Impaired synaptic plasticity and learning in aged amyloid precursor protein transgenic mice

Alzheimer's disease (AD) is identified by progressive deficits in cognitive and emotional function that are associated with extensive forebrain neuropathology. Several mutations affecting -amyloid, either to APP^{1, 2, 3} or to presenilins^{4, 5}, are capable of producing familial forms of AD^{6, 7, 8, 9, 10, 11, 12}. APP₆₉₅SWE mice expressing transgenic human APP with the two-point (Lys⁶⁷⁰Asn, Met ⁶⁷¹Leu) 'Swedish' mutation show age-dependent impairments in several histopathological features, including -amyloid deposits and amyloid plaques, neuritic dystrophy, astrogliosis, reactive microgliosis and abnormal tau phosphorylation^{13, 14, 15}. In additional to structural and neurochemical abnormalities, these aged APP transgenic mice perform poorly in the water maze, a task involving spatial learning, whereas young APP ₆₉₅SWE mice show no such impairment¹³. However, even in behaviorally impaired animals, loss of neurons has been difficult to demonstrate¹⁴. We therefore sought evidence that attributed brain dysfunction to a specific disruption of neuronal function.

Figure 1. Aged APP695SWE mice demonstrate abnormal long-term synaptic plasticity despite otherwise normal synaptic physiology. (a) Input/output curves generated by stimulating the Schaffer collaterals and recording in CA1 stratum radiatum (left) or by stimulating the perforant pathway and recording in stratum moleculare of the dentate gyrus (right) indicate no significant differences in baseline synaptic responses to low-frequency (0.2 Hz) stimulation. Examples of field EPSPs recorded using extracellular electrodes (top) are taken from single slices at stimulus intensities from 30−100 A (in 10 A increments). (b) Following ten minutes of baseline stimulation at 0.067 Hz, tetanic stimulation was delivered to either the Schaffer collateral/commissural pathway (CA1) or to the perforant pathway (dentate gyrus). Theta-burst stimulation was used as tetanus in the Schaffer collaterals; the perforant pathway was tetanized with shorter bursts of higher-frequency stimulation17. Example responses are from 16-month-old control and transgenic mice before (dashed lines) and 50 minutes after tetanic stimulation (solid lines). Although baseline responses were of comparable size, control slices were significantly enhanced in both regions (increase s.e., 43 10%, n = 5 slices from 5 mice for CA1; 28 8%, n = 6/5), whereas aged transgenic slices were not (14 15%, n = 7/6 for CA1; 0 4%, n = 8/4 for dentate gyrus). LTP in CA1 between transgenic mice was no different and controls when tested between two and eight months of age. (c) Paired-pulse facilitation in CA1 did not differ between control and transgenic animals. The fEPSPs illustrated were evoked with an interpulse interval of 30 ms, which on the average produced facilitation of 32 10% for aged controls and 35 3% for aged transgenics. Full Figure and legend (44K)

Figure 2. LTP is impaired in the dentate gyrus in aged APP695SWE transgenic mice in vivo. Following 20 minutes of baseline, theta-burst stimulation17 was delivered to the perforant pathway. Aged control mice demonstrated LTP of both the fEPSP slope (b) and the population spike (c), whereas transgenic mice showed only potentiation of the population spike, which was nonetheless significantly smaller than controls. Representative responses (a) demonstrate the degree of potentiation in transgenic (left) and non-transgenic littermate control (right) brains, and also indicate the comparability of baseline responses between transgenics and controls. Full Figure and legend (38K)

To determine whether the deficits in LTP were related to behavioral and neurochemical abnormalities, we tested APP₆₉₅SWE mice on a spatial working-memory task in the T-maze before doing slice electrophysiology experiments (Fig. 3), and assayed the brains for A concentrations and for amyloid plaques (Fig. 4) after electrophysiological testing. Sixteen- to seventeen-month-old APP ₆₉₅SWE mice and their non-transgenic littermate controls were tested for 12 consecutive days on a forced-choice alternation task in the T-maze. The percentage of correct choices (that is, alternations) were recorded for each daily session. Fifty percent correct responses represented chance levels of performance, and we considered a mouse at criterion for learning if 80% or more of responses were correct over any two consecutive days. A three-way analysis of variance (ANOVA) revealed significant main effects of genotype (p < 0.005), indicating the transgene produces performance impairments, age (p < 0.0001) and training day (p < 0.0001), indicating overall improvement during training. There were also significant interactions of genotype by age (p < 0.05) and genotype by age by training ( p < 0.01), indicating that the performance of old transgenic animals failed to improve, whereas littermate controls and young transgenic and control mice were learning (Fig. 3a). These performance deficits are underscored by the observation (Fig. 3b) that only 33% of aged transgenic mice reached the learning criterion in 12 days of training, compared with at least 75% of aged non-transgenic littermates and both transgenic and non-transgenic young mice.

Figure 3. Behavioral deficits in aged APP695SWE transgenic mice are correlated with deficits in hippocampal synaptic plasticity. (a) Control mice, trained for 12 days (six pairs of trials per day, two days per trial block) on a forced-choice alternation task in the T-maze, increased percent of correct choices until performance was better than 80% correct. Sixteen-month-old APP695SWE transgenic mice failed to improve over the course of training and, by the last three blocks, were significantly impaired relative to littermate controls (*p < 0.05; **p < 0.0001). In contrast, the performance of two month-old transgenics was indistinguishable from littermate controls. There was also no difference between young and aged controls, indicating that age alone does not contribute to the deficit of 16-month-old transgenic mice. (b) The performance differences are highlighted by the inability of 16-month-old transgenic mice to reach the criterion of 80% correct over two sessions. (c) Plots of the magnitude of LTP in CA1 (left) and dentate gyrus (right) against percent correct performance in the T-maze over the last two training sessions reveal a correlation between behavioral performance and neuronal plasticity. Each point represents a single value for each animal that received a full course of training in the T-maze and a full LTP experiment on at least one slice. If more than one LTP experiment was completed per animal (that is, on multiple slices), the results from all slices were averaged. The correlation coefficients displayed on the figure are for transgenic mice alone; the values for all animals are 0.65 for CA1 and 0.85 for dentate gyrus. Full Figure and legend (28K)

Figure 4. A immunostaining with red cy3-labelled biotinylated 3D6 (ref. 19) in a 250 m-thick section used for electrophysiology studies. (a) A deposits in the hippocampal dentate gyrus (dg) and CA. (b) A higher-power view of the field in (a) demonstrates the predilection of the deposits for the outer molecular layer (oml) of the dentate gyrus. iml, inner molecular layer; gc, granule cell layer. Scale bars (a), 250 m; (b), 50 m. Full Figure and legend (26K)

Table 1. Tests of sensorimotor and emotional behavior. Full Table

Elevated concentrations of A and A deposition are key pathological features of both AD and APP₆₉₅SWE mice. We therefore measured A concentrations and determined the prevalence of amyloid plaque deposition to ascertain whether the aged transgenic mice that demonstrated impairments in learning and synaptic plasticity had histopathology comparable to that of human patients with AD. Aged transgenic APP₆₉₅SWE mice ( n = 6) had mean total brain A₁₋₄₀ concentrations of 3,933 pmol/g and a mean total brain A₁₋₄₂ concentration of 941 pmols/g (Table 2). These levels are consistent with previous observations of enhanced A concentrations in aged APP₆₉₅SWE transgenic mice. Moreover, they indicate that these animals have overall A concentrations in brain that are comparable to those of human AD patients³. To draw a further parallel between behaviorally and physiologically impaired mice and human AD, we examined the number and distribution of A deposition in the hippocampal formation of aged APP₆₉₅SWE transgenic mice. Plaque-like accumulations of A were found throughout the hippocampal formation, particularly in the outer molecular layer of the dentate gyrus (Fig. 4). Once again, these results are consistent with previous observations of APP ₆₉₅SWE mice^{13, 14}, and with the general pattern of amyloid pathology in human AD¹⁹.

Table 2. Concentrations of A in transgenic mouse brains. Full Table

	Top

Both spatial working memory and LTP are normal in young transgenic mice at an age when A levels are not significantly elevated. Once A concentrations have risen and deposits have formed, however, both synaptic plasticity and behavioral performance deteriorate in a highly correlated fashion. Hippocampal neurons exhibit a selective deficit in long-lasting plasticity; baseline synaptic transmission and short-term plasticity remain essentially intact at all ages (although we cannot rule out the possibility that there might be differences in paired-pulse facilitation at longer interpulse intervals than were tested). Although there is some evidence of increased excitability to higher-intensity stimulation in some of the CA1 responses in aged transgenic slices (Fig. 1a), this never produced synchronous discharge or even multiple population spikes. This activity was never observed during in-vivo recordings (Fig. 2c), nor was it found at the stimulus intensities used to induce LTP. Finally, behavioral evidence of seizures was never seen in the course of observation of mouse behavior during the mini-neurological exam and the T-maze testing. Taken together, these data support the hypotheses not only that LTP and learning are related, but also that processes initiated or exacerbated by a combination of aging and expression of mutated human APP can affect both plasticity and behavior. Precisely how APP or its cleavage products affect the function of neurons in the brains of APP₆₉₅SWE transgenic mice is unknown. These mice provide an opportunity to address this question directly for the first time.

Several pieces of evidence suggest that structural abnormalities (that is loss of neurons or synapses) are not key factors in the behavioral or physiological plasticity demonstrated by these mice. First, although they present A concentrations and amyloid deposition patterns that are similar to those of human AD brains, the mice demonstrate no significant loss of cells or synaptophysin immunoreactivity in the hippocampus¹⁴. Why neurons are preserved in these transgenic mice and lost in AD brains is unknown; this may be due to species or strain differences in resistance to cell death^{20, 21}. Furthermore, two major glutamate receptor subtypes (GluR1 and NMDAR1) are expressed normally in the hippocampus of aged APP₆₉₅SWE mice (M. Irizarry and B. Hyman, unpublished observations). These observations suggest that the isolated deficits in long-lasting plasticity occurr at levels beyond elementary postsynaptic (glutamate receptors and neuron cell bodies) and presynaptic (synaptophysin) structures.

Fifteen-month-old Tg2576 mice have dramatically elevated concentrations of A (Table 1) and significant A depositions (Fig. 3), whereas three month-old Tg2576 mice have neither. Therefore, because of the ages at which mice were tested in these studies, we cannot easily determine whether the effects we have measured result from elevated concentrations of soluble A, deposited A or both. It is unlikely that amyloid plaques are the direct cause of the LTP deficits, as they only cover 4−6% of the neuropil in 15−16 month-old Tg2576 mice¹⁴, although they could affect a wider area indirectly. Moreover, amyloid-derived diffusible ligands (ADDLs), which are not deposited, can affect both neuronal viability in culture and a form of plasticity induced by high-frequency stimulation in tissue slices²². Although others²³ have suggested that cell death induced by the presence of C-terminal fragments of APP can reduce LTP while increasing thigmotaxis (wall-hugging) in the water maze, this cannot explain our data, because of both the lack of cell loss in Tg2576 mice and the differences in phenotype (that is, spatial memory deficit versus enhanced thigmotaxis; loss of LTP in both dentate and CA1 versus reduction of LTP in CA1). The most reasonable explanation for the data presented here is that the LTP deficits (and, possibly, the behavioral impairments) observed in aged Tg2576 mice result not from neuronal cell death, but from a direct effect of A on one or more mechanism of use-dependent synaptic plasticity.

The data we present here provide some constraints on the nature of these functional abnormalities. Any pathology involving the mechanisms of synaptic transmission must target its ability to be modified by activity, because baseline synaptic responses are normal. Although paired-pulse facilitation is normal, longer-term modifications of presynaptic function that affect neurotransmitter release could be altered²⁴. Alternately, the deficits could involve changes in one of the numerous enzymes, second messengers, transcription factors or cell-adhesion molecules that have been reported to affect plasticity of postsynaptic neurotransmitter receptors without modifying baseline synaptic responses ^{25, 26, 27, 28, 29, 30, 31, 32}. Defining the relationships among APP, A and mechanisms known to contribute to the induction or expression of LTP will therefore be critical to understanding the development of AD.

	Top

All mice studied were Tg(HuAPP₆₉₅SWE)2576 in a hybrid background of C57Bl/6j with SJL. On each individual test, transgenic mice were always compared to littermate controls so that age and background strain were comparable. There were, however, small generational differences in the relative percentage of each of these background strains. The founder of the Tg2576 line was a C57Bl/6j SJL F3, which was subsequently crossed twice into C57Bl/6j. Subsequent generations were then crossed to C57Bl/6j SJL F1. Thus, in all cases, the average contribution of C57Bl/6j ranged from 59% to 88%, with the remainder from SJL.

400-m hippocampal slices were prepared using standard methods and media and maintained in a submersion chamber at 32 0.5°C. Dissection was done at 0−2°C in standard recording medium (119 mM NaCl, 2.5 mM KCl, 1.3 mM MgSO₄, 1.0 mM NaH₂PO₄, 26.2 mM NaHCO₃, 2.5 mM CaCl ₂ and 11 mM glucose) also containing 1 mM kynurenic acid to protect neurons from the possible harmful effects of slicing. Slices were incubated in this solution for at least 1 hour while slowly coming to room temperature before being transferred one at a time to the recording chamber where they were superfused with drug-free recording medium. The experimenter was blind to the animal's genotype throughout the experiments. Extracellular field potentials were recorded in stratum radiatum of the CA1 region of hippocampus or in the stratum moleculare of the dentate gyrus using carbon fiber electrodes in response to stimulation of the Schaffer collateral/commissural pathway (for CA1) or the perforant pathway (for dentate gyrus) with monopolar electrodes passing constant current. For LTP experiments, baseline stimulation was delivered at an intensity (typically 30−40 A) that evoked a response approximately 25−30% of the maximum evoked response. LTP was induced using theta-burst stimulation in both CA1 and the dentate gyrus. In CA1 the bursts consisted of 4 pulses at 100 Hz, with the bursts repeated at 5 Hz and each tetanus including 3 ten-burst trains separated by 15 seconds. In the dentate gyrus, the bursts were 6 pulses delivered at 400 Hz, and 6 trains of bursts were delivered 6 times, with 20 seconds between trains¹⁷. LTP in dentate gyrus was always attempted in the presence of 10 M bicuculline to block GABA _A-mediated inhibition. These parameters were chosen because they produce robust LTP both in vivo and in vitro. This therefore provides the opportunity to distinguish between a block of LTP and a change in the LTP induction threshold, while using a protocol that mimics important patterns of activation observed in the hippocampus in vivo.

Measuring -amyloid concentration.

Brain tissue was sequentially extracted first in Tris-buffered saline (TBS) containing protease inhibitors (1 tablet Complete [Boehringer Ingelheim]/50 ml TBS [137 mM NaCl, 20 mM Tris base, pH 7.6]), and then in 1% Triton X-100 (TBS with 1% Triton X-100), 2% SDS in water and 70% formic acid in water. Brain tissue was sonicated initially in TBS at a concentration of 150 mg wet weight per ml, and pellets were prepared for each round of sequential extraction by centrifuging at 100,000 g for 1 h at 4°C. The TBS, Triton X-100, SDS and formic acid extracts were diluted in EC buffer³ at dilutions of 1:20, 1:40, 1:2000, and 1:2000 respectively before the A assay. A ₁₋₄₀ and A₁₋₄₂ were analyzed by sandwich ELISA as described³. The total values for A₁₋₄₀ and A₁₋₄₂ shown in Table 1 were calculated by summing the TBS, Triton X-100, SDS and formic acid fractions. Virtually all of the A₁₋₄₀ and A₁₋₄₂ in these brains was found in the SDS or formic acid fractions.

-amyloid staining.

After electrophysiology studies, hippocampal slices were fixed in 4% paraformaldehyde for 48 hours, maintained in 15% glycerol in TBS, permeabilized with 0.5% Triton X-100 in TBS for 1 h, blocked with 3% milk, then sequentially probed with bi-3D6 (anti-A) 1:750 overnight and cy3-streptavidin 1:750 before dehydration in graded alcohols and mounting with permount.

	Top

1. Citron, M. et al. Mutation of the beta-amyloid precursor protein in familial Alzheimer's disease increases beta-protein production. Nature 360, 672−674 (1992). | Article | PubMed | ISI | ChemPort |
2. Cai, X., Golde, T. E. & Younkin, S. G. Release of excess amyloid protein from a mutant amyloid protein precursor. Science 259, 514−516 (1993). | PubMed | ISI | ChemPort |
3. Suzuki, N. et al. An increased percentage of long amyloid beta protein secreted by familial amyloid beta protein precursor (beta APP717) mutants. Science 264, 1335−1340 (1994).
4. Borchelt, D. R. et al. Familial Alzheimer's disease-linked presenilin 1 variants elevate A1−42/1−40 ration in vitro and in vivo. Neuron 17, 1005−1013 (1996). | Article | PubMed | ISI | ChemPort |
5. Duff, K. et al. Increased amyloid-beta 42(43) in brains of mice expressing presenilin 1. Nature 383, 710−713 (1996). | Article | PubMed | ISI | ChemPort |
6. Goate, A. M. et al. Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer's disease. Nature 349, 704−706 (1991). | Article | PubMed | ISI | ChemPort |
7. Mullan, M. et al. A pathogenic mutation for probable Alzheimer's disease in the APP gene at the N-terminus of -amyloid. Nat. Genet. 1, 345−347 (1992). | Article | PubMed | ISI | ChemPort |
8. Chartier-Harlin, M. C. et al. Early-onset Alzheimer's disease caused by mutation at codon 717 of the -amyloid precursor protein gene. Nature 353, 844−846 (1991). | Article | PubMed | ChemPort |
9. Murrell, J., Farlow, M., Getti, B. & Benson, M. D. A mutation in the amyloid precursor protein associated with hereditary Alzheimer's disease. Science 254, 97−99 (1991). | PubMed | ISI | ChemPort |
10. Hendriks, L. et al. Presenile dementia and cerebral haemorrhage linked to a mutation at codon 692 of the -amyloid precursor protein gene. Nat. Genet. 1, 218−221 (1992). | Article | PubMed | ISI | ChemPort |
11. Eckman, C. B. et al. A new pathogenic mutation in the APP gene (I716V) increases the relative proportion of A-beta 42(43). Hum. Mol. Genet. 6, 2087−2089 (1997). | Article | PubMed | ISI | ChemPort |
12. Hardy, J. Amyloid, the presenilins and Alzheimer's disease. Trends Neurosci. 20, 154−159 (1997). | Article | PubMed | ISI | ChemPort |
13. Hsiao, K. K. et al. Correlative memory deficits, A elevation and amyloid plaques in transgenic mice. Science 274, 99−102 (1996). | Article | PubMed | ISI | ChemPort |
14. Irizarry, M. C., McNamara, M., Fedorchak, K., Hsiao, K. K. & Hyman, B. T. APP_SW transgenic mice develop age-related A deposits and neuropil abnormalities, but no neuronal loss in CA1. J. Neuropathol. Exp. Neurol. 56, 965−973 (1997). | PubMed | ISI | ChemPort |
15. Frautschy, S. A. et al. Microglial response to amyloid plaques in APP_SW transgenic mice. Am. J. Pathol. 152, 307−317 (1998). | PubMed | ISI | ChemPort |
16. Bliss, T. V. P. & Collingridge, G. L. A synaptic model of memory—long-term potentiation in the hippocampus. Nature 361, 31−39 (1993). | Article | PubMed | ISI | ChemPort |
17. Errington, M. L., Bliss, T. V. P., Morris, R. J., Laroche, S. & Davis, S. Long-term potentiation in awake mutant mice. Nature 387, 666−667 (1997). | Article | PubMed | ISI | ChemPort |
18. Zucker, R. S. Short-term synaptic plasticity. Annu. Rev. Neurosci. 12, 13−31 (1989). | Article | PubMed | ISI | ChemPort |
19. Hyman, B. T., Van Hoesen, G. W., Kromer, L. J. & Damasio, A. R. Perforant pathway change and the memory impairment of Alzheimer's disease. Ann. Neurol. 20, 473−482 (1986).
20. Finch, C. E. Neuron atrophy during aging: programmed or sporadic? Trends Neurosci. 16, 104−110 (1993). | Article | PubMed | ISI | ChemPort |
21. Schauwecker, P. E. & Steward, O. Genetic determinants of susceptibility to excitotoxic cell death: implications for gene targeting approaches. Proc. Natl. Acad. Sci. USA 94, 4103−4108 (1997). | Article | PubMed | ChemPort |
22. Lambert, M. P. et al. Diffusible, nonfibrillar ligands derived from A₁₋₄₂ are potent central nervous system neurotoxins. Proc. Natl. Acad. Sci. USA 95, 6448−6453 (1998). | Article | PubMed | ChemPort |
23. Nalbantoglu, J. et al. Impaired learning and LTP in mice expressing the carboxy terminus of the Alzheimer amyloid precursor protein. Nature 387, 500−505 (1997). | Article | PubMed | ISI | ChemPort |
24. Südhof, T. C. The synaptic vesicle cycle: a cascade of protein−protein interaction. Nature 375, 645−653 (1995). | Article | PubMed | ISI | ChemPort |
25. Luthi, A., Laurent, J. P., Figurov, A., Muller, D. & Schachner, M. Hippocampal long-term potentiation and neural cell adhesion molecules L1 and NCAM. Nature 372, 777−779 (1994). | Article | PubMed | ISI | ChemPort |
26. Dragunow, M. A role for immediate-early transcription factors in learning and memory. Behav. Genet. 26, 293−299 (1996). | PubMed | ISI | ChemPort |
27. Ikezu, T. et al. Negative transactivation of cAMP response element by familial Alzheimer's mutants of APP. EMBO J. 15, 2468−2475 (1996). | PubMed | ISI | ChemPort |
28. Castillo, P. E. et al. Rab3A is essential for mossy fibre long-term potentiation in the hippocampus. Nature 388, 590−593 (1997). | Article | PubMed | ISI | ChemPort |
29. Kang, H., Welcher, A. A., Shelton, D. & Schuman, E. M. Neurotrophins and time: different roles for TrkB signaling in hippocampal long-term potentiation. Neuron 19, 653−664 (1997). | Article | PubMed | ISI | ChemPort |
30. Giese, K. P., Federov, N. B., Filipkowski, R. K. & Silva, A. J. Autophosphorylation at threonine 286 of the calcium-calmodulin-kinase II in LTP and learning. Science 279, 870 (1998). | Article | PubMed | ISI | ChemPort |
31. Klann, E., Chen, S. J. & Sweatt, J. D. Persistent protein kinase activation in the maintenance phase of long-term potentiation. J. Biol. Chem. 266, 24253−24256 (1991). | PubMed | ISI | ChemPort |
32. Winder, D. G., Mansuy, I. M., Osman, M., Moallem, T. M. & Kandel, E. R. Genetic and pharmacological evidence for a novel, intermediate phase of long-term potentiation suppressed by calcineurin. Cell 92, 25−37 (1998). | Article | PubMed | ISI | ChemPort |

	Top