Neuron-specific activation of necroptosis signaling in multiple sclerosis cortical grey matter,Picon,Acta2021

Multiple Sclerosis Cortical Grey Matter and Necroptosis

Introduction

Progressive stages of multiple sclerosis (MS) are characterized by:
- Chronic demyelinated lesions in the white matter (WM) $[31]$
- Axon damage and loss in the WM $[26]$
- Diffuse changes in normal-appearing WM $[17, 27]$
- Increasing cortical grey matter (GM) pathology $[3, 18, 27, 45]$
The degree of GM pathology correlates better with clinical progression, physical disability, and cognitive impairment than WM lesion burden on MRI $[4, 11, 16, 37]$
Neurodegeneration in the GM is thought to be a main driver of irreversible clinical disability.
Neuronal loss, axon/neurite damage, and synaptic dysfunction contribute to GM neurodegenerative pathology $[14, 23, 35, 36, 45, 58]$
Mechanisms underlying cortical neurodegeneration and GM atrophy are still unclear but suggested to be stimulated by chronic inflammation $[3, 12]$
Neuronal loss in cortical GM associates with compartmentalized inflammation within the leptomeninges $[14, 18, 35]$
Leptomeningeal inflammation is suggested to drive underlying pathology, possibly by diffusion of pro-inflammatory cytokines into the GM $[21]$
Tumour necrosis factor (TNF) stimulates cell death in non-CNS chronic inflammatory conditions and is elevated in active lesions, CSF, and meninges of MS patients, and is associated with increased GM pathology $[34, 50, 53]$
Transcriptomics analysis of cortical GM suggested a dysregulation of TNF signaling towards activation of necroptosis in the presence of increased meningeal inflammation $[33]$
Soluble TNF (sTNF) binding to TNF receptor 1 (TNFR1) can regulate both pro-survival and cell death pathways, while transmembrane TNF (tmTNF) binding to TNF receptor 2 (TNFR2) mediates mainly cell survival and tissue regeneration $[1, 32, 46, 57, 61]$
Death signaling following sTNF/TNFR1 interaction is initiated when receptor-interacting protein kinase 1 (RIPK1) is deubiquitinated, leading to caspase-8 activation, which leads to apoptosis $[40, 41, 43, 51]$
However, when caspase-8 is inhibited, RIPK1 interacts with receptor-interacting protein kinase 3 (RIPK3), inducing autophosphorylation $[24, 56]$ , recruitment and phosphorylation of mixed lineage pseudokinase ligand (MLKL), leading to formation of the necrosome $[28, 55]$
MLKL oligomers translocate to the plasma membrane and execute plasma membrane rupture $[5, 7]$
Hypothesis: TNF produced in the meninges diffuses into the underlying grey matter to cause TNFR1-dependent necroptotic neuronal cell death, either directly, or by activating glial cells.
Report: Upregulation in multiple stages of TNFR1 signaling towards necroptosis activation in cortical neurons in progressive MS brain, indicating neurodegeneration is occurring via necroptosis rather than apoptosis. Chronically increased levels of TNF in CSF in an in vivo model reproduced findings in MS tissue and were associated with neurodegeneration and necroptosis activation in cortical neurons.
Findings reveal an inflammatory mechanism of cell death in cortical neurons that might result in new potential therapeutic avenues to target neurodegeneration in progressive MS.

Materials and Methods

Tissue samples were provided by the UK MS Society Tissue Bank at Imperial College London with informed consent under ethical approval (08/MRE09/31).
Demographic, clinical, and neuropathological features of SPMS cases and controls are shown in suppl. Table 1.
MS diagnosis was confirmed as previously reported $[49]$
For mRNA and protein level analysis, snap-frozen blocks were obtained from the cingulate gyrus, precentral gyrus, insula, and temporal gyrus from 28 SPMS cases (median post-mortem delay (PMD) = 17 h, range: 8–27; median age at death = 53 years, range: 38–70) and ten control brains (median PMD = 21 h, range: 10–48; median age at death = 63 years, range: 35–82).
For quantitative immunohistochemical analysis, 6 μm thick, formalin-fixed, paraffin-embedded sections were cut from the superior frontal cortex and cingulate gyrus from 17 SPMS cases (median PMD = 18 h, range 8–27; median age at death = 54 years, range: 38–27) and 11 controls (median PMD = 21 h, range: 10–48; median age at death = 63 years, range: 35–82), based on tissue availability.

Immunohistochemistry

Individual antibody details are listed in suppl. Table 2.
Human paraffin-embedded sections were deparaffinised, rehydrated, and subjected to epitope retrieval.
Sections were blocked with 10% normal horse serum (NHS), incubated with primary antibody overnight, and then incubated with ImmPRESS HRP-conjugated secondary antibodies (Vector Laboratories).
Slides were visualized with ImmPACT-DAB (Vector Laboratories) as chromogen, counterstained with hematoxylin, and DePex mounted.
Dual colour IHC was performed sequentially, detecting primary antibody with ImmPACT-DAB, followed by incubation with the second primary antibody and detection using the ABC-alkaline phosphatase detection system, using Vector Blue as substrate.
Snap-frozen tissue for IHC was pre-treated with cold methanol and followed the same steps as for paraffin tissue.
For immunofluorescence, sections were fixed with 100% methanol at −20 °C, blocked, and incubated overnight with primary antibodies, then incubated with appropriate secondary antibody conjugated to a fluorochrome, and nuclei counterstained with DAPI (Sigma-Aldrich) and mounted with Vectashield Anti-fade Mounting Media (Vector Laboratories).

Protein Extraction

Grey matter samples from MS and control tissue blocks were dissected carefully on a Leica cryostat, homogenized in RIPA buffer (Thermo Scientific) containing protease and phosphatase inhibitors (Thermo Scientific), and incubated on ice for 20 min at 4 °C.
The protein extract was centrifuged at $120,000 g$ for 15 min at 4 °C. The resulting supernatant was taken as the RIPA soluble fraction.
Pellets were washed in TBS and homogenized in 6 M urea/5% SDS for 30 min at room temperature (RT). The samples were stored at −80 °C until used for Western blotting.
To extract proteins in native condition, tissue was homogenized in TBS containing 0.1% of Triton X-100, incubated for 10 min at RT, and centrifuged at $16,000 g$ for 10 min.
The cytoplasmic and nuclear fraction was extracted using the NE-PER™ Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific) according to the manufacturer’s guidelines.

Western Blotting

Protein concentration was measured using a Pierce BCA protein assay kit (Thermo Scientific).
Subsequently, 10–50 μg of protein was loaded onto 4–12% Bis–Tris gels (Thermo Scientific) and transferred to polyvinylidene difluoride (PVDF) membranes for 1.30 h.
The membranes were incubated in 5% BSA or milk for 1 h at room temperature and incubated overnight at 4 °C with the appropriate primary antibodies. Individual antibodies used are listed in suppl. Table 2.
The blots were washed three times with TBS-T for 10 min and incubated with the specific secondary antibodies (1:20,000, Jackson laboratory) for 1 h at RT.
The blots were then washed with TBS-T and imaged/quantified using a Syngene G:Box.
For detection of MLKL monomers and dimers, proteins were run on 3–8% Tris–acetate gels (Thermo Scientific) in non-reducing conditions.

RNA Extraction and RT-PCR

Grey matter regions in each brain tissue block were dissected from serial sections as for protein extraction.
The grey matter tissue samples were then homogenized and processed for total RNA extraction using PureLinkTM RNA Mini kit (Life Technologies Corporation) as per the manufacturer’s protocol.
Purified total RNA (100 ng) was used from each sample for One-step real-time reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) using the iTaqTM Universal SYBR® Green One-Step kit (Bio-Rad) in the StepOnePlusTM Real-Time PCR system (Applied Biosystems).
The primers used for mlkl, ripk3, tnfr1, ripk1 and xpnpep1 were the commercially available PrimePCRTM SYBR® Green assay primers (Biorad).
For each sample, reactions were set up in triplicate with the following cycling protocol: 50 °C for 10 min, 95 °C for 1 min, 40 cycles with a 3-step protocol (95 °C for 15 s, 60 °C for 1 min), and a final melting curve analysis with a ramp from 65 to 95 °C.
Relative quantification of mRNA levels from various treated samples was determined by the $\Delta\Delta$ Ct method $[29]$ , after normalizing with the corresponding xpnpep1 levels $[8]$ from the samples.

Immunoprecipitation

Samples from MS and control brain grey matter were homogenized in lysis buffer (EDTA 1 mm, EGTA 1 mm, 0.1%, Triton X-100, 2 mm $MgCl_2$ , 150 mm NaCl, 50 mm pH 7.5 Tris–HCL) containing protease and phosphatase inhibitors.
50 μg of protein per sample was incubated with 6 μg of anti-RIPK3 (R&D) and 50 μl of Dynabeads Protein G (Thermo Scientific) while rotating overnight at 4 °C.
The beads were then washed three times with lysis buffer and mixed with loading NuPAGE LDS sample buffer (Thermo Scientific).
The beads were boiled at 70 °C for 10 min and placed on ice. Samples were loaded onto a 10% Bis–Tris gel (Thermo Scientific).
Gels were transferred to a PVDF membrane and incubated overnight at 4 °C with anti-MLKL followed by 1 h incubation at 25 °C with anti-rabbit IgG-HRP secondary antibody.
Blots were then washed and developed with Clarity (BioRad).

Animals

Eight- to ten-week-old female Dark Agouti rats (140–160 g; total n=9) were obtained from Janvier Labs (France) and maintained in groups of four in a 12 h light/dark cycle and had ad libitum access to food and water.
The UK Home Office approved all procedures.

Lentiviral Vector Production

Lentiviral vectors carrying the human TNF and IFNγ genes were produced as described elsewhere $[21]$
293T cells were transiently transfected with the HIV-1 vector plasmid pRRLsincppt-CMV-TNF-WPRE or pRRLsincppt-CMV-IFNƴ-WPRE), the packing vector containing HIV-1 gag/pol gene (pMD2-LgRRE), HIV-1 Rev (pRSV-Rev) and VSVG envelop plasmids using calcium phosphate.
After 16 h the medium was replaced with fresh medium supplemented with 10 mM sodium butyrate. At 36 h, vector-containing medium was harvested and filtered through a 0.45-μm filter.
The supernatant was then centrifuged overnight. The pellet was concentrated by ultracentrifugation and resuspended with TSSM (10 mM Tromethamine, 100 mM NaCl, 10 mg/ml sucrose and 10 mg/ml mannitol) over several hours.
The genome copy number was calculated using the Clontech Lenti-X qRT-PCR Titration kit (Takara).

Intracerebral Injection of Lentiviral Vectors

Surgeries were performed as previously reported $[13]$
Rats were anaesthetized with isofluorane and a 2 mm hole was drilled in the midline 0.9 mm caudal to bregma. A finely calibrated glass capillary needle attached to a Hamilton 10 μl syringe was then inserted stereotactically to a depth of 2.3 mm below the dural surface, down the sagittal sulcus.
The rats were then injected with 4 µl of a lentiviral mixture ( $5 \times 10^8$ genomic copies (GCs) of TNF and $5 \times 10^7$ GCs of INFγ) a rate of 0.20 μl/ml. The needle was left in place for 5 min to allow diffusion of the sample from the area and then slowly withdrawn.
Animals were perfused after 28 days with 4% paraformaldehyde in PBS (PFA) under sodium pentobarbitone anaesthesia. The brains were removed, further fixed overnight in 4% PFA and then cryoprotected in 30% sucrose in PBS before embedding in OCT and immersion in isopentane cooled on dry ice.

Primary Cortical Neuron Culture

Primary cortical neurons were obtained from P1 Sprague–Dawley rat pups. Cerebral cortices were dissected in ice-cold HBSS –Ca2+/–Mg2+ (plus 10 mM HEPES, pH 7.3), treated with 0.25% papain, washed in plating medium, and incubated for 1 h at 37 °C, pipetting up and down every 15 min until no remaining clumps were observed.
The resulting cells were seeded at $1 \times 10^5$ cells/well on Poly-d-lysine-coated 96 well plates (10 mg/ml, Sigma-Aldrich) in media containing 10% heat-inactivated FBS (Invitrogen), DMEM (Invitrogen) and penicillin/streptomycin (Invitrogen).
After 2 h, the plating media was changed to Neurobasal™ Plus medium (Invitrogen) supplemented with 2% B27 Plus (Invitrogen), Glutamax (Invitrogen) and penicillin/streptomycin (Invitrogen).
The culture was maintained in a humidified atmosphere of 5% $CO_2$ in air at 37 °C. All experiments were performed at 10–12 days in vitro (10–12 DIV) at which point the purity of the cultures was 98.7% NeuN+ cells (CL=96–99% from n=6 experiments). The only contaminating cells were a small number of GFAP+astrocytes.
Treatment with rat TNF (Peprotech), SMAC mimetics (Tocris) and Z-VAD-FMK (R&D Systems) was performed at the concentration indicated in the text. Necroptosis inhibitors GSK-547 (Sigma-Aldrich), GSK-872 (Sigma-Aldrich) and necrosulfonamide (Sigma-Aldrich) were added to the cultures for 24 h at the concentration indicated in the text.

LDH Assay

Cell cytotoxicity was determined by measuring the lactate dehydrogenase (LDH) release (LDH Assay kit) on primary cultured neurons. Supernatant (20 μl) was collected from the neuronal cultures under different conditions and samples were processed following the kit instructions.
The LDH released was measured in a TECAN reader at 450 nm.

Image Acquisition and Analysis

Immunohistochemistry slides stained for MOG, HuC/D, HLA-DR, RIPK1, pRIPK3, pMLKL and TNFR1 were digitized by whole slide scanning using an Aperio SCF400F scanner (Leica, Wetzlar, Germany). Image files were handled using QuPath v0.2.0. software to measure areas of demyelination.
The entire GM fraction was traced and measured for each MOG-stained slide before measuring the area of individual GM lesions. The mean GM lesion area was reported per section and per case as the percentage of total GM (suppl. Fig. 1).
HLA-DR+ and TNFR-1 immunoreactive cells were counted manually from strips of sulcal grey matter extending from the pial surface to the WM, separating the data into subpial layer I, layers II–III and layers V–VI. HLA-DR+ cells were also counted within the subarachnoid space as a marker of leukocyte infiltration (suppl. Fig. 1).
HuC/D, RIPK1, pRIPK3 and pMLKL cell densities were measured with the “Positive Cell Detection of a Region of Interest” tool in QuPath. Cortical layers II–III and V–VI were identified, and regions of interest were selected in each section. To identify only neurons, we restricted the analysis to immunostained features with an area > 60 μm2.
To determine the levels of perivascular and meningeal inflammation, the total number of CD3+ cells/mm and CD20+ cells/mm was assessed as previously described and digitized using an Olympus BX63 microscope (suppl. Fig. 1) $[18]$ Three regions of the meninges were selected in the subarachnoid space and 8-perivascular space profiles were selected for each slide and cells counted manually.
Cases showing more than 50 (CD3+ or CD20+) cells packed within the subarachnoid spaces were considered as the high meningeal inflammation group (high-inf) and less than 50 as the low inflammation group (low-inf). Sections from human post-mortem tissue labelled by double immunohistochemistry for necroptotic proteins were imaged with an Olympus BX63 scanning microscope.
Scanning of the cortical ribbons was performed at 20×magnification and regions of interest were selected in layers II–III and V–VI and cells co-localizing NeuN with pRIPK3, cleaved caspase 3 and pMLKL counted manually with ImageJ software.
Immunofluorescence sections from the post-mortem human tissue and rat sections were imaged with either an epifluorescence Olympus BX63 scanning microscope or a SP8 Leica confocal microscope. Neuronal numbers in rat tissue were counted using NeuN+staining with automatic cell counting using the software QPath, as described above. Regions of interest were drawn outlining layers II and layer III and layers V and VI within the cingulate cortex.
Iba-1+ and GFAP+ cells were counted manually at 20×using QPath, in the same regions of interest as the NeuN staining. To analyse the number of NeuN+ cells co-expressing TNFR1, pRIPK3 and pMLKL, we selected regions of interest in layers II–III and V–VI and counted manually at 20×using ImageJ software.
For all the immunofluorescence analyses, the digital image settings were fixed during acquisition from each experiment. The total number of cells for all the studies is given as total cells/mm $^2$ .
In vitro experiments were imaged with an Olympus BX63 scanning microscope. Quantitative analysis of pixel intensity (NfH-20, NeuN, pRIP1, pMLKL) was performed with ImageJ using regions of interest at 20×. To minimize variability between images, pixel intensity was normalized to an unstained area and the exposure time and microscope setting were fixed throughout the acquisition. The neurite degeneration index was assessed in immunofluorescence images of 200 KDa neurofilament protein-immunostained primary cortical neurons using ImageJ. Relative neurite integrity was based on the ratio of fragmented to total neurites longer than 150 µm.

Statistical Analysis

All human post-mortem data was assumed to be sampled from a non-Gaussian distribution and non-parametric analysis methods applied. The difference between two groups was compared using the unpaired Mann–Whitney test, whilst the Kruskal–Wallis test with Dunn’s multiple comparison post-test was used when comparing three or more groups. Spearman correlation was used to test for associations between groups and the Spearman r and p values reported in each instance.
To analyse in vivo and in vitro experiments we used one-way ANOVA with Bonferroni’s post-hoc correction for multiple comparisons. A two-sided p value < 0.05 was considered significant.

Data Availability

All the data have been made available as part of the manuscript.

Results

TNFR1/RIPK1 Signaling is Upregulated in Neurons in MS Grey Matter

A previous transcriptomic analysis of cortical MS GM identified a possible dysregulation in TNF signaling at the mRNA level in secondary progressive MS (SPMS) brains.
Detailed changes in TNF signaling pathways at the protein level were studied in 28 brains from secondary progressive MS patients (SPMS) and ten non-neurological controls (detailed clinical and neuropathological characterization in suppl. Table 1 and suppl. Fig. 1).
The cohort of MS brains was chosen to have a wide range of cortical GM demyelination (mean: 30.5 ± 3.3%), perivascular inflammation (mean: 79.9 ± 12.9 cells/mm $^2$ ), meningeal inflammation (mean: 114.0 ± 17.8 cells/mm $^2$ ), microglia activation (mean: 156.0 ± 28.9 HLA-DR+ cells/mm $^2$ ) and neuronal density (mean: 409.5 ± 18.1 HuC/D+ cells/ mm $^2$ ) (suppl. Fig. 1a–g).
Neuronal loss was evident through all layers but only reached significance in layers II–III (suppl. Fig. 1g).
Immunohistochemistry (IHC) demonstrated a substantial and significant upregulation of TNFR1 expression in cortical macroneurons in layers II–III and V–VI in MS GM, with no apparent reactivity in microglia, astrocytes, or oligodendrocytes (Fig. 1a–c).
Neuronal TNFR1 expression was almost absent in the normal control cortical GM (Fig. 1a). The percentage of total neurons expressing TNFR1 was increased by 12.4-fold in MS and constituted 14.9 ± 3.0% in MS and only 1.2 ± 0.7 in controls (Fig. 1d).
TNFR1 protein and mRNA levels were both significantly increased in the GM of MS cases (Fig. 1e, f; suppl. Fig. 2a). In contrast, TNFR-II protein and mRNA levels did not vary between groups (Fig. 1e, f; suppl. Fig. 2a).
Next, the link between TNFR1 and activation of cell death by studying the expression of three key initial regulators: Fas ligand associated death receptor (FADD), cylindromatosis (CYLD), and receptor-interacting protein kinase 1 (RIPK1) [43] was investigated.
A significant upregulation of FADD was found in MS grey matter compared to controls, while CYLD expression was unchanged (Fig. 1e, f; suppl. Fig. 2b).
RIPK1 is the key molecular switch for TNF signaling pathways and its protein levels were significantly increased by a mean of sevenfold in MS grey matter, this result being also verified for mRNA levels (Fig. 1e, f; suppl. Fig. 2a).
There was a non-significant trend towards the association between RIPK1 protein levels and the extent of cortical demyelination and meningeal inflammation (suppl. Fig. 2c), but no correlation with any clinical outcome variables (not shown).
RIPK1 was found to be expressed mainly in a discrete population of neurons spread throughout the entire cortex in MS GM, with only low levels of expression in astrocytes and microglia (Fig. 1g, h; suppl. Fig. 2d). The most striking upregulation was in pyramidal neurons in layers II–III and V–VI of the MS cortex, (Fig. 1h) and it was present as both diffuse cytoplasmic staining and large aggregates (Fig. 1g). In contrast, expression in the control brain was very low (Fig. 1g, h).
These results indicate a neuronal-specific activation of TNFR1-mediated signaling in MS GM.

Cleaved Caspase-8 Levels and Apoptotic Signaling are Downregulated in MS Grey Matter

To determine whether apoptotic signaling was activated in neurons in MS cortex, the activation of caspase-dependent apoptosis was studied.
The protein levels of the cleaved active p18 subunit were significantly downregulated by 5.3-fold in MS GM, whereas the subunit p43 levels were unchanged (Fig. 2a, b).
The p18 subunit activates the downstream pathway of apoptosis, beginning with the cleavage of procaspase-3 into the p17 and p19 subunits. A low level of cleaved caspase-3 (CC3) staining was detected in MS and controls with no significant differences between groups (Fig. 2c, d). Additionally, labelled cells were mainly located in the immediate subpial layer (Fig. 2c, d).
In MS GM, the vast majority of CC3+cells co-expressed the astrocytic marker GFAP, with a small number co-expressing the oligodendrocyte marker Olig-2 (Fig. 2e). NeuN+CC3+double-positive neurons were extremely scarce and no differences were found between MS and controls (0.08 ± 0.04% of NeuN+cells in MS and 0.18 ± 0.14% in controls) (suppl. Fig. 3a, b).
These results indicate the downregulation of caspase-8-dependent apoptosis in neurons in MS.

Activation of Necroptosis in MS Cortical Neurons

Next, whether necroptotic signaling was activated in cortical neurons in MS, which relies on the phosphorylation of first RIPK3 and then MLKL, was investigated.
An increase in the mean gene expression for both RIPK3 and MLKL in MS cortex, 2.2- and 2.3-fold, respectively was found (Fig. 3a).
The levels of the non-activated RIPK3 and MLKL proteins were not significantly altered (Fig. 3b, c). However, pMLKL protein levels were remarkably 89.3-fold upregulated in the grey matter of MS cases compared to control (MS: 8.93 ± 2.65; Ctrl: 0.1 ± 0.1), together with a trend towards increased levels of pRIPK3 (MS: 2.2 ± 0.55; Ctrl: 0.9 ± 0.4) (Fig. 3b, c).
PhosphoMLKL expression in MS cortex appeared to be highly heterogenous, suggesting that pMLKL levels could be related to the microenvironment in individual brains (Fig. 3c).
When MS cases were classified according to the severity of meningeal inflammation, pMLKL and MLKL levels were both significantly increased in cases with more abundant meningeal inflammation when compared to those with low levels of meningeal inflammation (6.5- and 8.0-fold, respectively) (Fig. 3d). Significant moderate correlations were observed between the levels of both MLKL and pMLKL and cortical demyelination and inflammation (Fig. 3e, f).
Within the cortical layers, pRIPK3-expressing cells morphologically resembled pyramidal neurons and their density was increased in both layers II–III and V–VI in MS compared to controls (Fig. 4a). Such cells were rarely seen in the control brain. Immunolabeling was mainly diffuse in the cytoplasm together with some nuclear localisation (Fig. 4a).
The cellular identity of the majority of pRIPK3-positive cells was confirmed as neuronal using double immunohistochemistry with NeuN (4.36% ± 1.98% of total NeuN+neurons in MS; 0.56 ± 0.47% in control brain) (Fig. 4b). Some low-level immunoreactivity was also seen in astrocytes, but none in microglia (Fig. 4c). Co-localisation of pRIPK3 together with TNFR1 (Fig. 4c) strongly suggests a TNFR1-mediated activation of necroptosis signaling in cortical neurons in MS GM.
Immunohistochemistry to further validate the expression of pMLKL in the GM of MS cases demonstrated an 8.6-fold increase in the number of pMLKL+cells in comparison to controls (Fig. 4d). Indeed, a small number of pMLKL+neurons were only seen in one of the control brains (Fig. 4d). The highest numbers of pMLKL+neurons were found in layers II–III, with a 30-fold increase (Fig. 4d).
In line with findings for RIPK1 and pRIPK3, double immunohistochemistry revealed that pMLKL was mainly expressed by NeuN+pyramidal cortical neurons in the GM, with 1.98% ± 0.5% of total neurons expressing pMLKL in MS cases compared to 0.06% ± 0.06% in controls (Fig. 4e).
Next, whether the surrounding inflammatory microenvironment was associated with the presence of pMLKL in neurons was investigated. A strong significant correlation with the number of HLA-DR+ (a marker of activated microglia/macrophages) cells in the GM (r = 0.6; p = 0.01) and a weaker non-significant correlation with HLA-DR+cells within the meninges (r = 0.4; p = 0.09) was foun (suppl. Fig. 4).

Necrosome Formation Occurs in the MS Cortex

Upon phosphorylation, MLKL is reported to oligomerize in the cytoplasm and migrate to the plasma membrane, where it causes membrane rupture and ultimately cell death. Both MLKL and pMLKL were variably expressed in both the cytoplasmic and nuclear/perinuclear compartment in neurons (Fig. 5a, b), which was validated by subcellular fractionation and WB (Fig. 5d).
Importantly, pMLKL protein co-localizing with TNFR1 in neurons was found, supporting a link between TNFR1 and necroptosis activation in MS (Fig. 5c).
In addition, previous studies have shown that the necrosome complex, consisting of RIPK1, RIPK3 and MLKL, assembles as a fibrillar structure that can be extracted by chaotropic agents such as urea $[28]$ . In keeping with this, RIPK3 co-immunoprecipitated with MLKL in MS GM but not in controls (Fig. 5e) and the presence of RIPK1 and MLKL in the urea soluble fraction by WB was identified and this was markedly increased in MS samples (Fig. 5f).
Using non-reducing conditions to probe the oligomeric state of MLKL, the presence of MLKL oligomers of approximately 250 KDa only in the MS cortex was found, most likely representing tetramers (Fig. 5g).

The Number of pMLKL Expressing Neurons is Associated with More Rapid Disease Progression

To determine the relationship between the activation of necroptosis in neurons and disease progression, the density of neurons expressing pRIPK3 and pMLKL with clinical parameters (Fig. 6) was correlated.
The density of pRIPK3+ cells correlated inversely with the time in the progressive phase of the disease (r = −0.6; p = 0.02) and there was a similar non-significant trend with the disease length (r = −0.4; p = 0.06) (Fig. 6a).
The clinical variables with the density of pMLKL+neurons were further compared and strong significant inverse correlations between the density of pMLKL+cells and the age at progression (r = −0.55; p = 0.03), the age at death (r = −0.7; p = 0.008) and the disease length (r = −0.5; p = 0.04) was found (Fig. 6b).

Chronic Exposure to TNF and IFNγ in the CSF Induces Neurodegeneration

To determine whether TNF-mediated induction of necroptotic neurodegeneration could be induced by chronic expression of pro-inflammatory cytokines in the meningeal compartment, a novel animal model in which lentiviral vectors containing the transgenes for human TNF and IFNγ were injected into the brain midline subarachnoid space of Dark Agouti (DA) rats, thereby inducing meningeal cells to persistently produce these cytokines was used.
Previous studies had ascertained that it was necessary to combine TNF with IFNγ to induce TNFR1 expression within CNS cells $[13, 21]$ After 28 days post injection (dpi), the presence of leukocyte aggregation in the sagittal sulcus in rats injected with cytokine vectors (suppl. Fig. 5a, b), with meningeal infiltrates enriched in CD4/CD8 T-lymphocytes and CD79a B-lymphocytes was observed.
Injected rats displayed an increased number of microglial cells (Iba-1+cells) in the GM upper layers with no evident changes in the number of astrocytes (GFAP+cells) (suppl. Fig. 5c, d). These results were accompanied by a significant 20% decrease in the number of neurons located in cortical layer II, with smaller non-significant decreases in the deeper cortical layers III and V (suppl. Fig. 5e).
I n T N F / I F N γ v e c t o r i n j e c t e d a n i m a l s , TNFR1+ NeuN+ neurons were observed predominantly in the upper cortical subpial layers (Fig. 7a).
Quantitative analysis of the TNFR1+NeuN+neurons after 28 dpi showed a significant 125-fold increase within layers II/III (Fig. 7a). Negligible numbers of TNFR1+cells were seen in the cortical layers of the GFP vector control group. An increase in pRIPK3+ NeuN+ and pMLKL+ NeuN+ neurons was also seen in layers II–III (1.7- and 7.9-fold, respectively), whereas pMLKL+cells were rarely present in GFP vector injected animals (Fig. 7b, c).
Triple immunofluorescence staining for pMLKL, TNFR1 and NeuN demonstrated the co-localization of the three markers only in TNF/IFNγ vector injected animals, with a significant increase and higher proportion of neurons co-localizing pMLKL and TNFR1 in layers II–III, but not layers V–VI (Fig. 7d, e).
Immunofluorescence images for MLKL and pMLKL with the neuronal marker NeuN showed that neurons were expressing MLKL within both the cytoplasm and nucleus, whereas pMLKL staining was in large aggregates within the cytoplasm (suppl. Fig. 5f, g).In keeping with the human tissue results, cleaved caspase-3 immunoreactivity was only observed in GFAP+astrocytes and not in NeuN+neurons (Fig. 7f).
Thus, the persistent production of TNF and INFγ in the cerebral subarachnoid space of DA rats induces neurodegeneration associated with an increase in expression of necroptotic signaling molecules in neurons in the underlying upper cortical layers.

In Vitro Stimulation with TNF Induces Necroptosis Activation in Cortical Neurons when Apoptosis is Inhibited

To evaluate the activation of necroptosis by TNF in cortical neurons, dissociated primary cortical neurons were exposed to TNF (100 ng