NEMO reshapes the α-Synuclein aggregate interface and acts as an autophagy adapter by co-condensation with p62

Abstract

NEMO is a ubiquitin-binding protein which regulates canonical NF-κB pathway activation in innate immune signaling, cell death regulation and host-pathogen interactions. Here we identify an NF-κB-independent function of NEMO in proteostasis regulation by promoting autophagosomal clearance of protein aggregates. NEMO-deficient cells accumulate misfolded proteins upon proteotoxic stress and are vulnerable to proteostasis challenges. Moreover, a patient with a mutation in the NEMO-encoding IKBKG gene resulting in defective binding of NEMO to linear ubiquitin chains, developed a widespread mixed brain proteinopathy, including α-synuclein, tau and TDP-43 pathology. NEMO amplifies linear ubiquitylation at α-synuclein aggregates and promotes the local concentration of p62 into foci. In vitro, NEMO lowers the threshold concentrations required for ubiquitin-dependent phase transition of p62. In summary, NEMO reshapes the aggregate surface for efficient autophagosomal clearance by providing a mobile phase at the aggregate interphase favoring co-condensation with p62.

Fig. 1. NEMO is associated with pathological protein aggregates.

a Widespread mixed brain proteinopathy in a patient expressing mutant Q330X NEMO. Both low and high magnification images show the presence of aggregated proteins, such as α-synuclein, hyperphosphorylated tau, TDP-43, and amyloid beta in different brain regions. Many structures resembling Lewy body and Lewy neurites in pigmented neurons in the substantia nigra pars compacta (SNpc) are positive in immunostaining for α-synuclein, hyperphosphorylated tau, and ubiquitin. Scale bars, as indicated. b Domain structure of wildtype (WT) human NEMO and mutant Q330X NEMO. DD dimerization domain, CC1 coiled-coil 1 domain, CC2 coiled-coil 2 domain, UBAN ubiquitin-binding in ABIN and NEMO, LZ leucine zipper, ZF zinc finger. c, d M1-linked ubiquitin and NEMO co-localize with α-synuclein, tau, and TDP-43 aggregates in human brain. Immunofluorescent stainings of cortical or midbrain sections from patients with Parkinson’s disease (PD), Alzheimer´s disease (AD), or frontotemporal dementia (FTD). Brain sections were stained with antibodies against M1-ubiquitin (c), NEMO (d), and α-synuclein (PD), tau (AD), or TDP-43 (FTD); DAPI (blue). Scale bar, 10 µm.

Fig. 2. NEMO protects from proteotoxic stress.

a b NEMO-deficient cells are prone to protein aggregation under proteotoxic stress. a NEMO wildtype (WT) and knockout (KO) embryonic fibroblasts (MEFs) were heat stressed (42 °C, 1 h) or treated with the proteasomal inhibitor MG-132 (0.5 µM, 48 h) or the lysosomal inhibitor Bafilomycin A1 (BafA, 100 nM, 48 h) and then stained by Proteostat® to detect protein aggregates. Scale bar: 20 µm. b Cells positive for aggregates were quantified. All data are displayed as mean ± SD based on 3 independent experiments, analyzed by two-way ANOVA followed by Šídák’s multiple comparisons test. At least 150 cells were assessed per condition. Heat stress: *p = 0.0283, BafA: *p = 0.0414. c, d Wildtype NEMO but not Q330X NEMO decreases misfolding of the folding sensor FlucDM-EGFP-luciferase. NEMO KO MEFs transiently expressing FlucDM-EGFP-luciferase and either wildtype (WT) NEMO or Q330X NEMO were subjected to a heat stress (HS, 43 °C, 20 min) 48 h after transfection or left untreated. c The cells were then analyzed by immunocytochemistry and fluorescence microscopy. Shown is the fraction of NEMO-expressing cells with EGFP-positive foci. Data represent mean ± SEM based on 3 independent experiments. At least 900 transfected cells have been analyzed per condition. Statistics: one-tailed Mann–Whitney U-tests *p ≤ 0.05. d In parallel, luciferase activity of control and heat stressed cells were analyzed luminometrically. Data represent mean ± SEM based on 7 independent experiments. Statistics: One-way ANOVA with Bonferroni’s multiple comparison posthoc test; ***p ≤ 0,001. e, f Transient heat stress does not activate NF-κB signaling. e HEK293T cells transiently expressing an NF-κB-luciferase reporter construct were heat stressed for 20, 40, or 60 min (42 °C) and 8 h later luciferase activity was quantified. As a positive control, one set of cells was treated with TNF (10 ng/ml, 8 h). Data are shown as normalized mean ± SD based on 5 independent biological replicates. **p ≤ 0.01. f SH-SY5Y cells were heat stressed (42 °C) for the indicated time and then nuclear translocation of the NF-κB subunit p65 was analyzed by immunocytochemistry and fluorescence microscopy using antibodies against p65. As a positive control, one set of cells was treated with TNF (25 ng/ml, 15 min). Scale bar, 10 µm (overview) and 10 µm (inset).

Fig. 3. NEMO and LUBAC components are recruited to aSyn aggregates.

a M1-linked ubiquitin is enriched at aSyn aggregates formed in the cellular aSyn seeding model. SH-SY5Y cells stably expressing aSyn A53T-GFP were treated with aSyn A53T seeds, fixed 72 h after seeding, and analyzed by immunocytochemistry and fluorescence SR-SIM using M1-ubiquitin-specific antibodies. Scale bar, 10 µm (overview) and 5 µm (inset). b M1-linked ubiquitin colocalizes with pS129-aSyn-positive neurites in primary neurons. Primary cortical neurons were treated with aSyn A53T seeds at day 5 in vitro to induce aggregation of endogenous aSyn, fixed 7 days (rows 1 and 2) or 10 days (rows 3 and 4) after seeding, and analyzed by immunocytochemistry and fluorescence SR-SIM using antibodies against pS129-aSyn, M1-linked ubiquitin, and βIII-Tubulin. Scale bar, rows 1, 2, and 3: 10 µm (overview) and 5 µm (inset), row 4: 20 µm. c M1-linked ubiquitin co-immunoprecipitates with aSyn. SH-SY5Y cells stably expressing A53T-GFP aSyn were treated with aSyn A53T seeds for 72 h, lysed in 1% Triton X-100 in PBS, and aSyn-GFP was immunoprecipitated using GFP-trap beads. An immunoprecipitation with anti-HA beads was used to control for nonspecific binding. The pellet was analyzed by immunoblotting for M1-linked ubiquitin and GFP. The input was immunoblotted against aSyn and β-actin. d Endogenous NEMO is enriched at aSyn aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were treated as described in a and analyzed by immunocytochemistry and fluorescence SR-SIM using antibodies against NEMO and M1-linked ubiquitin. Scale bar, 20 µm (overview) and 10 µm (inset). For quantification of the relative NEMO fluorescent signals, the NEMO-specific signal at aSyn-GFP aggregates was compared to a cytoplasmic area outside aSyn aggregates. The mean intensity within the cytoplasm was set to 1. Data represent mean ± SEM (n = 10). Statistics: Two-tailed t-test, **p = 0.0026. e LUBAC components are recruited to aSyn aggregates. SH-SY5Y cells stably expressing α-Synuclein A53T-GFP were transiently transfected with plasmids encoding either HA-HOIP, HA-HOIL-1 L, or HA-SHARPIN, or HA-HHARI as a control. One day after transfection, the cells were treated with aSyn A53T seeds, fixed 48 h after seeding, and analyzed by immunocytochemistry and fluorescence SR-SIM using antibodies against the HA-tag. Scale bar, 10 µm (overview) and 5 µm (inset).

Fig. 4. In contrast to Q330X NEMO, WT NEMO is recruited to pathological protein aggregates and increases the abundance of M1-linked ubiquitin.

a In contrast to WT NEMO, Q330X NEMO is not present at aSyn aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were transiently transfected with either HA-tagged WT NEMO (left panel) or HA-tagged Q330X NEMO (right panel). After 24 h, the cells were treated with aSyn A53T seeds, fixed 48 h after seeding, and analyzed by immunocytochemistry and fluorescence SR-SIM using an antibody against the HA-tag. Scale bar, 20 µm (overview) and 10 µm (inset). The graph displays the percentage of cells showing colocalization of WT NEMO or Q330X NEMO with aSyn aggregates per field of view. Data are shown as mean ± SEM based on 8 technical replicates within 2 biological replicates. At least 41 cells per condition were quantified. Statistics: non-parametric Mann–Whitney U-test. ***p ≤ 0.001. b Q330X NEMO is not recruited to Htt-Q97-GFP aggregates. SH-SY5Y cells were transiently transfected with Htt-Q97-GFP and either HA-tagged WT NEMO (left panel) or HA-tagged Q330X NEMO (right panel). Cells were fixed after 72 h and analyzed by immunocytochemistry and fluorescence SR-SIM using antibodies against the HA-tag. Scale bar, 10 µm (overview) and 5 µm (inset). c Q330X NEMO is not recruited to insoluble Htt-Q97-GFP aggregates. NEMO KO MEFs were transiently transfected with Htt-Q97-GFP and either WT HA-NEMO, Q330X HA-NEMO or K285R K309R HA-NEMO. After 72 h, the cells were lysed and the SDS-insoluble fractions were analyzed by a filter retardation assay. The cellulose acetate membranes were immunoblotted for Htt and NEMO (using an HA antibody). The input was immunoblotted using antibodies against HA and β-actin. d Q330X NEMO does not bind to M1-linked ubiquitin. HEK293T cells were transiently transfected with either WT HA-NEMO or Q330X HA-NEMO. After 24 h, the cells were lysed and lysates were incubated with recombinant M1-linked tetra-ubiquitin (4×M1-ub) for 2 h at 4 °C. NEMO was immunoprecipitated using anti-HA beads. Immunoprecipitated NEMO was immunoblotted using antibodies against M1-linked ubiquitin. The input was immunoblotted for M1-linked ubiquitin, NEMO and β-actin. e Q330X NEMO is not M1-ubiquitinated upon TNF treatment or increased LUBAC expression. HEK293T cells were transiently transfected with WT HA-NEMO or Q330X HA-NEMO ± LUBAC (HOIP + HOIL-1L + SHARPIN) as indicated. After 24 h, one set of cells was treated with TNF (25 ng/ml, 30 min) to stimulate linear ubiquitination. After cell lysis under denaturing conditions, NEMO was immunoprecipitated using anti-HA beads. Immunoprecipitated NEMO was immunoblotted using antibodies against M1-linked ubiquitin. The input was immunoblotted for NEMO and β-actin. f In contrast to WT NEMO, Q330X NEMO does not co-immunoprecipitate with endogenous HOIP. HEK293T cells were transiently transfected with either WT HA-NEMO or Q330X HA-NEMO. After cell lysis, NEMO was immunoprecipitated using anti-HA beads. Anti-c-myc beads were used to control for nonspecific binding. Immunoprecipitated NEMO was immunoblotted for endogenous HOIP. The input was immunoblotted for HOIP, NEMO, and β-actin. g, h M1-ubiquitin chains at aSyn aggregates are increased by WT but not Q330X NEMO. CRISPR/Cas9 NEMO KO SH-SY5Y cells were transiently transfected with aSyn A53T-GFP and either WT NEMO or Q330X NEMO. One day after transfection, the cells were treated with aSyn A53T seeds, and fixed 48 h after seeding and analyzed by immunohistochemistry and fluorescence SR-SIM using antibodies against aSyn, M1-ubiquitin and NEMO. Colocalization of M1-ubiquitin, NEMO and aSyn aggregates (g) or M1-ubiquitin and aSyn aggregates (h) was quantified using the Pearson colocalization coefficient. g Data are displayed as mean ± SEM, n = 16 individual cells. Statistics: two-tailed Mann–Whitney U-test. ***p = 0.0010. h Data are displayed as mean ± SEM. n = 18, 23 individual cells. Statistics: two-tailed student´s t-test. ***p = 0.0003. i NEMO Q330X and NEMO D311N do not increase linear ubiquitination at Htt-polyQ aggregates. HEK293T cells were transfected with Htt-Q97-GFP and either wildtype HA-NEMO, HA-NEMO Q330X, or HA-NEMO D311N. The cells were lysed 72 h after transfection under denaturing conditions in 1.5 % (w/v) SDS. SDS-insoluble pellets were dissolved in formic acid. Formic acid-dissolved aggregates and the SDS-soluble fractions were analyzed by immunoblotting using antibodies against M1-linked ubiquitin, Htt, NEMO, and β-actin.

Fig. 5. A local NF-κB signaling platform is assembled at aSyn aggregates that does not promote nuclear translocation of p65.

a IKKα/β and p65 are recruited to aSyn aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were treated with aSyn A53T seeds, fixed on day 3 after seeding, and analyzed by immunocytochemistry and fluorescence SR-SIM using antibodies against phospho-IKKα/β, p65, phospho-p65, and M1-linked ubiquitin. Scale bar, 10 µm (overview) and 5 µm (inset). b, c Assembling of an NF-κB signaling platform at aSyn aggregates does not result in p65 nuclear translocation and impairs TNF-induced NF-κB activation. b Samples were prepared as described in a, treated with TNF for 15 min (25 ng/ml) on day 1, 2, and 3 after seeding, fixed and quantified for nuclear translocation of p65 by immunocytochemistry and fluorescence microscopy using antibodies against p65. For the seeded samples, only cells with aSyn aggregates were used for quantification. Data represent the mean ± SD of five independent experiments. Statistics: One-way ANOVA followed by Tukey’s multiple comparison test. ***p ≤ 0.001. c Representative immunofluorescence images of the experiment described in b (day 2 after seeding). Scale bar, 20 µm. d TNF-induced p65 nuclear translocation is impaired in cells with Htt-polyQ aggregates. SH-SY5Y cells were transiently transfected with GFP-tagged Htt-25Q or Htt-Q97 and either vector (co), wildtype HOIP or catalytically inactive C885A HOIP, as indicated. On day 3 after transfection, the cells were treated with TNF (20 ng/ml, 20 min), and nuclear translocation of p65 was analyzed as described in (b). Expression of HOIP was analyzed by immunoblotting, actin was used as input control. Data represent the mean ± SD of three independent experiments each performed in triplicates. At least 600 transfected cells were assessed per condition. Statistics: One-way ANOVA with Tukey’s Multiple Comparison posthoc test; ***p ≤ 0.001.

Fig. 6. NEMO decreases the number of cells with aSyn aggregates in a p62-dependent manner.

a WT NEMO but neither Q330X NEMO nor D311N NEMO decreases the number of cells with aSyn aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were transiently transfected with either WT HA-NEMO, Q330X HA-NEMO, D311N HA-NEMO, or mCherry as a control and treated with aSyn A53T seeds 24 h after transfection. The cells were treated 16 h after seeding with Bafilomycin A1 (BafA, 25 nM). 40 h after seeding, the cells were fixed and analyzed by immunocytochemistry and fluorescence microscopy using anti-HA antibodies. The fraction of HA- or mCherry-positive cells containing aSyn-GFP aggregates was quantified. Data are shown as mean ± SD based on 5 independent experiments. At least 750 cells per condition were quantified. Statistics were applied to the entire dataset. For better comparability, the control is shown for each NEMO construct. Statistics: One-way ANOVA followed by Tukey’s multiple comparison test. ***p ≤ 0,001. b Catalytically active HOIP decreases the number of cells with aSyn aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were transiently transfected with either WT HA-HOIP, catalytically inactive C885A HA-HOIP, or mCherry as a control, and treated as described in a. The fraction of HA- or mCherry-positive cells containing aSyn-GFP aggregates was quantified. Data are shown as mean ± SD based on 5 independent experiments. At least 750 cells per condition were quantified. Statistics were applied to the entire dataset. For better comparability, the control is shown for each HOIP construct. Statistics: One-way ANOVA followed by Tukey’s multiple comparison test. ***p ≤ 0,001. c NEMO and HOIP silencing increase the number of cells with aSyn-GFP aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were transiently transfected with either NEMO- or HOIP-specific siRNAs, treated with aSyn A53T seeds as indicated, and fixed 48 h after seeding. The fraction of cells containing aSyn aggregates was quantified. Data are shown as mean ± SD from 5 biological replicates. At least 750 cells per condition were quantified. Statistics: One-way ANOVA followed by Bonferroni’s multiple comparison test. *p ≤ 0,05, **p ≤ 0,01. d p62 is present at aSyn aggregates. SH-SY5Y cells stably expressing aSyn A53T-GFP were treated with aSyn A53T seeds, fixed 72 h after seeding, and analyzed by immunocytochemistry and SR-SIM fluorescence microscopy using anti-p62 antibodies. Scale bar, 10 µm (overview) and 5 µm (inset). e p62 co-immunoprecipitates with WT NEMO but not with Q330X NEMO. HEK293T cells were transiently transfected with either WT HA-NEMO, Q330X HA-NEMO, or WT FLAG-NEMO to control for unspecific binding. After cell lysis, HA-tagged proteins were immunoprecipitated using anti-HA-beads. Immunoprecipitated proteins were immunoblotted using antibodies against p62, M1-linked ubiquitin, and NEMO. The input was immunoblotted for p62, NEMO, and β-actin. f NEMO and HOIP reduce the number of aSyn aggregates in a p62-dependent manner, which requires the UBA domain of p62. p62 KO MEFs were transiently transfected with aSyn A53T-GFP, HA-NEMO, HA-HOIP or mCherry as a control, and p62 or p62ΔUBA, as indicated. One day after transfection, the cells were treated with aSyn A53T seeds, and fixed 48 h after seeding. The fraction of cells containing aSyn aggregates was quantified as described in a. Data are shown as mean ± SD based on 5 individual experiments. Statistics: One-way ANOVA followed by Tukey’s multiple comparison test. **p ≤ 0,01, ***p ≤ 0,001. g The p62-dependent effect of NEMO on aSyn aggregates is sensitive to lysosomal inhibition. p62 KO MEFs were transiently transfected with aSyn A53T-GFP and HA-NEMO, or HA-NEMO and p62, or mCherry as a control, as indicated. The next day, cells were treated with aSyn A53T seeds, and after 16 h treated with Bafilomycin A1 (BafA, 25 nM). 40 h after seeding, cells were fixed and the fraction of cells containing aSyn aggregates was quantified as described in A. Data are shown as mean ± SD based on 5 individual experiments. Statistics: One-way ANOVA followed by Tukey’s multiple comparison test. **p ≤ 0,01, ***p ≤ 0,001.

Fig. 7. p62 binds to aSyn aggregates in a NEMO-dependent manner.

a, b, c Colocalization of p62 and aSyn aggregates is decreased in the Q330X NEMO patient brain despite increased p62 expression. a Paraffin-embedded brain sections from control, DLBL (Dementia with Lewy Bodies) or the Q330X NEMO patient brain were analyzed by immunohistochemistry and fluorescence SR-SIM using antibodies against aSyn and p62. Scale bar, 200 µm. b Foci staining positive for both aSyn and p62 were quantified in 28-30 fields of view per brain section. Data are shown as mean ± SEM, n = 28/30 individual cells. c Foci staining positive for p62 only were quantified in 10 fields of view per brain section. Data are displayed as mean ± SEM from 10 fields of view. Statistics: Kruskal–Wallis test followed by Tuckey’s multiple comparison test. ***p ≤ 0.001. d, e Foci-like concentration of p62 and NBR1 at aSyn aggregates is reduced in NEMO-deficient cells. CRISPR/Cas9 NEMO KO or wildtype (WT) SH-SY5Y cells were transiently transfected with aSyn A53T-GFP. One day after transfection, the cells were treated with aSyn A53T seeds, fixed 48 h after seeding, and analyzed by immunohistochemistry and fluorescence SR-SIM using antibodies against p62 (d) and NBR1 (e). 3D-reconstructions were performed using the surface module of Imaris 10.0.1 image analysis software. Scale bars, D/E: 1 µm. f, g The abundance of LC3 and LAMP2 at aSyn aggregates is decreased in NEMO-deficient cells. CRISPR/Cas9 NEMO KO or WT SH-SY5Y cells were transiently transfected with aSyn A53T-GFP. One day after transfection, the cells were treated with aSyn A53T seeds, fixed 48 h after seeding, and analyzed by immunohistochemistry and fluorescence SR-SIM using antibodies against LC3 (f) and LAMP2 (g). 3D-reconstructions were performed using the surface module of Imaris 10.0.1 image analysis software. Coverage of aSyn-GFP aggregates by LC3 and LAMP2 was analyzed using the Imaris 10.0.1 surface modules and a surface-to-surface MatLab Plugin. The coverage of the reconstructed aSyn-GFP surface was quantified and plotted as percentage of the total aSyn-GFP aggregate surface for LC3 (f) and LAMP2 (g). Data are displayed as mean ± SEM, n = 11/20 individual cells for LC3 (*p = 0.0422) and n = 14/13 individual cells for LAMP2 (*p = 0.0143). Statistics: two-tailed student´s t-test. *p ≤ 0.05. Scale bars, f 2 µm (left panel), 1 µm (right panel), g 0.5 µm (left panel), 1 µm (right panel).

Fig. 8. NEMO promotes the local concentration of p62 by co-condensation with M1-linked ubiquitin.

a, b NEMO forms a mobile phase at immobile aSyn aggregates. CRISPR/Cas9 NEMO KO HeLa cells were transiently transfected with aSyn A53T-GFP and NEMO-Halo7. One day after transfection the cells were incubated with aSyn A53T seeds for 48 h. a For labeling of the Halo tag, the cells were incubated for 30 min with 2.5 µM TMR dye and washed for 30 min with cell culture medium prior to live cell imaging. Fluorescence recovery after photobleaching (FRAP) was performed by 5 consecutive bleaching pulses using the 561 nm laser at 100% intensity within a defined region of interest (white circles) at an aSyn-GFP aggregate (green). Fluorescence recovery was measured for 3 min and plotted as a percentage of baseline fluorescence. Scale bar, 1 µm b aSyn-GFP aggregates were bleached at two independent positions at the aggregate center or aggregate periphery (white circles) using 5 consecutive bleaching pulses using the 488 nm laser at 100% intensity within defined regions of interest. Fluorescence recovery was measured for 3 min and plotted as a percentage of baseline fluorescence. Scale bar, 5 µm. c p62 colocalizes with NEMO and M1-linked ubiquitin at aSyn aggregates. HeLa cells were transiently transfected with aSyn A53T-GFP. One day after transfection, the cells were treated with aSyn A53T seeds, fixed 48 h after seeding and analyzed by immunohistochemistry and fluorescence SR-SIM using anti-antibodies against p62, NEMO and M1-ubiquitin (all at endogenous expression). Shown are representative images of p62 co-localizing with NEMO (left panel) and p62 co-localizing with M1-ubiquitin (right panel) at aSyn-GFP aggregates. Scale bar, 1 µm. d The interaction of NEMO with p62 is enhanced in presence of linear ubiquitin chains. 2.5 µM recombinant MBP-NEMO WT or Q330X and 2 µM recombinant mCherry-p62 were incubated with or without 1 µM recombinant linear tetra-ubiquitin (4×M1-ub) with 25 µL MBP affinity agarose beads. Proteins were immunoblotted using antibodies against p62, and NEMO. e p62 and NEMO co-condensate in the presence of M1-linked polyubiquitin. 2.5 μM recombinant mCherry-p62 (red) mixed with 5 μM recombinant wildtype NEMO-GFP (green) were supplemented with M1-linked ubiquitin to induce phase separation. Top lane: No M1-linked ubiquitin added. Middle lane: 2.5 μM M1-linked tetra-ubiquitin (4×M1-ub). Bottom lane: 1.0 μM M1-linked octa-ubiquitin (8×M1-ub). Shown are laser scanning microscopy images. Scale bar, 10 μm. f NEMO reduces the threshold concentrations required for ubiquitin-dependent p62 phase separation. Phase diagrams depicting concentration-dependent phase separation of p62 and M1-linked octa- or tetra-ubiquitin (8×M1-ub or 4×M1-ub) with or without 5 μM recombinant wildtype NEMO. p62 was incubated in presence of recombinant M1-linked ubiquitin at the concentrations indicated and analyzed by laser scanning microscopy. Black empty circles: no phase separation; red/yellow solid circles: phase separation.