ApoE4 disrupts interaction of sortilin with fatty acid-binding protein 7 essential to promote lipid signaling

Abstract

Sortilin is a neuronal receptor for apolipoprotein E (apoE). Sortilin-dependent uptake of lipidated apoE promotes conversion of polyunsaturated fatty acids (PUFA) into neuromodulators that induce anti-inflammatory gene expression in the brain. This neuroprotective pathway works with the apoE3 variant but is lost with the apoE4 variant, the main risk factor for Alzheimer's disease (AD). Here, we elucidated steps in cellular handling of lipids through sortilin, and why they are disrupted by apoE4. Combining unbiased proteome screens with analyses in mouse models, we uncover interaction of sortilin with fatty acid-binding protein 7 (FABP7), the intracellular carrier for PUFA in the brain. In the presence of apoE3, sortilin promotes functional expression of FABP7 and its ability to elicit lipid-dependent gene transcription. By contrast, apoE4 binding blocks sortilin-mediated sorting, causing catabolism of FABP7 and impairing lipid signaling. Reduced FABP7 levels in the brain of AD patients expressing apoE4 substantiate the relevance of these interactions for neuronal lipid homeostasis. Taken together, we document interaction of sortilin with mediators of extracellular and intracellular lipid transport that provides a mechanistic explanation for loss of a neuroprotective lipid metabolism in AD.

Keywords: Alzheimer's disease; Fatty acid binding protein; Polyunsaturated fatty acid; Protein sorting; VPS10P domain receptor.

Fig. 1.

Surface proteome analysis identifies fatty acid-binding protein 7 as novel sortilin target in neurons but not astrocytes. (A) Workflow of surface proteome analysis in primary neurons from wild-type (WT) and Sort1^−/− (KO) mice. (B) Comparison of the surface proteomes of WT and KO primary neurons (n=3 biological replicates/group; two technical replicates per biological replicate). Dashed lines show threshold values for log2 (fold change) and −log10 (P-value), ±0.3 and 1.3, respectively. Proteins with increased (green) or decreased (red) levels in the surface proteome of KO neurons are color-coded. Selected proteins with altered surface exposure in KO neurons are highlighted (see Table S1 for details). (C) Data as in B but comparing the surface proteomes of WT and KO primary astrocytes (n=6 biological replicates/group). Selected proteins with altered surface exposure in KO astrocytes are highlighted (see Table S1 for details).

Fig. 2.

Sortilin and FABP7 interact in cells. (A) Chinese hamster ovary (CHO) cells stably overexpressing sortilin (CHO-S) or sortilin and FABP7 tagged with a Myc epitope (CHO-S/F) were used for co-immunoprecipitation (IP) experiments. Expression of sortilin and FABP7–Myc in total cell lysates of both cell lines is shown in the panel Input (15 μg of total cell lysate). In panel IP-FABP7, co-immunoprecipitation of sortilin with anti-Myc affinity resin (+) is seen in lysates from CHO-S/F but not from CHO-S cells. No immunoprecipitation of sortilin with anti-Myc affinity resin is seen in CHO-S cells, or in CHO-S/F in the absence of the affinity resin (−). (B) Co-immunoprecipitation of sortilin with FABP7-myc from CHO-S/F cells as described in A. Prior to immunoprecipitation with anti-Myc affinity resin, cells were treated with conditioned medium from HEK293 cells containing 5 µg/ml of apoE3 (+E3) or apoE4 (+E4; see Materials and Methods for details) or blank medium (−) for 24 h. (C) Proximity ligation assay (PLA) to assess close spatial proximity of sortilin and FABP7 in CHO cells. Primary antibodies were directed against sortilin or the Myc epitope in FABP7–Myc. Close proximity is detected in CHO cells expressing both sortilin and FABP7 (red signal). No PLA signal is seen in cells expressing sortilin only. Cell nuclei were counterstained with DAPI (blue). The inset shows a higher magnification image of cells positive for PLA signals for sortilin and FABP7–Myc. Scale bar: 50 µm. (D) Subcellular fractionation of CHO-S/F cells using gradient ultracentrifugation. Fractions were identified based on markers for endoplasmic reticulum (ER; GRP78), plasma membrane (PM; β-integrin), trans-Golgi network (TGN; VTI1b), and early endosomes (Endo; EEA1). FABP7 colocalizes with sortilin to the PM (fractions 3–4), TGN (fractions 7–8), and endosomes (fraction 11). Images shown are representative of results from three independent experiments.

Fig. 3.

Sortilin stabilizes cellular levels of FABP7 in CHO cells. (A,B) Parental Chinese hamster ovary (CHO) cells or CHO cells stably expressing sortilin (CHO-S) were transiently transfected with expression constructs encoding for FABP7 and GFP. A representative western blot of documenting expression of sortilin and FABP7 in replicate lysates of CHO and CHO-S cells is shown in A. Detection of GFP and tubulin served as transfection and loading controls, respectively. B shows FABP7 levels in CHO and CHO-S transfectants as determined by densitometric scanning of replicate western blots (n=9 replicates from 3 independent experiments per cell line). Data are mean±s.e.m. given as percentage of FABP7 levels in CHO cells (set to 100%). Levels of FABP7 (but not of GFP) are significantly increased by the presence of sortilin in CHO-S compared with CHO cells. **P<0.01 (unpaired, two-tailed Student's t-test). (C,D) CHO and CHO-S cells were transiently transfected with a FABP7-myc expression construct. At 48 h post transfection, replicate cultures of transfected cells were treated with 10 µg/ml of cycloheximide and collected at time points 0, 4, 8, and 12 h later. Levels of FABP7 were determined by western blotting (C). Detection of GAPDH served as loading control. The decrease in FABP7 levels was significantly faster in CHO than in CHO-S cells as determined by densitometric scanning of replicate blots (D; n=9 replicates per condition from three independent experiments). Data are mean±s.e.m. given as percentage of FABP7 levels in CHO or CHO-S at 0 h of treatment (set to 100%). **P<0.01; ****P<0.0001 (two-way ANOVA, followed by Bonferroni post-hoc analysis).

Fig. 4.

Sortilin deficiency decreases FABP7 levels in brains of apoE3 but not apoE4 mice. (A,B) Western blot analysis of FABP5 and FABP7 levels in brain cortices of apoE3 (A) and apoE4 (B) targeted replacement mice, either wild-type (WT) or homozygous for the Sort1 null allele (KO) (at 3 months of age). Na/K ATPase and GAPDH served as loading controls for detection of FABP5 and FABP7, respectively. Detection of sortilin served as genotype control. (C) Quantitative analysis of FABP7 levels in brain cortices of apoE3- and apoE4-expressing mice of the indicated Sort1 genotype using densitometric scanning of replicate western blots. Values are mean±s.e.m. given as percentage of the respective WT control (set to 100%); n=16 for E3/WT, n=18 for E3/KO, n=12 for E4 groups. **P<0.01 between genotypes (unpaired, two-tailed Student t-test). (D) Levels of Fabp7 transcripts in brain extracts (cortex and hippocampus) of apoE3- and apoE4-expressing mice of the indicated Sort1 genotypes as determined by quantitative RT-PCR. Values are given as mean±s.e.m. (n=5 for E3 groups, n=6 for E4/ WT, n=8 for E4/KO). No statistically significant differences in transcript levels were observed using unpaired, two-tailed Student's t-test. (E,F) Western blot analysis of FABP5 and FABP7 levels in prefrontal cortex specimens of AD patients homozygous for APOEε3 or APOEε4 (pathological characteristics given in Table S2). A representative western blot is shown in E. Detection of GAPDH served as loading control. F shows the result of densitometric scanning of replicate blots. Values are mean±s.e.m. given as percentage of APOEε3/3 genotype (mean value set to 100%); n=12–34 individuals per group. *P<0.05 (Welch's t-test).

Fig. 5.

Expression of FABP7 in neurons of mouse and human brains. (A) Immunohistological detection of FABP7 (red) in brain cortical sections from apoE3 mice, either wild-type (WT) or genetically deficient for Sort1 (KO). Additionally, the sections were stained for the neuronal marker NeuN (blue) and sortilin (green) as well as DAPI (white; in insets). Merged images show co-expression of FABP7 with NeuN. As a positive control, the insets document expression of FABP7 in glia in the cerebellum of E3/WT and E3/KO mice. Representative images from analysis of three mice per genotype are shown. Scale bar: 100 µm. (B) Staining for FABP7 in hippocampal subfield CA1 and temporal cortex specimens of AD patients, showing immunoreactivity in glial cells (arrowheads) and light positivity in sparse neuronal cells (arrows, and insets). Representative images from one of three AD cases analyzed are shown (pathological characteristics given in Table S3). Scale bars: 100 µm.

Fig. 6.

Sortilin deficiency decreases FABP7 levels in primary neurons of apoE3 mice. (A) Immunodetection of FABP7 (red) and sortilin (green) in primary neuronal cultures from apoE3 mice either (WT) or genetically deficient for Sort1 (KO). FAPB7-stained cells are identified as neurons by expression of MAP2 (blue). Merged images show co-expression of FABP7 and sortilin. Images shown are representative of results from three independent experiments. Scale bar: 20 µm. (B) Detection of apoE3 and apoE4 in medium and lysate of neuronal cultures of the indicated genotypes. Detection of recombinant apoE4 served as control (ctr). (C) Western blot analysis of sortilin as well as FABP5 and FABP7 in primary neuronal cultures from apoE3 or apoE4 expressing mice, either wild-type (WT) or genetically deficient for Sort1 (KO). Detection of GAPDH served as loading control. (D) Levels of FABP7 are reduced in primary neurons from E3/KO and E4/WT mice as compared with neurons from E3/WT animals as determined by densitometric scanning of replicate blots (exemplified in panel C). Data are given as mean±s.e.m. with the E3/WT levels set to 100% (n=8 for E3 groups, n=5 for E4/WT, n=4 for E4/KO). **P<0.01; ***P<0.001 for expression levels between E3/WT and all other genotypes (unpaired, two-tailed Student's t-test; *P<0.05 for the interaction of APOE and sortilin genotypes (two-way ANOVA with Bonferroni post test). (E) Levels of sortilin are comparable in primary neurons from E3/WT and E4/WT animals as determined by densitometric scanning of replicate blots (exemplified in panel C). Data are given as mean±s.e.m. with the E3/WT levels set to 100% (n=5–6 biological replicates per genotype). (F) Quantitative RT-PCR analysis of Gfap transcripts in primary neuronal cultures from E3/WT and E3/KO mice. Values are given as log₂ fold changes relative to E3/WT set to value 0 (n=4 independent cultures per genotype). (G) Levels of Fabp7 transcript were determined in mouse primary neuronal cultures of the indicated genotypes by quantitative RT-PCR (n=10 for E3/WT, n=7 for E3/KO, n=5 for E4/WT, n=6 for E4/KO). Values are given as log₂ fold change relative to transcript levels in the respective WT (set to value 0).

Fig. 7.

Levels and intracellular sorting of FABP7 are impacted by application of apoE4. (A,B) CHO cells stably co-expressing sortilin and FABP7 (CHO-S/F) were treated for 24 h with conditioned medium containing 5 µg/ml of apoE3 (+E3) or apoE4 (+E4; see Materials and Methods for details). Western blot analysis (A) and densitometric scanning of replicate blots (B) document reduced levels of FABP7 in CHO-S/F cells in the presence of apoE4 as compared with apoE3 (n=10 replicates for each condition from four independent experiments). Data are mean±s.e.m. given as percentage of FABP7 levels in apoE3-treated cells (set to 100%). *P<0.05 (unpaired, two-tailed Student's t-test). (C,D) Experiment as in A and B but using conditioned medium from primary astrocytes secreting apoE3 or apoE4 (n=3 replicates from one culture). Data are mean±s.e.m. given as percent of FABP7 levels in apoE3-treated cells (set to 100%). *P<0.05 (unpaired, two-tailed Student's t-test). (E) Proximity ligation assay (PLA) to visualize the intracellular localization of sortilin–FABP7 complexes (red signal) in CHO-S/F cells treated with apoE3- or apoE4-conditioned HEK293 medium for 24 h (left panel). For comparison, immunostaining of total sortilin (middle panels) and FABP7 (right panels) in treated cells are shown as well. Cell nuclei were counterstained with DAPI (blue). PLA signals for sortilin–FABP7 complexes change from a dispersed vesicular pattern with apoE3 to a perinuclear pattern with apoE4 (arrowheads). A similar change is seen for total sortilin, whereas the pattern for total FABP7 remains unaffected by apoE4. Images shown are representative of results from three independent experiments. (F,G) Colocalization studies in CHO-S/F cells documenting the presence of sortilin–FABP7 complexes (as deduced by PLA; red signal) in early endosomes, marked by antibodies against Rab5 (green signal; F) or recycling endosomes marked by Rab11 (green signal, G). (H,I) Extent of colocalization of PLA for sortilin and FABP7 with Rab5 (H) (n=34 cells for apoE3 and 33 cells for apoE4 treatment) and Rab11 (I) (n=38 cells for each condition) as documented by thresholded Manders’ coefficient tM1. This experiment was replicated three times. *P<0.05; **, P<0.01 (unpaired, two-tailed Student's t-test). Scale bars: 20 µm.

Fig. 8.

Interaction of sortilin and FABP7 in PPARγ-dependent gene expression. (A) CHO cells stably expressing sortilin (CHO-S) were transfected with reporter gene constructs encoding a PPAR-responsive firefly luciferase gene (Luc) and a constitutively expressed Renilla luciferase gene (Ren). Where indicated, transfectants were also treated with rosiglitazone (+RGS) for 24 h. At 48 h after transfection, the activities of firefly and Renilla luciferases were determined in cell lysates using a luminometer (n=3–8 replicates per cell line). Values are given as ratio of firefly to Renilla luciferase activity (mean±s.e.m., condition without RGS as 100%). ***P<0.001; ****P<0.0001 (unpaired, two-tailed Student's t-test). (B) Replicate layers (n=4) of CHO cells stably expressing sortilin (CHO-S), FABP7 (CHO-F), or both proteins (CHO-S/F) were transfected with constructs encoding a PPAR-responsive firefly luciferase and a constitutively expressed Renilla luciferase and analyzed for luciferase activity as described above. Values are given as ratio of firefly to Renilla luciferase (mean±s.e.m., CHO-S set to 100%). **P<0.01; ***P<0.001 (unpaired, two-tailed Student's t-test). (C) CHO-S/F were transfected with reporter gene constructs encoding a PPAR-responsive firefly luciferase reporter gene and a constitutively expressed Renilla luciferase. Then, transfectants were treated with conditioned medium containing 5 µg/ml of human apoE3 or apoE4 for 24 h. At 48 h after transfection, the activities of firefly and Renilla luciferases were determined in cell lysates using a luminometer (n=9 replicates from 3 independent experiments per condition. Values are given as ratio of firefly to renilla luciferase (mean±s.e.m., +apoE3 set to 100%). *P<0.05 (unpaired, two-tailed Student's t-test). (D,E) Proposed model of the interaction of sortilin with FABP7 and apoE3 (D) or apoE4 (E) in cellular lipid metabolism (see Discussion section for details).