Alzheimer's Disease Related Markers, Cellular Toxicity and Behavioral Deficits Induced Six Weeks after Oligomeric Amyloid-β Peptide Injection in Rats

Alzheimer’s disease (AD) is a neurodegenerative pathology associated with aging characterized by the presence of senile plaques and neurofibrillary tangles that finally result in synaptic and neuronal loss. The major component of senile plaques is an amyloid-β protein (Aβ). Recently, we characterized the effects of a single intracerebroventricular (icv) injection of Aβ fragment (25–35) oligomers (oAβ25–35) for up to 3 weeks in rats and established a clear parallel with numerous relevant signs of AD. To clarify the long-term effects of oAβ25–35 and its potential role in the pathogenesis of AD, we determined its physiological, behavioral, biochemical and morphological impacts 6 weeks after injection in rats. oAβ25–35 was still present in the brain after 6 weeks. oAβ25–35 injection did not affect general activity and temperature rhythms after 6 weeks, but decreased body weight, induced short- and long-term memory impairments, increased corticosterone plasma levels, brain oxidative (lipid peroxidation), mitochondrial (caspase-9 levels) and reticulum stress (caspase-12 levels), astroglial and microglial activation. It provoked cholinergic neuron loss and decreased brain-derived neurotrophic factor levels. It induced cell loss in the hippocampic CA subdivisions and decreased hippocampic neurogenesis. Moreover, oAβ25–35 injection resulted in increased APP expression, Aβ1–42 generation, and increased Tau phosphorylation. In conclusion, this in vivo study evidenced that the soluble oligomeric forms of short fragments of Aβ, endogenously identified in AD patient brains, not only provoked long-lasting pathological alterations comparable to the human disease, but may also directly contribute to the progressive increase in amyloid load and Tau pathology, involved in the AD physiopathology.


Introduction
Alzheimer's disease (AD) is the most common cause of dementia in the elderly and is characterized by a progressive impairment in cognitive functions, resulting from synapse and nerve cell destruction in the brain. AD symptoms include memory loss, alteration of the individual's personality and failure to communicate or perform routine tasks. The histopathological hallmarks of AD include the presence of extracellular senile plaques, intracellular neurofibrillary tangles (NFT), reduction and dysfunction of synapses, neuronal death and reduction in overall brain volume. Senile plaques are composed of insoluble extracellular aggregates consisting mainly of amyloid-b (Ab) peptides, which are generated by enzymatic cleavages of the amyloid precursor protein (APP), while NFT are the result of hyper-and abnormal phosphorylation of the microtubule-stabilizing protein Tau [1,2].
There is no doubt that progressive Ab accumulation contributes to AD. A correlation between the total amount of Ab in human brain and cognitive decline indicates that the amount of Ab, but not necessarily plaque formation, is important for AD progression [3,4]. Transgenic APP mice demonstrate cognitive decline before plaque formation [5], and soluble oligomers can inhibit cognitive function [6] and long-term potentiation [7,8]. In fact, it is possible that extracellular amyloid deposits are only one aspect of the larger pathological cascade and an indirect consequence of possible protective responses intended to sequester toxic soluble Ab oligomers [4]. The degree of dementia in AD correlates better with Ab assayed biochemically, than with the histologically determined number of plaque. The concentration of soluble Ab species, which cannot be detected through an immunohistochemical analysis, appears to be more closely correlated with cognitive deficits [3,4]. In fact, Ab deposits may not even carry the most aggressive toxicity, but instead represent a reserve of toxicity from where toxic oligomeric fragments could be released [9][10][11].

Animals
Adult male Sprague-Dawley rats (Depré, France) weighing 280-300 g at the beginning of the experiments were housed for 1 week before experiments under standard laboratory conditions (12 h/ 12 h light/dark cycle with lights on at 7:00 AM; 2161uC, food and water ad libitum). The animals were treated in accordance with the European Community Council Directive (EEC/86/609). The Animal Welfare Committee at the University of Montpellier 2 approved all protocols and all efforts were made to minimize the number of animals used and potential pain and distress. All surgery was performed under Ketamine/Xylazine mixture, and all efforts were made to minimize suffering. All experiments were performed in conscious rats between 9:00 AM and 2:00 PM, i.e. during the diurnal trough of the circadian rhythm.

Locomotor Activity and Body Temperature Variations
The day of the icv injection and during the same anesthetic session, a single telemetric transmitter (PhysioTel, TA 10TA-F40; DSI, USA) was implanted intra-peritoneally. The corresponding receiver (RA1010; DSI) was fixed under the animal's cage and connected via a BMC100 consolidation matrix (DSI) to a Dataquest III computerized data analyzer (DSI). The animals were then recorded during the 6 th week following peptide injection and telemetric data were analyzed. This system allows measurement of continuous locomotor activity and body temperature variations, as previously reported [20].

Spatial Short-term Memory (Delayed Alternation in the Tmaze)
As previously detailed [20,36], delayed alternation was tested in the T-maze and the results were expressed as ratio of the time spent in the initially closed novel arm over the time spent in the previous arm and as a ratio of the number of entries into the novel arm over the familiar one.
Spatial Long-term Memory (Place Learning in the Watermaze) As previously reported [20,25], spatial reference memory was tested using a place learning procedure in the water-maze. Training consisted of three swims per day for 5 days. Each rat was allowed a 90 s swim to find the platform and was left for a further 30 s on the platform. The median latency was determined for each training session. A probe test was performed 4 h after the last training session. The platform was removed and each rat was allowed a free 60 s swim. The percentage of time spent in the training quadrant was determined by videotracking (Viewpoint, France).

Endocrine Stress
Blood samples were collected after killing the rats by decapitation, as previously reported [37]. Plasma corticosterone (CORT) was assayed with a radioimmunoassay kit (Biotrak, GE-Healthcare, France) in 50 ml plasma sample diluted (1:5) with the assay buffer. The intra-and inter-assay coefficients of variation were 5% and 7%, respectively. The assay sensitivity was 0.6 ng/ml.

APP Processing
As previously reported [20], 60 mg from each sample was taken to western blot analysis following the same procedures detailed in 2.8. The primary antibody used to detect APP (125 kDa) and C99 fragment (13 kDa) was a rabbit anti-Amyloid Precursor Protein (PA1-84165: 1/750, ABR-Thermo-Scientific, France).

Tau Phosphorylation
To determine the levels of Tau phosphorylation at specific sites, equal amounts of protein (varying from 60 and 80 mg depending on the antibody used) from each sample were taken to western blot analysis following the same procedures detailed before. The primary antibodies to detect phospho-Tau epitopes (50 kDa) were a mouse AT8 (S 199 /S 202 /T 205 ) and AT100 (T 212 /S 214 /T 217 ) antibodies (MN1020: 1/3000 and MN1060: 1/3000, respectively; ThermoScientific, France), and to detect total Tau (50 kDa) was a mouse anti-Tau antibody (MA1-38710: 1/5000, ThermoScientific, France).

BDNF Content
Rats were sacrificed by decapitation and structures of interest were weighed, immediately frozen in liquid nitrogen and stored at 220uC until assayed. Brain-derived neurotrophic factor (BDNF) content was measured with a conventional ELISA assay (BDNF Emax; Promega, France), as previously reported [40,41]. The assay sensitivity was 15 pg/tube. The BDNF concentration was expressed as pg/g wet weight. The intra-and inter-assay coefficients of variation were 3% and 6%, respectively.
Histology (Cresyl Violet Staining; GFAP, Iba-1, PSA-NCAM, VAChT Immunolabeling) Animals were anaesthetized using an intramuscular injection of ketamine/xylazine solution and perfused intracardially (50 ml of NaCl 0.9% and 100 ml PO 4 0.2 M containing 4% paraformaldehyde). Brains were removed and postfixed in the same fixative for 48 h (4uC) and then in a solution of sucrose (30%) for 3 days. Thereafter, tissues were included in a block of OCT compound (Tissue-Tek, Sakura Finetek, USA) and quickly frozen in acetone chilled on dry ice. Frozen brains were mounted on a cryostat (Leica, France) and serially cut into 10 mm coronal sections. For histology, they were stained with 0.2% cresyl violet reagent, dehydrated and mounted. The method used for neuronal count is a classical method used with thin brain sections (10 mm) to quantify undamaged hippocampic cells [20,29,42,43]. The counting was made using images captured with Leica DFC495 highresolution camera (Leica Microsystem, Nanterre, France) attached to Leica DM2500 microscope (Leica) and the Leica LAS Core image analysis software (Leica). For this purpose, digitized images acquired using a 640 objective were transformed into TIFF files and brought to the same level of contrast and sharpness using the software. Four sections were studied from each brain, taken from the anterior hippocampus level (23.0 to 24.0 from the bregma) [35], with intervals of 250 mm. Sections were selected on a subjective random basis. Three fields were analyzed per hippocampus for CA1, one for CA2 and CA3 and two for DG. Counts of undamaged cells were made using ImageJ software on TIFF captured images. Neuron densities on slices (number of neurons in the optical field expressed as the number of cells per mm 2 ) were calculated as the arithmetic mean number of neurons in the two hemispheres and, for each animal, as the arithmetic mean of results obtained for each of the four slices. The count of undamaged cells in the CA1, CA2, CA3 and DG fields of the hippocampus was done by two different scientists unaware of the experimental conditions and independently from each other using display projections of the images (each person performed cell count for all animals included into the experiment and no difference was observed between the two independent and unaware analyses).

Results
Regional Distribution of Ab25-35-HLF Showed Persistent Presence of oAb25-35 after 6 Weeks in the Brain The regional distribution of oAb 25-35 -HLF 6 weeks after injection is presented in Fig. 1A-I. Control rat sections were treated and examined in the same conditions as injected rat sections and displayed no specific labeling. oAb 25-35 -HLF labeling was relatively discreet, suggesting a progressive clearance of the peptide after 6 weeks in comparison to previous study [20]. However, oAb 25-35 -HLF was again found at the injection site level where it was trapped by local cells (Fig. 1A). The fluorescent peptide was also found in brain ventricles, particularly in the lateral ventricle (LV) (Fig. 1B-C), and at the dorsal (D3V) and ventral (3V) parts of the third ventricle ( Fig. 1D-F). At this level, oAb 25-35 -HLF was found in ependymal cells bordering ventricles and in the surrounding brain structures (Fig 1B-F). The fluorescent peptide was found in the walls of blood vessels particularly in the amygdala, frontal (Fig. 1G) and parietal cortex, but also in hypothalamus and thalamus (Fig. 1H) regions. As previously reported, in addition to ependymal cells, oAb 25-35 -HLF was found in neurons ( Fig. 1H-I) and in glial cells, particularly in Figure 6. ER stress. Variations in pro-and activated caspase-12 levels in the frontal cortex, amygdala, hippocampus and hypothalamus, determined in rats by western blot 6 weeks after oAb [25][26][27][28][29][30][31][32][33][34][35] icv injection (10 mg/rat). Pro-caspase-12 (50 kDa) and activated caspase-12 (25 kDa) variations were normalized with b-tubulin (b-tub, 55 kDa) variations and compared with untreated rats (control group: C). The results are expressed as means 6 SEM. *p,0.05 and **p,0.01 vs. control group, +p,0.05 and ++p,0.01 vs. scrambled treated rats. Note that scrambled peptide injection (10 mg/rat) served as negative control and did not induce any modifications in pro-and activated caspase-12 levels. The number of animals in each group is indicated within the columns. doi:10.1371/journal.pone.0053117.g006 the median eminence (Fig. 1E), but also in nerve fibers, principally in the hippocampus (Fig. 1I).

4-days Incubation of Ab25-35 Led Almost Exclusively to Soluble Oligomers
In addition to a previous qualitative study [20], where we have evidenced the amyloid property of the aggregated Ab 25-35 peptide, we have determined in the present work the respective quantity of each particle species contained in the solution injected in rats. The characterization of the aggregated Ab 25-35 solution was realized by PCS and is presented in Figure 1J. This figure shows the size distribution of the oligomeric species in the aggregated Ab [25][26][27][28][29][30][31][32][33][34][35] solution. After 4-days incubation at 37uC, the sample is composed of particles with an average diameter of 103.4 nm (weighted average, min/max: 50.7/712.4 nm) (red curve). In order to determine the size of particles that populated the aggregated Ab 25-35 solution, a low-speed centrifugation was carried out at 1 000 g for 15 min. PCS analysis of the pellet resuspended in water (purple curve) indicated that the particle size extended from 295.3 to 825 nm with a peak maximum at 458.7 nm and an average diameter of 479.5 nm. For the supernatant (green curve), the weighted average diameter of particles was 60.0 nm with a peak maximum at 50.7 nm. Similar size of particles (weighted average diameter of 52.8 nm) was measured in the supernatant after centrifugation at 16 000 g (blue curve). Note that these particles had a size with at least one order of magnitude higher than the monomer form of Ab 25-35 peptide in HFIP (black curve). This suggests that the aggregated Ab [25][26][27][28][29][30][31][32][33][34][35] solution that has been used for icv injection (red curve) is mainly composed of a mixture of soluble oligomer species whose sizes extended from 52.8 to 295.3 nm (98.1%). Some high-density aggregates with a diameter greater of equal at 458.7 nm were also detectable but in low proportion (1.8%).
The spatial reference memory was analyzed using a water-maze procedure. When rats started training 6 weeks after peptide injection (Fig. 2D), acquisition profiles decreased with training. Two-way repeated measure ANOVA showed an effect of training trials and a treatment effect: F 2,260 = 28.7, p,0.0001 for the treatment, F 4,260 = 131, p,0.0001 for trials and F 8,260 = 3.96, p,0.001 for the interaction (Fig. 2D). These data outlined an alteration of acquisition performances. However, when animals were submitted to the probe test (Fig. 2E), oAb 25-35 -treated rats showed a preferential presence in the training quadrant similarly as scrambled peptide-treated and control rats. The peptide injection therefore only slowed down place learning acquisition but did not impeded it.
Increased GFAP immunoreactivity, indicative of astrogliosis was noted, essentially throughout the amygdala and hypothalamus (Fig. 8E), while no modification was observed in the parietal cortex and CA1 region of the hippocampus and a decrease was noted in the frontal cortex and other hippocampal regions (DG and CA3). Scrambled peptide icv injection did not induce any astrogliosis modification in the structures of interest (Fig. 8E).

Long-term AD-like Toxicity after oAb25-35 Peptide Injection in Rats
The main finding of this study is that a single icv injection of oAb [25][26][27][28][29][30][31][32][33][34][35] provoked important physiological, behavioral, biochemical and morphological alterations 6 weeks after injection. The results revealed a clear similarity with numerous relevant signs of the pathology and were in line with the amyloid cascade hypothesis, while also suggesting the possible involvement of soluble oligomeric Ab fragments in the etiology of AD [3,44].
The impact of oAb 25-35 on cholinergic neurons observed at earlier stages [20] was maintained after 6 weeks. VAChT immunoreactivity was decreased in the hippocampus, parietal cortex and basal nuclei of Meynert, but not in the hypothalamus. The cholinergic deficits induced by oAb [25][26][27][28][29][30][31][32][33][34][35] injection therefore seemed consistent with the well-characterized pathological hallmarks described in AD [54,55]. An effect that could be explained in part by the high levels of circulating glucocorticoids evidenced here that was shown to increase Ab 1-42 and NMDA-induced neurodegeneration in cholinergic neurons from the nucleus basalis in rat [56].

Involvement of Soluble Ab Short Fragments in the AD Physiopathology
For the first time in this study, we have characterized the particles composition of the aggregated Ab 25-35 solution injected icv in rats. Indeed, while in previous study [20] we detailed qualitatively each component of the injected solution, here we showed that the majority of particles (more than 98%) were small soluble particles, suggesting that the toxicity observed after the icv injection of this peptide (oAb [25][26][27][28][29][30][31][32][33][34][35] ) could be due to a mixture of soluble oligomers. However, in a previous study [25], we have showed that the injection of a non-aggregated solution of Ab [25][26][27][28][29][30][31][32][33][34][35] induced less toxicity than the aggregated solution, suggesting that bigger particles would be necessary to the toxicity. This hypothesis is reinforced by a previous result using an electronic microscopy approach and showing that bigger particles (amyloid fibrils) seemed to stabilize the smaller soluble particle forms [20].
In addition, the long-lasting presence of oAb 25-35 -HLF tagged peptide in the brain not only showed that such short fragments have in vivo, an important lifespan within brain tissues, but also strongly suggested that they could participate in the maintenance of the toxicity as observed here.
Six weeks after oAb [25][26][27][28][29][30][31][32][33][34][35] injection, APP processing, and particularly the amyloidogenic pathway (C99 levels), was increased in the frontal cortex, amygdala and hippocampus. This long-term effect of oAb [25][26][27][28][29][30][31][32][33][34][35] could contribute to the global toxicity always observed after 6 weeks. Interestingly, the increase in C99 level is accompanied with a decrease in APP level in the hippocampus. In the other structures examined, both APP and C99 levels were concomitantly increased. This difference could be due to a specific activation of BACE in the hippocampus, as previously suggested [82], while, in the other structures, oAb [25][26][27][28][29][30][31][32][33][34][35] could only induce an increase in APP expression and processing. Furthermore, as previously discussed [56,62,63], the relation between the high corticosterone levels induced by oAb [25][26][27][28][29][30][31][32][33][34][35] and the activation of amyloidogenic pathway must therefore be further analyzed to clarify the contribution of glucocorticoids deregulation in the amyloid toxicity and more largely in AD etiology. oAb [25][26][27][28][29][30][31][32][33][34][35] injection also modified Tau phosphorylation. Previous studies showed an increase of Tau phosphorylation up to 3 months after intra-amygdala injection of Ab [25][26][27][28][29][30][31][32][33][34][35] or 4 weeks after icv injection [32,83]. In these studies, the authors did not perform the distinction between the different phosphorylation epitopes of Tau. Here, we used two antibodies directed against AT8 and AT100 epitopes, both considered as markers of AD-related Tau phosphorylation [84]. We evidenced clearly a difference of sensitivity to oAb 25-35 among the brain regions considered. An increase of AT8 phosphorylation was noted only at the frontal cortex and amygdala levels, while it was decreased in hippocampus and unchanged in the hypothalamus. Concomitantly, in the same rats, the AT100 phosphorylation was increased in the amygdala and hippocampus and unchanged in the frontal cortex and hypothalamus. In fact, one explanation of these differences of sensitivity to amyloid peptide comes from recent study, where the authors evidenced an intrinsic specific regulation of Tau phosphorylation. Indeed, it seems that Tau phosphorylation occurred in a sequential order of events and that feedback mechanisms exist within neurons where the phosphorylation of certain sites would induce the dephosphorylation of other sites, in order to constantly maintain a phosphorylation level [85]. Thus in our study, it could be suggested that each brain regions could be at a different stage of Tau phosphorylation, and for instance at the hippocampal level that the phosphorylation decrease observed of the AT8 epitope could be under a negative feedback loop exerted by the phosphorylation of the AT100 epitope shown in this structure 6 weeks after oAb [25][26][27][28][29][30][31][32][33][34][35] injection.
There is no doubt that progressive Ab accumulation contributes to the AD pathology and that extracellular amyloid deposits are a hallmark of AD. However, the view that the mature amyloid fibril was the only toxic species of Ab is now challenged. Indeed, experimental studies have shown for a range of peptides and proteins that amyloid fibril formation is preceded by the appearance of organized molecular assemblies, usually termed protofibrils [86]. In addition, detailed biophysical studies are currently identifying the formation of smaller oligomeric species at earlier stages of the aggregation process, in vitro, in animal models and in patient brains. In fact, it appears highly conceivable that amyloid deposits may only be one aspect of a larger pathological cascade and indirect consequences of protective responses geared towards sequestering toxic soluble Ab molecules within plaques, from which oligomeric toxic fragments could be released by proteolysis [4,9,10,87]. The peculiar potent aggregative ability and neurotoxicity of oAb [25][26][27][28][29][30][31][32][33][34][35] , its capability to induce de novo the Ab 1-40/42 protein synthesis and the abnormal phosphorylation of Tau, now even discovered physiologically to be present in elderly people [14][15][16], reinforce its potential involvement in the pathogenesis of AD.