Fig 2: ATAD3A and CYP46A1 cooperate to promote APP processing.a Brain sections from 3-month-old mice of the indicated genotypes were stained with anti-CTxB (green) and anti-APP (red) antibodies. The intensity of CTxB in the mouse cortex was quantified from three separate fields of each mouse and shown in the histogram (n = 3 mice/group). b Brain sections from 9-month-old mice of the indicated treatment groups were stained with anti-CTxB (green) and anti-APP (red) antibodies. The intensity of CTxB in the mouse cortex was quantified from three separate fields of each mouse and shown in the histogram (n = 4 mice/group). c Stable APPwt-expressing Neuro2a (N2a) cells were transfected with ATAD3A-WT-Flag, ATAD3A-?N50-Flag, ATAD3A-?CC-Flag, or control vector for 24 h, followed by treatment with DA1 or control peptide TAT for 48 h (1 µM, each). WB was performed. Histogram: the densities of C99 and APP relative to actin (n = 4 independent experiments). d Total brain lysates were harvested from the cortex of 8-month-old 5XFADhet;CMV;ATAD3Afl/+ group or 9-month-old mice treated with DA1. The APP proteolytic product C99 was assessed by WB (Indicated by red arrow). Histogram: the density of C99 relative to actin (n = 6 mice for DA1 treated groups, n = 9 mice for 5XFADhet;CMV;ATAD3Afl/+ groups). e CYP46A1 was overexpressed in stable APPwt- and APPswe-expressing Neuro2a cells, which were then treated with DA1 or control peptide TAT (1 µM for 3 days). WB was performed. Ctl and CYP on the images indicate Control Vector and CYP46A1, respectively. The relative densities of the ATAD3A oligomers and C99 were quantified (n = 5). Representative images and blots from at least three independent experiments are shown. All data are presented as the mean ± SEM. The data in a and b were compared by the unpaired Student’s t test (two-tailed), and the data in c–e were compared by one-way ANOVA with Tukey’s multiple comparisons test.

Fig 3: ATAD3A oligomerization inhibits cholesterol turnover in AD models.a The cortex of 3-month-old mice (n = 3 mice/group) was subjected to Label-free tandem mass spectrometry. Left: the number of proteins changed in mice. Middle: KEGG database analysis on proteins from the pool marked as *. Right: GO biological pathway analysis of the proteins enriched in the “metabolic pathway”. Heat map: proteins involved in the “lipid metabolic process”. Brain sections from b 8-month-old mice and d 6-month-old mice were stained with anti-CYP46A1 and anti-NeuN antibodies. The intensity of CYP46A1 in NeuN+ cells from mouse cortex was quantified from three separate fields per mouse in c (WT and 5XFADhet;ATAD3A+/+: n = 4; CMV;ATAD3Afl/+ and 5XFADhet;CMV;ATAD3Afl/+: n = 3) and e (n = 6 mice/group). Scale bar: 20 µm. The total cholesterol content was measured in f the cortex from 8-month-old mice and g 9-month-old mice (WT: n = 8; CMV;ATAD3Afl/+: n = 6; 5XFADhet;ATAD3A+/+: n = 8 and 5XFADhet;CMV;ATAD3Afl/+: n = 8. WT/TAT: n = 10; the rest of groups: n = 9). h Left: Brain sections from 9-month-old mice were stained with the filipin probe. The immunodensity of filipin+ cholesterol was quantified from three separate fields per mouse (n = 3 mice/group). Right: Stable APPwt- and APPswe-expressing Neuro2a cells were treated with DA1 or TAT (1 µM/day for 4 days) and stained with the filipin probe. The intensity of filipin+ cholesterol was quantified from five separate fields per sample (n = 5 independent biological experiments). Representative images are shown in Supplementary Fig. 6c. i Plasma 24-OHC levels from 9-month-old mice were measured (WT/TAT: n = 13; WT/DA1: n = 9; 5XFADhomo/TAT: n = 12; 5XFADhomo/DA1: n = 11). j CYP46A1 mRNA was measured by qPCR (n = 6). k Left: HT-22 cells were treated with TAT or DA1 (1 µM) and Aß1–42 peptides (5 µM, 9 h). Right: Stable APPwt-expressing Neuro2a cells were treated with TAT or DA1. CYP46A1 mRNA was measured by qPCR (n = 3). l Total cholesterol content was measured (n = 6). m The concentration of 24-OHC was measured (n = 6). All data are presented as mean ± SEM from at least three independent experiments and compared by the unpaired two-tailed Student’s t test (k-right) or one-way ANOVA with Tukey’s multiple comparisons test (c–m).

Fig 4: Suppression of ATAD3A oligomerization by DA1 is neuroprotective in AD mice.5XFAD and age-matched WT mice were subcutaneously treated with either TAT or DA1 beginning at six weeks of age (1 mg/kg/day) using an Alzet pump (treatment timeline shown in Supplementary Fig. 5b). a Total protein lysates were harvested from 6-month-old mouse brains, and ATAD3A oligomerization was analyzed by WB under non-reducing conditions (ßME: ß-mercaptaethanol). Histogram: the density of ATAD3A oligomers relative to total ATAD3A protein levels (n = 8 mice/group). b Brain sections from 6-month-old mice were stained with anti-SigmaR1 and anti-VDAC or anti-ATAD3A and anti-FACL4 antibodies and then subjected to PLA analysis. Histogram: the number of PLA-positive puncta (red) (n = 4 mice/group). c The Y-maze test was administered to 6-month-old mice (WT/TAT and 5XFADhomo/TAT: n = 26 mice/group; n = 23 mice for the WT/DA1 group and n = 24 mice for the 5XFADhomo/DA1 group). d The Barnes maze test was administered to 8-month-old mice (WT/TAT: n = 16 mice/group; WT/DA1: n = 14 mice/group; 5XFADhomo/TAT: n = 13 mice/group; 5XFADhomo/DA1: n = 13/12 mice/group in day5/ day12 test). e Brain sections from 6-month-old mice of the indicated groups were stained with an anti-6E10 antibody. The area covered by 6E10+ Aß plaques and the immunodensity of 6E10 in cortex were quantified from three separate fields of each mice (n = 4 mice/group). f Brain sections from 6-month-old mice of the indicated groups were stained with anti-Iba1 (green) and anti-GFAP (red) antibodies. The intensity of Iba1 and GFAP was quantified from four separate fields of each mice (n = 4 mice/group). Scale bar: 50 µm. g Brain sections from 6-month-old mice were stained with Jade C probe (green) and anti-NeuN antibody (red). The Jade C intensity in NeuN+ cells in the cortex and hippocampus was quantified and shown in the histograms (n = 3 mice/group). Scale bar: 20 µm. Representative images and blots are at least three independent experiments are shown. All data are presented as the mean ± SEM and were compared by one-way ANOVA with Tukey’s multiple comparisons test.

Fig 5: ATAD3A oligomerization impairs MAM integrity under AD conditions.a Brain sections from 3-month-old WT and 5XFAD mice (n = 4 mice/group) and b postmortem cortex sections from normal subjects (Nor) and AD patients (n = 3 subjects/group) were stained with anti-SigmaR1 and anti-VADC antibodies or anti-ATAD3A and anti-FACL4 antibodies.Human subject information is presented in Supplementary Fig. 2b. The In situ Duolink proximity ligation assay (PLA) was performed. Histogram: the number and size of PLA-positive puncta (red). The number and size of PLA puncta signals were quantified from six separate fields of each sample. c HT-22 cells were treated with oligomeric Aß1–42 peptides (5 µM) for 18 h. Control (Con): n = 6; Aß1–42: n = 5. d HT-22 cells were infected with control or ATAD3A shRNA lentivirus and then treated with oligomeric Aß1–42 peptides (5 µM, 18 h). Cells were stained with the indicated antibodies and subjected to PLA analysis. Histogram: the number and size of PLA-positive puncta (red). n = 10 for shATAD3A-1/Aß1–42 group, n = 8 for the rest of three groups. At least 200 cells/group were analyzed. e Total protein lysates from the indicated groups were analyzed by WB. Histograms: the densities of FACL4 and IP3R3 relative to actin. n = 4 for Control group (NC) and n = 7 for shATAD3A group. f HT-22 cells were transfected with the indicated plasmids for 48 h. Cells were stained with the indicated antibodies and then subjected to PLA analysis. Histogram: the number and size of PLA-positive puncta (red). n = 10 for SigmaR1/VDAC group and n = 7 for IP3R3/VADAC group. At least 200 cells/group were analyzed. Scale bar: 10 µm. The data are presented as the mean ± SEM. Representative images and blots from at least three independent experiments are shown. The data in a–c were compared by the unpaired Student’s t-test (two-tailed), and the data in d–f were compared by one-way ANOVA with Tukey’s multiple comparisons test.