Counteracting the effects of TNF receptor-1 has therapeutic potential in Alzheimer's disease

Abstract

Alzheimer's disease (AD) is the most common form of dementia, and neuroinflammation is an important hallmark of the pathogenesis. Tumor necrosis factor (TNF) might be detrimental in AD, though the results coming from clinical trials on anti-TNF inhibitors are inconclusive. TNFR1, one of the TNF signaling receptors, contributes to the pathogenesis of AD by mediating neuronal cell death. The blood-cerebrospinal fluid (CSF) barrier consists of a monolayer of choroid plexus epithelial (CPE) cells, and AD is associated with changes in CPE cell morphology. Here, we report that TNF is the main inflammatory upstream mediator in choroid plexus tissue in AD patients. This was confirmed in two murine AD models: transgenic APP/PS1 mice and intracerebroventricular (icv) AβO injection. TNFR1 contributes to the morphological damage of CPE cells in AD, and TNFR1 abrogation reduces brain inflammation and prevents blood-CSF barrier impairment. In APP/PS1 transgenic mice, TNFR1 deficiency ameliorated amyloidosis. Ultimately, genetic and pharmacological blockage of TNFR1 rescued from the induced cognitive impairments. Our data indicate that TNFR1 is a promising therapeutic target for AD treatment.

Keywords: Alzheimer's disease; TNF receptor 1; blood‐CSF barrier; choroid plexus; therapy.

Figure 1. Ingenuity pathway analysis analysis of microarray results of human choroid plexus of late‐stage AD patients

Ingenuity pathway analysis (IPA) was used to identify the pathways of differentially expressed genes in the choroid plexus of patients with late‐stage Alzheimer's disease (AD) compared to age‐matched healthy controls.

1. Identification of upstream cytokine mediators in the choroid plexus of AD patients. The upstream mediators are ranked according to their z‐score (left axis) and P‐value (right axis).

2. Network of differentially expressed genes downstream of TNF in the choroid plexus of late‐stage AD patients compared to age‐matched healthy controls (red = upregulated, green = downregulated).

3. Venn diagram of the top 20 differentially expressed genes and the differentially expressed genes downstream TNF.

4. Overlay of the dataset on the canonical pathway of the TNF/TNFR1 [red = upregulated (with log ratio), green = downregulated (with log ratio)].

Figure 2. Analysis of NF‐κB‐dependent genes in APP/PS1 mice and in Aβ_1–42 oligomer (AβO)‐injected wild‐type (WT) mice

1. A–D

   Fold change in mRNA gene expression in the choroid plexus and hippocampus determined by qPCR of late‐stage APP/PS1^tg/wt mice compared with age‐matched control mice (A, B) (n = 3–4/group) and of C57BL6/J WT mice 6 h after intracerebroventricular (icv) injection with AβO (1 μg/ml) compared to icv injection with scrambled peptide (C, D) (n = 5–7/group).

Data information: Bars represent mean ± SEM. qPCR was normalized to stable housekeeping genes, determined by GeNorm. Statistics between control mice and AD mice were done with an unpaired t‐test, *0.01 ≤ P < 0.05; **0.001 ≤ P < 0.01; ***0.001 ≤ P < 0.0001, ****P < 0.0001.

Figure 3. TNFR1 deficiency abrogates inflammation in choroid plexus and hippocampus

1. A–H

   Relative mRNA gene expression of Il6, Nos2, Cxcl9, and Lcn2 in the choroid plexus (A–D) and of Il1β, Nos2, Cxcl9, and Tnf in the hippocampus (E–H) of late‐stage AD APP/PS1^tg/wt mice in a TNFR1^+/+ and TNFR1^−/− background compared to age‐matched APP/PS1^wt/wt mice (n = 2–4/group).

2. I–P

   Relative mRNA gene expression of Il6, Nos2, Cxcl9, and Lcn2 in the choroid plexus (I–L) and of Il1β, Nos2, Cxcl9, and Tnf in the hippocampus (M–P) of C57BL/6J TNFR1^+/+ and TNFR1^−/− mice 6 h after intracerebroventricular (icv) injection with scrambled peptide or with AβO (1 μg/ml) (n = 4–5/group).

Data information: Bars represent mean ± SEM. qPCR was normalized to stable housekeeping genes, determined by GeNorm. Statistics were performed with an unpaired t‐test, *0.01 ≤ P < 0.05; **0.001 ≤ P < 0.01; ***0.001 ≤ P < 0.0001, ****P < 0.0001.

Figure 4. TNFR1 deficiency protects against AβO‐induced morphological alterations of the choroid plexus determined by TEM

1. A–D

   Representative conventional transmission electron microscopy (TEM) images of the choroid plexus 6 h after intracerebroventricular (icv) injection of scrambled peptide (control) (A, B) or Aβ_1–42 oligomers (AβO, 1 μg/ml) (C, D) in C57BL/6J TNFR1^+/+ (A, C) and TNFR1^−/− (B, D) mice (n = 2/group). In scrambled‐injected control mice (A, B), the cuboidal structure of the choroid plexus epithelial (CPE) cells is preserved and the microvilli are aligned and structured (insert). The cuboidal structure of the CPE cells in AβO‐injected WT mice (C) is altered, and the microvilli are shortened (insert). In contrast (D), CPE cells of TNFR1^−/− mice icv injected with AβO have a normal morphology and the microvilli are organized (insert). The TEM images were taken at a magnification of 1,000×, and scale bar represents 10 μm; inserts were taken at a magnification of 3,000×, and scale bar represents 2 μm.

Figure 5. TNFR1 deficiency protects against AβO‐induced morphological alterations in the choroid plexus determined by SBF‐SEM

1. A–D

   Representative serial block‐face scanning electron microscopy (SBF‐SEM) images of the choroid plexus 6 h after intracerebroventricular (icv) injection of scrambled peptide (control) (A, B) or Aβ_1–42 oligomers (AβO, 1 μg/ml) (C, D) in C57BL/6J TNFR1^+/+ (A, C) and TNFR1^−/− mice (B, D) (n = 2/group), derived from Movies EV1, EV2, EV3 and EV4. CSF: cerebrospinal fluid; Mv, microvilli; Nu, nucleus. Scale bar, 5 μm.

Figure EV1. TNFR1 deficiency protects against morphological alterations in the choroid plexus of APP/PS1^tg/wt mice determined by TEM

1. A–D

   Representative conventional transmission electron microscopy (TEM) images of the choroid plexus of 18‐week‐old C57BL/6J APP/PS1^tg/wt mice in a TNFR1^+/+ and TNFR1^−/− background compared to age‐matched non‐transgenic controls. In non‐transgenic controls (A, B), the cuboidal structure of the choroid plexus epithelial (CPE) cells is maintained. The nuclei have regular shapes and mitochondria look normal (zoom). The loss of the cuboidal shape is more enhanced in CPE cells of APP/PS1^tg/wt TNFR1^+/+ mice (C). The capillaries are swollen and filled with plenty of red blood cells, the nuclei have irregular shapes, and some CPE cells are at a degenerative state (zoom). In contrast (D), CPE cells of APP/PS1^tg/wtTNFR1^−/− mice have the same cellular shape as non‐transgenic littermates. The capillaries are less swollen, and the mitochondria and nuclear shape are normal (zoom). The TEM images were taken at a magnification of 1,000×, scale bar represents 10 μm; zooms were taken at a magnification of 3,000×, scale bar represents 2 μm.

Figure EV2. TNFR1 deficiency protects morphological alterations in the choroid plexus of APP/PS1^tg/wt mice determined by SBF‐SEM

1. A–D

   Representative serial block‐face scanning electron microscopy (SBF‐SEM) images of the choroid plexus of 18‐week‐old C57BL/6J APP/PS1^tg/wt mice in a TNFR1^+/+ (A, C) and TNFR1^−/− (B, D) background compared to age‐matched non‐transgenic controls (n = 1/group), derived from Movies EV5, EV6, EV7 and EV8. CSF, cerebrospinal fluid; Mv, microvilli; Nu, nucleus. Scale bar, 5 μm.

Figure 6. TNFR1 deficiency protects the blood–cerebrospinal fluid (CSF) barrier by preventing MMP increase and preserving tight junctions

1. A

   Blood–CSF barrier permeability was determined by measuring leakage of FITC‐labeled dextran in the CSF of C57BL/6J TNFR1^+/+ and TNFR1^−/− mice 6 h after intracerebroventricular (icv) injection of either scrambled peptide or Aβ_1–42 oligomer (AβO 1 μg/ml) (n = 11–14/group).

2. B–E

   Relative mRNA gene expression of Mmp8, Mmp3, Cldn5, and Ocln in choroid plexus of TNFR1^+/+ and TNFR1^−/− mice 6 h after icv injection of 1 μg/ml AβO (n = 5–6/group).

3. F

   Representative images of CLDN1 staining in choroid plexus of the fourth ventricle of TNFR1^+/+ (left and middle image) and TNFR1^−/− mice (right image) 6 h after icv injection with either 1 μg/ml AβO (middle and right image) or scrambled peptide (left image). Arrows indicate preserved CLDN1 tight junctions, and arrowheads indicate affected CLDN1 tight junctions (n = 3/group). Scale bar represents 15 μm.

Data information: Bars represent mean ± SEM. qPCR was normalized to stable housekeeping genes determined by GeNorm. The experiments are done in duplicates, and pooled results are shown in (A) and representative results in (B–E). Statistics were performed with a one‐way ANOVA for the permeability data (A) and an unpaired t‐test for the qPCRs (B–E), *0.01 ≤ P < 0.05; **0.001 ≤ P < 0.01; ***0.001 ≤ P < 0.0001, ****P < 0.0001.

Figure 7. TNFR1 deficiency reduces Aβ pathology in APP/PS1 mice

1. A–D

   Brain sections of late‐stage C57BL/6J APP/PS1^tg/wt and APP/PS1^tg/wtTNFR1^−/− mice were stained with Thioflavin‐S to detect Aβ disposition in the whole brain. The amount of plaques was quantified (A), and a morphometric analysis was performed (B) (n = 5 and 6 mice/group, respectively). Representative images (C, D) of Thioflavin‐S (ThioS) staining of the brain containing the hippocampus of late‐stage APP/PS1^tg/wt and age‐matched APP/PS1^tg/wtTNFR1^−/− mice (scale bar represents 200 μm).

2. E–H

   ELISA analysis of soluble and insoluble Aβ_1–40 and Aβ_1–42 in the cortex of late‐stage APP/PS1^tg/wt in a TNFR1^+/+ and TNFR1^−/− background compared to age‐matched controls (n = 8, 9, and 3 mice, respectively).

3. I

   Relative mRNA expression of Bace1 in the hippocampus of late‐stage APP/PS1^tg/wt mice in a TNFR1^+/+ and TNFR1^−/− background compared to age‐matched APP/PS1^wt/wt mice (n = 5/group).

Data information: Bars represent mean ± SEM. qPCR was normalized to stable housekeeping genes determined by GeNorm. Statistics were performed with an unpaired t‐test (A), a two‐way ANOVA assay (B) or a one‐way ANOVA (E–I), *0.01 ≤ P < 0.05; ***0.001 ≤ P < 0.0001; ****P < 0.0001.

Figure 8. TNFR1 deficiency prevents microglia activation in APP/PS1 mice and upon icv AβO injection

1. A, B

   IBA1 staining for microglia on whole‐brain sections of late‐stage C57BL/6J APP/PS1^tg/wt in a TNFR1^+/+ and TNFR1^−/− background compared to age‐matched controls. (A) Representative images of a region around the fourth ventricle (microglia indicated with arrowheads) in age‐matched controls (upper panel, n = 3), APP/PS1^tg/wtTNFR1^+/+ (middle panel, n = 4), and APP/PS1^tg/wtTNFR1^−/− (lower panel, n = 6) mice. (B) Quantification of IBA1⁺ cell count (determined by brown staining).

2. C, D

   IBA1 staining for microglia on whole‐brain sections of C57BL/6J TNFR1^+/+ mice 6 h after intracerebroventricular (icv) injection with either scrambled peptide or Aβ_1–42 oligomer (AβO, 1 μg/ml) or TNFR1^−/− mice icv injected with AβO. (C) Representative images of a region around the fourth ventricle (microglia indicated with arrowheads) in scrambled‐injected mice (upper panel, n = 2) and in AβO‐injected TNFR1^+/+ (middle panel, n = 3) and TNFR1^−/− mice (lower panel, n = 3). (D) Quantification of IBA1⁺ cell count (determined by brown staining).

Data information: Scale bars represent 100 μm and insert 20 μm. Bars represent mean ± SEM. Statistics were performed with a one‐way ANOVA assay, *0.01 ≤ P < 0.05; **0.001 ≤ P < 0.01.

Figure 9. Genetic or therapeutic blockage of TNFR1 prevents cognitive decline in APP/PS1^tg/wt mice and upon icv AβO injection

Analysis of short‐term memory (STM) and long‐term memory (LTM) evaluated by the novel object recognition (NOR) test.

1. A, B

   STM (A) and LTM (B) assessed in C57BL/6J APP/PS1^wt/wtTNFR1^+/+ mice (n = 19) and APP/PS1^tg/wt mice in a TNFR1^+/+ (n = 10) and TNFR1^−/− background (n = 11) aged 30–34 weeks.

2. C, D

   STM (C) and LTM (D) assessed 24 h after intracerebroventricular (icv) injection of either scrambled peptide or Aβ_1–42 oligomer (AβO, 1 μg/ml) in C57BL/6J TNFR1^+/+ or TNFR1^−/− mice (n = 6–11/group).

3. E, F

   STM (E) and LTM (F) assessed in human TNFR1 transgenic (hTNFR1 Tg) mice 24 h after icv injection of scrambled peptide, AβO alone (1 μg/ml), or AβO combined with a trivalent anti‐TNFR1 Nanobody (TROS, 1.55 μg/μl) (n = 6–8/group).

Data information: Min‐to‐max box plots represent median with the 25 and 75% percentiles, and whiskers are the min and max values ± SEM. The mean values are indicated above the graphs. All experiments were done in duplicates, and pooled results are shown. Statistics were performed with a one‐way ANOVA assay, *0.01 ≤ P < 0.05; **0.001 ≤ P < 0.01; ****P < 0.0001.

Figure EV3. Icv injection of TROS in WT mice does not affect inflammation induced by Aβ_1–42 oligomers (AβO)

1. A–E

   Relative mRNA gene expression of Il1β, Tnf, Il6, Mmp3, and Tnfrsf1a of C57BL/6J wild‐type (WT) mice 6 h after intracerebroventricular (icv) injection with scrambled peptide or with AβO (1 μg/ml) together with PBS or with TROS 1.55 μg/μl (n = 6/group). Bars represent mean ± SEM. qPCR was normalized to stable housekeeping genes determined by GeNorm. Statistics were performed with an unpaired t‐test, **0.001 ≤ P < 0.01; ***0.001 ≤ P < 0.0001; ****P < 0.0001.

Figure 10. Schematic hypothesis of the roles of TNFR1 in the pathology of Alzheimer's disease

Amyloid precursor protein (APP) is cleaved into pathogenic amyloid‐beta 1–42 (Aβ_1–42) species by BACE1, the expression of which can be activated by TNFR1‐induced NF‐κB in neurons of the hippocampus. Choroid plexus epithelial (CPE) cells are tightly connected with tight junctions. The presence of Aβ oligomers (AβO) in the CSF leads, via TNFR1, to induction of chemo‐ and cytokines (e.g., TNF) by CPE cells and other neuronal cells such as microglia and neurons. Upon AβO exposure, CPE cells also express matrix metalloproteinases (MMPs), which disrupt tight junctions (right lower panel). Eventually, this results in leakage of the blood–CSF barrier. In the brain parenchyma, TNF triggers TNFR1 molecules on the surface of microglia and consequently induces its own release. Additionally, TNF also promotes neuronal death via TNFR1. Ultimately, all these events lead to further aggravation of neuroinflammation and to cognitive impairment.