Address correspondence and reprint requests to Jochen Herms, Zentrum für Neuropathologie und Prion Forschung, Ludwig Maximilians Universität, Feodor-Lynen-Street-23, 81377 München, Germany. E-mail: firstname.lastname@example.org
Consecutive cleavages of amyloid precursor protein (APP) generate APP intracellular domain (AICD). Its cellular function is still unclear. In this study, we investigated the functional role of AICD in cellular Ca2+ homeostasis. We could confirm previous observations that endoplasmic reticulum Ca2+ stores contain less calcium in cells with reduced APP γ-secretase cleavage products, increased AICD degradation, reduced AICD expression or in cells lacking APP. In addition, we observed an enhanced resting cytosolic calcium concentration under conditions where AICD is decreased or missing. In view of the reciprocal effects of Ca2+ on mitochondria and of mitochondria on Ca2+ homeostasis, we analysed further the cellular ATP content and the mitochondrial membrane potential. We observed a reduced ATP content and a mitochondrial hyperpolarisation in cells with reduced amounts of AICD. Blockade of mitochondrial oxidative phosphorylation chain in control cells lead to similar alterations as in cells lacking AICD. On the other hand, substrates of Complex II rescued the alteration in Ca2+ homeostasis in cells lacking AICD. Based on these observations, our findings indicate that alterations observed in endoplasmic reticulum Ca2+ storage in cells with reduced amounts of AICD are reciprocally linked to mitochondrial bioenergetic function.
A central hypothesis of the Alzheimer’s disease (AD) pathogenesis is that the amyloid-β peptide (Aβ) and/or Aβ-containing plaques play an important role in the development of disease. Mutations in either the amyloid precursor protein (APP), presenilin 1 (PS1), or presenilin 2 (PS2) are linked to the inherited forms of the disease and result in an increased production of Aβ, particularly of the Aβ42 isoform (Hardy 1997). Most of these mutations in APP, PS1, or PS2 alter the γ-cleavage of APP, affecting the ratio of Aβ40/Aβ42 species (Selkoe 1998; Sastre et al. 2001; Sato et al. 2003). Moreover, the generation of the corresponding APP C-terminal fragment called APP intracellular domain (AICD) is strictly dependent on γ-secretase and can also be affected by FAD mutations (Moehlmann et al. 2002). Recently, several studies indicate that not only Aβ42 but also the C-terminal fragments may alter neuronal survival. The AICD may have a critical pathophysiological role by modulating the neuronal Ca2+ homeostasis. AICD has been identified in cell lines, in rat and in transgenic mouse brains (Gu et al. 2001; Sastre et al. 2001; Yu et al. 2001; Weidemann et al. 2002) as well as in AD brain tissues (Passer et al. 2000). AICD is however very difficult to detect either because its generation is under strict regulation (Evin et al. 2003) or it is very unstable and rapidly degraded possibly by insulin-degrading enzyme (IDE) (Edbauer et al. 2002). However, it has been shown to be stabilized under certain conditions like binding to the cytosolic adaptor protein Fe65 (Kimberly et al. 2001) or by binding to cholesterol (George et al. 2004). This complex was found to interact with the transcription factor CL2/LSF/LBP1 (Fiore et al. 1995; Minopoli et al. 2001) and Tip60, a histone acetyltransferase (Cao and Sudhof 2001).
Genetic ablation of the PSs or pharmacological inhibition of γ-secretase activity (and thereby AICD production) attenuates the Ca2+ release from the endoplasmic reticulum (ER). Cells expressing mutant PS1, reveal an enhanced generation of AICDs and show enhanced ER Ca2+ filling. Moreover, cells lacking APP were shown to exhibit similar deficits in Ca2+ storage that could be reversed by transfection with APP constructs containing an intact AICD. Constructs lacking the AICD domain were not found to rescue the phenotype, strongly indicating that this domain is critically involved in ER Ca2+ filling (Leissring et al. 2002). The underlying mechanisms, however, have not yet been resolved. In the present study, we sought to elucidate the nature of the causal link between γ-secretase activity, AICD generation and intracellular cytosolic Ca2+ concentration ([Ca2+]i) homeostasis. By using three different cell types under various experimental conditions, we show that loss of AICD enhances the basal resting cytosolic Ca2+ levels and reduces intracellular ER Ca2+ concentration ([Ca2+]ER) storage. Furthermore, by analysing cellular ATP synthesis as well as the mitochondrial membrane potential we gather strong evidence that the mitochondrial function is compromised in cells lacking AICD.
Materials and methods
Fluorescent dyes Fura-2-acetoxymethylester [fura-2AM; 5 μmol/L in dimethylsulfoxide (DMSO)] and rhodamine-123 (Rh123; 10 μmol/L in DMSO) were purchased from Molecular Probes (Göttingen, Germany). Pluronic acid (0.08%), bovine serum albumin (0.1%), carbonyl cyanide 4-trifluoro-methoxyphenylhydrazone (FCCP; 2 μmol/L in DMSO), oligomycin (10 μmol/L in DMSO), doxycycline (Dox; 100 ng/mL), pyruvate (5 mmol/L), malate (5 mmol/L), ascorbate (1 mmol/L), 2,3,5,6-tetramethyl-p-phenylenediamine (200 μmol/L in DMSO) and succinate (10 mmol/L) were from Sigma (Deisenhofen, Germany). γ-Secretase inhibitor IV (γ-IV; 10 μmol/L) was purchased from Calbiochem (Darmstadt, Germany). DAPT (1 μmol/L in DMSO) was a gift from Boehringer Ingelheim (Ingelheim, Germany).
Cell lines and cell culture
Human embryonic kidney (HEK293) cells, HEK293 stably over-expressing wild type (Wt) APP (WtAPP), expressing APP695 containing the Swedish mutation of APP (APPsw), co-expressing APPsw and functionally inactive PS1 D385N, co-expressing swAPP and IDE, were cultured as described previously (Sastre et al. 2001; Edbauer et al. 2002). Human neuroglioma cells (H4) which express an inducible allele of AICD fragment with a C-terminal VSV-tag under the control of the ‘Tet-Off’ system (Gossen and Bujard 1992) were cultured in Dulbecco’s Modified Eagle's Medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. The expression of AICD-VSV could be turned off using Dox (100 ng/mL) in the growth medium. Mouse astrocytes were used as primary cultures prepared from cortical hemisphere of newborn C57 × Sv129 and APP0/0 (generated from C57 × Sv129 genetic background) by mechanical and enzymatic dissociation. Astrocytic cultures were maintained in Dulbecco’s Modified Eagle’s Medium supplemented with 5% fetal bovine serum, 1% N2 supplement, 0.2% MITOTM +serum extender (BD Bioscience, Belford, MA, USA) and 1% penicillin/streptomycin at 37°C for 1 week before using for experiments.
Cytosolic Ca2+ was measured using the fura-2AM dye (Molecular Probes) as described previously (Herms et al. 2003). In brief, fura-2AM was dissolved in Hank’s Balanced Salt Solution (HBSS) buffer (145 mmol/L NaCl, 2.5 mmol/L KCl, 1 mmol/L MgCl2, 20 mmol/L HEPES, 10 mmol/L glucose and 1.8 mmol/L CaCl2) containing 1% bovine serum albumin to improve solublization of the ester form of the dye and pluronic acid (0.08%) to serve as an additional dispersion aid. Cells on coverslip were loaded at 37°C for 30 min and washed one to two times with normal HBSS buffer before starting measurements. For Ca2+-free buffer, it was simply replaced by 50 μmol/L of EGTA. Cells were viewed by upright microscope (BX50WI; Olympus, Hamburg, Germany) using 20× water immersion objective. Excitation of the cells was performed at 340 nm for the bound form and 380 nm for the unbound form of Ca2+ and the ratio of the emissions could be obtained using digital imaging system (Till Photonics, Munich, Germany). Digital fluorescence images were acquired and displayed as pseudo colour images and subsequently analysed by choosing defined region of interest so as to exclude any area not covered by a cell. Values were acquired at 3 s intervals at both excitation wavelengths (340 and 380 nm). Ca2+ concentrations were calculated from fluorescence intensities ratio using the following equation (Grynkiewicz et al. 1985):
where Kd is the dissociation constant of fura-2 for Ca2+ (Kd = 224 nmol/L), β is the fluorescence ratio of the 380 nm signal in the absence of Ca2+ to that in the presence of saturating Ca2+, and Rmin and Rmax are the minimal and maximal fluorescence ratios, respectively, obtained using a calcium calibration buffer kit (Molecular Probes). Calcium values calculated with this approach have to be understood as estimations rather than as absolute values (Neher 1995).
All experiments were performed at 25°C and drugs were applied by bath perfusion. ER stores were depleted by adding cyclopiazonic acid (CPA), a potent reversible Ca2+ATPase inhibitor known to deplete Ca2+ stores passively, in a Ca2+-free HBSS buffer.
Measurement of mitochondrial membrane potential
Mitochondrial membrane potential was measured using Rh123 dye using the same microscope and software as described above. Loading protocol was similar to the one used for fura-2AM. Cells were excited at 470 nm and fluorescence data was acquired at an interval of 3 s. Rh123 distributes itself preferentially across the mitochondrial membrane strictly as a function of membrane potential, therefore quenching the fluorescent signal. On depolarisation, dye is released from the mitochondria, thereby increasing the cytosolic fluorescent signal. The data presented here is normalized to the maximal increase in Rh123 signal in response to complete mitochondrial depolarisation by FCCP.
Determination of ATP levels
ATP was determined luminometrically using ATP bioluminescent somatic cell assay kit (Sigma) according to the manufacturer’s protocol. This kit is based upon the luminometric method employing the luciferin-luciferase reaction. The reaction involves the ATP-dependent oxidation of luciferin catalysed by firefly luciferase yielding oxyluciferin, AMP, CO2 and pyrophosphate (PPi) as products. It is accompanied by emission of light, which is linearly related to ATP concentration and is measured using a luminometer (Fluostar optima, BMG Labtech, Jena, Germany). ATP content was calculated using internal standards and the data represented is the mean ± SEM for four determinations.
Crude mitochondrial isolation
Mitochondria were prepared from H4 cells by differential centrifugation. Cells were grown in 80-cm2 flasks to about 95% confluency, washed with phosphate-buffered saline, collected and pelleted down at 700 g for 12 min at 4°C. The pellet was resuspended in 700 μL of resuspension buffer containing10 mmol/L Tris–Cl (pH 7.5), 5 mmol/L EDTA and 1.5 mmol/L CaCl2 and incubated for 10 min on ice. The cell suspension was homogenized by 20 strokes using teflon potter homogenizer. Homogenate was centrifuged twice at 700 g for 12 min at 4°C to obtain post-nuclear supernatant. A crude mitochondrial pellet was isolated by high-speed centrifugation of post-nuclear supernatant at 16 500 g for 12 min at 4°C.
Western blot analysis
Western blot analysis were performed as previously described (Herms et al. 1999). Samples were separated by sodium dodecyl sulfate (SDS) –polyacrylamide gel electrophoresis on 10–20% Tricine–SDS gels (Invitrogen, Karlsruhe, Germany) and were analysed by immunoblotting with mouse anti-VSV antibody (Sigma). For detection of AICD from cell lysates, cells were lysed in radioimmunoprecipitation buffer supplemented with protease inhibitors (10 mmol/L Tris pH 8.0, 150 mmol/L NaCl, 1% Nonidet-P40, 0.5% deoxycholate, 0.1% SDS, 1x complete inhibitor mix (Roche, Mannheim, Germany), 5 mmol/L EDTA and 2 mmol/L 1.10-phenanthroline (Sigma). Equal amount of protein from each cell lysate was separated on 16.5%T 6%C Tricine–SDS gels containing 6 mol/L urea (Schagger and von Jagow 1987), then proteins were transferred to polyvinylidene fluoride membrane. AICD was detected with the anti-APP-C-terminal polyclonal antibody A8717 (Sigma), anti-rabbit-horseradish peroxidase secondary antibody (Promega, Mannheim, Germany) and Super Signal West Pico reagents (Pierce, Bonn, Germany).
Data analysis was performed offline using Excel and area under curve was calculated using Sigma Plot software. All results are expressed as mean ± SEM. Statistical data was assessed by one-way analysis of variance (anova). Appropriate post hoc tests i.e. Tukey test and Dunn’s method were performed wherever required for the pairwise comparisons.
Loss of APP or its C-terminal cleavage products modulate cytosolic free Ca2+ concentration and ER Ca2+ filling
In order to investigate the role of APP and its cleavage products in cellular Ca2+ homeostasis, we studied Ca2+ signalling in three different cell models: HEK293 cells, H4 cells and primary cultures of mouse astrocytes. Independent of the cell model analysed, loss of APP or its cleavage products were found to alter both resting (or basal) free [Ca2+]i as well as the amount of Ca2+ that can be released from the ER [Ca2+]ER (Fig. 1). Untransfected HEK293 cells (Wt), HEK293 cells that express APPsw or HEK293 cells over-expressing APP (WtAPP) show an average [Ca2+]i of ∼20 nmol/L (Figs 1a and c). On the other hand, APPsw cells expressing dominant negative PS1 mutation (PS1 D385N), exhibit more than four times higher basal cytoplasmic Ca2+ values, i.e. 92 ± 6 nmol/L (Figs 1b and c). Similar enhancement was obtained in APPsw cells over-expressing IDE (60 ± 4 nmol/L), and in APPsw cells that were treated with the potent γ-secretase inhibitor gamma-IV (77 ± 7 nmol/L) (Fig. 1c). Subsequently, we analysed the Ca2+ storage within the ER by the application of CPA which is known to release Ca2+ from the ER by blocking the Ca2+ATPase present within the ER membrane. We observed significantly reduced [Ca2+]ER levels in PS1 D385N cells, IDE, or APPsw cells that were treated with the γ-secretase blocker gamma-IV (Fig. 1d). To confirm that the observed changes in basal and ER Ca2+ were due to decreased AICD levels, endogenous AICD levels were detected in IDE and PS1 D385N cells. A decreased level of AICD was found in IDE and was even more reduced in PS1 D385N cells as compared with APPsw cells (Fig. 1e).
In order to verify that the effects observed in PS1 D385N or IDE expressing cells were indeed due to an alteration in APP cleavage products and not due to other γ-secretase cleavage products, we analysed [Ca2+]i and [Ca2+]ER in APP0/0 astrocytes (Fig. 2). Indeed, we observed highly significant enhanced resting calcium in astrocytes lacking APP compared with littermate control cells (Figs 2a and b). Moreover the [Ca2+]ER was found to be reduced, however to a lesser extent compared with the results obtained from HEK293 cells (Figs 2a and c). In order to evaluate the role of other γ-secretase cleavage products produced for example by the two other members of the APP protein family, namely the amyloid precursor-like protein 1 and 2, we performed additional experiments in the presence of a γ-secretase blocker, DAPT (1 μmol/L). Indeed, the treatment of APP0/0 astrocytes with DAPT further enhanced the [Ca2+]i to 74 ± 7 nmol/L (Figs 2a and b). Also, consistent with the results described above, the [Ca2+]ER was further reduced to 364 ± 10 nmol/L compared with Wt astrocytes (532 ± 14 nmol/L) (Fig. 2c). Taken together, our results demonstrate that cytosolic baseline Ca2+ was altered as a consequence of loss of APP cleavage products, in addition to previously described results that showed ER Ca2+ storage was influenced (Leissring et al. 2002).
Based on the observations that [Ca2+]i was enhanced and [Ca2+]ER was reduced in IDE expressing cells and in APP0/0 astrocytes treated with DAPT, one can hypothesize the involvement of AICD. To ascertain that AICD was responsible for modulating the phenomenon observed above, we extended our studies to analyse the effect of AICD using H4 neuroglioma cells expressing AICD with a VSV tag (AICD-VSV) under the control of ‘Tet-Off’ system. We compared cells which express AICD-VSV in the absence of Dox to cells where AICD-VSV expression was turned off by adding Dox (Fig. 3a). These experiments were performed in presence of DAPT (1 μmol/L) to exclude any influence of Aβ or AICD produced endogenously. As illustrated in Fig. 3 relative to controls, the cells lacking AICD-VSV fragment markedly enhanced the [Ca2+]i and significantly reduced [Ca2+]ER, very similar to the observations obtained in transfected HEK293 cells. In order to confirm that the effect seen was that of AICD-VSV only, another control experiment was made under the same experimental conditions in H4 cells expressing only the regulator plasmid for the Tet-Off system and not the response plasmid. As shown in Fig. 3b, there was no significant differences in [Ca2+]i and [Ca2+]ER in both Dox-treated and -untreated cells, both measured in the presence of DAPT (1 μmol/L). However, H4 cells expressing only regulator plasmid, when measured without DAPT, showed significantly reduced [Ca2+]i and enhanced [Ca2+]ER (Fig. 3b), indicating that endogenous level of AICD was sufficient to maintain normal Ca2+ homeostasis.
Although shown in different cell types, these results consistently and robustly indicated that the γ-secretase-derived carboxyl terminal cleavage product of APP, known as AICD, modulates both the resting cytosolic-free Ca2+ concentration and ER Ca2+ storage.
Cells lacking AICD have a lower ATP level and hyperpolarisation of the inner mitochondrial membrane potential
The [Ca2+]i is tightly regulated and mainly controlled by the Ca2+ATPase that is located within the ER membrane. Furthermore, in order to investigate if the alterations in the cellular Ca2+ homeostasis in cells lacking AICD were related to mitochondrial function, we analysed the cellular ATP level and the mitochondrial membrane potential (ψm).
As shown in Fig. 4a, both H4 cells lacking AICD and APP0/0 astrocytes revealed significantly reduced ATP levels of about 56% and 53%, respectively, as compared with control cells. In order to test if these alterations in the cellular ATP content were related to changes in the ψm, the latter was measured using Rh123 dye. Rh123 distributes itself preferentially across the mitochondrial membrane strictly as a function of ψm (Nicholls and Budd 2000; Toescu and Verkhratsky 2000). Fig. 4b shows H4 cells with and without AICD expression under basal conditions (i) and after FCCP treatment (ii), an uncoupler that collapses the proton gradient and fully depolarises the membrane potential. As shown in Fig. 4b (ii), collapse of the ψm by application of FCCP causes an increase in Rh123 fluorescence due to unquenching and release of Rh123 from the mitochondrial matrix into the cytosol. As shown in Fig. 4c, we observed significant lower values in H4 cell lacking AICD than in the control cells. This meant that ψm in cells lacking AICD was hyperpolarized compared with the control cells that express AICD. Analogous results were obtained in APP0/0 astrocytes (Fig. 4c).
Cytosolic-free calcium was increased and ER calcium was reduced on inhibition of mitochondrial oxidative phosphorylation
In order to ascertain whether mitochondrial ATP generation was altered due to lack of AICD, we blocked the glycolytic ATP synthesis by substitution of glucose with pyruvate. Under these conditions, cellular ATP generation was entirely dependent on mitochondrial oxidative phosphorylation. As shown in Fig. 5, the omission of glucose in both cells with or without AICD expression was not found to affect [Ca2+]i or [Ca2+]ER significantly. We then analysed the effect of oligomycin, a pharmacological agent that blocks ATP synthesis via inhibition of ATP synthase (Complex V). In addition, we investigated the effect of NaCN, an inhibitor of Complex IV and ATP synthesis. On treatment with oligomycin and NaCN, AICD-VSV over-expressing H4 cells (analysed in presence of DAPT) and Wt astrocytes indeed revealed enhanced [Ca2+]i levels (Figs 5a and c) and reduced [Ca2+]ER levels (Figs 5b and d), similar to the results obtained in cells lacking AICD (Figs 2 and 3). The treatment of cells with oligomycin and NaCN were not found to have a significant effect in cells lacking AICD both in H4 cells (Figs 5a and b) and in APP0/0 astrocytes, also analysed in presence of DAPT (data not shown).
Responses of AICD lacking cells was reversed on application of Complex I and II substrates
As shown in the preceding experiments, inhibition of oxidative phosphorylation chain mimics the Ca2+ responses from cells lacking AICD or APP. We therefore explored the action of substrates for mitochondrial complexes on cells with suppressed AICD-VSV expression and APP0/0 astrocytes, analysed in presence of DAPT, in order to reveal if it rescues the calcium phenotype to that shown by cells expressing AICD. For mitochondrial Complex I, we used pyruvate and malate (5 mmol/L each) as a substrate, methyl-succinate (10 mmol/L) was used as substrate for Complex II and 2,3,5,6-tetramethyl-p-phenylenediamine (200 μmol/L) in combination with ascorbate (1 mmol/L) was used as a substrate for Complex IV (Abramov et al. 2004). As shown in Fig. 6, pyruvate/malate and succinate significantly reduced [Ca2+]i (Fig. 6a) in cells with suppressed AICD production (Dox treated). On the other hand, in these cells, the [Ca2+]ER was significantly enhanced on application of succinate (Fig. 6b). However, no significant reversal was seen by the application of Complex IV substrates. These substrates, however, did not affect the [Ca2+]i or [Ca2+]ER in the control cells (No Dox) that express AICD-VSV (Figs 6a and b). In APP0/0 astrocytes only the succinate administration was found to rescue both the [Ca2+]i and [Ca2+]ER (Figs 6c and d), while there was no effect on Wt astrocytes (data not shown). The observation that the alteration in Ca2+ recovered almost completely by application of substrates of Complex I and II both in H4 cells lacking AICD-VSV (Dox treated) and in APP0/0 astrocytes robustly supports the notion that AICD influences the cellular Ca2+ homeostasis by modulating mitochondrial bioenergetic function.
ATP synthase activity was found to be independent of AICD
In general, inhibition of ATP synthase inhibits the flow of protons through F0 particles eventually increasing mitochondrial membrane potential. Therefore, in order to investigate if the hyperpolarisation of mitochondrial membrane in AICD lacking cells (Dox treated) was due to reduced ATP synthase activity, we assayed ATP synthase activity in AICD-VSV expressing (Dox untreated) and AICD-VSV non-expressing (Dox treated) H4 cells. The amount of inhibition by oligomycin (10 μmol/L) was used to quantify ATP synthase activity in mitochondria-enriched cellular fractions. As shown in Fig. 7, oligomycin inhibited the activity by about 60% in both AICD-VSV expressing and non-expressing H4 cells, indicating a similar amount of total activity in both cell types.
Mutations in PS1 have been shown to be associated with increased levels of Ca2+ within the ER, whereas cells deficient in PS1 and PS2 showed decreased levels (Leissring et al. 2000; Yoo et al. 2000; Schneider et al. 2001; Herms et al. 2003). These alterations, observed in a variety of cell types including neuronal cells, strongly indicate that PS1 modulates the filling of the ER with Ca2+. In a previous study, we gained strong evidence that in mutant PS1 the enhanced ER filling is indeed due to γ-secretase-derived APP cleavage products (Herms et al. 2003). Using three independent well-controlled cell culture systems, we now show for the first time that the resting cytosolic Ca2+ is affected by AICD. Our study using CPA-mediated induction of ER Ca2+ release shows a remarkably strong correlation between ablation/suppression of cellular AICD expression and enhanced basal [Ca2+]i. Thus, PS1 D385N cells having the lowest steady state levels of AICD also manifested the highest resting cytosolic Ca2+ concentration. At the same time, we found a reduced [Ca2+]ER in AICD ablated/suppressed cells. The exact mechanism through which endogenous AICD modulates resting cytosolic Ca2+ is still under investigation. However, upon change to zero Ca2+ medium we observed an immediate drop in the elevated cytosolic Ca2+ down to control resting levels. This implies a possible alteration at the level of the plasma membrane induced by absence of AICD.
Given the well-documented role of mitochondria in [Ca2+]i storage and release as well as the reciprocal effects of Ca2+ on mitochondria (Carafoli 2003), we turned to the functional study of these organelles. In summary, we found that lack of AICD results in reduced ATP levels and is associated with hyperpolarisation of the mitochondrial inner membrane potential. In general, a mitochondrial hyperpolarisation is either due to an alteration of the mitochondrial ATP synthase or a compensatory phenomenon due to an enhanced ATP hydrolysis. To test this, we analysed the activity of the ATP synthase in crude mitochondrial fractions. However, we did not find any significant differences in the ATP synthase activity in cells with or without AICD expression. Therefore, it is very unlikely that the reduced cellular ATP content in cells lacking AICD is due to a reduced ATP production by mitochondria. More likely, the consumption of ATP is enhanced when AICD is lacking. Indeed, hydrolysis of ATP (or oxidation of respiratory substrates) further drives the electrochemical potential gradient across the inner mitochondrial membrane, allowing mitochondria to take up additional amounts of Ca2+ into the mitochondrial matrix by a ‘uniporter’ (Rizzuto et al. 2004). Therefore, it is reasonable to assume that the high basal cytosolic Ca2+ concentrations in AICD-deficient cells induce increased ATP hydrolysis (and hence ATP depletion) as a consequence of mitochondrial hyperpolarisation. Thus, mitochondria attempt, unsuccessfully, to buffer the cytosolic Ca2+ overload. The depletion in ATP would have further direct sequelae on Ca2+ uptake by the ER. Basically, ER Ca2+ refilling depends strongly on sarcoplasmic/endoplasmic reticulum Ca2+ ATPase (SERCA) activity. A reduced ATP supply for the SERCA pumps would inhibit ER calcium uptake and explain the low Ca2+ ER stores we observed in AICD-deficient cells. A reduced ability of the ER to buffer [Ca2+]i concentrations in turn contributes positively to higher cytosolic overload of this cation.
A mechanistic insight into the correlative evidence outlined above is provided by our experiments using mitochondrial respiratory inhibitors on the one hand and respiratory substrates on the other. For instance, cells over-expressing AICD and exposed to the mitochondrial oxidative phosphorylation inhibitors oligomycin and sodium cyanide re-capitulated the abnormal calcium phenotype of AICD-deficient cells, i.e. enhanced [Ca2+]i levels and reduced [Ca2+]ER levels. In this study, ATP depletion presumably directly lowered SERCA activity with the consequent alterations in calcium concentrations. Not surprisingly, use of the pharmacological respiratory inhibitors had no additional effect on the already altered calcium homeostasis in non-AICD expressing cells. Importantly, the AICD-deficient calcium phenotype was simply rescued by the addition of mitochondrial substrates known to enhance ATP production. This was shown to occur in two-independent cell lines, namely H4 cells lacking AICD and in the APP-knockout astrocytes. The mechanism here is that mitochondria accumulate larger amounts of cytosolic calcium, fueled by oxidation of respiratory substrates, while at the same time permitting SERCA activity to drive uptake of the cation to fill the ER. Taken together, our findings indicate for the first time that the alteration of the Ca2+ storage by ER is not merely due to an alteration in the mechanism involved in ER Ca2+ release through the phosphoinositol cascade (evaluated by bradykinin-induced Ca2+ release), as suggested previously (Leissring et al. 2002). Our results favour the hypothesis that AICD affects Ca2+ buffering within the ER at least in parts by affecting mitochondrial bioenergetic function and hence the activity of calcium ATPase. A recent report shows that PS1-ΔE9 FAD mutant has increased ER Ca2+ leak channel activity leading to an enhanced basal cytosolic Ca2+ and reduced ER storage (Tu et al. 2006). In context of AICD, this mutation has been previously shown to produce reduced amount of AICD fragments (Chen et al. 2002). Together, these two reports further strengthen our finding of a potential link between AICD and cellular Ca2+ homeostasis. Of course, one cannot exclude other additional mechanistic possibilities that link AICD expression to changes in the cytoplasmic and ER calcium levels. For instance, Ca2+ can directly affect oxidative phosphorylation as the activation of many mitochondrial dehydrogenases are Ca2+ dependent (Hansford and Chappell 1967; Denton et al. 1972, 1978; McCormack and Denton 1979; Nichols and Denton 1995). Also, AICD could influence cellular ATP metabolism and Ca2+ homeostasis via a number of possible mechanisms such as the formation of a transcriptionally active complex (Cao and Sudhof 2001;) which regulates the expression of genes involved in cellular ATP homeostasis and calcium storage like the actin cytoskeleton including α2-actin and transgelin (Mueller et al. 2007). A recent report by Kasri et al., shows a decrease in the [Ca2+]ER in PS double-knockout mouse embryonic fibroblast due an increased expression of type I IP3 receptors (Kasri et al. 2006). This modulation of ER Ca2+ by increased expression of IP3 receptors, which is independent of its role in mitochondrial function, may additionally alter the Ca2+ homeostasis in AICD deficient cells and could hold for the phenomenon that a total block of mitochondrial ATP production in AICD expressing cells does not alter the Ca2+ storage as much as the lack of AICD expression. Although we have no direct evidence of AICD controlling the IP3 receptor expression, a role of AICD in the transcription of glycogen synthase kinase (GSK)3β has also been observed in a study of nuclear extracts of AICD-transfected PC12 cells (Kim et al. 2003) as well as in HEK293 cells (Von Rotz et al. 2004). Interestingly, transgenic mice expressing AICD show an enhanced activation of the GSK3β (Ryan and Pimplikar 2005). The active form of GSK3β has been found in mitochondria (Bijur and Jope 2003) and is known to be an important modulator of mitochondrial function (Hoshi et al. 1996). Moreover GSK3β was found to modulate the expression of the IP3 receptor (Beals et al. 1997), a receptor critically involved in ER Ca2+ storage and release. This modulation of ER Ca2+ by GSK3β, which is independent of its role in mitochondrial function, may additionally alter the Ca2+ homeostasis in AICD deficient cells and could hold for the phenomenon that a total block of mitochondrial ATP production in AICD expressing cells does not alter the Ca2+ storage as much as the lack of AICD expression. However, our results did not show any difference in the expression of phosphorylated and non-phosphorylated forms of GSK3β in both AICD expressing and non-expressing H4 cells (data not shown).
In conclusion, it is most likely that the alteration in Ca2+ homeostasis in cells lacking AICD is intrinsically linked to alterations in mitochondrial function and ATP consumption. AICD might be an important link for the understanding of the cascade of events leading to nerve cell dysfunction in AD. In light of current therapeutic concepts that aim to affect γ-secretase and β-secretase function, both affecting the generation of AICD as well, the role of AICD in mitochondrial bioenergetic function is warranted to be analysed further in order to avoid potential side effects arising from manipulations of these secretases.
This work was supported by the DFG (SFB 596) to JH. We are grateful to Dr Rüdiger Schmalzbauer and Dr Bjarne Krebs for their valuable suggestions on this study. We would like to thank the late Doris Schechinger for her excellent technical assistance.