Loosening ER–Mitochondria Coupling by the Expression of the Presenilin 2 Loop Domain

Presenilin 2 (PS2), one of the three proteins in which mutations are linked to familial Alzheimer’s disease (FAD), exerts different functions within the cell independently of being part of the γ-secretase complex, thus unrelated to toxic amyloid peptide formation. In particular, its enrichment in endoplasmic reticulum (ER) membrane domains close to mitochondria (i.e., mitochondria-associated membranes, MAM) enables PS2 to modulate multiple processes taking place on these signaling hubs, such as Ca2+ handling and lipid synthesis. Importantly, upregulated MAM function appears to be critical in AD pathogenesis. We previously showed that FAD-PS2 mutants reinforce ER–mitochondria tethering, by interfering with the activity of mitofusin 2, favoring their Ca2+ crosstalk. Here, we deepened the molecular mechanism underlying PS2 activity on ER–mitochondria tethering, identifying its protein loop as an essential domain to mediate the reinforced ER–mitochondria connection in FAD-PS2 models. Moreover, we introduced a novel tool, the PS2 loop domain targeted to the outer mitochondrial membrane, Mit-PS2-LOOP, that is able to counteract the activity of FAD-PS2 on organelle tethering, which possibly helps in recovering the FAD-PS2-associated cellular alterations linked to an increased organelle coupling.


Introduction
Presenilin 1 and 2 (PS1, PS2) are two highly conserved, widely expressed, multipass transmembrane proteins mainly known as constituents of the catalytic core of the γ-secretase complex that is involved in the pathogenesis of Alzheimer's disease (AD). The high molecular weight aspartyl-protease complex is formed by either PS1 or PS2 (cleaved into the N-and C-terminal fragments, NTF and CTF, respectively) and three other additional proteins, anterior pharynx defective 1 (APH1), presenilin enhancer (PEN-2) and nicastrin [1,2]. The enzymatic complex catalyses endomembranous proteolysis of different type I membrane proteins, with Amyloid Precursor Protein (APP) being the most studied γ-secretase substrate because of its involvement in the formation of amyloid β (Aβ) peptides. Indeed, the toxic effects of intracellular Aβ accumulation, as well as its extracellular deposition into Aβ plaques in the brain, are typical hallmarks associated with AD [3]. Importantly, different mutations in the genes encoding PS1, PS2 and APP are responsible for early-onset familial AD (FAD; [3]), indicating the crucial part played by the γ-secretase-dependent APP-processing pathway in disease onset. Accumulating evidence, however, supports the existence of additional and yet unidentified pathogenic mechanisms, further sustained by the fact that the majority of clinical trials targeting the Aβ pathway have failed to modify disease progression [4] (but see recent promising results for Lecanemab and Aducanumab; https://www.alzforum.org/therapeutics/lecanemab; accessed on 20 July 2021).
Beyond their well-defined catalytic role within the γ-secretase complex, both PS1 and PS2, as individual proteins, exert specific activities that modulate additional important cell Rv: CCGGAATTCTCACAGGTCTTCTTCAGAGATCAGTTTCTGTTCGATGTAGAGCT GATGGGA The PCR product was digested with the BamHI and EcoRI restriction enzymes (Fast-Digest, Thermo Scientific) and ligated into pcDNA3.
The PS2-LOOP was amplified from a plasmid bearing the cDNA of PS2 WT with the following primers (5 -3 ): Fw: CGCGATATCGGTGGAGGTGGAGGTGACTACAAAGACGACGACGACAA-GATGGCGAAGCTGGACC Rv: GGCCTCGAGTTAGCGGCCGCTCTTCACGCCCCTTTC In the forward primer, before the region annealing with PS2-LOOP, a disordered linker was inserted (containing a FLAG-tag) to allow proper folding.
The PCR product was digested using EcoRV and XhoI restriction enzymes (Thermo Scientific).
The mitochondrial targeting sequence of AKAP1 was amplified from a previously described AKAP1-bearing plasmid (OMM-RFP, a gift from G. Hajnoczky, [33]) using the following primers (5 -3 ): Fw: CGCAAGCTTGCCACCATGG Rv: GCGGGTACCGCGCCCACTGCCTTTTCCTTTTCCAGATCCACCAGATTTAGATAG GATAGCACCAGC In the reverse primer, a disordered linker was included to permit all portions of the construct to fold independently.
The PCR product was digested using HindIII and KpnI restriction enzymes (Thermo Scientific). The digested products were ligated in a pcDNA3 vector.
To generate ER-tdTomato, the ER targeting sequence from CYP450-2C1, which anchors the fusion protein to the ER membranes [34], was amplified from the ER-ABKAR construct (a gift from Takanari Inoue and Jin Zhang, Addgene plasmid # 61508) with the following primers (5 -3 ): Fw: CGCAAGCTTGCCACCATGGACCCTGTGGTG Rv: GCCGATATCGGTACCTCCAGATCCACCAGACCCTCCCCCATAGC In the reverse primer, a disordered linker was included to permit all portions of the construct to fold independently.
Where indicated, to partially pre-deplete control cells of their ER Ca 2+ content, cells were treated for 120 s in Ca 2+ -free EGTA-containing mKRB supplemented with CPA (10 µM), before stimulation with BK.
Images were collected with a Leica SP5 confocal microscope (DM IRE2) by WLL laser, acquiring the different colour channels independently. Alexa 488, cytosolic and mitochondrial GFP, along with SPLICSs, were excited at 488 nm, whereas for TdTomato it was at 555 nm.
The collected images were analysed by ImageJ or Fiji (NIH). To evaluate lipid droplet or SPLICSs staining, multi-stack images were collected, keeping the same parameters during acquisition (zoom, laser intensity, PMT gain). For SPLICSs analysis, images were background-subtracted and the Convolve filter was applied to minimize dot fragmentation. Dots were counted using the 3D Objects Counter plugin, considering only objects in the range 5-500 pixels, to avoid recording noise or aggregates, respectively.
Stained lipid droplets were excited at 485 nm and their number, volume and surface were calculated through all stacks by the 3D Objects Counter plugin after calculating a threshold (corresponding to 2× the mean fluorescence intensity of each cell), considering only objects in the range 5-500 pixels.
The mitochondrial morphological analysis was performed on single-stack images subtracting the background. Mean and Noise Despeckle ImageJ plugins were applied and a binary image was obtained by setting an automatic threshold, to better resolve the mitochondrial network. Mitochondrial mean area, perimeter, circularity and aspect ratio values were calculated by the Analyze Particle ImageJ plugin, applying a cut-off of 3 pixels.
In the representative images, after quantification performed as detailed above, the signals were enhanced with the automatic ImageJ plugins Brightness/Contrast and Enhance Image, to better appreciate low fluorescent signals.

Statistical Analysis
All data are representative of at least 3 independent experiments. Significance was calculated by unpaired Student's t-test for normal distributions and Wilcoxon Mann-Whitney test for not normally distributed data. * = p < 0.05, ** = p < 0.01, *** = p < 0.001. Values are reported as mean ± SEM. For additional information, see Table S1.

The PS2 C-Terminal Fragment, but Not Its Loop Domain, Retains the Effects of Full-Length PS2 on ER-Mitochondria Ca 2+ Transfer
To define the PS2 domain mediating the effects of the protein on ER-mitochondria Ca 2+ shuttling, we expressed in SH-SY5Y cells the PS2-CTF (aa 298-448, physiologically generated upon the autoproteolytic cleavage of the full-length protein [36]) fused to a Myc-tag, as well as the entire Myc-tagged-PS2, for comparison. Immunofluorescence staining revealed that both PS2 and PS2-CTF localize in endomembranes, mostly in the ER ( Figure 1A). Importantly, upon maximal IP3R-dependent ER Ca 2+ release, the expression of either PS2 or PS2-CTF induced a similar decrease in cytosolic Ca 2+ peaks, compared to control cells ( Figure 1B). We previously reported that this phenomenon depends on a reduced ER Ca 2+ content induced by PS2 expression [14,15,21], suggesting that PS2-CTF conserves the same effect. On this line, since mitochondrial Ca 2+ uptake is deeply affected by the amount of Ca 2+ released from the ER through IP3Rs, lower mitochondrial Ca 2+ peaks were observed in PS2-or PS2-CTF-expressing cells ( Figure 1C), compared to controls.  To better investigate the process of mitochondrial Ca 2+ uptake independently of the different Ca 2+ levels reached in the cytosol upon cell stimulation, the amplitude of cytosolic Ca 2+ peaks in control cells were matched with those observed in PS2-or PS2-CTFexpressing cells by a pre-depletion protocol (CTRL pre-depleted, Figure 1B; see Materials and Methods). Notably, the mitochondrial Ca 2+ peaks in PS2-or PS2-CTF-expressing cells were substantially increased compared to those of pre-depleted controls ( Figure 1C), hinting at a higher efficiency of ER-mitochondria Ca 2+ shuttling. This is in line with our previous finding of increased physical and functional ER-mitochondria tethering in cells expressing PS2, in particular FAD-PS2 mutants [18,19], again indicating that PS2-CTF exerts a similar function to the full-length protein.
We previously demonstrated that the activity of PS2 on ER-mitochondria coupling depends on its binding to MFN2 [19]. Moreover, we reported that PS2-CTF, as well as a truncated form of PS2 (PS2-∆374-448, lacking the final part of PS2-CTF but containing its large cytosolic loop (PS2-LOOP; Figure 2A), interacts with MFN2, whereas the PS2-NTF does not [19]. These data suggest that PS2-LOOP might be critical for PS2-MFN2 interaction and the potentiation of ER-mitochondria coupling. We therefore tested whether the expression of PS2-LOOP alone (PS2 aa 305-361; Figure 2A), without the other PS2 protein domains, affects organelle coupling. To visualize its subcellular localization, PS2-LOOP was fused with the fluorescent protein tdTomato (Cyt-PS2-LOOP, Figure 2B); a construct encoding only the tdTomato (Cyt-CTRL) was also generated, as a control ( Figure 2B). Importantly, both constructs are efficiently expressed at comparable levels, as revealed by Western Blot (WB) analysis ( Figure 2C) and present a similar intracellular distribution, as shown by the co-expression of either Cyt-CTRL or Cyt-PS2-LOOP with a cytosolic GFP in SH-SY5Y cells ( Figure 2D). Representative cytosolic (B) and mitochondrial (C) Ca 2+ traces in control (black), pre-depleted control (red), PS2-CTF-(green) and PS2 WT-(orange) expressing SH-SY5Y cells stimulated with BK (100 nM) and CPA (10 μM). On the right of each panel, the bar graph represents the mean [Ca 2+ ] peak values obtained upon cell stimulation (mean ± SEM; number of independent experiments: cytosolic data: CTRL n = 5; CTRL pre-emptied n = 15; PS2-CTF n = 8; PS2-WT n = 7; mitochondrial data: CTRL n = 5; CTRL pre-depleted n = 11; PS2-CTF n = 9; PS2-WT n = 9). **= p < 0.01, *** = p < 0.001  We then evaluated whether Cyt-PS2-LOOP maintains an effect on ER-mitochondria Ca 2+ transfer. When expressed in SH-SY5Y cells, no significant differences in the cytosolic ( Figure 2E) or in the mitochondrial ( Figure 2F) Ca 2+ peaks (obtained by either bradykinin (BK) or fetal calf serum (FCS), to induce respectively, a maximal or milder, more physiological, IP3R-dependent ER Ca 2+ release [20]), were observed compared to control cells. The lack of significant effects on cellular Ca 2+ handling suggests that Cyt-PS2-LOOP, when expressed free in the cytosol, does not maintain the capacity of PS2 (as well as of PS2-CTF; see Figure 1C) to impact on ER-mitochondria coupling.

The Mitochondria-Targeted PS2 Loop Domain Exerts an Opposite Effect Compared to Full-Length PS2 on ER-Mitochondria Coupling
Since PS2 (as well as PS2-CTF) is enriched in MAM [19,29], whereby its binding to MFN2 and its activity on organelle tethering might be favoured by an advantageous stoichiometry, we reasoned that the PS2 loop domain could also benefit from a proper localization/enrichment. To test this hypothesis, we therefore targeted, by an appropriate targeting sequence (see Materials and Methods), the fused protein tdTomato-PS2-LOOP (Mit-PS2-LOOP), or the tdTomato protein alone (Mit-CTRL) as a control, to the outer mitochondrial membrane (OMM) ( Figure 3A). Both Mit-CTRL and Mit-PS2-LOOP correctly localized to the mitochondrial network ( Figure 3B and Figure S1A) and were expressed at similar levels within cells ( Figure 3C). Upon BK-or FCS-induced stimulation of SH-SY5Y cells, the cytosolic Ca 2+ peaks that resulted were unaffected by the expression of MIT-PS2-LOOP ( Figure 3D). Importantly, however, the corresponding mitochondrial Ca 2+ peaks, obtained by the same stimulations, were significantly lower in cells expressing Mit-PS2-LOOP, compared to those observed in Mit-CTRL-expressing cells (Figure 3E), suggesting a reduced efficiency of ER-mitochondria Ca 2+ transfer in this condition, and implying an opposite effect of Mit-PS2-LOOP with respect to the full-length PS2 protein or its CTF (see Figure 1C). Indeed, the reduced mitochondrial Ca 2+ rises induced by Mit-PS2-LOOP expression do not depend on an intrinsically defective capacity of mitochondria to take up Ca 2+ , as revealed by the similar cytosolic and mitochondrial Ca 2+ peaks observed upon activation of store-operated Ca 2+ entry (SOCE; Figure 3F), i.e., a condition in which cytosolic and mitochondrial Ca 2+ elevations are induced by cation entry through the plasma membrane, rather than by its release from the ER.
A similar reduction in IP3R-dependent mitochondrial Ca 2+ peaks was observed in Mit-PS2-LOOP-expressing HeLa cells ( Figure S1B). Importantly, by a recently reported sensor for close (<8-10 nm) ER-mitochondria contacts (SPLICSs; [32]), a lower number of dots (representing organelle close contacts) was retrieved in Mit-PS2-LOOP-expressing cells, compared to controls ( Figure S1C), suggesting that a reduced ER-mitochondria tethering decreases the efficiency of Ca 2+ shuttling between the two organelles. Of note, mitochondrial morphology was not affected by Mit-PS2-LOOP expression ( Figure S1D).
We previously demonstrated that PS2 modulates ER-mitochondria tethering by binding to MFN2 and tuning its negative activity on inter-organelle coupling [19]. We therefore speculated that Mit-PS2-LOOP might compete with endogenous, WT or mutated PS2 for MFN2 binding, thus triggering opposite effects on ER-mitochondria coupling. In line with this hypothesis, we found that Mit-PS2-LOOP, but not Mit-CTRL, coimmunoprecipitates with endogenous MFN2 ( Figure 3G).

The Mitochondria-Targeted PS2 Loop Domain Normalizes ER-Mitochondria Tethering in FAD-PS2-N141I Patient-Derived Fibroblasts
We previously demonstrated that FAD-PS2-N141I patient-derived fibroblasts show increased ER-mitochondria tethering and functional Ca 2+ transfer compared to control cells obtained from age-matched healthy individuals [19], indicating that mutated PS2 potentiates organelle coupling at endogenous levels of expression. Both WT and FAD-PS2 forms interact with MFN2, with an increased efficiency of FAD-PS2 because of its enrichment in MAM, compared to the WT counterpart [19].
We thus wondered whether Mit-PS2-LOOP could be able to hamper the PS2-N141Imediated ER-mitochondria tethering in FAD patient-derived fibroblasts. To this purpose, both ER-mitochondria contact sites and organelle Ca 2+ transfer were investigated in control and FAD-PS2-N141I fibroblasts, expressing either Mit-PS2-LOOP or Mit-CTRL. The SPLICSs probe was expressed in both control and FAD-PS2-N141I fibroblasts to detect ER-mitochondria close contacts ( Figure 4A). As already reported [32], FAD-PS2-N141I patient-derived cells showed an increased number of SPLICSs-dependent dots ( Figure 4B), confirming the increased ER-mitochondria tethering found in different FAD-PS2 cell models [14,18,19], compared to controls. Importantly, the expression of Mit-PS2-LOOP in FAD-PS2-N141I fibroblasts fully corrects the parameter, lowering the number of SPLICSs dots to control values ( Figure 4B). Mit-CTRL or Mit-PS2-LOOP. MFN2 was precipitated by a specific monoclonal antibody, and the coprecipitated samples were analyzed by SDS-PAGE and probed with α-tdTomato and α-MFN2 antibodies, as indicated. In the negative controls, an irrelevant antibody (IgG) was used for the IP (n = 3 independent experiments).

The Mitochondria-Targeted PS2 Loop Domain Normalizes ER-Mitochondria Tethering in FAD-PS2-N141I Patient-Derived Fibroblasts
We previously demonstrated that FAD-PS2-N141I patient-derived fibroblasts show increased ER-mitochondria tethering and functional Ca 2+ transfer compared to control cells obtained from age-matched healthy individuals [19], indicating that mutated PS2 potentiates organelle coupling at endogenous levels of expression. Both WT and FAD-PS2 forms interact with MFN2, with an increased efficiency of FAD-PS2 because of its enrichment in MAM, compared to the WT counterpart [19].  As to organelle Ca 2+ handling, the BK-induced maximal ER Ca 2+ release caused blunted cytosolic Ca 2+ rises in FAD-PS2-N141I fibroblasts compared to controls ( Figure 4C), confirming the reduced ER Ca 2+ content found in these and other FAD-PS2-expressing cells [13][14][15]19]. Consequently, mitochondrial Ca 2+ rises induced by the same stimulation were also reduced in FAD-PS2 patient-derived cells compared to controls ( Figure 4D). In accordance with ER-mitochondria tethering data ( Figure 4A-B and Figure S1C), Mit-PS2-LOOP expression induced a further decrease in mitochondrial Ca 2+ rises of FAD-PS2-N141I fibroblasts ( Figure 4D), because its expression keeps the two organelles further apart, thus making their Ca 2+ transfer less efficient.

The Mitochondria-Targeted PS2 Loop Domain Rectifies the Increased Lipid Droplet Formation Found in FAD-PS2-N141I Patient-Derived Fibroblasts
Lipid droplets (LDs) are key ubiquitous eukaryotic organelles providing essential lipid precursors for membrane biogenesis (thus supporting organelle and cell growth) and acting as a sink for toxic fatty acids. They are connected to different organelles within the cells [37], being also identified in the proximity of both ER and mitochondria [38]. Interestingly, an increased LD formation has been linked to an altered ER-mitochondria tethering in sporadic and familial AD patient-derived fibroblasts [39]. We thus wondered whether the Mit-PS2-LOOP, being able to modulate ER-mitochondria tethering in FAD-PS2 cells, could also modify the formation of these structures in FAD patient-derived fibroblasts.
We first visualized LDs in control and FAD-PS2-N141I fibroblasts, finding an increase in the number and volume of these organelles in diseased cells, compared to controls ( Figure 4E-G), and in line with previous results [39]. Of note, when the Mit-PS2-LOOP was expressed, a reduction, although not significant, in LD number and a complete recovery in LD size was observed in FAD-PS2 fibroblasts ( Figure 4F,G).

Discussion
The dynamic interplay between intracellular organelles is crucial to carry out several physiological activities, from modulation and decoding of Ca 2+ signals to lipid synthesis and metabolism [40,41]. Specifically, the regions of close juxtaposition between organelle membranes host most of the enzymes/proteins mediating these cell functions. Moreover, the exchange of metabolites and/or information between organelles is favoured by their physical proximity, thus making contact areas, i.e., MAM domains in the case of ERmitochondria contacts, the ideal hubs whereby key cellular tasks take place. Therefore, it is not surprising that alterations of organelle contacts and/or functionality have been associated with different pathological conditions, such as diabetes, cancer and neurodegeneration [27,40,[42][43][44][45]. In particular, increasing evidence suggests that disturbances in ER-mitochondria connectivity are early events in neurodegenerative disorders such as Parkinson's disease, amyotrophic lateral sclerosis and AD [46]. On this latter disorder, the majority of studies have highlighted an upregulated MAM functionality in different experimental models [14,18,19,39,47,48], although the underlying mechanisms, as well as its impacts on the neurodegenerative process, are still debated. For instance, we previously demonstrated that PS2 (in particular its FAD mutants), but not PS1, directly strengthens ER-mitochondria coupling [14,18], by interacting with MFN2 and tuning its negative activity on organelle juxtaposition [19,25]. Here, on the one hand, we further defined the PS2 protein domain mediating its functional interaction with MFN2 and, on the other, we generated a tool to correct the alterations in ER-mitochondria tethering triggered by FAD-PS2 mutants.
First, we found that PS2-CTF retains the activity of PS2, sustaining the efficiency of ER-mitochondria Ca 2+ shuttling, while dampening the overall amplitude of IP3-linked cytosolic (and thus, mitochondrial) Ca 2+ rises, likely as consequence of a reduced ER Ca 2+ content. This result might be surprising, because different FAD-PS2 mutants, among those we previously demonstrated to potentiate ER-mitochondria coupling, display point mutations in the PS2-NTF. However, it should be considered that PS2-NTF and PS2-CTF originate from an autocleavage of full-length PS2, likely during its incorporation into γ-secretase [36]. Therefore, it is possible that while PS2-CTF modulates per se ER-mitochondria coupling, FAD-linked mutations enhance this activity by altering the subcellular localization of PS2-NTF/PS2-CTF, regardless of their position along the PS2 sequence. In line with this hypothesis, we previously observed that in the presence of FAD-PS2 mutants, both PS2-NTF and PS2-CTF are more enriched in MAM, compared to WT [19]. The reason underlying this distinct subcellular distribution of FAD-PS2 is unknown, but changes in the interaction with specific molecular partners might be involved.
Importantly, the PS2-CTF is composed of a large cytosolic loop followed by the remaining three TMD of the protein. Our data revealed that ER-mitochondria coupling is affected by the expression of the PS2-LOOP alone when targeted to the OMM (through an OMM-anchoring sequence), but not when expressed free in the cytosol. This further suggests that the enrichment of PS2 at specific subcellular domains is crucial to stoichiometrically favour its interaction with key molecular partners. Notably, we found that the OMM-targeted PS2-LOOP domain coimmunoprecipitates with MFN2, suggesting that it contains the minimal sequence mediating PS2-MFN2 interaction. Surprisingly, however, Mit-PS2-LOOP exerts opposite effects on ER-mitochondria tethering, compared to the entire PS2. This implies that, after binding to MFN2, PS2 performs an additional activity on this protein (such as the re-localization or prevention of MFN2 interaction with additional molecular partners), which is not conserved by Mit-PS2-LOOP. We speculated that Mit-PS2-LOOP might compete with PS2, in particular FAD-PS2, for MFN2 binding, thus hinting at the possibility to take advantage of this property to counteract the increased ER-mitochondria coupling observed in different FAD-PS2-expressing models. We tested this possibility by comparing the effects of Mit-PS2-LOOP expression in primary human fibroblasts from either a healthy donor or a FAD-PS2-N141I patient, wherein we previously reported, and we here confirmed (Figure 4), a larger physical and functional coupling between ER and mitochondria [19,32]. We indeed found that Mit-PS2-LOOP was able to rescue organelle tethering in FAD-PS2-N141I fibroblasts to levels similar to those observed in cells from a healthy donor (Figure 4), suggesting it might be particularly effective in the presence of FAD-PS2 mutants.
Importantly, to what extent the altered ER-mitochondria coupling observed in different AD models is involved in the neurodegenerative process remains an outstanding question. Nevertheless, increasing evidence suggests that pronounced alterations of lipid metabolism, such as hypercholesterolemia, are associated with AD onset and progression [49,50]. As a proof of this concept, the ε4 variant of apolipoprotein E (APOE), which works as a cholesterol/lipid transporter in the central nervous system, is considered the most frequent risk factor for AD development (reviewed in [51]). Remarkably, MAM are key regions for lipid synthesis [26], in particular for cholesterol metabolism. Indeed, free cholesterol is converted into cholesteryl esters (CEs) by acyl-CoA: cholesterol acyl transferase (ACAT1), an enzyme highly enriched in MAM [52], and key steps of phospholipid synthesis are typical activities localized to MAM [26]. In particular, an upregulated ACAT1mediated esterification of excessive free cholesterol to CEs, leading to the accumulation of LDs in the cytosol, has been reported in different AD models [39,53] and associated with an increased MAM functionality [39]. Consistently, we confirmed an increased number and size of LDs in FAD-PS2-N141I fibroblasts and, remarkably, we found that correction of ER-mitochondria tethering by Mit-PS2-LOOP expression correlates with a rescue of their size and a tendency to a reduction in their number, to levels similar to those observed in control cells. Additional investigations will be required to test whether these observations are maintained in cell types with a higher pathological relevance (such as neurons differentiated from FAD patient-derived induced pluripotent stem cells) and their possible role on disease progression.
As to mitochondrial Ca 2+ homeostasis, which is modulated by ER-mitochondria coupling, alterations in this signalling pathway have been reported in different neurodegenerative diseases, including AD [54]. In the specific case of FAD-PS2 mutants, upon IP3-dependent cell stimulation, a complex balance between a lower ER Ca 2+ content (which weakens the amplitude of Ca 2+ transfer to mitochondria) but a strengthened ERmitochondria tethering (which increases the efficiency of ER-mitochondria Ca 2+ shuttling), results in a decreased mitochondrial Ca 2+ signal [14,[18][19][20]. We associated this lower signal with a reduced mitochondrial ATP synthesis, affecting the overall cell bioenergetics [20]. Therefore, on this specific cell pathway, Mit-PS2-LOOP is unlikely to be helpful in the presence of FAD-PS2-mutants, because it further decreases mitochondrial Ca 2+ peaks by decoupling mitochondria from the ER. In this pathological context, we foresee that a combination of treatments recovering ER Ca 2+ content (such as by enhancing SERCA activity) with Mit-PS2-LOOP expression (which rescues organelle tethering) might be necessary to fully restore the ER-mitochondria axis.
We believe the modulation of ER-mitochondria juxtaposition is worthy of being harnessed as a promising pharmacological target in AD, because it has the potential to tune key pathways recently associated with disease onset and progression, such as altered mitochondrial Ca 2+ homeostasis and lipid metabolism.