Enhanced Aβ1–40 Production in Endothelial Cells Stimulated with Fibrillar Aβ1–42

Amyloid accumulation in the brain of Alzheimer’s patients results from altered processing of the 39- to 43-amino acid amyloid β protein (Aβ). The mechanisms for the elevated amyloid (Aβ1–42) are considered to be over-expression of the amyloid precursor protein (APP), enhanced cleavage of APP to Aβ, and decreased clearance of Aβ from the central nervous system (CNS). We report herein studies of Aβ stimulated effects on endothelial cells. We observe an interesting and as yet unprecedented feedback effect involving Aβ1–42 fibril-induced synthesis of APP by Western blot analysis in the endothelial cell line Hep-1. We further observe an increase in the expression of Aβ1–40 by flow cytometry and fluorescence microscopy. This phenomenon is reproducible for cultures grown both in the presence and absence of serum. In the former case, flow cytometry reveals that Aβ1–40 accumulation is less pronounced than under serum-free conditions. Immunofluorescence staining further corroborates these observations. Cellular responses to fibrillar Aβ1–42 treatment involving eNOS upregulation and increased autophagy are also reported.


Introduction
Alzheimer's disease (AD) is characterized by neuronal degeneration and accumulation of senile plaques, which are composed of amyloid-b (Ab) peptides predominantly consisting of 40 and 42 amino acids [1,2]. The excessive accumulation of Ab peptides in AD may be due to enhanced endoproteolytic cleavage of membrane bound amyloid precursor protein (APP), overexpression of APP and/or decreased clearance of Ab from the central nervous system (CNS) [3][4][5].
Postmortem analyses of AD subjects reveal that amyloid plaques in the brain suffuse vascular cells in addition to the parenchymal. The implications of this vascular infiltration for AD has been less well studied than the parenchymal Ab, but has generated considerable interest with studies that b-amyloid fibrils accumulate in small arteries, arterioles and capillaries of the brain [6][7][8]. Cerebrovascular amyloid toxicity generally manifests itself in the breach of the blood-brain-barrier and enhanced inflammation in the cerebrovasculature [9,10].
The mechanism for the onset of pathological vascular changes has yet to be elucidated [11]. Two mechanisms that have been proposed involve: (1) The production of excess superoxide by amyloid-b induced oxidative stress [12,13] and (2) the formation of amyloid aggregates whose resistance to protease degradation turns them into cellular ''tombstones'' that impair blood flow and cellular function [14,15].
The oxidative stress mechanism was used to explain an in vitro investigation where the interaction with Ab fibrils resulted in the endothelial lining of the rat aorta undergoing rapid damage leading to exposure of smooth muscle cells and connective tissue [16]. In agreement with these observations, it has been shown that antioxidant treatment and superoxide dismutase (SOD) treatment can reduce damage of endothelial cells caused by amyloidb [17,18].
The ''tombstone'' mechanism is consistent with biochemical and biophysical studies of synthetic Ab peptides indicating that the more toxic Ab peptides ending at residue 42 aggregate more rapidly than peptides of 39 or 40 amino acids [19][20][21][22]. This feature of Ab 1-42 makes it less susceptible to proteolytic degradation [23][24][25]. Further support for this mechanism comes from studies on the APP mutation found in HCHWA-Dutch type, which results in the production of Ab with enhanced tendency to aggregate relative to that of wild type Ab. Fibril formation in this mutation is limited to the cerebrovasculature and amyloidosis leads to cerebral hemorrhage [26,27]. APP synthesis and processing to Ab normally takes place only to a limited extent in endothelial cells [28]. It has, however been shown that amyloids can alter the expression pattern of specific proteins. Thus, the accumulation of Ab 1-42 in lysosomes down regulates the catabolism of APP resulting in enhanced production of amyloidogenic fragments [29]. Production of individual isoforms of Ab induced by the same isoform has been shown in smooth muscle cells [30]. On the basis of these studies we investigated the possibility that amyloid fibrils can induce synthesis of more APP and amyloid of its other isoforms. Such a process would provide a synergistic mechanism whereby amyloid fibrils in circulation potentiate damage to the blood brain barrier endothelium. For this purpose we used synthetic, preformed Ab 1-42 fibrils and established the resultant accumulation of APP and Ab 1-

Preparation of Amyloid Fibrils
Synthetic Ab 1-42 was disaggregated by pretreatment with trifluoroacetic acid (TFA) followed by treatment with trifluoroethanol (TFE) three times to remove traces of TFA. After each step, solvents were evaporated to form a film. Ab 1-42 fibrils were formed from disaggregated Ab 1-42 by following a known procedure [31]. Briefly, Ab 1-42 (250 mM) was dissolved in phosphate buffered saline, (PBS), pH 7.5, and polymerized in Eppendorf tubes (1.5 ml), at 37uC for 2 days. Newly formed fibrils were then separated from monomeric peptides by centrifugation at 4uC for 90 min at 15,000 rpm, using a Sorvall RC-5B centrifuge. The fibril pellet was re-suspended in PBS and used for in vitro studies [31,32].

Cell Culture and Flow Cytometry
Human endothelial (Hep-1) cell line is a kind gift from Dr. John Kusiak at the National Institute on Aging. This is an immortal human endothelial cell line derived from an adenocarcinoma of the liver [33]. Cells were grown in DMEM medium containing 10% fetal bovine serum (FBS Prime, Biofluids), 16MEM nonessential amino acids, 2 mM L-glutamine, 100 units/ml penicillin, and 100 mg/ml streptomycin in a humidified atmosphere at 37uC in the presence of 5% CO 2 . 1610 6 cells were incubated with varying concentrations of Ab 1-42 fibrils in the presence or absence of FBS for 12 hrs. Cells were then harvested, washed with PBS, fixed in 2% glutaraldehyde, and probed with biotinylated monoclonal antibodies against Ab 1-40 . This antibody detects low molecular weight Ab 1-40 , but not fibrils. Cells were counterlabeled with 20 mg/ml phycoerythrin-streptavidin for 15 to 30 min. Labeled cells were analyzed on a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose CA) equipped with a 15 mW argon laser. Phycoerythrin labeled cells were excited with 488 nm light from the argon laser and emitted fluorescence was collected using the FL-2 (585/42) bands pass filter. A total of 20,000 events were analyzed for each sample. The amount of Ab 1-40 present in the endothelial cells was determined by mean fluorescence values. Light scatter signals were also collected in the forward and side scatter detectors. Experi-ments were repeated three times. The mean and standard error of the mean values were calculated.

Fluorescence Microscopy
Cells grown on glass cover slips were incubated at 37uC for 12 hrs, and then washed with PBS. Cells with and without Ab treatment were subsequently fixed in 2% glutaraldehyde for 10 min and washed with 100 mM glycine in PBS containing 2% bovine serum albumin. After washing, samples were incubated for 1 hr with a biotin-conjugated anti-Ab 1-40 antibody (1:500 dilution) at room temperature, washed with PBS and stained with FITCstreptavidin conjugate. Immunofluorescence images were captured with an Axiovert S100 microscope (Zeiss, Munich, Germany). Images were recorded at the same exposure time using a digital SPOT camera (Diagnostic Instruments, Sterling Heights, MI).

APP Upregulation by Fibrillar Ab 1-42
Ab 1-42 fibrils were prepared from synthetic Ab 1-42 peptides and morphologically characterized by TEM (Fig. 1A). Western blotting was used to determine whether Ab 1-42 fibrils can increase synthesis of APP. For this purpose human endothelial cells (Hep-1) were incubated with Ab 1-42 fibrils for 12 hrs and then harvested in a lysis buffer for Western blotting. As shown in Fig. 1B and C, the amount of APP was increased in cells incubated with Ab 1-42 in a dose dependent manner.

Ab 1-40 Production in Amyloidic Condition
The association of the increased APP synthesis with an increase in Ab was investigated in the same cell model. It was possible to discriminate between newly synthesized amyloids and the Ab 1-42 used to treat Hep-1 cells by using antibodies that are specific for Ab  , which only detects low molecular weight Ab 1-40 , but not fibrils. The latter sequence accounts for about 90% of newly processed amyloid but no cross reactivity between Ab 1-42 and Ab 1-40 was observed, in agreement with previous reports [34].
The flow cytometry histograms, of cultures grown with and without FBS ( Fig. 2A and B, respectively), show the amyloid-induced accumulation of Ab 1-40 , which is further accentuated by the stress associated with the absence of serum. In both cases, it is clear that Ab 1-42 fibrils increase Ab 1-40 synthesis in a concentration dependent manner as shown by the plots of the mean fluorescence in Fig. 3. The concentration dependence further highlights the effect of stress associated with the absence of serum. Thus, in the presence of serum a gradual raise in fluorescence takes place over the range of amyloid concentrations tested (Fig. 3A), while more dramatic changes occur in the absence of serum (Fig. 3B) with the relative fluorescence increasing from 250 to 800 at the lowest concentration tested (1 mM). At higher concentrations only minor additional increases in fluorescence are detected.
Because the antibody used only detects low molecular weight amyloid, the observed Ab 1-40 cannot be derived from cleavage of added Ab 1-42 fibrils. To confirm that a cleaved fibril cannot be detected, Hep-1 endothelial cells were treated with 20 mM protofibrillar Ab 1-40 (10 hrs aged peptide) and analyzed by flow cytometry. Only a negligible increase in Ab 1-40 was observed and the fluorescence histogram was similar to that of control (data not shown).

Cellular Morphology
Flow cytometry provides scattering data in addition to the fluorescence. The intensity light scattered at small angles (0.5-2.0u) from the incident laser beam (referred to as forward scattering) is proportional to the cell size. The side scattering at larger angles is a measure of structural changes such as granulated structure on the membrane or cytoplasm that increase the scattering. Examination of changes in scattering properties of Ab 1-42 treated cells revealed no difference in the forward scattering consistent indicating no change in cell size. On the other hand, changes in the granulated structure were apparent as evidenced by an increase in the intensity of side scattering (Fig. 4), which was even more pronounced in the absence of serum.
Morphological changes as well as increased production of Ab 1-40 in Ab 1-42 treated cells were further corroborated by light and  confocal microscopy (Fig. 5). As found by flow cytometry, no effect was observed by microscopy when the endothelial cells were treated with Ab 1-40 instead of Ab 1-42 ( Fig. 5A and B). In contrast, shape changes in some of the cells as well as numerous intracellular and superficial granular patches (tentatively attributed to big amyloid aggregates) were clearly visible by light microscopy when cells were treated with Ab 1-42 fibrils (Fig. 5C and D). Treated and non-treated endothelial cells were stained with antibodies against Ab 1-40 and observed under confocal microscopy (Fig. 5E, F, G, H). A fluorescence signal was only detected in those cells that were exposed to Ab 1-42 ( Fig. 5G and H).

Ab 1-42 Treatment Affects eNOS Expression and Autophagy
To further study the cellular response to Ab 1-42 treatment, we examined the protein levels of the nitric oxide synthases (NOS) which catalyze the production of NO, a known modulator of CNS vascular and neuronal function. After the cells were treated with 5 and 10 mM fibrillar Ab 1-42 , Western blot of different NOS isoforms showed a mild but statistically significant increase of endothelial NOS (eNOS) and no change in neuronal NOS (nNOS) level ( Fig. 6A and B). Inducible NOS (iNOS) was not detected with or without treatment (data not shown). Since caveolin-1 is shown to interact with eNOS within endothelial plasmalemmal caveolae and affect eNOS function [35], we also examined caveolin-1 protein level but found no change (Fig. 6A  and B). Since autophagy is induced under conditions of cell stress and shown to be involved in APP processing [36], we studied the state of autophagy after Ab 1-42 treatment. Increased ratio of LC3B-II/I (Fig. 6A and C), a reliable marker of autophagosomes, indicated activation of autophagy.

Discussion
Major attention has been devoted to the role played by neuronal cells in AD development. Most of the amyloid found in AD subjects is produced in the brain and contained in plaques with a high content of highly aggregated amyloids. However, cognitive function does not generally correlate with the level of cerebral plaques. It was only in recent years that the contribution of the cerebrovascular system has been appreciated [37][38][39]. In this respect, it should be noted that a large proportion of AD patients have cerebral amyloid angiopathy (CAA) involving vascular amyloidosis within the cerebral circulation. It has in fact been proposed that vascular pathology and the resultant impairment of oxygen delivery to the brain may play a primary role in the development of AD [40]. In this study, we have delineated a new role for the cerebrovascular system involving Ab 1-42 fibril induced APP synthesis in the human endothelial cell line Hep-1.
We first demonstrate that incubation of endothelial cells with fibrillar Ab 1-42 results in elevated levels of the APP (Fig. 1). Using an antibody specific for Ab 1-40 , we also report an increase in the concentration of Ab 1-40 peptides. This finding is indicated both by flow cytometry (Fig. 2 and 3) and confocal microscopy (Fig. 5E, F,  G, H). The increased levels of APP together with increased Ab 1-40 cannot be explained just by altered processing of APP and requires increased synthesis of APP by the endothelial cells.
Such an effect would seem to require that the Ab 1-42 fibrils are endocytosed by cells. Evidence for amyloid uptake and transcytosis in endothelial cells was provided more than a decade ago [41]. Among the potentially physiologically relevant amyloid receptors that can be involved in amyloid uptake, the RAGE (receptor for advanced glycation end products) [42][43][44] was shown to bind Ab 1-40 even in the fibrillar form with a high affinity [45,46]. Support, from our data, for a contribution of stress to the observed increase in APP and Ab comes from the augmentation of the effects under serum-deprived conditions [47]. An analogous stimulation of amyloid production has been reported [48] in serum-deprived human primary neuron cultures.
Altered cellular function can be triggered by the aggregation state of the amyloids in addition to stress. Our observation that only Ab 1-42 fibrils but not Ab 1-40 fibrils induce cellular changes implies that this phenomenon is triggered by the highly aggregated fibrillar form that is favored by the Ab peptide ending at residue 42 [49]. Additional studies will be required to explain how the cellular changes produced by stress and/or amyloid aggregation induce APP synthesis and its processing. Two processes induced by exogenous Ab fibrils are likely to be involved: inflammatory response and oxidative stress. As for increased APP synthesis, we have previously shown that cellular uptake of fibrillar Ab induces interleukin-1a (IL-1a) expression [50]. The link between IL-1a and APP can be found in an earlier study which showed an IL-1a  [51]. Interestingly, the APP mRNA level was also unchanged after the Hep-1 cells were treated with Ab 1-42 (Rajadas unpublished data) indicating the increase of APP could also be translational in our study. As for increased Ab production, there is ample evidence that in neuronal cells, Ab accumulation and oxidative stress, each accelerating the other, generate a vicious circle of more Ab production and oxidation [52]. Based on our observation, it is possible that the similar process is also taking place in endothelial cells. This notion is supported by an Ab immunotherapy study which indicated that Ab depositions in the brain parenchyma and blood vessels occur independently [53].
Once inside the cell, the amyloid can affect many other cellular responses. One of the cellular responses we observed is the upregulation of eNOS (Fig. 6) which is consistent with a previous study showing a Ca 2+ dependent induction of eNOS by hydrogen peroxide in endothelial cells. This upregulation has been attributed to a compensatory mechanism for increased superoxide formation [54,55]. Elevated eNOS expression has been observed in cardiovascular and other vascular pathologies, wherein increased levels of ROS have been detected [56]. Another cellular response we observed is increased autophagy (Fig. 6). Since autophagy is involved in APP processing as a protective mecha-nism [36], this response is expected and also explains the increased Ab 1-40 production. The intricate relations between oxidative stress and autophagy and its implication in AD vascular pathogenesis awaits further elucidation.
The increased cellular granulation indicated both by changes in side-scattering in the flow cytometry experiment (Fig. 4) as well as by microscopy (Fig. 5) seem to indicate damage to the cells. Thus, coupled with the altered cellular function, and the new synthesis of APP and its metabolism, amyloid toxicity results in damage to the endothelial cells.
In conclusion, our observation of increased APP and amyloid in ECs incubated with Ab 1-42 fibrils establishes, for the first time, that the exposure of endothelial cells to Ab 1-42 fibrils results in an elevated amyloid load in endothelial cells. This process provides a new potential pathway for amyloidosis with these newly formed amyloids in the endothelial cells able to contribute to both vascular amyloidosis and/or parenchyma amyloidosis.