Necroptosis inhibition counteracts neurodegeneration, memory decline, and key hallmarks of aging, promoting brain rejuvenation

Abstract Age is the main risk factor for the development of neurodegenerative diseases. In the aged brain, axonal degeneration is an early pathological event, preceding neuronal dysfunction, and cognitive disabilities in humans, primates, rodents, and invertebrates. Necroptosis mediates degeneration of injured axons, but whether necroptosis triggers neurodegeneration and cognitive impairment along aging is unknown. Here, we show that the loss of the necroptotic effector Mlkl was sufficient to delay age‐associated axonal degeneration and neuroinflammation, protecting against decreased synaptic transmission and memory decline in aged mice. Moreover, short‐term pharmacologic inhibition of necroptosis targeting RIPK3 in aged mice, reverted structural and functional hippocampal impairment, both at the electrophysiological and behavioral level. Finally, a quantitative proteomic analysis revealed that necroptosis inhibition leads to an overall improvement of the aged hippocampal proteome, including a subclass of molecular biofunctions associated with brain rejuvenation, such as long‐term potentiation and synaptic plasticity. Our results demonstrate that necroptosis contributes to age‐dependent brain degeneration, disturbing hippocampal neuronal connectivity, and cognitive function. Therefore, necroptosis inhibition constitutes a potential geroprotective strategy to treat age‐related disabilities associated with memory impairment and cognitive decline.


Quantification of GSK'872 in brain and plasma
Brain and plasma samples were obtained from aged mice treated for 1h with GSK'872 (10mg/kg i.p.). The bioanalysis of plasma and brain samples was conducted by LC-MS/MS with a QTRAP 4500 triple quadrupole mass spectrometer (Applied Biosystems SCIEX) in the negative ion mode and interfaced with an Ekspert ultraLC 100-XL UHPLC System (Eksigent). Calibration standards (0.003 to 10 μM) and quality controls (0.02, 0.2 and 2.0 μM) were prepared from naïve mouse plasma in parallel with mouse plasma study samples (60 μL) by precipitation with three volumes of ice-cold acetonitrile containing 20 μM of theophylline. The precipitated samples were centrifuged at 6,100 g for 30 min at 4°C. Following centrifugation, an aliquot of each supernatant was transferred to an autosampler vial and diluted with two volumes of aqueous mobile phase (0.2% formic acid in water). Samples were injected onto a reverse phase analytical column (YMC Triart C18; 2.0 x 50 mm; 1.9 μm; YMC CO) and eluted with a gradient of 0.2% formic acid in Acetonitrile. GSK'872 was monitored by a multiple reaction monitoring (MRM) experiment using an Analyst software (v1.6.2, Applied Biosystems SCIEX). Quantitation was conducted using a MultiQuant software (v2.1, Applied Biosystems SCIEX) and the resulting calibration curve was fitted with a linear regression and 1/x weighting. The lower limit of quantitation (LLOQ) was 0.010 μM.

Tunel Assay
Apoptosis was evaluated by using the Click-iT TM TUNEL Alexa Fluor-488 Imaging Assays kit (Invitrogen, Thermo Fisher Scientific, USA) in accordance with the manufacturer's instructions. DNaseI was used as a positive control for DNA fragmentation.

Histological Analysis
Free-floating sections were processed for immunohistochemistry as previously described (Oñate et al. 2020).
Finally, sections were co-stained with Nissl (cresyl-violet staining) to detect nuclei and with Eriochrome-C staining for myelinated-axons detection, and finally mounted on glass slides with Entellan medium (Merck).

Fluoro-Jade C staining
Brain tissue was mounted on positively charged slides and rehydrated on decreasing concentrations of ethanol.
After rehydration, the tissue was pre-treated for 10 min in potassium permanganate and then incubated for 10 min in the dark at 25 °C in Fluoro-Jade C and DAPI (Biosensis, TR-100-FJ). The tissue was then washed with water and let dry overnight. The next day the slides were cleared in xylene and coverslipped with DPX-new mounting solution (Merck Millipore).

Electrophysiology
Electrophysiological recordings were performed as described before (Carvajal et al. 2018). Briefly, transverse slices (400 μm) from the dorsal hippocampus were cut under cold artificial cerebrospinal fluid (ACSF, in mM: 124 NaCl, 2.6 NaHCO3, 10 D-glucose, 2.69 KCl, 1.25 KH2PO4, 2.5 CaCl2, 1.3 MgSO4, and 2.60 NaHPO4) using a Vibratome (BSK microslicer DTK-1500E, Ted Pella, Redding, CA, USA) and incubated in ACSF for 1 hour at room temperature. In all experiments, 10 μM PTX was added to suppress inhibitory GABAA transmission. Slices were transferred to an experimental chamber (2 ml), superfused (3 ml/min, at room temperature) with gassed ACSF (using 95% O2/5% CO2) and visualized by trans-illumination with a binocular microscope (Amscope, Irvine, CA, USA). To evoke field excitatory post synaptic potentials (fEPSPs), Schaffer collaterals were stimulated with bipolar concentric electrodes (Tungsten, 125 μm OD diameter, Microprobes) connected to an isolation unit (Isoflex, AMPI, Jerusalem, Israel). The stimulation was performed in the stratum radiatum within 100-200 μm from the recording site. Recordings were filtered at 2.0-3.0 kHz, sampled at 4.0 kHz using an A/D converter (National Instrument, Austin, TX, USA), and stored with the WinLTP program. The basal excitatory synaptic transmission was measured using an input/output curve protocol with 10 s of interval between stimuli. Data were collected and analyzed offline with pClamp 10 software (Molecular Devices, San Jose, CA, USA). To generate LTP, we used high-frequency stimulation (HFS) protocol, which consisted of 3 trains at 100 Hz of stimuli with an inter-train interval of 10 s. Data were collected and analyzed offline with pClamp 10 software (Molecular Devices, San Jose, CA, USA).

Golgi-Cox staining
Golgi-Cox impregnation method was used to analyze dendritic spine density in hippocampal slices by using the FD Rapid GolgiStain™ Kit, following manufacturer instructions (FD Neurotechnologies Inc, MD, USA). See detailed protocol in Supporting Information. Briefly, fixed brains by PFA perfusion were immersed in impregnation solution (Solution A/ B), and store at room temperature for 2 weeks in the dark. After 72 h in precipitation Solution C, brains were quickly frozen with dry ice and immediately sectioned with a cryostat at a thickness of 120 µm. Coronal brain sections were mounted on gelatin-coated slides, stained with solution D/E, dehydrated with sequential rinses of 50%, 75%, 95% and 100% ethanol and finally mounted with Entellan medium (Merck).
Dendritic spines were imaged as Z-stacks images with a Nikon Eclipse E200 microscope by using a 100X objective with immersion oil. Image analyses were performed with the Image J software by Z-projection of the stacks (sum stacks) followed by the measurement of dendrite length with the segmented line tool. Dendritic spines were manually counted in the defined dendrite length and plotted as spine number normalized to 10 µm of dendrite. Between 8 and 10 dendrites were imaged and quantified per mouse, considering n=3 mice per experimental group.

Luminex Assay
Cytokines levels were analyzed by Luminex Mouse Discovery Assay (R&D Systems, MN, USA) using a selfdesigned panel of 12 selected cytokines (plate code: LXSAMSM-12), based on color-coded beads, pre-coated with analyte-specific capture antibodies that permits simultaneous analysis of the analytes. Table S1 detailed cytokines and chemokines of the panel, bead region, sensitivity, and the main functions. Analysis and detection were performed in a Dual-laser flow-based detection instrument, Luminex 200 analyzer by Proyecto Luminex, Programa de Virología, Redeca, ICBM, Facultad de Medicina, Universidad de Chile.

S-Trap processing of samples
Samples were processed using S-trap mini protocol (Protifi) (for 310 ug and 110 ug samples) and S-trap micro protocol (for low conc samples) as recommended by the manufacturer with little modification. After, application of the samples on the S-trap mini spin column, trapped proteins were washed 5 times with S-TRAP binding buffer.
A double digestion with trypsin (1:40) was carried out first overnight at 37 o C in TEAB at a final concentration of 50 mM, and then for another 4 hrs (1:40) in 50mM TEAB. Elution of peptides from S-trap mini spin column was achieved by centrifugation at 1000 x g for 1 min by adding 50 mM TEAB, then 0.2% aqueous formic acid and finally 50% acetonitrile/0.2% formic acid. Resulting tryptic peptides were pooled, dried, and quantified using Pierce Quantitative fluorometric Peptide Assay (Thermo Scientific).

LC-MS methods
1.5 µg peptide was analysed per sample. Samples were injected onto a nanoscale C18 reverse-phase chromatography system (UltiMate 3000 RSLC nano, Thermo Scientific) then electrosprayed into an Q Exactive Plus Mass Spectrometer (Thermo Scientific). For liquid chromatography buffers were as follows: buffer A (0.1% formic acid in Milli-Q water (v/v)) and buffer B (80% acetonitrile and 0.1% formic acid in Milli-Q water (v/v). Sample were loaded at 10 μL/min onto a trap column (100 μm × 2 cm, PepMap nanoViper C18 column, 5 μm, 100 Å, Thermo Scientific) equilibrated in 0.1% trifluoroacetic acid (TFA). The trap column was washed for 5 min at the same flow rate with 0.1% TFA then switched in-line with a μPAC C18 nano-LC column (200 cm, inter-pillar distance-2.5 μm, pore size-100-200 Å, PharmaFluidics). The peptides were eluted from the column at a constant flow rate of 300 nl/min with a linear gradient from 3.8% buffer B to 12.5% buffer B in 22 mins, then from 12.5% buffer B to 41.3% buffer B in 95 mins, then from 41.3% buffer B to 61.5% in 23 mins and finally to 100% buffer B in 10 mins. The column was then washed with 100% buffer B for 10 min and re-equilibrated in 1% buffer B for 38 mins. Two blanks were run between each sample to reduce carry-over. The column was kept at a constant temperature of 50 o C. The data was acquired using a uPAC-compatible easy spray emitter source operated in positive mode with spray voltage at 2.2 kV, and the ion transfer tube temperature at 275 o C. The MS was operated in DIA mode. A scan cycle comprised a full MS scan (m/z range from 345-1155), with RF lens at 60%, AGC target 3E6, orbitrap resolution 70,000, maximum injection time at 200 ms and source fragmentation disabled. The MS survey scan was followed by MS/MS DIA scan events using the following parameters: collision energy mode set to linear with a normalized HCD collision energy set to 25, orbitrap resolution 17500, first fixed mass 200 m/z, AGC target 3E6, maximum injection time 55 ms, isolation windows were variable from 5-66 m/z. The inclusion list (DIA windows) and windows widths are shown in Table S2. Data for both MS and MS/MS scans were acquired in profile mode. Mass accuracy was checked before the start of samples analysis.

Data filtering and generation of expression ratios
Raw data files from single-shot label-free experiments were converted into Microsoft Excel workbooks and utilised to generate ratios of protein expression within each animal relative to mean expression of n=4 control animals within each comparison (ie. adult: adult (expression ratio=1), aged: adult, Mlkl-KO: aged-WT; aged GSK'872: aged vehicle). Proteins identified by fewer than 2 unique peptides were excluded from subsequent analyses in order to ensure maximum identification confidence ( Figure S9). Average abundances were calculated for each protein within each condition. As the proteomic analysis is confirmatory of the morphological, functional and behavioral assessments carried out in this study, to retain a as high a coverage of the proteome as possible, averages within group could include proteins with an n of 1 or more absent abundance values. Statistics for grouping consistency have been provided (see Figure S15b). Raw data are available online for further analysis.
Following this, relative expression ratios per study (ie. aged vs. adult, Mlkl-KO vs. aged WT, and aged GSK'872 vs. aged vehicle) were used for subsequent expression profile clustering analyses. UniProt Accession numbers of proteins identified by 2 or more unique peptides with accompanying expression ratios generated as described above were subjected to expression profile clustering in BioLayout Express 3D . BioLayout utilises a userdetermined Pearson correlation and the Markov Clustering Algorithm to cluster input data based on userdetermined parameter(s) (Enright 2002;Theocharidis et al. 2009). Pearson correlation was set to 0.97 to cluster datasets into distinct subsets based on similarity in expression profile. Discrete clusters exhibiting biologically relevant expression profiles-ie. opposing directionality between aged versus both MLKL-KO and GSK'872 expression ratios (Fig S10) were identified and exported as .txt files containing and identifier column and expression ratios, for subsequent analyses in IPA. See Supporting Information for detailed procedure.

Ingenuity Pathway Analysis (IPA)
The Ingenuity Pathway Analysis (IPA) application (Ingenuity Systems, Silicon Valley, CA) was used to visualise and explore the cellular and molecular pathways that may have been altered as result of genetic (Mlkl-KO) or pharmacological (GSK'872) inhibition of necroptosis. Without user-directed manipulation, IPA's statistical predictions and annotations are approximately 90% based off on peer-reviewed publications; the remaining 10% of stored interactions have been identified by other in silico techniques. The analyses were performed only using experimentally reported interactions published in peer-reviewed publications stored within the "hand-curated" and continually updated Ingenuity Knowledge database (Ingenuity Systems, Silicon Valley, CA). For more information on the computational methodology underpinning IPA, please refer to http://www.ingenuity.com/.
Prior to all analyses within IPA, input datasets comprising, as described above, mean expression ratios of n=4 animals per experimental group, were converted to fold-change values, and a ±20% cut-off in expression change respective to control was applied within each respective study. Individual analyses of aged vs. adult, Mlkl-KO vs.
WT, and GSK'872 vs. vehicle were performed prior to a comparative analysis in order to gain insight into potential biological networks distinguishing "normal" versus "necroptosis-inhibited" aging processes.
For canonical pathway analysis, p-values of canonical pathway scores and subsequent ranking for all analyses performed in this study were derived from a Fisher's Exact Test calculating overlap between molecules in each respective input dataset and number of molecules comprising canonical pathway as defined by the Ingenuity Systems Database. Predicted activation z-scores were calculated by weighing the predicted expression change of target molecules as defined by Ingenuity Knowledge Database against the actual expression change of target molecules reported in input dataset. An activation z-score >2 or <−2 is considered statistically significant (Ingenuity Systems, Silicon Valley, CA). Constituent molecules within pathway were colourized with intensity of colour corresponding to magnitude of change.
For diseases and functions analysis, predicted activation z-scores of associated downstream diseases and functions were calculated by weighing the predicted expression change of target molecules associated with specific "diseases or functions" annotation as defined by Ingenuity Knowledge Database against the actual expression change of target molecules reported in input datasets. An activation z-score >2 or <−2 is considered statistically significant. P-values of overlap is derived from a Fisher's exact test were derived from a Fisher's Exact Test calculating overlap between molecules in each respective input dataset and number of molecules comprising the known interactome of each regulator as defined by Ingenuity Systems Database. In graphical format, target molecules present within each proteomic dataset predicted to be activated or inhibited to mediate the associated "diseases or functions" annotation were visualised in relation to their associated predicted regulator and were colourised with intensity of colour corresponding to magnitude of change. were measured in the hilus of the DG at different ages, bar: 25 µm. Non-significant differences were observed between groups. One-way ANOVA and Tukey's test for multiple comparison, p=0.519 for adult vs aged, n=5-6 mice. (b) Representative image of cleaved-caspase-3 (green) and pRIPK3 (red) co-staining in the DG of aged mice. Bar, 25 µm. Magnification (x4) shows pRIPK3 positive cells co-labeling with activated caspase-3. (c) Fluorescent Tunel assay performed in mouse brain along aging. Images and quantification show an upward trend of Tunel staining (green) along aging, without significant differences (p= 0.345 for adult vs aged, n=5-6).

SUPPLEMENTARY FIGURES
pRIPK3/Tunel co-staining and magnification show increasing levels of pRIPK3 (red) during aging, which overlap with Tunel in the DG of aged mice. One of the values from the adult group was identified as an outlier and was excluded from the analysis. Oultiers were identified with the "Identify Outliers" tool from GraphPad Prism v8.0.2.  5 µm 2 and 0<0.3 circularity). Calibration bar, 10 µm. Degeneration Index (DI) was calculated as the ratio between the area of fragmented axons over the total axonal area. Immunohistochemistry against pMLKL (brown) and Eriochrome-C staining (myelinated axons, blue) in the DG and CA3-CA1 axons, respectively; bar: 40 µm. Dotted boxes correspond to image magnification, bar: 15 µm.

Increased pMLKL in neuronal somas (dark brown/+Nissl, violet) is observed in the both hippocampal subfields.
Aggregation pattern of pMLKL is observed in the Schaffer collateral projections, indicated with red arrowheads.

Fig S5. Necroptosis activation is correlated with axonal degeneration in diverse areas of the aging brain.
Necroptosis activation was measured through the phosphorylation of MLKL (pMLKL in green) in axon-enriched subfields of different brain areas. Pan-axonal NF antibody (red) was used to stain axons, and DAPI (blue) to detect nuclei and to identify granular cell layers and gray matter regions in the striatum, cerebellum and spinal cord, respectively. (a,b) Phosphorylated-MLKL mean intensity increases in axonal tracts of the striatum along aging. One-way ANOVA with Tukey analysis for multiple comparisons, *p:<0.05; **p<0.01, ***p<0.005. (b) Plots represent absolute cytokine levels (pg/ml) for IL-6 and TNF-α in serum samples of different mice groups. Error bars, mean ± SEM, *p:<0.05; **p<0.01; ****p<0.001. Statistical significance was determined by one-way ANOVA with Tukey analysis for multiple comparisons. One of the values corresponding to IL-6 levels in a GSK'872 mouse was considered outlier and excluded from the analysis. Outliers were identified with the "Identify Outliers" tool from GraphPad Prism v8.0.2. Clusters (groupings of proteins delineated by color) can be further analyzed with other in silico tools such as DAVID thereby allowing the data to be broken down into more manageable groupings. Cluster 1 is shown in green as clustering example. Fig S12. Proteomic data clustering and analysis. (a) Table containing information regarding the total number of proteins identified, and those belonging to each cluster of grouping proteins classified as "up" or "down" clusters.