The β-amyloid peptide (Aβ) is thought to play a critical role in the pathophysiology of Alzheimer's disease (AD). To study the effects of Aβ on the brain, transgenic mouse models have been developed that express high levels of Aβ. These mice show some features of AD, including amyloid plaques and mild cognitive impairment, but not others such as progressive neurodegeneration. We investigated the age-dependent effects of Aβ on synaptic physiology in Tg2576 mice that express human Aβ. We report that both basal synaptic activity and long-term potentiation (LTP), as measured in the CA1 region of the hippocampus, were compromised by 7 months of age before plaque deposition. Despite a persistent increase in Aβ levels with age, LTP recovered in 14-month-old mice, with no further loss of basal activity compared with activity measured in 7-month-old mice. Previous work has shown that inhibitors of γ-secretase, an enzyme critical for Aβ synthesis, can significantly reduce Aβ production and plaque formation in Tg2576 mice. Our data demonstrate that 7-month-old Tg2576 mice treated with an orally available γ-secretase inhibitor showed a significant improvement in synaptic function and plasticity within days, and the effect was correlated with the extent and duration of Aβ reduction. These results indicate that recovery from Aβ-mediated synaptotoxicity can occur rapidly with Aβ-lowering therapies. These findings highlight some of the strengths and limitations of using Aβ-overexpressing mouse models for Alzheimer's drug discovery.
Alzheimer's disease (AD) is characterized clinically by progressive loss of memory and the pathological accumulation of neurofibrillary tangles and amyloid plaques in the brain. Tangles are composed of a fibrillized form of the microtubule-binding protein tau. Amyloid lesions contain β-amyloid peptide (Aβ) produced endogenously by the proteolytic cleavage of the amyloid precursor protein (APP) by β-secretase and γ-secretase (Wolfe, 2006; Cole and Vassar, 2008). Mutations near the cleavage sites of APP cause early onset AD as do variants in the γ-secretase subunit presenilin 1. These mutations typically elevate Aβ production or increase Aβ42/40 ratios, which result in a more aggregation-prone form of the peptide. Essential to developing disease-modifying therapies for AD is the identification and characterization of relevant preclinical animal models.
Transgenic mice that express human disease variants of APP or presenilin 1 develop amyloid plaques, memory impairment, deficits in synaptic function, gliosis, and dystrophic neurites (Ashe, 2006; Gotz and Ittner, 2008; Morrissette et al., 2009). However, despite high levels of Aβ expression, these mice typically lack the overt tau pathology, neurodegeneration, and show only modest cognitive impairment. The full value of these preclinical models of Alzheimer's disease requires an explanation for the incomplete pathology. Tg2576 mice that express the APPsw mutation (K770N, M671L) have been widely studied (Hsiao et al., 1996; Ashe, 2006) as a model for plaque deposition and Aβ toxicity. These mice develop diffuse plaques after approximately 8 months and compact plaques after approximately 12 to 13 months (Kawarabayashi et al., 2001). Although investigators continue to debate the precise forms of Aβ that contribute to neuronal dysfunction (Walsh and Selkoe, 2007; Viola et al., 2008; Wasling et al., 2009), considerable evidence indicates that soluble oligomeric Aβ is particularly toxic, with larger insoluble fibrils and plaques possibly being neuroprotective (Cheng et al., 2007).
Results from numerous studies indicate that hippocampal long-term potentiation (LTP), a measure of synaptic plasticity that is considered a cellular correlate of memory, is vulnerable to acute application of oligomeric Aβ (Townsend et al., 2006; Lacor et al., 2007; Klyubin et al., 2008). Likewise, Tg2576 mice show synaptic dysfunction and behavioral changes before significant plaque formation, a finding that further implicates soluble Aβ (Chapman et al., 1999; Fitzjohn et al., 2001; Jacobsen et al., 2006). However, the precise mechanisms by which chronically elevated Aβ affects synaptic transmission and plasticity are not clear. We used hippocampal slice recordings to compare basal synaptic activity and LTP in young (3–4 months old), middle-aged (6–7 months, preplaque), and old (14–15 months, plaque) Tg2576 mice. Remarkably, LTP deficits were observed only in 6- to 7-month-old mice but not in young or old mice. Basal activity was also depressed by 6 to 7 months and persisted (but did not deteriorate further) in older mice. The deficit in LTP in 6- to 7-month-old mice was reversed within 3 days of oral dosing with a γ-secretase inhibitor, whereas basal synaptic activity improved after 7 days of treatment. We conclude that there is a transition in Aβ-induced synaptic deficits in Tg2576 mice because plaque deposition begins and pharmacological intervention to reduce Aβ at this stage can rapidly restore function.
Materials and Methods
Male wild-type (WT) and Tg2576 mice were acquired from Taconic Farms (Hudson, NY) and housed individually. Tg2576 mice overexpress human APP with the Swedish K670N, M671L mutation on a B6 background. For oral dosing of the γ-secretase inhibitor MRK-560 (Best et al., 2007), the compound was formulated in 0.5% methylcellulose. Body weight was monitored daily, and no adverse compound-related side effects were observed.
Animals were euthanized with isoflurane according to protocols approved by the Institutional Animal Care and Use Committee. Brains were rapidly dissected and placed in oxygenated, 4°C artificial cerebral spinal fluid (ACSF) containing 2.8 mM KCl, 1 mM MgCl2, 2 mM MgSO4, 1.25 mM NaH2PO4, 26 mM NaHCO3, 10 mM d-glucose, 206 mM sucrose, and 0.4 mM sodium ascorbate (pH 7.4, osmolarity 295) (Moyer and Brown, 1998). Brains were sectioned on a Leica VT 1000S Vibratome (Leica, Wetzlar, Germany) into 350-μm coronal slices. These sections were allowed to recover in normal ACSF containing 124 mM NaCl, 2.8 mM KCl, 2 mM MgSO4, 1.25 mM NaH2PO4, 26 mM NaHCO3, 10 mM d-glucose, and 0.4 mM sodium ascorbate (pH 7.4, osmolarity 295) for 2 h (unless indicated otherwise) at room temperature and continuously aerated with 95% oxygen. For experiments with kynurenic acid, a 1 mM solution was prepared in double-distilled H2O and continuously heated and stirred until dissolved. The solution was then cooled to room temperature before the remaining salts were added. Neither GABA nor glycine receptor antagonists were used in any experiments. All reagents for ACSF solutions were purchased from Sigma-Aldrich (St. Louis, MO).
Slices were immersed in slice chambers (Harvard Apparatus Inc., Holliston, MA) and visualized with a Zeiss Stereo Discovery V8 dissecting microscope (Zeiss Inc., Oberkochen, Germany) mounted on a TMC Micro-g air table (TMC, Peabody, MA). Solution exchange in the chambers was controlled by a Valve Bank 4 solution handler (Automate Scientific, Berkeley, CA).
Field potential recordings (Sarvey et al., 1989) were obtained using a CyberAmp 380 and individual AI401 10x preamplifiers. Data were digitized with a Digidata 1440A and analyzed with pClamp software (Clampex and Clampfit) (Molecular Devices, Sunnyvale, CA). The data were sampled at 10 kHz and filtered at 2 kHz. An A360 stimulus isolator was used to deliver electrical stimuli to a concentric tungsten electrode (World Precision Instruments, Inc. Sarasota, FL) placed in the Schaeffer collateral fiber tract. Excitatory postsynaptic potentials (EPSPs) were recorded using a glass electrode (2–4 MΩ) placed approximately 300 to 500 μm distally in the stratum radiatum filled with ACSF. Test stimuli were delivered every 20 s (0.05 Hz) at an intensity of 10 to 20 microamps, which elicited an EPSP that was 20 to 30% of the maximal field potential response. LTP was induced by two high-frequency stimulations (100 Hz, 1 s) spaced 5 min apart. The slope of the field potential was estimated using approximately 10 to 80% of the total response. A stable baseline was established for ≥20 min before experiments were initiated. Individual slices were excluded from analysis if there was a disruption in the adequate perfusion of the brain tissue. Individual slices were also excluded from analysis if the baseline shifted by more than 25%, if the standard deviation of the EPSP over the last 10 min of the recording was >50%, and if the minimal EPSP slope recorded was <30% baseline.
Brain slices were lysed in buffer containing 50 mM NaCl, 0.2% diethylamine (DEA), and Complete protease inhibitors (Roche, Nutley, NJ) at a ratio of 1 ml:100 mg with use of a T8 homogenizer (IKA Works, Inc., Wilmington, NC). Samples were boiled and centrifuged in siliconized microcentrifuge tubes. The supernatants were neutralized with 0.5 M Tris-HCl, pH 6.8. MesoScale 96-well plates were coated with Biotin-4G8 and detected with ruthenium-labeled 12F4 (Aβ42) or G210 (Aβ40). Plate preparation and detection were performed according to the manufacturer's recommended protocol (MesoScale Discovery, Gaithersburg, MD). Antibodies were from Signet/Covance (Princeton, NJ).
Five coronal brain sections per Tg2576 mouse were continuously perfused in aerated ACSF for 30 min. The ACSF was collected and treated with or without 0.25 mM BS3 protein cross-linking reagent (Pierce ThermoScientific Rockford, IL) at 4°C for 30 min while nutating. The cross-linking agent was inactivated by the addition of Tris-HCl (pH 7.4) to a final concentration of 20 mM. Protease inhibitors were then added (Roche Complete protease inhibitors and 1 mM 1,10-phenanthroline). The collected medium was immunoprecipitated with a proprietary antibody directed toward the midregion of Aβ (amino acids 13–28) and protein G-Sepharose overnight. Beads were washed three times in 50 mM high-salt buffer and eluted with sodium dodecyl sulfate sample buffer containing 1 μM dithiothreitol. Samples were heated to 65°C before being loaded onto a 10 to 20% Tricine gel (Invitrogen, Carlsbad, CA) probed with the anti-Aβ antibody 6E10 (Signet/Covance). Blots were detected using a Licor Odyssey system (Lincoln, NE).
Analyses were carried out using Prism Statistical software (GraphPad Software Inc., La Jolla, CA). For input/output curves, the absolute value of the peak fiber-volley and excitatory postsynaptic potential were measured using Clampfit software (Molecular Devices, Sunnyvale, CA). The exponential of each value was then calculated to avoid small numbers in the denominator (exp input/exp output).
Despite the extensive deposition of amyloid plaques in aged Tg2576 mice, only modest deficits have been observed in synaptic plasticity and cognition (Ashe, 2006). To understand the interaction between the age-dependent increase in Aβ and its effects on synaptic plasticity, we compared the LTP with changes in soluble Aβ40 and Aβ42 in ex vivo brain slices from Tg2576 mice at three different stages of maturity (Fig. 1). Consistent with previous reports, we observed an increase in both soluble Aβ40 and Aβ42 between 3 to 4 months, 6 to 7 months, and 14 to 15 months of age, as measured by ELISA (Fig. 1A). These ages were selected to reflect distinct stages in amyloid accumulation and deposition.
LTP was measured in the CA1 region of hippocampal brain slices made from WT and Tg2576 littermates. At 3 to 4 months of age, the LTP was not significantly different between genotypes (Fig. 1, B and C). Soluble Aβ increased by approximately 50% at 6 to 7 months of age relative to 3 to 4 months of age, but LTP was severely impaired in Tg2576 compared with WT (Fig. 1, D and E). By 14 to 15 months of age, soluble Aβ increased by approximately 1000% relative to 3 to 4 months of age. Remarkably, LTP in aged mice had improved compared with younger mice and was not significantly different from that of age-matched WT mice. These data demonstrate that before plaque deposition (diffuse plaques began at approximately 8 months, and compact plaques began at approximately 12 to 13 months), soluble Aβ is inversely correlated with LTP. However, after plaque deposition began, soluble Aβ (as measured by ELISA) no longer predicted the severity of the LTP deficit.
As an independent measure of synaptic efficacy, input/output (I/O) (fiber volley/excitatory postsynaptic potential) data were collected on the same brain sections (Fig. 2). The I/O ratio reflects the number of active axons required to elicit a given postsynaptic response and is a sensitive measure of synaptic connectivity. At 3 to 4 months of age, the I/O ratio was not significantly different between WT and Tg2576. By 6 to 7 months, the I/O ratio in Tg2576 was significantly increased, indicating that a stronger stimulus was required to elicit a postsynaptic response of the same size. The I/O ratio in Tg2576 at 14 to 15 months remained elevated compared with WT but did not undergo further change compared with Tg2576 at 6 to 7 months of age. These results demonstrate that a measure of synaptic efficacy (I/O) exhibits a deficit in Tg2576 mice as Aβ levels increase with age. However, the deficit appears to plateau by 6 to 7 months and does not become more severe in older animals (Fig. 2).
Because soluble Aβ levels may fluctuate during the preparation and subsequent perfusion of brain slices, over the course of an experiment we measured Aβ40 and Aβ42 in brain slices of 6- to 7-month-old Tg2576. Figure 3A shows that the Aβ levels changed significantly during the brain-slice preparation and during the subsequent recovery period. Compared with brain tissue that was snap-frozen, brain slices that were frozen shortly after sectioning on a Vibratome (only approximately 4 min after decapitation) had a near doubling of soluble Aβ. This result may reflect an increase in both acute proteolysis of APP and activity-dependent exocytosis of Aβ. As the slices were perfused in the recovery chambers, soluble Aβ dissipated to preslicing levels by 1 h and stabilized thereafter.
To determine whether the washout of Aβ affects LTP, recordings were started 1, 3, or 5 h after brain sectioning (Fig. 3, B and C). After a 20-min baseline, the slices were stimulated, and the EPSPs were tracked for 60 min (see Materials and Methods). The LTP in Tg2576 showed time-dependent improvement, with the greatest deficit seen in recordings starting 1 h after slicing and a nearly complete recovery in LTP in recordings starting 5 h after sectioning. In contrast, WT brain slices showed the strongest LTP at shorter time intervals and a progressive loss of LTP with longer recovery periods. The latter observation is consistent with the slow loss of viability in ex vivo slices over time. Thus, the LTP deficit seen in 6- to 7-month-old Tg2576 mice was variable depending on the period the slices were perfused. With a slight delay, the recovery of LTP is inversely correlated with return of total Aβ to resting levels. The specific source and aggregation state of the different Aβ species known to exist were not determined. Because of this finding, all experiments in Tg2576 mice were initiated 2 h after slice preparation (including those shown in Fig. 1) when the deficit was reproducibly significant.
To understand whether the deficits in LTP and I/O ratio were attributable to Aβ, 6- to 7-month-old mice were dosed with a potent γ-secretase inhibitor, N-[cis-4-[(4-chlorophenyl)sulfonyl]-4-(2,5-difluorophenyl)cyclohexyl]-1,1,1-trifluoromethanesulfonamide (MRK-560), for 1, 3, or 7 days. Previous work had shown that this compound is well tolerated in mice (Best et al., 2007). MRK-560 shows maximal Aβ lowering at approximately 16 h after dosing and a brain/plasma ratio of 0.3. Animals were sacrificed approximately 16 h after the last dose. Soluble Aβ was collected from brain slices that were snap-frozen immediately after sectioning and measured by ELISA (Fig. 4). Aβ40 was reduced by 67% (approximately 33% remaining) after 1 day of dosing. MRK-560 was slightly less effective at lowering Aβ42 (approximately 48% remaining). After 3 days of dosing with MRK-560, Aβ40 was reduced by 73%, and Aβ levels partially recovered when mice were drug-free for either 1 or 2 days after the third dose. With 7 days of dosing at 30 mg/kg, Aβ40 levels were reduced by 81%, and 1 mg/kg was sufficient to reduce Aβ40 by 64%. These data demonstrate that MRK-560 is effective at modulating Aβ levels in Tg2576 mice.
In the same mice in which we measured soluble Aβ levels, we performed electrophysiology experiments to determine basal synaptic activity (I/O ratio) and LTP. As expected, the LTP in vehicle-treated Tg2576 mice was impaired to a similar extent as in untreated animals, and this was significantly different from WT. After 1 day of dosing with 30 mg/kg MRK-560, there was a nominal improvement in the LTP. The recovery of LTP was significant after 3 days of dosing. With 3 days of dosing followed by either 1 or 2 days of washout, the improvement in LTP was partially maintained compared with untreated controls. A week of dosing at 30 mg/kg resulted in a significant recovery in LTP, whereas the 1 mg/kg dose showed an intermediate effect. These results demonstrate that LTP in Tg2576 brain slices is highly correlated with and is rapidly modified by ambient Aβ levels.
The I/O ratio was also affected by Aβ-lowering treatments (Fig. 5B). Vehicle-treated Tg2576 mice showed I/O ratios similar to those in untreated mice and this ratio was significantly elevated compared with WT littermates. After 3 days of treatment with 30 mg/kg MRK-560, there was no change in the I/O ratio. With 3 days of dosing followed by either 1 or 2 days without treatment, there was a further increase in the I/O ratio of Tg2576 mice that fell short of significance. Animals that had been dosed for 7 days with 1 mg/kg showed little change in the I/O ratio, whereas mice that received 30 mg/kg showed a decrease in the I/O ratio, which approached significance. These data demonstrate that the I/O ratio is also affected by Aβ. Although not having achieved statistical significance, these results suggest that the recovery of Aβ after cessation of treatment may further suppress basal synaptic activity, whereas sustained Aβ lowering shows a trend toward recovery to WT levels. The latter finding is consistent with a partial recovery of synaptic connectivity and demonstrates dissociation between the effects of Aβ on synaptic plasticity versus basal synaptic transmission.
For each mouse in the study, we plotted the LTP versus Aβ load (Fig. 5C). It is clear that LTP was consistently improved in mice in which Aβ levels were lowered by >65%. Thus, the deficits in LTP observed in Tg2576 mice are likely caused by Aβ rather than the soluble APP ectodomain. These results support the notion that Aβ-lowering strategies (at least in preplaque mice) can lead to recovery of this form of synaptic plasticity.
To determine whether aged animals would also benefit from γ-secretase inhibition, Tg2576 mice were treated with daily dosing of 2.5 mg/kg MRK-560 between the ages of 12 and 15 months. This dose was chosen to ensure that the mice could tolerate the compound for the duration of the study. Because LTP at 15 months was unperturbed, the analysis focused on I/O ratios, which continued to be elevated in Tg2576 mice at this age. Figure 6A demonstrates that soluble Aβ40 was effectively lowered by MRK-560 by 63% (37% remaining) as was Aβ42. The I/O ratio of vehicle-treated Tg2576 mice was not significantly different from untreated, age-matched Tg2576 mice. In mice that received MRK-560, I/O ratio showed no improvement after 3 months of dosing and continued to be significantly different from that of age-matched WT.
Because previous electrophysiology studies in Tg2576 mice typically use the ionotropic receptor antagonist kynurenic acid during the slice preparation, we performed similar experiments with this inhibitor in 6- to 7-month-old animals. As shown in Fig. 7A, 1 mM kynurenic acid in the cutting solution (used during brain-slice preparation) resulted in a small improvement in the LTP, although this effect did not reach significance. Kynurenic acid had no effect on the I/O ratios of Tg2576 mice (WT + average dose of kynurenic acid = 0.87 ± 0.10, n = 3 mice, 8 slices; Tg2576 + average dose of kynurenic acid = 1.66 ± 0.23, n = 5 mice, 24 slices; data not shown).
Kynurenic acid can reduce excitotoxicity during the Vibratome sectioning by blocking ionotropic receptors and potentially reducing synaptic vesicle release. Therefore, we reasoned that Aβ levels in the brain slices may also be affected. We observed a small reduction in soluble Aβ in slices treated with kynurenic acid during the brain sectioning (Fig. 7B), although the effect was not significant and the relative change was similar. Thus, kynurenic acid may reduce but not eliminate the impairment in LTP or I/O ratio observed in 6- to 7-month-old Tg2576 mice.
These results demonstrate that despite the exponential increase in soluble Aβ and deposition of plaques in aged Tg2576 mice, the deficits in LTP and I/O ratio are most pronounced at 6 to 7 months. Because a γ-secretase inhibitor alleviates both these deficits, they are likely to be attributable to Aβ. Moreover, because the deficit in LTP is mitigated by prolonged washout of Aβ from the slices, we reasoned that a synaptotoxic form of Aβ may be removed from these brain slices. To test this theory, we perfused Tg2576 brain slices for 30 min in the recovery chambers. The ACSF was then collected and either treated with the cross-linker bis(sulfosuccinimidyl)suberate, which reacts via primary amines, or left untreated. The collected medium was immunoprecipitated with a proprietary monoclonal antibody to Aβ, run by SDS-PAGE, and probed with the anti-Aβ antibody 6E10. We compared brain slice samples from 6- to 7-month-old mice and 14- to 15-month-old mice because the former showed LTP deficits, whereas the latter did not.
Distinct immunoreactive bands at approximately 18 and 30 kDa were visible in the cross-linked samples (Fig. 8). A simple ladder of bands increasing in 4-kDa increments was not observed, as might be expected if the cross-linking agent was inducing stepwise oligomerization. Higher molecular weight bands (such as Aβ*56) were difficult to detect because of contamination with antibody protein. Surprisingly, there was essentially no monomer detected in these samples. Lysates prepared from slices that were snap-frozen at the time of sectioning showed a distinct Aβ pattern that consisted primarily of approximately 4- and 8-kDa bands. In contrast, the uncross-linked samples showed a more prominent monomeric Aβ band with a weaker signal at approximately 14 and 18 kDa. Because immunoprecipitation is not a quantitative technique, it is not possible to conclude that certain Aβ species are found only in 6- to 7-month-old Tg2576 mice. Nevertheless, this approach may help to determine whether specific Aβ species are correlated with deficits in LTP and I/O ratio.
We investigated changes in synaptic activity in hippocampal brain slices isolated from Tg2576 (APPsw) transgenic mice (one of the best studied mouse models of AD) at three stages of Aβ accumulation. These mice produce toxic forms of Aβ, including soluble oligomers, fibrils, and plaques that result in glial proliferation, decreased synapse density, and behavioral changes (Ashe, 2006). We demonstrated that, at 7 months of age, these mice have a reduced capacity for LTP, which corresponds to the age at which they first display deficits in hippocampal memory tests (Westerman et al., 2002). We also found that the slice preparation itself caused a transient release of soluble Aβ. This Aβ may contribute to the profound LTP deficits in 6- to 7-month-old mice compared with the relatively modest changes in learning in these mice. In addition to the impairment in LTP, a decrease in basal synaptic activity (I/O) became profound at 7 months. Unlike LTP that recovered in aged mice, this synaptic deficit persisted in transgenic mice through 15 months of age, with no change in severity. We also found that treatment with a γ-secretase inhibitor was sufficient to reverse the LTP deficit in 6- to 7-month-old mice.
These experiments address some previously unresolved issues reported in the literature. In their original report, Chapman et al. (1999) showed that Tg2576 mice have normal fast synaptic transmission but impaired LTP in both the CA1 region and dentate gyrus of the hippocampus in 15- to 17-month-old but not in 2- to 8-month-old mice. Our experiments detected a deficit in CA1 hippocampal by 6 to 7 months that was not apparent at 4 months. However, this deficit was sensitive to factors such as slice recovery time and the use of kynurenic acid. These variables may account for why Jacobsen et al. (2006) reported impairments in the dentate gyrus as early as 4 months. We unexpectedly observed a restoration of LTP in 12- to 15-month-old mice despite a nearly 10-fold increase in soluble Aβ. Similar to our findings, Fitzjohn et al. (2001) reported normal LTP in similarly aged mice and a comparable impairment in basal synaptic activity.
Modeling the early stages of familial Alzheimer's disease, Tg2576 mice exhibit Aβ-induced synaptic dysfunction before plaque deposition. As the mice continue to age, plaque deposition proceeds aggressively. However, it is not clear that behavioral deficits are progressive beyond what can be attributed to aging (Middei et al., 2006). Consistent with this, we observed deficits in synaptic function before Aβ deposition that either corrected or stabilized as the animals aged and plaque formation began. Two potential explanations for this finding include a change in the toxic forms of Aβ at this age (Lesné et al., 2008) or an improved ability to compensate with age, perhaps due to age-dependent changes in the molecular mechanisms of LTP (Kumar et al., 2007; Gruart et al., 2008).
Several other mouse models that contain distinct APP or PS1 mutations and promote excessive Aβ production show impairments in learning and synaptic function (Ashe, 2005; Gotz and Ittner, 2008; Morrissette et al., 2009) similar to those observed in the Tg2576 model. For instance, Dewachter et al. (2002) reported that APP[V717] transgenic mice had deficits in LTP by 10 months that were dependent on PS1 activity. These mice displayed hyperexcitability resulting in hidden seizure activity, decreased LTP, and decreased basal activity in the CA1 region (Palop et al., 2007). Nevertheless, other lines of transgenic mice are resilient to memory and synaptic deficits despite high levels of Aβ and plaque deposition (Saganich et al., 2006, Van Vickle et al., 2007; Gruart et al., 2008), which raises the possibility that the form of Aβ (monomer, oligomer, proto-fibril, etc.) may be important for toxicity and behavioral phenotype (Cheng et al., 2007). Our direct measurements of soluble Aβ in slices indicate that LTP deficits correlate well with soluble Aβ in 6- to 7-month-old mice but not in aged mice. As shown by our Western blot analysis, at different ages soluble Aβ exists in various forms, only some of which may be toxic to synapses (Westerman et al., 2002, Lesne et al., 2008).
γ-Secretase is a critical enzyme for regulated intramembrane cleavage of many membrane proteins and may be a tractable target for Aβ-lowering therapies for AD (Marjaux et al., 2004; Wolfe, 2006). Not surprisingly, the γ-secretase inhibitor MRK-560 lowers Aβ in both transgenic mouse models and normal rats (Best et al., 2006). Chronic treatment with the γ-secretase inhibitor MRK-560 for 3 months was previously shown to lower insoluble Aβ and reduce total plaque area in Tg2576 mice (Best et al., 2007), and acute γ-secretase inhibition can improve memory (Comery et al., 2005). PS1 conditional knockouts, which eliminate γ-secretase activity, can lead to recovery of both memory and synaptic deficits related to Aβ (Saura et al., 2005). Other Aβ-lowering strategies, such as immunization by use of immunogenic Aβ42 or direct intracerebroventricular injection of anti-Aβ antibodies, have improved learning capacity and LTP deficits in other Aβ models (Klyubin et al., 2005; Chen et al., 2007). Likewise, removing the other major Aβ-cleaving enzyme β-secretase has shown that lowering Aβ levels correlates with improvement of certain forms of memory (Ohno et al., 2007).
Our results suggest that γ-secretase inhibition by MRK-560 is likely to exert its positive effects on both LTP and basal synaptic activity via reduction of Aβ (although we cannot exclude a role for the intracellular APP fragment AICD, which also is disrupted by γ-secretase inhibition). Treatment of 6- to 7-month-old Tg2576 mice with MRK-560 for as little as 3 days was sufficient to reverse the LTP deficit in 7-month-old mice, although no improvement in I/O ratio was seen. However, a partial improvement in this ratio was observed after 7 days of treatment. This differential sensitivity of LTP and basal synaptic activity to γ-secretase inhibition suggests that there may be two separate Aβ toxicities manifest in these synapses that can be distinguished by age and treatment paradigm. Alternatively, synaptic plasticity may recover first, followed by synaptic efficacy. In conclusion, therapeutic γ-secretase inhibitors may alleviate Aβ toxicity in multiple ways, each of which could require a different extent or duration of Aβ lowering.
In contrast to younger mice, we did not observe any improvement in the basal synaptic transmission of 15-month-old Tg2576 mice after 3 months of treatment with MRK-560. It is possible that the Aβ effects on basal synaptic transmission become irreversible in aged Tg2576 mice. An alternative explanation, however, may be that the absolute levels of Aβ are more significant than the percentage lowering. For instance, although Aβ40 levels were reduced to 37% of vehicle-treated littermates in 15-month-old mice, the level of soluble Aβ40 was still approximately twice as high as in untreated 7-month-old mice. Measures of total soluble Aβ in human AD cases vary between studies depending on the extraction protocol. Nevertheless, they are typically in the range of 10 to 70 pmol/g (Lue et al., 1999; McLean et al., 1999). Comparable levels were found in 6- to 7-month-old Tg2576 mice (Fig. 1), whereas 14- to 15-month-old animals exhibit levels that far surpass those found in human disease.
It remains to be determined how Aβ mediates synaptic toxicity and what forms of Aβ are relevant to disease. Understanding both the early phase of Aβ-induced deficits in Tg2576 mice as well as this deviation in aged animals may provide important insights into the biological processes of Alzheimer's disease and aid in the design of novel therapeutics. Our data are consistent with the prevailing opinion that early intervention to disrupt Aβ production can arrest and perhaps reverse Aβ-mediated synaptic toxicity.
We recognize the Laboratory Animal Resources staff at Merck Research Laboratories (Boston, MA) for their care for the animals used in the study and assistance in dosing animals. We also acknowledge the efforts by Phieng Siliphaivanh to prepare and supply MRK-560 for our dosing studies.
- Received November 17, 2009.
- Accepted January 6, 2010.
This work was supported by Merck and Co.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- Alzheimer's disease
- β-amyloid peptide
- amyloid precursor protein
- Tg2576 or APPsw
- transgenic mouse with K670N/M671L mutation in APP
- long-term potentiation
- wild type
- artificial cerebral spinal fluid
- excitatory postsynaptic potential
- enzyme-linked immunosorbent assay
- N-[cis-4-[(4-chlorophenyl)sulfonyl]-4-(2,5-difluorophenyl) cyclohexyl]-1,1,1-trifluoromethanesulfonamide
- input/output ratio.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics