Inhibition of the mitochondrial pyruvate carrier in astrocytes reduces amyloid and tau accumulation in the 3xTgAD mouse model of Alzheimer's disease

Alzheimer’s Disease (AD) is characterized by an accumulation of pathologic amyloid-beta (Aβ) and Tau proteins, neuroinflammation, metabolic changes and neuronal death. Reactive astrocytes participate in these pathophysiological processes by releasing pro-inflammatory molecules and recruiting the immune system, which further reinforces inflammation and contributes to neuronal death. Besides these neurotoxic effects, astrocytes can protect neurons by providing them with high amounts of lactate as energy fuel. Astrocytes rely on aerobic glycolysis to generate lactate by reducing pyruvate, the end product of glycolysis, through lactate dehydrogenase. Consequently, limited amounts of pyruvate enter astrocytic mitochondria through the Mitochondrial Pyruvate Carrier (MPC) to be oxidized. The MPC is a heterodimer composed of two subunits MPC1 and MPC2, the function of which in astrocytes has been poorly investigated. Here, we analyzed the role of the MPC in the pathogeny of AD, knowing that a reduction in overall glucose metabolism has been associated with a drop in cognitive performances and an accumulation of A  and Tau. We generated 3xTgAD mice in which MPC1 was knocked-out in astrocytes specifically and focused our study on the biochemical hallmarks of the disease, mainly A  and neurofibrillary tangle production. We show that inhibition of the MPC before the onset of the disease significantly reduces the quantity of A  and Tau aggregates in the brain of 3xTgAD mice, suggesting that acting on astrocytic glucose metabolism early on could hinder the progression of the disease.


Introduction
Glucose consumption in the brain provides the energy needed for neurons to function properly.
Astrocytes are one of the main types of brain cells and are, among others, interposed between the blood-brain barrier (BBB) and neurons, taking up glucose directly from capillaries by covering 99% of their surface with astrocytic feet.Thus, astrocytes are glucose's main consumers in the brain (Bonvento and Bolanos, 2021;Weber and Barros, 2015).Whereas in neurons, most pyruvate generated by glycolysis is transported into mitochondria to boost oxidative phosphorylation and ATP synthesis (Herrero-Mendez et al., 2009;Lopez-Fabuel et al., 2016), in astrocytes, ~60% pyruvate is converted into lactate by the lactate dehydrogenase, which restricts the amount of pyruvate oxidized within mitochondria (Iglesias et al., 2017).Furthermore, 33-50% of the pyruvate that enters astrocytic mitochondria is not used for ATP synthesis but rather anaplerosis to replenish TCA intermediates (Weber and Barros, 2015).In addition, the catabolism of mitochondrial pyruvate is mainly mediated by the pyruvate carboxylase that leads to gluconeogenesis and possibly lactate production (Rose et al., 2020).Moreover, astrocytes do not vitally depend on their mitochondrial energy metabolism (Rose et al., 2020;Supplie et al., 2017).Indeed, mice with a knockout (KO) for the mitochondrial cytochrome C oxidase in astrocytes did not show changes in microglial reactivity compared to controls, or neuronal or astrocyte death, nor astrocytic changes in terms of density or morphology (Supplie et al., 2017).Altogether these results indicate that astrocytes' metabolism mainly relies on aerobic glycolysis and is accompanied by high production of lactate.(Herrero-Mendez et al., 2009;Takahashi, 2021) According to the so-called astrocyte-neuron lactate shuttle model, astrocytes release lactate to provide neurons with fuel (Machler et al., 2016;Pellerin and Magistretti, 1994), even if this model is still contested (Dienel, 2019).Astrocytic lactate has been shown to play a fundamental role in memory mechanisms since blocking the astrocyte-neuron lactate transfer inhibits memory formation and synaptic plasticity (Descalzi et al., 2019;Suzuki et al., 2011).Lactate has also been shown to exhibit neuroprotective properties (Won et al., 2012) and high lactate levels are associated with overall better cognitive performances (Harris et al., 2016).In addition to lactate, it has been recently shown that, in drosophila, glycolysis-derived alanine from glia could be an alternative to lactate as a fuel for neurons during cognitive functions such as memory (Rabah et al., 2023).
Pyruvate, the end product of glycolysis, is imported into mitochondria through the mitochondrial pyruvate carrier (MPC) (Bricker et al., 2012;Herzig et al., 2012).This transporter is a heterodimer composed of two subunits MPC1 and MPC2 (Tavoulari et al., 2019).Genetic or pharmacological inhibition of the MPC in the J o u r n a l P r e -p r o o f Journal Pre-proof brain has been shown to affect diverse brain activities (Buchanan and Taylor, 2020;Zangari et al., 2020).For example, pharmacological inhibition of the MPC is neuroprotective and stimulates long-term potentiation and cognition (Mansour et al., 2021;Zhong et al., 2023), and increases lactate production (Iglesias et al., 2017;Zhong et al., 2015).Genetic inhibition of the MPC in glutamatergic neurons renders them hyperexcitable (De La Rossa et al., 2022).Recently, inhibition of the MPC in GFAP-expressing cells specifically was reported to stimulate neurogenesis in the dentate gyrus (Petrelli et al., 2023a).The effects of MPC inhibition thus seem to be cell dependent, showing that a better understanding of its functions is necessary.
In AD, a decreased cerebral glucose consumption has been shown to precede the onset of cognitive disorders.Indeed, reduced astrocyte lactate production and astrocytic glycolysis (i.e.hexokinase activity) have been observed in both familial and sporadic forms of the disease, making this inhibition a ubiquitous mechanism in AD (Bell et al., 2023;Le Douce et al., 2020;Zhang et al., 2022).In vitro addition of A in the culture medium reduces astrocytic glucose uptake (Tarczyluk et al., 2015), demonstrating the close relationship between A and glucose metabolism.Conversely, in a vicious circle, a reduction of glucose consumption induces an increase in phospho-Tau and A accumulation (Fu et al., 2015;Liu et al., 2004).Thus, in addition to playing a role as a marker of pathology, altered astrocyte glycolysis could play an active part in the progression of AD, and may therefore represent a therapeutic target.
Here, given the central role of pyruvate metabolism in neurons and astrocytes, we decided to test whether inhibition of the MPC in astrocytes would have an impact on the onset of AD's pathological markers and their accumulation over time using a murine model of AD, focusing mostly on the early pathology stages.To this end, a crossbreeding between 3xTgAD mice, a well-described AD model exhibiting both amyloid and phospho-Tau accumulation (Oddo et al., 2003b), and Mpc1 Flox;GFAP-Cre mice (Petrelli et al., 2023a) gave rise to the tamoxifen-inducible MPC1 knockout in GFAP-expressing cells in 3xTgAD mice (3xTgAD.MPC1 GFAP in the rest of the document).Amyloid and Tau were quantified in young and old mice to underline the potential of astrocytic MPC1 to regulate AD pathology.In addition, in young mice, glial cells, ATP production and doublestrand breaks were analyzed.

J o u r n a l P r e -p r o o f
Journal Pre-proof

Animals.
12-and 18-month-old female 3xTgAD (harboring the APP SWE , PS1 M146V and Tau P301L transgenes), 3xTgAD.MPC1 GFAP and WT mice were housed in a 12:12 light:dark cycle with water and food ad libitum.All models come from a C57BL/6 genetic background.All experimental procedures were approved by the Ethics Committee for Animal Experimentation of the Canton of Geneva, Switzerland.
Additionally, to visualize the recombined cells and their progeny, we crossed these inducible Mpc1 cKO mice with an inducible tdTomato lineage tracing line [tdTom (Madisen et al., 2010)], resulting in triple transgenic mice referred to as MPC1 cKO-tdTom mice.Control mice were generated in the same manner but carried WT alleles of the Mpc1 gene (MPC1 WT-tdTom mice).Cre expression was induced by administering 4hydroxytamoxifen at postnatal day 5 (P5) (Zehnder et al., 2021), leading to the deletion of Mpc1, which has previously been reported to lead to complete inhibition of the MPC (De La Rossa et al., 2022;Petrelli et al., 2023a) in GFAP-expressing cells from MPC1 cKO-tdTom mice, including astrocytes and radial glial cells (Petrelli et al., 2023a).tdTomato-GFAP expressing cells include astrocytes (all brain) and radial glial cells in the J o u r n a l P r e -p r o o f Journal Pre-proof SVZ and dentate gyrus.References (Hirrlinger et al., 2006;Petrelli et al., 2020;Petrelli et al., 2023b;Zehnder et al., 2021) provide detailed information on which cells are GFAP-expressing.

J o u r n a l P r e -p r o o f
Journal Pre-proof Following pentobarbital euthanasia, the brain was removed and the hippocampus was isolated and frozen at -80°C.A solution of triton (50mM Tris HCl, 150mM NaCl, 1%Triton X100, protease and phosphatase inhibitors 1x, pH=7.4) was added to the sample before sonication and centrifugation (20 000 g, 20 min, 4°C).The supernatants (containing Triton-solubilised proteins) were stored at -80°C until use.The pellets were dissolved in a solution of guanidine (5M guanidine, 50mM Tris HCl, protease, and phosphatase inhibitors 1x, pH=8) by gentle agitation (3h, 4°C).After centrifugation (20 000 g, 20 min, 4°C), the collected supernatants (containing Guanidine-solubilised proteins) were stored at -80°C until use.Protein concentrations were measured by BCA and kept for further western blots and ELISA quantifications.

ELISA tests.
Ab40, Ab42, sAPPa, sAPPb and phosphorylated Tau Thr231 (pT231) concentrations were determined using ELISA kit according to the manufacturer's instructions (Ab40, Ab42, pT231 ELISA kits from Thermofisher; sAPPa and sAPPb ELISA Kits from Mybiosource).The OD reading was performed at 450 nm and scaled on a standard curve, specific to each kit.The absence of amyloid and phospho-Tau in WT animals was previously validated (data not shown).

Western Blot.
Triton-solubilized proteins were denaturated at 70°C for 10 min in a 1x Laemmli buffer/2.5% β-mercaptoethanol buffer.20µg of proteins and a protein ladder (All blue precision plus ladder, Biorad) were loaded onto in Criterion™ TGX™ precast midi protein gel (Biorad) and migrated at 150 V for 55 min with the manufacturer's buffer (BioRad).After the transfer of proteins on a LF-PVDF membrane (7 min at 2.5A constant, and up to 25V in the manufacturer buffer using the Trans-blot Turbo machine, BioRad), the membrane was saturated in 5% non-fat dry milk/TBST (20 mM Tris, 150 mM NaCl, 0.1% Tween20, pH=7.4) for 45 min.The primary antibodies (ACTIN: 1/2500, Santa Cruz; ApoE: 1/500, Abcam; IDE: 1/250, Abcam) were sequentially incubated for 48 h at 4 °C in 5% milk/TBST.After three 10-min washes in TBST, the membrane was immerged in the appropriated Alexa Fluor-conjugated secondary antibody (Alexa-Fluor 488 or 555; Invitrogen) in 5% milk/TBST for 90 min.After three 10-min washes, the fluorescence was detected using the iBright imaging system (ThermoFisher Scientific).The same membrane was reused after a new saturation step.Densitometry analysis was performed using ImageJ and protein levels were normalized to ACTIN levels.Whole blots can be visualized in Supplemental figures 1 and 2.

J o u r n a l P r e -p r o o f
Journal Pre-proof ATP synthesis.
Following pentobarbital euthanasia, the brain was removed and the frontal cortex was isolated and directly processed for ATP measurements, using the CellTiter-Glo Luminescent Cell Viability Assay (Promega), as previously described (Gruz-Gibelli et al., 2016).The tissue was immersed in Neurobasal medium. 1 mg was passed in polytron for 10 seconds and this suspension was mixed with 100 µl of CellTiter-Glo substrate and buffer to induce ATP release for 10 min.Luminescence was measured with a luminometer (GloMax 20/20, Promega).Medium without tissue samples resulted in background luminescence.

Double-strand breaks.
Following pentobarbital euthanasia, the brain was removed, and the frontal cortex was isolated, mixed with 2mM EDTA, the pellet separated in PBS and processed for double-strand breaks measurements, using the Trevigen CometAssay kit (AMS Biotechnology) with modifications previously described (Authiat et al., 2022).
Briefly, cells were individualized, diluted in low-melting point agarose, spread on a comet slide and the DNA fragments were separated by electrophoresis.The DNA was stained with SYBR green I dye and at least 30 comets per genotype were photographed using an Olympus digital camera attached to an epifluorescent Zeiss Axioplan microscope (Axio vision rel 4.6) and the comet tail lengths were measured.

Amyloid Phagocytosis
1321N1 cells were seeded at 1.0x10 4 cells/well density in 96-well plate and incubated 4h at 37 °C, 5% CO 2 .The medium was then removed and replaced with fresh complete medium containing the MPC1 inhibitor (UK5099) at 10 μM and incubated 24h at 37 °C, 5% CO 2 .The amyloid-β solution (HiLyte Fluor 555, AS-60480-01, AnaSpec) was first diluted to 37.28 µM in 1,1% acetic acid in culture medium and then diluted to 0.25 µM in J o u r n a l P r e -p r o o f Journal Pre-proof either complete medium or complete medium containing the MPC1 inhibitor was applied for different period of time.At the end of the incubation period, cells were washed, detached from the 96-well plate with 100µl of trypsin, diluted with 500 µl of complete medium, centrifuged (500g, 2 min) for an additional wash and diluted in a final volume of 600 µl of complete medium and store on ice.2µl of Draq7 (Invitrogen, D15106) was added before analysis by flow cytometry.The Cytoflex LX cytometer (Beckman Coulter) was used to sort live (Draq7cells) and amyloid-positive cells.We checked for interferences between fluorochromes, but no compensation was necessary.Cells were sorted based on their forward and side scatter from all possible events.Single cells were then sorted based on Draq7 and Aβ-555.The median intensity was used as an index of Aβ-555 phagocytosis."

Statistics
AT8 + neurons and extracellular amyloid deposits were manually counted.Staining intensities were measured using ImageJ in manually defined regions of interest.The comet tail lengths were manually measured using ImageJ.All analyses were performed by experimenters blind to the genotype.One-way ANOVA was used to identify a genotype effect with, when significant, the LSD post hoc test.Two-way ANOVA with the Sidak's  Reduction in A deposits and Tau pathology in 12-and 18-month-old 3xTgAD.MPC1 GFAP mice.
At 12 months, 3xTgAD mice showed the onset of pathology with the appearance of extracellular amyloid plaques positive for Ab40-and 4G8-immunoreactivity restricted to the subiculum (Fig. 2A-H).At 18 months, the situation was amplified with a greater number of extracellular deposits.Conversely, 3xTgAD.MPC1 GFAP mice showed no accumulation of extracellular deposits, resulting in a significant difference between the two groups for 4G8 at 18 months (p<0.001) and for Ab40 at 12 (p<0.01)and 18 months (p<0.0001)(Fig. 2D-H).

J o u r n a l P r e -p r o o f
Journal Pre-proof Moreover, the number of AT8 + Tau neurons was higher in 3xTgAD mice than in 3xTgAD.MPC1 GFAP mice at 12 (p<0.01)and 18 (p<0.01)months.As the difference in accumulation of extracellular plaques was already present at 12 months of age, we then set out to determine the origin of this difference, in order to identify early alterations in the mice.Considering that the intracellular amyloid labeling was previously reported to appear before extracellular deposits (Oddo et al., 2003a;Oddo et al., 2006) and to be associated with cognitive deficits (Billings et al., 2005), intracellular 4G8 staining intensity was also quantified (Fig. 3).The intensity of the 4G8 staining was higher in 3xTgAD in the dorsal hippocampus (p<0.01) and the soma of the pyramidal cell layer (p<0.0001), as compared 3xTgAD.MPC1 GFAP mice, showing improvement as early as 12 months of age in 3xTgAD.MPC1 GFAP mice.

Reduction in A aggregates in 3xTgAD.MPC1 GFAP mice
To better characterize A and Tau changes between 3xTgAD and 3xTgAD.MPC1 GFAP mice, ELISA tests were performed.Because the monomeric and fibrils/aggregated forms of amyloid and Tau could not be distinguished from one another using immunohistochemistry, ELISA was used to measure Ab40, Ab42 and Tau pT231 in proteins solubilized by triton (Tx-, slightly aggregated forms) and guanidine (Gu-, highly aggregated forms) buffers in other cohorts of animals.There was no difference between the two groups for the Tx-soluble A (Fig. 4 A, B).In contrast, a reduction of Gu-Ab40 (Fig. 4 C, p<0.0001) and Gu-Ab42 (Fig. 4 D, p<0.001) levels was observed in 3xTgAD.MPC1 GFAP as compared to 3xTgAD mice.No differences were observed for Tx-Tau and Gu-Tau (Fig. 4 E,F).

J o u r n a l P r e -p r o o f
Journal Pre-proof

3xTgAD.MPC1 GFAP mice showed reduced levels of soluble APPb in the hippocampus.
To go further in the understanding of the underlying mechanisms of amyloid load reduction, we quantified markers of the non-amyloidogenic and amyloidogenic APP catabolic pathways and the soluble APPa and APPb forms respectively.In triton-solubilized proteins, no difference between groups was observed (Fig. 5 A-C) but in guanidine-solubilized proteins, a reduction in Gu-sAPPb (p<0.05) and a trend to an increase in the Gu-sAPPa/Gu-sAPPb ratio (p=0.057) in 3xTgAD.MPC1 GFAP mice were observed (Fig. 5 D-F).The insulindegrading enzyme and the ApoE levels did not show any difference between groups (Fig. 5 G,H, one-way ANOVA: F 2,17 =1.039; p>0.05 and F 2,17 =0.41; p>0.05, whole raw images of the blots are given in Supplemental Figures 1 and 2).

Inhibition of MPC increased A phagocytosis in the human 1321N1 cell cultures.
We raised the hypothesis that the decreased deposits of amyloid observed in MPC-deficient mice may be due to increased phagocytosis and clearance by astrocytes.To test this hypothesis, we used the human 1321N1 cells and tested their capacity to internalize fluorescent amyloid in the presence or absence of 10 mM UK5099, an inhibitor of MPC.A time-course experiment of the amyloid peptide internalization was performed.As shown in

Discussion
Our study is the first to investigate the effect of astrocyte-specific MPC inhibition on pathological markers of Alzheimer's disease in the 3xTgAD murine model, which combines amyloid and Tau pathologies.
We show that MPC1 knockout induces a large decrease in the number of Tau-positive neurons and amyloid plaque density, at both early and late stages of the pathology.In addition, MPC inhibition reduces both astrocytic cell reactivity and drop in overall ATP levels.Our data (i) show that blocking pyruvate transport into mitochondria could represent an interesting therapeutic approach to decrease the formation of neurofibrillary tangles and amyloid deposits, (ii) further demonstrate the early involvement of astrocytes in the appearance of markers of pathology, and (iii) highlight the possibility that early intervention on astrocytes could modify the long-term progress of the pathology.These data are summarized in the figure 9.In physiological conditions, astrocytes uptake glucose from the blood, convert it to pyruvate, some of it is internalized by the mitochondria through the MPC dimer but most of it is converted to lactate for neuron consumption.In 3xTgAD model, there is a reduced glucose uptake by reactive astrocytes, which leads to a decrease in lactate release.In the 3xTgAD.MPC1 GFAP mice model, the pyruvate cannot be internalized by the mitochondria and we observed a reduction in amyloid and Tau deposition as well as an increase in amyloid phagocytosis by astrocytes.
It has been widely demonstrated in AD that glycolysis is downregulated and that this change precedes cognitive dysfunction by decades.A drop in brain glucose uptake was reported to be associated with cognitive decline and increased astrocyte reactivity (Nam et al., 2021).The hippocampus of 3xTgAD mice also displays this hypometabolism (Le Douce et al., 2020).This reduced glucose consumption, which partially reflects a reduction in glucose uptake by astrocytes, could therefore lead to a reduction in lactate production by astrocytes, as previously shown in different AD models (Le Douce et al., 2020;Lu et al., 2019;Zhang et al., 2018).The combination of a deficit in astrocytic glucose uptake and neuronal lactate transporter density (Zhang et al., 2018), could lead to an energetic crisis and cognitive decline due to less lactate being supplied to neurons.This glucose hypometabolism could be responsible for decreased ATP levels in the brain (Rossi et al., 2020) as we observed in our study, a drop in cognitive performances (El Hayek et al., 2019) and an upregulation of the main pathological markers in AD (Fu et al., 2015;Liu et al., 2004).Zhang et al. (2022) proposed that a decrease in lactate levels in the hippocampus of AD leads to an increase in BACE1 activity and therefore an increase in the accumulation of Aβ, as well as an increase in tau hyperphosphorylation (Ghosh and Osswald, 2014;Liu et al., 2004;Xu et al., 2016;Zhang et al., 2022).In turn, Aβ, in a vicious circle, induces a decrease in glucose uptake, notably by reducing the availability of GLUT1, the plasma membrane transporter of glucose, which amplifies the decrease in glycolysis (Hendrix et al., 2021;Tarczyluk et al., 2015), and this effect is reduced by the stimulation of glucose uptake by glucagon (Zheng et al., 2021).Blocking glycolysis amplifies the effects of A on human astrocytes in terms of GFAP reactivity, A accumulation and apoptosis (Fu et al., 2015).It is likely that astrocyte death (linked to A accumulation) (Saha and Biswas, 2015) contributes to the amplification of glucose hypometabolism in later stages of the disease.
Insulin resistance in AD is a well-known phenomenon.Insulin directly interacts with A peptide, regulating clearance through lipid metabolism (Arnold et al., 2018) but also indirectly as insulin resistance in J o u r n a l P r e -p r o o f Journal Pre-proof peripheral tissue precedes A accumulation (Ekblad et al., 2018).Furthermore, peripheral and central insulin resistance can be induced by accumulation of A (Clarke et al., 2015).As insulin resistance impedes glucose uptake in individual cells, it reinforces neuronal as well as glial dysfunctions.Recently the insulin sensitizers thiazolidinediones (TZD), primarily developed to treat diabetes, have been tested in AD patients and reported to reestablish glucose regulation but also reduce inflammation and A peptide accumulation as well as increase cerebral blood flow (Sato et al., 2011).Pioglitazone, which belongs to the TZD class of compounds, has been shown to reduce the amyloid load in an ApoE and/or IDE-dependent process, neurofibrillary tangle formation and decline in cognitive function in 3xTgAD mice (Mandrekar-Colucci et al., 2012;Mansour et al., 2021;Searcy et al., 2012;Zhong et al., 2023).TZDs and their derivatives are mostly known as agonists of peroxisome proliferator-activated gamma-type (PPAR-γ) receptors.Interestingly, some of these compounds have been discovered to be mild inhibitors of the MPC (Tavoulari et al., 2022).How these compounds could exert beneficial effects in models of AD remains elusive given that their ability to cross the BBB is limited (Zhang et al., 2019).Nevertheless, these studies with TZDs confirm an important role of glucose and pyruvate metabolism in the pathophysiology of AD and raise the question of the role of MPC in the first stages of this disease.Here we found that MPC inhibition in astrocytes specifically has an impact on the biomarkers of AD in the mouse, thereby confirming an important role of astrocytes in the apparition of pathological markers of AD and pointing to MPC as a potentially relevant therapeutic target.Similar effects have been shown in Parkinson's disease, with the use of TZDs to modulate energy production from the use of pyruvate as a substrate for the TCA cycle, lowering production of ROS and generally showing neuroprotective effects (Quansah et al., 2018).
In line with our study, a deficit in L-serine, produced by glycolysis, has been shown to be aggravating and present very early in the disease (i.e.before the glial cells become reactive) (Le Douce et al., 2020).Inhibition of the 18kDa translocator protein (TSPO), another mitochondrial protein that plays a role in controlling astrocyte glucose consumption in inflammatory conditions (Tournier et al., 2023), also improves the neuropathological symptoms of AD and reduces astrocyte reactivity (Ceyzériat et al., 2022).
Our results at 12 months of age show a difference in glial activity specific to astrocytes, backed up by studies who suggested that astrocytes undergo changes in their activity in AD before microglia.In fact, based on the kinetics of appearance of reactivity markers, previous studies concluded that astrocyte alterations would precede microglia reactivity (Chun et al., 2018;Gordon et al., 2002;Tournier et al., 2020).In addition to being J o u r n a l P r e -p r o o f Journal Pre-proof early markers of AD, pathological alterations in astrocytic functions appear to play a role in the development of AD.
Astrocytes have a dual role in the development of AD.On the one hand, their initial reactivity is the physiological process of inflammation, allowing them to clear more efficiently the pathological Aβ and Tau aggregates.It has been demonstrated that GFAP knockout results in an increase in Aβ load, thereby implicating the filament protein that is required for astrocyte activation in the response to AD pathology (Kraft et al., 2013).
However, this result was not replicated in a more recent study using the same models but a different genetic background (Kamphuis et al., 2015).Another study demonstrated that pharmacological inhibition of astrocytes in organotypic brain culture slices resulted in an increase in Aβ (Davis et al., 2021), thereby confirming the hypothesis of a correlation between astrocyte activation and Aβ degradation.On the other hand, blocking astrocyte reactivity has demonstrated positive effects on pathology (Ceyzeriat et al., 2018;Reichenbach et al., 2019).Our results indicate that mice with an MPC inhibition express less GFAP but also less Aβ, sustaining the idea of a functional link between astrocyte activation, MPC1 activity and Aβ degradation.In a mice model of Creutzfeldt-Jakob disease, an increase in astrocyte activation (Gfap mRNA) is concomitant with an increase in Mpc1 mRNA levels (Andres-Benito et al., 2021) that may suggest that astrocyte reactivity and MPC1 function are co-regulated.Importantly, we also demonstrated that the inhibition of MPC1 in human astrocytes clearly improves their phagocytic activity, demonstrating a functional impact of MPC1 activity on A degradation.
Astrocytes can change morphologies and reactivity states and in turn influence those of surrounding cells such as microglia, and reciprocal and multiple interactions lead to the neuropathological pattern of AD (Chun et al., 2018).Reciprocal stimulation of glial cells leads to an inflammatory reaction, with the release of cytokines such as Tumor necrosis factor a (TNFa).Interestingly, TNFa induces a decrease in lactate release by astrocytes which is abrogated by MPC inhibition (Chao et al., 2019), suggesting that both A and the neuroinflammation reaction (linked to A) induced a decrease in the lactate production by astrocytes.Moreover, as only a slight decrease of sAPPβ in the Gua-soluble fraction was observed without a shift in favor of the non-amyloidogenic pathway, an increase of amyloid degradation seems to be the main mechanism involved in the reduction of highly aggregated forms of Aβ observed in our model.In agreement with this hypothesis, we demonstrated that the inhibition of MPC1 in human astrocytes clearly improves their phagocytic activity.
Astrocyte participation to tau degradation, and as a result, tau propagation, is now well described (Fleeman and Proctor, 2021).In our study, we demonstrated that MPC inhibition led to a total absence of AT8 + J o u r n a l P r e -p r o o f Journal Pre-proof neurons.Surprisingly, no such effect was measured on pT231 tau levels measured by ELISA.We hypothesize that either MPC1 KO impacts only some phosphorylation pathways or it affects the accumulation process in neurons.In the globular glial tauopathy linked to MAPT P301T mutation, MPC1 is increased (Ferrer et al., 2020) that may suggest a link between Tau and MPC1 activity.Further studies are needed to investigate those mechanisms.
Thus, the use of very early therapies could cut this chain of events and have positive effects on the progression of AD.Among the non-drug therapies aimed at improving cognitive performance, numerous studies have reported a beneficial effect of physical exercise.In the search for anatomo-functional correlates of this stimulation, lactate has been proposed to be the key molecule mediating this improving effect, even if this idea is not fully accepted (Ballester-Ferrer et al., 2022;Cho et al., 2020;Chycki et al., 2021;El Hayek et al., 2019).In AD, it is recognized that physical exercise has beneficial effects on the progression of the disease (Cass, 2017;Jia et al., 2019;Mrakic-Sposta et al., 2018;Revilla et al., 2014;Shen and Li, 2016).We can therefore hypothesize that stimulating physical activity stimulates lactate production, which in turn acts as a neuroprotective molecule and stimulates cognitive abilities.Unfortunately, it was not possible to directly measure lactate in our studies and therefore further investigations will be necessary to confirm the direct role of lactate in the MPC-inhibition ameliorative effects.It should also be mentioned that although inhibition of the MPC improved the biochemical markers of AD, the increase of double strand breaks in 3xTgAD mice was not reversed by inhibition of astrocytic MPC1, indicating that certain facets of the pathology were not improved.In addition, the lack of cognitive and neurodegenerative measures in our study could limit the clinical interpretations of our findings and additional experiments should address this point.However, intracellular A has previously been associated with cognitive decline (Billings et al., 2005), which may indicate a better cognitive maintenance in mice lacking astrocytic MPC activity.Finally, it could be interesting to study the effect of MPC depletion at later stages, once the pathological hallmarks are in place, to see if the functional changes induced in astrocytes are sufficient to reverse the disease.
In conclusion, although further studies are needed to better understand the molecular mechanisms underlying the decrease in amyloid and Tau induced by the inhibition of astrocytic MPC, our data pave the way for targeted research into astrocyte metabolism as a tool for limiting AD progression.Finally, the results of our study suggest that inhibition of astrocytic MPC may be a therapeutic target in AD and could also be relevant to other degenerative neuropathologies where hypometabolism is observed.
multiple comparison test was used to analyze A phagocytosis.Unpaired two-tailed Student's t-tests were used to compare two groups.Data are presented as individual values and mean ± SD.J o u r n a l P r e -p r o o f Journal Pre-proof Results Validation of the absence of MPC1 in astrocytes in the MPC1 GFAP knock-out mice.The absence of MPC1 in the MPC1 GFAP knock-out mice was validated by immunofluorescence in the cortex of adult mice, 2 months after tamoxifen induction of MPC1 deletion.As shown in the figure 1A-B, MPC1 is no longer expressed in astrocytes in the KO model as compared to WT, whereas it is clearly visible in neurons.

Figure 1 .
Figure 1.MPC1 GFAP knock-out mice show reduced MPC1 in astrocytes.A, Overview confocal images of TAM-induced tdTom recombination (tdTom, red) and MPC1 (green) immunostaining in the cortex of Control-tdTom and MPC1cKO-tdTom.Scale bars 20 μm.B, High magnification confocal images of TAM-induced tdTom recombination (tdTom, red) and MPC1 (green) immunostaining in the cortex of Control-tdTom and MPC1cKO-tdTom.Scale bars, 10 μm.In A and B, arrows indicate the co-staining for Tomato and MPC1 in control mice, whereas astrocytes in the KO are only Tomato positive.

Figure 5 .
Figure 5. Reduction of soluble APP fragment in 12-month-old 3xTgAD.MPC1 GFAP mice.The byproducts of the APP catabolism, APPa and APPb and the APPa/APPb ratio were quantified in triton-(Tx-, A-C) and guanidine-(Gu-, D-F) solubilized proteins.G-H, Representative illustration and western blot quantification of insulin-degrading enzyme (IDE) and ApoE in triton-solubilized proteins.Unpaired two-tailed Student's t-tests at *p<0.05 (A-F) and one-way ANOVA (G-H).

Figure 8 .
Figure 8.The MPC1 antagonist UK5099 increases the phagocytosis capacity of the human 1321N1 astrocyte cells.A time-course of the median fluorescence intensity (MIF) of intracellular A addition of UK5099 (10µM) in the cell culture media was performed.(A) Representative images of the alive cell population depending of size (forward scatter, FSC) and granulometry (side scatter, SSC).(B-C) Representative images of Draq7 and A-555

Figure 9 .
Figure 9. Summary diagram of the observed effects of MPC inhibition on glucose, amyloid and astrocytes functions.