Hippocampal synaptic activity, pattern separation ... - Semantic Scholar

902

Biochemical Society Transactions (2011) Volume 39, part 4

Hippocampal synaptic activity, pattern separation and episodic-like memory: implications for mouse models of Alzheimer’s disease pathology Alice Palmer and Mark Good1 School of Psychology, Cardiff University, Park Place, Cardiff CF10 3AT, U.K.

Abstract The present review summarizes converging evidence from animal and human studies that an early target of amyloid pathology is synaptic activity in the DG (dentate gyrus)/CA3 network. We briefly review the computational significance of the DG/CA3 network in the encoding of episodic memory and present new evidence that the CA3/DG pattern of activation is compromised in a mouse model of amyloid pathology. In addition, we present a new behavioural method to test the prediction that amyloid-related synaptic pathology will disrupt the formation of an integrated episodic-like (what, where and when) memory in mice.

Introduction There is converging evidence from both human and animal studies that the initial stages of AD (Alzheimer’s disease) are associated with synaptic pathology in medial temporal lobe structures that are necessary for episodic memory. To be more specific, recent work has highlighted functional deficits in the DG (dentate gyrus)/CA3 hippocampal subfields in animal models of AD pathology and in patients in the early stages of dementia. Here we briefly describe computational theories of hippocampal function that assign a pattern separation role to the DG/CA3 network, a process that is key to the encoding of episodic memory. We present evidence that activity in the DG/CA3 network is compromised during learning in transgenic mice expressing an APP (amyloid precursor protein) mutation. In addition, we describe a behavioural method to test the prediction that disruption to this network will lead to a deficit in episodic memory for what, where and when an item is presented to mice. Our preliminary results support the hypothesis that pattern separation processes supported by the entorhinal cortex and DG/CA3 network are an early target of amyloid-induced synaptic pathology and this leads to the disruption of episodic memory.

Hippocampus, information processing and memory The medial temporal lobe is generally acknowledged to be one of the earliest targets of pathology in AD. Characterization of AD pathology initially emphasized cell loss (starting in the Key words: Alzheimer’s disease (AD), amyloid β-peptide (Aβ), dentate gyrus (DG), episodic-like memory, hippocampal synaptic activity, pattern separation. Abbreviations used: AD, Alzheimer’s disease; APP, amyloid precursor protein; Aβ, amyloid βpeptide; DG, dentate gyrus; MF, mossy fibre; PP, perforant path; RC, recurrent axon collateral; SC, Schaffer collateral; WT, wild-type. 1 To whom correspondence should be addressed (email [email protected]).

 C The

C 2011 Biochemical Society Authors Journal compilation 

parahippocampal/entorhinal cortex) as a diagnostic feature of the disease [1]. However, subsequent work has shown that synaptic loss, and subsequent network dysfunction, is a major correlate of cognitive decline in AD ([2–5]; see also [6,7] for reviews). Evidence that the parahippocampal cortex (including the entorhinal cortex) undergoes pathological changes during the early stages of the disease is theoretically significant, as this region provides the main input pathway for cortical information into the hippocampus [8,9]. Thus, during early stages of AD, the hippocampus or to be more precise the DG and the CA3 region receive a degraded input from the entorhinal cortex. Neural network theories of hippocampal function provide a valuable framework to evaluate putative computational significance of such a degraded input. Before we summarize these views, we briefly outline the key anatomical features of the hippocampus that have informed computational models of hippocampal information processing. The hippocampus receives highly processed multimodel information from the cortex, which includes the parahippocampal cortex and the perirhinal cortex, and reward-related information from the amygdala and orbitofrontal cortex, via PP (perforant path) projections from the entorhinal cortex (for reviews see [10,11]). PP inputs from layer 2 of the entorhinal cortex synapse on granule cells in the DG (see Figure 1A). The granule cells in turn send a sparse projection [via the MFs (mossy fibres)] to the CA3 region. The CA3 region also receives a direct projection from the entorhinal cortex. However, the largest number of synaptic connections in the CA3 stems from RC (recurrent axon collaterals) of CA3 neurons. The CA3 region projects, in turn, to the CA1 sub-field via the SC (Schaffer collaterals). CA1 neurons provide the output from the hippocampus and project to the subiculum, entorhinal cortex (and related parahippocampal structures), frontal cortex and subcortical structures via the fornix. Biochem. Soc. Trans. (2011) 39, 902–909; doi:10.1042/BST0390902

Models of Dementia: the Good, the Bad and the Future

Figure 1 Hippocampal networks and memory (A) Summary of connections between the major excitatory input pathway (entorhinal cortex) and output system of the hippocampus (CA1, subiculum and fornix). The projections between the putative pattern separation system (DG region) and the auto-associative network (CA3 region) are sparse. (B) Mean normalized c-Fos expression in the DG and CA3 regions of normal C57Bl6, 12-month-old WT and Tg2576 mice, following exposure to a familiar or novel experimental room sharing common features. C57Bl6 and WT mice showed elevated levels of c-Fos expression (a marker of neuronal activity) in the DG following exposure to the novel environment. *P < 0.05, results are means ± S.E.M.

this feature coupled with the small, but relatively strong, number of projections from the DG to the CA3 region generates a relatively sparse pattern of activity within the CA3 network. The end result of this orthogonalization of inputs is that different events will be represented by a different pattern of activity in the CA3 network. Furthermore, the sparse nature of the input to the CA3 network acts to reduce interference between similar input patterns (e.g. event memories with overlapping content). The DG thus acts as a competitive network that performs a pattern separation function, condensing the cortical input from the entorhinal cortex to a relatively sparse code [15]. The role of the CA1 region is to receive a sparse representation from the CA3 network that is then passed back to the cortex responsible for the provision of the original inputs to the hippocampus.

Amyloid and hippocampal synaptic function

A number of influential computational theories of the hippocampus assign an important function to the interaction between the DG and CA3 networks in memory encoding ([12–14]; see [15] for review and references therein). These theories posit that the CA3 region acts as an auto-associative network that enables the binding (association) of multiple inputs to the network that represent an event or episode. Extensive recurrent collateral projections from CA3 neurons play an important role not only in encoding but also in the retrieval of the pattern of activity established during encoding, when one or more components of the input pattern (representing the episode) is available during retrieval. The direct input from the PP to the CA3 is thought to relay information that initiates retrieval from the CA3 network. Critical to the generation of the unique pattern of CA3 activity during memory encoding is the sparse input pattern received from the DG via MF projections. Electrophysiological recording studies carried out in rats have revealed that the firing activity of dentate granule cells appears to be relatively sparse [16]. According to Rolls [15],

The amyloid cascade hypothesis proposes that the excess production or accumulation of Aβ (amyloid β-peptide) leads eventually to the pattern of cell loss that characterizes AD. A key stage in this process is the disruption of synaptic connections caused by soluble, rather than fibrillar, forms of Aβ [17,18]. Support for the synaptic deficit theory has emerged from animal studies showing that certain oligomeric forms of soluble amyloid can reduce hippocampal synaptic density and cause memory deficits [19,20]). Indeed, spine loss is a common feature of many genetic animal models of amyloid pathology (e.g. [21]). Other evidence has shown that soluble oligomeric forms of Aβ (including extracts of ex vivo samples of cerebrospinal fluid or a post-mortem aqueous solution of the cerebral cortex from patients with AD) disrupt the induction of long-term synaptic potentiation and enhance the induction of long-term synaptic depression in animals (both are cellular models of synaptic processes thought to underpin memory (see [22,23] for reviews). The pattern of synapse disruption in the hippocampus of mice expressing human APP mutations is of particular theoretical interest. For example, Jacobsen et al. [24] reported that Tg2576 mice, expressing the human Swedish APP mutation, showed an accelerated age-related decline in synaptic density in the DG region of the hippocampus. This was accompanied by a decline in basal (α-amino-3-hydroxy5-methylisoxazole-4-propionic acid receptor-mediated) synaptic transmission and a deficit in the induction of longterm potentiation in the DG. Functional and morphological changes noted by Jacobsen et al. [24] were also accompanied by impairments in spatial (context) memory (see also [25]). Chapman et al. [26] also noted age-related changes in DG synaptic plasticity that correlated with performance of a spatial working memory task in Tg2576 mice (see [27] for review). As noted earlier, the entorhinal cortex is the major input pathway to the hippocampus and is one of the sites affected earliest in AD. Very recently, Harris et al.  C The

C 2011 Biochemical Society Authors Journal compilation 

903

904

Biochemical Society Transactions (2011) Volume 39, part 4

[28] restricted the expression of mutant Aβ to the entorhinal cortex in a mouse. This resulted in elevated levels of soluble Aβ and synaptic loss in the DG as well as in abnormalities within the hippocampal CA1 region. This interesting study indicates that elevated Aβ in the entorhinal cortex can lead to the trans-synaptic disruption of hippocampal circuits and thus compromise information processing supported by this structure.

DG-CA3 activation in normal and transgenic APP mice Computational models have suggested that the DG/CA3 network is vital for the disambiguation of overlapping patterns of information and that disruption to this network may be central to memory deficits observed in normal aging and disease (cf. [15]). Electrophysiological studies have shown that DG granule firing is sensitive to small/subtle changes in the environment [29]. Furthermore, mice with a specific deficit in DG synaptic plasticity showed a marked impairment in discrimination between contexts with overlapping features (i.e. pattern separation [30]), and this was accompanied by a reduction in the context-dependent firing rate of CA3 pyramidal cells. In addition, aged rats show impaired discriminative firing of CA3 neurons in relation to two similar (but different) environments; this suggests that this system is a focus of age-related synaptic dysfunction [31]. These findings suggest that the disruption of dentate synaptic plasticity and normal aging can disrupt pattern separation processes supported by the hippocampus. To our knowledge, functional characteristics of the DG/CA3 network during memory encoding have not been examined in APP mutant mice. The evidence reviewed previously leads to the prediction that this network in mutant mice should show aberrant activity during memory encoding. To investigate the functional properties of this network, we carried out an analysis of immediate early gene (cFos) expression, as a marker of neuronal activity in normal and Tg2576 APP mice, following exposure to a familiar or a novel environment. In this study, normal and transgenic mice were placed in a square clear Perspex arena for 30 min in one experimental room for 3 days. On the fourth day, half the mice were placed in the arena in the familiar room and the remainder were placed in an identical arena located in a novel experimental room (the room contained novel extramaze landmarks, but was otherwise similar to the pre-exposure room in terms of its overall shape and dimensions). After exposure to these environments, brain tissue was collected and subjected to standard c-Fos immunohistochemistry procedures (see [32] for details). Figure 1(B) shows the pattern of c-Fos activation in the DG and CA3 network in the dorsal hippocampus of normal C57BL6 mice. Inspection of this Figure shows that, following exposure to a novel room, c-Fos expression in the DG of C57Bl6 mice, but not in the CA3 region, in comparison with animals in the control (familiar room) condition. Figure 1(B) also shows a similar pattern of  C The

C 2011 Biochemical Society Authors Journal compilation 

c-Fos expression in 12-month-old WT (wild-type) mice. In contrast, 12-month-old Tg2576 mice showed a reversed pattern of c-Fos expression: there was a marked increase in c-Fos expression in the CA3 region, but no change in the DG. These preliminary data clearly indicate the differential activation of the DG-CA3 network in Tg2576 mice and confirm that expression of mutant human APP disrupts the functional characteristics of a network thought to be crucial for pattern separation and the encoding of episodic memory.

Episodic memory in animals The extant literature together with our own preliminary immediate early gene studies have shown that the entorhinal cortex → DG → CA3 network is abnormal in mice expressing human APP mutations. This parallels findings concerning aged humans and patients diagnosed with amnesic mild cognitive impairment, who show functional abnormalities [fMRI (functional magnetic resonance imaging) hyperactivity] and structural changes in the CA3/DG network [33,34]. Of course, a straightforward prediction based on these observations and computational theories [15] is that elevated Aβ production should be sufficient to disrupt episodic memory in transgenic mice. However, evaluation of this prediction has been hampered by the fact that behavioural tests of episodic memory in mice are still in their infancy (and in some cases are subject to alternative theoretical explanations). Episodic memory is generally defined in terms of its content: an integrated memory of what, where and when an event occurred [35]. A defining characteristic of episodic memory in humans is that the person consciously experiences recollection of prior events. Of course, the assessment of this property of episodic memory is not possible in nonverbal animals. However, Clayton et al. [36,37] have proposed a set of behavioural criteria (based on their seminal work with food-storing birds) that emphasize the original tripartite definition of episodic memory formulated by Tulving [35]. To be more specific, the ability to form an integrated memory of what, where and when an item was encountered. This type of memory in animals is generally referred to as episodic-like memory. Work on memory in food-storing birds by Clayton et al. provided the catalyst for a raft of behavioural studies of episodic-like memory in rodents. These experiments have been reviewed extensively elsewhere (see [38] for a recent review) and we shall confine ourselves to an outline of the main experimental approaches that have been followed to examine episodic-like memory in animals. Broadly speaking, the assessment of episodic-like memory in rodents has taken three different experimental approaches [38]. The first approach attempted to adapt procedures used by Clayton et al. for use with rodents. Babb and Crystal [39] assessed the ability of rats to discriminate where and when a valued food item was encountered in a radial arm maze. During this procedure, rats demonstrated the ability to discriminate the arm (location) where a valued

Models of Dementia: the Good, the Bad and the Future

Figure 2 Memory for what, where and when (A) Illustration of the object recognition procedure used by Good et al. [46] to investigate the effects of APP overexpression in Tg2576 mice on episodic-like (what, where and when) memory. (B) The performance of 12-month-old WT and Tg2576 mice (data taken from [46]). Tg2576 mice showed a preference for exploration of old objects (relative recency effect C/D compared with A and B), but, unlike WT mice, this preference was not modified by memory for the spatial location of object A. (C) The three sub-panels illustrate alternative patterns of connections (associations) that may arise following exposure to two cues (A and B) and their pairing with a third event (such as a food reward). Features of the two cues (A and B) are represented in the network by a and b respectively. In the upper panel, representations of a and b form direct links with a third stimulus (represented by c) with which they are paired (a simple elemental model). In the middle panel, the representations of a and b activate a configural unit (black circle) that in turn activates c. The lower panel shows a hybrid model (combining elemental and configural associations) that forms the basis of configural theories of hippocampal function.

food item was available depending on the length of a retention interval between a study and test phase. Crystal et al. have also shown that rats are able to encode detailed sensory features concerning food rewards, their specific spatiotemporal features and that they are able to adapt their search behaviour when new information on the value of rewards is made available. These procedures have not been used in mice, as far as we are aware; nevertheless, they serve to demonstrate that rodents are capable of encoding episodiclike information (for a review of these studies see [40]). The second approach to the study of episodic-like memory in rodents has exploited the natural tendency of rodents to explore novel objects. In the standard object recognition procedure, animals are first presented with two identical copies of an object in a familiar arena. After a variable delay, the animals are then returned to the arena and presented with a copy of the familiar object (from the study phase) and a novel object. Normal rats show a preference for exploration of the novel object (see Winters et al. [41] for review). A wealth of evidence suggests that the discrimination of novel

objects from familiar ones relies upon the perirhinal cortex [41]. Apart from encoding information about the features of objects, rodents also encode information about the location of objects and their temporal order, features mediated by the hippocampus and the frontal cortex respectively (see [42] for a recent review). Two (similar) procedures have been published that examined the ability of rats and mice to encode the spatial and temporal orders in which objects were presented [43– 46]. In both procedures, animals are first exposed to two sets of objects at two different time periods. At the test stage, the animal is presented with objects that are either old (presented in the first study phase) or recent (presented in the second study phase) and are located in either the same or a different spatial location relative to the sample phase (see Figure 2A for a summary of the procedure used by Good et al. [46]). In studies conducted by Good et al. [45,46] rats and mice preferentially explored an old object that was presented in a new (but familiar) spatial location (object A in Figure 2A). These studies indicate that the exploratory activity of rodents  C The

C 2011 Biochemical Society Authors Journal compilation 

905

906

Biochemical Society Transactions (2011) Volume 39, part 4

can be influenced by the temporal order (relative recency) and the spatial location of objects. Furthermore, episodiclike recognition memory is severely impaired in aged Tg2576 mice ([46] see Figure 2B), whereas the same mice show a normal tendency to explore novel objects (normal object novelty detection). However, additional experiments showed that aged Tg2576 mice were impaired specifically in terms of detecting the movement of familiar objects to a new spatial location. As a result, it comes as no surprise that Tg2576 mice were impaired in the episodic-like recognition memory task. Two issues prevent one from drawing conclusions regarding the effects of amyloid production on episodiclike memory from this study. First, Tg2576 mice failed to react to changes in the spatial location of the objects. This may reflect a performance deficit or that the Tg2576 mice failed to encode information on the location of objects [47]. Secondly, the preference of normal mice for the old object in a different location is subject to an alternative theoretical account that does not rely upon the mice forming an integrated representation of what, where and when. Instead, the performance of control mice may reflect elemental associative processes whereby the separate effects of a change in spatial location and temporal features of objects summate to influence exploratory activity in rats [46].

Configural memory for what, where and when We argue that the latter point is critical to the valid assessment of episodic-like memory processes in animals, particularly models of human dementia. The defining feature of episodic memory is that it reflects an integrated memory of spatiotemporal features associated with an item(s). This specific property of episodic-like memory has led to a third approach that examines the formation of an integrated configural representation of what, where and when an item was presented. Traditional associative theories of learning posit that animals form associations between representations of events ([48]; see Figure 2C top panel). Although this view has been highly successful in capturing the processes of learning in animals and humans, there are occasions where such accounts fail to predict the pattern of learning: for example, under conditions where the presentation of a compound cue (formed by the simultaneous presentation of two cues, A and B) signals a different reinforcement outcome from when each cue is presented individually [49]. In contrast with elemental theories, configural theories of learning posit that learning involves the formation of associations between configural representations (that specify that two or more events occur together) and reward ([50], see Figure 2C middle panel). The view that the hippocampus is central to the formation of an integrated/configural memory of a pattern of stimulation was expressed elegantly by the configural association theory proposed by Rudy and Sutherland [51,52]. The original theory posited that the hippocampus was  C The

C 2011 Biochemical Society Authors Journal compilation 

Figure 3 A novel test of mouse memory for episodic information (A) An illustration showing the key stages of the experimental design used by Iordanova et al. [54,55]. Normal C57Bl6 mice (B) show more freezing in the context in which the conditioned auditory cue was presented in the morning and similarly so for the afternoon. Performance is shown as an elevation ratio generated by applying this formula: freezing in the target context in the morning/freezing in both contexts. This formula was applied to morning and afternoon datasets and this leads to the below-0.5 ratio observed in the afternoon session. Results are means ± S.E.M.

required for the formation of a configural memory that represents the co-occurrence of two or more (spatial and nonspatial) cues. According to this view, when different sensory inputs are presented together, they are not linked directly, but are stored as individual inputs that are also linked to an independent shared/or configural representational unit (memory). This form of representation was contrasted against an associative or elemental system, whereby binary associations are formed between representations of stimuli (see Figure 2C lower panel). The application of this theory in explaining the role of the hippocampus in forming configural memory of non-spatial cues has been challenged on several grounds (e.g. [53]). Furthermore, until recently, there were relatively few tests that sought to determine whether the formation of a configural memory for the spatial and temporal properties of cues is reliant upon the hippocampus. Iordanova et al. [54,55] developed a procedure in which rats were seen to integrate information on where and when a cue was presented. The basic design of the task is shown in Figure 3(A). At the first stage of training, rats were exposed to two contexts [conditioning chambers with either a spotted (context A) or striped (context B) wall-paper pattern on the walls of the chamber] in the morning as well as in the afternoon. However, in the morning the rats were presented with a tone in the spotted context and a clicker in the striped context. In the afternoon, the pattern of presentations in each context was reversed (the clicker was presented in context A and the tone presented in context B). Note that the animals were not required to learn these contingencies at any stage.

Models of Dementia: the Good, the Bad and the Future

After 4 days of exposure to the context, time of day and cue patterns, the animals received a set of conditioning trials at mid-day on days 5 and 6 in a novel context. Here, one of the auditory cues (e.g. the tone) was paired with a footshock and the remaining cue was presented on its own in the absence of a footshock. The logic behind this manipulation was to establish a conditioned response to the auditory cue that could be used to interrogate a memory with regard to where and when the cue was presented. After conditioning was completed, the rats received test trials in which they were simply placed in each context in the morning and afternoon, as they were during exposure, except that no auditory cues were presented. If the rats had formed a representation that specified the where and when combinations of each auditory cue, they would show greater freezing in the context where the conditioned auditory cue (the tone in our example) had been presented during exposure in the morning and would likewise exhibit greater freezing in the context in which the same auditory cue had been presented in the afternoon. In this instance, the rats should show more freezing in the striped context in the morning and exhibit freezing in the spotted context in the afternoon. In terms of the configural model illustrated in Figure 2(C), the context and time-ofday features should activate a configural unit specifying the morning/striped context/tone combination, and thus elicit freezing. It is important to note here that it is because the rats had received exposure to both contexts, in the morning and the afternoon, and had experienced both auditory cues in each context and at each time of day, that a simple binary or elemental associative representation (e.g. tone-morning) would not yield the predicted pattern of results described above. Normal rats display the predicted pattern of freezing and thus formed a configural (or episodic-like) representation that specified the time and place where each individual auditory cue had been presented [54]. It is significant that lesions of the hippocampus in rats disrupted the formation of episodic-like memory of what, where and when, but that they did not disrupt the formation of binary associations between a cue and a context or between a cue and the time of day [55]. Thus damage to the hippocampus severely impairs the formation of a configural episodic-like memory of what, where and when information. In a recent series of experiments, we used the same procedure to assess whether C57Bl6 mice are able to form an integrated memory of what, where and when. As inspection of Figure 3(B) shows, C57Bl6 mice showed more freezing in the context associated with the conditioned auditory cue in the morning and the afternoon. The critical question is whether the performance of this task is impaired in the case of mice expressing human APP mutations. We can report that a preliminary assessment of performance in 3and 10-month-old Tg2576 mice shows an age-related deficit with regard to this task. This is consistent with the view that Aβ production disrupts neural networks necessary for the formation of a memory of what, where and when events occur (A. Palmer, R. Honey and M. Good, unpublished work).

Pattern separation and pattern completion Computational models not only explain how a compromised DG/CA3 network may disrupt the pattern separation required for episodic memory, they also suggest that such a deficit will bias the CA3 network towards a pattern completion processes. The latter refers to the process of eliciting a pre-existing pattern of CA3 activity in response to a partial or degraded input. A putative shift in the CA3/DG network towards pattern completion will result in failure to separate memories of events with overlapping elements and will generate considerable interference in long-term memory. Interestingly, a shift in the CA3 region activity has recently been reported in patients diagnosed with amnesic mild cognitive impairment ([5], see also [56]), a major risk factors in the case of AD. Whether such changes are present in mouse models of amyloid pathology remains to be determined, although the preliminary result of our c-Fos study is consistent with observation of modified learningrelated activity in the CA3 subregion in APP mice.

Conclusion The aim of this review was to highlight the putative impact of amyloid pathology on computational processes carried out by the hippocampus in humans and other animals and its consequences with regard to memory formation. Convergent evidence from rodents, transgenic models of amyloid pathology and human participants indicates that aging and dementia disrupt functional characteristics of the DG/CA3 hippocampal network and thus affect memory encoding. The early stages of AD (i.e. amyloid-induced synaptic pathology) may exacerbate the normal age-related decline in pattern separation processes supported by DG and CA3 regions and result in a parallel rapid decline in episodic memory. Further work on animals together with human imaging studies will refine these ideas and stimulate the design of cognitive tests that will aid in the detection of abnormalities within this network. This in turn may contribute to the early diagnosis, treatment and monitoring of patients with AD and those at risk of the disease.

Acknowledgement We thank Robert Honey and Mihaela Iordanova for insightful discussions on this topic.

Funding This work was funded by the Wellcome Trust and the Biotechnology and Biological Sciences Research Council.

References 1 Braak, H. and Braak, E. (1991) Neuropathological stageing of Alzheimer-related pathology. Acta Neuropathol. 82, 239–259  C The

C 2011 Biochemical Society Authors Journal compilation 

907

908

Biochemical Society Transactions (2011) Volume 39, part 4

2 Terry, R.D., Masliah, E., Salmon, D.P., Butters, N., DeTeresa, R., Hill, R., Hansen, L.A. and Katzman, R. (1991) Physical basis of cognitive alterations in Alzheimer’s disease: synapse loss is the major correlate of cognitive impairment. Ann. Neurol. 30, 572–580 3 DeKosky, S.T. and Scheff, S.W. (1990) Synapse loss in frontal cortex biopsies in Alzheimer’s disease: correlation with cognitive severity. Ann. Neurol. 27, 457–464 4 Kerchner, G.A., Hess, C.P., Hammond-Rosenbluth, K.E., Xu, D., Rabinovici, G.D., Kelley, D.A., Vigneron, D.B., Nelson, S.J. and Miller, B.L. (2010) Hippocampal CA1 apical neuropil atrophy in mild Alzheimer disease visualized with 7-T MRI. Neurology 75, 1381–1387 5 Mueller, S.G., Schuff, N., Yaffe, K., Madison, C., Miller, B. and Weiner, M.W. (2010) Hippocampal atrophy patterns in mild cognitive impairment and Alzheimer’s disease. Hum. Brain Mapp. 31, 1339–1347 6 Nimmrich, V. and Ebert, U. (2009) Is Alzheimer’s disease a result of presynaptic failure? Synaptic dysfunctions induced by oligomeric β-amyloid. Rev. Neurosci. 20, 1–12 7 Gong, Y. and Lippa, C.F. (2010) Disruption of the postsynaptic density in Alzheimer’s disease and other neurodegenerative dementias. Am. J. Alzheimer’s Dis. Other Dement. 25, 547–555 8 Hyman, B.T., Van Hoesen, G.W., Kromer, L.J. and Damasio, A.R. (1986) Perforant pathway changes and the memory impairment of Alzheimer’s disease. Ann. Neurol. 20, 472–481 9 Witter, M.P. (2007) The perforant path: projections from the entorhinal cortex to the dentate gyrus. Prog. Brain Res. 163, 43–61 10 Amaral, D.G. and Witter, M.P. (1989) The three-dimensional organization of the hippocampal formation: a review of anatomical data. Neuroscience 31, 571–591 11 Lavenex, P. and Amaral, D.G. (2000) Hippocampal–neocortical interaction: a hierarchy of associativity. Hippocampus 10, 420–430 12 Marr, D. (1971) Simple memory: a theory for archicortex. Philos. Trans. R. Soc. London Ser. B 262, 23–81 13 McNaughton, B.L. and Morris, R.G.M (1987) Hippocampal synaptic enhancement and information storage within a distributed memory system. Trends Neurosci. 10, 408–415 14 Ceresti, E. and Treves, A. (2010) How informative are aspatial CA3 representations established by the dentate gyrus. PLoS Comput. Biol. 4, 1–16 15 Rolls, E.T. (2010) A computational theory of episodic memory formation in the hippocampus. Behav. Brain Res. 215, 180–196 16 Jung, M.W., Weiner, S.I. and McNaughton, B.L. (1994) Comparison of spatial firing characteristics of units in dorsal and ventral hippocampus of the rat. J. Neurosci. 14, 7347–7356 17 Terry, R.D. (1996) The pathogenesis of Alzheimer disease: an alternative to the amyloid hypothesis. J. Neuropathol. Exp. Neurol. 55, 1023–1025 18 Lue, L.F., Kuo, Y.M., Roher, A.E., Brachova, L., Shen, Y., Sue, L., Beach, T., Kurth, J.H., Rydel, R.E. and Rogers, J. (1999) Soluble amyloid β peptide concentration as a predictor of synaptic change in Alzheimer’s disease. Am. J. Pathol. 155, 853–862 19 Cleary, J.P., Walsh, D.M., Hofmeister, J.J., Shankar, G.M., Kuskowski, M.A., Selkoe, D.J. and Ashe, K.H. (2005) Natural oligomers of the amyloid-β protein specifically disrupt cognitive function. Nat. Neurosci. 8, 79–84 20 Haass, C. and Selkoe, D.J. (2007) Soluble protein oligomers in neurodegeneration: lessons from the Alzheimer’s amyloid β-peptide. Nat. Rev. Mol. Cell Biol. 8, 101–112 21 Aoki, C., Mahadomrongkul, V., Fujisawa, S., Habersat, R. and Shirao, T. (2007) Chemical and morphological alterations of spines within the hippocampus and entorhinal cortex precede the onset of Alzheimer’s disease pathology in double knock-in mice. J. Comp. Neurol. 505, 352–362 22 Shankar, G.M. and Walsh, D.M. (2009) Alzheimer’s disease: synaptic dysfunction and Aβ. Mol. Neurodegener. 4, 48 23 Ondrejcak, T., Klyubin, I., Hu, N.W., Barry, A.E., Cullen, W.K. and Rowan, M.J. (2010) Alzheimer’s disease amyloid β-protein and synaptic function. Neuromol. Med. 12, 13–26 24 Jacobsen, J.S., Wu, C.C., Redwine, J.M., Comery, T.A., Arias, R., Bowlby, M., Martone, R., Morrison, J.H., Pangalos, M.N., Reinhart, P.H. and Bloom, F.E. (2006) Early-onset behavioral and synaptic deficits in a mouse model of Alzheimer’s disease. Proc. Natl. Acad. Sci. U.S.A. 103, 5161–5166 25 Dong, H., Yuede, C.M., Coughlan, C.A., Murphy, K.M. and Csernansky, J.G. (2009) Effects of donepezil on amyloid-β and synapse density in the Tg2576 mouse model of Alzheimer’s disease. Brain Res. 1303, 169–178 26 Chapman, P.F., White, G.L., Jones, M.W., Cooper-Blacketer, D., Marshall, V.J., Irizarry, M., Younkin, L., Good, M.A., Bliss, T.V.P., Hyman, B.T. et al. (1999) Impaired synaptic plasticity and learning in aged amyloid precursor protein transgenic mice. Nat. Neurosci. 2, 271–276  C The

C 2011 Biochemical Society Authors Journal compilation 

27 Palop, J.J. and Mucke, L. (2010) Amyloid-β-induced neuronal dysfunction in Alzheimer’s disease: from synapses toward neural networks. Nat. Neurosci. 13, 812–818 28 Harris, J.A., Devidze, N., Verret, L., Ho, K., Halabisky, B., Thwin, M.T., Kim, D., Hamto, P., Lo, I., Yu, G.Q. et al. (2010) Transsynaptic progression of amyloid-β-induced neuronal dysfunction within the entorhinal–hippocampal network. Neuron 68, 428–441 29 Leutgeb, J.K., Leutgeb, S., Moser, M.B. and Moser, E.I. (2007) Pattern separation in the dentate gyrus and CA3 of the hippocampus. Science 315, 961–966 30 McHugh, T.J., Jones, M.W., Quinn, J.J., Balthasar, N., Coppari, R., Elmquist, J.K., Lowell, B.B., Fanselow, M.S., Wilson, M.A. and Tonegawa, S. (2007) Dentate gyrus NMDA receptors mediate rapid pattern separation in the hippocampal network. Science 317, 94–99 31 Wilson, I.A., Ikonen, S., Gallagher, M., Eichenbaum, H. and Tanila, H. (2005) Age-associated alterations of hippocampal place cells are subregion specific. J. Neurosci. 25, 6877–6886 32 Poirier, G.L., Amin, E., Good, M.A. and Aggleton, J.P. (2011) Early-onset dysfunction of retrosplenial cortex precedes overt amyloid plaque formation in Tg2576 mice. Neuroscience 174, 71–83 33 Yassa, M.A., Lacy, J.W., Stark, S.M., Albert, M.S., Gallagher, M. and Stark, C.E. (2010) Pattern separation deficits associated with increased hippocampal CA3 and dentate gyrus activity in nondemented older adults. Hippocampus, doi:10.1002/hipo.20808 34 Yassa, M.A., Stark, S.M., Bakker, A., Albert, M.S., Gallagher, M. and Stark, C.E. (2010) High-resolution structural and functional MRI of hippocampal CA3 and dentate gyrus in patients with amnestic mild cognitive impairment. Neuroimage 51, 1242–1252 35 Tulving, E. (1972) Episodic and semantic memory. In Organization of Memory (Tulving, E. and Donaldson, W., eds), pp. 381–403, Academic Press, New York 36 Clayton, N.S. and Dickinson, A. (1998) Episodic-like memory during cache recovery by scrub jays. Nature 395, 272–274 37 Clayton, N.S., Griffiths, D.P., Emery, N.J. and Dickinson, A. (2001) Elements of episodic-like memory in animals. Philos. Trans. R. Soc. London Ser. B 356, 1483–1491 38 Crystal, J.D. (2010) Episodic-like memory in animals. Behav. Brain Res. 215, 235–243 39 Babb, S.J. and Crystal, J.D. (2006) Episodic-like memory in the rat. Curr. Biol. 16, 1317–1321 40 Crystal, J.D. (2009) Elements of episodic-like memory in animal models. Behav. Proc. 80, 269–277 41 Winters, B.D., Saksida, L.M. and Bussey, T.J. (2008) Object recognition memory: neurobiological mechanisms of encoding, consolidation and retrieval. Neurosci. Biobehav. Rev. 32, 1055–1070 42 Warburton, E.C. and Brown, M.W. (2010) Findings from animals concerning when interactions between perirhinal cortex, hippocampus and medial prefrontal cortex are necessary for recognition memory. Neuropsychologia 48, 2262–2272 43 Dere, E., Huston, J.P. and De Souza Silva, M.A. (2005) Integrated memory for objects, places, and temporal order: evidence for episodic-like memory in mice. Neurobiol. Learn. Mem. 84, 214–221 44 Kart-Teke, E., De Souza Silva, M.A., Huston, J.P. and Dere, E. (2006) Wistar rats show episodic-like memory for unique experiences. Neurobiol. Learn. Mem. 85, 173–182 45 Good, M.A., Barnes, P., Staal, V., McGregor, A. and Honey, R.C. (2007) Context- but not familiarity-dependent forms of object recognition are impaired following excitotoxic hippocampal lesions in rats. Behav. Neurosci. 121, 218–223 46 Good, M.A., Hale, G. and Staal, V. (2007) Impaired ‘episodic-like’ object memory in adult APPswe transgenic mice. Behav. Neurosci. 121, 443–448 47 Good, M.A. and Hale, G. (2007) The ‘Swedish’ mutation of the amyloid precursor protein (APPswe) dissociates components of object-location memory in aged Tg2576 mice. Behav. Neurosci. 121, 1180–1191 48 Rescorla, R.A. and Wagner, A.R. (1972) A theory of Pavlovian conditioning; variations in the effectiveness of reinforcement and nonreinforcement. In Classical Conditioning II: Current Research and Theory (Black, A.H. and Prokasy, W.F., eds), pp. 64–99, Appleton Century Crofts, New York 49 Rescorla, R.A. (1972) ‘Configural’ conditioning in discrete-trial bar pressing. J. Comp. Psychol. 79, 307–317 50 Pearce, J.M. (1994) Similarity and discrimination: a selective review and a connectionist modelPsycholRev. 101, 587–607

Models of Dementia: the Good, the Bad and the Future

51 Rudy, J.W. and Sutherland, R.J. (1989) The hippocampal formation is necessary for rats to learn and remember configural discriminations. Behav. Brain Res. 34, 97–109 52 Rudy, J.W. and Sutherland, R.J. (1995) Configural association theory and the hippocampal formation: an appraisal and reconfiguration. Hippocampus 5, 375–389 53 Coutureau, E., Killcross, S.A., Good, M., Marshall, V.J., Ward-Robinson, J. and Honey, R.C. (2002) Acquired equivalence and distinctiveness of cues: II. Neural manipulations and their implications. J. Exp. Psychol. Anim. Behav. Proc. 28, 388–396 54 Iordanova, M.D., Good, M.A. and Honey, R.C. (2008) Configural learning without reinforcement: integrated memories for correlates of what, where and when. Q. J. Exp. Psychol. 61, 1785–1792

55 Iordanova, M.D., Burnett, D.J., Aggleton, J.P., Good, M. and Honey, R.C. (2009) The role of the hippocampus in mnemonic integration and retrieval: complementary evidence from lesion and inactivation studies. Eur. J. Neurosci. 30, 2177–2189 56 Mueller, S.G., Chao, L.L., Berman, B. and Weiner, M.W. (2011) Evidence for functional specialization of hippocampal subfields detected by MR subfield volumetry on high resolution images at 4T. Neuroimage 56, 851–857

Received 2 May 2011 doi:10.1042/BST0390902

 C The

C 2011 Biochemical Society Authors Journal compilation 

909