[1] 230w Small volumes of solutions injected into the hippocampus produce dramatic degeneration in dentate gyrus neurons, but not in neurons of the CA1 subfield. The aim of the present study was to ascertain whether solutions with different fragments of the β-amyloid protein (Aβ) could produce further degeneration in areas beyond the dentate gyrus. It was found that 5 days after injection of an aqueous solution containing the Aβ 1-40 fragment into the hippocampus, long stretches of the CA1 subfield were either deprived of neurons or most of the neurons were degenerating. By contrast, in animals with deposits containing Aβ 1-28, Aβ 1-42 or water, neuronal degeneration or depletion only occurred in a reduced area around the place where the implant needle penetrated the CA1 subfield. In animals injected with Aβ 1-40, many profiles in the CA1 subfield and dentate gyrus were undergoing apoptosis, as seen using preparations processed by routine histology or the TUNEL technique for detection of fragmented DNA. In addition, there was higher infiltration by ED1-positive, activated microglia-macrophagic cells in Aβ 1-42 deposits than in deposits of Aβ 1-40. The present results suggest that the intrahippocampal injection of toxic Aβ fragments produces neuronal degeneration in the rat CA1 subfield when using the appropriate protocol, and, thus, can provide an in vivo model to investigate the neurotoxic effects of Aβ and for the evaluation of drugs with potential anti-neurodegenerative activity.

[1] 599w The amyloid component of the senile plaques that appear in the brains of Alzheimer's disease (AD) patients is composed mainly of the β-amyloid protein (Aβ). This protein consists of 39-43 amino acids, and is a byproduct of the proteolysis of a larger membrane protein, the β-amyloid precursor protein (βAPP). The abundance of Aβ in heavily affected areas of the AD brain led to the hypothesis that this protein might contribute to the pathogenesis of AD [29,30], i.e., to the marked neuronal degeneration that occurs in the AD brain. This hypothesis was much reinforced when investigators from various laboratories localized families with inherited AD that was associated with a point mutation in the βAPP gene mapped on chromosome 21 [13]. This prompted other researchers to obtain direct evidence of neurodegenerative effects of Aβ fragments in vitro and in vivo. Using neuronal cultures of hippocampus, cortex or cerebellum, several studies have clearly established that Aβ, specially when aggregated into fibrils, produces degenerative alterations of the neurites close to the aggregates of Aβ, and eventually leads to death of cultured neurons [8,22,41]. In vivo, surprisingly, the evidence that injection of Aβ fragments causes neuronal death in the hippocampus has been less compelling, and some reports even deny neurotoxic effects of Aβ fragments injected into the rat brain [11]. One of the difficulties in showing neurotoxic effects of injections of Aβ in the hippocampus has been the demonstration that specific degeneration in the lateral blade of the dentate gyrus (lbgd), attributed in early reports to the action of Aβ from injected cores of senile plaques [9], is indistinguishable from the effects of simply injecting fluids such as distilled water into the hippocampus [11,28,33,39]. Kowall et al. [18] injected Aβ fragment 1-40 (Aβ 1-40) into the hippocampus and found that neurons in the CA1 subfield showed argyrophilic features indicative of degeneration, although no information was provided as to neuronal death by apoptosis and/or necrosis. Later experiments by Kowall et al. [19] included the cortex of rats and monkeys José Javier Miguel-Hidalgo • Ramón Cacabelos β-Amyloid(1-40)-induced neurodegeneration in the rat hippocampal neurons of the CA1 subfield as targets for Aβ deposits. However, other experiments appeared to rule out that specific neuronal degeneration occurred in any of the hippocampal areas as a consequence of small injections (1-2 µl) of solutions containing Aβ fragments [11,33]. A common finding with most of the fluid injections (with or without Aβ) in those investigations was the absence of specific damage in areas of the hippocampus other than the dentate gyrus, although some damage to the CA1 subfield surrounding the injection site was mentioned by Kowall et al. [18,19]. Recently, the successful development of some transgenic mice that express mutated forms of the human βAPP [11] or the mouse homologue of Aβ alone [20] has led to the observation that endogenous deposits of Aβ are associated with pathological features consistent with the neuropathology of AD. In addition, the expression of Aβ was concomitant with signs of apoptosis such as the one detected using terminal deoxynucleotide transferase-mediated dUTP-biotin nick end labeling (TUNEL) [20]. These signs of apoptosis and neuronal degeneration were associated with the appearance of intense β-amyloid immunoreactivity in the hippocampus, cortex, and amygdala among other brain areas [20]. Apoptosis has been found to follow exposure of cultured neurons to deposits of β-amyloid [3,8,22,44] and some researchers have provided evidence for the occurrence of apoptotic cell death in AD brains [21,32,34]. In another transgenic model of Alzheimer-like plaque formation [14], accumulation of Aβ into plaques and increases of Aβ 1-40 and Aβ 1-42 were shown to be correlated to memory deficits.

[2] 64w In line with the evidence for neurotoxic effects associated with β-amyloid, we examined whether injections of fragments Aβ 1-28, 1-40 and 1-42, which are known for differential toxicity in cultured neurons, would also produce differential degenerative effects after injection into the hippocampus in vivo. In addition, we used the TUNEL assay to ascertain whether any specific degenerative effects were accompanied by signs of apoptosis.

[1] 82w In concordance with the descriptions of other researchers, the injections of water into the hippocampus caused, in many cases, an extensive degeneration of cells in the dorsal blade of the dentate gyrus (Figs. 1a, b; 2). There was local gliosis in the zone depleted of normal granule cells and in the molecular layer. Only sporadic pyknotic bodies or cells containing fragmented DNA, as assessed by the TUNEL technique, were found. A few ED1-positive microglial cells appeared along the region of cell degeneration.

[2] 33w In the CA1 subfield injections of water caused only a comparatively small lesion, restricted to the immediate vicinity of the place where the needle of the Hamilton syringe had penetrated (Figs. 1b, 3).

[3] 240w Intrahippocampal injections of the Aβ fragments resulted in many cases in a well-delimited peptide deposit (Fig. 4). These deposits were identified as containing Aβ by means of the specific monoclonal antibody 10D5 that was able to label the three different peptides used in the present study. In all cases, the deposits were located variably between the stratum lacunosum moleculare and the granule cell layer of the lbgd. Sections containing deposits of aggregated Aβ were subjected to Congo red-hematoxylin staining. The Aβ deposits, when observed under polarized light, did not show the apple-green birefringence that is characteristic of β-amyloid fibrils. Regardless of the extent of the deposit itself, these injections also produced marked degeneration of granule cells of the lbgd well beyond the limits of the deposit (Figs. 1c, 4a). The group of rats injected with Aβ 1-40 included more animals with large lesions than the other groups, and the extent of neuronal degeneration in this group was significantly larger (P < 0.04; Fig. 2) than in the group injected with Aβ 1-28. The morphological appearance of this area of neuronal loss in the dentate gyrus was not different from that in animals injected with only water when examined after hematoxylin or cresyl violet staining. Deposits of Aβ were occasionally localized along the hippocampal fissure without evidence that they directly reached the granule cells of the lbgd (Fig. 1e). In these cases, however, there was also dramatic degeneration of granule cells.

[4] 143w Besides degeneration of lbgd neurons, we found that deposits of Aβ 1-40 produced a dramatic degeneration of Fig. 2 Graph illustrating the lateromedial extension and variability of neuronal damage in the dentate gyrus of the experimental animals. Note that there is a tendency for lesions to be larger in the group injected with Aβ 1-40 than in the other groups. This tendency is significant when compared to the group injected with Aβ 1-28 Fig. 4a-d Micrographs of deposits of Aβ fragments (arrows) in the rat hippocampus immunostained with antibody 10D5. a Aβ 1-28. The section was lightly counterstained with hematoxylin to observe the cellular layers; b Aβ 1-42; c Aβ 1-40. d A section adja-cent to that in c that was processed with an antibody previously preadsorbed with Aβ 1-40. The arrowheads point to topographically comparable sites in both sections. Bar 250 µm

[5] 229w neurons in a large part of the CA1 subfield (Fig. 1d-f). Degenerating areas contrasted sharply with the remaining part of the CA1 subfield, where neurons did not show morphological evidence of cellular pathology (as assessed by hematoxylin and cresyl violet staining). Affected CA1 areas contained either no recognizable neurons or these had considerably shrunken cell nuclei (Fig. 5a) and Nissl substance staining was absent. In addition, the parts of CA1 with neuronal degeneration or loss were also completely deprived of the dendritic marker MAP2 (not shown). Since the above-detailed features of neuronal degeneration in the CA1 permitted a sharp delimitation of the degenerating portions of CA1, we were able to use the maximal longitudinal lateromedial extension of CA1 lacking large neuronal nuclei and Nissl substance staining of the soma and devoid of dendritic MAP2 immunostaining in transversal sections to estimate the importance of neuronal degeneration in the lesioned hippocampus quantitatively in each of the groups of animals. The extent of neuronal degeneration so measured in the CA1 was significantly larger in the group of animals injected with Aβ 1-40 than in any of the other groups (Fig. 3). In the group injected with Aβ 1-42, although the lesion in the CA1 subfield was always located around the track left by the needle, this lesion was also significantly larger than in the groups injected with water alone or Aβ 1-28.

[6] 120w In hematoxylin-or cresyl violet-stained preparations, the lesioned lbgd in animals of all groups contained a few pyknotic or fragmented nuclei, indicating the occurrence of apoptosis. In addition, some cells in CA1 appeared to be in the early stages of apoptosis since they contained typical condensations of chromatin in the periphery of the nuclei (Fig. 5b). In other cells, chromatin was condensed into a few large, heavily stained clumps (Fig. 6a, b). The appearance of TUNEL-positive profiles was very sparse in most animals, except in the group with deposits of Aβ 1-40 in which they were conspicuously abundant in the dentate gyrus (Fig. 5c). Pyknotic or TUNEL-positive nuclei were also relatively abundant in the CA1 area of this group (Fig. 5d).

[7] 141w Close examination of Aβ deposits in hematoxylin-Congo red-stained sections or in sections immunostained with antibody ED1 revealed a dramatic difference in the amount of cells labeled within the deposit (Figs. 7, 8). In Aβ 1-40-injected animals, fewer cells appeared within and around the deposit (Fig. 7a) than in animals injected with Aβ 1-42, where the deposits were heavily infiltrated with cells (Fig. 7c). Immunostaining with activated microglia/macrophage-specific antibody ED1 revealed that most cells in the Aβ 1-42 deposit and the few cells present in the Aβ 1-40 aggregates were activated macrophagic cells (Fig. 7b, d). The graphs in Fig. 8 show that the percentage of area occupied by hematoxylin-stained and ED1-positive profiles was significantly larger and that the density of these profiles was higher in the intrahippocampal deposits of animals injected with Aβ 1-42 than in those injected with Aβ 1-40.

[1] 344w The present study has shown that the Aβ 1-40 fragment, known for being toxic to neurons in experimental approaches in vitro, can also be toxic to hippocampal neu- rons in the CA1 subfield when deposited in the hippocampus following a specific experimental protocol. In addition, we have shown that the neurotoxicity of other fragments is lower under the same experimental conditions. We cannot rule out that with different conditions the other fragments used in this study also would produce larger degeneration in the CA1 subfield. The extent of neuronal loss in the CA1 subfield produced by the Aβ 1-42 fragment was significantly larger than that caused by the Aβ 1-28 fragment. In the dentate gyrus, although all groups presented variable amounts of extensive neuronal degeneration, the lesion in the Aβ 1-40 group was significantly larger than in the Aβ 1-28 group. Since the lesion in the dentate gyrus is often large regardless of the type of fluid injected, it would not have been surprising if no significant differences were found in the extent of dentate gyrus lesion. It is probable that this is one important reason why some previous studies that used different experimental approaches for β-amyloid fragment deposition did not detect any specific effect of Aβ. Only one study [18] demonstrated specific neuronal pathology in cells of the CA1 and CA4 subfields after injection of β-amyloid into the hippocampus, although there was no description or quantification of the area affected by the degeneration. Other studies have not detected any neurotoxicity of βamyloid deposits in the hippocampus of living rats [11]. In vitro, however, it has been established that Aβ 1-40 and Aβ 1-42 are neurotoxic for cultured neurons and cause abnormal growth of their processes [22,26,42]. These toxic effects have been attributed mainly to aggregated fibrillary forms of β-amyloid, while soluble forms of those peptides do not seem to produce neurotoxicity. In our experiments we did not find evidence of fibrillary deposition of β-amyloid as assessed by birefringence under polarized light, but maximum damage appeared in animals with compacted deposits of Aβ 1-40.

[2] 308w The fact that Aβ 1-42 appeared to be less toxic to rat hippocampal cells than Aβ 1-40 is better understood in the light of the known lack of clinical signs of Alzheimertype dementia in some elderly individuals that carry abundant deposition of Aβ [4,5]. It seems that the behavior of the surrounding neuropil or the conditions in which a particular form of amyloid is deposited in the brain parenchyma might be important in the production of neuronal pathology in the neurons in contact with the deposits. On the other hand, some authors did not find specific neurodegenerative effects after injections of Aβ 1-40 into the hippocampus. The significant association of Aβ 1-40 and degeneration of CA1 neurons in the present experiments is likely related to differences in the protocol of Aβ 1-40 deposition and the differential ability of intrinsic factors of the rat neuropil to deal with differently deposited β-amyloid. In our protocol Aβ 1-40 was dissolved and left to stand for 3 h at 25 °C before storing it in the refrigerator. The second day was also kept at 25 °C for 3 h and the third day was injected into the experimental animals. It is likely that the 'aging' of Aβ 1-40 [26] in these conditions was sufficient to increase its toxicity. This aging could mean simply the time employed by the organic salt form of Aβ 1-40 (in the present case the trifluoroacetic acid salt form of Aβ 1-40, purchased from Sigma) to take on a β-structure together with the period necessary for the peptides themselves to develop their secondary and tertiary structures [6,16]. Consequently, we cannot rule out that under the appropriate conditions Aβ 1-42 would be highly toxic to hippocampal cells. In addition, the activation of cerebral microglia/macrophages in and around the Aβ peptide deposit was different according to the Aβ fragment considered.

[3] 15w Activated macrophage/microglial cells, as assessed by normal histology and immunostaining with ED1 antibody, a b

[4] 162w densely infiltrated the deposits of Aβ 1-42, while they were significantly more sparse in Aβ 1-40 and Aβ 1-28 deposits. This differential behavior of macrophagic cells when confronted with different Aβ fragments bears similarities with the fact that a single change of aspartic acid to isoaspartic acid in a residue of Aβ 1-42 is sufficient to produce the absence of complement pathway activation to this peptide [38]. Recently, Fukumoto et al. [10] have found that all senile plaques in the brains of a population of AD patients were Aβ 1-42 immunoreactive, while only a proportion of these contained Aβ 1-40-immunoreactive material. A majority of the Aβ 1-42 plaques with Aβ 1-40 immunoreactivity were rich in microglial cells, while only a few Aβ 1-42 plaques with no Aβ 1-40 immunoreactivity contained microglial cells. Whether these sequence-dependent variations in glial and immunological responsiveness to different forms of Aβ or their combinations are directly related to their differential toxic effects remains to be fully elucidated.

[5] 90w The presence of abundant profiles with characteristics suggestive of apoptosis associated with the deposits of Aβ 1-40 is consistent with the findings in transgenic mice that express murine Aβ. In these animals, Aβ deposition is concomitant with extensive TUNEL staining [20]. The findings of the present study are also compatible with the known induction of apoptosis in cultured neurons by Aβ fragments [22,41] and the demonstrated down-regulation of bcl-2, an anti-apoptotic protein, in cultured human fetal neurons following exposure to high physiological doses of Aβ 1-40 or Aβ 1-42 [24].

[6] 254w In all cases the amyloid deposit was seen to reach only the stratum lacunosum moleculare, while it never affected either the somata or the proximal parts of the main dendrites of pyramidal cells. In the animals injected with Aβ 1-40, this disposition appeared to be sufficient to cause degeneration of pyramidal cells in the CA1, which suggests that an Aβ-related signal in contact with the apical dendrites of the CA1 pyramidal cells was conveyed to their somata and triggered a program of neuronal death. One possible mechanism for the triggering of the death process could be related to the alteration of hippocampal physiology due simply to the lesion, which combined with the toxic properties of Aβ 1-40 would bring about neuronal death. Some evidence suggests that the toxicity of glutamate application or metabolic disturbance in cultured neurons is greatly increased in the presence of Aβ 1-40 [17,23,37]. Nevertheless, degenerative features in transgenic mice [20] and various in vitro studies suggest that β-amyloid, acting through specific intracellular second messenger systems [2,7,31,35,43], causes neuronal degeneration directly. Both views are compatible since common mechanisms such as intracellular calcium destabilization [1,23] and oxidative stress [2,27,36] appear to mediate Aβ neurotoxicity. Moreover, recent experiments with high physiological concentrations of Aβ [24] in neuronal cultures point to more basic mechanisms that involve down-regulation of bcl-2, an anti-death protein, and up-regulation of bax, a death-promoting protein, in the susceptibility of neurons to Aβ-induced oxidative stress or calcium destabilization and, consequently, in the causation of apoptosis and progressive loss of neuronal populations.

[1] 392w For these experiments 43 female Sprague-Dawley rats (250-275 g) were used. The animals were given a general anesthesia by i.p. injection of sodium pentothal (60 mg/kg), and local anesthesia with mepivacaine under the skin covering the skull. Rats were then placed on a stereotactic apparatus and microinjections of 2 µl were delivered into the right hippocampus. The fluid injected into the hippocampus consisted of one of the following solutions: deionized double-distilled water only, 3 nmol Aβ 1-40 in water, 3 nmol Aβ 1-28 in water or 3 nmol Aβ 1-42 in water. Twelve animals were employed for each experimental group except for that injected with Aβ 1-42 which comprised only seven animals. Injections were performed over 1 min with a 5 µl Hamilton syringe equipped with a 26S-gauge beveled needle. The hole in the tip of the needle was directed vertically down to 3.5 mm from the skull surface at 3.8 from bregma and 2 mm from the midline, following the coordinates of the atlas of Paxinos and Watson [25]. The needle was left in place for an additional 2 min and then withdrawn very slowly. The Aβ fragments (purchased from Sigma) were dissolved in water 48-72 h previously. Immediately after reconstituting the lyophilized peptide the solution was left 3 h at room temperature and then put in the refrigerator at 4 °C overnight. The next day the solution was again exposed to room temperature for 3 h and returned to the refrigerator for 21 h or longer. Finally, the solution was loaded into the Hamilton syringe and applied as explained above. The reason for this protocol was that in preliminary experiments we found that fresh Aβ solution or solutions with only 24-h preparation at 4 °C did not produce significant specific degeneration in the hippocampus. The 'aged' solutions appeared conspicuously more viscous and were slightly more turbid after the aging protocol than the freshly prepared solutions. However, we did not observe visually detectable precipitates nor did these solutions contain precipitates that stained with Congo red displaying the typical birefringence of amyloid fibrils under polarized light. Likewise, aggregates detected in the hippocampus after the injection did not show birefringence with Congo red staining. Five days after the injections, the rats were anesthetized with a lethal dose of sodium pentothal and perfused through the aorta with 4% formaldehyde in phosphate buffer pH 7.4.

[2] 71w The brains were removed from the skull and blocks were made through the hippocampus. These blocks were then dehydrated, cleared with D-limonene and embedded in plastic paraffin (Bioplast). Sections (7 µm) were cut from the blocks and mounted on gelatin-coated slides. The sections were deparaffinized, rehydrated, and subjected to either a Congo red-hematoxylin staining procedure, cresyl violet staining, immunohistochemistry or the TUNEL technique for the detection of fragmented DNA in situ.

[3] 195w Immunolocalization of the deposits of Aβ fragments was carried out using mouse monoclonal antibody 10D5 [15,40]. Deparaffinized and rehydrated sections were incubated with the Aβ antibody diluted 1:200 in 0.1 M TRIS-HCl buffer pH 7.4 containing 1% bovine serum albumin and 0.3% Triton X-100 (incubation solution). They were then washed three times for a total of 30 min in TRIS-HCl, incubated with a biotinylated anti-mouse IgG secondary antibody at room temperature for 90 min, washed again, and incubated with commercially obtained avidin-biotin-peroxidase complex (Vector) for 60 min. The complex attached to secondary and primary antibody was demonstrated using 3′-3′-diaminobenzidine tetrahydrochloride as chromogen and enhanced with ammonium nickel(II) sulfate. Preadsorption of the Aβ antibody with any of the Aβ fragments employed resulted in the abolition of specific immunostaining. For immunodetection of activated microglia/macrophages, we used the ED1 antibody (Serotec) that specifically labels a glycoprotein from membranes of activated microglial cells and macrophages. Microtubule-associated protein 2 (MAP2) was labeled using a monoclonal antibody from Sigma. The method for the detection of ED1 and anti-MAP2 was identical to that for the anti-Aβ antibody except that the dilutions for the ED1 and anti-MAP2 antibodies were 1:100 and 1:500, respectively.

[4] 179w In hematoxylin-and cresyl violet-stained sections we estimated the degree of neuronal degeneration in the lbgd or the CA1 subfield by measuring the maximum lateromedial longitudinal extent (in µm) of those layers that were deprived of neurons or that contained only small dense nuclei with loss of staining of Nissl substance in cell bodies and MAP2 immunoreactivity in dendrites. Measurements were made with the PC_IMAGE image analysis computer program from Foster-Findlay on images captured with a video camera attached to the microscope. Since in most cases the area of degeneration in the CA1 and dentate gyrus was delimited in the coronal plane by non-degenerating neurons, the length (in µm) between the non-degenerating groups of neurons was measured by taking advantage, first, of the much longer extension of the CA1 and dentate gyrus in the lateromedial direction than in the dorsoventral direction (at the rostrocaudal levels where the deposits of Aβ peptides were seen) and, second, of the easily traceable remains of the degenerated portion of the CA1 and dentate gyrus. Intergroup comparisons were carried out using the Mann-Whitney U test.

[1] 44w To detect signs of apoptosis, sections through the hippocampus were deparaffinized, washed three times in PBS, incubated with proteinase K and subjected to detection of fragmented DNA employing the commercial kit Apoptag from Oncor and following the steps and controls recommended by the manufacturer.