Assessment of spatial memory in mice

Abstract

Improvements in health care have greatly increased life span in the United States. The focus is now shifting from physical well-being to improvement in mental well-being or maintenance of cognitive function in old age. It is known that elderly people suffer from cognitive impairment, even without neurodegeneration, as a part of ‘normal aging’. This ‘age-associated memory impairment’ (AAMI), can have a devastating impact on the social and economic life of an individual as well as the society. Scientists have been experimenting to find methods to prevent the memory loss associated with aging. The major factor involved in these experiments is the use of animal models to assess hippocampal-based spatial memory. This review describes the different types of memory including hippocampal-based memory that is vulnerable to aging. A detailed overview of various behavioral paradigms used to assess spatial memory including the T-maze, radial maze, Morris water maze, Barnes maze and others is presented. The review also describes the molecular basis of memory in hippocampus called as ‘long-term potentiation’. The advantages and limitations of the behavioral models in assessing memory and the link to the long-term potentiation are discussed. This review should assist investigators in choosing suitable methods to assess spatial memory in mice.

Introduction

The average life expectancy in the United States has dramatically increased due to improved nutrition and health care measures (Harman 1999). The US Census Bureau projects that there will be 72 million people over the age of 65 by 2030 representing 20% of the total US population (Kinsella et al. 2001). This increase in life span reflects better health care, but at the same time it raises the concern about “health span”. Furthermore, it should be emphasized that the quality of life is key not only in terms of physical well-being, but also in terms of mental well-being.

It is well known that as individuals age physiological functions progressively decline and this decline is associated with slow adaptations in behavior, motivation, cognition, physical activity, body composition, energy expenditure and hormonal balance (Young 1997). However, one of the most significant problems people deal with, as they get older is a decrease in memory function. The very fact that elderly people are suffering from cognitive impairment as a result of “normal aging”, even in the absence of any overt neurodegenerative disease, makes it a very challenging problem.

Memory is one of the earliest cognitive functions to show decline during aging (Albert and Funkenstein 1992) and this decline has a devastating social and economic impact on individuals, families, the health care system, and society as a whole. The terminology used to define the decline in memory as we age (>50 years) is “age-associated memory impairment” (AAMI), when all other obvious causes of memory decline (for example—Alzheimer’s disease, cerebrovascular episodes, head injury, etc.) except normal aging, have been ruled out. It is surprising to note that the prevalence of AAMI is estimated to be 35–98% (Larrabee and Crook 1994). Apparently, AAMI seems to be a very prominent although not an inevitable consequence of normal aging in the population.

In this review, we will first describe different types of memory including hippocampal-based spatial memory, which is particularly vulnerable to the effects of aging. Secondly, various behavioral models used to determine hippocampal-based spatial memory in mice will be described. Lastly, an overview of the concept of long-term potentiation (LTP) as an important “experimental analog” for the molecular basis of memory will be provided.

What is learning and memory?

In one of the definitions proposed “Learning is referred to as a more or less permanent change in behavior that occurs as a result of practice,” (Kimble et al. 1961). According to Eric Kandel (Kandel et al. 2000; Yarasheski et al. 1995), learning and memory are thought to be a continuous process, and he stated that, “Learning is the process by which we acquire knowledge about the world and memory is the process by which that knowledge of the world is encoded, stored, and later retrieved.” Though used synonymously, learning is the process that modifies subsequent behavior while memory is the ability to remember past experiences. Cognition, on the other hand, is a broad term that applies to processes such as memory, association, language, attention, concept formation and problem solving (Coren et al. 1984).

Memory can be divided into short-term (working memory) and long-term memory. Short-term memory has a limited capacity and lasts only for a period of several seconds to a minute. In contrast, long­term memory can store larger quantities of information for potentially unlimited duration. Anderson (1976) divided long-term memory into declarative (explicit) or non-declarative (implicit) memory. Declarative memory answers the question “what”, and it includes knowledge of facts such as places, things and people, and the meaning of these facts. Declarative memory is further sub-divided into episodic memory, which is the personally experienced event specific to a particular context such as time and place; and semantic memory, which involves knowledge of these facts taken independent of the context in which they were learned (Miller 1956; Tulving 1972; Fig. 1). The major brain structure involved in declarative memory is the hippocampus along with other medial temporal lobe structures (Squire and Zola 1996). Formation of a new declarative memory is a sequential process that includes acquiring new knowledge (encoding), retaining the information (storage), and bringing it back (retrieval). Furthermore, memories are continually being consolidated in the neocortex (Solms and Turnbull 2002). Non-declarative or implicit memory, on the other hand, answers the question “how”. It is the acquisition of motor skills and habits and is mediated by neostriatum and cerebellum (Bechara et al. 1995; Knowlton et al. 1996; Salmon and Butters 1995). In addition, the amygdala mediates emotional memory and has been shown to be involved in memory consolidation (Cahill et al. 1995).

Fig. 1.

Classification of declarative memory (explicit memory).

Episodic memory, a type of declarative memory that depends on the ability to remember in a determined temporal and spatial context (Tulving 1972), is especially vulnerable to normal aging. Elderly people are more susceptible to forget the context related to an episodic event than the event per se (McIntyre and Craik 1987). It has been demonstrated that the elderly perform adequately in memory tasks that require less effort, as in implicit memory tasks that require them to repeat or recognize a stimulus (Craik and McDowd 1987; Light et al. 1992; Puckett and Stockburger 1988). But, their performance diminishes if they have to recollect or retrieve the information retained in their mind or carry out some actions without reminders (Dobbs and Rule 1989; Einstein et al. 1995; Schonfield and Robertson 1966).

Typically, spatial memory is conceptualized as a subtype of episodic memory because it stores information within the spatio-temporal frame (O’Keefe and Nadel 1978). Nadel and O’Keefe’s cognitive map theory suggested that spatial memory is dependent on the hippocampus. There were two lines of evidence suggesting that this was the case. The first was the most famous amnesic patient H. M. reported in late 1950s. This person had severe impairment in acquiring new memories after the medial temporal lobe was surgically removed (Scoville and Milner 1957). The other piece of evidence was the discovery of place cells. “Place cells are neurons that fire depending on animal’s place in the environment independent of any particular stimulus or ongoing behavior” (Eichenbaum 1996). Recently, the concept of cognitive mapping has been taken one step further to describe the role of the hippocampus in the construction of mental images (Bird and Burgess 2008).

During the past decades strong evidence has emerged showing that the hippocampus is critical to learning and memory as well as to age-related cognitive deficits (Corkin2002; Corkinetal. 1997; Gadian et al. 2000; Pigott and Milner 1993; Scoville and Milner 1957; Smith et al. 1999; Squire 1992; Stefanacci et al. 2000; Tulving and Markowitsch 1998; Vargha-Khadem et al. 1997). Studies of human subjects with hippocampal damage provide evidence that this brain region plays a critical role in spatial memory (Abrahams et al. 1997; Astur et al. 2002; Feigenbaum et al. 1996; Goldstein et al. 1989; Maguire et al. 1996; Rosenbaum et al. 2000; Teng and Squire 1999). Patients with bilateral medial temporal lobe damage exhibit severely impaired episodic memory, but have relatively preserved semantic memory (Bayley et al. 2008; Schmolck et al. 2002). They can perform well on a procedural skill and habit learning. Their vocabulary is normal as well as their knowledge of facts learned in the past. Functional imaging studies investigating the brain activity of people during navigation in a virtual reality town have demonstrated high activity in the hippocampus (Bohbot et al. 2004; Maguire et al. 1998). Similarly, an interesting study in licensed London taxi drivers showed that extensive navigational experience is associated with greater hippocampal gray volume as compared to controls (Maguire et al. 2006). Also, studies of non-human subjects (including rats and monkeys) with hippocampal lesions or age-associated changes in hippocampus demonstrate impaired learning on spatial memory tasks (Bannerman etal. 1999; de Bruin etal. 2001; Gallagher and Rapp 1997; Jarrard 1993; Murray et al. 1998; Pouzet et al. 2002).

All of this evidence is further supported by the fact that the hippocampus (store house of declarative memory) is particularly vulnerable to the effect of aging (Mesulam 1999; Miller and O’Callaghan 2005). The hippocampus appears to be an early target of age-related structural and physiological changes (Hasan and Glees 1973) and damage to the hippocampus results in similar cognitive impairments as experienced by elderly people (Barnes 1979, 1988; Gallagher and Nicolle 1993; Geinisman et al. 1986). Considerable evidence supports that aged humans have trouble navigating or finding their way in a large environment and remembering spatial relationships among landmarks (Bruce and Herman 1983; Caplan and Lipman 1995; Cherry and Park 1989; Evans et al. 1984; Flicker et al. 1984; Kirasic 1991; Kirasic et al. 1992; Kirasic and Bernicki 1990; Lipman and Caplan 1992; Moffat et al. 2001; Naveh-Benjamin 1987; Ohta et al. 1981; Park et al. 1990; Perlmutter et al. 1981; Sharps and Gollin 1987; Thomas 1985; Uttl and Graf 1993; Weber et al. 1978; Wilkniss et al. 1997; Zelinski and Light 1988). In some studies, humans show a 30–80% drop in performance of spatial memory tasks with advancing age (Cherry and Park 1993; Evans et al. 1984; Kirasic and Bernicki 1990; Moffat et al. 2001; Moore et al. 1984; Sharps and Gollin 1987).

There are abundant human studies that have administered various supplements to improve cognitive function (Arwert et al. 2003; Chandra 2001; Duffy et al. 2008; Fintelmann and Gruenwald 2007; Frick et al. 2002; Grodstein et al. 2007; Jorissen et al. 2001; Kotani et al. 2006; Krikorian etal. 2010; Kritz-Silverstein etal. 2003; Thorp et al. 2009) but with less definitive measures in sight. “Nootropics” or “smart drugs” are said to improve mental function such as cognition and memory. There are myriad of supplements including‒caffeine, Ginkgo biloba, Bacopa monniera, and drugs not limited to galantamine, memantine, donezepil, piracetam and rolipram (Bora et al. 2005; Brown et al. 2007; Capek and Guenther 2009; French et al. 2007; Gomer et al. 2007; Kennedy et al. 2007; Mecocci et al. 2009; Neyens et al. 1995; Rutten et al. 2008; Smith et al. 2003; Stough et al. 2008). Similarly, scientists are utilizing animals to understand the patho­physiology of brain aging to gain insight into the potential mechanism with the hope of finding ways to prevent or hopefully reverse memory loss (Barnes etal. 1996a,b; Blecharz-Klin et al. 2009; Costa et al. 2008; Dow-Edwards et al. 2008; Hebda-Bauer et al. 1999; Lei et al. 2003; Magnusson et al. 2007; Markowska et al. 1998; McDonald et al. 2005; Nishiyama et al. 1997; Wang et al. 2007; Yasui et al. 2002).

How to assess hippocampal-based spatial memory in mice?

In 1948, psychologist Edward Tolman was the first to study spatial behavior in rats. He put hungry rats at the entrance of a maze consisting of true paths and blind alleys and provided food at the end. Tolman observed that the error rate decreased with the number of trials (Tolman and Gleitman 1949). It was 3 decades later when the spatial reference memory system was proposed by Olton (Olton et al. 1979) to designate the type of memory process involved in obtaining spatial information over various trials. In contrast to spatial working memory, spatial reference memory has more capacity, lasts longer and resists interference (Olton et al. 1979).

Since the initial experiments by Tolman, evaluation of hippocampal-based spatial learning and memory has been assessed by numerous behavioral paradigms in rodents (D’Hooge and De Deyn 2001; Holcomb et al. 1998; Holmes et al. 2002; Koopmans et al. 2003; Kuc et al. 2006; Paul et al. 2009). Many of these behavioral memory paradigms are designed and established for rats. But, it is well known that mice are a valuable resource for researchers especially with the advancement of various genetic mouse models. In addition, mice are inherently different than rats in terms of physical and motivational factors when utilized in learning and memory tasks. For these reasons the behavioral tests used for rats cannot be directly used for mice (Paul et al. 2009). There is a critical need to design memory tasks specifically designed or adapted for mice. In this part of the review we will discuss various behavioral paradigms that can be used to assess hippocampal-based spatial memory in mice and include their advantages and disadvantages (Fig. 2).

Fig. 2.

Different types of mazes.

T-maze

The T-maze is a simple behavioral paradigm used to assess spatial memory. It consists of a capital T-shape design with a stem length of 35 cm and an arm length of 28 cm for mice (rats: 50 cm and 40 cm, respectively). There is a single choice point with only two alternatives. In this test the animals are allowed to freely explore each arm of the device; then the number and order of visits to the arms are recorded. The principle of alternation is based on the fact that the animals prefer to visit the less recently visited arm, thus implicating that it will need to recall which was the last arm visited. These tests can be performed in two manners: 1) free tests or 2) forced tests. In the latter, one of the arms of the device is blocked in order to favor alternation behavior (Bats et al. 2001; Deacon and Rawlins 2006; Lalonde 2002).

Alternatively a positive reinforcer may be placed in one of the arms so as to reward alternation behavior. This appetitively motivated learning task is utilized in various studies (Fanelli et al. 1983; Schlesinger et al. 1986). Moreover, by increasing the interval between tests it is possible to conduct studies designed to evaluate spatial working memory in which a decrease in alternation behavior can be observed (Dember and Fowler 1959). In our studies, we utilized an appetitively motivated task using a previously described protocol (Bizon et al. 2007; Deacon and Rawlins 2006). Briefly, since the mice need to learn to locate the food reward, they were food deprived to 85% of ad libitum body weight and maintained at this weight throughout the experiment. Mice were habituated to the maze on day 1 and arm preference was determined. The animals were trained against their preference in subsequent training trials (4 trials/day for 3 days) by baiting the other arm of the maze. Entries into the unbaited arm were scored as incorrect responses, and those into the baited arm were scored as correct responses. The time required to enter the baited arm and reach the target was recorded as latency. The T-maze was swabbed with 70% ethanol between animals to eliminate odors. In order to assess spatial memory mice are subjected to two probe trials (Days 5 and 12). During the probe trials, the maze was rotated 180°, but all cues, including the experimenter, remained stationary. Those mice entering the baited arm were designated as using a “place” strategy, which has been shown to involve the medial temporal lobe, including the hippocampus. Those mice entering the unbaited arm (i.e., making same body turn as used during training) were designated as using a “response” strategy that relies on caudate nucleus (Packard et al. 1989; Packard and McGaugh 1996). There are various protocols described that can be used in terms of number of trials, type of reward or methodology used (Barnes et al. 2004; Belzung et al. 2001; Crusio et al. 1990; Deacon 2006; Gerlai 1998; Lohninger et al. 2001; Wenk 2001). It has been shown that with aging associated loss of hippocampal function, aged animals use a response strategy more often as compared to adult animals who use place strategy more frequently (Barnes et al. 1980; Rapp et al. 1997).

In our lab, we have used a T-maze to study spatial memory in mice and found it to be useful to test Ames dwarf mice (Brown-Borg et al. 1996). In our experience the simplicity of the T-maze makes it one of the best paradigms to evaluate learning behavior in mice as compared to other complicated mazes (Sharma et al. 2010). We found that mice preferred a liquid reward (sweetened condensed milk) rather than normal chow or peanuts.

The T-maze has been used to evaluate spatial memory in mice in many studies and has been shown to be sensitive to the effects of drugs and toxins that enhance or impair spatial memory (Ferragud et al. 2010; Franowicz et al. 2002; Goulet et al. 2003; Incerti et al.; Ito and Canseliet 2010; Pierard etal. 2007; Pistell and Ingram 2010; Sanchez-Santed etal. 1997; Sanderson et al. 2009; White 1974). Recently, an aquatic version of the T-maze has been adapted and utilized (Del Arco et al. 2007; Locchi et al. 2007). One arm of the maze was painted black and the other arm white, with the correct arm leading to an escape ladder (Granholm et al. 2000). In another water escape motivated version, the mice are required to locate the hidden platform beneath the water in one of the arms thus utilizing natural tendency of mice to escape to a safer place (Belzung et al. 2001; Jaffard et al. 1991; Paylor et al. 1994). Another modification requires the animal to choose between a light and dark compartment in the T-maze (Raffalli-Sebille et al. 1990; Venault et al. 2006). Various studies have also used the T-maze to assess age-related cognitive decline (Caoetal. 2007; Heikkinenetal. 2004; von Bohlenund Halbach et al. 2006; Zhou et al. 2007).

The most important advantage of using a T-maze is that it is the simplest device to assess spatial working memory. The T-maze does not require automated video-recording systems (but does require constant observation by the experimenter) and also provides highly reproducible results. In contrast, the major disadvantage is that it has a single choice point with only two alternatives, which increases the possibilities of success (by default the probability of choosing the correct arm is 50%) or that the rodent may use a strategy other than spatial to solve the maze. It can therefore be used to assess working memory but not reference memory. Depriving the animal of food can be a potential problem. Another disadvantage with using the T-maze is that it requires constant handling of animals, which may induce stress and affect the test result. But, habituating the animal and handling by the same experimenter reduce this stress. One of the modifications performed to avoid manipulation and stress is the addition of removable doors at the entrance of each arm so that the unselected arm can be blocked (Gerlai 1998; Spowart-Manning and van der Staay 2004). The Y-maze is similar to the T-maze, but has three 35 cm long identical arms. The only difference is that the arms require a gradual turn as compared to the sharp turns in the T-maze, which might decrease the learning time in the Y-maze.

Multiple T-maze

The Multiple T-maze as the name suggests are several T-mazes put together to form one maze. This maze consists of a wooden platform of 150 cm×130 cm×15 cm and a path width of 8 cm with seven different choice points (Lohninger et al. 2001; Pangratz-Fuehrer et al. 2005). In another study, the maze arms were 3 cm wide and 55 cm above the floor, with 12 choice points (Lewejohann et al. 2004). These mazes add complexity (more choices/alternatives) to the T-maze and are being utilized as spatial learning tasks. With this maze, mice have to learn a complex route, which is stable from trial to trial and, thus, requires spatial reference memory (Ingram 1988). Similar to the T-maze, memory retention is evaluated as the ability of food deprived mice to locate the hidden food with a decrease in latency and an increase in correct decisions (Pangratz-Fuehrer et al. 2005). Mice are placed in a start box in a black cylindrical chamber (Prior et al. 1997). After 10 s have elapsed, the chamber is lifted and the first trial is started. Mice are allowed to search for the reward and the trial is completed when mice reach the goal box, or if failed, after 5 min. After each trial the entire maze is cleaned with 1% incidin or 70% alcohol solution to avoid odor cues. Mice are trained with three such trials per day for 4 days. Trials are carried out using 20-min intervals. The following parameters are recorded on a computerized tracking system—correct or wrong decision, path length, speed and latency to reach the goal box. Probe trials can be conducted on days 5 and 12 for short-term and long-term retention memory with no training during the time period between Days 5 and 12.

The Stone T-maze is a modified version of 14-unit maze and water-based T-maze for testing hippocampal-dependent memory (Del Arco et al. 2007; Locchi etal. 2007). It is modified from a version previously used in rats (Ingram 1988). In this version mice need to wade through water following a correct sequence of left and right turns to reach a dark dry goal box (Pistell and Ingram 2010).

Various studies have used the Multiple T-maze for assessing spatial memory in mice (Kunesova et al. 2008; Patil et al. 2009; Zheng et al. 2009). The advantage of using the Multiple T-maze is that it utilizes a more complex choice system with multiple-choice points and therefore can be utilized to test both working and reference memories. The disadvantage is that it requires a computerized tracking system that may not be available to all researchers. Like the T-maze, food deprivation and handling of animals can be a potential problem.

Radial-arm maze

The radial maze was developed by Olton and Samuelson (1976). It consists of 8 or 12 arms, 5 cm wide and 35 cm long for mice (rats 10 and 50 cm respectively) with a central platform. A pellet of food is placed at the end of each arm. Based on a principle similar to the T-maze, an animal who has been food deprived is placed in the center and is allowed to collect food pellets from each arm (sampling without replacement). The optimal strategy for obtaining all the food in the least amount of time is to visit each arm only once during a trial. The number of visits to empty arms (unbaited arms) is calculated as errors made (Wenk 2004). This original version tested spatial working memory. Olton also developed a variant that allowed assessing reference memory in addition to working memory. To test both, only three or four arms are baited with a food reward (Olton et al. 1979). Re-entry into a previously baited (now empty) arm is defined as a “working memory” error. Entries into never-baited arms are counted as “spatial reference memory” errors. Further studies have confirmed that the animal uses spatial cues to recognize the arms previously visited (Mazmanian and Roberts 1983; Suzuki et al. 1980). In addition, mice with induced hippocampal lesions exhibited impaired memory in the radial-arm maze (Olton 1983; Olton et al. 1979; Olton and Papas 1979). Moreover, learning in a radial-arm maze is sensitive to the effects of aging and aged rats have been shown to need more trials to reach a required performance (Barnes et al. 1980; Davis et al. 1983; Wallace et al. 1980).

An aquatic version of radial maze has been developed in which the stimulus driving behavior is linked to aversive stimuli rather than to appetite. Animals locate an escape platform hidden in some quadrant of the maze (Hyde et al. 1998). A variation of the radial-arm maze task (standard 8-arm, four baited arm task and cued version task) has also been used to study strain specific abilities (Ammassari-Teule and De Marsanich 1996; Ammassari-Teule et al. 1993).

The advantage of the radial-arm maze is that the training protocol and data interpretation of the basic version are simple and various studies having used the 8-arm radial maze for evaluation of spatial working memory and for spatial reference memory (Belotti and Galey 1996; Dudchenko2004; Hodges 1996; Schmitt etal. 2003; Zhang etal. 2005) as well as the 12-arm radial maze (Demas et al. 1998). The other advantage is that this test induces only a moderate level of stress (Hodges 1996). The radial maze has also been used to study age-related cognitive decline in various mouse models (Bernstein etal. 1985; Krazem etal. 2003; Lebrun etal. 1990; Lohmannand Riepe 2007; Marighetto et al. 2000; Nogues et al. 1996; Takahashi et al. 2009; Touyarot et al. 2002; Touzani et al. 2003). It has been suggested that mice use several types of information for task performance and as they mature they turn more often to distal cues for orientation (Chapillon et al. 1995).

The limitation of the radial maze is that an animal can use a serial strategy going from one arm to another instead of using spatial cues. There is also a possibility that olfactory cues can be used to detect those arms already visited (Wasserman and Jensen 1969). Some research groups have used the aromatic saturation strategy allowing the animals to freely explore the maze before the test without any food pellets with the aim to saturate the maze with aromatic cues (Dudchenko 2004; Roullet et al. 1993). Other strategies involve cleaning the device or/and rotating the arms while preserving the location of the food reward (Dudchenko 2004). One study failed to demonstrate learning in the radial maze (Mizumori et al. 1982) and some authors debate if the radial maze is actually measuring spatial memory (Dubreuil et al. 2003). However, the three or four arm-baited versions can be used to avoid this problem. Some studies indicate that radial devices are not sensitive enough to spot differences between animal strains or genders (Seymoure et al. 1996; van Haaren et al. 1990, 1987). But, various knockout mice have been successively trained on the radial maze (Dutar et al. 2002; Egashira et al. 2004; Mineur et al. 2002; Zlomuzica et al. 2009). Similarly, the radial maze has been used in both male and female mice to detect gender differences in learning behavior (LaBuda et al. 2002).

Morris water maze

The Morris water maze (Morris 1984) was invented by Richard G. M. Morris in 1981 as an alternative to the radial maze (Morris 1981). This device is one of the most widely used paradigms (D’Hooge and De Deyn 2001; Gallagher et al. 1993; Morris et al. 1982). It consists of a round pool filled with water made opaque using milk or white paint. The animal learns to locate the platform hidden under the water from four different starting points. Over a number of trials animals learn the location of the hidden platform based on distal cues and with time the latency to locate the platform decreases. The strength of learning is tested afterwards by a probe trial in which the hidden platform is removed and the amount of time spent in the former region of platform is measured (Morris 1981). In this maze, mice can use three different strategies (Brandeis et al. 1989) to locate the escape platform: a praxic strategy, when the animal learns the sequence of movements needed to reach the platform, a taxic strategy, when the animal uses cues or visual proximal guides to reach the platform, or a spatial strategy, when the animal reaches the target using information about the spatial location of the platform according to the spatial configuration of distal cues. The size of pool for mice can be 75–150 cm in diameter (rats 150–180 cm). The diameter may or may not affect the performance depending on the strain of animals used (Van Dam et al. 2006; van der Staay 2000).

Various parameters obtained with this maze include latency to find the platform, swim path length, swimming velocity, time spent in each quadrant, number of platform crossings, and percent of trials with failures (2 min without finding the platform). According to one study, the mean proximity to a former location of the platform is the most sensitive measure to assess in water maze performance (Maei et al. 2009). Other factors that may influence the performance of the mice on the Morris water maze include the number of trials per day, number of days of training, duration of each trial and the inter-trial interval (Zhou et al. 2009).

The Morris water maze permits the accurate and reproducible study of reference memory, spatial working memory and learning (D’Hooge and De Deyn 2001; Dudchenko 2004). The Morris water maze has been shown to be highly sensitive in the assessment of damage to the hippocampus (Bannerman et al. 1999; Morris et al. 1982; Sutherland et al. 1982). Furthermore, aged animals exhibit consistent impairment in learning the location of the escape platform (Foster et al. 1991; Gage et al. 1984; Gallagher et al. 1990). Recently, the assessment of age-related decline in spatial memory in mice was conducted using the Morris water maze to better elucidate the mechanisms involved (Francia et al. 2006; Harburger et al. 2007; Pan et al. 2008; Pawlowski et al. 2009; Polydoro et al. 2009; Tong et al. 2007; von Bohlen und Halbach et al. 2006; Zhao et al. 2009). A longitudinal study in mice indicated that previous learning in the Morris water maze could prevent some age-related spatial learning deficits (Vicens et al. 2002).

Learning has been shown to be faster on the Morris water maze compared to other mazes possibly due to aversive stimuli, with acquisition in 5 days of training (Hodges 1996). This maze does not require previous preparation (water or food deprivation) thus limiting the number of days needed to proceed with experimentation. The biggest advantage is that the use of water in this maze eliminates the possibility that animals use aromatic cues to orient themselves in the escape search, one potential confounding factor that occurs in the dry-land mazes. The Morris water maze, although simple to build and adapt, requires video-recording systems and software for complete analysis of behavioral parameters, equipment that may not be readily available to all researchers.

However, this behavioral paradigm when directly adapted for mice, is associated with problems. It is important to take into account that cognitive performance may be affected by additional stress on experimental animals due to swimming (de Quervain et al. 1998; Holscher 1999). In particular, swimming in the Morris water maze is associated with neurochemical changes (Kirby et al. 1997) that may interfere with testing cognitive function (Patil et al. 2009). Most of the researchers believe that previous habituation is needed before the actual trials are conducted to decrease the stress level (Morris 1981) while some do not (Avila-Costa et al. 2006). Furthermore, the variation in using visual ‘extramaze’ cues like salient cues, diffuse cues and absence of cues can also affect the spatial performance (Champagne et al. 2002).

Barnes maze

To avoid the stress induced by swimming, Carol Barnes developed a dry-land maze (Barnes 1979). This paradigm consists of a circular platform at a height of 140 cm, with holes along the perimeter (12, 20 or 40 depending on the diameter). It is surrounded by black curtains with visual cues on them. During testing, animals receive reinforcement to escape from the open platform surface to a small, dark, recessed chamber located under one of the holes called the “target box”. The principle is similar to that of the standard, non-cued Morris water maze but less stressful in the assessment of spatial learning (Deacon et al. 2002; Harrison et al. 2009, 2006; Pompl et al. 1999). It was initially designed to assess spatial learning and memory in rats, but was later adapted for mice (Pompl et al. 1999). This device has been considered appropriate for mice because of their propensity to find and escape through small holes (Barnes et al. 1980; Pompl et al. 1999) and the natural preference of rodents for a dark environment (Bach et al. 1995).

The method used in our lab was adapted from a recently published protocol (Sunyer et al. 2007). In the pre-training trial, the mouse is placed in the middle of the maze under a dark colored box allowing the mice to be in random orientation before each trial (similar to the different start quadrants in the Morris Water Maze). After 10 s have elapsed, the chamber is lifted, and the mouse is allowed to explore the maze for 5 min. Various types of aversive stimuli—sounds (78–108 dB) intense light, food, and air jets are often used to induce escape behavior (Ingram et al. 1994; Moscovitch et al. 2005). The following parameters are recorded (during acquisition and testing): errors, latency(s), and search strategy. Errors are defined as nose pokes and head deflections over any hole that does not have the target box. Latency is defined as the time it takes to locate the target box. Mice are trained for four trials per day for 4 days with an inter-trial interval of at least 15 min. After each trial the entire maze is cleaned with 70% alcohol. To further avoid odor cues, the maze can be rotated around its central axis. On days 5 and 12, a probe trial is conducted to evaluate short-term and long-term memory retention without any training between Days 5 and 12. Search strategies are determined by examining each trial and placing the mice into one of the three categories 1) Spatial‒moving directly to the target hole or to an adjacent hole before visiting the target hole (≤3 errors); 2) Random—hole searches separated by crossing through the center of the maze or an unorganized search; or 3) Serial‒the first visit to the target hole was preceded by a visit to adjacent holes in a serial manner, clockwise or counter clockwise in direction.

Aged animals are impaired in learning the location of the hidden escape tunnel, in retaining a memory of the location, and in learning a new location of the escape tunnel (Barnes 1979; Barnes and McNaughton 1985; Barnes et al. 1980). Results of another study revealed a similar pattern of learning using the Barnes maze in aged mice that was consistent with the idea that spatial learning is particularly vulnerable to effects of aging (Bach et al. 1999).

The main advantage of the Barnes maze is that being a dry-land maze, it does not induce stress as in the Morris water maze that involves swimming. In addition, like the Morris water maze, it allows evaluation of learning working memory and spatial reference memory (Sunyer et al. 2007). It is particularly suitable for mice since these animals exhibit a lower performance on the Morris water maze (Bach et al. 1995; Lukoyanov et al. 1999). Various studies in mice have successfully utilized the Barnes maze to assess spatial memory (Babovic et al. 2008; Bach et al. 1995; Barrett et al. 2010; Fabricius et al. 2008; Holmes et al. 2002; Inman-Wood et al. 2000; Koopmans et al. 2003; McAfoose et al. 2009; O’Leary and Brown 2009; Patil et al. 2009).

One disadvantage of the Barnes maze is that learning can be very slow or even absent in some cases. This can be explained by the lack of stressful stimuli, thereby producing more exploratory behavior than escape responses in animals that are not sufficiently motivated to escape (Sunyer et al. 2007). This can result in an increase in the number of errors due to further exploration of the maze although the mouse learned the association between the spatial cues and the escape location previously. Harrison et al. (2006) proposed a solution to this extra-exploratory behavior by calculating latency, path length and number of errors to the first encounter of the escape hole, called primary latency, primary path length and primary errors, respectively. Another disadvantage of the Barnes maze is that it can also stimulate non-spatial strategies like a serial strategy that can then affect performance. If the maze is not cleaned appropriately, the animals can use “aromatic cues” to solve the maze. One of the studies using multiple mazes to study spatial memory demonstrated that mice committed considerably more errors than expected by random and this could be due exploratory behavior of mice and may not indicate impairment in spatial memory (Lewejohann et al. 2004).

Cheeseboard maze

An alternative maze developed to avoid problems associated with Morris water maze in mice is called as the Cheeseboard maze (Llano Lopez et al. 2010). This maze involves a task equivalent to the cognitive demands of the Morris water maze but is more suitable to murine species as it takes into account their ecological behavior (Kesner et al. 1989). The Cheeseboard maze, like the Morris water maze, uses a circular arena, in which a single target (reward or escape) can be positioned in various predefined positions, allowing assessment of reference and working memory. It consists of a wooden circular board (110 cm diameter) with 32 holes in a regular radial fashion similar to 8-arm radial maze. Some studies have adopted the original regular square design of Kesner (Pillay et al. 2008) while others have adopted a random pattern (Yoshida et al. 2001). Animals in this task also need to be food deprived and the reward used is freshly prepared diluted condensed milk. After the initial habituation, each trial requires the animal to locate the well containing the reward within 2 min. Two such trials are given per day for 5 days with an inter-trial interval of 1 min. During the probe trial the animals are allowed to search on the cheeseboard in the absence of reward, thus allowing the assessment of spatial strategy.

In this study, the investigators also showed that Cheeseboard maze had a higher effect size for the main effect of days in the acquisition of reference memory as compared to Morris water maze experiments. They showed that animals were able to reach a stable level of performance in 4 days of training that was highly comparable (correlation significant) to the same animals performance in the subsequent working memory experiments (Llano Lopez et al. 2010). Additional time was required in the Cheeseboard maze experiments for the introduction and acclimatization to the food deprivation schedule but overall the total time was far less than that reported for a dry-land maze like the radial-arm maze (Schmitt et al. 2003). In addition, the use of a dry environment effectively prevents floating thus overcoming one major problem of the water mazes. Further, unlike the Morris water maze or the Barnes maze, the Cheeseboard maze task does not rely on negative reinforcement. Although the Barnes maze has been reported to be less stressful than the Morris water maze in terms of hypothalamo-pituitary-adrenal responses (Harrison et al. 2009) the Cheeseboard maze should be even less stressful because it does not require the use of either high light intensity, loud noise or wind to reinforce the escape response as in the Barnes maze (Nithianantharajah and Murphy 2009). The Cheeseboard maze is typically performed under conditions of dim lighting, without aversive stimuli, and with extensive pre-training habituation on the apparatus and in the testing room.

One of the limitations of the Cheeseboard maze is that the spatial distribution of the food wells where a palatable reward can be hidden has not been clearly defined. It is obvious that the number of potential reward sites as well as their relative spatial distribution could affect task difficulty. Food deprivation and use of the same starting position can be an issue with the Cheeseboard maze. However, when the animals are first placed and covered by a box, their head may not be in a specific direction when the box is lifted (random head direction). This can thus be comparable to different starting points used in Morris water maze.

Hole board discrimination test

The hole board apparatus consists of an open-field chamber with a 16-hole floor insert. This apparatus was originally used to look at exploratory behaviors (Belzung and Le Pape 1994; File 2001; Rogers et al. 1999). Hole board-learning paradigms were then used for rats by some researchers (Oades and Isaacson 1978; van der Staay et al. 1990a,b). It was later adapted to assess learning behavior in mice by retrieving a food pellet located in one hole out of four in a hole board task setup (Brosnan-Watters and Wozniak 1997; Brosnan-Watters et al. 1996). However, this task is not complex enough to assess reference and working memory. Recently, Kuc et al. (2006) have adapted this hole board-learning task for mice that allows for simultaneous assessment of spatial working and reference-memory performance. Across trials, animals have to remember the same four holes out of the 16 that are always baited.

Like other mazes, animals are food deprived and then habituated by allowing the mice to collect all the food pellets that were placed in every hole (one pellet per hole). For the learning trials the same four holes are baited with a food pellet during each trial. The software automatically ends a trial when all four pellets have been collected or after 3 min have elapsed. The start position of a mouse is randomly changed across trials. Mice readily acquire this task within 4 days when submitted to six trials per day or within 8 days when submitted to only four trials per day. The hole board floor insert is cleaned with alcohol to homogenize potential olfactory traces. In some experiments, reversal training is conducted under the same testing conditions as during acquisition training; however, the animals have to learn a different baited-hole configuration. The average number of errors per trial an animal made each day is used as a measure of cognitive performance. Errors consist of entering a hole that was never baited (reference-memory error), re-entering a hole (working-memory error) or missing a baited hole (error of omission). When appropriate, the task completion time and the number of head entries can also be analyzed.

In a similar test, another study used a four-hole board test to assess spatial delayed discrimination. This apparatus consists of a 4-hole board (30 cm×40 cm×30 cm) with spatial cues on it. Food deprived animals are placed on the center of the board and covered with a black cylinder for 30 s for initial random orientation. After the cylinder is removed, the mouse is allowed to explore and collect the reward. For measurement of retention, exploration frequency and time are recorded for each hole for 3 min with no food pellets in the apparatus (Galey and Jaffard 1992; Venault et al. 2006).

An advantage of this learning paradigm over other paradigms (i.e. the radial-arm maze and water maze) is that it is fully automated (infrared beams are used to track activity on the floor and nose pokes into the holes). Another distinct feature of hole board discrimination is that four different hole board chambers can be simultaneously used and several animals can be tested at the same time making it useful for high throughput testing. Moreover, this learning task appears to be sensitive to drug and experimental manipulations (van der Staay and Bouger 2005). One of the limitations, like the T-maze and the radial maze, is that the animals are food deprived which can induce stress and affect learning behavior.

Object localization

Object localization is a reference memory task dependent on cognitive mapping of environmental stimuli (Chapillon and Roullet 1997; Thinus-Blanc and Ingle 1985). It consists of a grey circular open field (diameter: 40 cm, wall height: 30 cm) made of polyvinyl chloride, with a floor covered by a sheet of white paper that is replaced after every trial. Spatial cues in the form of a black and white-striped cardboard pattern can be placed on the walls. Three identical round metal objects (diameter: 2 cm, height: 3 cm) are displayed in a V-shaped configuration in the open field. The mice are placed in center and are allowed to explore the maze during the habituation phase. After the habituation, during the learning phase, mice are placed in the center. The number of times the animals lean against the objects and the number of object sniffing episodes are recorded during four daily sessions lasting 15 min/day for three days. During the probe trial, the position of all three objects is changed to a different configuration. The animals are thus required to detect the novelty in a spatial context. Between each trial the objects are washed with water, and then dried in order to minimize olfactory cues. In another version, only two identical objects are used. For the learning phase, the two objects are kept side by side while during the probe trial the objects are kept diagonally opposite to each other (Wang et al. 2009). In another study, six different objects were used during the learning phase while the next few sessions used five objects followed by a session where one of the five objects was replaced by a new object (Lee et al. 2005).

The advantage of this behavioral paradigm is that it is simple and easy to design. It is not as stressful to mice as a dry-land maze and does not require food deprivation. However, in one study, the authors concluded that only object sniffing episodes could be considered as spatial tasks. Since object exploration as determined by the number of leanings, is not susceptible to intersession habituation, learning could not be measured (Belzung et al. 2001). Furthermore, this simple task to detect spatial novelty is not useful to assess spatial reference memory.

Complex alley maze

The complex alley maze was first described by McGaugh (1961). In this behavioral paradigm, mice have to learn a complex route through a maze leading to their home cage. The maze is comprised of a standard cage (37×21 ×15 cm), two tunnels and a larger central cage (56 × 32 ×18 cm). The cages are divided into several fields by Plexiglas walls with holes leading to adjacent fields. The animal is placed in the start box, which is the larger central cage. The mice have to learn the route through the tunnels leading to the home cage and avoid the dead ends. At the goal, the observer opens the door and allows the animal to enter the home cage as a reward. A trial is completed after mice reach the goal or after 20 min. The route is stable from trial to trial and thus requires spatial reference memory (Lewejohann et al. 2004). Observations are conducted using digital imaging techniques with automated animal tracking software. The number of errors (an error is counted for each time the mouse enters the first field of a path leading in the wrong direction), the number of total field entries and time used are calculated. After the acquisition of learning, long-term reference memory can be assessed after a delay of a few days.

The advantages of this maze include that it is dry, it involves the natural tendency of mice to go through tunnels and it does not need food deprivation as a motivator. As the route is stable from trial to trial, it is a good measure of reference memory. However, in the same study, one particular strain of mutant mice was not inclined to explore and thus remained in the start box for the 20-min trial (Lewejohann et al. 2004). Another limitation is that it needs a video tracking system that can add to research expenses.

Repeated acquisition and performance chamber

Most of the mazes utilized for evaluation of learning and memory measure the performance of a learned behavior. Thus these behavioral paradigms are not useful to assess the time course of the effects of experimental manipulations. A multiple schedule of repeated acquisition and performance was thus developed to assess learning vs rote performance within-behavioral test session and within-subject utilizing an apparatus modified from the rat (Furuya et al. 1988).

The repeated acquisition and performance chamber consists of a start box and a goal chamber separated by five compartments with one-way doors to prevent animals from returning to previous compartments. Following habituation, the animals are water deprived and then are allowed to explore for 20 min. In the repeated acquisition component, mice are required to learn a new sequence of door openings in each session. During the performance component, with an audio signal to discriminate it from repeated acquisition, the sequence of doors leading to the goal chamber containing saccharin is constant across sessions (Brooks et al. 2000). Latency to reach the goal box and sequence of door errors are recorded during the 10-min performance session.

Initially it takes considerable time to train the animals but there are several advantages. Repeated measures can be studied in same subject over long periods of time and each animal can be used as its own control (Thompson 1973; Wenger et al. 2004). This allows for comparison of the rate of learning at multiple levels of difficulty within the same test session. This paradigm has also been shown to be sensitive to hippocampal lesions resulting in longer latencies and increased errors in learning as well as it is useful to detect strain differences in spatial learning (Brooks et al. 2000).

Operant delayed response paradigms

Dunnettand colleagues (Dunnett 1985; Dunnett et al. 1988,1990; Dunnett and Martel 1990) developed a lever-pressing Delayed Matching (Nonmatching) to Position protocol for spatial working memory in rats by modifying a delayed conditional discrimination task (Herremans et al. 1994; Wallace et al. 1980). Recently, Goto et al. (2010) adapted the Delayed Matching to Position task for mice lever pressing using operant-conditioning chamber. The typical apparatus consists of a box with retractable levers. However, other response alternatives (e.g. illuminated response holes for nose-poke) have also been used (Delcasso et al. 2007; Wenger et al. 2004). The chamber is kept in a sound attenuating box in a test room. On pressing the correct lever, a dispenser delivers a food pellet as a reinforcer. In a typical Delayed Matching (Nonmatching) to Position task, the trial starts by the presentation of either a left or right front lever (sample) and the animal is required to press the lever. When the ratio requirement for presses to the lever is met, a delay interval is started. This is followed by the presentation of a choice of left or right front levers. In a matching condition, the choice of the same front lever as the sample is reinforced, whereas the other is not. In the non-matching version the animal is reinforced for pressing the lever that has not been presented before. The proportions of correct choices show a delay-dependent decrement. A higher ratio of response requirement to the sample resulted in increased accuracy, but the duration of inter-trial interval had no effect (Goto et al. 2010).

Delayed Matching to Position tasks use operant-conditioning chambers and have been a powerful means by which to study animal working memory (Brown and Wong 2007; D’Amato 1973; White 1985). Although the lever-pressing Delayed Matching to Position and other versions of delayed lever-pressing paradigms (Heise 1984; Heise etal. 1976; Pontecorvo 1983) are fundamentally similar to maze paradigms as a means of testing spatial working memory, lever-pressing procedures have numerous analytic advantages over maze procedures. Operant memory tests allow precise control of trial timing, retention and interval duration. The protocol can be arranged to include several trials per session. One potential problem with this paradigm is the possibility that animals may orient themselves to the correct lever during the delay and wait for the lever to be presented (Fletcher and Davis 1965). Another problem is that the stimulus to be remembered is not clearly defined (Pontecorvo 1983).

Summary of behavioral paradigms to assess learning and memory

Serial mazes are often employed in studies exploring aging processes, as well as in pharmacological, toxicological and neuro-developmental models, however, their use is still moderate, and this is probably why their potential advantages are not well known (Ingram 1988; Ingram et al. 1994) (Table 1). These behavioral tasks have provided many insights into the neurobiological mechanisms and the neuroanatomical substrates of learning and memory in rodents and are now commonly used to assess the effects of novel and potential therapies.

Table 1:

Characteristics of different behavioral paradigms to assess spatial memory.

Behavioral Paradigm	Working/Reference Memory	Video Tracking System	Labor intensive	Use of aversive stimuli	Food/Water Deprivation	Advantages	Limitations
T-maze/ Y-maze	Working	Not Required	No	No	Yes	Very simple	Requires constant handling
Multiple T-maze	Working/ Reference	Required	No	No	Yes	Complex with multiple choice points	Requires constant handling
Radial Maze	Working/ Reference	Required	No	No	Yes	Moderate level of stress	Possible use of non-spatial strategy
Morris Water Maze	Working/ Reference	Required	Yes	Yes	No	Fast Learning	Stressful
Barnes Maze	Working/ Reference	Required	Yes	Mild	No	Non stressful, use natural motivation	Slow or absent learning
Cheese Board	Working/ Reference	Required	Yes	No	Yes	Non stressful	Spatial distribution not clear
Hole board	Working/ Reference	Required	Yes	No	Yes	Fully automated	Depends on exploratory behavior
Object Localization	Working	Required	No	No	No	Simple and easy to design	Depends on exploratory behavior
Complex Alley Maze	Working	Required	Yes	No	No	Use natural motivation	Depends on exploratory behavior
Repeated Acquisition & Performance	Working/ Reference	Required	No	No	Yes	Long term and within subject comparison	Time consuming
Operant Delayed response	Working	Required	No	Sometimes	Yes	Precise control of timing	Response may be predictable

Maze-based tests are rather inexpensive, require minimal technological sophistication and are rapidly acquired. However, they are space-consuming and labor intensive without the video tracking system. In addition, mice may use non-spatial strategies to solve the maze defeating the very purpose of the task thus; careful study designs are needed to avoid such learning patterns.

In order to be trained to properly use these spatial tasks, animals may be food or water deprived. Animals can be maintained at a certain body weight, to prevent them from reaching satiety. Alternatively, they can be deprived for a certain number of hours; it has been shown that the speed of learning might be related to the time of food deprivation. If food deprivation is not sufficient, the learning process can be delayed. Similarly, food deprivation in the Multiple T-maze is considered to be more stressful than the Barnes maze (Inman-Wood et al. 2000). In addition, this task requires a significant motor component, a potential confound for studies involving mice that have motor or muscle dysfunction resulting from age or genetic manipulation.

It has been demonstrated that stress-related or anxiety-related behavior of two strains could result in differences between performance on water mazes and land mazes (Ennaceur et al. 2006; Kinsey et al. 2007). The stress produced by motivators and the physical fatigue produced by swimming exercise has an effect on the performance of mice (Mizunoya etal. 2004). Swimming in water mazes is considered an additional stress on experimental animals and therefore may have profound effects on cognitive performance (de Quervain et al. 1998; Holscher 1999). Mice differ from rats and thus need special consideration (Whishaw and Tomie 1996). Unlike rats, mice are not natural swimmers. The surface to mass ratio is also higher for mice than rats, leading to more rapid and substantial chilling when exposed to cool water (Gordon 2007). In addition to the stress induced by swimming, mice can adopt non-spatial strategies like thigmotaxis (wall hugging) that can alter the test results (Lipp and Wolfer 1998; Whishaw and Tomie 1996). Mice are also not fond of swimming and have often been reported to resort to passive floating in the Morris water maze (Nakazawa et al. 2003; Westerman et al. 2002). Mice perform better in tests not involving aversive or stressful stimuli (Francis et al. 1995) and they are comparable to rats in non-aquatic paradigms (Whishaw and Tomie 1996).

Another important variable to be considered is the gender of the animals (Frick et al. 2002; Harburger et al. 2008). Estrogen has been shown to affect memory in mice throughout a lifespan (Gresack et al. 2007). In addition, it has been demonstrated that an age-related decrease in spatial memory begins at an earlier age in female mice than in male mice and may be related to the cessation of estrous cycling (Frick et al. 2000). In a metanalysis, gender differences in mice were discussed in terms of spatial memory and showed that there is a small female advantage in the water maze and a small male advantage in the radial maze (Jonasson 2005).

The Morris water maze is simple and learning is rapid as compared to other behavioral paradigms such as the radial-arm maze (Becker et al. 1980). Studies comparing the radial and Morris mazes have shown that the number of trials required to achieve acceptable performance levels on some tasks in the radial maze is often two times that required for the Morris water maze, thus the acquisition process is slower (Ormerod and Beninger 2002). These differences are attributed to the fact that in contrast to the aquatic maze, animals in the radial maze may use a spontaneous alternation strategy. In comparative terms, both radial and circular Barnes mazes generate low levels of stress during the progress of the test (McLay et al. 1998) but these and the Morris water maze are sensitive to the deleterious effects of augmented corticosterone levels associated with stress (Holscher 1999; Sandi 1998; Wenk 2004).

The age of the animals also plays an important role in these tests. It is widely accepted that learning capacity declines with age, and this can be observed in the performance on these mazes. With age, swimming exploratory behavior and locomotor behavior decrease. These motor deficits have to be separated from the cognitive deficits that may be present in order to conclude that structural or functional changes occurring in the aged brain are contributing to the decrement of visual-spatial skills (Geinisman et al. 1995). It is therefore extremely important to select a test of spatial memory that induces minimum stress in old animals such as those that do not involve food restriction. In addition, it is important to take into account treatment-induced changes in sensori-motor function and motivational levels that can indirectly influence learning and memory.

A recent study proposed that at least two mouse strains and at least a land and a water maze paradigm should be applied when reliable data needs to be generated. This statement may be particularly relevant when drugs are studied for the improvement of learning and memory (Patil et al. 2009). The T-maze was used in one study, followed one week later by the Morris water maze (Bizon et al. 2007), while in another study, the Multiple T-maze was used with other measures of spatial memory like the Y-maze (Kunesova et al. 2008) or the Morris water maze (Weitzdoerfer et al. 2004). The inclusion of multiple paradigms, such as the parallel examination of appetitive and aversive conditioning in the same animals, can yield a more comprehensive and accurate characterization of the phenotypic profile of genetically modified mice (Yee et al. 2004).

Considerable evidence shows that the most useful approach is to use a set of tests rather than using single “hallmark” tests for evaluation of spatial memory. The findings from only one test can be misleading but comparing the results from different tests, one can obtain convincing results (Lewejohann et al. 2004).

What is the neurochemical basis of memory?

Synaptic plasticity, the changes in the structure or biochemistry of synapses that alter their post-synaptic effects, is considered to be the cellular mechanism underlying learning and memory. In late 1940, psychologist Donald Hebb hypothesized that permanent memory traces could be laid down by the sustained activation of neural networks (Hebb 1949). Bliss and coworkers, thirty years later, observed a long-lasting increase of synaptic strength in the rabbit dentate gyrus in response to high-frequency stimulation (Bliss and Gardner-Medwin 1973; Bliss and Lomo 1973). This synaptic enhancement eventually became known as long-term potentiation and is most extensively studied in the hippocampus (Douglas and Goddard 1975). Long-term potentiation displays many of the characteristics thought to be required for the molecular mechanism of memory formation (deToledo-Morrell et al. 1988; Gallagher and Nicolle 1993; Landfield and Lynch 1977; Moore et al. 1993; Shankar et al. 1998; Ward et al. 1999a,b).

Long-term potentiation (LTP) is extensively studied in thinly cut hippocampal slices that are kept alive in a bath of artificial cerebrospinal fluid. A brief, high-frequency train of action potentials in any one of the three major anatomical pathways within the hippocampus (the perforant pathway, the mossy fiber pathway or the Schaffer collateral pathway) produces long-term potentiation (Bliss and Collingridge 1993). The Schaffer collateral pathway lies between the presynaptic CA3 neurons and the CA1 postsynaptic target cells and is the best studied synaptic pathway in the hippocampus. Hippocampal lesions in this pathway disrupt long-term potentiation and lead to memory deficits.

Long-term potentiation, like the memory storage in intact animals, occurs in two phases: the early phase of LTP (usually described < 1 h after stimulation) that does not require protein synthesis and gene transcription; and the late phase of LTP (usually described >3 h after stimulation) that involves protein synthesis and gene transcription (Abel et al. 1997; Bliss and Collingridge 1993). Age-related deficits have been reported in the late phase of LTP (Eckles-Smith et al. 2000; Lanahan et al. 1997; Moore et al. 1993). Barnes showed that after multiple daily LTP induction sessions, LTP decayed more quickly in aged rats as compared to adult rats (Barnes 1979).

A great deal of evidence indicates that the N-methyl D-aspartate (NMDA) receptor is critical for most forms of LTP (Collingridge et al. 1983; Errington et al. 1987; Morris etal. 1990a,b; Shankar etal. 1998). Activation of NMDA receptors leads to calcium influx into the cell. When calcium enters the post-synaptic cell it activates protein kinases such as calcium/calmodulin protein kinase II (Malenka et al. 1989; Malinow et al. 1989). Calcium/calmodulin protein kinase then activates the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor, which increases conductance to sodium and potassium followed by increased responsiveness to glutamate (Lynch 2004). Calcium influx also leads to activation of adenyl cyclase, which in turn activates ATP to cAMP. This then activates protein kinase C leading to the phosphorylation of cAMP responsive transcription factor (CREB), protein synthesis and structural changes (Eichenbaum and Cohen 2001).

N-methyl D-aspartate (NMDA) receptor activation also plays an important role in the acquisition of spatial memory (Morris et al. 1986; Tsien et al. 1996). Pharmacological blockade of NMDA receptors or deletion of the NMDA receptor 1 (NR1) subunit of the NMDA receptor leads to a substantial impairment of spatial memory reference tasks (Danysz et al. 1995; Morris et al. 1990a,b; Tsien et al. 1996). Similarly, NR1 knockout mice show impairment of LTP (Nakazawa et al. 2003). In addition, it is important to note that with aging there is decrease in the expression of certain subunits of the NMDA receptor and their function in the hippocampus (Barnes et al. 1997; Clayton et al. 2002; Eckles-Smith et al. 2000; Magnusson 2001; Magnusson et al. 2006; Sonntag et al. 2000).

In 1990, an NMDA-receptor-independent form of LTP was discovered in CA1 pyramidal cells (Grover and Teyler 1990). When NMDA receptors are blocked, this form of LTP can be induced if very high-intensity stimuli (in terms of frequency, duration or current amplitude) are applied. This form of synaptic plasticity appears to be mediated by calcium influx through voltage-dependent, L-type calcium channels (vdcc) and has been called vdccLTP (Aniksztejn and Ben-Ari 1991; Cavus and Teyler 1996; Coussens and Teyler 1996a,b; Grover and Teyler 1990; Johnston et al. 1992; Morgan et al. 2001; Morgan and Teyler 2001a,b; Teyler et al. 1994). Like nmdaLTP, vdccLTP appears to be synapse-specific, that is, only the synapses active during vdccLTP induction are strengthened (Grover and Teyler 1992). Unlike nmdaLTP, which requires serine-threonine kinases (Lovingeretal. 1987; Malenka et al. 1989; Malinow et al. 1989), vdcc LTP appears to require tyrosine kinase (Cavus and Teyler 1996).

Previous studies of LTP in aged animals (using high-intensity stimulus parameters) observed no LTP-induction deficits (Barnes et al. 1996a,b; Chang et al. 1991). However, one study observed that vdccLTP if measured in isolation, was increased in magnitude in aged animals (Shankar et al. 1998). In the light of this new evidence, the authors suggested that the expected decrease in nmdaLTP in aged animals was probably offset by the increased vdccLTP. When lower intensity stimuli were used or vdcc LTP was blocked, the nmdaLTP deficits were unmasked (Boric et al. 2008; Deupree et al. 1993; Moore et al. 1993; Rosenzweig et al. 1997).

It has also been shown that LTP induction is due to NMDA receptor activation while LTP maintenance involves modification of AMPA receptors. The ampakines, positive modulators of AMPA receptors, have been shown to be cognitive enhancers (Ingvar et al. 1997) indicating that up-regulation of AMPA receptor-mediated synaptic responses facilitate LTP formation. Another neurotransmitter Gamma Amino butyric acid (GABA) has been shown to be involved in learning and memory. The benzodiazepine potentiates the effect of GABA and produces impairment of learning and memory while reverse agonists and antagonists of benzodiazepine receptors has been shown to enhance memory (Lister 1985; McGaugh 1989). Similarly, noradrena­line, serotonin and histamine are also shown to be possibly involved in learning and memory processes (D’Hooge and De Deyn 2001). Acetylcholine and its interaction with other neurotransmitters and neuromodulators—including norepinephrine, dopamine, serotonin, GABA, opioid peptides, galanin, substance P and angiotensin II are also involved in modulation of learning and memory processes (Decker and McGaugh 1991).

The strong link between spatial memory and synaptic plasticity has been extensively reviewed (Martin et al. 2000). Furthermore, there has been experimental proof that learning induces LTP-like processes in the hippocampus (Foster et al. 2000; Foster et al. 1996; Power et al. 1997). Like the performance on the mazes, LTP is a feature of the hippocampus and is influenced by stress (Shors et al. 1990; Shors and Thompson 1992). A study found that stress impairs both spatial memory and LTP suggesting that a common mechanism is shared by the two assessment parameters (Baker and Kim 2002). Moreover, it has been demonstrated that animals with more durable LTP in a given age group tended to show the best spatial learning, providing support to the hypothesis that LTP at hippocampal synapses and spatial learning may depend on similar mechanistic processes (Barnes 2003).

Though a great deal is understood about spatial memory from research over the past few decades, there are still several unanswered questions. Experimental paradigms of memory are very important in the elucidation of the pathophysiology of age-associated memory impairment as they can be utilized to study memory-enhancing drugs. Spatial memory problems such as the inability to find your way back home or forgetting where you left your car keys, can be devastating in old age. In conclusion, to better understand the age-related impairment of hippocampal-based spatial memory and to tease apart the mechanisms, it is important to select a set of appropriate behavioral paradigms. In addition, to generate meaningful data and convincing interpretation of these results, additional experimental studies using long-term potentiation are equally important. It is anticipated that such studies will make it possible to develop therapeutic treatments to prevent or alleviate memory decline in normal aging and provide a greater quality of life to those getting older.