Age-Related Loss of Synaptophysin Immunoreactive Presynaptic Boutons within the Hippocampus of APP751^SL, PS1^M146L, and APP751^SL/PS1^M146L Transgenic Mice

Abstract

Neuron and synapse loss are important features of the neuropathology of Alzheimer’s disease (AD). Recently, we observed substantial age-related hippocampal neuron loss in APP751^SL/PS1^M146L transgenic mice but not in PS1^M146L mice. Here, we investigated APP751^SL mice, PS1^M146L mice, and APP751^SL/PS1^M146L mice for age-related alterations in synaptic integrity within hippocampal stratum moleculare of the dentate gyrus (SM), stratum lucidum of area CA3 (SL), and stratum radiatum of area CA1–2 (SR) by analyzing densities and numbers of synaptophysin-immunoreactive presynaptic boutons (SIPBs). Wild-type mice, APP751^SL mice and PS1^M146L mice showed similar amounts of age-related SIPB loss within SM, and no SIPB loss within SL. Both APP751^SL mice and PS1^M146L mice showed age-related SIPB loss within SR. Importantly, APP751^SL/PS1^M146L mice displayed the severest age-related SIPB loss within SM, SL, and SR, even in regions free of extracellular Aβ deposits. Together, these mouse models offer a unique framework to study the impact of several molecular and cellular events caused by mutant APP and/or mutant PS1 on age-related alterations in synaptic integrity. The observation of age-related SIPB loss within SR of PS1^M146L mice supports a role of mutant PS1 in neurodegeneration apart from its contribution to alterations in Aβ generation.

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder comprising cognitive and memory deterioration, progressive impairment of activities of daily living, and several neuropsychiatric symptoms.1 The neuropathology of AD is characterized by disturbances in neural circuits, such as loss of neurons and synapses, and by protein aggregations of β-amyloid and hyperphosphorylated tau.2–4 Accumulating evidence has indicated a crucial role for failure and loss of synapses in AD pathophysiology.2,4,5–7 Postmortem morphological studies on AD neuropathology have demonstrated significantly reduced synaptic connectivity in brain regions such as the neocortex and hippocampus.3,6,7 Reductions in synaptic densities showed a strong correlation with cognitive decline in AD.5 However, the reasons for the reduced connectivity in AD remain poorly understood.8 According to the “β-amyloid (Aβ) hypothesis of AD,” Aβ accumulation is the primary driving force in AD pathogenesis. This hypothesis is supported by the fact that mutations in the amyloid precursor protein (APP) and in presenilin (PS) 1 and PS2, causing early-onset cases of AD, modify APP processing and result in enhanced generation of Aβ.4 We have developed a transgenic mouse model of AD expressing human mutant APP751 (carrying the Swedish and London mutations KM670/671NL and V717I, Thy1 promoter; APP751^SL mice), human mutant presenilin-1 (PS1 M146L, HMG promoter; PS1^M146L mice) or both human mutant APP751 and human mutant presenilin-1 (APP751^SL/PS1^M146L mice) in neurons9–12 (see also Refs. 8 and 13). Similar to what happens in human AD, the amount of extracellular Aβ deposits as well as the level of reactive astrogliosis increase as these APP751^SL mice and APP751^SL/PS1^M146L mice age.9–12 Furthermore, the APP751^SL/PS1^M146L mice show substantial age-related loss of hippocampal pyramidal cells.12 Importantly, the amount of neuron loss exceeded the amount of Aβ deposits and associated reactive astrogliosis in these mice, indicating that the observed neuron loss was at least in part independent of Aβ deposits.12 Given the popular belief that extracellular Aβ deposits are the cause of neuronal degeneration in AD,4 these findings have prompted modification of the original “amyloid cascade” hypothesis for AD.13,14

It was the aim of the present study to test the hypothesis that these APP751^SL mice, PS1^M146L mice, and APP751^SL/PS1^M146L mice show region-specific, age-related alterations in synaptic integrity within the hippocampus, with region-specific differences among the investigated genotypes. Another aim was to determine to what extent these alterations in synaptic integrity were due to Aβ deposits within the hippocampus and to hippocampal neuron loss. The investigations were carried out by two-photon laser scanning microscopy, computerized image analysis and design-based stereology, focusing on densities and numbers of synaptophysin-immunoreactive presynaptic boutons (SIPBs). Analysis of SIPBs is a well-established approach for investigating alterations in synaptic integrity in vivo both in our15,16 and in many other laboratories.17–21

Materials and Methods

Animals

The following groups of either 4.5-month-old (M4.5) or 17-month-old (M17) female mice were examined in the present study: wild-type control mice (C57Bl6 background; M4.5, n = 6; M17, n = 6), APP751^SL mice (transgenic mice expressing human mutant APP751 carrying the Swedish and London mutations KM670/671NL and V717I; Thy1 promoter; C57Bl6 [87.5%] × CBA [12.5%] background; M4.5, n = 6; M17, n = 3), PS1^M146L mice (transgenic mice expressing human mutant presenilin-1 [PS1 M146L]; HMG promoter; back-crossed on a C57Bl6 background for more than six generations;; M4.5: n = 6; M17: n = 6) and APP751^SL/PS1^M146L mice (APP/PS1 double-transgenic mice carrying the aforementioned APP and PS1 mutations; M4.5, n = 6; M17, n = 6). All mice were generated at the SanofiAventis Centre de Recherche de Paris (Vitry sur Seine, France) (for details, see Refs. 9–12). The APP751^SL/PS1^M146L mice were generated by crossing homozygous PS1 mice to hemizygous APP751^SL mice. In the present study, hemizygous PS1^M146L littermates were used. All experiments were performed in accordance with the German animal protection law.

Tissue Processing

Mice were anesthetized with chloral hydrate and sacrificed by intracardial perfusion fixation with tyrode followed by fixative containing 4% paraformaldehyde, 15% picric acid and 0.05% glutaraldehyde in phosphate buffer (for details, see 12). Brains were removed rapidly, halved in the medio-sagittal line, and postfixed for 2 hours at 4°C in the fixative, omitting the glutaraldehyde. Brain tissue was then cryoprotected by immersion in 30% sucrose in Tris-buffered saline at 4°C overnight. Afterward, brains were quickly frozen and stored at −80°C until further processing. The right hemispheres were cut into entire series of 30-μm-thick frontal sections on a cryostat (Leica CM 3050; Leica, Nussloch, Germany). Afterward, these series were divided into 10 subseries of every 10th section each. The hippocampus was found on five to eight sections per subseries, depending on the individual rostrocaudal extension of the hippocampus. Four subseries of sections per animal (from the wild-type control mice, the PS1^M146L mice, and the APP751^SL/PS1^M146L mice but not from the APP751^SL mice) were used in a former study12 to investigate age-related alterations in neuron numbers, the amount of Aβ deposits, apoptotic cell death, and microglia activation within the hippocampus. The left hemispheres were not used for the present study.

Investigation of Alterations in SIPB Morphology Due to Aβ Deposits

To investigate normal SIPB morphology and its alterations due to Aβ deposits, one subseries of sections per animal was used for immunohistochemical detection of both synaptophysin and β-amyloid (monoclonal mouse anti-synaptophysin antibody, 1:2000; Chemicon, Temecula, CA; rabbit anti-730 antibody detecting β-amyloid and P3,22 1:5000; generous gift from Dr. G. Multhaup). Briefly, fluorescent immunohistochemistry was carried out in a free-floating manner using standard protocols. Sections were incubated overnight at 4°C after rinsing with Tris-buffered saline (TBS) and TBS with Triton-X 100 and blocking of unspecific staining with 10% fetal calf serum and 4% nonfat-dry-milk in TBS. On the following day, the sections were rinsed and incubated with secondary antibodies for 1.5 hours at room temperature. Donkey anti-mouse IgG Alexa Fluor 488 antibody (1:100; Molecular Probes, Eugene, OR) and donkey anti-rabbit IgG Alexa Fluor 594 antibody (1:100; Molecular Probes), were used as secondary antibodies. Sections were mounted on gelatinized glass slides, dried, and coverslipped using 80% glycerol in PBS.

Sections were analyzed with standard fluorescence microscopy and two-photon laser scanning microscopy (TPLSM). TPLSM was performed as previously described23 using a microscope objective (60×; water dipping; numerical aperture [NA] = 1.0) connected to an upright Nikon E600FN microscope (Nikon Corporation, Tokyo, Japan). If needed, further magnification was achieved by optical zoom in the scan head. To remove noise, each image was filtered applying the Kalman filtering procedure on three subsequent images during experiments. The fluorescent secondary antibody Alexa 488 was mainly visible in the green channel, whereas Alexa 594 was only visible in the red channel, and the obtained images (coded green and red, respectively) were combined into single images if warranted.

Analysis of SIPB Densities

To analyze SIPB densities in various hippocampal subregions, one subseries of sections per animal was used for immunohistochemical detection of synaptophysin (monoclonal mouse anti-synaptophysin antibody; 1:2000; Chemicon; avidin-biotin immunoperoxidase labeling) (Figure 1, A to D). The M.O.M. immunodetection kit (Vector Laboratories, Peterborough, UK) was used according to the manufacturer’s protocol to minimize background labeling and labeling of intraparenchymal IgG molecules. Control experiments comprised incubations without primary antibody and incubations without use of the M.O.M. immunodetection kit (Vector). All incubations were performed in a free-floating manner under exactly identical conditions. After labeling procedures, sections were mounted on gelatinized glass slides, dehydrated, coverslipped, and coded. The code was not broken until all analyses were completed.

Figure 1.

Representative photomicrographs of frontal sections through the hippocampus of a 17-month-old wild-type control mouse (A, E to H, and K), a 17-month-old APP751^SL mouse (B), a 17-month-old PS1^M146L mouse, and a 17-month-old APP751^SL/PS1^M146L mouse (D and I), showing immunoreactivity for synaptophysin. Aβ deposits within the hippocampus of the APP751^SL mouse and the APP751^SL/PS1^M146L mouse are indicated by arrows in B and D, respectively. The small squares in E represent the areas where the high-power photomicrographs for quantitative SIPB analysis were taken within SM, SL, and SR. Corresponding high-power photomicrographs depict larger SIPBs within SL (G) than within SM (F) and SR (H). Aβ deposits were clearly recognizable as round/oval structures not immunoreactive for synaptophysin and surrounded by multiple enlarged SIPBs and disturbed neuropil (I). The dotted line in I indicates the delineation of Aβ deposits as performed for the analysis of SIPB densities in regions free of Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology. In K, the projection areas of SM, SL, and SR are shown. Scale bar = 250 μm in A to E and K, 20 μm in F to H, and 10 μm in I.

SIPB densities were analyzed as previously described15,16 within three subregions of the hippocampus: stratum moleculare of the dentate gyrus (SM), stratum lucidum of area CA3 (SL), and stratum radiatum of area CA1–2 (SR) (Figure 1E). Three randomly chosen areas per section were evaluated within SM and SR each, and two randomly chosen areas within SL (Figure 1E). Three high-power photomicrographs were taken from each evaluated area at approximately 2 μm below the upper surface of the sections (ie, when the SIPBs were clearly in focus) with a digital camera (F-view; Olympus, Tokyo, Japan) attached to an Olympus AX-70 microscope (100× oil objective; Olympus UplanApo, NA = 1.35). This procedure resulted in 45 to 72 photomicrographs for SM and SR each per animal (3 photomicrographs × 3 areas × [5 to 8] sections), and in 30 to 48 photomicrographs for SL per animal (2 photomicrographs × 3 areas × [5 to 8] sections) (Figure 1, F to H). In case of the APP751^SL mice and the APP751^SL/PS1^M146L mice, photomicrographs were exclusively taken at regions free of Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology. Aβ deposits were clearly recognizable as round/oval structures not immunoreactive for synaptophysin and surrounded by multiple enlarged SIPBs and disturbed neuropil (Figure 1I). SIPBs were detected by image analysis using AnalySIS-pro software (Soft Imaging System, Münster, Germany), slightly modified for detection of grayscale punctae as previously described.16 All settings were kept identical for all analyses. Shading error correction was performed before measurements were done to correct for irregularities in illumination of the microscopic fields. Furthermore, background levels were equalized. From these data, mean particle densities per square micrometer were calculated.

To enable the elimination of noise signal and to differentiate between possible artifacts and specific SIPBs, the normal size of SIPB was evaluated separately for SM, SL, and SR in a pilot experiment. For doing so, the area occupied by individual particles with a morphological appearance of SIPBs was measured separately within SM, SL, and SR in three wild-type mice using the AnalySIS-pro software (Soft Imaging System). A total of 400 particles was measured per hippocampal subregion. SIPB areas were, on average, 0.33 ± 0.12 μm² (mean ± SD) in SM, 1.02 ± 0.38 μm² in SL, and 0.36 ± 0.13 μm² in SR. The different SIPB size can be seen in Figure 1, F to H. A region-specific range of mean ± (2 × SD) was considered as normal SIPB size, and these settings were kept identical for all analyses. Size measurements of all SIPBs investigated in the present study indicated no differences in average SIPB size between the various groups of mice. Additional pilot studies (not described here) indicated that average SIPB size was independent from the plane of section (ie, identical results were obtained on frontal, sagittal, and horizontal sections of mouse brains).

Measuring Volumes of SM, SL, and SR

Volumes of SM, SL and SR were evaluated on the sections immunohistochemically processed for synaptophysin detection with avidin-biotin immunoperoxidase labeling with the aid of another subseries of sections per animal stained with cresyl-violet (Nissl) staining (0.01%) as previously described.12,16 Briefly, the Nissl-stained sections were laid onto the corresponding adjacent sections processed with immunohistochemistry and were exactly aligned. Then SM, SL, and SR were delineated according to the literature24 (Figure 1K). Delineation was carried out with a stereology workstation and StereoInvestigator software (MicroBrightField, Williston, VT) as previously described.16 For doing so, the Nissl-stained sections were brought into focus. After delineation, these sections were removed; the sections processed with immunohistochemistry were brought into focus; and the boundaries of SM, SL, and SR were verified. Volumes of SM, SL, and SR were calculated with Cavalieri’s principle25,26 from the projection area measurements and the average actual section thickness after histological processing. The latter was measured with the electronic microcator of the stereology software and a 40× oil objective (Olympus UplanApo; NA = 1.00) as described previously.26

Calculation of SIPB Numbers

Hippocampal subregion-specific SIPB numbers were calculated by multiplying the individual SIPB density data with the corresponding volume data. Note that these SIPB numbers were not unbiased because SIPB densities were analyzed in one focal plane and thus in two dimensions instead of in three dimensions.26 Thus, the calculations did not take the dimensions of the investigated parameters (ie, 1/μm² [SIPB density] and mm³ [volume]) into account. Importantly, this did not affect the results of the present study because for all groups of mice, similar distribution plots were obtained for individual SIPB size (not shown), and the average SIPB size was independent from the plane of section. Calculating un-biased total numbers of SIPBs was not the aim of the present study.

Calculation of Corrected SIPB Numbers

For the analysis of the APP751^SL mice and the APP751^SL/PS1^M146L mice the calculation of SIPB numbers resulted in theoretical numbers assuming that Aβ deposits did not influence normal SIPB morphology (because SIPB densities were exclusively measured at regions free of Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology as explained above). To take the impact of Aβ deposits on SIPB numbers into account, the percent volumes of SM, SL, and SR occupied by Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology (as depicted in Figure 1I) were analyzed with point-counting methods26,27 on the sections immunohistochemically processed for synaptophysin detection with avidin-biotin immunoperoxidase labeling. Point counting was performed on the stereology workstation with the 40× oil objective and a distance of 50 μm between the points in mutually orthogonal directions X and Y for SM and SR (and 20 μm for SL, respectively). Corrected SIPB numbers for the APP751^SL mice and the APP751^SL/PS1^M146L mice were calculated by subtracting the volume occupied by Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology from the total volume of the investigated hippocampal subregions and by multiplying this value with the mean SIPB density found within the corresponding hippocampal subregion. Thus, corrected SIPB numbers for the APP751^SL mice and the APP751^SL/PS1^M146L mice were smaller than the corresponding uncorrected SIPB numbers. For the wild-type control mice and the PS1^M146L mice, the corresponding uncorrected SIPB numbers could serve as corrected SIPB numbers as well. This was due to the fact that these mice did not develop Aβ deposits.

Statistical Analysis

For each group of mice mean and SEM were calculated for all investigated variables (hippocampal subregion-specific SIPB density, volume of the subregion, SIPB number, and corrected SIPB number if applicable). Comparisons between the groups were performed separately for SM, SL, and SR with two-way analysis of variance (two-way ANOVA; four genotypes and two age groups per genotype) followed by Bonferroni posttests to compare replicate means focusing on age-related effects. Statistical significance was established at P < 0.05. All calculations were performed using GraphPad Prism (Version 4.00 for Windows; GraphPad Software, San Diego, CA).

Photography

Pictures shown in Figure 1, A to E and K, were generated with the stereology workstation using the Virtual Slice tool of the StereoInvestigator software, which automatically created high-resolution seamless image montages composed of approximately 100 individual fields-of-view visualized at high resolution. Photomicrographs shown in Figure 1, F to I, were taken with the Olympus AX-70 microscope. Pictures shown in Figures 2 and 3 were acquired using TPLSM. Image stacks were reconstructed into three-dimensional images followed by colocalization analysis with Imaris software (Bitplane, Zurich, Switzerland). Final figures were constructed using Corel Draw Version 11.0 (Corel, Ottawa, Canada) and Adobe Photoshop Version 7.0 (Adobe, San Jose, CA). Only minor adjustments of contrast and brightness were made, which in no case altered the appearance of the original materials.

Figure 2.

Representative high-power photomicrographs obtained by TPLSM, showing alterations in synaptic integrity caused by Aβ deposits within hippocampal subregion SR of a 17-month-old APP751^SL/PS1^M146L mouse. Aβ immunoreactivity is shown in red in A, D, and G; synaptophysin immunoreactivity at the corresponding locations and focal planes in green in B, E, and H; and merged pictures in C, F, and I. Note the disturbed SIPB morphology within and in the near vicinity of the Aβ deposits (arrows). The SIPB-free regions around the Aβ deposits (arrowheads in E and F) are most probably caused by astrocytes surrounding the Aβ deposits. Interestingly, several enlarged SIPBs were entirely surrounded by β-amyloid (arrows in I). In K to M, a three-dimensional reconstruction of a representative image stack acquired with TPLSM is shown, displaying synaptophysin (green) and Aβ (red) immunoreactivity within SR of a 17-month-old APP751^SL/PS1^M146L mouse. The dotted lines in K represent the position within the X-Y view at which the Y-Z view (L) and the X-Z view (M) were generated. The depicted enlarged synaptophysin-immunoreactive presynaptic bouton is surrounded by a rim of Aβ immunoreactivity. Scale bar = 100 μm in A to C, 50 μm in D to I, and 5 μm in K to M.

Figure 3.

Results of the quantitative investigations in the hippocampus. Analysis of hippocampal SM (A, D, G, and K), SL (B, E, H, and L), and SL (C, F, I, M) of wild-type control mice (WT), APP751^SL mice (APP), PS1^M146L mice (PS1), and APP751^SL/PS1^M146L mice (APP/PS1) for age-related alterations in SIPB densities within regions free of Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology (A to C); volumes of SM, SL, and SR (D to F); SIPB numbers within regions free of Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology (G to I); and corrected SIPB numbers considering the amount of Aβ deposits within SM, SL, and SR (K to M). Results of 4.5-month-old mice (M4.5) are represented by open bars; those of 17-month-old mice (M17) by closed bars. Data are given as mean and SEM. Comparisons between the groups were performed separately for SM, SL, and SR with two-way ANOVA (P values summarized in Table 1) followed by Bonferroni posttests to compare replicate means focusing on age-related effects (P values provided as *P < 0.05, **P < 0.01, or ***P < 0.001 in A, B, C, D, G, H, I, K, L, and M). Significant differences in mean data are indicated (M17 versus M4.5) above each graph. ROI, region of interest; Corr., corrected. SIPB numbers were calculated by multiplying the individual SIPB density data with the corresponding volume data not considering the dimensions (see Materials and Methods for details).

Results

Alterations in SIPB Morphology Due to Aβ Deposits

Aβ deposits within the hippocampus were detected in APP751^SL mice at M17 and in APP751^SL/PS1^M146L mice at M4.5 and M17. No synaptophysin immunoreactivity was observed within the core of the Aβ deposits. Regions directly adjacent and surrounding the Aβ deposits showed substantial alterations in synaptophysin immunoreactivity. Specifically, considerably enlarged SIPBs were detected (up to 8 μm in diameter) close to the Aβ deposits (Figure 2). Three-dimensional reconstruction of TPLSM image stacks showed that some of these enlarged SIPBs were surrounded by Aβ (Figure 2).

SIPB Densities

Two-way ANOVA showed statistically significant age- and genotype-related differences in mean SIPB densities within SM, SL, and SR (all P values of two-way ANOVA are summarized in Table 1). Bonferroni posttests showed a significant age-related decline in mean SIPB densities within SM for the PS1^M146L mice (on average, −19%) and the APP751^SL/PS1^M146L mice (−33%), within SL only for the APP751^SL/PS1^M146L mice (−27%), and within SR for the PS1^M146L mice (−24%) and the APP751^SL/PS1^M146L mice (−31%) (data and P values of Bonferroni posttests shown in Figure 3, A to C).

Table 1.

Results of Two-Way ANOVA

		SM	SL	SR
SIPB density	Age	<0.001	0.009	0.001
	Genotype	<0.001	0.014	0.001
	Interaction	0.001	0.004	0.015
Volume of the subregion	Age	0.001	0.002	<0.001
	Genotype	<0.001	0.290	0.036
	Interaction	0.114	0.755	0.440
SIPB no.	Age	<0.001	0.004	<0.001
	Genotype	<0.001	0.077	<0.001
	Interaction	0.018	0.059	0.024
Corrected SIPB no.	Age	<0.001	0.001	<0.001
	Genotype	<0.001	0.016	<0.001
	Interaction	0.004	0.056	0.002

Data are expressed as P values. 

Volumes of SM, SL, and SR

Two-way ANOVA indicated significant age-related differences in mean volumes of SM, SL, and SR (see Table 1 for all P values). In the case of SM and SR, differences in mean volumes were also significant among the investigated genotypes. However, Bonferroni posttests showed a significant age-related decline only in the mean volume of SM in the APP751^SL/PS1^M146L mice (on average −26%) (Figure 3, D to F).

SIPB Numbers

As for mean SIPB densities and the mean volumes of SM, SL, and SR, two-way ANOVA indicated significant age-related differences in mean SIPB numbers within all investigated hippocampal subregions (see Table 1 for all P values). In the case of SM and SR, differences in mean SIPB numbers were also significant among the investigated genotypes. Bonferroni posttests showed a significant age-related decline in mean SIPB numbers within SM for the wild-type control mice (on average, −19%), the PS1^M146L mice (−24%) and the APP751^SL/PS1^M146L mice (−49%), within SL only for the APP751^SL/PS1^M146L mice (−33%), and within SR for the PS1^M146L mice (−30%) and the APP751^SL/PS1^M146L mice (−38%) (for data and P values of Bonferroni posttests, see Figure 3, G to I).

Corrected SIPB Numbers

Two-way ANOVA revealed statistically significant age- and genotype-related differences in mean corrected SIPB numbers within SM, SL, and SR (the P values of two-way ANOVA are given in Table 1). Bonferroni posttests showed a significant age-related decline in mean corrected SIPB numbers within SM for all investigated genotypes (wild-type control mice, −19% on average; APP751^SL mice, −19%; PS1^M146L mice, −24%; and APP751^SL/PS1^M146L mice, −56%), within SL only for the APP751^SL/PS1^M146L mice (−42%) and within SR for all investigated genotypes except for the wild-type control mice (APP751^SL mice, −32%; PS1^M146L mice, −30%; APP751^SL/PS1^M146L mice, −51%) (data and P values of Bonferroni posttests shown in Figure 3, K to M).

Discussion

Alterations in SIPB Morphology Due to Aβ Deposits

In the present study, enlarged SIPBs were found in the near vicinity of Aβ deposits within the hippocampus of the APP751^SL mice and the APP751^SL/PS1^M146L mice. This finding is in line with several studies in the literature reporting the presence of enlarged SIPBs in AD patients28 as well as in APP transgenic and APP/PS1 transgenic mice.21,29–33 Furthermore, dystrophic neurites with swollen varicosities in the near vicinity of Aβ deposits as well as spine loss and shaft atrophy have been reported in APP transgenic and APP/PS1 transgenic mice.29,30,34,35

Of particular interest was the observation of a close spatial relationship between aggregated Aβ and enlarged SIPBs in the present study. One may speculate that Aβ deposits are involved in the formation of enlarged SIPBs. In this respect, a study by Phinney et al36 indicated that Aβ deposits in APP23 mice may attract dystrophic axonal terminals. However, it is also possible that enlarged SIPBs are involved in the production or secretion of Aβ. Recent evidence indicating that axonally transported APP can give rise to Aβ and that Aβ is subsequently released from presynaptic sites and deposited in Aβ deposits in vivo37 may argue for the latter speculation. Furthermore, APP, presenilins, and synaptophysin have been shown to co-localize at the growth cones of developing cultured neurons.38 In summary, it is reasonable to conclude that Aβ deposits play an important role in degeneration of presynaptic structures and therefore have considerable impact on local synaptic connectivity.

Alterations in SIPB Densities and SIPB Numbers in APP751^SL and APP751^SL/PS1^M146L Transgenic Mice

The wild-type control mice investigated in the present study did not show significant differences in mean SIPB densities within hippocampal subregions SM, SL, and SR and in mean volumes of these subregions between 4.5 months and 17 months of age. Furthermore, an age-related decline in mean SIPB number was only observed within SM but not within SL and SR of the wild-type control mice. These data are in line with previous reports in the literature.19,20,39,40

Transgenic expression of human mutant APP and human mutant PS1 in the investigated mice had significant, region-specific effects on mean SIPB densities and SIPB numbers within the investigated hippocampal subregions. Compared with the age-matched wild-type control mice, the APP751^SL mice showed transgene-induced age-related loss of SIPBs only within SR (results of the PS1^M146L mice are separately discussed below). Within SM, a similar amount of age-related loss of SIPBs was observed in the wild-type control mice, the APP751^SL mice, and the PS1^M146L mice. Furthermore, none of these mice showed an age-related loss of SIPBs within SL. In contrast, the APP751^SL/PS1^M146L mice displayed transgene-induced age-related loss of SIPBs within SM, SL, and SR. Besides this, there was no difference between the wild-type control mice and the APP751^SL mice with respect to age-related loss of SIPBs within SR as long as Aβ deposits were not taken into account. In contrast, the APP751^SL/PS1^M146L mice showed a transgene-induced age-related loss of SIPBs even in regions within SM, SL, and SR free of Aβ deposits (discussed separately below).

Based on the analysis of SIPB densities within the hippocampal subregion SM of APP_Wt I63 and APP_Ind H6 transgenic mice, Mucke et al20 suggested evidence for Aβ-induced synaptotoxicity even in the absence of Aβ deposits in APP transgenic mice. This is in contrast to the findings of the present study in which age-related loss of SIPBs in the absence of Aβ deposits was found only for the APP751^SL/PS1^M146L mice but not for the APP751^SL mice. Methodological differences might explain this discrepancy between the findings of Mucke et al20 and the present study. Mucke et al20 correlated their SIPB density data with the overall hippocampal plaque load and found no significant correlation between these parameters in their APP transgenic mice. Based on this lack of correlation, these authors concluded that Aβ is synaptotoxic even in the absence of plaques in their APP transgenic mice.20 However, in the present study, no correlation between SIPB densities within SM and overall hippocampal plaque load was found in the APP751^SL mice as well as in the APP751^SL/PS1^M146L mice (not shown). Rather, correlation analyses between plaque load and other neuropathologic findings in transgenic mouse models of AD may be misleading, as outlined recently.12 Another reason for the discrepancy between the findings of Mucke et al20 and the present study might be differences in the transgenic constructs, ie, in the introduced mutations and/or the promoters used. In this regard, it is important to note that the APP transgenic mice investigated by Mucke et al20 showed a significantly reduced SIPB density within SM compared with wild-type controls already at 2 to 4 months of age. This might reflect a specific neurodevelopmental defect in the APP transgenic mice investigated by these authors. In contrast, in the present study the APP751^SL mice had the highest mean SIPB densities as well as (except for SL at M4.5) the highest mean SIPB numbers within SM, SL, and SR among the investigated genotypes at both M4.5 and M17. This might indicate synaptotrophic effects of APP in the APP751^SL mice investigated in the present study as known from other transgenic mice expressing different forms of human APP.41,42 In this regard, it is important to note that Takeuchi et al33 found slightly higher SIPB den-sities within hippocampal subregions SM and SR of APP^K670N/M671L transgenic mice and of APP^K670N/M671L/PS1^M146L double transgenic mice than within SM and SR of both wild-type control mice and PS1^M146L transgenic mice at 12 months of age. These data are fully in line with the results of the present study at M4.5. Wong et al43 found, compared with wild-type controls, increased densities of vesicular acetylcholine transporter immunoreactive presynaptic boutons within SR of APP^KM670/671NL transgenic mice but neither within SR of PS1^M146L transgenic mice nor of APP^K670N/M671L/PS1^M146L double transgenic mice at 8 months of age. Except for the results for the APP^K670N/M671L/PS1^M146L double transgenic mice, these data are in line with the findings of the present study at M4.5 as well. Takeuchi et al33 and Wong et al43 restricted their analyses to 12-month-old animals (or to 8-month-old animals, respectively). Thus, these studies cannot be compared with the present study with respect to age-related alterations in hippocampal SIPB densities based on transgenic expression of human mutant APP and/or human mutant PS1 in mice. The same holds true for a study by Dodart et al44 in which SIPB densities within several hippocampal subregions of 12-month-old APP^V717F transgenic mice were investigated (on average, no difference between APP^V717F transgenic mice and wild-type control mice).

Furthermore, it should be mentioned that King and Arendash21 observed a decline of approximately 40% in mean SIPB densities within hippocampal subregions SM and SR of wild-type control mice between 3 and 19 months of age and an approximately 40% increase in SIPB densities within SM and SR of APP⁶⁹⁵ transgenic mice. Accordingly, these authors found SIPB densities within SM and SR of 19-month-old APP⁶⁹⁵ transgenic mice approximately twice as high as within SM and SR of 19-month-old wild-type controls.21 These data are in line neither with the results of the present study nor with the findings of Mucke et al20 and other reports in the literature concerning age-related alterations in synaptophysin immunoreactivity within the hippocampus of wild-type rodents.19,20,39,40 Besides differences in the transgenic constructs, the reason for this discrepancy might be due to different methodologies. Seemingly, the image analysis approach of King and Arendash21 did not focus on the analysis of SIPB densities within regions free of Aβ deposits and Aβ aggregation-related disturbances in normal SIPB morphology. Furthermore, it cannot be excluded that the intense synaptophysin staining demonstrated by King and Arendash21 for APP⁶⁹⁵ transgenic mice at 19 months of age was affected by intraparenchymal IgG molecules (blocked by using the M.O.M. immunodetection kit (Vector) in the present study).

Finally, Phinney et al36 observed that Aβ deposits in APP23 mice attracted dystrophic axonal terminals to ectopic locations, which may lead to disruption of normal neuronal connectivity. Accordingly, the intense synaptophysin immunoreactivity at the periphery of neuritic plaques seen in the APP751^SL mice and the APP751^SL/PS1^M146L mice investigated in the present study as well as described in the literature (discussed above) might reflect neurotropic properties of Aβ deposits. One might even ask the question of whether this mechanism could result in reduced SIPB numbers in areas that should have received the projections. The results obtained on the APP751^SL mice in the present study may answer this question. Provided that the main mechanism in SIPB loss in regions free of Aβ deposits would have been the result of inappropriate redirection of projections to cluster around the Aβ deposits (and consequently resulting in deficits in the areas that should have received the projections), the APP751^SL mice should have shown significant differences in SIPB densities and numbers (compared with the wild-type control mice) in regions free of Aβ deposits. However, as shown in Figure 3, this was not the case.

In summary, the results of the present study are in line with several reports in the literature, and differences with other reports can be explained. Importantly, to the best of our knowledge, the results of the present study are the first ones demonstrating both quantitative and qualitative differences between the simultaneous transgenic expression of human mutant APP and human mutant PS1 and the single transgenic expression of either human mutant APP or human mutant PS1 in mice on age-related loss of SIPBs within the hippocampus. Potential reasons for these differences are discussed in the next paragraphs.

Alterations in SIPB Densities and SIPB Numbers in PS1^M146L Transgenic Mice

Of particular importance was the finding of a transgene-induced age-related reduction in both mean SIPB density and mean SIPB number within the hippocampal subregion SR (CA1–2) of the PS1^M146L mice in the present study. To the best of our knowledge, these are the first results demonstrating a region-specific effect of transgenic expression of human mutant PS1 on age-related loss of SIPBs within the mouse hippocampus.

Presenilins are considered to be the catalytic subunits of the γ-secretase complex, but they are also essential for the fine tuning of the immunological system and for memory and synaptic plasticity.45 AD-associated presenilin mutations clearly affect γ-secretase cleavage and thereby enhance Aβ generation.10 According to previous observations by Duff et al,46 expression of mutant PS1 itself may generate murine Aβ. We have previously investigated the presence of human Aβ in the PS1^M146L mice and found no Aβ signal in ELISA blots using a monoclonal antibody (WO-2) against human Aβ.10 However, we cannot exclude that there was at least some generation of endogenous murine Aβ (possibly not detectable by the antibody used in our previous study) in the PS1^M146L mice. On the other hand, we detected considerable amounts of Aβ in the APP751^SL mice.10 The fact that we found more age-related SIPB loss within SM and SR of the PS1^M146L mice than within SM and SR of the APP751^SL mice would argue that other factors rather than Aβ were the driving force in SIPB loss in the PS1^M146L mice. In this respect, it is important to address that presenilins affect intracellular calcium stores, protein trafficking and protein turnover, and neuronal signaling.47 Consequently, it has been proposed that destabilization of calcium signaling has a role in neurodegeneration in AD.47 Mutations in PS1 and PS2 are known to perturb endoplasmic reticulum calcium homeostasis such that greater amounts of calcium are released on stimulation.48,49 Furthermore, Herms et al50 showed deregulation of intracellular calcium homeostasis by decreased capacitative calcium entry in hippocampal neurons dissociated from newborn mice transgenic for mutant PS1^A246E (capacitative calcium entry being a refill mechanism allowing entry of extracellular calcium ions through plasma membrane channels that are tightly regulated by and even physically linked to intracellular stores51). Importantly, calcium-induced formation of reactive oxidative species and mitochondrial swelling were found to be higher in isolated rat mitochondria from the CA1 than from the CA3 region.52 Furthermore, synaptosomal proteins from PS1^M146V knock-in mice displayed increased oxidative stress compared with synaptosomal proteins from wild-type mice.53 Begley et al54 observed evidence of disturbed calcium homeostasis and mitochondrial dysfunction in synaptosomes prepared from mice transgenic for mutant PS1; it is reasonable to hypothesize that similar to the knock-in mice studied by LaFontaine et al,53 synaptosomal proteins were also affected by increased oxidative stress in the transgenic mice investigated by Begley et al.54 Moreover, it was recently shown that expression of the PS1 mutations M146V, I143T, and deletion of exon 9 in human neuronal cells in vitro significantly increased the activation of glycogen synthase kinase 3-β (GSK3-β).55 In vivo, GSK3-β phosphorylates kinesin light chains and causes the release of kinesin-I from membrane-bound organelles, leading to a reduction in kinesin-I-mediated fast axonal transport.56 Consistent with such a deficit in kinesin-I-mediated fast axonal transport, densities of synaptophysin-containing vesicles and mitochondria were found reduced in neuritic processes of hippocampal neurons expressing mutant PS1^M146V in vitro.55 On the basis of these findings, it is attractive to speculate that the observed transgene-induced age-related loss of SIPBs within SR of the PS1^M146L mice in the present study was connected to transgene-induced alterations in cellular calcium homeostasis, fast axonal transport, and the formation of reactive oxidative species. At least the latter mechanism might occur preferentially within the hippocampal CA1 region, therefore triggering age-related SIPB loss preferentially within SR. Interestingly, perturbations in calcium homeostasis and alterations in GSK3-β activation caused by mutations in PS1 might be connected to each other. Calcium signaling activates the phosphatase calcineurin and induces movement of NFATc proteins into the nucleus, whereas nuclear import of NFATc proteins is opposed by kinases such as GSK3-β.57

Besides these mechanisms, recent studies indicated that presenilins also mediate cleavage of synaptic-associated adhesion molecules such as nectin1α58 and cadherins.59 However, it remains to be determined whether PS1 cadherin/catenin interaction or PS-mediated cleavage of adhesion molecules play a functional role in synapse formation or modulate synaptic activity.60 Thus, one can only speculate whether altered PS1-mediated cleavage of synaptic-associated adhesion proteins influenced synaptic function and integrity and was involved in the reported age-related loss of SIPBs within SR of the PS1^M146L mice.

Finally, it should be mentioned that in a recent study on 3-month-old mice in which the M146V mutation of PS1 was knocked in into the endogenous mouse PS1 gene, Wang et al61 found impaired hippocampus-dependent contextual learning and reduced adult neurogenesis in the dentate gyrus. The latter result is in line with earlier findings of increased cell cycle arrest caused by PS1 mutations overexpressed in cells in vitro.62 On the other hand, total numbers of hippocampal granule cells were found unchanged compared with wild-type controls in mice transgenic for the same PS1 mutation as investigated by Wang et al61 as well as in mice in which the human PS1 mutations M233T and L235P were knocked in into the mouse PS1 gene.12,63 Thus, the possible impact of reduced neurogenesis in the dentate gyrus of PS1^M146V knock-in mice on hippocampus-dependent contextual learning remains elusive.

In summary, the finding of a transgene-induced age-related loss of SIPBs within the hippocampal subregion SR of the PS1^M146L mice in the present study can be explained by potential mechanisms of mutant PS1 reported in the literature. One can regard the findings of the present study as among the first in vivo evidence for a role of mutant PS1 in neurodegeneration without concomitant neuron loss, apart from its contribution to alterations in Aβ generation. This might serve as the basis for further modifying the amyloid hypothesis of neurodegeneration in AD.14

Alterations in SIPB Densities and SIPB Numbers in APP751^SL/PS1^M146L Transgenic Mice Independent of Aβ Deposits

As explained above, there was no difference between the wild-type control mice and the APP751^SL mice with respect to age-related loss of SIPBs as long as Aβ deposits were not taken into account. It can therefore not be excluded that local development of Aβ deposits fully accounted for the loss of SIPBs seen within hippocampal subregion SR of the APP751^SL mice at M17. In contrast, the APP751^SL/PS1^M146L mice showed a transgene-induced age-related loss of SIPBs even in regions within SM, SL, and SR free of Aβ deposits. This phenomenon can be due to several interacting mechanisms: 1) The aforementioned potential actions of mutant PS1 (at least within SR). 2) Local neuron loss. Our previous study on the same APP751^SL/PS1^M146L mice showed a substantial age-related loss of hippocampal pyramidal cells (on average approximately 30%).12 Loss of hippocampal pyramidal cells resulted most probably in a decreased number of dendrites in SL and SR. Thus, one cannot exclude that the combined action of hippocampal pyramidal cell loss and Aβ deposits within SL and SR fully accounted for the loss of SIPBs seen within these regions of the APP751^SL/PS1^M146L mice. At M17, these mice showed a transgene-induced age-related reduction in mean SIPB numbers of 33% in regions within SL (and of 38% in regions within SR, respectively) free of Aβ deposits and an approximately 30% loss of pyramidal cells within the hippocampus.12 3) Neuron loss in regions with heavy projections to the areas analyzed. In this regard, we have previously shown that age-related neuron loss occurred within the hippocampal pyramidal cell layer but not within the hippocampal granule cell layer of the APP751^SL/PS1^M146L mice,12 ie, exactly the same mice as investigated in the present study. The mossy fibers (ie, the axons of the hippocampal granule cells) project to the hippocampal pyramidal cells (mainly to CA3). Thus, the effects seen in the SL of the APP751^SL/PS1^M146L mice cannot be explained by neuron loss in the region projecting to SL. Furthermore, age-related loss of SIPBs was found in SR of the PS1^M146L mice, which did not show age-related neuron loss in any hippocampal area.12 Preliminary evidence from microscopic inspection suggested age-related neuron loss in the entorhinal cortex of the APP^751SL/PS1^M146L mice but not in the PS1^M146L mice (not shown) that showed SIPB loss within SM. Together, these data indicate that loss of neurons at regions with heavy projections toward the areas analyzed in the present study was of limited impact on age-related SIPB loss. 4) Action of soluble Aβ. Soluble Aβ species have been demonstrated to induce neurotoxicity, which is likely to provoke neuropathology reminiscent of AD64,65 (see also Refs. 2 and 66–68). In this regard, a 12% reduction in total dendritic length of hippocampal granule cells was found in 3-month-old APP^V717F transgenic mice, ie, before amyloid deposition, in a recent study by Wu et al.69 Certain subsets of granule cells (superficially located granule cells in the dorsal blade of the dentate gyrus) even exhibited a 23% loss of total dendritic length at 3 months of age.69 The authors of this study hypothesized that these alterations were due to elevated levels of soluble forms of Aβ, induced by the overexpression of APP.69 Comparably, one may speculate that at least part of the 30% (49% minus 19% age-related loss of SIPB numbers seen within SM of the wild-type control mice) transgene-induced age-related reduction in SIPB numbers in regions free of Aβ deposits within SM of APP751^SL/PS1^M146L mice is due to elevated levels of soluble Aβ. 5) Intraneuronal Aβ. Previous reports have shown that intraneuronal Aβ preceded the formation of Aβ deposits in APP751^SL/PS1^M146L mice9 and that elevated levels of intraneuronal Aβ were associated with several alterations in normal cell homeostasis.11,14,64,70 Moreover, intraneuronal Aβ has been reported to substantially affect synaptic function26 and integrity.71 6) A chronic inflammation status. Previously, we reported widespread microglia activation within the hippocampal neuropil of APP751^SL/PS1^M146L mice that was only partly related to Aβ deposits.12 Thus, one could argue that microglia activation might be an early pathogenic event preceding loss of SIPBs and possibly having a causative role in SIPB loss.72,73

With respect to the potential impact of both soluble forms of Aβ and intraneuronal Aβ on SIPB loss in hippocampal regions free of Aβ deposits, analysis of SIPB densities and numbers in the hippocampus of APP^SL/PS1KI mice70 will be important. This is due to the fact that mutations in PS1 have been shown to enhance γ-secretase activity toward elevated Aβ42 generation.74–76 The Aβ42/Aβ ratio was approximately 0.2 in the APP751^SL mice and approximately 0.4 in the APP751^SL/PS1^M146L mice investigated here11 and was increased to approximately 0.85 in APP^SL/PS1KI mice that showed profound hippocampal neuron loss and intraneuronal accumulation of Aβ already at 10 months of age.70 We hypothesize that in the hippocampus of APP^SL/PS1KI mice, profound SIPB loss will also be found in regions free of Aβ deposits, and the comparison of results from APP751^SL/PS1^M146L mice and APP^SL/PS1KI mice will provide novel insights into the impact of transgenic expression and “knock-in” expression of mutations of PS1 on impairments of synaptic integrity in AD.

Relevance of the Findings of the Present Study to Human AD

Loss of hippocampal synaptophysin immunoreactivity was established as an early marker in human AD but did not show spatial correlation with Aβ deposits.77 By simultaneously investigating the formation of neurofibrillary tangles and the relative amount of synaptophysin mRNA within the same neurons, a number of studies indicated evidence for a tangle-dependent mechanism of hippocampal synaptophysin loss in AD.78–80 However, the formation of tangles accounts for only a small proportion of loss of hippocampal neurons in AD, at least within area CA1.81 Accordingly, the formation of tangles is most probably not the predominant mechanism of hippocampal synaptophysin loss in AD. This is supported by the finding of no local correlation between the degree of synaptophysin-like immunoreactivity and CD-1 immunoreactive neurons within the hippocampus of brains from patients with AD (with the CD-1 antibody recognizing tangles, dystrophic neurites in and around Aβ deposits, and neuropil threads).77 Rather, Callahan et al80 recently found that hippocampal CA1 neurons free of both tangles and immunohistochemical detectable tau phosphorylation at Ser 262, 356 and 396/404 in brains from patients with AD (Braak stage V-VI82) showed a statistically significant reduction of more than 20% in the relative amount of synaptophysin mRNA compared with corresponding neurons in brains from controls (Braak stage I-II).

These data indicate that hippocampal synaptophysin loss in AD seems to be multifactorial. In this regard, the simultaneous analysis of the APP751^SL mice, the PS1^M146L mice, and the APP751^SL/PS1^M146L mice investigated in the present study offers a unique framework to test hypotheses on the impact of several molecular and cellular events caused by transgenic expression of human mutant APP and/or human mutant PS1 on age-related alterations in synaptic integrity. First, the transgene-induced age-related SIPB loss seen within hippocampal subregion SR of the APP751^SL mice might be almost completely due to Aβ deposits. This hypothesis can be tested by vaccination or other approaches focusing on the prevention of Aβ aggregation. Second, the transgene-induced age-related SIPB loss seen within SL and SR of the APP751^SL/PS1^M146L mice might be almost completely due to the combined action of Aβ deposits and neuron loss. This offers for the first time the possibility to address the question of whether alterations in synaptic integrity precede neuron loss in a transgenic animal model of AD. Third, the transgene-induced age-related SIPB loss seen within SR of the PS1^M146L mice cannot be attributed to Aβ deposits or to neuron loss. This offers, for the first time, the possibility to study actions of mutant PS1 in neurodegeneration apart from its contribution to altered APP processing and Aβ generation. Finally, the transgene-induced age-related SIPB loss seen within SM of the APP751^SL/PS1^M146L mice can also neither be attributed to Aβ deposits nor to neuron loss. As age-related SIPB loss within SM of the PS1^M146L mice did not exceed age-related SIPB loss within this hippocampal subregion of the wild-type control mice, one might speculate that the age-related SIPB loss within SM of the APP751^SL/PS1^M146L mice that could not be explained by Aβ deposits and neuron loss was not due to actions of mutant PS1. Thus SIPB loss in APP751^SL/PS1^M146L is most probably due to complex interactions between the Aβ deposits, neuronal loss locally or at a distance, generation of intraneuronal and soluble Aβ, and other PS1 mediated actions.

In summary, the transgenic mouse models as investigated in our study can be seen among the best animal models of AD and provide, for the first time, the possibility to investigate hippocampal subregion-specific differences in interactions between neuron loss, Aβ deposits, soluble Aβ, intraneuronal Aβ, and PS1-mediated actions on alterations in synaptic integrity. Future experiments are likely to have further significant impact on the field of AD research.