Aβ42-induced Increase in Neprilysin Is Associated with Prevention of Amyloid Plaque Formation in Vivo *

Brain β-amyloid plaques are principal targets for the development of treatments designed to slow the progression of Alzheimer's disease. Intracranial injections of synthetic β-amyloid peptide (Aβ42) in transgenic mice expressing the Alzheimer's disease-causing Swedish APP double mutations increased neuronal levels of neprilysin, a metalloendopeptidase that degrades Aβ42 in vivo, on mRNA and protein level. This increase was associated with significant reductions in brain levels of Aβ and with almost complete prevention of amyloid plaque formation throughout the brain. In addition, astrogliosis normally associated with amyloidosis was significantly reduced. Our results suggest that up-regulation of neprilysin in brain may represent an opportunity to reduce or prevent amyloid plaque formation in vivo.

To characterize mechanisms involved in A␤ clearance in vivo, we injected aggregated A␤ 42 into brains of 11-week-old transgenic mice that expressed the disease-causing Swedish double mutation of APP (SwAPP) under the control of the hamster PrP promoter as well as into non-transgenic littermates (18,19). At this age, our SwAPP mice had not developed amyloid plaques (18,19). We report here that A␤ 42 aggregates caused sustained increases in brain levels of NEP and that these increases were associated with dramatically reduced brain concentrations of A␤ as well as with prevention of brain amyloid plaque formation and reduced astrogliosis.

EXPERIMENTAL PROCEDURES
Animals-Transgenic mice expressing the AD-causing Swedish double mutant of the human APP (SwAPP) were generated and bred as described previously (18,19). All mice used in this study were the progeny of a single SwAPP male cross-bred to non-transgenic (non-tg) littermates. The colony was housed under a light cycle of 12 h with dry food and water ad libitum. The presence of the transgene was determined by PCR on genomic DNA isolated from tail biopsies using specific primers 5Ј-GTG GAT AAC CCC TCC CCC AGC CTA GAC CA-3Ј and 5Ј-CTG ACC ACT CGA CCA GGT TCT GGG T-3Ј to the transgene (18). Fibrillar aggregated synthetic A␤ 42 was prepared by resuspending lyophilized A␤ 42 (Bachem) in phosphate-buffered saline (PBS), pH 7.4, by shaking for 48 h at 37°C. At 11 weeks of age, SwAPP (n ϭ 5) and non-tg littermates (n ϭ 4) were anesthetized, 1 l of fibrillar A␤ 42 (350 M stock concentration) was stereotaxically injected unilaterally into the parietal cortex Ϫ1.7-mm anterior, 1-mm lateral, and 1-mm ventral to the bregma above the hippocampus within 1 min (20). After the injection, the needle was left in place for one more minute and then was slowly withdrawn over a period of 1 min. Control groups consisted of SwAPP mice and non-tg mice that received PBS injections (n ϭ 4 each). In a second set of experiments, we used C57BL/6 mice purchased from RCC Ltd. Biotechnology & Animal Breeding (Fü llingsdorf, Switzerland) at the age of 6 weeks and housed for 5 weeks in our animal facility. At 11 weeks of age, these mice received bilateral injections of A␤ 42 , reversed A␤ 42 -peptide (A␤ 42 R, Bachem, aggregated as described for A␤ 42 ), bovine serum albumin (BSA, 1.61 mg/ml), or PBS as described above and were sacrificed 1, 3, 12, and 20 weeks post-injection for combined analysis of mRNA and protein levels (n ϭ 4 for each group and time).
Tissue Preparation-For combined A␤-ELISA and immunohistochemistry, SwAPP and non-tg littermates injected with A␤ 42 aggregates or PBS were deeply anesthetized 20 weeks following the injections by intraperitoneal injection of a mixture containing 60 g/g Ketaminol (Veterina, Zurich, Switzerland) and 20 g/g Rompun (Provet, Lyssach, Switzerland) body weight and were perfused transcardially using ice-cold PBS, pH 7.4, followed by 4% paraformaldehyde and post-fixed in the same solution overnight. Tissue was washed several times in PBS, and the caudal parts of the brains reaching from interaural 0 to Ϫ1 excluding the cerebellum (20) were dissected and prepared for A␤-ELISA. The remaining parts of the brain containing both hippocampi were embedded in paraffin and processed for IHC. 5-m thick paraffin sections were probed with antibodies against A␤ (4G8 and 6E10, Serotec; 6H1 and 9G10, Evotec), NEP, IDE, angiotensin-converting enzyme (ACE), 24.15 protein, and GFAP according to providers' protocols and counterstained with Nuclear Fast Red (Sigma).
For combined Western blotting and reverse transcription-PCR, mice were killed and the frontal parts of the brain containing interaural region 6 -4 were processed immediately for RNA isolation. The rest of the brains excluding the cerebellum were homogenized in lysis buffer containing 250 mM sucrose, 10 mM Tris, 1% Triton X-100, 1% SDS, and 1ϫ Complete proteinase inhibitor mixture, pH ϭ 8 (Roche Molecular Biochemicals), for Western blotting as described previously (1). Total RNA was isolated by the peqGOLD RNAPure TM kit (Axon Laboratories) according to the manufacturer's protocol. NEP mRNA levels were determined by quantitative real time reverse transcription-PCR performed on total RNA in the LightCycler (Roche Diagnostics) using the NEP-specific primers 5Ј-TAA GCA GCC TCA GCC GAA ACT ACA A-3Ј and 5Ј-GAC TAC AGC TGC TCC ACT TAT CCA CTC A-3Ј and CY-BRGreen (Roche Molecular Biochemicals) in the reaction mixture. As a standard, mRNA levels of ␤-actin was amplified in same samples using 5Ј-TGG AAC GGT GAA GGT GAC A-3Ј and 5Ј-GGC AAG GGA CTT CCT GTA A-3Ј primers. The levels of NEP mRNA were normalized to the corresponding amount of ␤-actin mRNA in each mouse. The average was calculated for each group and compared with levels of uninjected mice for each time point.
A␤-ELISA-For quantitation of A␤ content in mouse brain, protein extracts of brain tissue reaching from the stereotaxic coordinates interaural 0 to interaural Ϫ1 were prepared. This area lies outside of the site of the brain that was injected with A␤ 42 . Tissue samples were homog-enized in a 25-fold wet weight amount of 70% formic acid, homogenates were centrifuged at 200,000 ϫ g for 1 h at 4°C, and supernatant fluids were neutralized by adding the 20-fold volume of 1 M Tris base. To measure the total content of brain A␤, microtiter plates (Maxi Sorb, Nunc, Belgium) were coated with 150 l of monoclonal mouse antibody 22C4 directed against the C terminus of A␤ 40 and A␤ 42 (amino acids 30 -42) at a concentration of 20 g/ml PBS. Plates were blocked with 1% BSA, 1% gelatin in 100 mM Tris, 5 mM EDTA, and 0.1% Tween 20, pH 7.6, for 4 h at 37°C and washed three times with PBS containing 0.02% Tween 20. 150-l diluted samples or standards (A␤ 40 , Bachem) were incubated overnight, washed, and detected with biotinylated monoclonal mouse antibody 6H1 directed against amino acids 1-17 of human A␤ (Evotec), incubated at a concentration of 1 g/ml in blocking buffer for 4 h at 37°C, and visualized with a peroxidase reaction using 3,3Ј,5,5Ј-tetramethylbenzidine as substrate and detected at 450 nm. The linear range (r 2 Յ 0.99) of the used ELISA system was between 2.5 and 100 ng of A␤/ml. Serial dilutions of samples were used for measurements in the linear range. In parallel, the concentrations of A␤ 42 in each sample were determined using a commercially available kit (Innogenetics) according to provider's instructions.
Titer Assays-Blood was taken from the tail veins at the times of injections and before sacrificing the mice. Anti-A␤ antibody titers were measured by incubating of serial dilutions of blood sera on A␤ 42 -coated microplates that were blocked with 100 mM Tris-HCl, pH 7.6, 5% milk powder, and 0.1% Tween-20 and washed three times with PBS containing 0.02% Tween 20. Serum samples were diluted in blocking buffer and incubated overnight at 4°C while shaking. Serial dilutions of monoclonal anti-A␤ antibody 6E10 were used as positive control. Plates were washed four times and incubated with goat anti-mouse biotinylated IgG (heavy and light) antibody (Vector Laboratories) in PBS with 1% BSA for 2 h at 37°C. Plates were then washed four times and incubated with peroxidase-conjugated streptavidin (Jackson Laboratories) for 30 min at room temperature. After five washes, plates were incubated with 3,3Ј,5,5Ј-tetramethylbenzidine for 5 min at room temperature, and the reactions were stopped with 1 M sulfuric acid. Optical densities were read at 450 nm in a microplate reader (Victor2 Multilabel, EG&G Wallac, Turku, Finland).
NEP Peptidase Activity Assay-Brain protein extracts were dialyzed against 4000 volumes of PBS for 24 h at 4°C by using Slide-A-Lyzer dialysis cassettes (Pierce) with a 10-kDa cut off. Three independent samples of 10 g of protein from each brain were incubated for 1 h at 37°C with 50 M Z-Ala-Ala-Leu-p-nitroanilide (ZAAL-pNA, Bachem) in 50 mM HEPES buffer, pH 7.2. Thereafter, 0.4 milliunits of leucine aminopeptidase (Sigma) were added to the reaction mixtures, incubated for additional 20 min at 37°C (21), and optical densities were measured at 405 nm. To inhibit NEP and related enzyme activities, 10 M thiorphan (Sigma) was added for 5 min at room temperature before the addition of ZAAL-pNA (21).
Statistical Analysis-Data were collected by investigators blinded to the genotype and treatment of the mice and were analyzed by nonparametric Mann-Whitney U tests. -tg (a, b, e, f, i, and j) and SwAPP mice (c, d, g, h, k, and l). Twenty weeks after the injections, A␤ 42 -induced increases in NEP immunoreactivity were detected in cell bodies and neurites of many pyramidal cells of the cerebral cortex (b, d, i, and k) and in the cell bodies of pyramidal neurons in hippocampus (f, h, j, and l) of the A␤ 42 -injected non-tg (b, f, i, and j) and SwAPP (d, h, k, and l) mice. In contrast, PBS-injected non-tg (a and e) and SwAPP mice (c and g) showed only weak and diffuse NEP staining in these structures (a, b, c, d, i, and k: cerebral cortex; e, f, g, h, j, and l: hippocampus). Scale: 300 and 100 m for i, j, k, and l. 42 is involved in the regulation of NEP levels in vivo, we injected 1 l of PBS or a suspension of aggregated synthetic A␤ 42 into the brains of SwAPP mice and non-tg littermates and analyzed NEP by immunohistochemistry. Strongly stained NEP-immunopositive neurons were present at 20 weeks following one single injection of A␤ 42 (Fig. 1). In these mice, NEP immunoreactivity was high in cellular compartments of many pyramidal cells of the cerebral cortex in all cortical layers (Fig. 1, b, d, i, and k) and in the cell bodies of pyramidal hippocampal neurons (Fig. 1, f, h, j, and l). In contrast, neuronal NEP staining was weak and diffused in cerebral cortex (Fig. 1, a and c) and hippocampus (Fig. 1, e and g) of non-tg littermates (Fig. 1, a and e) and SwAPP mice (Fig. 1, c and g) injected with PBS. Together these data show that A␤ 42 injections increased neuronal levels of NEP protein in most brain regions. For detailed analysis of tissue levels of NEP, we injected wild type mice bilaterally with A␤ 42 and analyzed them by Western blotting at 1, 3, 12, and 20 weeks following the injections. Control mice were injected with same amounts of A␤ 42 R, BSA, or were injected with PBS only. The levels of NEP protein were high in all injected groups 1 week after the injections (Fig. 2a). However, three weeks after the injections, 3 of 4 A␤ 42 -injected mice showed a higher NEP protein level than other injected mice (Fig. 2a). At 12 weeks post-injection, all A␤ 42 -injected mice showed higher NEP concentrations in brain tissue (4 of 4) when compared with their control counterparts, whereas at 20 weeks post-injection, 2 of 4 mice that had been injected with A␤ 42 exhibited a remarkably higher NEP protein level in the brain (Fig. 2a). NEP protein levels in the brains of PBS-or BSA-injected groups were similar to A␤ 42 R group for each time point (data not shown). The persistent elevation of NEP levels was specific to A␤ 42 -injected mice, because none of the control injected mice showed elevated NEP protein at 3 weeks post-injection or later (Fig. 2, a and b). ␤-Actin (Fig. 2, a and b) and glyceraldehyde-3-phosphate dehydrogenase (data not shown) Western blotting was performed for all groups as loading controls giving identical results. We next tested the NEP enzymatic activity in brain homogenates of mice 20 weeks after the injections using the ZAAL-pNA peptide as the substrate (21). This assay revealed significantly higher NEP activities in brains that had been injected previously with A␤ 42 and that had higher brain levels of NEP on Western blots. The NEP enzyme inhibitor thiorphan blocked these increases (Fig. 2c).

A␤ 42 Increased Brain Concentrations of NEP-To determine whether A␤
To determine whether the increases in brain levels of NEP were paralleled by increased levels of NEP message, we analyzed by real time reverse transcription-PCR the NEP mRNA prepared from same mouse brains. One week after injections, the levels of NEP mRNA were 4 -5-fold higher in all injected groups when compared with uninjected mice of the same age. This up-regulation was a consistent feature of the A␤ 42 -injected mice. At 20 weeks post-injection, the mRNA level of NEP in all other groups had dropped to the level of uninjected mice, whereas A␤ 42 -injected mice showed a 4.9-fold increase in NEP mRNA levels (Table I).
A␤ 42 -induced Increase in NEP Was Associated with Reduced Amyloid Plaque Formation-To determine whether the A␤ 42induced increase in neuronal NEP was associated with reduced amyloid plaque formation in transgenic SwAPP mice, we counted the number of brain amyloid plaques stained by immunohistochemistry at 20 weeks after one single injection of A␤ 42 . Intracranially injected A␤ 42 suspension generated 4G8immunoreactive deposits that were detectable up to 20 weeks following the injections (Fig. 3, d and e, open arrow). As ex-pected, all of the control SwAPP mice (4 of 4) tested at the end of the experiment exhibited numerous amyloid plaques that immunoreacted with the monoclonal antibody 4G8 directed against the amino acids 17-24 of human A␤ (Fig. 3, a-c). In particular, amyloid plaques were abundant throughout the cerebral cortex and, to a somewhat lesser extent, in the hippocampus. Other A␤-specific antibodies including 6E10, 6H1, and 9G10 gave identical results. In striking contrast, 4 of 5 42 and A␤ 42 R control-injected wild-type mice are shown. Levels of NEP increased in all injected mice as early as 1 week after the injections. Three weeks after the injections, 3 of 4 mice injected with A␤ 42 showed a higher NEP protein level than the control-injected mice. At 12 weeks post-injection, all A␤ 42 -injected mice expressed a higher amount of NEP in the brain. At 20 weeks after A␤ 42 injections, 2 of 4 mice showed significantly higher NEP levels. Note that only A␤ 42 -injected mice show long term elevated NEP protein levels. b, representative Western blot analysis of mice 12 weeks following the injections. The levels of NEP protein of control mice injected with either A␤ 42 R-peptide, PBS, or BSA did not differ from that of uninjected mice and was much lower than the A␤ 42 -injected mice. ␤-Actin was used to control for equal protein loading. c, twenty weeks after the injections, A␤ 42 -injected mice with high brain levels of NEP on Western blots exhibited also high NEP peptidase activity in brain protein extracts. This activity was blocked by thiorphan. NEP peptidase activity in brain protein homogenates was measured by using ZAAL-pNA as a substrate in absence or presence of 10 M thiorphan. Enzymatic NEP activities in brains of A␤ 42 -injected mice (solid bars) were higher than the activities in brains of two randomly chosen control mice that were injected with the A␤ 42 R peptide (open bars). Bars represent means Ϯ S.D. of triplicate measurements. A␤ 42 -injected SwAPP mice were free of amyloid plaques at 31 weeks of age or at 20 weeks after the injection (Fig. 3, d-f). These four mice were evaluated further in this study. In these mice, 4G8-immunoreactive material was observed only around the needle tract; this material most probably represented the previous injected aggregated synthetic peptide (Fig. 3, d-e,  open arrow).

FIG. 2. A␤ 42 caused a marked increase in NEP levels in injected brains. a, Western blots of all A␤
To quantify the numbers of amyloid plaques, we stained every fiftieth coronal section (250 m apart) of all mice beginning with a random section of the olfactory bulbs. We found only a few 4G8-immunoreactive plaques in SwAPP mice at 20 weeks after A␤ 42 injections in addition to marked staining of the fimbria and occasional vascular amyloid deposits (data not shown). Plaque counts showed a significant decrease of average numbers of 4G8-stained amyloid plaques in A␤ 42 -treated SwAPP mice as compared with control SwAPP littermates (Fig.  4a). To determine whether the reduced amyloid pathology was associated with reduced concentrations of transgenic A␤, we measured the A␤ concentrations in brain tissue by a sandwich ELISA that specifically recognized intact human A␤. No human A␤ was detected by ELISA in wild type mice. A␤ levels measured in control SwAPP mice were significantly higher than in A␤ 42 -treated SwAPP mice (Fig. 4b).
In parallel, we measured serum levels of total A␤ and A␤ 42 in SwAPP mice before and 20 weeks after the injections. We found that neither total A␤ nor A␤ 42 levels differed among these two time points in any group (data not shown).
Prevention of Amyloid Plaque Formation Was Accompanied by Reduced Reactive Astrocytosis-Because brain amyloidosis in SwAPP mice is accompanied by reactive astrocytosis, we analyzed GFAP-reactive astrocytes in response to A␤ 42 injections (Fig. 5, a-c). Twenty weeks after injections, GFAP staining of reactive astrocytes was significantly lower in brains of A␤ 42 -injected SwAPP mice (Fig. 5c) as compared with control SwAPP littermates (Fig. 5b) but significantly higher than in wild type littermates (Fig. 5a). Moreover, the clusters of GFAPpositive cells commonly found in SwAPP mice were completely absent in response to A␤ 42 injections, paralleling the absence of amyloid plaques in A␤ 42 -treated mice (Fig. 3, d-f).
Prevention of Amyloid Formation Was Independent of IgG-An IgG-mediated immune response was implicated in the prevention of amyloid formation after active immunization of PDAPP mice against A␤ (10). Therefore, we next tested our mice for the presence of endogenous immunoglobin antibodies at the amyloid injection sites and on the few plaques that had developed after A␤ 42 injections by immunochemistry with mouse-specific anti-immunoglobulin antibodies. Neither the injected A␤ 42 remaining around the needle tract nor the few 4G8-immunoreactive plaques reacted with the anti-mouse antibodies. This was in sharp contrast to peripherally immunized mice that had abundant IgG-positive plaques (this study and  42 injections A sustained overexpression of NEP mRNA was observed after A␤ 42 injections. When compared to those of uninjected group (set as 100%), the levels of NEP mRNA were significantly higher one week after the injections in all groups. 20 weeks after the injections, however, mRNA levels of the control-injected mice did not differ from those of the uninjected mice, whereas this value was still high in A␤ 42 -injected mice. *, p Յ 0.05 for comparison to uninjected group; n.s., not significant. and hippocampus (c). Twenty weeks after injection of A␤ 42 aggregates into cortex of SwAPP mice (d-f), a significant reduction of 4G8-positive amyloid plaques was observed throughout the brain (d) and in cortex (e) and hippocampus (f). Twenty weeks after the injections, the injected fibrillar A␤ 42 was still detectable by 4G8 immunostaining (open arrows). Scale: 1 mm for a and d; 200 m for b, c, e, and f.   FIG. 4. Numbers of amyloid plaques and brain levels of A␤ were significantly lower 20 weeks after intracranial injection of A␤ 42 . a, randomly chosen sections of A␤ 42 -injected SwAPP (SwAPP/ A␤ 42 ) mice and control age-matched SwAPP littermates (SwAPP) were examined for the presence of 4G8-positive amyloid plaques in randomly selected visual fields covering cortical areas equal to 5 mm 2 on 5-m thick paraffin sections. Plaque counts were significantly reduced in SwAPP mice after A␤ 42 treatment. b, total A␤ amounts were measured in formic acid extracted brain tissue. Twenty weeks following the injections, brain A␤ levels in A␤ 42 -injected SwAPP mice were also significantly reduced (*, p Յ 0.05, Mann-Whitney U test. Mean values Ϯ S.D.).
Ref. 10). We also determined serum levels of anti-A␤ antibodies in SwAPP mice and non-tg littermates collected before and 20 weeks after the injections. We found no significant changes of anti-A␤ antibody levels because of the A␤ 42 injections in any group (Fig. 6).
Although these data do not exclude a role of the immune system in the prevention of amyloid formation in our experiments, they strongly suggest that the mechanisms involved in prevention of amyloid formation differ between the intracerebral injections of A␤ 42 fibrils and the peripheral immunization protocols with added adjuvants and repeated boosts.
Role of Other A␤-degrading Enzymes-To determine whether other A␤-degrading proteases were involved in preventing amyloid formation in our mice, we also analyzed brain tissue levels of IDE, a 100-kDa Zn-metalloproteinase that is a major A␤-degrading enzyme in tissue culture (11,22). 20 weeks after the injections, the IDE-1 antibody (11) revealed very little if any differences in tissue levels of IDE in response to A␤ 42 injections (data not shown). Moreover, metalloprotease 24.15, another A␤-degrading enzyme (12), was also unchanged in response to A␤ injections. Together, these data do not exclude the possible roles of IDE and 24.15 in degrading A␤ in vivo, but they clearly suggest an important role of NEP in preventing amyloid formation in our experimental model. As an additional control experiment, we tested brain tissue levels of ACE, an unrelated neuronal Zn-metalloendopeptidase (23)(24)(25) with no known affinity to A␤ (26). The ACE levels also did not differ between treated and untreated following an A␤ 42 injection. DISCUSSION Our data show that injection of synthetic fibrillar A␤ 42 into mouse brains caused sustained increases of NEP over a period of 20 weeks was associated with a dramatic reduction in brain tissue levels of A␤ and with inhibition of amyloid plaque formation in our mice. As claimed by the amyloid cascade hypothesis (3,27), the overproduction of A␤ or the failure to remove it can lead to the formation of amyloid plaques and subsequently to neuronal damage and is crucial to the development of AD. A␤ 42 fibrils are probably formed by initial "seeding" followed by increased precipitation of additional A␤ molecules into amyloid plaques (28). However, our data show clearly that the injection of pure synthetic fibrillar A␤ 42 did not accelerate amyloid formation in SwAPP mice.
These data are possibly at variance with reports that found evidence for accelerated amyloid plaque formation after intracerebral infusion of AD brain extracts into SwAPP mice (29). This apparent discrepancy may be related to the fact that in those experiments a complex mixture of plaque-bearing human brain extracts was injected into SwAPP mice, whereas our injected material consisted of chemically pure synthetic A␤ 42 fibrils. The possibility of existence of additional amyloidogenic factors in human brain extracts is further supported by the finding that similar seeding effect, caused by human brain extracts also occurred after immunodepletion of Ͼ80% A␤ from the extracts. 2 Thus, the effects of injections of brain extracts on amyloid formation may be independent of A␤ and principally different from these described here. Amyloidogenic factors in human brain extract may include the heptapeptide spinorphin, the endogenous brain-specific NEP inhibitor, that was shown to slow A␤-degradation in vivo (1,30), thus accelerate amyloid plaque formation.
Intracranially injected A␤ can be cleared from the brain within minutes (31), possibly via receptor-mediated transport through ventricle and choroid plexus cells (32,33). These data are consistent with two observations we made during this study, a sustained astrogliosis in hippocampus along with high 4G8 immunoreactivity in the fimbria of A␤ 42 -treated SwAPP mice. Taken together, these data suggest an active role of the hippocampus and the ventricular system in clearing A␤ from the brain.
We found no evidence for an involvement of IgG in mediating A␤ clearance, because in contrast to active immunization by peripheral injections of A␤, we found no IgG in any injected brain. In addition, serum levels of anti-A␤ antibodies were unchanged before and 20 weeks after the injections. Moreover, serum levels of total A␤ and A␤ 42 were identical before and after the injections. Therefore, a decrease in brain A␤ levels in our experiments were not the result of immune responses similar to those found after intramuscular or intraperitoneal immunizations against A␤ 42 (10,34,44).
These results are in agreement with the interpretation that increased brain levels and proteolytic activities of NEP resulted in the cleavage and removal of A␤ and consequently in prevention of amyloid plaque formation in these mice. Our results are consistent with previous data showing that deleting the NEP gene in mice caused deficient cleavage of either endogenous or exogenous A␤ (35) and that elevating the expression of NEP in neurons reduced both secreted and cell-associated A␤ (36).
The mechanism that links A␤ 42 to increased NEP levels is 2 L. Walker, personal communication.
FIG. 5. A␤ 42 injections reduced astrogliosis associated with amyloid formation in SwAPP mice. Representative sections through the cortex and hippocampus are shown. a-c, 20 weeks after A␤ 42 injections, the overall astrogliosis was lower in SwAPP mice treated with A␤ 42 (c) as compared with control SwAPP littermates (b) but higher than non-tg littermates (a). Note the nearly complete absence of astrogliosis in the cortex of treated SwAPP mice (c), whereas many reactive astrocytes could be found in hippocampus. Scale: 1 mm.
FIG. 6. Serum levels of anti-A␤ antibodies did not differ significantly 20 weeks before and after the injections of A␤ 42 or PBS in brains of SwAPP mice, and they were similar to the corresponding levels in non-tg littermates.
unknown. Our data favor the possibility of a transcriptional activation of NEP gene expression in response to injections and a sustained stimulation of this mechanism by highly insoluble A␤ 42 aggregates, ensuring a long term elevation of NEP protein levels in the brain. This scenario is supported by the fact that the synthetic A␤ aggregates injected into mouse brains were remarkably stable over 20 weeks following the injection and seem to be resistant to the clearing mechanism triggered here, constantly causing the induction of the unknown signal for the NEP up-regulation. This may explain the long term increases in NEP levels in response to A␤ injections.
Altered gene expression as a response to unilateral neuronal injury is often not restricted to the ipsilateral site and is also found contralaterally (37)(38)(39). Most interestingly, an increase in the expression of NEP-related genes such as tachykinins and bradykinin receptors in response to unilateral neuronal injury is also found contralaterally to the lesion site (40,41). Therefore, increased NEP expression throughout the brain in our study may very well be initiated by the damage done to the tissue by our injections during the deposition of the fibrillar A␤ 42 and sustained because of the stability of the injected A␤ 42 fibrils in brain tissue.
NEP expression is reduced on mRNA and protein levels in AD brains, indicating a crucial role of NEP down-regulation in the pathophysiology of AD (27,42,43). Moreover, neprilysin deficiency led to reduced degradation of exogenously administered A␤ and in the metabolic suppression of the endogenous A␤ levels in a gene dose-dependent manner in mice (35). Our data suggest the possibility that the increase in NEP is related to the inhibition of amyloid plaque formation by decreasing the brain levels of A␤. As a consequence, amyloid plaque-related astrogliosis was also reduced. Further research is necessary to characterize the mechanisms of NEP function in brain amyloid formation, and the role of physiological and pharmacological regulators including spinorphin needs to be characterized. In particular, possible strategies of increasing NEP levels or activating its functions may contribute to the development of treatments designed to reduce or prevent amyloid formation in subjects at risk for AD.