C. Cecchi, Department of Biochemical Sciences, University of Florence, viale Morgagni 50, 50134 Florence, Italy Fax: +39 055 459 8905 Tel: +39 055 459 8320 E-mail: email@example.com
Considerable attention has been paid to the high cytotoxic potential of small, prefibrillar aggregates of proteins/peptides, either associated or not associated with amyloid diseases. Recently, we reported that different cell types are variously affected by early aggregates of the N-terminal domain of the prokaryotic hydrogenase maturation factor HypF (HypF-N), a protein not involved in any disease. In this study, we provide detailed information on a chain of events triggered in Hend murine endothelial cells and IMR90 fibroblasts, which have previously been shown to be highly vulnerable or very resistant, respectively, to HypF-N aggregates. Initially, both cell lines displayed impaired viability upon exposure to HypF-N toxic aggregates; however, at longer exposure times, IMR90 cells recovered completely, whereas Hend cells did not. In particular, significant initial mitochondrial permeability transition (MPT) pore opening was found in IMR90 cells followed by a sudden repair of membrane integrity with rapid and efficient inhibition of cytochrome c and AIF release, and upregulation of Bcl-2. The greater resistance of IMR90 fibroblasts may also be due to a higher cholesterol content in the plasma membrane, which disfavours interaction with the aggregates. In contrast, Hend cells, which have less membrane cholesterol, showed delayed MPT opening with prolonged translocation of cytochrome c into the cytosol. Finally, the caspase 9 active fragment was increased significantly in both Hend and IMR90 cells; however, only Hend cells showed caspase 8 and caspase 3 activation with DNA fragmentation. From our data, the different responses of the two cell types to the same aggregates appear to be associated with two key events: (a) aggregate interaction with the plasma membrane, disfavoured by a high level of membrane cholesterol; and (b) alterations in mitochondrial functionality, leading to the release of pro-apoptotic stimuli, which are counteracted by upregulation of Bcl-2.
The amyloidoses are a group of protein-folding diseases in which specific peptides or proteins, which are either incorrectly folded or unfolded, aggregate intra- or extracellularly into polymeric assemblies rich in β sheet, and are eventually deposited in tissue as amyloid fibrils [1,2]. Amyloid diseases include a number of sporadic, familial or transmissible degenerative pathologies affecting either the central nervous system (Alzheimer's, Parkinson's and Creutzfeldt–Jakob diseases) or a variety of peripheral tissues and organs (systemic amyloidoses and type II diabetes) . Since 1998, a growing number of peptides and proteins not associated with known protein deposition diseases have been shown to aggregate in vitro, under suitable experimental conditions, into fibrils that are indistinguishable from those associated with pathological conditions [3,4]. This has led to the proposal that the ability to form amyloid assemblies can be considered a generic property inherent in any polypeptide chain .
Currently, considerable attention is focused on the cytotoxic potential of small prefibrillar protein aggregates arising initially in the protein fibrillization pathway. This cytotoxic potential appears to be higher than that of mature fibrils [2,3]. These early assemblies share basic structural features that, in most cases at least, seem to underlie the common biochemical mechanisms of cytotoxicity [5,6]. Cells exposed to toxic prefibrillar aggregates apparently die as a consequence of apoptosis [7–9] or, less frequently, by secondary necrosis [10–13]. Recent studies have shown that cells experiencing prefibrillar aggregates undergo similar early biochemical modifications; these include interaction between the aggregates and cell membranes and, possibly, interaction with membrane receptors [14–16], followed by an imbalance in the intracellular redox status [13,15] and ion levels [1,17], and mitochondria impairment [9,18], together with other modifications such as lipid homeostasis. Prefibrillar aggregates of a number of peptides associated with amyloid diseases can also induce mitochondrial permeability transition (MPT) pore opening in exposed cells, allowing molecules smaller than 1500 Da to diffuse freely between the matrix and the cytosol [18–23]. These modifications can result in the collapse of the transmembrane electrochemical gradient with loss of solutes from the matrix, mitochondrial swelling, release of proapoptotic factors such as cytochrome c and AIF, and activation of procaspase 2, 3 and 9. Cytochrome c, in complex with the cytosolic factor Apaf-1 activates the caspase-dependent apoptotic pathway, whereas AIF translocates to the nucleus inducing chromatin condensation and large-scale fragmentation of DNA [23,24].
Similar modifications have also been found in cells exposed to prefibrillar amyloid aggregates of proteins that are not associated with disease, including the N-terminal domain of the prokaryotic hydrogenase maturation factor HypF (HypF-N) [5,25]. In particular, when added to the cell culture media, early HypF-N aggregates can be internalized by the cells , where they induce modifications in intracellular free Ca2+ and reactive oxygen species (ROS) levels [10–13,26], reducing the potential across the inner mitochondrial membrane. In turn, ROS trigger the intrinsic or extrinsic apoptotic pathways , or in some cases lead to cell death by necrosis [13,26]. Data on the toxicity of HypF-N prefibrillar aggregates suggest a mechanism of cell death that is possibly shared with the prefibrillar aggregates of most peptides and proteins .
Much research is currently being carried out into molecules that are able to avoid the appearance of misfolded proteins and their initial aggregates in tissue. Notwithstanding the validity of such an approach, better knowledge of the biochemical basis of cell vulnerability to protein aggregates may also provide clues to possible interventions aimed at increasing the resistance of cells to these toxic assemblies. We sought to provide information on the chain of events that leads to death in cells experiencing toxic aggregates by investigating features of the apoptotic pathways triggered in two different cell lines upon exposure to toxic HypF-N prefibrillar aggregates. Although different cell types often show similar biochemical alterations, they are variously affected by exposure to the same toxic protein aggregates, such that only specific cell populations are stressed [14,15,28,29]. Such differences in vulnerability reflect the inherent ability of any cell to disfavour aggregate interaction with the plasma membrane, and possibly other membranes, and the subsequent early modifications by using its specific biochemical equipment. This equipment includes the specific membrane lipid composition, the total antioxidant defences (TAC), the efficiency of Ca2+ extrusion membrane pumps and the energy load (ATP availability).
A recent study showed large variations in the toxic effects of HypF-N prefibrillar aggregates on a panel of cultured cell lines , leading us to rank the cell lines according to their vulnerability. This study was carried out using murine endothelial Hend cells and human IMR90 fibroblasts; these were chosen as examples of cells that are very vulnerable or very resistant to toxic HypF-N aggregates, respectively. The different vulnerability of the cell lines was associated with different plasma membrane cholesterol content, which has been shown to disfavour membrane interaction with aggregates . We found that both cell lines showed early activation of a programmed cell death following exposure to the aggregates; however, IMR90 cells were able to counteract the toxic insult and recover despite initial impairment. Details of the apparent differences in the specific apoptotic pathways in the two cell lines are also discussed.
Hend and IMR90 cells are differently impaired upon exposure to toxic HypF-N aggregates
We recently reported that different cell types exposed for 24 h to HypF-N prefibrillar aggregates are variously impaired, as assessed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT) cell viability assay . In this study, we performed a more detailed time-course analysis of the viability of two cell lines when exposed to the same aggregates: Hend endothelial cells and IMR90 fibroblasts, previously shown to be highly vulnerable and highly resistant to HypF-N toxic aggregates, respectively . Our analysis was carried out using a highly sensitive test based on Resazurin reduction by mitochondrial oxidoreductases. A significant early decrease (∼ 27%) in Hend and IMR90 cell viability was evident after 3 h exposure to the toxic aggregates. However, Hend viability was increasingly impaired over 24 h of aggregate treatment, whereas IMR90 cells had recovered completely after 24 h exposure (Fig. 1A). Hend cells were not able to recover even at longer (48 h) exposure times (data not shown). We also investigated the reversibility of the cell damage. Hend cells were exposed to toxic aggregates for different times, following transfer into aggregate-free fresh medium for 24 h. Figure 1A shows that cell damage appears to be almost completely reversible when aggregate exposure was for relatively short lengths of time (< 6 h) following cell transfer into aggregate-free medium; under these conditions, it can be assumed that cell damage is not so great that it hinders complete recovery. At longer exposure times (to 24 h) cell viability recovered only partially under our experimental conditions. In both cell lines, global cell impairment was not due to the necrosis of individual cells. In fact, lactate dehydrogenase (LDH) activity, measured in the culture media, remained substantially unchanged compared with control cells exposed to the same amount of a harmless monomeric soluble protein (Fig. 1A, inset). The differences in vulnerability seen in the two cell types is not due to differing sensitivity to the aggregates in terms of dose–response; in fact, IMR90 cells, unlike Hend cells, appeared resistant to exposure to a wide range (0.02–20 µm) of aggregate concentrations (Fig. 1B).
Early modifications of the intracellular redox status, free Ca2+ and ATP levels: possible role of membrane cholesterol
There is strong experimental evidence that oxidative stress is one of the earliest biochemical modifications in cells exposed to toxic prefibrillar aggregates [13–15]; therefore, we carried out a time-course confocal analysis of ROS production in Hend and IMR90 cells exposed to toxic HypF-N aggregates. As shown in Fig. 2, intracellular ROS levels increased over time in Hend cells, reaching a maximum at 24 h exposure, whereas in IMR90 cells ROS levels were substantially unchanged with respect to control values, showing only a negligible increase. It is widely reported that in different cell types oxidative stress matches a sharp increase in the levels of free cytosolic Ca2+. This is in agreement with our time-course analysis of the intracellular Ca2+ content of Hend and IMR90 cells exposed to HypF-N aggregates, plated on glass coverslips and fixed at various exposure times (Fig. 3). Indeed, in Hend cells we found an increase in free Ca2+ that was earlier and stronger than the increase in ROS and was followed by a partial reduction at 30 min; it then remained substantially unchanged and higher than in controls, increasing slowly between 3 and 24 h. In contrast, in IMR90 cells the Ca2+ increase was much smoother and smaller, and without an early peak. The data suggest that the more vulnerable Hend and more resistant IMR90 cells are provided, respectively, with a poorly or highly efficient biochemical machinery that is aimed at counteracting any uncontrolled increase in the levels of ROS and free Ca2+.
The differing vulnerability of the two cell lines was confirmed by the different changes in intracellular ATP content upon exposure to the aggregates. The ATP load provides cells with the energy needed to counteract early biochemical modifications, such as any imbalance in Ca2+ homeostasis, induced by exposure to prefibrillar aggregates and/or to sustain apoptosis . In IMR90 cells exposed to HypF-N aggregates for 3 h intracellular ATP was significantly decreased (to ∼ 65% with respect to control cells exposed to the same amount of monomeric soluble protein), with complete energy recovery at longer incubation times (Fig. 4A). A much stronger and more prolonged ATP depletion was seen in Hend cells treated with HypF-N aggregates, which suggests more serious mitochondrial impairment, with substantial recovery to starting values at longer exposure times (16 and 24 h). In any case, basal ATP levels were significantly higher in untreated IMR90 fibroblasts than in untreated Hend cells.
Despite controversy about the role of blood cholesterol levels and neuronal membrane cholesterol content in the pathogenesis of amyloid diseases , our recent investigation on a wide range of cell lines supports the increasingly accepted idea that membrane lipid composition is a key biochemical feature affecting protein aggregation, interactions between aggregates and cells, and the response of cells to the presence of aggregates [14,31]. According to our previous data, the differing extent of alterations in ROS and free Ca2+ in IMR90 and Hend cell lines exposed to HypF-N aggregates was inversely correlated with membrane cholesterol content. In particular, we found a significantly higher level of basal cholesterol in the resistant IMR90 cells (14.9 ± 1.6 µg·mg−1 of protein; P ≤ 0.05) than in the vulnerable Hend cells (9.0 ± 1.2 µg·mg−1 of protein). Cholesterol may increase the resistance of membranes to the destabilizing effects of aggregates by reducing the interaction between the membrane and the aggregates or by changing membrane fluidity . These effects provide a possible explanation for the different responses of the two exposed cell lines in terms of increases in free Ca2+ and ROS.
Aggregate-induced stress is associated with typical apoptotic features
It is widely reported that protein aggregates are able to interact with cell membranes thus impairing fundamental cellular processes, and eventually resulting in apoptotic or, less frequently, necrotic cell death . In our previous study on a panel of different cell lines, aggregate-induced cellular stress was associated with typical apoptotic features rather than with a necrotic pattern . A distinct feature of apoptotic cells is the exposure of phosphatidylserine (PtdSer) on the outer membrane surface. PtdSer, normally found in the inner membrane leaflet, flips to the outer leaflet during the early stages of apoptosis . We used annexin V-FITC and propidium iodide (PI) double labelling to detect PtdSer externalization and membrane integrity in Hend and IMR90 cells exposed to HypF-N aggregates. A larger fraction of Hend cells, with respect to IMR90 cells, was double stained with high annexin-V and low PI positivity, indicating, in the former, a progressive apoptotic (PtdSer exposure) rather than a necrotic (membrane rupture) outcome (Table 1). In contrast, neither cell line when treated with the monomeric soluble HypF-N displayed annexin V-FITC or PI binding until 24 h exposure. Only in a minority of cells exposed to the HypF-N aggregates was a high positivity to both annexin-V and PI observed, suggesting a very low percentage of plasma membrane ruptures (Table 1). The data agree with those reported in Fig. 1A (inset) showing a substantial lack of LDH release from both Hend and IMR90 cells exposed to aggregates.
Table 1. Annexin V assay. The data are reported as a per cent of the value determined in the total population and are means ± SD of three independent experiments.
IMR90 Apoptotic cells (%)
Necrotic cells (%)
Hend Apoptotic cells (%)
Necrotic cells (%)
0.05 versus cells treated with soluble monomeric protein.
We then sought to explain the different degree of recovery in the two cell lines shown in Fig. 1A. We therefore analysed the mitochondrial status and some apoptotic markers in our cellular models exposed for 24 h to toxic HypF-N aggregates. It has recently been reported that β-amyloids gradually impair mitochondrial structure and function via changes in membrane viscosity, energy load, ROS production and cytochrome c release . One well-described consequence of aggregate toxicity is induction of the MPT, a Ca2+-dependent process characterized by the opening of pores in the inner mitochondrial membrane and by ATP depletion . Figure 4B shows changes in the fluorescence of mitochondria loaded with calcein in the presence of Co2+, a method that allows detection of MPT . The presence of HypF-N aggregates in the cell culture media resulted in a large initial (at 15 min) decrease in calcein fluorescence in IMR90 mitochondria due to MTP opening, followed by a rapid (at 30 min), almost complete, recovery of membrane integrity. In contrast, Hend cells showed a delayed progressive decrease in calcein fluorescence. It therefore appears that in IMR90 cells, mitochondria are initially heavily affected by the aggregates, but they are able to recover rapidly; whereas in Hend cells mitochondrial involvement is delayed, but is progressively more severe and without any possibility of recovery. The data may explain the more severe loss of ATP load seen in Hend cells compared with IMR90 cells exposed to the toxic aggregates (Fig. 4A).
It has been reported that MPT opening triggers the release of cytochrome c from mitochondria, which, in turn, activates procaspase 9 and then the effector caspases that amplify programmed cell death . Under these conditions, other mitochondrial proteins, including AIF can be released . The early apoptotic steps in either cell line exposed to the toxic HypF-N aggregates were investigated using a time-course analysis of cytochrome c and AIF translocation. As shown in Fig. 5A, in Hend cells cytochrome c was significantly released into the cytosol at 30 min exposure and was maintained at significantly higher levels than controls up to 24 h exposure. In contrast, IMR90 cells showed earlier and sharper cytochrome c translocation to the cytosol followed by recovery to basal levels (Fig. 5A). Significant early, although delayed with respect to cytochrome c, AIF translocation from the mitochondria to the nuclear fraction was also seen in IMR90 cells at up to 3 h exposure, whereas in Hend cells AIF did not appear to be involved in the apoptotic response (Fig. 5B).
It is well known that cytochrome c can activate cleavage of procaspase 9 into its active fragment by forming a complex with the cytosolic factor Apaf-1 . We therefore measured the levels, in the total homogenates, of active caspase 9, a marker for the activation of the intrinsic apoptotic pathway. A sharper increase in caspase 9 active fragment was seen in IMR90 cells than in Hend cells at early exposure times (Fig. 6A). However, in IMR90 cells, caspase 9 returned to control levels after 3 h of treatment, whereas in Hend cells the caspase 9 content appeared to increase slowly, reaching significant activation only after 24 h exposure to the aggregates (Fig. 6A). This agrees with data relative to the cytochrome c translocation in both cell lines up to 24 h exposure (Fig. 5A). In contrast, levels of the caspase 8 active fragment, a marker of the extrinsic apoptotic pathway, were significantly increased after 20 min and from 1 to 16 h of treatment in Hend cells, whereas in IMR90 cells caspase 8 remained at control levels (Fig. 6B). This agrees with data showing a lower interaction of the aggregates with the plasma membrane in IMR90 cells than in Hend cells, possibly due to the different cholesterol content.
Early caspase 3 activation followed by a reduction and further increase at longer exposure times (16 and 24 h) was seen in Hend cells (Fig. 7A). This possibly reflects the initial activation, in these cells, of caspase 8, followed by later activation of caspase 9. In IMR90 cells only moderate activation of caspase 3 was found as a consequence of the modest activation of caspase 9 and the lack of activation of caspase 8 (Fig. 7A). It is known that the activated caspase 3 fragment may cleave poly(ADP-ribose) polymerase (PARP; EC 220.127.116.11), which functions primarily as a DNA damage sensor in the nucleus . Accordingly, we found that, in Hend cells, early caspase 3 activation triggered cleavage of PARP (data not shown), resulting in a significant decrease of PARP activity early and late during aggregate treatment (0.5, 16 and 24 h) (Fig. 7B). In contrast, exposed IMR90 cells showed significant activation of PARP resulting in the enhancement of its DNA repair function. Data on the caspase active fragments and the different impairment of mitochondria were further confirmed by the analysis of the antiapoptotic factor Bcl-2 in either exposed cell line. Interestingly, in IMR90 cells Bcl-2 was significantly and progressively upregulated up to 1 h of aggregate treatment and persisted at the highest levels until 16 h of treatment, whereas it was significantly reduced in Hend cells at 3 and 24 h of treatment (Fig. 8). Consequently, Bcl-2 levels were significantly higher in IMR90 cells than in Hend cells at all exposure times and increased by > 100% in cells exposed for 1–3 h. Finally, IMR90 and Hend cells exposed to HypF-N aggregates displayed a typical DNA fragmentation pattern as evaluated in terms of enrichment of histone-associated oligonucleosomes released into the cytoplasm. As expected from the susceptibility scale and from the extent of caspase 3 activation, a greater increase was found in Hend cells (224 ± 35%) than in IMR90 cells (116 ± 18%) after 24 h exposure to the toxic aggregates.
It is known that only specific cell types are impaired in tissues facing amyloid deposits [29,36] and that cell stress eventually leads to cell death by apoptosis or, in some cases, to secondary necrosis [12,37]. We previously reported that the vulnerability of different cell lines to toxic HypF-N prefibrillar aggregates appears to be related to intrinsic biochemical features of the cells . We also provided data suggesting that the choice between an apoptotic and a necrotic outcome depends on the timing and severity of mitochondria impairment . In this study, we investigated the apoptotic pathways activated in two different cell lines, Hend and IMR90, chosen as examples of cells that are highly vulnerable or highly resistant to insult by toxic prefibrillar aggregates, respectively. The differing susceptibility to the damage by the aggregates was not an artefact due to a different dose–response in each cell line, as shown by the substantial resistance of IMR90 cells to much higher amounts of aggregates than those impairing Hend cells. Both cell lines appeared significantly stressed after 3 h exposure to the aggregates. At this time, cell damage appeared substantially reversible even for the most heavily affected Hend cells; however, at longer exposure times cell recovery was increasingly less complete, indicating a progressive deterioration in cell viability. At longer exposure times, IMR90 cells recovered completely despite early activation of the apoptotic programme, whereas a significant fraction of Hend cells underwent apoptotic death at 24 h exposure. Therefore, the differing vulnerability seen in the two cell lines following 24 h exposure to the aggregates appears to be associated with the greater ability of IMR90 cells to counteract the early biochemical modifications underlying activation of the apoptotic pathway, rather than an effect of a lower sensitivity to similar amounts of aggregates.
Severe alterations in many biochemical parameters, including intracellular redox status, energy load and free Ca2+ homeostasis , as well as membrane lipid composition [14,38], appear to be key factors in favouring cell impairment or resistance to the toxic aggregates of peptides and proteins either associated [1,39] or not associated with amyloid diseases [13,14]. It is also well known that protein prefibrillar aggregates can interact with the plasma membrane of exposed cells inducing modifications in the lipid or proteolipid structure, or self-assembling into pores thus inducing alterations in membrane selective permeability . In this scenario, it is conceivable that cells endowed with higher basal antioxidant defences and efficient Ca2+ pumps are better suited to resist any increase in free Ca2+ (or other ion) and the consequent biochemical modifications .
We found that the highly vulnerable Hend cells exposed to HypF-N toxic aggregates displayed earlier and greater increases in both intracellular ROS and free Ca2+ when compared with the more resistant IMR90 cells. The early Ca2+ increase may induce ROS overproduction by speeding up oxidative metabolism to supply energy for the increased activity of the membrane Ca2+ pumps . The resulting oxidative stress may subsequently favour entry of Ca2+ into the cell with endoplasmic reticulum stress and mitochondrial impairment eventually targeting the cell for apoptotic death [40,41]. Resistance of IMR90 to aggregate damage was previously found to be significantly related to the high efficiency of these cells in counteracting early modifications of the intracellular free Ca2+ and redox status . Under our experimental conditions, both exposed cell lines displayed ATP depletion supporting mitochondria involvement; however, Hend cells, endowed with a lower basal energy load, showed much more serious and prolonged loss of ATP than IMR90 cells, indicating that the former were less suited to counteracting ion balance derangement, which may explain their higher vulnerability to apoptotic death.
The higher resistance of IMR90 fibroblasts to toxic insult by the aggregates may also result from a significant upregulation of Bcl-2. Such an antiapoptotic factor acts as an endogenous inhibitor of MPT pore opening and mitochondrial apoptotic channel (MAC) formation by Bax and Bak [42,43], resulting in the release of proapoptotic factors such as AIF and cytochrome c[1,23] and inhibition of the proteolytic processing of AIF . Interestingly, nuclear AIF was unchanged in Hend cells, suggesting that it is not involved in the apoptotic cascade. The partial release of cytochrome c not associated with AIF release found in Hend cells agrees with previous data on infrared-irradiated human fibroblasts . AIF was significantly increased in the nuclei of IMR90 cells after 3 h exposure, where it matched, although in a delayed manner, cytochrome c release. However, the release of AIF and cytochrome c was not sustained at longer exposure times, where upregulation of Bcl-2 occurred. The latter could disassemble MAC, the proposed channel allowing cytochrome c to translocate to the cytosol , thus explaining the complete recovery in mitochondrial function, which is also supported by the recovery in ATP levels, and hence cell viability.
As pointed out above, both exposed cell lines displayed early translocation of cytochrome c from the mitochondria to the cytosol. However, cytochrome c release was much higher and decreased rapidly in IMR90 cells, whereas in Hend cells it increased progressively up to 24 h exposure. Once released from the mitochondria, cytochrome c, in association with Apaf-1, is involved in the activation of caspase 9. Indeed, both Hend and IMR90 cells showed a significant increase in the caspase 9 active fragment, however, the latter occurred earlier and was higher in IMR90 cells than in Hend cells, where a sharp increase in caspase 9 activation was observed after 24 h exposure. Moreover, the extrinsic apoptotic pathway triggered by caspase 8 cleavage was activated only in Hend cells after just 20 min exposure to aggregates. It is known that activation of the effector caspase 3 occurs downstream of caspase 8 and caspase 9 cleavages in response to differing apoptotic stimuli; once activated, caspase 3 can activate caspase 9 directly in a feedback loop, and caspase 8 indirectly . Indeed, in Hend cells exposed to HypF-N aggregates the increase in caspase 3 active fragment was earlier and sharper than in IMR90 cells and was probably responsible for the late activation of caspase 8 in Hend cells. It seems likely, therefore, that the different response of either cell line to the same toxic insult can be traced to, among others, a differing interaction between the aggregates and the plasma membrane, resulting in differing activation of the apoptotic extrinsic pathway.
We have previously shown that HypF-N prefibrillar aggregates interact with the plasma membrane of Hend cells more extensively than with the membrane of more resistant cell lines, apparently due to a different lipid composition, including cholesterol . In this study, we confirmed the dependence of cell resistance on membrane cholesterol content, which is considerably higher in IMR90 cells than in Hend cells. Therefore, according to our previous results, we expect a reduced interaction between the aggregates and the plasma membrane in IMR90 cells compared with Hend cells. In IMR90 cells, reduced membrane fluidity with increased resistance to the destabilizing effects of the aggregates can also be hypothesized. These considerations may explain the lack of activation of the apoptotic extrinsic pathway and cell recovery after the initial insult; they also agree with recent findings indicating that cholesterol can modulate membrane-associated Aβ fibrillogenesis and neurotoxicity, and that decreasing the fluidity of brain lipid bilayers reduces the interaction of Aβ40 with the bilayer surface and insertion of the latter inside the bilayer itself .
Caspase 3 activation targeted exposed Hend cells to apoptotic death. In these cells, the amount of histone-associated oligonucleosomes released into the cytoplasm confirmed a significant increase in DNA fragmentation, in agreement with the severe impairment of cell viability and the increases in ROS and active caspase 3 fragment. Caspase 3 activation resulted in PARP cleavage and inactivation. In contrast, IMR90 cells appeared more resistant to DNA oxidative attack, possibly because of their higher basal antioxidant capacity  and/or reduced damage and permeabilization of the cell membrane by the aggregates.
Overall, the data support differing scenarios for the responses of Hend and IMR90 cells to the toxic aggregates. These differences can, at least in part, be traced to differing involvement of the plasma membranes with the aggregates (Fig. 9). In Hend cells, reduced membrane cholesterol content favours interaction of the aggregates with the plasma membrane  leading to membrane destabilization and permeabilization. The structural and biochemical modifications of the plasma membrane result in early and transient increases in cytosolic free Ca2+ that appears sufficient to trigger the extrinsic apoptotic pathway. In this case, the metabolic efficiency of the mitochondria appears to be impaired at late exposure times, possibly following the oxidative stress accompanying the initial high energy demand to counteract the altered membrane permeability. However, mitochondria do not release large amounts of proapoptotic factors, nor is the antiapoptotic Bcl-2 upregulated, and activation of the effector caspase 3 appears to result mainly from the early activation of caspase 8.
In IMR90 cells, whose plasma membranes are richer in cholesterol, the extrinsic pathway does not appear to be activated. Nevertheless, early involvement of the mitochondria is apparent in this cell line, with a significant but transient release of cytochrome c and, slightly later, of AIF, suggesting that a signal arising from the plasma membrane may trigger transient MAC organization in the mitochondria outer membrane. At the same time, a significant but transient activation of caspase 9 was seen with subsequent recovery to basal values (Fig. 9). The ability of these cells to recover after an initial insult can be tentatively traced to the higher resistance of a plasma membrane rich in cholesterol to the destabilizing effects of the aggregates, and to the early upregulation of Bcl-2, possibly counteracting the initial formation of MAC in the outer mitochondrial membrane.
All reagents were of analytical grade or the highest purity available. Unless stated otherwise, chemicals were purchased from Sigma (Milan, Italy). HypF-N was expressed and purified as previously described . In brief, the domain was made to aggregate upon incubation for 48 h at 0.3 mg·mL−1 protein concentration in 50 mm acetate buffer, pH 5.5 in the presence of 30% (v/v) trifluoroethanol (TFE) at room temperature. Under these conditions, granular aggregates about 4–8 nm wide and 20–60 nm long are formed; the presence of such aggregates in the preparations used for the experiments was checked by transmission electron microscopy (not shown).
Cell culture and HypF-N-aggregate treatment
Human IMR90 pulmonary fibroblasts were obtained from the ATCC (Manassas, VA), murine endothelium cells (Hend) were a kind gift of F. Bussolino (University of Turin, Italy). IMR90 and Hend cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 1.0 mm glutamine and antibiotics and 10% fetal bovine serum, (FBS) (Mascia Brunelli, Milan, Italy) or 10% FBS (Sigma), respectively. Cell cultures were maintained and used for experiments at ∼ 90% confluence in a 5% CO2 humidified atmosphere at 37 °C.
Aliquots of solutions containing prefibrillar HypF-N aggregates were centrifuged, dried under N2 to remove the TFE when necessary, dissolved in the appropriate cell media at 200 µm and immediately added to cultured cells at 2.0 µm final concentration. No microscopic differences in aggregate structures were observed following dilution in the cellular culture medium. Cells were incubated for differing times in the presence of prefibrillar aggregates or of monomeric soluble protein as a control.
Cell viability assay and the reversibility of cell damage
The cytotoxicity of the HypF-N aggregates was assessed using the CellTiter-Blue™ Cell Viability Assay (Promega, Milan, Italy) based on the reduction of the indicator dye Resazurin into Resurfin by viable cells. Cells were plated on 96-well plates (∼ 1.0 × 104 cells per well) and after 24 h HypF-N prefibrillar aggregates (2.0 µm) were added to fresh culture media for differing times (0.5, 1, 3, 6, 16, 24 h). Hend and IMR90 cells were also exposed to various concentrations (0.02, 0.2, 2.0 or 20 µm) of prefibrillar HypF-N aggregates and to the soluble monomeric protein for 24 h. To analyse the reversibility of cell impairment the cells were incubated with culture medium containing 2 µm prefibrillar aggregates for differing times (0.5, 1, 3, 6, 16, 24 h), washed, and cultured in fresh culture medium for 24 h. After aggregate treatment, 100 µL CellTiter-Blue™ reagent solution in RPMI (1 : 6) was added to each well at, shacked for 10 s and incubated at 37 °C for 1 h. Sample fluorescence was measured by using a Fluoroscan Ascent FL (Thermo Electron Corporation, Vantaa, Finland) with 544 nm excitation and 590 nm emission wavelengths, after subtracting the average fluorescence values of the culture media background . Cell viability was expressed as per cent increase of Resazurin reduction in cells treated with prefibrillar aggregates respect to the cells treated with the soluble protein (assumed to be 100%).
Assessment of necrotic cells
The presence of necrotic cells was assessed by assaying the activity of LDH, a typical necrotic marker released to the cell culture medium after rupture of the plasma membrane. LDH activity was measured in the culture media of cells exposed 0.5, 1, 3, 6, 16, 24 h to the HypF-N aggregates by using the LDH assay kit (Roche Diagnostics) at 490 nm, after blank subtraction at 595 nm.
Confocal measurement of intracellular ROS
Intracellular ROS were measured in intact cells treated with 5.0 mm ROS-sensitive fluorescent probe DCFH-DA using a confocal Bio-Rad MCR 1024 ES scanning microscope (Bio-Rad, Hercules, CA) equipped with a Kr/Ar laser source (15 mW) as previously described . Briefly, the cells were cultured on glass coverslips, exposed for differing times to 2.0 µm HypF-N prefibrillar aggregates (monomer protein concentration) and loaded with the dye upon incubation for 20 min at 37 °C with DCFH-DA added to the culture media. Then the cells on coverslips were fixed with 2.0% paraformaldehyde for 10 min, mounted on glass and analysed by confocal microscopy. Twenty optical sections (512 × 512 pixels) were taken for each examined sample through the depth of the cells with a thickness of 1.0 µm at intervals of 0.8 µm and then projected as a single composite image by superimposition.
Confocal analysis of calcium transients
Levels of free cytosolic Ca2+ in Hend and IMR90 cells were imaged at differing exposure times as reported previously . Briefly, the cells were incubated with serum-free DMEM containing 0.1% BSA, 10 µm (final concentration) Fluo3-AM as the fluorescent calcium indicator, 0.1% dimethylsulfoxide and Pluronic acid F-127 (0.01% w/v) for 10 min. The cells were washed, fixed with 2.0% paraformaldehyde for 20 min, mounted on glass and examined by confocal analysis. Cell fluorescence was monitored at 488 nm excitation by collecting the emitted fluorescence with a Nikon Plan Apo X60 (Nikon, Florence, Italy) oil-immersion objective through a 510 nm long-wave pass filter confocal microscope.
Assay of cholesterol content in the cell-surface membrane
Cholesterol in the surface membrane of untreated cells was assayed as previously described . Briefly, 2.0 × 106 cells were washed twice and resuspended in phosphate buffer, pH 7.5, containing 310 mm sucrose. Cells were incubated at 37 °C for 2 h with cholesterol oxidase (2.5 U·mL−1) and phospholipase C (0.2 U·mL−1 in 1.3 mm CaCl2) to make cholesterol in the intact membranes available for cholesterol oxidase . Then, 600 µL of Dole Reagent (78% isopropanol, 20% heptane, 2.0% H2O) was added to the reaction mixture, followed by 300 µL of heptane to stop the reaction and to reduce the background by lipid extraction. Total oxidized cholesterol was assayed as cholest-4-en-3-one at 235 nm in the upper phase obtained upon vortexing and centrifuging for 10 min at 2000 g and 4 °C. Controls consisted of cells supplemented with all the reagents except cholesterol oxidase. Cholesterol content was determined by comparison with a reference curve built by reacting differing amounts (1–50 µg) of cholesterol dissolved in isopropanol.
Intracellular ATP assay
ATP determination was performed using a highly sensitive bioluminescence assay (Kit HS II, Roche Diagnostics, Mannheim, Germany) by which extremely low concentrations of ATP can be detected on the basis of the ATP dependency of the light emission of the luciferin oxidation catalysed by luciferase . Briefly, IMR90 and Hend cells, plated on P60 wells (5.0 × 105 cells per plate), were exposed to fresh media supplemented with 2.0 µm HypF-N prefibrillar aggregates for differing time periods (0.5, 1, 3, 6, 16, 24 h). The cells were harvested and resuspended in the dilution buffer provided with the kit, following three freeze–thaw cycles. The lysis buffer was added to each sample in a 1 : 1 ratio and the samples were stored 5 min at room temperature and finally centrifuged at 750 g for 15 min at 4 °C. Total protein content was measured in the supernatant according to the method of Bradford . ATP measurement were carried out on this fraction by using a luminometer Lumat LB 9507 (EG & G Berthold, Bad Wildbad, Germany).
Annexin V-FITC and PI labelling
Annexin V-FITC and PI labelling (Bender MedSystems, Vienna, Austria) were used to detect the externalization of PtdSer during the apoptotic progression in Hend and IMR90 cells exposed to the 2.0 µm HypF-N aggregates for differing lengths of time . To measure the number of apoptotic cells, HypF-N-treated cells were washed twice with NaCl/Pi and resuspended in binding buffer (10 mm Hepes/NaOH, pH 7.4, 140 mm NaCl, 2.5 mm CaCl2) at a density of 5.0 × 105 cells·mL−1. Then, 2.5 µg·mL−1 of annexin V-FITC (PharMingen, San Diego, CA) and 0.2 µg·mL−1 PI (Sigma) were added. Within 60 min the labelled cells were determined using a FACScalibur flow cytometer (Becton Dickinson, Milan, Italy).
Measurement of MPT pore opening
MPT was measured by calcein fluorescence according to the method of Petronilli et al. [34,53] with minor modifications. In brief, calcein/AM freely enters the cell and becomes fluorescent upon de-esterification. Coloading of cells with cobalt chloride quenches the fluorescence in the cell except in mitochondria, because cobalt cannot cross mitochondrial membranes. During induction of the MPT, cobalt can enter the mitochondria where it is able to quench calcein fluorescence. Thus, decreased mitochondrial calcein fluorescence can be taken as a measure of the extent of MPT induction. At differing times of exposure to 2.0 µm HypF-N prefibrillar aggregates, cell cultures were washed in HBSS (10 mm Hepes buffer, pH 7.4, containing 144 mm NaCl, 2.0 mm CaCl2, 1.0 mm MgCl2, 5.0 mm KCl and 10 mm glucose) and incubated for 20 min at 37 °C in HBSS containing 1.0 mm calcein/AM and 1.0 mm cobalt chloride. Following cobalt quenching, cultures were washed with HBSS and then analyzed using a flow cytometer FACScalibur (Becton Dickinson) with 488 nm excitation and 590 nm emission filters.
Subcellular fractionation was achieved by the cytosol/mitochondria fractionation kit (Oncogene Research Products, San Diego, CA). The homogenized samples were treated as indicated by the manufacturer. Briefly, after treatment cells were harvested and resuspended in the cytosol extraction buffer supplemented with 1.0 mm dithiothreitol and a mix of proteases inhibitors. Plasma membrane rupture was achieved by three freeze–thaw cycles, the samples were then centrifuged at 750 g for 10 min at 4 °C. The resultant pellet, representing the nuclear fraction, was resuspended in 20 mm Hepes buffer, pH 7.4, containing 250 mm sucrose, 2.0 mm EGTA, 1.0 mm EDTA and sonicated twice for 5 s in ice. The supernatant was centrifuged at 10 000 g for 30 min at 4 °C and the pellet, containing the mitochondrial fraction, resuspended in the mitochondria extraction buffer supplemented with 1.0 mm dithiothreitol and a mix of protease inhibitors. The supernatant of the centrifugation was representative of the cytosolic fraction. Nuclear, mitochondrial and cytosolic fractions were used to assess Bcl-2, cytochrome c and AIF compartmentalization in both cell lines.
Another set of experiments was carried out on samples of cell homogenates. Cells were harvested and resuspended in 20 mm Tris/HCl buffer, pH 8.0, containing 2.0 mm EDTA, 1.0% Triton X-100, 10% glycerol, 137 mm NaCl, 6.0 m urea, 0.2 mm phenylmethanesulfonyl fluoride, 10 µg·mL−1 aprotinin and leupeptin. After three freeze–thaw cycles, the samples were sonicated twice for 5 s in ice and then centrifuged at 14 000 g for 10 min at 4 °C. The homogenates were used to assess the content of caspase 3 (the 11 kDa active fragment), caspase 8 (the 43 kDa active fragment) and caspase 9 (the 37 kDa active fragment). Total protein content in the nuclear and cytosolic fraction and in cell homogenates was measured according to Bradford .
Western blot analysis
To assess the intracellular levels of various apoptotic markers, equal amounts of cellular fractions were diluted in Laemmli sample buffer and boiled for 5 min. Samples containing ∼ 50 µg of proteins were run on SDS/PAGE and then blotted onto poly(vinylidene difluoride) (PVDF) Immobilio-P Transfer Membrane (Millipore Corp., Bedford, MA). The native form of the mitochondrial Bcl-2 (25 kDa) was determined in blots of 12% SDS/PAGE gels using anti-(Bcl-2) monoclonal sera (Santa Cruz Biotechnology, San Diego, CA). The cytosolic fractions of cytochrome c (16 kDa) were quantified in blots of 15% SDS/PAGE gels by using mouse anti-(cytochrome c) monoclonal sera (Oncogene Research Products). After washing, the membranes were incubated with peroxidase-conjugated anti-(mouse Ig) secondary sera for 1 h and the immunolabelled bands were detected using a SuperSignal West Dura (Pierce, Rockford, IL).
The nuclear fractions of AIF (57 kDa) were determined in blots of 12% SDS/PAGE gels by using anti-AIF polyclonal sera (Santa Cruz Biotechnology). The caspase 3 (11 kDa active fragment) content in the total homogenate fractions was determined by 15% SDS/PAGE, western blotting and blot incubation with anti-(caspase 3) polyclonal sera (Santa Cruz Biotechnology). Determination of caspase 8 and caspase 9 fragments was carried out in cell homogenates run in 12% SDS/PAGE, and blotted, by using rabbit anti-(caspase 8) and anti-(caspase 9) polyclonal sera (Chemicon Int., Temecula, CA), respectively. After washing, membranes were incubated with peroxidase-conjugated anti-rabbit secondary antibodies for 1 h and the immunolabelled bands were detected using a SuperSignal West Dura (Pierce).
All band densities were measured as densitometric units·µg−1 total proteins using the image analysis and densitometric program quantity one (Bio-Rad, Milan, Italy). For each band of interest, values relative to cells treated with soluble HypF-N were taken as 100%, whereas the values relative to cells treated with the prefibrillar aggregates were calculated as a percentage of the control within the same blot. β-Tubulin (for homogenate and cytosolic fraction) (Santa Cruz Biotechnology), prohibitin (for mitochondria) (Abcam Ltd, Cambridge, UK) and histones (for nuclei) (Chemicon Int.) were used to normalize the samples for equal amount of protein loading.
PARP activity measurement
PARP activity was assessed by an immunodot blot which detects poly(ADP-ribosylated) proteins . Aliquots of nuclear suspensions were diluted in 0.4 m NaOH containing 10 mm EDTA and loaded onto a Hybond N+ nylon membrane (Amersham Life Science, UK). The membrane was washed once with 0.4 m NaOH, blocked in NaCl/Pi–MT (NaCl/Pi, pH 7.4, containing 5.0% notfat dried milk and 0.1% Tween 20) and then incubated overnight with anti-poly(ADP-ribose) polyclonal sera (96-10-04, Alexis, San Diego, CA). The membrane was washed with NaCl/Pi–MT and incubated for 30 min with peroxidase-conjugated anti-(rabbit IgG) (Amersham Life Science). At the end of the incubation the blot was washed with NaCl/Pi–MT followed by chemiluminescence analysis. The membrane was subsequently immunoblotted with anti-histone sera (Chemicon Int.) to assess that equal amounts of proteins were loaded. Band (dot) densities were expressed as densitometric units·µg−1 protein (the constant protein amount applied on Hybond N+ nylon membrane) using the program for image analysis and densitometry quantity one software (Bio-Rad). Cells treated with soluble HypF-N were taken as 100%, whereas the values relative to HypF-N aggregate treated cells were calculated as a percentage of the control within the same immunodot blot.
DNA fragmentation analysis
DNA fragmentation was determined using an immunometric method (Cell Death Detection ELISAPLUS, Roche Diagnostics) according to the manufacturer's instructions. Briefly, after 24 h exposure to HypF-N prefibrillar aggregates, 20 µL of the cell homogenates were placed in a streptavidin-coated microtitre plate and incubated with a mixture of biotinylated anti-histone sera, peroxidase-labelled anti-DNA sera and incubation buffer (1% BSA, 0.5% Tween, 1 mm EDTA in NaCl/Pi) for 2 h. After a washing step, the retained peroxidase-linked complexes were incubated with ABTS for 10 min, resulting in colour development proportional to the number of nucleosomes captured in the antibody sandwich. DNA fragmentation was expressed as the enrichment of histone-associated mono- and oligonucleosomes released into the cytoplasm by measuring the absorption at 405 nm. The enrichment factor was proportional to the number of apoptotic cell present in the population.
All data were expressed as mean ± SD. Comparison between the different groups were performed by anova followed by Bonferroni's t-test. A P-value < 0.05 was set as significant.
We thank Daniele Nosi and Francesca Toni for technical advice. This study was supported by grants from the Italian MIUR (project numbers 2003054414-002 and 2005054147-001) and from Fondazione Cassa di Risparmio Pistoia e Pescia (project number 2004,0213).