|Home | About | Journals | Submit | Contact Us | Français|
Mutations in presenilins (PS) are the major cause of familial Alzheimer’s disease (FAD) and have been associated with calcium (Ca2+) signaling abnormalities. Here, we demonstrate that FAD mutant PS1 (M146L) and PS2 (N141I) interact with the inositol 1,4,5-trisphosphate receptor (InsP3R) Ca2+ release channel and exert profound stimulatory effects on its gating activity in response to saturating and sub-optimal levels of InsP3. These interactions result in exaggerated cellular Ca2+ signaling in response to agonist stimulation as well as enhanced low-level Ca2+ signaling in unstimulated cells. Parallel studies in InsP3R-expressing and -deficient cells revealed that enhanced Ca2+ release from the endoplasmic reticulum as a result of the specific interaction of PS1-M146L with the InsP3R stimulates amyloid beta processing, an important feature of AD pathology. These observations provide molecular insights into the “Ca2+ dysregulation” hypothesis of AD pathogenesis and suggest novel targets for therapeutic intervention.
Alzheimer’s disease (AD) is a common dementia involving slowly developing and ultimately fatal neurodegeneration. Most AD is sporadic and usually develops at age > 60, but ~10% of AD is inherited as an autosomal dominant trait (familial AD, FAD) and disease develops as early as the late 30 years of age. Mutations in amyloid precursor protein (APP) and presenilins (PS1, PS2) cause early onset FAD (Hutton and Hardy, 1997). Hallmark features of AD include accumulations of extracellular β amyloid (Aβ) plaques and intracellular neurofibrillary tangles with neuronal atrophy or loss (Hardy, 2006; Mattson, 2004). The “amyloid hypothesis” of AD postulates that accumulation of oligomeric or fibrillar Aβ, due to production of more amyloidogenic forms of Aβ and defective processing and clearance, leads to pathological sequelae associated with the disease (Haass and Selkoe, 2007; Hardy and Selkoe, 2002). This hypothesis has nevertheless been questioned (De Strooper, 2007; Hardy and Selkoe, 2002; Shen and Kelleher, 2007), and others have also been proposed to describe pathological origins of the disease (reviewed in (Blennow et al., 2006; Mattson, 2004)). Accumulating evidence suggests that sustained disruption of intracellular Ca2+ signaling may play an early proximal, and perhaps central, role in AD pathogenesis (Gandy et al., 2006; LaFerla, 2002; Mattson and Chan, 2003; Smith et al., 2005a; Stutzmann, 2005). Ca2+ is involved in many facets of neuronal physiology, including activity, growth and differentiation, synaptic plasticity and learning and memory, as well as pathophysiology, including necrosis, apoptosis and degeneration (Berridge et al., 2000). Before identification of PS, Ito (Ito et al., 1994) discovered that fibroblast lines derived from AD patients, later shown to harbor the A246Q mutation in PS1, without exception, generated exaggerated intracellular Ca2+ concentration ([Ca2+]i) responses to sub-maximal concentrations of two G-protein coupled receptor (GPCR) agonists that activate phospholipase C (PLC). Activation of GPCR, including bradykinin, 5HT2A and metabotropic glutamate receptors, stimulates PLC activity that produces inositol 1,4,5-trisphosphate (InsP3), which binds to its receptor (InsP3R), an endoplasmic reticulum (ER)-localized Ca2+ channel, resulting in release of Ca2+ from the ER and elevation of [Ca2+]i (Foskett et al., 2007). Enhanced Ca2+ release could not be attributed to altered amounts of Ca2+ in the ER or number of agonist receptors, or to plasma membrane influx pathways (Ito et al., 1994). It was suggested that alteration of InsP3R-mediated Ca2+ release was a fundamental defect in AD, although the molecular mechanisms were undefined. Subsequently, it was demonstrated that enhanced InsP3-mediated Ca2+ signaling was a highly predictive diagnostic feature of AD-derived peripheral cells (Hirashima et al., 1996). Many subsequent studies have confirmed that mutant PS expression is associated with exaggerated ER Ca2+ release in several cellular and animal model systems including cells from FAD patients (Etcheberrigaray et al., 1998; Hirashima et al., 1996; Ito et al., 1994), neuronal and non-neuronal cells engineered to express recombinant mutant PS proteins (Cedazo-Minguez et al., 2002; Guo et al., 1996; Johnston et al., 2006; Leissring et al., 2001; Leissring et al., 1999a; Leissring et al., 1999b; Smith et al., 2002), and cells from mutant PS transgenic animals (Leissring et al., 2000), including brain neurons (Barrow et al., 2000; Mattson et al., 2000; Schneider et al., 2001; Stutzmann et al., 2004; Stutzmann et al., 2006) long before the appearance of plaques and tangles (Stutzmann et al., 2004). A “Ca2+ overload” hypothesis has been widely invoked to account for exaggerated Ca2+ release in mutant PS-expressing cells (discussed in (Stutzmann, 2005), but increased ER Ca2+ stores have not been consistently observed (Giacomello et al., 2005; Ito et al., 1994; Lessard et al., 2005; Zatti et al., 2006; Zatti et al., 2004) nor is there a consensus regarding possible molecular mechanisms involved (Gandy et al., 2006; LaFerla, 2002; Smith et al., 2005a; Stutzmann, 2005). Here, we have discovered a mechanism that can account for intrinsic altered Ca2+ signaling in AD cells that involves a biochemical and functional interaction of WT and FAD mutant PS with the InsP3R Ca2+ release channel. The biochemical interactions of FAD mutant PS1 or PS2 with the InsP3R profoundly enhance the activity (gating) of the channel, that results in and can account for exaggerated cellular Ca2+ signaling. By use of InsP3R deficient cells, we show that this enhancement is directly involved in mutant PS-mediated APP processing (Aβ generation), an important feature of AD pathology (Mattson and Guo, 1997).
The effects of PS expression on InsP3R channel activity were examined in native ER membranes by single channel patch clamp electrophysiology of the outer membrane of isolated Sf9 cell nuclei (Ionescu et al., 2006). Insect cells express a single InsP3R isoform, most closely related to the mammalian type 1 channel, the predominant brain isoform (Foskett et al., 2007), that has permeation and gating properties and ligand regulation very similar to those of mammalian InsP3R channels (Ionescu et al., 2006). In membrane patches from control nuclei, InsP3R channels exposed to optimal ligand conditions (10 μM InsP3, 1 μM Ca2+) were consistently detected and they gated with high open probability (Po = 0.76 ± 0.05, n = 10; Fig. 1B and Supp. Table 1). Human wild-type PS1 (PS1-WT) and FAD mutant PS1 M146L (PS1-M146L) were expressed in Sf9 as full-length holoproteins (Fig. S1) throughout the ER (not shown) and in the nuclear envelope (Fig. S2). In nuclei from either FAD mutant PS1-M146L or PS1-WT expressing cells, no novel ion channels were detected (n = 59; Fig. 1A). Nor were channel activities observed in the absence of InsP3 (n = 10) (Fig. 1A) or in the presence of InsP3 together with its competitive inhibitor heparin (100 μg/ml; n = 7, not shown). InsP3R channels observed in the presence of saturating 10 μM InsP3 in membrane patches from PS1-WT-infected cells had Po similar to those from mock-infected or uninfected cells (p > 0.05; Fig. 1B & D). In contrast, Po was elevated (p < 0.05) in PS1-M146L infected cells (Po = 0.86 ± 0.03, Fig. 1B & D), with many channels observed to be ‘locked open’ for long periods (as in Fig. 1B), a gating behavior only extremely rarely observed in control channels (unpublished observations). PS1-M146L enhanced InsP3R gating in saturating [InsP3] by prolonging the channel open time (Fig. 1E). Interestingly, PS1-WT also enhanced the mean open time, although to a lesser extent than mutant PS1 (Fig. 1E), and it also enhanced the mean closed time (Fig. 1F).
With [InsP3] lowered to 33 nM, InsP3R channel Po in patches from control nuclei was reduced to 0.27 ± 0.01, reflecting the [InsP3]-dependence of gating (Ionescu et al., 2006). Strikingly, Po was elevated by nearly 3-fold in patches from nuclei isolated from the FAD mutant PS1-expressing cells (Po = 0.75 ± 0.06), to a level that was comparable to that observed in saturating [InsP3] (Fig. 1C & D). Enhanced Po was caused by a marked destabilization of the channel closed state (enhanced opening rate) as well as prolongation of the open time (Fig. 1E & F). In contrast, PS1-WT was without effect on Po (Fig. 1D), although it appeared to influence channel gating, reflected as a small increase of the channel mean closed time (Fig. 1F).
These results demonstrate a gain-of-function effect of mutant PS1 on InsP3R channel activity observed at the single channel level. Because the effects of FAD PS1 expression were observed in isolated nuclei, we speculated that this functional effect was caused by an association of PS with the InsP3R in the ER membrane patches. PS1-WT and PS1-M146L immunoprecipitates from Sf9 cells that had been co-infected with rat types 1 or 3 InsP3R contained both InsP3R channel isoforms (Fig. 2A). Similarly, PS1 immunoprecipitates from mouse whole brain lysates contained the type 1 InsP3R (Fig. 2C). These results demonstrate a biochemical interaction between InsP3R and both WT and mutant PS1.
Similar studies were undertaken in chicken DT40 cells, using analogous protocols, but with polyclonal lines engineered to stably express either WT or mutant PS1 proteins (Fig. S3). A DT40 line with all three InsP3R isoforms genetically deleted (InsP3R-KO DT40) constitutes the only InsP3R-null cell line available (Sugawara et al., 1997). We reasoned that if PS1 similarly affected gating of InsP3R channels in native InsP3R-expressing DT40 cells, the generality of this as a molecular mechanism associated with FAD mutant PS1 could be established, and furthermore that use of InsP3R-KO DT40 cells would be valuable for investigating the physiological relevance of the PS-InsP3R interaction. As in the Sf9 cell recordings, no novel ion channel activities were observed in nuclei from the wild type or mutant PS1-expressing native DT40 cells (not shown). Importantly, highly similar effects of PS expression on InsP3R channel behavior were observed as in the Sf9 cells (Fig. 3 and Supp. Table 2). Thus, in saturating [InsP3], PS1-WT was without effect (Po = 0.53 ± 0.05 in control vs. 0.57 ± 0.07 in PS1-WT transfected cells), whereas PS1-M146L expression increased channel activity (Po = 0.83 ± 0.04; p <0.01) (Fig. 3A & C) by locking the channel open for long periods (Fig. 3D). In sub-optimal [InsP3] (100 nM), mutant PS1 stimulated channel activity by 4-fold (Po = 0.63 ± 0.07 vs. 0.16 ± 0.02 for control cells; Fig. 3B & C) to levels similar to those observed for control channels in saturating [InsP3]. Also as in Sf9 cells, PS1-WT was without effect on Po, although it influenced channel gating, as evidenced by small effects on both mean open and closed times (Fig. 3D & E).
The similar behaviors of InsP3R gating from different species in different cell systems in response to expression of FAD mutant and WT PS1 strongly suggests that this is a fundamental channel regulatory mechanism.
Altered [Ca2+]i signaling has been observed in cells expressing either FAD mutant PS1 or PS2 (Gandy et al., 2006; LaFerla, 2002; Smith et al., 2005a; Stutzmann, 2005). Like PS1, wild-type PS2 (PS2-WT) and FAD mutant PS2 N141I (PS2-N141I) also localized to the nuclear envelope (Fig. S2) and interacted biochemically with the InsP3R (Fig. 2B & C). With 10 μM InsP3 in the pipette solution, InsP3R channels from PS2-WT-infected cells (Fig. S1) had Po similar to those from mock-infected cells (P > 0.05; Fig. 4A & C and Supp. Table 3). In contrast, Po was elevated (p < 0.05) in PS2-N141I expressing cells (Po = 0.81 ± 0.02, Fig. 4C). As observed for PS1-M146L, many channels observed in PS2-N141I nuclei were ‘locked open’ (as in Fig. 4A). With sub-saturating [InsP3], Po in patches from PS2-N141I nuclei was elevated by nearly 3-fold compared with Po observed in nuclei from mock or PS2-WT-expressing cells (Fig. 4C). Similar to the effect of PS1-M146L, elevated Po was caused primarily by marked enhancement of the opening rate and prolongation of the mean open time (Fig. 4D & E). PS2-WT was without effect on channel Po (Fig. 4B & C), but it influenced gating, reflected as increases in the mean closed and open times (Fig. 4D & E). These effects of WT and FAD mutant PS2 on InsP3R gating are highly similar to those observed for WT and mutant PS1, respectively, strongly suggesting that aberrant modulation of InsP3R channel gating is a general property of FAD mutant PS.
To address whether these effects of FAD mutant PS observed at the single channel level account for altered [Ca2+]i signaling in AD cells, InsP3R-mediated Ca2+ signals were recorded in the DT40 cell lines that were used for the single channel studies and that had comparable levels of InsP3R expression (Fig. S3). Physiological InsP3R-mediated Ca2+ signals were elicited by cross-linking the B-cell receptor (BCR). A high concentration of anti-IgM (5 μg/ml) triggered a rapid increase in [Ca2+]i in control cells (Fig. 5A) due to InsP3-mediated Ca2+ release from intracellular stores (Sugawara et al., 1997). In PS1-WT-expressing cells, the magnitude and kinetics were similar to those observed in control cells. In contrast, exaggerated [Ca2+]i responses were observed in PS1-M146L-expressing cells (Fig. 5A), with the peak response over 1.5-fold higher than in control or PS1-WT-expressing cells (Fig. 5B). Thus, the DT40 cell system recapitulates the exaggerated [Ca2+]i responses observed in peripheral cells from FAD patients and neuronal and non-neuronal animal and model cells expressing FAD mutant PS. In response to a low anti-IgM concentration (0.05 μg/ml) expected to generate less InsP3, repetitive [Ca2+]i oscillations were triggered in 52 ± 5 % of control cells (Fig. 5C & D), due to periodic Ca2+ release through the InsP3R (White et al., 2005) after a long lag period (Fig. 5F) as [InsP3] increased to levels sufficient for channel stimulation. In cells expressing PS1-M146L, the peak amplitude of the oscillations was similar to those observed in control and PS1-WT expressing cells (Fig. S4), whereas both the oscillation frequency and number of responding cells were increased, and the latency between application of agonist and the first [Ca2+]i response was decreased (Fig. 5C-F). Of note, the latency was nearly abolished in a significant subset (~30%) of cells expressing PS1-M146L (Fig. 5C & F). This response is highly reminiscent of that of control cells to saturating concentrations of BCR antibody (Fig. 5A), and was never observed in control or PS1-WT-expressing cells. Increased number of oscillating cells, enhanced oscillation frequency and diminished latencies are all consistent with a heightened InsP3 sensitivity of InsP3R-mediated Ca2+ release in the FAD mutant PS1 expressing cells. Interestingly, the number of responding cells was increased and the latency was shortened in the PS1-WT-expressing cells (Fig. 5D & E), although to a lesser extent than the more exaggerated responses observed in cells expressing FAD mutant PS1.
Spontaneous InsP3R-dependent [Ca2+]i oscillations were observed in ~6% of control cells perfused with Hank’s balanced salt solution without stimulation (Fig. 5G & H), as observed previously (White et al., 2005). Expression of PS1-WT approximately doubled the percentage of cells displaying this behavior (Fig. 5H). In contrast, a 4-fold higher percentage, ~25%, of the PS1-M146L-expressing cells exhibited spontaneous [Ca2+]i oscillations (Fig. 5H). Furthermore, the oscillation frequency in these cells was also enhanced (Fig. 5I). In a subset of the mutant PS1-expressing cells (~ 4%), spontaneous exaggerated [Ca2+]i transients were observed that were never seen in control and PS1-WT-expressing cells (Fig. 5G).
These results are congruent with those obtained in the single channel studies. In both sets of experiments, exaggerated responses to InsP3, particularly under conditions of low [InsP3], were observed in the context of FAD mutant PS1 expression, with PS1-WT expression also having effects, but much less exaggerated. The congruence of two very different sets of data suggests that the observed exaggerated [Ca2+]i responses are due to the observed exaggerated InsP3R single-channel responses.
Exaggerated [Ca2+]i responses have been a consistent observation in cells expressing mutant PS proteins (LaFerla, 2002; Smith et al., 2005a; Stutzmann, 2005; Yoo et al., 2000), but it has been suggested that they are caused by enhanced expression of release channels (Chan et al., 2000; Kasri et al., 2006; Schneider et al., 2001; Smith et al., 2005b; Stutzmann et al., 2006) or overfilling of ER Ca2+ stores (Leissring et al., 2000; Leissring et al., 2001; Schneider et al., 2001; Tu et al., 2006). We therefore examined whether either factor contributed to the exaggerated [Ca2+]i responses observed here in DT40 cells. All three cell lines expressed approx. equal levels of the types 1 and 3 InsP3R (Fig. S3). Thus, the exaggerated responses cannot be accounted for by altered InsP3R expression in our studies. To investigate the filling state of intracellular Ca2+ stores, we evaluated [Ca2+]i responses to the Ca2+ ionophore ionomycin or the SERCA inhibitor thapsigargin applied in the absence of extracellular Ca2+ so that the observed [Ca2+]i responses were due entirely to Ca2+ derived from intracellular compartments. Ionomycin (5 μM) triggered a rapid release of Ca2+ that was diminished in the PS1-M146L- but not PS1-WT-expressing cells (Fig.S5). Similarly, Ca2+ released in response to thapsigargin (1 μM) was also significantly reduced in the mutant PS1-expressing cells, and to a lesser extent, in the PS1-WT-expressing cells (Fig.S5). These results suggest that the Ca2+ stores are not overloaded in mutant PS1-expressing cells and may in fact even be reduced. To examine the filling state of the ER Ca2+ stores more directly, ER [Ca2+] was measured with the low-affinity fluorescence indicator Mag-Fura2 (Laude et al., 2005). Addition of MgATP to a solution bathing permeabilized cells with Ca2+ stores depleted (not shown), enhanced the fluorescence ratio as a consequence of SERCA-mediated loading of intracellular stores with Ca2+ (Fig.6). At steady-state, the stores in the FAD PS1-expressing cells were loaded less fully than in control or PS1-WT-expressing cells (Fig.6A; p < 0.01), confirming that the Ca2+ stores are not overloaded in the mutant PS1-expressing cells. In contrast, when loading was performed in the presence of the InsP3R inhibitor heparin, the steady-state level of ER Ca2+ was similar in all the cell lines (Fig. 6B). Thus, the exaggerated [Ca2+]i responses observed in mutant PS1-expressing DT40 cells (Fig. 5) cannot be accounted for by overfilling of ER Ca2+ stores. Rather, a reduced ER Ca2+ was observed in the FAD-PS1 expressing cells that appeared to be due an activated heparin-sensitive Ca2+ leak through the InsP3R. This was more directly examined by use of a different protocol, in which the Ca2+ leak permeability of the ER membrane was measured following the addition of thapsigargin to cells with ER stores filled with Ca2+ to equivalent levels. The Ca2+ leak rate was similar in control and PS1-WT cells, whereas it was greater in the FAD-mutant PS1 expressing cells (Fig. 6C & D). This enhanced Ca2+ leak observed in the PS1-M146L cells was eliminated by addition of heparin (Fig. 6E & F), suggesting that it was mediated by the InsP3R. To test this rigorously, we generated stable PS1-expressing DT40 cells with genetic disruption of all three InsP3R genes (Sugawara et al., 1997). In the absence of InsP3R expression (Fig. S3), both anti-IgM-induced and spontaneous [Ca2+]i signals were absent (not shown). Notably, the sizes of the intracellular Ca2+ stores in the PS1-M146L or PS1-WT-expressing cells were similar in the InsP3R-deficient cells (Fig. 6G). Importantly, the enhanced PS1-M146L induced ER Ca2+ leak rate was absent in the InsP3R-deficient cells (Fig. 6H). These results demonstrate that exaggerated Ca2+ signaling, reduced Ca2+ store size and enhanced ER Ca2+ leak permeability are specific properties of PS1-M146L-expressing cells and furthermore, that these features are completely dependent on the InsP3R. Together, these results strongly suggest that mutant PS1-M146L diminishes the size of the ER Ca2+ store by a mechanism that involves its enhancement of InsP3R channel activity, a process that disrupts the normal Ca2+ pump/leak balance in favor of enhanced leak mediated by InsP3-dependent InsP3R-mediated Ca2+ permeability. We conclude therefore, that the observed altered InsP3R single-channel activity most likely accounts for the altered [Ca2+]i signals observed.
Whereas FAD mutant PS associated exaggerated Ca2+ signaling has been observed in many cell types, AD pathophysiology is manifested primarily in brain neurons. To determine if FAD PS1 expression was associated with altered InsP3R-mediated Ca2+ release in brain neurons, cortical neurons were isolated from E15-E16 mouse brains and the Ca2+ permeability properties of the ER membranes were measured as above (Fig. 7A). The steady-state level of Ca2+ accumulated in the ER lumen was not different between control cells and neurons expressing either WT- or M146L-PS1 (Fig. 7B). The Ca2+ leak permeability measured upon addition of thapsigargin was somewhat higher in PS1-M146L expressing neurons than in those expressing PS1-WT (p < 0.05), although it was not different compared with control cells (Fig. 7C). In a separate set of experiments (Fig. 7D), 33 nM InsP3 failed to elicit Ca2+ release in control or PS1-WT cells, whereas it simulated Ca2+ release from the ER of PS1-M146L cells (Fig. 7E), with an initial rate that was comparable to that achieved by addition of saturating (10 μM) InsP3 to the control cells (Fig. 7F). Thus, as in the non-neuronal cells examined, FAD mutant PS1 specifically sensitizes InsP3R-mediated Ca2+ release in brain cortical neurons.
Identification of a molecular mechanism that links FAD mutant PS to altered [Ca2+]i signaling provides an opportunity for insights into relationships between pathological features of AD and altered [Ca2+]i signaling. PS1 is the core subunit of the γ-secretase that enzymatically cleaves APP into amyloid peptides, including Aβ40 and Aβ42 (Edbauer et al., 2003). FAD PS mutations alter secretase function by either modifying its sequence specificity or absolute activity, such that the relative proportion or amount of Aβ42 produced is increased (Citron et al., 1997; Scheuner et al., 1996). To determine the relevance of the functional interaction of PS1 and InsP3R for APP processing, we engineered DT40 cells to stably express APP harboring Swedish mutations (APPSWE) that enhance production of Aβ species (Scheuner et al., 1996), together with either PS1-WT or PS1-M146L (Fig. 8A). PS1-M146L specifically enhanced Aβ40 and Aβ42 by ~2- and ~3-fold, respectively, compared with control cells. This result is consistent with observations in other cell types (Citron et al., 1997), validating again the use of this model system. Of note, the Aβ42/Aβ40 ratio was enhanced in the FAD mutant PS1-expressing cells (Fig. 8), as observed in AD patients. To determine the role of the InsP3R in PS1-dependent APP processing, APP and PS1-expxressing cells were generated in the InsP3R-KO background, and APP processing was similarly evaluated. Remarkably, the mutant PS1 enhancement of Aβ secretion observed in the InsP3R-expressing cells was abolished. Furthermore, the absolute levels of Aβ peptides detected were strongly reduced in all control and PS1-expressing InsP3R-KO lines (Fig. 8B). These results indicate that altered APP processing by mutant PS1-M146L PS1 has a strong dependence on the InsP3R.
The underlying pathogenic mechanisms of Alzheimer’s Disease remain obscure. In this study, we considered that dysregulated [Ca2+]i signaling is a proximal mechanism in AD. We used non-neuronal model cell systems because they provided unique advantages for deciphering the molecular mechanisms involved, and confirmed the major results in primary brain neurons. Both the InsP3R (Foskett et al., 2007) as well as presenilins (Hebert et al., 2004) are widely distributed throughout all tissues investigated, with the highest levels of PS expression outside the brain (Hebert et al., 2004), and similar Ca2+ signaling abnormalities have been observed in several peripheral and neuronal cell types that express mutant PS (LaFerla, 2002; Smith et al., 2005a; Stutzmann, 2005). We have demonstrated that FAD mutant presenilins interact biochemically and functionally with the InsP3R Ca2+ release channel and exert profound stimulatory effects on its gating activity that result in exaggerated Ca2+ signaling in intact cells, including brain neurons. Our results indicate that this functional interaction has physiological implications that may be relevant in AD, including APP processing.
We observed that expression of two FAD mutant PS (PS1-M146L and PS2-N141I) each had strong and similar effects on the gating of single InsP3R Ca2+ release channels. We attempted to ensure equal levels of expression of InsP3R and the different PS proteins among the different groups of cells, but this could not be achieved perfectly. Nevertheless, because very similar results were obtained in two very distinct cell systems, Sf9 and DT40 cells, this was likely not a significant confounding variable. In both systems, mutant PS expression caused the InsP3R channel to gate as actively in low sub-maximal [InsP3] as it normally does in saturating [InsP3]. Elevated Po in low [InsP3] by FAD mutant PS was caused primarily by a substantial increase of channel opening rate, the major mechanism whereby InsP3 activates the channel (Foskett et al., 2007), indicating that mutant PS enhances InsP3R gating by sensitizing it to low [InsP3]. This mechanism is likely an allosteric one mediated by interaction of PS with the InsP3R that is preserved in isolated nuclei. Both membrane proteins localized to the nuclear envelope and mutant- as well as WT-PS co-immunoprecipitated to the same extent with mammalian InsP3R, suggesting that PS mutations do not strongly affect the biochemical interaction. Although WT PS1 or PS2 did not affect channel Po, small effects on InsP3R channel gating were nevertheless observed. It has been debated whether FAD PS mutations are gain- or loss-of-function (Shen and Kelleher, 2007). Here, we show an apparent gain in InsP3R function by FAD mutant PS expression. However, channel dwell time analyses suggest that WT PS1 and PS2 also influence InsP3R channel gating. It is possible therefore that the gain-of-function phenotype we observe is due to disruption by mutant PS of normal PS-WT regulation of InsP3R channel activity. Future studies of the InsP3R gating behaviors in PS-deficient cells will be informative in this regard.
Analysis of our single channel data suggests that mutant PS1 and PS2 stimulate InsP3R gating by sensitizing the channel to InsP3. One of the most consistent observations of the effects of FAD PS on Ca2+ signaling is potentiation of InsP3-mediated Ca2+ release (Smith et al., 2005a; Yoo et al., 2000). Nevertheless, the molecular mechanisms have remained obscure. InsP3-mediated [Ca2+]i signals in FAD patient fibroblasts were exaggerated in response to low agonist concentrations (Ito et al., 1994), consistent with an enhanced sensitivity to InsP3 in AD. The magnitude and rate of [Ca2+]i rise in response to photo-release of InsP3 were enhanced by FAD mutant PS in Xenopus oocytes (Leissring et al., 1999a; Leissring et al., 1999b). Furthermore, Ca2+ puffs, low-level Ca2+ release events mediated by clusters of InsP3R in response to sub-maximal [InsP3], were both more frequent and likely to generate [Ca2+]i waves in the FAD mutant PS-expressing cells (Leissring et al., 2001). In brain slices from PS1-M146V knock-in mice, photo-release of InsP3 caused 3-fold greater Ca2+ release than in non-transgenic neurons, and enhanced the number of neurons responding and those exhibiting strong responses (Stutzmann et al., 2004). By demonstrating that mutant PS sensitize the InsP3R to InsP3, our new results provide a molecular mechanism that is consistent with these previous observations and can therefore possibly account for them. To test this idea, the relevance of the observed effects of FAD PS1 expression on InsP3R single-channel gating was determined by measuring [Ca2+]i responses in PS-expressing DT40 cells, the same cells used for the single channel studies, that had comparable levels of InsP3R expression. Exaggerated Ca2+ release responses to both high as well as low threshold concentrations of agonist were observed in the cells that expressed the PS1-M146L protein. In a significant proportion of the FAD mutant PS-expressing cells, the [Ca2+]i responses to low agonist concentrations mimicked the responses of normal cells to a saturating agonist concentration. Furthermore, a high percentage of PS1-M146L-expressing unstimulated cells displayed spontaneous low-level [Ca2+]i signals in the absence of agonist stimulation. Most importantly, the exaggerated Ca2+ responses were mediated specifically by the InsP3R, since they were absent in InsP3R knockout cells. All these features are consistent with an enhanced InsP3-sensitivity of the InsP3R in intact cells expressing FAD mutant PS, even to [InsP3] that may exist in resting, unstimulated cells. The conclusions based on the effects of PS1-M146L expression on [Ca2+]i signals are therefore in strong agreement with those reached from the single channel studies. Thus, these results suggest that exaggerated [Ca2+]i signals in cells can be accounted for by exaggerated InsP3R gating as a result of its interaction with FAD mutant PS. Importantly, these insights appears to be relevant for brain neurons, since InsP3R-mediated ER Ca2+ permeability was enhanced by FAD mutant PS1 expression in mouse brain cortical neurons, enabling Ca2+ release to be triggered by low [InsP3] that were without effect in either control cells or cells expressing WT PS1.
It has been suggested that exaggerated Ca2+ release in AD cells is due to increased filling of ER Ca2+ stores (discussed in (LaFerla, 2002; Mattson and Chan, 2003; Stutzmann, 2005). Although this “Ca2+ overload” hypothesis has been widely invoked, many studies, including ours here, have observed either no alteration or reduced ER Ca2+ stores in FAD PS-expressing cells (Giacomello et al., 2005; Lessard et al., 2005; Zatti et al., 2006; Zatti et al., 2004). The absence of elevated ER [Ca2+] suggests that the effects of FAD PS on InsP3R gating observed here cannot be accounted for by possible effects of elevated luminal [Ca2+] on channel function. Furthermore, we have observed in excised luminal side-out nuclear patches that InsP3R channel gating is insensitive to bath (luminal) [Ca2+] between 100 nM and 500 μM (data not shown). Ca2+ release is affected by many variables, including activities and expression of release channels, amount of ER Ca2+, Ca2+ buffering, and more. Enhanced spontaneous Ca2+ release activity observed in our PS1-M146L-expressing DT40 cells reduced ER Ca2+, but [Ca2+]i responses to agonists were nevertheless enhanced because of the strong effect of mutant PS on InsP3R gating. Thus, exaggerated Ca2+ release responses cannot provide unambiguous measurements of the amount of Ca2+ in stores. Our results demonstrate that effects of FAD mutant PS1 on both [Ca2+]i signaling and ER Ca2+ content are InsP3R dependent, with mutant PS1and PS2-stimulated InsP3R channels remaining regulated by InsP3. Thus, cell-type specific differences in concentrations of InsP3R ligands and expression levels can contribute to the magnitude of the effect of mutant PS1 expression on ER Ca2+, which might reconcile discrepant published observations. The enhanced Ca2+ release responses observed in PS1-M146L-expressing DT40 cells are highly reminiscent of those in peripheral and neuronal mutant PS-expressing cells in many previous studies. The absence of these responses in InsP3R-knockout cells, together with the identification of a molecular mechanism observed at the single channel level, strongly suggest that mutant PS-mediated enhancement of InsP3 sensitivity of the InsP3R is a fundamental underlying mechanism that accounts for many observations of dysregulated Ca2+ signaling in AD cells and model cell systems.
It has been reported that exaggerated Ca2+ responses in cells expressing FAD mutant PS are associated with enhanced expression or activities of ryanodine receptor (RyR) Ca2+ release channels (Chan et al., 2000; Smith et al., 2005b; Stutzmann et al., 2006). In AD-transgenic mouse cortical neurons, exaggerated responses to InsP3 were mediated in part by RyR activated by Ca2+ released through InsP3R (Stutzmann et al., 2006) suggesting that exaggerated RyR responses could be a secondary effect. Because Ca2+ signaling can influence InsP3R (Cai et al., 2004; Genazzani et al., 1999) and RyR (P. Nicotera, personal comm.) expression, it is possible that mutant PS-mediated enhanced InsP3R Ca2+ signaling drives transcriptional programs, with RyR expression up-regulated as a result. Conversely, Aβ exposure increased RyR expression in mouse cortical neurons (Supnet et al., 2006), suggesting that changes in RyR expression may be more downstream compared with a more proximal InsP3R-mediated process described here. Alternately, PS and RyR may functionally interact similar to the PS-InsP3R interaction. PS2 has been reported to interact with RyR2, the cardiac isoform, and PS2 knockout results in cardiac Ca2+ signaling abnormalities (Takeda et al., 2005). In addition, WT and mutant PS1 have been reported to interact with RyR3 (Chan et al., 2000).
It was proposed that reconstituted PS proteins form divalent cation-permeable ion channels in bilayer membranes, and that loss of this function in FAD PS enhanced ER Ca2+, resulting in exaggerated Ca2+ release responses (Nelson et al., 2007; Tu et al., 2006). In contrast, we did not detect in native ER membranes a cation permeability associated with either WT or FAD mutant PS expression. Furthermore, we found that FAD mutant PS expression either reduced (DT40 cells) or had no effect (cortical neurons) on the amount of Ca2+ in the ER. Thus, our experiments failed to discover evidence in support of the hypothesis that PS form ion channels.
Finally, it should be emphasized that the Ca2+ signaling abnormalities we observed represent proximal mechanisms distinct from effects of Aβ on Ca2+ signaling (Mattson and Guo, 1997). Our studies indicate the presence of a complex involving PS and InsP3R that results in hyper-activation of the Ca2+ release channel in a process that is independent of Aβ.
Disrupted Ca2+ signaling in AD cells has been well documented, but its physiological implications and the roles that these changes play in AD pathogenesis have not been well studied. Altered Ca2+ signaling could impinge on synaptic plasticity, membrane excitability, oxidative stress and APP processing. Identification of the InsP3R as a molecular mechanism of Ca2+ disruption associated with FAD PS now suggests specific hypotheses that can be tested to evaluate the relevance of altered Ca2+ signaling through this pathway on disease-associated processes.
Our data suggest that one function of PS proteins may be to regulate the activity of the InsP3R. Is there a relationship between this activity and the activity of PS as secretases? APP processing can be enhanced by elevations of [Ca2+]i (Buxbaum et al., 1994; Jolly-Tornetta et al., 1998; Querfurth and Selkoe, 1994) and diminished by inhibition of ER Ca2+ release (Buxbaum et al., 1994). APP processing was examined in DT40 cells lines engineered to stably express PS1 and APPSWE, exploiting DT40 KO cells as the only cell type available that completely lacks InsP3R expression. Expression of PS1-M146L caused a nearly 3-fold increase in Aβ42 compared with control and PS1-WT expressing cells. Consequently, Aβ42/Aβ40 increased ~2-fold, recapitulating an important feature of AD (Haass and Selkoe, 2007). Importantly, APP processing appeared to be strongly dependent on the InsP3R, since production of Aβ40 and Aβ42 were substantially lower in InsP3R-deficient lines. Furthermore, the PS1-M146L enhancement Aβ42 and Aβ42/Aβ40 were eliminated in the InsP3R-KO cells. These results suggest that the γ-secretase activity of WT and FAD mutant PS may be regulated by either Ca2+ released through the InsP3R, or their biochemical interaction with the channel. Our studies suggest that InsP3R activity affects APP processing, but pharmacological inhibition of γ-secretase is without effect on InsP3R-mediated Ca2+ signaling (Oh and Turner, 2006). Thus, the γ-secretase activity of PS is likely not involved in this mechanism by which mutant PS regulates InsP3R channel gating.
The dependence of Aβ production on InsP3R expression observed here suggests that mutant PS-mediated exaggerated Ca2+ signaling could be a proximal mechanism in AD. Brain Aβ production is driven by neuronal activity (Cirrito et al., 2005; Kamenetz et al., 2003), suggesting that activity and metabolic patterns may contribute to Aβ production and amyloid deposition (Buckner et al., 2005). As yet unspecified activity- or metabolism-dependent mechanisms may cause preferential accumulation of amyloid that, over many years, may participate in AD pathology (Buckner et al., 2005). Because of a central role of Ca2+ in regulating both neuronal excitability (Verkhratsky, 2005) and cell metabolism (Balaban, 2002; McCormack et al., 1990), active brain regions are likely sites of higher Ca2+ signaling activity. Mutant PS-mediated exaggerated Ca2+ release activity of the InsP3R may therefore provide a mechanism for preferential accumulation of amyloid. Other mechanisms that do not involve either PS or InsP3R, but that similarly cause chronic low-level exaggerated Ca2+ signaling, might also be expected to result in AD pathology. Although speculative, such mechanisms could possibly be involved in sporadic forms of AD.
In summary, our results indicate that PS interact with the InsP3R Ca2+ release channel and modulate its gating activity. FAD mutant PS1 and PS2 exert stimulatory effects on InsP3R channel activity that result in perturbed cellular Ca2+ signaling. These data provide molecular insights into the mechanisms of enhanced InsP3-mediated Ca2+ signals observed in cells that express FAD mutant presenilins, including those from AD patients. Enhanced Ca2+ release from the ER as a result of this interaction has physiological implications that may be relevant in AD, including enhanced APP processing. These observations may provide unique molecular insights into the “Ca2+ dysregulation hypothesis” of AD pathogenesis and suggest novel targets for therapeutic intervention.
Spodoptera frugiperda cells (Sf9, BD Biosciences, San Jose, CA) were maintained as described (Ionescu et al., 2006). Human PS baculovirus constructs (PS1-WT, PS1-M146L, PS2-WT and PS2-N141I) were subcloned into pFastBac1; baculoviruses were generated using the Bac-to-Bac system (Invitrogen). Expression was confirmed by western blotting with anti-PS1 (monoclonal MAB5232, polyclonal MAB1563, Chemicon Int., Inc.) and -PS2 (EMD Chemicals Inc., San Diego, CA) antibodies. Localization of PS in Sf9 cells was confirmed by immunocytochemistry with anti-PS1 or -PS2 antibodies. Sf9 nuclei were counterstained with TOTO-3 nuclear dye (Molecular Probes, Eugene, OR).
DT40 cells were maintained as described (White et al., 2005). Human wild-type (WT) PS1 and M146L cDNAs were subcloned into pIRES2-EGFP (Clontech, Palo Alto, CA). Cells were transfected using a Nucleofector Device (Amaxa, Gaithersburg, MD). To select stable polyclonal lines, transfected cells were cultured for 2 weeks in 2 mg ml-1 Geneticin (Invitrogen). PS expression was confirmed by western blot. Human APP harboring Swedish mutations (APPSWE) was introduced into PS1-expressing DT 40 lines by retrovirus infection. APPSWE cDNA was subcloned into pΔMX-IRES-dsRED retrovirus vector. APPSWE retrovirus was generated using Retro-X system (Clontech). Expression of APPSWE was confirmed by western blotting using a polyclonal anti-APP antibody. Primary cortical neurons were prepared from embryonic day 15 (E15-16) C57BL/6J mice, as described (Meberg and Miller, 2003). Transfections were performed on 7-14 day-old cultures with pIRES2-EGFP-PS1WT, pIRES2-EGFP-M146L or pIRES2-EGFP empty vector by n-FectTM reagent (Neuromics, Edina, MN). Experiments performed 48 hr after transfection.
Preparation of isolated nuclei from Sf9 or DT40 cells was as described (White et al., 2005). Nuclei were studied in standard bath solution (in mM): 140 KCl, 10 HEPES and 0.5 BAPTA (free [Ca2+] = 300 nM), pH 7.3. The pipette solution contained (in mM): 140 KCl, 0.5 ATP, 10 HEPES, 1 μM free Ca2+, pH 7.3. Free [Ca2+] in all solutions was adjusted, as described (Mak et al., 1998). Data were acquired at room temperature and analyzed as described (Ionescu et al., 2006).
[Ca2+]i was measured in fura-2 loaded DT40 cells as described (White et al., 2005). ER Ca2+ content was estimated from changes in [Ca2+]i in response to thapsigargin or ionomycin, as described (White et al., 2005), or with the low affinity Ca2+ indicator Mag-Fura2 (Invitrogen) following procedures similar to those described in (Laude et al., 2005) modified for single cell imaging. Following loading of cells on coverslips with Mag-Fura2 (10 μM for 60 min), the plasma membranes were permeabilized by exposure to 10 μg/ml digitonin for 2 min in a medium that contained 220 nM Ca2+ and lacked MgATP . Following 30 min of continous perfusion with the MgATP-free bath to wash out the digitonin, cells were alternately illuminated with 340/380 nm light and fluorescence intensity at 510 nm was collected with a Perkin Elmer Ultraview imaging system. Changes in [Ca2+]ER are presented as changes in fluorescence ratio (R/R0). R0 was verified to reflect the ratio for Ca2+ depleted stores by using ionomycin (1 μM), as described (Laude et al., 2005). Ca2+ leak rates expressed as Δ(R/R0) sec-1.
Aβ levels in culture media were measured by sandwich ELISA (Suzuki et al., 1994). Briefly, plates were coated with Aβ N-terminal antibody Ban50 prior to application of cell media. The concentration of Aβ was determined using horseradish peroxidase (HRP)-conjugated BA27 or BC05 antibodies to detect Aβ40 or Aβ42, respectively. Aβ40 or Aβ42 levels were normalized to total protein concentration. Standard curves, constructed from serial dilutions of synthetic Aβ40 or Aβ42, were generated in each experiment.
Data were summarized as the mean ± s.e.m. with statistical significance of differences between means was assessed using unpaired t-tests or analysis of variance (ANOVA) for repeated measures at the 95% level (p < 0.05).
We thank Ikuo Hayashi for recombinant baculoviruses, and Eric Swanson for help with neuron isolation. Supported by NIH GM56328 and MH059937 to JKF and a Pilot Project from the Alzheimer’s Disease Core Center at the University of Pennsylvania AG 10124.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.