Sting and p53 DNA repair pathways are compromised in Alzheimer's disease

Abstract

Alzheimer's disease (AD) is the most common cause of dementia. A common finding in AD is DNA damage. Double-strand DNA breaks (DSBs) are particularly hazardous to neurons because their post-mitotic state forces neurons to rely on error-prone and potentially mutagenic mechanisms to repair DNA breaks. However, it remains unclear whether DNA damage results from increased DNA damage or failure of DNA repair. Oligomerization of the tumor suppressor protein p53 is an essential part of DSB repair, and p53 phosphorylated on S15 is an indicator of DNA damage. We report that the monomer:dimer ratio of phosphorylated (S15) p53 is increased by 2.86-fold in temporal lobes of AD patients compared to age-matched controls, indicating that p53 oligomerization is compromised in AD. In vitro oxidation of p53 with 100 nM H₂O₂ produced a similar shift in the monomer:dimer ratio. A COMET test showed a higher level of DNA degradation in AD consistent with double-strand DNA damage or inhibition of repair. Protein carbonylation was also elevated (190% of control), indicating elevated oxidative stress in AD patients. Levels of the DNA repair support protein 14-3-3σ, γ-H2AX, a phosphorylated histone marking double strand DNA breaks, and phosphorylated ataxia telangiectasia mutated (ATM) protein were all increased. cGAS-STING-interferon signaling was impaired in AD and was accompanied by a depletion of STING protein from Golgi and a failure to elevate interferon despite the presence of DSBs. The results suggest that oxidation of p53 by ROS could inhibit the DDR and decrease its ability to orchestrate DSB repair by altering the oligomerization state of p53. The failure of immune-stimulated DNA repair may contribute to cell loss in AD and suggests new therapeutic targets for AD.

Figure 1

DNA damage is increased in AD. Each panel shows ten randomly selected nuclei isolated from human temporal lobes from two AD and two age-matched control patients. Isolated nuclei were plated onto agarose slides, lysed, and incubated in high-pH unwinding solution. After agarose electrophoresis (4°, 21 V, 1 h), they were stained with SYBR Gold, illuminated with a 402-nm laser, and emission at 530 nm was photographed at a fixed distance from the slide. Representative images are shown. Fifty comets from each image (5 AD and 5 control) were analyzed. Analysis of all ten samples showed that AD samples contained consistently higher percentage of DNA in the tail and increased tail moment (Fraction of DNA in tail: AD = 0.83 ± 0.02; Control = 0.75 ± 0.02, n = 5, p = 0.0091. Tail moment: AD = 1.62 ± 0.16 of control, n = 5, p = 0.0034, two tailed Student’s t test), indicating greater dsDNA breakage in the AD samples.

Figure 2

Changes in DNA damage repair-related proteins in brains of AD patients. (a) Protein carbonylation measured by ELISA. AD is 1.90 ± 0.26 × control. N = 20, p = 2.5 × 10⁻⁵. Protein carbonylation was confirmed using the 2,4-dinitrophenylhydrazine Western blot method (Fig. S01l,m). (b,c) Phospho- and total (pan-) ATM measured by ELISA. (b) Total ATM, AD = 0.655 ± 0.0017, Con = 0.752 ± 0.002, p = 0.0019. (c) Ratio of phospho-S1981-ATM/pan-ATM. AD = 0.996 ± 0.044, Con = 0.776 ± 0.058, p = 0.00045. (d–h) DNA repair markers measured by Western blot. (d) C/EBPβ normalized to control, AD = 79.53 ± 6.01, Control = 100 ± 6.40, N = 17, p = 0.027. (e) 14-3-3σ, AD = 85.6 ± 6.6, Con = 126.1 ± 16. N = 10, p = 0.034. (f) phospho-S15 p53 normalized to control, AD = 135.0 ± 12.0, Control = 100 ± 6.6. N = 15, p = 0.017. (g) γH2AX, a marker of DSBs, normalized to control, AD = 171.6 ± 10.3, Control = 100 ± 9.76. N = 20, p = 1.2 × 10^–5. (h) Representative Western blot of γH2AX in AD and control samples. The combined γH2AX results are shown in the scatterplot (f). Western blots were run and developed in parallel under identical conditions. After imaging, blots were stripped and re-stained with histone H3 and actin. Images shown have been cropped for clarity of presentation. Original full-size blots and confirmatory Western blots of the ELISA results are presented in Supplementary Fig. S01a–s. Samples are temporal lobe, Brodmann’s area 38. Ages: AD = 77.2 ± 1.9 yr, Control = 76.3 ± 1.9 yr. PMI: AD = 14.33 ± 0.73 h, Control = 15.05 ± 0.98 h, N = 20, p = 0.56 ($\overline{\text{x}}$ ± SEM).

Figure 3

Changes in ATR in AD. Scatterplots show the levels of (a) ATR, a marker of ssDNA damage, (b) T1989-phosphorylated ATR, and (c) ratio of p-ATR to total ATR. No significant changes are evident in temporal lobe of AD patients. Measured by ELISA. This was confirmed by Western blots (Fig. S02).

Figure 4

Phosphorylated (S15) p53 oligomerization state is changed in AD. Extracts from human AD and control temporal lobe were chromatographed on size-exclusion HPLC and analyzed by p53-pS15 dot blot followed by densitometry. (a) Averaged chromatograms from temporal lobe of five AD and five age-matched controls. Inset shows a chromatogram from a sample that was deliberately concentrated to induce unfolded p53 and aggregation. (b) Ratio of monomer/dimer peak areas calculated from the chromatograms. AD = 1.29 ± 0.27, Con = 0.45 ± 0.06, Con + H₂O₂ = 1.09 ± 0.20, $\overline{\text{x}}$ ± SEM. n = 5, AD vs. Con: p = 0.017, N = 5. (c) Oxidation by 100 nM hydrogen peroxide for 30 min increases p53 monomer/dimer ratio. (d) Ratio of monomer/tetramer peak areas calculated from the chromatograms. AD = 4.17 ± 1.10, Con = 1.22 ± 1.35, Con + H₂O₂ = 17.9 ± 1.0, n = 3, n.s.

Figure 5

STING protein is depleted from Golgi in AD. (a) Representative Western blots of the 52 kDa isoform of STING in enriched Golgi fractions from AD and control temporal lobe, measured by densitometry. STING was 2.82 ± 0.54 × lower in AD patients than controls (AD = 14.3 ± 1.18, control = 40.3 ± 6.9, $\overline{\text{x}}$ ± SEM, n = 10, p = 0.0027). (b) The 30 kDa isoform of STING was moderately increased in cytosol extracted from the same samples (AD = 83.7 ± 16.2, control = 39.4 ± 6.7, n = 13, p = 0.026, $\overline{\text{x}}$ ± SEM). (c) Representative Western blot of 30 kDa isoform of STING protein from isolated Golgi vesicles from AD and control temporal lobe. No statistically significant change was observed. (d) Representative Western blot of 52 kDa isoform of STING protein from isolated nuclei from AD and control temporal lobe. No statistically significant change was observed. (e) Scatterplots of combined data from Western blots of 52 kDa and 30 kDa STING protein in Golgi, cytosol, and nuclei. Western blots in (a–c) and d are different blots. Blots were stripped and re-probed with actin (a–c) or histone H3 (d). Colored MW markers were photographed on the corresponding blot at the time of initial immunostaining. Western blot images shown have been cropped for clarity of presentation. Original full-size blots are presented in Supplementary Fig. S04a–n. Scatterplots represent combined densitometry results from 10 to 13 AD samples and an equal number of age-matched unaffected patients (Con).

Figure 6

Total and membrane-associated interferon α and β measured by ELISA show little or no change in AD compared to control temporal lobe samples. (a,b) Membrane-associated IFNβ is moderately decreased in AD (ratio of particulate/cytosol: AD = 0.592 ± 0.045 pg/mg protein, control = 0.797 ± 0.073, p = 0.030, n = 10/group, $\overline{\text{x}}$ ± SEM) while total IFNβ shows no effect in AD. (c,d) Membrane-associated IFNα and total IFNα do not differ between AD and unaffected patients.

Figure 7

Phosphorylation of IRF3 is decreased in homogenates of AD temporal lobes compared to age-matched controls. (a) Representative Western blot. (b) Scatterplots of all pIRF3 samples measured by densitometry of Western blots. IRF3 phosphorylation (S396) was decreased in AD to 76.5 ± 4.5% of control (n = 18, p = 0.00016). The ratio of pIRF3/total IRF3 was also decreased (AD = 0.582 ± 0.038, control = 1.131 ± 0.152, n = 11, p = 0.0030), suggesting that phosphorylation of IRF3, which is required for formation of the trimeric complex, is inhibited in AD. All three panels are from the same blot. Western blot images shown have been cropped for clarity of presentation. Original full-size blots are presented in Supplementary Fig. S05. After imaging, the pIRF3 blot in (a) was stripped and re-probed for actin, then re-probed for total IRF3. Colored MW markers were photographed on the blot at the time of pIRF3 staining. Scatterplots represent combined densitometry results from 11 to 18 AD samples and an equal number of age-matched unaffected patients (Con).

Figure 8

Role of p53 in DNA repair pathway. The DNA damage repair pathway involves activation of ATR for ssDNA breaks and ATM for dsDNA breaks. These kinases act on CHK1 and CHK2 to phosphorylate p53. Dephosphorylated p53 is ubiquitinated by MDM2 and rapidly degraded at the proteasome. Phosphorylation of p53 at S15 inhibits interaction with MDM2. p53 induces MDM2 transcription, ensuring that the lifetime of p53 is short. Phosphorylation and oligomerization state determine p53’s effect so that p53 acts as a switch to control the cell’s response to DNA damage. The tetramerized form most strongly binds DNA. Many other post-translational modifications of p53 have been described.