucleotide microarrays. In complementary DNA microarray platforms, two RNA samples can be reverse transcribed, labeled with different fluorochromes (e.g., Cy3 and Cy5), and simultaneously cohybridized to arrays. At the end of the hybridization, the image produced by the dye is collected by a laser scanner. Intensity values from each

More complex computational methods are needed to monitor several gene expression profiles, such as those arising from time course studies, and various clustering techniques have been applied to the identification of patterns in gene expression data. Cluster analysis is a commonly used method to investigate and interpret gene expression data sets. By grouping together genes that have similar expression profiles, cluster analysis can be used for extraction of regulatory motifs, inference offunctional annotation, and classification ofcell types or tissue samples.

1. Cluster Analysis

The term clustering stands for a method that makes it possible to partition a set of objects (genes) into subgroups with similar features called clusters. These partitions have to satisfy the following features: homogeneity in the cluster (the objects that belong to the same cluster have to be as similar as possible) and heterogeneity among clusters (the objects that belong to different clusters have to be as different as possible).

Briefly, a clustering method generally consists of two distinct components: a distance measure (or similarity coefficient) that indicates how similar two gene expression patterns are (or more generally, two clusters) and a clustering algorithm, which uses some heuristics to identify clusters of similar gene expression patterns, based on the distance measure.

a. Measure of Distance or Similarity Coefficient. Many of the advanced analysis techniques are based on measures of gene similarity. Similarity or ''nearness'' between genes is usually based on the correlation between the expression profiles of the genes. For expression data, we can solve the problem of ''similarity'' mathematically by defining an ''expression vector'' for each gene that represents its location in ''expression space.'' In this way, expression data can be represented in n-dimensional expression space, where n is the number of experiments and where each gene expression vector is represented as a single point in that data space.

b. Clustering Algorithms. After providing a means of measuring distance between genes, clustering algorithms sort the data and group genes together on the basis of their separation in expression space. Various clustering techniques have been applied to the identification of patterns in gene expression data. Most cluster analysis techniques are hierarchical, the resultant classification has an increasing number of nested classes, and the result resembles a phylogenetic classification. Nonhierarchical clustering techniques also exist, such as k-means clustering, which simply partitions objects into different clusters without trying to specify spot are calculated and then analyzed by specific software. Data can be represented graphically by a scatter plot, with the values of sample one plotted on the x-axis and the values of sample two plotted on the y-axis. Data obtained under different conditions (e.g., different time points) can be analyzed with different algorithms such as hierarchical or k-mean clustering. (See Color Insert.)

the relationship between individual elements. Clustering techniques can further be classified as divisive or agglomerative. A divisive method begins with all elements in one cluster that is gradually broken down into smaller and smaller clusters. Agglomerative techniques start with (usually) single-member clusters and gradually fuse them together. Finally, clustering can be either supervised or unsupervised. Supervised methods use existing biological information about specific genes that are functionally related to guide the clustering algorithm. However, most methods are unsupervised.

Although cluster analysis techniques are extremely powerful, great care must be taken in applying this family of techniques. Even though the methods used are objective in the sense that the algorithms are well defined and reproducible, they are still subjective in the sense that selecting different algorithms, different normalizations, or different distance metrics will place different objects into different clusters. Furthermore, clustering unrelated data would still produce clusters, although they might not be biologically meaningful.

c. Hierarchical Clustering. Hierarchical clustering is advantageous because it is simple and the result can be easily visualized. It has become one of the most widely used techniques for the analysis of gene expression data. Hierarchical clustering is an agglomerative approach in which single expression profiles are joined to form groups, which are further joined until the process has been carried to completion, forming a single hierarchical tree. Relationships among objects (genes) are represented by a tree, called dendrogram, whose branch lengths reflect the degree of similarity between the objects. An example is reproduced in Fig. 1.

There are several variations on hierarchical clustering that differ in the rules governing the way distances are measured between clusters as they are constructed. Each of these will produce slightly different results, as will any of the algorithms if the metric distance is changed. One potential problem with many hierarchical clustering methods is that as clusters grow, the expression vector that represents the cluster might no longer represent any of the genes in the cluster. Consequently, as clustering progresses, the actual expression patterns of the genes themselves become less relevant. Furthermore, if a bad assignment is made early in the process, it cannot be corrected. An alternative, which can avoid these artifacts, is to use a divisive clustering approach, such as k-means or self-organizing maps, to partition data (either genes or experiments) into groups that have similar expression patterns.

d. k-Means Clustering. Having a priori knowledge about the number of clusters that should be represented in the data, k-means clustering is a good alternative to hierarchical methods. In k-means clustering, objects are partitioned into a fixed number (k) of clusters, so the clusters are internally similar but externally different; no dendrograms are produced. An example is reproduced in Fig. 1.

Some implementations of k-means clustering allow not only the number of clusters, but also seed cases (or genes) for each cluster, to be specified. This has the potential to allow, for example, use of previous knowledge of the system to help define the cluster output. For example, an attempt to classify patients with two morphologically similar but clinically distinct diseases using microarray expression patterns can be imagined. By using k-means clustering on experiments with k = 2, the data will be partitioned into two groups. The challenge then faced is to determine whether there are really only two distinct groups represented in the data or not. The main disadvantage of the k-means algorithm is that the number of clusters, k, must be supplied as a parameter. A simple validity measure based on the intracluster and intercluster distance measures can be used to determine automatically the number of clusters.

e. Semantic Clustering. Cluster analysis is a methodology to identify groups of genes that share expression characteristics and behaviors. It has been frequently exploited in the analysis of genome-wide expression data as the experimental observation that a set of genes that is coexpressed implies that the genes share a biological function and are under common regulatory control.

Frequently, the clustering is used to group genes considering only similar expression profiles, but it does not consider other well-known features of the gene properties. Actually, genes with a different profile expression could have similar functions as well and the classic clustering methodologies do not put it in evidence.

To extract knowledge from gene expression information, cluster analysis can be organized in two approaches: numerical and semantic clustering.

The numerical clustering method is applied to the levels of gene expression. It tends to group genes with a similar expression profile in the same clusters and makes sure that genes having different profiles with similar semantic features fall in different clusters. These considerations suggest that simple numerical clustering algorithms are inadequate to infer the genes and proteins role.

To discover more complex relationships among gene sequences, semantic clustering is used. It allows to group genes showing common biological characteristics. The term semantic clustering indicates methods of clustering based on semantic characteristics such as gene ontologies. Before performing the semantic clustering, the features have to be turned in to numerical values. At the beginning, the semantic clustering turns the features in numerical values to transform features that are similar functionally in near values. After that, methods of classic numerical clustering can be applied to that data set. In this way, the cluster analysis could make groups with similar semantic features but different profiles.

f. Visual Representation of Clustering. To interpret the results from any analysis of multiple experiments, it is helpful to have an intuitive visual representation. A commonly used approach relies on the creation of an expression matrix in which each row of the matrix represents the expression vector for a particular gene and each column represents a single experiment. Coloring each of the matrix elements on the basis of its expression value creates a visual representation of gene expression patterns across the collection of experiments. The most commonly used method colors genes on the basis of their relative expression level in each experiment (Fig. 1). For each element in the matrix, the relative intensity represents the relative expression, with brighter elements being more highly differentially expressed.

III. Use of DNA Microarrays for Studying Learning and Memory

This section focuses on the use of DNA microarray technology to dissect and analyze the pathophysiology of learning and memory in the mammalian brain. We start with experiments performed in different behavioral paradigms to study the physiology of learning and memory, and then we move to an animal model of cognition disorders.

A. Physiology of Learning and Memory

1. Eyelid Conditioning

To begin a comprehensive survey of the molecular mechanisms that underlie LTM, we have used cDNA microarray technology to perform genome-wide expression analysis after classical conditioning of the rabbit's nictitating membrane response (NMR), a uniquely well-controlled associative learning paradigm (Fig. 1) (Cavallaro et al., 2001). Classical conditioning of the rabbit NMR involves the presentation of an innocuous stimulus such as a tone followed by a noxious stimulus such as air puff to or electrical stimulation around the eye (Gormezano et al., 1962). Extensive lesion and recording data have implicated the cortex of the cerebellum and in particular, lobule HVI, in classical conditioning of the rabbit NMR (Berthier and Moore, 1986; Gould and Steinmetz, 1996; Gruart and Yeo, 1995; Schreurs et al., 1991; Yeo et al., 1985). Although the hippocampus may not be necessary for NMR conditioning, recording data do show consistent eye blink conditioning-specific hippocampal changes (Coulter et al., 1989; Sanchez-Andres and Alkon, 1991). In addition, imaging studies have implicated both structures in human eye blink conditioning (Blaxton et al., 1996; Logan and Grafton, 1995; Molchan et al., 1994; Schreurs et al., 1997).

Messenger RNA levels from cerebellar lobule HVI and hippocampus of unpaired and paired rabbits were simultaneously analyzed with high-density cDNA microarrays containing more than 8700 cDNAs (Cavallaro et al., 2001). When gene expression patterns were compared, mRNA levels of 79 (0.9%) and 17 (0.2%) genes differed more than twofold in lobule HVI and hippocampus, respectively (Fig. 2B and C). Approximately 50% (eight) of the genes differentially expressed in the hippocampus were also differentially expressed in the HVI lobule, suggesting common mechanisms of memory storage in the two areas.

A majority of differentially expressed genes were downregulated, whereas only two genes that differed by a factor greater than 2 were upregulated in lobule HVI of paired animals (Fig. 2B). Because LTM can be blocked by transcription and protein synthesis inhibitors, most previous reports have focused on the identification of proteins whose expression is upregulated (Davis and Squire, 1984). The preponderant reduction of gene expression during LTM, therefore, would not have been predicted and provides new and unexpected insights into the molecular mechanisms that underlie it. The specific role of the downregula-tion of these genes following learning remains a matter of speculation. Down-regulation of a gene may be the end-point in a dynamic gene expression process that begins with upregulation during acquisition of the learned response. Alternatively, memory storage may require a balance of upregulation of some genes and downregulation of genes that exert inhibitory constraints on memory formation (Alberini et al., 1994). These latter genes might be termed memory suppressor genes (Abel and Kandel, 1998).

Although our data represented the average gene expression from separate microarray analyses of cerebellar and hippocampal tissue obtained from a group of paired and a group of unpaired rabbits, there could be differences in gene expression between individual rabbit-derived tissues or between trained and sit control animals. To address these questions and confirm the microarray results, we selected eight (~10%) of the genes that were differently expressed and performed in situ hybridization in cerebellum and forebrain tissue sections from individual paired, unpaired, and sit rabbits. In addition to corroborating the microarray data, the in situ hybridization analysis revealed distinct spatial distribution patterns of the genes. Figure 2D shows the regional mRNA expression of EST W18585.1, insulin-like growth factor-I (IGF-I), and Bach 2. All these mRNAs were abundantly expressed in the cerebellar cortex and were reduced in the lobule HVI of paired rabbits. In addition to this lobule, downregulation of EST W18585.1 was also found in other cerebellar lobules. In paired animals, a marked downregulation of Bach 2 was also revealed in the dentate gyrus, CA1, and CA3 areas of the hippocampus.

A majority of the differential expressed genes implicated have no recognized function and are not yet named. Complete nucleotide sequence determination, conceptual translation, expression monitoring, and biochemical analysis are underway and should provide a detailed functional understanding ofthese genes.

Seventeen genes have significant similarity to known genes and can be grouped into three classes (Fig. 2E): (1) signal transduction, (2) protein modification, and (3) DNA transcription regulation. It is important to note that some of these genes have been previously related to synaptic plasticity, memory, or cognitive disorders.

Fig. 2. Microarray analyses of eye-blink—conditioned rabbits. (A) Mean percent conditioned responses in paired, unpaired, and sit control rabbits as a function of three training sessions. To relate changes in gene expression to a learning task, we used pairings of a tone and periorbital electrical

a. Signal Transduction. The first group of genes encodes proteins involved in signal transduction and includes growth factors and proteins engaged in phosphorylation. One of the identified growth factors is IGF-I, a peptide with trophic and neuromodulatory actions. In the cerebellum, IGF-I is locally synthesized by Purkinje cells but also originates from climbing fibers, which are thought to convey information to the cerebellum about the reinforcing properties of the unconditioned stimulus. IGF-I modulates the size of dendritic spines on Purkinje cells (Nieto-Bona et al., 1997) and inhibits glutamate-induced 7-aminobutyric acid (GABA) release by Purkinje cells (Castro-Alamancos and Torres-Aleman, 1993). Interestingly, IGF-I levels have been correlated with cognitive test performance in aging humans (Aleman et al., 2000) and administration of IGF-I has been shown to ameliorate age-related behavioral deficits in rats (Markowska et al., 1998). Two differentially expressed growth factors whose functions in the central nervous system (CNS) are not known were growth differentiation factor-9 (Fitzpatrick et al., 1998), a member of the transforming growth factor-^ (TGF-/3) family, and a fibrinogen/angiopoietin-related protein (Kim et al., 2000). In lobule HVI, we observed the combined downregulation of a leukocyte common antigen-related (LAR) protein-tyrosine phosphatase and liprin-beta 2, a LAR-interacting protein-like gene. The LAR gene is a transmembrane protein tyrosine phosphatase (PTPase) with sequence similarity in the extracellular region to cell adhesion molecules such as the neural cell adhesion molecule NCAM (Zhang et al., 1994). Liprins function to localize the LAR tyrosine PTPase at specific sites stimulation in a standard delay conditioning procedure, training rabbits to asymptotic levels of conditioning over 3 consecutive days. Paired rabbits (n — 12) acquired conditioned responses to the tone and reached a mean terminal level of 94.7% conditioned responses, whereas the unpaired control rabbits (n — 12) responded to the tone at mean levels of less than 1.3% across the 3 days of stimulus presentations and sit control rabbits (n — 5) had spontaneous blink rates of less than 1% (p < .001). Without further training or testing, rabbits show a level of 80% conditioned responses as long as 1 month after the 3 days of the stimulus pairings used in the present experiments. Consequently, harvesting cerebellar and hippocampal tissue 24 hours after 3 days of pairings ensured that rabbits were still at an asymptotic level of conditioning. Scatter plot of gene expression levels for paired and unpaired animals in (B) cerebellar lobule HVI and (C) hippocampus. Messenger RNA (mRNA) levels from cerebellar lobule HVI and hippocampus of unpaired and paired rabbits (n — 7 per group) were simultaneously analyzed with high-density complementary DNA (cDNA) microarrays containing more than 8700 cDNA mouse clones with a length of 500—5000 bp and with averages in the 1-kb region. The estimated percentage of homology between mouse clones and rabbit genes is 88.98 ± 3.7 (mean ± SD). The cross-species similarity and a complete list of the differentially expressed genes are available online at www.ct.isn.cnr.it/genomic-center/microarray-data/eye-blink.htm as supplementary information. (D) In situ hybridization validation of microarray results. Specific riboprobes labelled with [a-35S] for insulin-like growth factor-I, Bach-2, and EST W18585.1 were hybridized with brain sections of sit paired and unpaired rabbits. Labeled mRNA signals were revealed with autoradiography. The color spectrum on the right side of each panel represents the pixel value of gray levels. (E) DiVerentially expressed genes with a known function are ordered into functional groups. (See Color Insert.)

on the plasma membrane, possibly regulating their interaction with the extracellular environment and their association with substrates (Serra-Pages et al., 1998). Although the extracellular ligands and physiological substrates of LAR-PTPase are not known, it may be part of specific signal transduction cascades that have effects on neuronal plasticity by functioning as signal transducers of cell contact phenomena. The final identified gene that may play a role in signal transduction is phocein, a protein that binds striatin, a Ca2+/calmodulin-binding protein mostly found in dendritic spines where it is essential for the maintenance and growth of dendrites (Bartoli etal., 1999).

b. Protein Modification. The group of proteins involved in protein degradation includes a protein similar to CD156, a transmembrane glycoprotein with metalloprotease activity (Kataoka et al., 1997); the F-box protein FBX8, a specificity-conferring component of the ubiquitin protein ligase SCFs complex, which functions in phosphorylation-dependent ubiquitination of a wide array of regulatory molecules (Winston et al., 1999); and hippostasin, a brain-related serine protease of unknown function. Although the substrates of these protein-degrading enzymes are unknown, their differential expression may play a critical role in synaptic plasticity and axonal remodeling.

c. Transcription Regulation. Among the group of differentially expressed genes involved in transcription regulation, 7SK is a small nuclear RNA involved in the control on transcription (Krause, 1996), TR2 is an orphan receptor belonging to the family of steroid/thyroid hormone receptors (Young etal., 1998), whereas the rest have functions related to transcription factors. One of these is similar to the CCAAT enhancer-binding protein (C/EBP) family of transcription factors, which have been implicated in LTM consolidation after inhibitory avoidance learning (Taubenfeld et al., 2001) and longterm facilitation, a synaptic mechanism that in Aplysia is thought to contribute to LTM (Alberini et al., 1994). Interestingly, selectively enhanced contextual fear conditioning (24 hours after training) has been shown in mice lacking the transcriptional regulator C/EBP-delta, implicating some isoforms of this family of proteins in specific types of learning and memory as memory suppressor genes (Sterneck et al., 1998). WBSCR11 is a putative transcription factor gene that is commonly deleted in Williams-Beuren syndrome and may contribute to the spectrum of developmental symptoms that includes mental retardation and profound impairment of visuospatial cognition (Osborne et al., 1999). The bifunctional protein dimerization cofactor of transcription factor HNF1/pterin-4-alpha-carbinolamine dehydratase (DCoH/PCD) is both a dimerization cofactor of transcription factor HNF1 and a cytoplasmatic enzyme PCD involved in the regeneration of tetrahydrobiopterin, the cofactor for aromatic amino acid hydroxylases (Strandmann et al., 1998). Bach 2 is a transcription factor almost exclusively expressed in neurons that forms hetero-dimers with MafK and may play important roles in coordinating transcription activation and repression (Oyake et al., 1996). Pirin is a putative nuclear factor

I—interacting protein (Wendler et al., 1997). DRG11 is a paired homeodomain protein specifically expressed in sensory neurons and a subset of their CNS targets (Saito et al., 1995).

The data reported previously were the first reported in the literature to demonstrate the feasibility and utility of a cDNA microarray system as a means of dissecting the molecular mechanisms of associative memory. Further studies, however, were required at different times and behavioral conditions to better understand the role of the implicated genes. To perform such studies, we changed animal species and moved to rat, which is better suitable for genomic studies than rabbit in terms of sequenced genes and available microarrays. In the following two sections, we review studies obtained in rats following water-maze and passive-avoidance training using the same microarray platform.

2. Water-Maze Learning

We measured hippocampal gene expression profiles in naive, swimming control and water-maze—trained animals using microarrays containing more than 1200 genes relevant to neurobiology (Fig. 3) (Cavallaro et al., 2002). When gene expression profiles in naive and swimming control animals 1, 6, and 24 hours after swimming sessions were compared, 345 genes (27.3%) were found differentially expressed more than twofold in at least two of the four conditions (Fig. 3C). These genes, operationally defined as ''physical activity—related genes'' (PARGs), indicate that physical activity and mild stress associated with behavioral training has a significant impact on hippocampal gene expression.

When gene expression levels in swimming control animals were compared to water-maze—trained animals 1, 6, or 24 hours after training, 140 genes (11%) were found differentially expressed and operationally defined as ''memory-related genes'' (MRGs) (Fig. 3C). Most of these MRGs (110 out of 140) were also PARGs, that is, influenced by physical activity. Among MRGs, 55 genes were upregulated in the hippocampus of water-maze—trained animals (Fig. 1D, G, I, and M), whereas 91 genes were downregulated (Fig. 1E, F, H, and L).

Most of the MRGs, those differentially expressed between the swimming and spatial learning animal groups, were also affected during swimming alone but with entirely different temporal patterns of expression (Fig. 3F—M). Although learning and physical activity involves common groups of genes, the behavior of learning and memory can be distinguished from unique patterns of gene expression across time.

Genes implicated by gene expression profiling participate in various stages of learning and memory, and further studies are required to fully characterize their exact role. Their encoded proteins, however, may represent potential drugs or molecular targets whose activity and modulation may improve successive stages of memory (e.g., learning, consolidation, and long-term retention), under normal conditions and in disorders that affect cognitive functioning, such as

Alzheimer's disease. A promising example is represented by fibroblast growth factor-18 (FGF-18), one peptide whose sustained increase during memory retention was implicated by microarray analysis (Fig. 1G, I, and M). To explore the effect of FGF-18 in spatial learning, we tested the effects of a single exogenous dose of FGF-18. As shown in Table I, animals treated intracerebroventricularly with 0.94 pmol of FGF-18 displayed significantly improved spatial learning behavior (p < .05) compared with vehicle-injected control animals. FGF-18 treatment induced a 49% reduction in the escape latency but no significant changes in motor activity.

All of the ''LRGs'' identified have a recognized function and can be classified into six major groups based on their translated product (a complete list of the differentially expressed genes and their function is available online at www.ct.isn.cnr.it/genomic-center/microarray-data/water-maze.htm as supplementary information): (1) cell signaling, (2) synaptic proteins, (3) cell—cell interaction and cytoskeletal proteins, (4) apoptosis, (5) enzymes, and (6) transcription or translation regulation. Some of these genes have been previously related to synaptic plasticity, memory, or cognitive disorders, whereas others provide a significant number of unique and novel entry points. The exact role and functional relationships of the genes and proteins implicated, however, are

Fig. S. Continued.
0 0

Post a comment