Increased Mitochondrial Calcium Uptake and Concomitant Hyperactivity by Presenilin Loss Promotes mTORC1 Signaling to Drive Neurodegeneration

We show that both genetic and pharmacological inhibition of mTORC1 greatly improves proteostasis in sel-12 mutants. Furthermore, we demonstrate that sel-12 mutants have reduced autophagy that results from mTORC1 signaling activity. Finally, we show that improvements to proteostasis, neuronal function and neuronal health from mTORC1 inhibition in sel-12 mutants require the induction of autophagy. Our data indicate that elevated mitochondrial calcium uptake and mitochondrial hyperactivity resulting from SEL-12 loss disrupts proteostasis through mTORC1 activation and mTORC1-mediated suppression of autophagy, resulting in neurodegeneration.

function concerns the identi cation of the molecular pathways connecting mitochondrial calcium uptake to neurodegeneration and loss of proteostasis.
Here, we report that the increased mitochondrial calcium uptake and associated mitochondrial hyperactivity accompanying SEL-12 loss drives activation of the mTORC1 (mechanistic target of rapamycin complex I) pathway, which in turn impacts proteostasis and promotes neuronal dysfunction. We show that both genetic and pharmacological inhibition of mTORC1 greatly improves proteostasis in sel-12 mutants. Furthermore, we demonstrate that sel-12 mutants have reduced autophagy that results from mTORC1 signaling activity. Finally, we show that improvements to proteostasis, neuronal function and neuronal health from mTORC1 inhibition in sel-12 mutants require the induction of autophagy. Our data indicate that elevated mitochondrial calcium uptake and mitochondrial hyperactivity resulting from SEL-12 loss disrupts proteostasis through mTORC1 activation and mTORC1-mediated suppression of autophagy, resulting in neurodegeneration.

C. elegans maintenance and strains
For all experiments, C. elegans strains were grown on E. coli OP50 seeded NGM plates at 20°C. Animals were age synchronized by bleaching gravid worms to obtain the eggs, which were then incubated in M9 for 24-48 hours before being allowed to hatch. Afterward, L1 larvae were grown to adulthood on NGM plates for further experiments. Day 1 adults were analyzed for all experiments unless otherwise indicated.
For Q35 aggregation experiments, L4 animals were sterilized by moving them to plates containing 0.5 mg/ml 5-uorouraci1-2'-deoxyribose (Sigma) until the age required for the experiment was reached.

Analysis of neuronal morphology
The structure of the mechanosensory neurons was observed in either mec-7p::GFP(zdIs5) or mec-4p::mCherry(bzIs166)-expressing animals. Presence or absence of wave-like bending in the axon, lesions sprouting off the axon, or a beading-like pattern indicating breaks was determined. The number of sprouts stemming from the ALM soma were counted. The worms were immobilized in 0.3% sodium azide on 2% agarose pads, and imaged using the 63X oil objective on a Zeiss Axio Observer microscope equipped with an Andor Clara charged-coupled device (CCD) camera, and Metamorph software was used to compile the images.

Mitochondrial Organization Analysis
The organization of the mitochondria in the ALM mechanosensory neuronal soma was observed in animals expressing mec-4p::MLS::GFP(jsIs609). Animals were immobilized in 1M levamisole on 2% agarose pads, then imaged using 100X oil objective on a Zeiss Axio Observer microscope equipped with an Andor Clara CCD camera, and Metamorph software was used to compile the images. Mitochondrial organization was scored on a binary scale, where continuous mitochondria did not show breaks and discontinuous mitochondria appeared fragmented and showed breaks.
Heat stress assay For each strain, approximately 50 age-synchronized day 1 adults were placed in a water bath at 37°C for 1,2,3, and 4 hours, then allowed to recover at 20°C for 2 hours. Animals were scored as dead if they did not show any movement when prodded with a wire pick.

Mechanosensation assay
Day 1 adults' response to soft touch was determined using an eyebrow hair attached to a Pasteur pipette, with ten touches per worm, and with each stroke alternating across the anterior and posterior half of the worm as previously described (23). A positive response was scored when the animal moved away from the hair. The mean percentage of positive responses per worm was then determined.
Locomotion assay Day 1 adults were placed in a well that was coated with 2% agarose and lled with 150 uL M9 buffer. A single worm was transferred to an unseeded plate to remove OP50, then picked into the M9 buffer. After waiting one minute for the worm to acclimate, the number of full body bends was counted for one minute.
Quantitation of polyQ aggregates unc-54p::Q35::YFP(rmIs132)-expressing day 3 adult animals were immobilized in 0.1 µm diameter polystyrene microspheres (Polysciences) on 2% agarose pads, and were imaged using the 10X objective on a Zeiss Axio Observer microscope equipped with an Andor Clara CCD camera. Metamorph software was used to compile the images. The number of polyQ aggregates was counted, each aggregate determined as a structure fully discernible from any surrounding aggregates.

Autophagosome assay
We utilized animals expressing the lgg-1p::lgg-1::GFP (sqIs13) construct to label autophagosomes (24). Animals were immobilized in 0.1 µm diameter polystyrene microspheres (Polysciences) on 2% agarose pads. The number of aggregates per visible seam cell or per visible muscle cell was counted. Animals were viewed at 63x magni cation using Zeiss AxioObserver microscope equipped/captured with an Andor Clara CCD camera, and Metamorph software was used to compile the images.
Proteosome function assay rpl-28p::ub(G76V)::GFP(mgIs77)-expressing day 1 adult animals were immobilized in 1M levamisole on 2% agarose pads, and were imaged using the 10X objective on a Zeiss Axio Observer microscope equipped with an Andor Clara charged-coupled device CCD camera. Metamorph software was used to compile the images. The uorescence intensity was quanti ed using ImageJ.

Calcium Imaging
Mitochondrial calcium was measured in the mechanosensory neurons in animals expressing mec-7p::mito-GCaMP6f::SL2::wrmScarlet as previously described (23). In brief, animals were immobilized on 1M levamisole on 2% agarose pads. Images were taken using a 100X objective lens on a Zeiss Axio Observer microscope equipped with an Andor Clara CCD camera, and images were compiled with Metamorph software. Fluorescence intensity of GCaMP6 was normalized to wrmScarlet uorescent intensity, which was used as an expression control, and quanti ed using ImageJ. Cytoplasmic calcium levels were measured similarly in the mechanosensory neurons of animals expressing mec-4p::SL1::GCaMP3.35::SL2::mKate2, with uorescence intensity of GCaMP6 normalized to mKate2 intensity using ImageJ.

RNAi
RNAi was delivered by feeding as previously described (25). L1 animals were grown to adulthood on NGM plates seeded with HT115 bacteria expressing lgg-1 or bec-1 double stranded RNA, both from the Ahringer library (26), or empty RNAi feeding vector. RNAi bacteria strains were veri ed by PCR and DNA sequencing. Furthermore, to show speci city lgg-1(RNAi) and bec-1(RNAi) treatment abolished or reduced lgg-1::GFP, respectively.

Drug treatments
Rapamycin was prepared in dimethyl sulfoxide (DMSO) and added to plate agar at 100 µM. Equivalent DMSO was added to control plates. Bortezomib (LC Laboratories) was prepared in DMSO and added to plate agar at 10 μM. Doxycycline was added as described (23), with doxycycline prepared in and added to plates at a concentration of 10 µg/ml. Equivalent DMSO was also used in control plates. For rapamycin, doxycycline, and bortezomib treatments, L4 animals were grown on treated plates overnight, and the assays performed the following day on day 1 adults.

Statistical Analyses
All statistical analyses were conducted using Graph Pad Prism software. A p value of less than 0.05 is considered to be signi cant. Statistical difference was determined using a Student's T test for comparing two variables and a one-way analysis of variance with a Tukey post hoc analysis for comparing more than two variables.
Results sel-12 mutants show decreased autophagy that is rescued by reducing mitochondrial calcium uptake Like other age-related diseases, AD is associated with a general decline in proteostasis that contributes to the buildup of toxic protein aggregates, leading to neuronal dysfunction and death (27,28). It has been postulated that initial impairments to proteostasis pathways promote the accumulation and aggregation of misfolded proteins observed in neurodegenerative disorders, including the accumulation of the aggregation-prone Aβ and tau, whose plaque deposition and tangle formation, respectively, are considered the hallmarks of AD (29)(30)(31)(32)(33). Utilizing several models of proteotoxicity, we previously reported that sel-12 mutants have a severe defect in proteostasis resulting from elevated ER to mitochondrial calcium signaling (16). Two major systems are responsible for degrading misfolded or damaged proteins and are critical for maintaining integrity of the proteome: the ubiquitin-proteasome system and the autophagy-lysosomal pathway. To investigate the activity of these pathways in sel-12 mutants, we rst examined proteasome activity. As a reporter for proteasome function, we analyzed animals expressing ubiquitin (ub(G76V)) tagged to GFP, which is readily degraded by the proteasome in wild type animals, but accumulates if proteasome activity is perturbed (34) (Fig. 1A). Indeed, unlike animals treated with a proteasome inhibitor (bortezomib), animals carrying a sel-12 null allele, sel-12(ty11), showed similarly low uorescent intensity as wild type animals ( Fig. 1A-B), thus, indicating typical proteasomal degradation of the ub(G76V) tagged GFP. These data are consistent with normal proteasome activity of sel-12 mutants previously reported (16).
Next to investigate the activity of autophagy-lysosomal pathway in sel-12 mutants, we utilized GFP tagged LGG-1 to visualize autophagosome formation. LGG-1 is the C. elegans ortholog of LC3/Atg8, which is incorporated into and decorates pre-autophagosomal and autophagosomal membranes and organizes into puncta during autophagy and, thus, is a widely used autophagy marker (35)(36)(37). Compared to wild type animals, we observed a signi cant reduction in the number of LGG-1::GFP puncta in the hypodermal seam cells (Fig. 1C-D) and body wall muscle ( Fig. 1E-F) of sel-12 mutants. To determine whether elevated ER-mitochondrial calcium signaling in sel-12 mutants is responsible for decreased puncta formation, we introduced into the sel-12 mutant background a null mutation in the mitochondrial calcium uniporter, mcu-1, which reduces mitochondrial calcium uptake (23,38) and has been shown to reduce mitochondrial calcium levels, neurodegenerative phenotypes and proteostasis defects in sel-12 mutants (16,23). We found that introduction of an mcu-1 null mutation in the sel-12 mutant background increases puncta formation to levels indistinguishable from wild type animals (Fig. 1G). This suggests that altered mitochondrial calcium signaling is responsible for reducing autophagy in sel-12 mutants. As an alternative method to measure autophagic ux, we immunoblotted for GFP in the GFP::LGG-1 animals to assess lipidation of LGG-1 with phosphatidylethanolamine (PE). When autophagy is induced, PE is conjugated to LGG-1 to anchor it to the autophagosome. Thus, LGG-1-PE detection via western blotting is a reliable method to quantify the rate of autophagy (39). Consistent with this, GFP::LGG-1-PE levels signi cantly increase in raga-1 null mutants, which have been shown to have elevated autophagy via inhibition of mTORC1 signaling (40) (Fig. S1A,B). Consistent with our autophagosome puncta quanti cation, we found the amount of processed GFP::LGG-1-PE was signi cantly decreased in sel-12 mutants, and the level of processed GFP::LGG-1-PE was restored in mcu-1; sel-12 animals (Fig. 1H,I). Therefore, unlike proteasome activity in sel-12 mutants, these data suggest that autophagy is defective, which is consistent with previous observations of disrupted autophagy in sel-12 mutants as well as other models studying presenilin function (6,9,16,41,42) and also implicate a critical role of mitochondrial calcium in this defect.
Inhibition of mTORC1 in sel-12 mutants increases autophagy A central inhibitor of autophagy is the serine/threonine protein kinase mTORC1 signaling pathway (43,44). To investigate whether mTORC1 has a role in inhibiting autophagy in sel-12 mutants, we genetically ablated two key positive mediators of the mTORC1 pathway in sel-12 mutant animals. These include the gene encoding the RagA GTPase ortholog, raga-1, which is critical for the activation of mTORC1; and a gene encoding a key effector protein of mTORC1 signaling, ribosomal protein S6 kinase (rsks-1). Analysis of LGG-1::GFP puncta in raga-1(ok386); sel-12 and rsks-1(ok1255); sel-12 double mutant animals reveals that sel-12 mutants with mTORC1 signaling inhibited, unlike sel-12 mutants alone, show robust accumulation of LGG-1::GFP puncta (Fig. 1C-F). Moreover, consistent with mTORC1 acting as a strong inhibitor of autophagy, blocking mTORC1 signaling resulted in elevated puncta formation and GFP::LGG-1-PE levels in wild type animals ( Fig. 1C-F, Fig. S1A,B). Notably, a similar number of puncta is observed in sel-12 mutants with compromised mTORC1 signaling (Fig. 1C-F). These data indicate that sel-12 mutants have the capacity to carry out autophagy when mTORC1 signaling is disrupted. However, without mTORC1 inhibited, autophagy is blunted in sel-12 mutants, suggesting a role of activated mTORC1 in mediating sel-12 phenotypes. mTORC1 signaling is upregulated in sel-12 mutants.
Given that the mTORC1 pathway is a central inhibitor of autophagy and is a critical metabolic sensor (43,44) and we have observed increased mitochondrial metabolic activity due to elevated ER to mitochondria calcium signaling in sel-12 mutants (23), we asked whether mTORC1 signaling is elevated in sel-12 mutants. To assess mTORC1 activity, we examined phosphorylation levels of the central mTORC1 target RSKS-1/S6 kinase in sel-12 mutants. Strikingly, phosphorylated RSKS-1 was signi cantly increased in sel-12 animals compared to wild-type animals ( Fig. 2A,B), indicating that mTORC1 signaling is elevated in sel-12 mutants. Next, we sought to determine whether the increased mitochondrial activity observed in sel-12 mutants is leading to the elevation in mTORC1 signaling. Previously, we demonstrated that loss of sel-12 function promotes calcium uptake into the mitochondria from the ER, a process that increases mitochondrial activity and leads to the subsequent proteostatic collapse and neurodegeneration observed in sel-12 mutants (16,23). Indeed, reducing ER calcium release or mitochondrial calcium uptake in sel-12 mutants, as well as reducing oxidative phosphorylation, suppresses the proteostasis defects and neurodegeneration phenotypes observed in sel-12 mutants (16,23). To determine whether this altered calcium signaling pathway observed in sel-12 mutants is responsible for mTORC1 hyperactivation, we examined phospho-RSKS-1 levels in mcu-1; sel-12 double mutants and found that introduction of the mcu-1 null mutation leads to reduced phopsho-RSKS-1 levels in sel-12 mutants compared to sel-12 mutants alone (Fig. 2C,D). It is likely that mitochondrial calcium levels are responsible for changes to mTORC1 activation, as the mcu-1 mutation does not increase cytosolic calcium levels relative to the sel-12 mutation alone (Fig. S2). To determine whether mitochondrial hyperactivity due to elevated mitochondrial calcium promotes mTORC1 activation, we treated sel-12 worms with doxycycline to reduce mitochondrial respiration (23,45). Similar to blocking mitochondrial calcium uptake, sel-12 mutants treated with doxycycline abrogated the increase in phospho-RSKS-1 levels (Fig. 2E,F). Collectively, these data suggest that the altered ER-mitochondrial calcium signaling in sel-12 mutants causes aberrant activation of the mTORC1 pathway by increasing mitochondrial activity.
Reduction of mTORC1 signaling suppresses neuronal defects in sel-12 mutants.
We next asked whether mTORC1 activity contributes to the behavioral and neuronal defects seen with sel-12 loss. We rst examined the structure of the C. elegans mechanosensory neurons, which show age dependent structural decline and neurodegeneration (46)(47)(48). Previously, we found that these structural aberrations associated with aging develop precociously in sel-12 mutants (23). Day 1 adult sel-12 mutants display numerous ectopic neurite sprouts stemming off the ALM neuronal soma which are absent in wild type animals, and also show defects in the structure of the ALM and PLM axons, exhibiting abnormal lesions at a higher frequency relative to wild-type animals at day 1 ( Fig. 3A-C, Fig. S3A,C). To determine whether mTORC1 inhibition can suppress these neuronal morphological defects, we examined mechanosensory neuron structure in rsks-1; sel-12 and raga-1; sel-12 animals, as well as in aak-2(uthIs248); sel-12 animals, which carry a mutation that results in constitutive activation of the catalytic subunit of 5' adenosine monophosphate-activated protein kinase (AMPK/AAK-2), a global energy sensor and a major inhibitor of mTORC1 activity (49). raga-1; sel-12, aak-2(uthIs248); sel-12, and rsks-1; sel-12 animals each showed substantially improved neuronal morphology, with reduced ectopic neurite processes stemming off the soma (Fig. 3B) and reduced frequency of lesions (Fig. 3C), wave-like processes (Fig. S3B) and breaks (Fig. S3C) in the ALM and PLM axons. In addition, we treated the sel-12 animals with rapamycin, a clinical grade drug that speci cally inhibits mTORC1, and examined neuronal morphology. Consistent with the genetic manipulations, treatment with rapamycin showed similar improvements to the ALM soma (Fig. 3D) and axonal structure (Fig. 3E, Fig. S3D,E).
To determine whether the structural improvements we observed in the mechanosensory neurons translates to functional improvement, we examined the effects of genetic and pharmacological mTORC1 inhibition on soft touch behavior, which is controlled by the mechanosensory neurons. Wild type day 1 adult animals, when touched on the anterior or posterior half of the body with an eyebrow hair, will reverse their progression and move away from the stimulus. Consistent with increased morphological defects in the mechanosensory neurons of aged animals, mechanosensation declines with age (47). This reduced response rate happens prematurely in sel-12 mutants and continues to worsen with age (23). Indeed, day 1 adult sel-12 mutants show pronounced defects in soft touch response (Fig. 3F). Consistent with mechanosensory neuronal structural improvements, the raga-1; sel-12, aak-2(uthIs248); sel-12, and rsks-1; sel-12 double mutants all showed signi cant improvements to soft touch response (Fig. 3F). Moreover, rapamycin treatment recapitulated these improvements (Fig. 3G). Furthermore, this improvement is speci c to mTORC1, as inhibition of mTORC2 signaling in sel-12 mutants showed no improvement to sel-12 mutant neuronal defects. Indeed, genetic ablation of RICTOR, rict-1, which is required for the activation of mTORC2, in the sel-12 mutant background did not show signi cant improvements to mechanosensory neuron morphology or soft touch behavior (Fig. S4). These data indicate that neurodegeneration in sel-12 mutants can be suppressed with mTORC1 inhibition.
Additionally, we found that expression of raga-1 cDNA driven under the pan-neuronal rab-3 promoter (50) was su cient to fully prevent any improvements to soft touch behavior in raga-1; sel-12 animals (Fig.  3H), indicating a cell autonomous role of mTORC1 hyperactivity in the nervous system of sel-12 mutants. Additionally, unlike in wild type animals, pan-neuronal expression of raga-1 in the neurons of sel-12 animals signi cantly aggravated their touch defect. These data demonstrate a central role for neuronal mTORC1 activity in mediating and exacerbating the neurodegenerative behavioral defect in sel-12 mutants.
Inhibition of mTORC1 improves proteostasis in sel-12 mutants. mTORC1 activity may further explain impaired proteostasis in sel-12 animals. In fact, many studies show that modulation of mTORC1 activity widely impacts proteostasis (51). To de ne a role of mTORC1 in the collapse of proteostasis in sel-12 mutants (16), we rst examined animals with body wall expression of polyglutamine (polyQ) Q35::YFP fusion protein (rmIs132), which aggregates as the animals age (52).
While expression of Q35::YFP remains soluble and evenly distributed in day 3 adult wild type animals, adult sel-12 mutants show premature Q35 aggregation by day 3. However, analyses of raga-1; sel-12 and rsks-1; sel-12 animals show a signi cant reduction in polyQ aggregates at day 3 compared to sel-12 mutants, suggesting that mTORC1 inhibition improves proteostasis in these animals (Fig. 4A,B).
Consistent with these results, rapamycin treatment also reduced polyQ aggregates in day 3 adult sel-12 adult animals (Fig. 4C).
As an alternate method to assess the state of proteostasis in sel-12 mutants, we examined animals expressing human Aβ1-42 (dvIs100), which generates proteostatic stress and causes progressive paralysis in the transgenic animals (15). Previously, we found that that sel-12 mutants expressing Aβ1-42 have severely reduced motility relative to either mutant background alone, suggesting that the sel-12 mutation promotes Abeta1-42 toxicity and enhances proteostasis defects (16). To determine the effect of mTORC1 inhibition on motility in this background, we examined swimming behavior of raga-1; sel-12 and rsks-1; sel-12 mutants expressing Aβ1-42, and found they had signi cantly higher motility compared to sel-12 mutants expressing Aβ1-42 (Fig. 4D). Furthermore, treating sel-12 mutants expressing Aβ1-42 with rapamycin showed similar improvements in motility (Fig. 4E).
To evaluate the state of proteostasis of endogenous proteins, we subjected animals to heat stress (exposure to 37°C) to induce protein misfolding (53) and then examined animal survival. Previously, we found that sel-12 mutants have reduced resistance to heat stress and reduced survival (16). Thus, we examined the survival rate of wild type, sel-12 and raga-1; sel-12 mutants, as well as rapamycin treated sel-12 animals after 1, 2, 3, and 4 hours of exposure to 37°C. We found that the survival rate after heat stress at each time point was increased in the sel-12 mutants with mTORC1 signaling inhibited either genetically or pharmacologically relative to sel-12 mutants (Fig. 4F,G). Altogether, these data suggest that mTORC1 impacts proteostasis in sel-12 mutants and that the defects in proteostasis due to loss of SEL-12 are improved through mTORC1 inhibition.
Improvements to proteostasis and neuronal function through mTORC1 inhibition require the induction of autophagy.
To further investigate the mechanism by which mTORC1 contributes to proteostasis and neuronal defects in sel-12 mutants, we rst examined the condition of mitochondrial morphology in sel-12 mutants with mTORC1 signaling abrogated. Previously, we found that sel-12 mutants have severe defects in mitochondrial morphology and function due to elevated mitochondrial calcium (23,54). To determine whether mTORC1 activity contributes to mitochondrial disorganization in sel-12 mutants, we examined sel-12 and raga-1; sel-12 mutants expressing a mitochondrial localization signal tagged with GFP in the mechanosensory neurons (23,55,56). We found that compared to sel-12 mutants, raga-1; sel-12 double mutants had signi cantly improved mitochondrial organization, suggesting that elevated mTORC1 activity contributes to aberrant mitochondrial structure in sel-12 mutants (Fig. 5A,B).
Next, since we previously found that disorganized mitochondrial structure in sel-12 mutants is caused by elevated ER to mitochondrial calcium signaling (23, 54), we asked whether the improvements we observed in raga-1; sel-12 mutants are due to reduced ER-mitochondrial calcium signaling. Using a genetically encoded GCaMP6 calcium indicator localized to the mitochondrial matrix (23), we assessed mitochondrial calcium levels in the mechanosensory neurons. Notably, we found that calcium levels were unchanged after mTORC1 inhibition (Fig. 5C,D), indicating that mTORC1 activation is likely a downstream consequence of the altered calcium signaling observed in sel-12 mutants. This data is consistent with our pRSKS-1 western blot data showing that reduction of mitochondrial calcium uptake reduced pRSKS-1 levels in the sel-12 mutant background (Fig. 2C,D).
Thus, since mTORC1 is likely not playing a role in mitochondrial calcium signaling in the sel-12 mutants, it is possible mTORC1's inhibition of autophagy is responsible for the defects in proteostasis and neuronal function observed in sel-12 mutants. To determine whether the improvements in behavior and proteostasis through mTORC1 inhibition is primarily due to promoting autophagy, we knocked down inducers of autophagy lgg-1 and bec-1 using RNA interference (RNAi). Importantly, RNAi directed to lgg-1 or bec-1, which encodes the C. elegans beclin 1 ortholog, have been shown to inhibit autophagy (37,57). From these analyses, we found that the improvements to soft touch behavior in raga-1; sel-12 mutants is abrogated when treated with lgg-1 or bec-1 RNAi (Fig. 6A), and these animals resemble sel-12 mutant animals. Similarly, the improvements to proteostasis as measured by the number of Q35 aggregates in raga-1; sel-12 double mutants are lost with either lgg-1 or bec-1 RNAi treatment (Fig. 6B). Additionally, the increase in swimming rate is lost in raga-1; sel-12 animals expressing Aβ1-42 when treated with lgg-1 or bec-1 RNAi (Fig. 6C). These data suggest mTORC1 primarily impacts proteostasis and neuronal function in sel-12 animals by inhibiting autophagy. Altogether, our data identify activation of mTORC1 as a critical pathway by which SEL-12 loss results in neurodegeneration, and de ne an important role of the mTORC1 pathway in exacerbating the defects in proteostasis and autophagy following loss of SEL-12.

Discussion
AD is characterized by the pathophysiological buildup of protein aggregates such as amyloid plaques and neuro brillary tangles. The decline in the ubiquitin-proteasome and the autophagy-lysosomal proteostasis pathways that are responsible for degrading misfolded or damaged proteins has been postulated to underlie age related diseases such as AD, Huntington's disease, Lewy body dementia and Parkinson's disease (1)(2)(3). Here, we have shown that the premature decline in proteostasis in animals lacking SEL-12/presenilin function is due, at least in part, to decreased autophagy. Our ndings demonstrate that in C. elegans the loss of SEL-12/presenilin function leads to elevated mTORC1 signaling and this increase is due to the disruption of mitochondrial calcium homeostasis and mitochondrial hyperactivity. Furthermore, this increase in mTORC1 signaling inhibits autophagy ultimately resulting in the collapse of proteostasis and progression of neuronal dysfunction. Thus, our data indicate a crucial function of SEL-12/presenilin in mediating ER-mitochondrial calcium homeostasis that is vital for proteostasis and disruption of this role leads to neurodegeneration.
Presenilins are a family of highly conserved proteins that are commonly found on endomembranes, such as the ER and endosomes (4). The functional role of the presenilin family on these endomembrane structures is not well understood. The best studied function of presenilin is its role as the catalytic subunit of the gamma secretase, an intramembrane aspartyl protease. In addition to presenilin, the gamma secretase is composed of three other well-conserved components, including APH-1, APH-2/Nicastrin and PEN-2 (58,59). Gamma secretase cleaves several type I transmembrane proteins most notably Notch and APP (60). It is presenilin's role in the proteolytic cleavage of APP that has drawn the most intensive research efforts. This is because mutations in PSEN1 or PSEN2 or mutations or duplications in APP cause early onset familial AD. Due to the accumulation of Aβ peptides in patients with AD and the ability of the gamma-secretase to process APP to generate Aβ peptides, research on presenilin function has primarily focused on APP processing and the amyloid hypothesis, which centers on Aβ peptide accumulation causing AD. Despite intensive work on the amyloid hypothesis and a deep understanding into the processing of APP, the cause of AD is still not clear and a successful treatment for AD is not available. While smaller than the body of literature covering APP processing, there are many reports of gamma-secretase independent functions of presenilin in regulating several biological pathways that are disrupted in AD, such as calcium homeostasis, mitochondrial and lysosomal function and autophagy (5)(6)(7)(8)(9)61).
Presenilins found on the ER are enriched in membrane fractions that are associated with mitochondrial contact, the mitochondrial associated membranes (62). Accordingly, disruption of presenilin function results in more frequent ER mitochondrial contacts and signaling shared between these organelles, which leads to altered lipid synthesis and calcium uptake into the mitochondria (5,12,17,(20)(21)(22). This altered calcium uptake results in mitochondrial functional changes and elevated oxidative stress (18,23,54). In C. elegans, this oxidative stress leads to neuronal dysfunction, which can be suppressed by limiting ER calcium release, mitochondrial calcium uptake or mitochondrial respiration (23). Importantly, this ERmitochondrial activity of presenilin has been found to be gamma-secretase independent (16,23,61).
Moreover, in C. elegans, we have found that this altered ER-mitochondrial calcium signaling promotes the collapse of proteostasis in sel-12 mutants, which can be reversed by reducing ER calcium release or mitochondrial calcium uptake (16).
Here, we nd that the proteostasis defects associated with loss of SEL-12/presenilin are due to a reduction of autophagy caused by elevated mTORC1 activity. Both in vivo models of AD and post-mortem brain samples from AD patients have revealed hyperactivated mTORC1 signaling (63)(64)(65)(66). mTORC1 activity promotes cellular growth and metabolism by activating many anabolic pathways, such as the biosynthesis of protein and lipids, and inhibiting catabolic processes, such as autophagy. Despite reports indicating mTORC1 hyperactivation in AD, the cause of mTORC1 activation is not known. Several studies have suggested Aβ peptides directly or indirectly activate mTORC1 (64,67) and other studies have suggested that the activation of mTORC1 is an enhancer of Aβ generation and deposition (68,69). Interestingly, several laboratories have shown that hyperactive mTORC1 contributes to tau pathology (e.g. neuro brillary tangles), which in addition to amyloid plaques, is a classic hallmark of AD (68,70,71). In C. elegans sel-12 mutants, we nd that altered ER-mitochondrial calcium signaling and concomitant mitochondrial hyperactivity increases mTORC1 signaling, which subsequently leads to decreased autophagy. This in turn promotes proteostatic stress in sel-12 mutants driving neuronal dysfunction.
Since C. elegans does not generate Aβ peptides, the mechanism leading to mTORC1 activation in driving neurodegeneration in C. elegans is not mediated through Aβ peptides but through altered mitochondrial calcium homeostasis and mitochondrial hyperactivity. Indeed, we demonstrate that reduction of mitochondrial calcium or mitochondrial activity in sel-12 mutants reduces mTORC1 signaling. Importantly, we have shown that reduction of ER calcium release or mitochondrial calcium uptake rescues the proteostasis defects and neurodegeneration observed in sel-12 mutants (16,23). Calcium uptake into the mitochondrial matrix is critical for calcium buffering and in uencing mitochondrial metabolic activity (e.g. generation of NADH, ATP and superoxide). Alterations in mitochondrial calcium uptake can lead to oxidative stress or calcium overload triggering cell death (72)(73)(74). Interestingly, several recent studies have revealed a role of increased mitochondrial calcium levels in promoting neurodegenerative phenotypes (75)(76)(77)(78)(79). For example, in Drosophila and zebra sh, it was found that mutations in the genes encoding the PINK1 ortholog results in elevated mitochondrial calcium, and when mitochondrial calcium was reduced in these mutants neurodegeneration was prevented (80,81). Moreover, it was recently demonstrated in a mouse model of AD that neuronal mitochondrial have increased calcium levels that precedes neurodegeneration and that inhibition of mitochondrial calcium in ux prevents neurodegeneration (82).
Our data show that loss of SEL-12/presenilin function causes proteostatic collapse and neurodegeneration, at least in part, by the hyperactivation of mTORC1 and the ensuing inhibition of autophagy in C. elegans. Of note, hyperactivation of mTORC1 signaling is associated with several neurological disorders, such as autism, AD, and tuberous sclerosis, and suppression of mTORC1 using compounds that inhibit mTOR signaling has been shown to be an effective treatment in several clinical trials (83)(84)(85). Using C. elegans, we have found that treatment with rapamycin, a classic mTORC1 inhibitor, improves the proteostasis and neurodegeneration defects associated with sel-12 mutants.
Additionally, several studies have demonstrated the effectiveness of rapamycin on ameliorating neuropathology in mouse AD models overexpressing amyloid beta or hyperphosphorylated Tau (86)(87)(88)(89). Thus, these data suggest that treatment with mTOR inhibitors may provide some therapeutic bene t to patients suffering from AD. In summary, our study shows that loss of SEL-12/presenilin results in mTORC1 activation, caused by exacerbated mitochondrial calcium uptake and concomitant mitochondrial hyperactivity, which further contributes to loss of proteostasis and neurodegeneration in sel-12 mutants.

Conclusions
Collectively, our data reveal a novel connection between dysregulated calcium signaling and mTORC1 hyperactivation in presenilin/sel-12 mutants. Increased mitochondrial calcium uptake and concomitant mitochondrial hyperactivity accompanying presenilin/SEL-12 loss drives mTORC1 pathway activation, which subsequently leads to a decrease in autophagy. This, in turn, promotes proteostatic stress and neuronal dysfunction. Since C. elegans does not generate Aβ peptides, the mechanism leading to mTORC1 activation in driving neurodegeneration in C. elegans is not mediated through Aβ peptides but through altered mitochondrial calcium homeostasis and mitochondrial hyperactivity. Indeed, we demonstrate that reduction of mitochondrial calcium or mitochondrial activity in sel-12 mutants reduces mTORC1 signaling to wild type levels. Genetic and pharmacological inhibition of mTORC1 also greatly improves proteostasis and neuronal function in sel-12 mutants. Our data identify a crucial function of SEL-12/presenilin in mediating ER-mitochondrial calcium homeostasis that is vital for proteostasis and neuronal health.