Fig 2: Ku70 phosphorylation at Ser77/78 impairs binding to DNA damage foci.a Increased phosphorylation of Ku70 at Ser77/78 or Thr90 was observed commonly in human AD brains (mean, N = 5) and HMGB1-treated human U2OS cells (mean, N = 16). Co-addition of anti-HMGB1 antibody suppressed the increase of phosphorylation at Ku70 at Ser77/78 or Thr90. Phosphorylation at Thr401 and Ser520 was increased in human AD brains but not stimulated by HMGB1 in U2OS cells. Phosphorylation at Ser550 was stimulated by HMGB1 in U2OS cells but not increased in human AD brains. b Based on Ku70 structure at PDB (ID: 1JEY) (left), structure of Ku70 phosphorylated at Ser77 and Ser78 was modeled by PyMOL (Shrödinger, LLC) (right), suggesting that phosphorylation at Ser77 and Ser78 increases their polarity and interrupts hydrogen bond at the interaction surface between Ku70 and DNA. The positions of Thr90, Thr401, and Ser520 in the Ku70–DNA complex model are close to Ku70–DNA interaction surface as indicated in Supplementary Fig. 2. c Molecular dynamics (MD) simulation. Left panel shows root mean square difference (RMSD) reflecting the squared average of 3D positional changes (coordinate values) of all atoms in the Ku70–Ku80 protein complex. The graphs of non-phosphorylated Ku70 and pSer77/78 Ku70 (SP1) revealed stabilities of their structures. Middle panels show the distance (nm) and relative frequency of distance between Ser77(O) and DNA-thymidine4(P) during 30 ns, and right panels show those of Ser77(O) and DNA-thymidine4(O1P). MD simulation suggested that the distance of pSer77/78 Ku70 from DNA was larger than that of non-phosphorylated Ku70. d Gel shift assay was performed with GST-fusion proteins of phospho-mimetic and non-phospho-mimetic mutants of Ku70 at Ser77/78.

Fig 3: Ku70 phosphorylation at Ser77/78 impairs Ku70 foci formation after DNA damage and prevents deacetylation by SIRT1.a Ku70 phosphorylation at Ser77/78 impairs dynamics to DNA damage foci. Upper panels show accumulation of wild-type, non-phospho-mimetic mutant, and phospho-mimetic mutants of Ser77/78 or Ser77 to DNA damage foci in U2OS cells after microirradiation. Values in each group are summarized by mean ± SEM. b Co-immunoprecipitation of phospho-mimetics of Ku70 reveals a physical interaction between Ku70 and either SIRT1 or 14-3-3. Phosphorylation of Ku70 results in a change of acetylation state. The phosphorylation state of Ku70 does not affect interactions with Ku80 or DNA PKcs. Ku70 does not interact with TDP43 or VCP. Right graphs show quantitative analyses of three independent experiments (n = 3, ##p < 0.01 in Dunnett’s test). c Positional relationship between Ser77/78 and Lys331 in the Ku70 structure when bound to DNA (PDB: 1JEY). d Hypothetical model depicting the competitive interactions of SIRT1 and 14-3-3 with Ku70 and the resultant acetylation of Lys331.

Fig 4: Interruption of HMGB1 signal recovers neuronal DSB.a The abundance of DNA double-strand breaks was evaluated using two markers (?H2AX and 53BP1) in HMGB1 antibody-treated or control IgG-treated 5xFAD mice and non-transgenic sibling mice (B6/SJL). The graphs on the right show quantifications of the intensity/nuclei for the four groups. N = 3 mice, n = 90 cells. Tukey’s HSD test revealed increased DNA damage in 5xFAD mice and a rescue via treatment with human monoclonal anti-HMGB1 antibody. b Western blot analysis of the entire cerebral cortices from the four groups with anti-?H2AX and anti-53BP1 antibodies (N = 3). c Immunohistochemical staining of ?H2AX and 53BP1 in occipital cortex neurons of 5xFAD mice that received subcutaneous or intravenous administration of human monoclonal anti-HMGB1 antibody (#129) or control human IgG. The graphs on the right show quantitative analyses of the ?H2AX and 53BP1 signals in MAP2-positive neurons. d Western blot analyses of ?H2AX and 53BP1 levels in whole cortex samples. e Immunohistochemical staining of human postmortem brain tissue samples (occipital lobe) from patients with no neurological diseases and AD patients with anti-?H2AX antibody. f The dynamics of Ku70 recruitment to DNA damage foci induced via microirradiation were evaluated by transiently expressing EGFP-Ku70 in U2OS cells cultured in the presence/absence of HMGB1 (0.24 nM). Human monoclonal anti-HMGB1 antibody (#129) but not control human IgG or mock treatment rescued the delayed EGFP-Ku70 accumulation at DNA damage foci. g Quantitative analyses of the EGFP-Ku70 intensities at damage sites from time-lapse images of U2OS cells (n = 5) from which the EGFP-Ku70 intensities from a non-damaged area of the nucleoplasm were subtracted. The rate of EGFP-Ku70 accumulation was reduced in the presence of HMGB1 but recovered upon co-addition of human anti-HMGB1 monoclonal antibody to the culture medium. Values in each group are summarized by mean ± SEM. Box plots show the median, quartiles, and whiskers that represent data outside 25th to 75th percentile range.

Fig 5: Therapeutic effects of a PKC inhibitor in vivo.a Experimental protocol. Protocols for human monoclonal anti-HMGB1 antibody (s.c. or i.v.) were similar to those in Supplementary Figs. 4a and 11a. Injection was started after the symptomatic onset of 5xFAD mice. A PKC inhibitor (Gö6976) was continuously injected intrathecally via an Alzet micro-osmotic pump (6.6 µM, 0.15 µL/h) from 1.5 to 6 months of age. Y-maze tests were performed at 6 months of age and brain samples were dissected. b Immunohistochemistry of mice treated by the protocol in a. ?H2AX and 53BP1, the markers of DSB, revealed that DNA damage increased in 5xFAD mice was recovered by PKC inhibitor (Gö6976). c Left panels: western blot analysis of Gö6976-treated and non-treated mice confirmed the increase of DNA damage in 5xFAD mice and the recovery by Gö6976. Western blotting of various PKCs confirmed the activation of PKCa and the increase of pSer77/78 Ku70 in the cortex of 5xFAD mice. Gö6976 treatment similarly suppressed PKCa and pSer77/78 Ku70 in the cortex of 5xFAD mice. Middle panels: full-scan of another set of samples reconfirmed the increase of pSer77/78 Ku70, which matched the pattern of PKCa-mediated Ku70 phosphorylation (Fig. 4a, b). Right graphs show quantitative analysis of the ?H2AX and 53BP1 band signals. d Gö6976 treatment (shown in a) rescued cognitive impairment of 5xFAD mice in Y-maze test alteration ratio. e Human monoclonal anti-HMGB1 antibody also recovered cognitive impairment of 5xFAD mice in Y-maze test alteration ratio. f Rescuing effect of Gö6976 and human monoclonal anti-HMGB1 antibody on the secondary necrosis in 5xFAD mice. Left panels show the immunohistochemistry with anti-pSer46-MARCKS antibody that differentiate active necrosis, secondary necrosis, and Aß extracellular aggregates reflecting the ghost of neuronal necrosis37. Right graphs show quantitative analysis in each group of active necrosis, secondary necrosis, and Aß-positive ghost of cell death (Student’s t-test). Active necrosis was not significantly changed, whereas secondary necrosis and the following ghost of cell death were decreased. Especially, secondary necrosis was remarkably decreased in multiple-group comparison (Tukey’s HSD test). Box plots show the median, quartiles, and whiskers that represent data outside 25th to 75th percentile range.