|Home | About | Journals | Submit | Contact Us | Français|
Individuals with Down syndrome develop β-amyloid deposition characteristic of early-onset Alzheimer's disease (AD) in mid-life, presumably due to an extra copy of the chromosome 21-located amyloid precursor protein (App) gene. App mRNA and APP metabolite levels were assessed in the brains of Ts65Dn mice, a mouse model of Down syndrome, using qPCR, Western blot analysis, immunoprecipitation, and ELISAs. In spite of the additional App gene copy, App mRNA, APP holoprotein, and all APP metabolite levels in the brains of 4-month-old trisomic mice were not increased compared to the levels seen in diploid littermate controls. However starting at 10 months of age, brain APP levels were increased proportional to the App gene dosage imbalance reflecting increased App message levels in Ts65Dn mice. Similar to APP, sAPPα and sAPPβ levels were increased in Ts65Dn mice compared to diploid mice at 12 months, but not at 4 months of age. Brain levels of both Aβ40 and Aβ42 were not increased in Ts65Dn mice compared with diploid mice at all ages examined. Therefore, multiple mechanisms contribute to the regulation towards diploid levels of APP metabolites in the Ts65Dn mouse brain.
In mid-life, Down syndrome (DS) or trisomy 21 individuals develop early-onset Alzheimer's Disease (AD)-like dementia and age-related AD pathologies (Casanova et al. 1985; Mann et al. 1989), including deposition of the Aβ peptide as insoluble β-amyloid plaques in the brain parenchyma and around the cerebrovasculature (Hardy and Selkoe 2002). Since the amyloid precursor protein (App) gene is located on human chromosome 21 (HSA21), App triplication is thought to contribute to the early-onset AD phenotype in DS patients. Analyses of families with small duplications of a very small region containing the App gene on HSA21 support the view that App triplication alone is sufficient to cause AD pathology in humans (Rovelet-Lecrux et al. 2006; Sleegers et al. 2006). While this finding implicates App gene dosage in the eventual development of AD pathology in DS patients, the relationship between App gene copy levels, APP protein levels, and DS neuropathology - including, but not limited to β-amyloidosis - is aging-dependent and is likely to be multifactorial.
A widely used experimental mouse model of human DS is the Ts65Dn mouse, which is trisomic for a segment of murine chromosome 16 (MMU16) orthologous to the DS critical region of HSA21, which includes the App gene (Reeves et al. 1995). Ts65Dn mice exhibit developmental delay (Holtzman et al. 1996) and abnormal behaviors (Reeves et al. 1995) that appear to be analogous to mental retardation in DS patients. These learning deficits correlate with age-related neuronal atrophy, neurodegenerative changes, and loss of nerve growth factor (NGF) retrograde transport that results in the degeneration of basal forebrain cholinergic neurons (BFCNs) (Cooper et al. 2001; Granholm et al. 2003; Salehi et al. 2006). Similar to human DS, the Ts65Dn mouse also develops AD-like neuronal endosomal pathology (Cataldo et al. 2003), pathological changes which are likely to underlie the failure of NGF-mediated trophic support in this model through signaling endosomes (Wu et al. 2009).
Crossing the Ts65Dn mouse with a mouse carrying an APP null allele to restore App gene copy to 2N levels has shown that triplication of the App gene is necessary for the development of the abnormally large neuronal endosomes (Cataldo et al. 2003) and for the defective retrograde NGF trophic support seen in the Ts65Dn mouse (Salehi et al. 2006). While reduction of App gene dosage to 2N levels in Ts65Dn mice reduces neuronal endosomal pathology and BFCN degeneration, the interpretation of these genetic experiments is complex given the proteolytic processing of APP into multiple, and potentially both neurotrophic and neurotoxic, metabolites. Proteolytic processing of APP by both α- and β-cleavage generates soluble, amino-terminal fragments (sAPPα and sAPPβ), which are abundant and stable in the brain and have been shown to have neurotrophic effects in vitro in conjunction with NGF (Wallace et al. 1997; Wang et al. 2000). In contrast to the neurotrophic sAPP fragments, β-cleavage followed by γ-cleavage yields various Aβ peptides, which have been shown to have neurotoxic effects in multiple experimental systems (Lin et al. 2000; Hardy and Selkoe 2002; Vetrivel and Thinakaran 2006). Since the Ts65Dn mouse has age-related decreased cognitive ability (Reeves et al. 1995; Demas et al. 1996; Holtzman et al. 1996; Demas et al. 1998; Hunter et al. 2003a) and App-gene dosage dependent loss of vulnerable BFCNs (Granholm et al. 2000; Cooper et al. 2001), it is a good model system in which to examine in vivo the potential interplay between App gene triplication, aging, neurodegeneration, and APP proteolysis and metabolism.
All mouse experimentation and animal care was approved by the Nathan S. Kline Institute's Institutional Animal Care and Use Committee. Ts65Dn mice (n = 37) were maintained on a mixed background (C57BL/6jEi x C3H/HeSnJ) and 2N littermates (n = 42) were used as control animals. APP null mice were purchased from Jackson Laboratory Mice and Services (Bar Harbor, Maine). Mice were euthanized and brains were immediately dissected and frozen on dry ice. For protein-based analyses, frozen hemibrains were homogenized as previously described with protease inhibitors (Schmidt et al. 2005b). For qPCR, frozen cortex was extracted in TRIzol LS Reagent (Gibco/Invitrogen, Carlsbad, California) prior to RNA isolation per the manufacturer's instructions. To block protein synthesis in the brain, a cohort of Ts65Dn mice and their control 2N littermates were injected intraperitoneally with 100 μl of 150 μg/ml cycloheximide prepared in pH-buffered saline (Gold and Sternberg 1978), and euthanized post injection as indicated prior to brain dissection (Morales-Corraliza et al., submitted).
The anti-APP C-terminal monoclonal antibody C1/6.1 recognizes both mouse and human APP holoprotein and C-terminal fragments (CTFs) (Mathews et al. 2002). The monoclonal m3.2 antibody is murine-specific and recognizes APP holoprotein, sAPPα, βCTF, and Aβ. In combination with Aβ40 and Aβ42 C-terminal specific monoclonal antibodies (JRF/cAβ40/10 and JRF/cAβ42/26), m3.2 was used as previously described to detect endogenous murine Aβ in Aβ40- and Aβ42-specific sandwich ELISAs (Schmidt et al. 2005a). To detect both species of sAPP, 22C11 (Millipore, Temecula, CA) was used. Neprilysin (EC 188.8.131.52) was detected with the monoclonal antibody 56C6 (CD10) (Novacastra, Newcastle, UK), and insulysin (insulin degrading enzyme, IDE, EC 184.108.40.206) with the rabbit polyclonal antibody anti-IDE1 (Qiu et al. 1998). Superoxide dismutase (SOD1, EC 220.127.116.11) was detected with a rabbit polyclonal antibody anti-SOD-1 (FL-154; Santa Cruz Biotechnology, Santa Cruz, CA).
For Western blot analysis, proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were incubated overnight in primary antibody, washed, and incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (MP Biomedicals, Irvine, CA) for 1 hour. ECL substrate (Amersham Biosciences, Piscataway, NJ) was added before exposure to x-ray film. Blots were quantified using Multigauge (Fujifilm, Stamford, CT) and the density of signal was normalized to 2N littermate signal levels. Data are presented as the ratio of Ts65Dn band intensity to that of 2N littermate (mean±SEM). Uniformity of loading was confirmed by Ponceau S staining, β-actin and β-tubulin probing.
Soluble proteins were extracted from hemibrain homogenates using 0.2% diethylamine followed by centrifugation at 135,000 x g for 1 hour (Schmidt et al. 2005a). To differentiate sAPPα and sAPPβ, the soluble protein extract containing both sAPP species was immunoprecipitated overnight using m3.2 (2 μg/μl) to isolate murine sAPPα. Equivalent amounts of immunoprecipitate and supernatant were electrophoresed, transferred to membrane, and probed with 22C11.
Aβ was measured as previously described following diethylamine extraction (Schmidt et al. 2005a). Sandwich ELISAs to detect murine Aβ40 or Aβ42 were done as previously reported with m3.2 antibody used for detection (Schmidt et al. 2005a). ELISA plates were developed using a color reaction (TMB Microwell Peroxidase Substrate System, Kirkegaard & Perry Laboratories, Gaithersburg, MD) and read against synthetic murine Aβ standards of known concentration.
App mRNA levels were analyzed by qPCR as previously described (Ginsberg 2005; Ginsberg et al. 2006; Ginsberg and Mirnics 2006; Alldred et al. 2009). qPCR was performed using Taqman PCR primers (Applied Biosystems, Foster City, CA) for the murine App gene (ABI Assay Mm00431827_m1), SOD1 gene (ABI Assay Mm01344232_g1), dual-specificity tyrosine(Y)-phosphorylation regulated kinase 1A (DYRK1A, ABI Assay Mm00432934_m1), and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH, ABI Assay Mm99999915_G1) as a control. Samples were run on a real-time PCR cycler (7900HT, Applied Biosystems, Foster City, CA) as per the manufacturer's instructions. Standard curves and cycle threshold (Ct) were measured using standards obtained from brain RNA. The ddCT method was employed to determine relative gene level differences. A total of 3-4 independent samples per subject were assayed in triplicate.
SPSS (Chicago, IL) was used to conduct one-way ANOVAs for all analyses with post hoc analysis (Neumann-Keuls test; level of statistical significance was set at 0.05) for individual comparisons. Error bars are SEM.
Hemibrains lacking cerebellum from 5-month-old (n=4 Ts65Dn, n=4 2N) and 12-month-old (n=4 Ts65Dn, n=4 2N) Ts65Dn mice and 2N littermate controls were analyzed via real-time qPCR using a Taqman assay (Ginsberg 2005; Ginsberg et al. 2006; Ginsberg and Mirnics 2006). App message levels in the aged 12-month-old Ts65Dn mice were 1.80±0.21-fold higher (p<0.005) than in 2N littermate controls while 5-month-old Ts65Dn mice had similar levels (0.92±0.06-fold) to 2N controls (Figure 1), in close agreement with our Western blot data on APP holoprotein levels.
Levels of APP in 4-month-old trisomic mice were similar to levels in 2N mice (1.1±0.03-fold of 2N, n=5 Ts65Dn, n=5 2N) (Figure 2A, lanes 2-7). Given the increased App gene dosage in these mice, it was unexpected that APP levels do not track with gene copy number in the 4-month-old trisomic mice. In contrast, a 1.8±0.08-fold (p<0.005; n=9 Ts65Dn, n=11 2N) increase in endogenous APP levels was seen in 12-month-old Ts65Dn mice compared to 12-month-old 2N mice (Figure 2A, lanes 8-13). We next examined APP levels in Ts65Dn mice at 2-month intervals between 2 and 12 months of age, and a representative Western blot is shown in Figure 2B. APP levels did not change as a function of age in control 2N mice. APP levels in hemibrain homogenates were similar in Ts65Dn mice up to 8 months of age when compared to 2N mice (2-month-old: Ts65Dn levels 1.1-fold of 2N, n=1 Ts65Dn, n=1 2N; 6-month-old: Ts65Dn levels 1.1±0.13-fold of 2N, n=2 Ts65Dn, n=3 2N; 8-month-old: Ts65Dn levels 1.0±0.08-fold of 2N, n=2 Ts65Dn, n=3 2N), yet APP levels differed significantly in 10-month-old (Ts65Dn levels 1.5±0.17-fold of 2N, n=3 Ts65Dn, n=1 2N, p<0.05) as well as the 12-month-old Ts65Dn mice when compared to 2N levels. Across multiple brain regions, APP levels were increased in 12-month-old Ts65Dn mice, with hippocampus (1.8±0.16-fold of 2N, lanes 1-4) and temporal cortex (1.6±0.08-fold of 2N, lanes 13-16) showing the greatest increase, and cingulate cortex (1.2±0.02-fold of 2N, lanes 5-8) and striatum (1.3±0.06-fold of 2N, lanes 9-12) showing a lesser increase (Figure 2C).
In addition to APP, we examined the brain expression with aging of two additional genes, SOD1 and DYRK1A, that are also triplicated in the Ts65Dn mouse as well as tripicated in human DS. Similar to APP, SOD1 protein levels were not elevated in hemibrain homogenates of 4-month-old Ts65Dn mice when compared to 2N mice (Figure 2D; 1.1±0.37-fold of 2N, n=3 Ts65Dn, n=3 2N), but was elevated in 12-month-old Ts65Dn mice (1.7±0.24-fold of 2N, n=3 Ts65Dn, n=3 2N, p<0.01). Quantitative PCR analysis of 12-month-old Ts65Dn mouse hemibrains showed a 1.34±0.06-fold increase of SOD1 message levels (p<0.05) compared to 2N littermate controls while 5-month-old Ts65Dn mice had similar levels (0.91±0.03-fold) compared to 2N controls (Figure 2E), in agreement with our SOD1 protein Western blot data. A similar finding for DYRK1A mRNA was seen (Figure 2F) with increased message levels in 12-month-old Ts65Dn mice (1.26±0.13-fold, p<0.05) when compared with age-matched littermate 2N controls and younger 5-month-old Ts65Dn and 2N mice (0.92±0.08-fold).
We next compared the levels of secreted, soluble APP (sAPP) in the brains of Ts65Dn mice to 2N littermate controls. Soluble proteins were examined by Western blot analysis using 22C11, which detects both α- and β-cleaved sAPP species (Figure 3A). In agreement with the levels of APP holoprotein (Figures 2A and 2B), the levels of total sAPP were found to be similar between Ts65Dn and 2N mice at 4 months of age (Ts65Dn 1.0±0.10-fold of 2N, n=3 Ts65Dn, n=3 2N). At 12 months of age, sAPP levels were increased (1.3±0.06-fold, n=3 Ts65Dn, n=3 2N) in the Ts65Dn compared to 2N mice. The specificity of the 22C11 signal for sAPP is demonstrated by the lack of any reactivity in an APP null mouse (Figure 3A, lane 13).
To determine whether the levels of sAPPα or sAPPβ are both similarly increased, or whether α- or β-cleavage is specifically altered in the Ts65Dn mice, a soluble protein extract containing both sAPP species was immunoprecipitated using the monoclonal antibody m3.2 to isolate sAPPα. This immunoprecipitation procedure, which fully recovers murine sAPPα (data not shown), was resolved along with an equal quantity of the sAPPβ-containing supernatant of the immunoprecipitation and sAPP species detected using 22C11 (Figure 3B and C). In 4-month-old mice (Figure 3B), the levels of both sAPPα and sAPPβ were found to be similar between Ts65Dn and 2N mice (sAPPα: Ts65Dn 1.0±0.09-fold of 2N, n=3 Ts65Dn, n= 3 2N; sAPPβ: Ts65Dn 1.0±0.07-fold 2N, n=3 Ts65Dn, n= 3 2N). In 12-month-old mice (Figure 3C), however, the levels of both sAPPα and sAPPβ were increased in the Ts65Dn brain relative to 2N mice (sAPPα: Ts65Dn 1.3±0.18-fold of 2N, n=3 Ts65Dn, n= 3 2N; sAPPβ: Ts65Dn 1.4±0.07-fold 2N, n=3 Ts65Dn, n= 3 2N). Thus, brain levels of both sAPPα and sAPPβ are increased in 12-month-old Ts65Dn mice, commensurate with increased brain levels of total sAPP in 12-month-old Ts65Dn mice. While the increase in APP holoprotein, total sAPP, and the β-cleaved sAPPβ in aged Ts65Dn mice predicts a similar increase in brain Aβ levels, unexpectedly no significant increase was seen in the Ts65Dn mouse brain compared to 2N tissue for CTFs (4-month-old αCTF: Ts65Dn 1.27±0.09-fold of 2N, n=3 Ts65Dn, n= 3 2N; 4-month-old βCTF: Ts65Dn 1.36±0.22-fold 2N, n=3 Ts65Dn, n= 3 2N; 12-month-old αCTF: Ts65Dn 1.21±0.35-fold of 2N, n=3 Ts65Dn, n= 3 2N; 12-month-old βCTF: Ts65Dn 0.93±0.24-fold 2N, n=3 Ts65Dn, n= 3 2N; Figure 3D), Aβ40 (Figure 3E upper panel), and Aβ42 (Figure 3E lower panel) at ages examined. The lack of an increase in brain Aβ levels, given the increase in APP, total sAPP, and the β-cleaved sAPPβ in aged Ts65Dn led us to examine the levels of two important Aβ degrading enzymes, IDE and neprilysin (Figure 3F). There were no significant differences in insulin degrading enzyme or neprilysin levels in the Ts65Dn mouse brain compared to 2N at 4 months of age or 12 months of age.
Although the age-dependent increase of brain APP levels in Ts65Dn is explained by gene-dosage driven changes in APP biosynthesis, the subsequent lack of increase in brain Aβ levels in Ts65Dn led us to examine the turnover of APP metabolites in the Ts65Dn mouse brain. To examine APP turnover in the intact brain, 12-month-old Ts65Dn and 2N littermate mice were injected with the protein synthesis inhibitor cycloheximide to block de novo synthesis of APP. APP has a short half-life in cultured neurons (LeBlanc et al. 1996) and it is substantially degraded within a few hours in the murine brain (Morales-Corraliza et al., submitted). Figure 4A shows a Western blot analysis of brain APP levels following protein synthesis inhibition. In agreement with the findings shown in Figure 2A, the initial levels of APP were greater in the Ts65Dn mice than in the 2N mice (compare lane 8 to lane 1 of Figure 4A). The rate of decline in the level of APP in the brain following cycloheximide treatment was found to be similar between 12-month-old Ts65Dn mice and their 2N littermates, although potentially more rapid in the Ts65Dn mouse brain (t½ ~1.5-2 hours) compared to 2N mouse brain (t½ ~2 hours).
In addition, we examined the levels of CTFs following cycloheximide treatment (Figure 4B). Although brain levels of both APP and sAPP were increased in the Ts65Dn mouse brain, CTF levels were not significantly increased in the Ts65Dn mouse brain at 12 months of age (12-month-old αCTF: Ts65Dn 1.2±0.20-fold of 2N, n=3 Ts65Dn, n= 3 2N; 12-month-old βCTF: Ts65Dn 0.9±0.14-fold 2N, n=3 Ts65Dn, n= 3 2N; compare lane 8 to lane 1 of Figure 4B; Figure 3D). Following cycloheximide treatment, the rate of clearance of CTFs in the 12-month-old Ts65Dn mouse brain was similar to that of the 2N littermates.
We also examined the clearance rate of total sAPP (Figure 4C). Since, sAPP is more stable than APP (Morales-Corraliza et al., submitted), we extended the time-course of cycloheximide treatment out to 7 hours. In agreement with the findings in Figure 3A, the initial levels of total brain sAPP in Ts65Dn are increased when compared to 2N. While 12-month-old 2N brain sAPP levels were found to be highly stable over the 7-hour period of cycloheximide treatment, the 12-month-old Ts65Dn brain sAPP turnover was more rapid (t½ ~7 hours). Finally, the clearance of brain Aβ40 and Aβ42 was examined in the 12-month-old Ts65Dn mouse following cycloheximide treatment (Figure 4D). The turnover rates were similar in the Ts65Dn mouse compared to 2N, although this methodology may exaggerate the half-life of Aβ in the brain due to the continued presence of APP during the initial time-points following the inhibition of protein synthesis.
Although copy number dependent overexpression of triplicated genes on the HSA21 locus in DS is observed (FitzPatrick et al. 2002), not all trisomic genes are overexpressed, and there is substantial variation in the increase of individual triplicated genes (Mao et al. 2005; Lockstone et al. 2007). Similarly, the Ts65Dn model of DS shows an increase in transcript levels of triplicated genes across multiple tissues (Saran et al. 2003; Kahlem et al. 2004). Nevertheless, there is variation of gene expression in the Ts65Dn mice, with many of the genes in dosage imbalance not expressing at the expected 1.5-fold increase in the brain at various ages (Lyle et al. 2004; Sultan et al. 2007). For example, our findings with DYRK1A are consistent with those published by Sultan and colleagues, which showed no increase in DYRK1A mRNA levels in 4-month-old Ts65Dn mice, although older mice were not examined (Sultan et al. 2007). In human DS, while APP protein (Rumble et al. 1989) and App mRNA levels (Oyama et al. 1994) appear to be increased in aged brain, other studies have shown limited changes in App message when DS individuals are compared to diploid (Lockstone et al. 2007) and contradictory findings have been reported from fetal tissue (Argellati et al. 2006). APP (Salehi et al. 2006) and combined APP/sAPP (Hunter et al. 2003a; Seo and Isacson 2005) protein levels have been shown to be increased in the brains of 12-month-old and older Ts65Dn mice. In addition, Hunter and colleagues have shown an aging-dependence to the increase of the combined APP/sAPP protein expression they detected in the Ts65Dn mouse brain, with changes apparent at an earlier age in the striatum and later in the cortex (Hunter et al. 2003b). At 12 months of age, we have shown that this increased signal in the brain consists not only of the parental APP protein, but also the two sAPP species. The reported findings on App mRNA levels in the Ts65Dn mouse brain are more variable, with some studies suggesting an increase in App mRNA in both young and old Ts65Dn mice (Lyle et al. 2004), while others have suggested that some brain regions show an increase in App mRNA in young Ts65Dn mouse brain (Sultan et al. 2007). Our findings suggest that age-dependent changes occur in the expression of multiple triplicated genes in the Ts65Dn brain, both at the level of the mRNA (App, SOD1, DYRK1A) and at the level of the protein (APP, SOD1).
Age-dependent increase of brain APP protein levels is intriguing given that BFCN degeneration in the Ts65Dn mouse is also age-dependent, initially detected at ~6 months of age (Holtzman et al. 1996; Granholm et al. 2000) with continued phenotypic loss of cholinergic neurons through 12 months of age (Cooper et al. 2001). The loss of this group of neurons has been correlated with decreasing spatial memory in 6-month-old Ts65Dn mice compared to 4-month-old Ts65Dn mice. Thus, while the Ts65Dn mouse has limited behavioral deficits at 4 months of age, these deficits increase in severity with aging (Hunter et al. 2003a), a phenotypic worsening that may drive or be driven by the apparent age-dependent dysregulation of various triplicated gene transcripts, including App. Additionally, multiple lines of evidence argue that BFCNs are particularly vulnerable to insults linked at some level to APP metabolite dysregulation in the brain. In AD, loss of BFCNs is a prominent feature of the disease (Whitehouse et al. 1982; Mufson et al. 2008). Failure of neurotrophic signaling required for BFCN maintenance and survival occurs through decrements in NGF levels, high-affinity NGF receptor-mediated endocytosis, and retrograde signaling endosome NGF trophic signaling (Delcroix et al. 2003; Nixon 2005; Wu et al. 2009). Collective dysfunction of these processes lead to frank BFCN degeneration (Mufson et al. 1995; Mufson et al. 1999), although how Aβ deposition or other aspects of AD pathobiology contributes to this process in AD is unclear. BFCNs in the Ts65Dn mouse show defective retrograde transport of NGF (Cooper et al. 2001; Sofroniew et al. 2001), and exogenous addition of NGF can prevent BFCN loss (Cooper et al. 2001; Sofroniew et al. 2001). NGF retrograde transport and BFCN survival are both improved in Ts65Dn mice made diploid for the App gene (Salehi et al. 2006), demonstrating genetically that dysregulation of some aspect of APP function, likely dependent upon APP overexpression, contributes critically to the degeneration of this neuronal population. Similarly, App gene triplication is necessary for the development of neuronal endosomal pathology in the Ts65Dn mouse (Cataldo et al. 2003), morphological changes hypothesized to be linked to signaling endosome dysfunction and NGF trophic failure.
The present biochemical analysis of APP metabolite expression was done using brain homogenates that contained cortical and subcortical regions, including white matter. It is possible that APP expression is initially increased locally in a limited neuronal population, an idea that is consistent with endosomal pathology being restricted at 4 months of age to a subpopulation of medial septal neurons (Cataldo et al. 2003) and with increased APP levels in striatum of Ts65Dn mice 6 to 8 months of age (Hunter et al. 2003b). Indeed, App mRNA level increases in the Ts65Dn mouse brain coincide broadly with the onset of cholinergic deficits (Holtzman et al. 1996; Granholm et al. 2000; Cooper et al. 2001) and increased endosomal pathology within the septohippocampal system (Cataldo et al. 2003). Despite an increase in App mRNA levels, which leads to increased APP holoprotein levels in aged Ts65Dn mice, our findings show that APP and sAPP turnover is not slowed in the Ts65Dn mouse brain. Strikingly, brain Aβ levels - in spite of increased APP expression and robust β-cleavage leading to a proportionate increase in sAPPβ - remain unchanged. Greatly reducing γ-cleavage leads to a stabilization of CTFs and an increase in CTF levels, which can be detected in brain (Rozmahel et al. 2002). We did not detect an increase in CTF levels in the aged Ts65Dn mice, nor was there an indication that the turnover of the CTFs was altered, which argues against a stabilization of CTFs as the underlying mechanism for the maintenance of brain Aβ levels in the aged Ts65Dn mouse brain. We did not detect an increased Aβ turnover rate or an increased level of two important Aβ degrading enzymes in the Ts65Dn mouse brain. Again, it is possible that critical aspects of Aβ production and turnover are altered within discrete subpopulations of neurons in the Ts65Dn mouse, and therefore not detected in a brain homogenate. Indeed it is possible, if not likely, that there are additional turnover mechanisms for APP metabolites, such as CTFs, which includes the lysosomal system, that can contribute to the turnover of CTFs without the generation of excess Aβ (Haass et al. 1992). While DS and triplication of small regions of HSA21 containing App inevitably lead to AD and β-amyloid pathology (Casanova et al. 1985; Mann et al. 1989; Rovelet-Lecrux et al. 2006; Sleegers et al. 2006), presumably through a mechanism that involves increased central Aβ and its accumulation over time, triplication of murine App does not lead to β-amyloid plaque pathology in the Ts65Dn mouse model. This may reflect that the more aggregative prone human Aβ has a greater tendency to accumulate and stabilize in the brain parenchyma than murine Aβ, and undoubtedly also reflects the difference in life span of the two species.
Overall, our findings argue that in the Ts65Dn mouse multiple mechanisms come into play to modulate APP and APP metabolite levels in the brain towards normal. In the young Ts65Dn mice, homeostasis of APP occurs at the level of gene transcription, with brain App mRNA levels identical to that of diploid in 4-month-old mice. Moreover, even in aged mice, where App mRNA, APP and sAPP levels approach App gene copy numbers, Aβ levels are maintained at close to normal levels. This complex regulation of APP in a well-established trisomic mouse model of DS is consistent with an important role for APP dysregulation in the development of the neurological phenotype in the mouse, including both endocytic alterations and BFCN neurodegeneration, and emphasize the potential for neuropathological processes to occur in some circumstances as a result of APP dysregulation in the absence of increased Aβ levels or β-amyloid accumulation.
We would like to thank Dr. Marc Mercken (Janssen Pharmaceutica, Beerse, Belgium) for anti-Aβ antibodies used in the ELISAs, Dr. Dennis Selkoe (Harvard Medical School, Boston, MA, USA) for the gift of the anti-IDE antibody, Ms. Nicole S. Diaz, Mr. Tae Lee, and Mr. Stephen D. Schmidt for technical and research support, and Dr. Monika Pawlik for her maintenance of the mouse colonies. This work was supported by the NIA (AG017617 and AG029787), the NINDS (NS045205), and the Alzheimer's Association (IIRG-07-60047).