The intracellular domain of CX3CL1 regulates adult neurogenesis and Alzheimer's amyloid pathology

Abstract

The membrane-anchored CX3CL1 is best known to exert its signaling function through binding its receptor CX3CR1. This study demonstrates a novel function that CX3CL1 exerts. CX3CL1 is sequentially cleaved by α-, β-, and γ-secretase, and the released CX3CL1 intracellular domain (CX3CL1-ICD) would translocate into the cell nucleus to alter gene expression due to this back-signaling function. Amyloid deposition and neuronal loss were significantly reduced when membrane-anchored CX3CL1 C-terminal fragment (CX3CL1-ct) was overexpressed in Alzheimer's 5xFAD mouse model. The reversal of neuronal loss in 5xFAD can be attributed to increased neurogenesis by CX3CL1-ICD, as revealed by morphological and unbiased RNA-sequencing analyses. Mechanistically, this CX3CL1 back-signal likely enhances developmental and adult neurogenesis through the TGFβ2/3-Smad2/3 pathway and other genes important for neurogenesis. Induction of CX3CL1 back-signaling may not only be a promising novel mechanism to replenish neuronal loss but also for reducing amyloid deposition for Alzheimer's treatment.

Figure 1.

Cleavage of CX3CL1 by β- and γ-secretases. (A and B) Myc-tagged CX3CL1 was expressed in BACE1-overexpressing HM cells in a dose-dependent manner. Two clonal stable cell lines (c1 and c2) were also selected from mouse neuroblastoma Neuro-2A cells transfected with myc-tagged CX3CL1. The myc antibody detected three C-terminal fragments: β and β′ fragments for BACE1 cleavages and the α-secretase–cleaved fragment. The arrowhead indicates nonspecific reacted bands with the myc antibody. (B) Both BACE1 and BACE2 similarly cleave CX3CL1, as demonstrated by coexpression of indicated plasmid DNA in HEK-293 cells. Inhibition of BACE activity by compound BACE IV diminished the levels of β and β′ fragments. (C) CX3CL1 was cleaved by γ-secretase, shown by inhibition of γ-secretase with the specific inhibitor DAPT to increase levels of fragments from BACE1 and α-secretase cleavages.

Figure 2.

Expression of CX3CL1-ct in cells. (A) Illustration of CX3CL1-ct construction and a potentially cleaved fragment by γ-secretase. (B) CX3CL1 or CX3CL1-ct constructs were expressed in BACE1-overexpressing cells, and all C-terminal fragments were detected in cytosolic fractions. The arrowhead indicates nonspecific reacted bands with the myc-antibody. m indicates the monomer of CX3CL1; d indicates a band that is presumably a dimer of CX3CL1-ct based on the molecular weight. (C) Cytosolic and nuclear fractions were separated from cells expressing either CX3CL1 or CX3CL1-ct. The γ-cleaved CX3CL1 fragment CX3CL1-cγ was detectable. β-Actin was detected mainly from the cytosol fractions, while histone H3 was enriched in the nuclear fractions. Nonspecific bands are indicated by arrowheads. (D) After transfection for 24 h, CX3CL1-ICD peptides Pep-34 and Pep-36, detected by CX3CL1 C-terminal antibody in green, were localized in mouse neuro-2a cell nucleus, marked by Topro in blue. Scale bar, 5 µm.

Figure 3.

Overexpression of CX3CL1-ct in 5xFAD transgenic mice reduces amyloid deposition. (A) Two different pairs of Tg-5xFAD and Tg-CX3CL1-ct/tTA/5xFAD mice were used for comparing the density of amyloid plaques by confocal or DAB staining. Antibody 6E10 recognizes human Aβ peptides and amyloid plaques, while HA-tagged CX3CL1-ct transgene was detected with HA antibody. Cell nucleus was marked by Topro. Scale bars, 200 µm. (B) Densities of amyloid plaques from subiculum or cortex were quantified from six mice in each pair. Both male and female mice were used for comparisons. Error bars are ± SEM. (C) APP and its cleavage product C99 and C83 were examined by antibody 8717 (n = 3 experiments). **, P < 0.01; ***, P < 0.001.

Figure 4.

Ectopic expression of CX3CL1-ct in 5xFAD mice reverses neuronal losses. (A and B) Neuronal loss was observed in the 5xFAD subiculum (A) and cortical layers (B) compared with WT controls. Higher neuronal density, marked by NeuN antibody, was noted in mice overexpressing CX3CL1-ct, suggesting a potential increase in neurogenesis. Ectopic expression of CX3CL1-ct in 5xFAD brains showed more neurons compared with 5xFAD comparable regions. Scale bar, 30 µm. (C) NeuN-marked neurons in the subiculum region were counted for comparison, one in every 10 sections per mouse. Each dot represents one animal (n = 6; **, P < 0.01; ***, P < 0.001, Student’s t test); error bars are ± SEM.

Figure 5.

Enhanced neurogenesis by CX3CL1-ct in transgenic mice. (A) Immunostaining of brain sections from BrdU pulse-labeled mice at P21 was performed using BrdU antibody. Scale bar, 30 µm. (B) Stereological quantification confirmed enhanced neurogenesis in the neurogenic niche of the SGZ of Tg-CX3cl1-ct/tTA mice. (C) Mice were BrdU-labeled starting at P11 for 5 d and then examined at 6 wk of age. Brain sections were fixed and costained with BrdU (red) and NeuN (green) antibodies. Scale bar, 30 μm. (D) Quantification of both BrdU-positive cells and mature neurons in granular cell layers in Tg-CX3cl1-ct/tTA mice and WT control littermates was conducted, and the results were plotted (n = 6 pairs; **, P < 0.01, Student’s t test). (E and F) Newly dividing cells in the SVZ were labeled for BrdU, marked with circles (E), and quantitatively compared from one in every six fixed sections (F; n = 4; ***, P < 0.001, Student’s t test). Error bars are ± SEM.

Figure 6.

CX3CL1-ct enhances neurogenesis in adult transgenic mice. (A) Tg-CX3CL1-ct/tTA mice that were treated with Dox for 2 mo (2M) and left untreated for another 4 mo (4M) were pulse labeled for examination of adult neurogenesis. Fixed brain sections were reacted with antibody for BrdU and Topro for the nucleus. (B) Semiquantification was conducted using one in every six sections to compare BrdU-positive cells in the Tg-CX3CL1-ct/tTA SGZ with that in WT (n = 4; ***, P < 0.001, Student’s t test). Each animal is shown in one dot. Error bars are ± SEM. (C) BrdU⁺ cells in Tg-CX3CL1-ct/tTA SVZ were significantly increased compared with WT SVZ. Scale bars, 60 µm.

Figure 7.

Molecular signatures of newborn mice in response to CX3CL1-ct expression. (A and B) Hippocampi from newborn pups were used for RNA-seq analyses; heatmap shows significant changes (P < 0.05) in genes between WT and Tg-CX3C1-ct/tTA (CX3) samples (A), and between Tg-Camk2a-tTA (CAMK) and Tg-CX3C1-ct/tTA samples (B). (C) Overlap statistical analysis uncovered similar transcriptional alterations shown in Fig. 8, A and B and revealed 233 up-regulated and 163 down-regulated genes in Tg-CX3C1-ct/tTA samples compared with both WT and Tg Camk2a tTA. (D) Among 233 commonly up-regulated genes, the top biological pathways indicates the control of system and multicellular organismal development. (E) TGFβ3 appears to be one of the top genes in the protein functional network involving system development. (F) Among 233 commonly up-regulated genes, ∼50 genes are important for the control of neurogenesis, including those well-documented genes highlighted by arrows. (G) Selected genes with significant changes (P < 0.001) between Tg-tTA and Tg-CX3C1-ct/tTA are listed, and some of them are known to control neurogenesis.

Figure 8.

Increased TGFβ signaling and neurogenesis by expressing CX3CL1-ct in adult mice. (A) Protein levels were compared between transgenic mice (Tg-CX3cl1-ct/tTA) and their control littermates (WT, Tg-Cx3CL1-ct and Tg-CamKIIa-tTA) using the indicated antibodies. Expression of transgene CX3CL1-ct was validated by HA and CX3CL1 antibody in compound mice. BACE1-cleaved C-terminal fragment was marked by β, and ADAM-cleaved fragment, by α. (B) Bar graphs confirm that neuronal expression of CX3CL1-ct significantly increases levels of TGFβ2, TGFβ3, p-Smad2, and p-Smad3 compared with controls (n = 3 experiments; **, P < 0.01; ***, P < 0.001, Student’s t test; error bars are ± SEM).