Fig 2: BACE1, PHF-tau, GFAP, and CLDN5 immunoreactivities in the CA1 area of dementia brains. (A) Representative image of location in human Control hippocampus of area ca1. Magnification: 10×. Scale bar: 25 µm. (B) Representative images of the BACE1, (C) PHF-tau, (D) GFAP, and (E) CLDN5 immunoreactivities in the CA1 area of human hippocampal tissue. Magnification: 10×. Scale bar: 50 µm. Inset: 25 µm scale bar. The values in the bar graph are expressed as a densitometric percentage of the BACE1 IR in the CA1 area. FAD, familial-type Alzheimer’s disease (presenilin 1 mutation E280A); SAD, sporadic Alzheimer’s disease; CADASIL, autosomal dominant cerebral arteriopathy with subcortical infarcts and leukoencephalopathy. The data are expressed as the means ± SEM. n = 4. *p < 0.05; *** p < 0.001; ****p < 0.0001.

Fig 3: Lack of IL12p40 but does not affect plaque-associated BACE1 immunoreactivity or gliosis in female APP23 mice Histological analysis of plaque-associated BACE1 immunoreactivity in female APP23 (n = 8) and APP23p40-/- (n = 10) mice. BACE1 area covered was normalised to 4G8-positive area covered of the same image (left). Right: representative images, scale bar = 50 µm. Mean ± SEM, statistical analysis: two-tailed unpaired t-test, P = 0.5402.Stereological quantification of the number of cortical GFAP-positive astrocytes in female APP23 (n = 8) and APP23p40-/- (n = 8) mice (left). Right: representative images of GFAP staining, scale bar = 200 µm. Mean ± SEM, statistical analysis: two-tailed unpaired t-test, P = 0.1579.Quantification of activated microglia within 30 µm from plaque borders. Top: numbers of Iba1-positive microglia were normalised to the size of the nearest 4G8-positive plaque and quantified in female APP23 (n = 8) and APP23p40-/- (n = 10) mice. Mean ± SEM, statistical analysis: two-tailed unpaired t-test, P = 0.8240. Bottom: Histogram representing Clec7a staining intensity within plaque-associated Iba1-positive microglia in female APP23 (n = 8) and APP23p40-/- (n = 10) mice. Mean ± SEM, statistical analysis: two-tailed unpaired t-test with Bonferroni correction for each single bin, P = N.S.. Right: representative images, scale bar = 40 µm.Radial intensity profiles of Iba1 and 4G8 around the centre of the nucleus of plaque-associated Iba1-positive microglia in female APP23 (n = 8) and APP23p40-/- (n = 10) mice. Iba1 intensity declines until a radius of ˜6 µm, marking the cell periphery. 4G8 intensity peaks inside the cell (˜4 µm), but stays high outside the cell. This is very likely due to the close proximity to 4G8-positive plaques. Mean ± SEM, statistical analysis: two-tailed unpaired t-test with Bonferroni correction for the number of binned radii shows no significant difference between both 4G8 traces. Source data are available online for this figure.

Fig 4: Brain safety of intravenously injected NLCs and NLC-siRNA complexes. The images show haematoxylin and eosin stained sections of cerebral cortex at 4 h, 24 h and 72 h after administration of saline (a), FAM-CGY peptide (10 µM) (b), FAM-CGY/BACE1-specific siRNA (c) and FAM-CGY/Cy5-siRNA complexes (d). In (c) and (d) FAM-CGY and siRNA concentrations were 10 µM and 48 nM, respectively. Magnified regions of box inserts are shown. E = cerebral endothelial cells, G = glial cells, M = microglial cells and N = Neurons. Scale bar = 100 µm. ELISA determination of pro-inflammatory markers TNF-a (e), IL-6 (f) and IL-1ß (g) in hippocampus on NLC and NLC-siRNA treatments compared with control (p > 0.05 in all cases; n = 3 determinations). Determination of microglial marker Iba-1 (h) and astrocyte marker GFAP (i) in hippocampus on NLC treatment (p > 0.05 in all cases; n = 3 determinations). Lipopolysaccharide (LPS) injection induced a small, but significant (*p < 0.05 in all cases compared with control and NLC treatments; n = 3 determinations) proinflammatory response at 24 h sacrifice point. All statistical analyses were performed with one-way ANOVA, using Tukey’s multiple comparison correction to calculate significance. Source data are available in the Source Data file

Fig 5: In male APP23p40-/- mice, Aß processing is unchanged compared to APP23 mice, while serum IFN? levels are reduced AWestern blot analysis of APP levels in SDS-soluble protein homogenates in male APP23 (n = 7) and APP23p40-/- (n = 7) mice. APP expression levels were normalised to ß-Actin. Mean ± SEM, statistical analysis: two-tailed unpaired t-test, P = 0.6702.BWestern blot analysis of insulin-degrading enzyme (IDE) (P = 0.9237), Neprilysin (Nep) (P = 0.7154) and BACE1 (P = 0.6421) levels in Triton-X-soluble protein homogenates in male APP23 (n = 4) and APP23p40-/- (n = 4) mice. Protein expression levels were normalised to GAPDH. Samples not showing a positive signal for GAPDH due to low protein content were excluded from analysis. Mean ± SEM, statistical analysis: two-tailed unpaired t-test.C–KV-PLEX analysis for (C) IFN? brain (P = 0.7587, plasma ***P = < 0.0001), (D) IL-10 (brain P = 0.0882, plasma P = 0.3591), (E) IL-1ß (brain P = 0.9520, plasma P = 0.6272), (F) IL-2 (brain P = 0.6889, plasma P = 0.3795), (G) IL-4 (brain P = 0.9515, plasma values undetected in APP23p40-/- group, P-value does not apply), (H) IL-5 (brain P = 0.9591, plasma P = 0.0958), (I) IL-6 (brain P = 0.6330, plasma P = 0.4225), (J) TNF-a (brain P = 0.1194, plasma P = 0.2246) and (K) CXCL1 (brain P = 0.5305, plasma P = 0.8743) protein levels in the TBS fraction of brain homogenates and plasma samples from male APP23 (brain n = 10, plasma n = 11–12) and APP23p40-/- (brain n = 8, plasma n = 7–8) mice. Total protein concentration of each sample was used as an internal reference and some plasma values removed based on the Grubbs's outlier test. Mean ± SEM, statistical analysis: two-tailed unpaired t-tests between brain and serum respectively. Source data are available online for this figure.