SUMMARY
Sex and apolipoprotein E ε4 (APOE4) interact to alter the risk for Alzheimer’s disease and other neurodegenerative disorders. Herein, we show sex-specific differences in immune activation and lymphatic function in the meningeal dura of humanized female and male mice expressing two alleles of APOE4 (E4/E4), when compared to their respective sex-matched E3/E3 controls. We also describe distinct effects of APOE4 on brain lipid composition and inflammation in females and males that were partially reverted upon CSF1R inhibition. Suppressing innate immunity reduced neuroinflammation and restored cognitive function in E4/E4 females, while exacerbating neuroinflammation and cognitive decline in E4/E4 males. Finally, in line with the E4/E4 humanized mouse model data, we show that APOE4 expression is linked to sexually dimorphic leukocyte activation profiles in the human brain. This study highlights the need for personalized therapies when targeting APOE, brain immunity, and meningeal lymphatics to promote cognitive resilience in either females or males.
Graphical Abstract
eTOC Blurb
Delivanoglou et al. show that expression of APOE4, a risk factor for Alzheimer’s disease, leads to sexually dimorphic lymphatic and inflammatory responses in mice and humans alike. Suppressing APOE4-induced innate immunity had contrasting effects on cognition in females and males, highlighting the need for immunotherapies tailored to each sex.
INTRODUCTION
Apolipoprotein E (APOE) is the main lipid carrier in the brain. The human APOE gene has three polymorphic alleles, namely APOE2, APOE3 (E3), and APOE4 (E4). The most frequent genotype found in the general population is E3/E3, with much lower incidence of the E3/E4 and E4/E4 genotypes.1,2 Of note, in Caucasians, expression of one or two alleles of E4 increases the risk for Alzheimer’s disease (AD) by approximately 3–4-fold or 9–15-fold, respectively.3–5 Women are also at a higher risk of developing AD. In fact, recent evidence suggests that AD might manifest earlier in female E4 carriers, despite the apparent lack of obvious clinical differences between females and males expressing the same number of E4 alleles.2,6 Sex and APOE4, independently, can modulate vascular, immune, and glial responses in the brain.7–16 Recent studies involving rodent models shined light on distinct effects of APOE4 in females and males. Neural and peripheral cells from rodents expressing E4 present a sexual dimorphism, a feature that was more evident in immune cells like microglia.9,13,15,17–19 Yet, it remains unclear why females are at a greater risk of developing AD and how E4 shifts the balance towards dementia in the elderly.
Humanized mice expressing either E3 or E4 in place of the endogenous murine Apoe gene allele represent unique models to investigate the effects of these two distinct APOE isoforms.20–22 Taking advantage of such humanized mouse models, numerous studies have focused on APOE4-linked changes in neuronal, glial, and vascular responses, either in the absence or presence of AD-like neuropathology. Differences in brain APOE levels, lipid composition, vascular and white matter integrity, blood flow, inflammation, and behavior have been reported in mice expressing one or two E4 alleles.8,10–14,16,20,21,23–34 Still, experiments have not been designed to evaluate the putative sex-dependent effects of E3/E3 or E4/E4 expression on brain immunity, and its implications for meningeal lymphatic drainage, neuroinflammation, and cognitive (dys)function.
Herein, we show that middle-aged female mice show more major histocompatibility complex class II (MHC-II)-expressing macrophages in the meningeal dura when compared with males. Expression of E4 led to sex-specific transcriptomic changes in meningeal innate and adaptive immune cells that were accompanied by increased lymphatic vessel coverage and reduced drainage of CSF into the deep cervical lymph nodes (LNs) in males. Male E4/E4 mice also showed higher levels of chemokines in the cervical LNs and lower levels of neuroinflammation in the brain, which were linked to cognitive resilience. Conversely, middle-aged female E4/E4 mice presented significantly higher levels of proinflammatory proteins in the cervical LNs and brain that were correlated with worsened cognitive function. Administration of the colony stimulating factor 1 receptor (CSF1R) inhibitor PLX5622 (PLX) led to similar degrees of brain and meningeal innate immune cell depletion in all groups of middle-aged mice. Yet the dural immune and vascular cells from female and male E4/E4 mice showed distinct transcriptional programs in response to PLX that were accompanied by a regression of meningeal lymphatics in E4/E4 males, and a reduction in lymphatic drainage in E4/E4 females. Sex- and APOE genotype-related brain lipidomic, glial, and neuroinflammatory profiles induced by CSF1R inhibition were linked to improved anxious-like behavior and cognitive function in E4/E4 females but worsened cognitive function in E4/E4 males. Finally, by integrating human brain transcriptomic data from 6 independent publicly available datasets, we show that E4 expression leads to a similar sexual dimorphic activation profile in different types of brain leukocytes, particularly in brain BAMs.
RESULTS
Females have more MHC-II+ macrophages and higher APOE protein content in the dura than males
Whole dorsal meningeal dura samples were isolated from 12–13-month-old (middle-aged) female and male mice either expressing two alleles of E3 (E3/E3) or E4 (E4/E4),22 instead of the murine Apoe gene allele, and used to determine the cellular sources of human APOE mRNA transcripts using RNAscope (Figures 1A and S1A–S1D). Labeling with probes against APOE, vascular gene platelet endothelial cell adhesion molecule 1 (Pecam1), and innate myeloid gene integrin subunit alpha M (Itgam), showed similar numbers of APOE+ cells in the vicinity of the dural superior sagittal sinus between groups (Figure S1A). Most APOE+ cells co-expressed Itgam rather than co-expressing Pecam1 or being negative for each gene transcript (Figures S1B–S1D). APOE was detected at a considerable concentration in the meningeal dura and was slightly higher in females than males (Figures 1B and 1C). This sex-specific difference in APOE levels was not detected in the forebrain samples (Figures S1E and S1F). Interestingly, APOE was detected at similar concentration magnitudes in protein extracts from meningeal dura and forebrain (Figures 1C and S1F). Due to the innate immune origin of the meningeal dural APOE mRNA transcripts, we then used flow cytometry to assess the frequencies of some of the most prevalent innate immune cell types and subtypes in this same brain border tissue (Figures 1D, 1E, and S1G–S1O). Despite the similar innate immune cell frequencies and numbers observed in the E3/E3 and E4/E4 groups within each sex, we observed an increase in macrophages expressing cluster of differentiation 206 (CD206) and MHC-II in females, when compared with males, independently of the APOE genotype (Figures 1E and S1M). Concurrently, the frequency, but not the number, of monocytes was increased in males compared with females (Figures S1K and S1O).
Figure 1. Lower APOE content and MHC-II+ macrophages in the meningeal dura of middle-aged E4/E4 male mice correlates with increased lymphatic vessel coverage, but reduced CSF lymphatic drainage.
(A) Meningeal dural whole mount showing the mRNA transcripts for the genes Pecam1 (cyan), APOE (green), and Itgam (red), and cell nuclei stained with 4,6-diamidino-2-phenylindole (DAPI, blue) in the vicinity of the superior sagittal sinus (peri-SSS).
(B and C) Whole protein extracts from samples of meningeal dura were used to measure APOE levels by ELISA, and plotted as shown in (C).
(D and E) Flow cytometry dot plots showing the expression of major histocompatibility complex class II (MHC-II) and cluster of differentiation 206 (CD206) in macrophages in (D), and frequencies of each macrophage subpopulation in (E).
(F) Meningeal dural whole mounts stained for lymphatic vessel endothelial hyaluronan receptor-1 (LYVE-1, green) and DAPI (blue).
(G) Graph showing the quantifications of LYVE-1+ lymphatic vessel length per area of region of interest (ROI).
(H) Representative in vivo stereomicroscopy images depicting the fluorescence signal of ovalbumin-Alexa Fluor 555 (OVA-AF555) in the deep cervical lymph node (LN) after intra-cisterna magna (i.c.m.) injection.
(I) Graph showing the gain in OVA-AF555 fluorescence (arbitrary units, A.U.).
Data in (C), (E), (G), and (I) are presented as mean ± standard error of mean (SEM); n = 5 mice per group in (C) and (E); n = 12–16 mice per group in (G); n = 11–12 mice per group in (I); data in (G) and (I) were pooled from 3 independent experiments; two-way ANOVA with Bonferroni’s multiple comparisons test was used in (C), (E), and (G); multiple comparisons test was performed between the E3/E3 and E4/E4 groups within each sex in (C) and (G); in (I), a linear regression model including covariates for APOE genotype, sex, and the interaction between APOE genotype and sex was used to examine whether the differences in OVA-AF555 levels over time between APOE genotypes was consistent for females and males; in (I), another linear regression model with covariates for sex and APOE genotype were used to examine differences in OVA-AF555 levels over time between females and males, and between E3/E3 and E4/E4 mice.
See also Figures S1, S2, and S3.
APOE4 affects meningeal lymphatic function in a sex-dependent manner
The higher density of myeloid APOE-expressing cells along the dural venous sinuses (Figure 1A) led us to investigate the putative effects of E3 or E4 expression on the nearby lymphatic vasculature, which are the only vessels co-expressing both CD31 and LYVE-1 in the dorsal dura (Figures 1F–1I and S2A–S2J). Assessments of meningeal dural whole mounts from mice at 2–4 or 12–13 months of age revealed an age- and sex-dependent effect of E4 expression on the total length of lymphatic vessel endothelial hyaluronan receptor 1 (LYVE-1)-expressing vessels (Figures 1F, 1G, S2B, and S2C). Increased length of LYVE-1+ vessels was seen in male, but not female, E4/E4 mice at 12–13 months, when compared with their respective sex- and age-matched E3/E3 control counterparts (Figures 1F and 1G). No differences in lymphatic morphology between groups were observed in the dorsal dura at the younger age of 2–4 months (Figures S2B and S2C), or in the ventral dura, namely around the cavernous, sigmoid, and petrosquamosal sinuses, at 12–13 months (Figures S2D–S2J). Likewise, we did not find differences between groups in terms of skin lymphatic vessel morphology at 12–13 months, which pointed to meningeal-specific, rather than organism-wide, sex-specific responses to E4 expression (Figures S2K and S2L). To determine if the differences in meningeal lymphatic vessel morphology observed in E4/E4 males were due to an APOE loss-of-function, we performed similar evaluations at 2–3 or 11–12 months of age in female and male mice expressing (Apoe+/+) or lacking (Apoe−/−) the murine Apoe gene alleles (Figures S3A–S3H). No changes between groups were detected in terms of lymphatic vessel length in the meningeal dura (Figures S3A–S3D) or the skin (Figures S3E–S3H).
Based on published evidence,35 we next hypothesized that the increased lymphatic vessel coverage observed in the dura of middle-aged male E4/E4 mice would be accompanied by a defect in lymphatic drainage of CSF (Figures 1H and 1I). Indeed, at 12–13 months of age, male E4/E4 mice showed a decreased outflow rate of fluorescent peptides from the CSF into the deep cervical LNs, when compared with their age-matched female E4/E4 counterparts. Of note, despite the similar rates of CSF lymphatic drainage observed in the E3/E3 groups, higher rates were detected in females, when compared with males (Figures 1H and 1I).
Overall, these data point to sex-specific and E4-dependent effects on meningeal dural APOE levels, innate immunity, and lymphatic morphology and function in middle-aged mice. Our data also suggests that the isoform-related effects of APOE on the morphology of meningeal lymphatics in males is not attributed to a loss of APOE physiological function.
APOE4 elicits sex-specific effects on meningeal dural immune cell transcriptomes
We turned to meningeal dural single-cell RNA sequencing (scRNA-seq) to decipher the cellular transcriptomes linked to altered meningeal lymphatic vascular function in middle-aged mice expressing E3 or E4 (Figure 2A). To further investigate the contribution of innate immune cell activation to the observed meningeal lymphatic responses, we included groups of age-, sex-, and APOE genotype-matched littermate mice that were exposed to regular chow diet (Ctrl) or a chow diet containing PLX (at 600 p.p.m.), a CSF1R inhibitor that depletes innate immune cells that rely on CSF1R signaling for their survival, namely microglia and BAMs in the central nervous system (CNS; Figure 2A).36,37 Using these approaches, we were able to identify and annotate 13 cell clusters, including blood endothelial cells (BECs), lymphatic endothelial cells (LECs), B cells, T & natural killer (NK) cells, macrophages, mast cells, and fibroblasts (Figures 2B, S4A, and S4B). We confirmed that the highest expression levels of APOE transcripts were observed in macrophages, which co-express the lineage-specific Mrc1, Csf1r, and Lyz2 genes (Figures S4C–S4F). Of note, elevated APOE expression was also detected amongst fibroblasts, and at a lower level in the clusters of mast cells, pericytes & smooth muscle cells (SMCs), and BECs (Figures 2B and S4C).
Figure 2. APOE4 expression is linked to sex-specific immune transcriptional profiles in the meningeal dura and responses to innate immune depletion by CSF1R inhibition.
(A) Whole dorsal meningeal dural tissues were processed into single-cell suspensions, CD31 expressing endothelial cells were enriched by magnetic activated cell sorting (MACS), and the cellular transcriptomes were analyzed via 10x Genomics single-cell RNA sequencing (scRNA-seq).
(B) Uniform manifold approximation and projection (UMAP) representation of the 13 clusters and respective cluster annotations. Smooth muscle cells (SMCs), blood endothelial cells (BECs), lymphatic endothelial cells (LECs), natural killer cells (NK cells), innate lymphoid cells (ILCs), bone marrow (BM) precursors.
(C and D) UpSet plots showing the number of unique or overlapping, up-regulated (Up) or down-regulated (Down), genes in E4/E4 Ctrl female or male (versus their respective E3/E3 groups) macrophages in (C) and B cells in (D).
(E and F) UpSet plots showing the number of unique or overlapping differentially expressed genes in the B cells of the PLX groups (versus their respective Ctrl groups). Up-regulated genes in (E) and down-regulated genes in (F).
(G–J) NicheNet circos plots depicting the target genes highly expressed in LECs and the corresponding links to cell type-specific ligand genes from BECs, B cells, T & NK cells, macrophages, or common to two or more cell types (common).
See also Figures S4, S5, S6, S7, and S8.
Macrophages were amongst the immune cell types with higher total number of differentially expressed genes (DEGs) in both female and male E4/E4 Ctrl mice, when compared with their respective sex-matched E3/E3 Ctrl counterparts (Figures 2C and S4G). Yet, E4 expression induced a sex-specific gene expression signature in macrophages, with little overlap between the up- and down-regulated genes in macrophages from the female and male groups (Figure 2C). Gene set enrichment analysis using the up- or down-regulated DEGs found in macrophages further corroborated the distinct effects of E4 expression on the gene functional pathways triggered in middle-aged female and male mice (Figures S4H–S4K). Except for the “R-MMU-6798695: neutrophil degranulation” gene enrichment term, which was equally up-regulated in E4/E4 Ctrl females and males (Figures S4H and S4J), all of the remainder terms were unique to each sex. The transcriptome of adaptive immune cells was also affected by APOE4 in a sex-dependent manner. The male E4/E4 Ctrl group showed a total of 289 and 126 DEGs in the clusters of B cells and of T & NK cells, respectively, when compared with E3/E3 Ctrl males. Whereas only 17 and 29 DEGs were found in the same cell clusters of the female E4/E4 Ctrl group, when compared with E3/E3 Ctrl females (Figures 2D and S4G).
Therefore, expression of two E4 alleles leads to significant, but distinct, gene expression changes in meningeal dural immune cells in middle-aged females and males.
CSF1R inhibition depletes innate immune cells and causes sex-specific gene expression changes in the dura of E4/E4 mice
We next wondered if female and male mice of the E3/E3 and E4/E4 groups would also present distinct gene expression profiles in the dura upon PLX-induced depletion of innate immune cells. A closer look at the scRNA-seq cluster frequencies showed, once again, that females present higher proportions of meningeal dural macrophages than males, and that, as expected, the 4-week exposure to PLX efficiently reduced the macrophage frequencies in all the PLX groups (Figure S4B). Because the scRNA-seq technique is not designed for accurate profiling of tissue-resident immune cell frequencies, we employed flow cytometry to determine the effects of the employed PLX regimen on meningeal dural immune cells (Figures S5A–S5I). We observed a consistent and selective depletion of CD206+MHC-II+ macrophages in all female and male groups (~75–80% reduction compared with Ctrl groups), regardless of their APOE genotype (Figures S5D and S5H). Discrete, but at times statistically significant, changes in the frequencies of the other meningeal dural immune cell populations (within total live CD45+ cells) were also detected in the PLX groups, when compared with the Ctrl groups, which we interpreted as compensations for the reduced frequencies of the CD206+MHC-II+ macrophages (Figures S5B and S5F).
Next, we assessed the effect of PLX-induced CSF1R inhibition on dural APOE gene expression and APOE protein levels in the different groups. Exposure to PLX led to reduced APOE expression levels in the clusters of macrophages, fibroblasts, mast cells, and BECs, regardless of sex, in both the E3/E3 PLX and E4/E4 PLX groups (Figure S6A). Accordingly, we detected lower concentrations of APOE protein in the meningeal dura of mice of the PLX groups at 2 weeks, regardless of sex or genotype (Figure S6B–S6D). Differential gene expression analysis also pointed to a wide impact of innate immune cell depletion by PLX on the transcriptomes of different meningeal dural cell types (Figures 2E–2J and S7A–S7G). As the cells expressing the highest levels of Csf1r in the dura (Figure S4E), and most sensitive to CSF1R inhibition, macrophages were amongst the cell types presenting the highest number of DEGs across all the comparisons between the PLX and Ctrl groups (Figure S7A). The number of DEGs in macrophages was particularly elevated in the female E4/E4 PLX group, when compared with the female E4/E4 Ctrl group. The same was true for BECs and B cells of the female E4/E4 PLX group, which presented a total of 4,760 and 492 DEGs, respectively (Figure S7A). Of note, the two cell types showing the highest numbers of DEGs in the male E4/E4 PLX group (compared with male E4/E4 Ctrl) were also BECs and B cells, with a total of 1,076 and 257 DEGs, respectively (Figure S7A). However, in contrast to the female PLX and Ctrl group comparisons, the male E4/E4 PLX group also showed substantially higher number of DEGs in T & NK cells (126 DEGs), innate lymphoid cells (72 DEGs), dendritic cells (197 DEGs), and fibroblasts (89 DEGs), when compared with the male E4/E4 Ctrl group (Figure S7A). We also noticed that, notwithstanding the changes in macrophage gene expression across all groups, overall, the dural cells from the E4/E4 groups showed higher number of DEGs after 4 weeks of CSF1R inhibition than the E3/E3 groups, regardless of sex (Figure S7A). Once again, we observed an insignificant overlap between the up- and down-regulated DEGs found in the B cells and BECs of female or male E4/E4 mice exposed to PLX (Figures 2E, 2F, S7B, and S7C). As expected, the largely unique up-regulated DEGs in B cells and BECs from females and males were part of different gene functional pathways (Figures S7D–S7G).
Altogether, these data show that CSF1R inhibition by PLX leads to distinct gene expression profiles in the immune and vascular cells residing in the meningeal dura of female and male E4/E4 mice.
Innate immune suppression differently affects lymphatic signaling and function in E4/E4 females and males
NicheNet38 was employed to shine light on the putative molecular interactions between LECs (whose transcriptomes were used as “target genes”) and BECs, B cells, T & NK cells, and macrophages (whose transcriptomes were used as “ligand genes”), that could underlie the sex-specific changes in meningeal lymphatic vessel coverage and CSF lymphatic drainage capacity (Figures 2G–2J). In the female E4/E4 Ctrl group, we observed a uniform crosstalk between the LECs and all cell types (Figure 2G). Some of the gene ligands were predicted to be upstream of Egr1 and Plau expression by LECs of the female E4/E4 Ctrl group (Figure 2G). The predicted cellular crosstalk in the dura of male E4/E4 Ctrl mice presented some similarities to female E4/E4 Ctrl mice, namely in terms of heightened Tgfb1 signaling from B cells, Il15 signaling from macrophages, and Tnf signaling from T & NK cells (Figure 2H). However, in the male E4/E4 Ctrl group, fewer gene ligands originated from BECs and macrophages, and Pdgfb appeared as a top ligand originating from BECs (Figure 2H). The Edn1, Fas, Ifit3b, and Myc genes appeared as the main targets of the putative vascular- and immune-derived ligands in the male E4/E4 group (Figure 2H). Upon exposure to PLX, the BECs, macrophages, and B cells from female E4/E4 mice no longer appeared as the main ligand gene contributors to LEC transcriptomic changes (Figure 2I). Compared with male E4/E4 Ctrl, the male E4/E4 PLX group showed reduced signaling from macrophages and expanded signaling from BECs, which overexpressed the ligand genes Cxcl12, Bmp6, and Tgfb2 (Figure 2J). The male E4/E4 PLX group also showed a shift from Tnf to Ifng signaling from T & NK cells. The main target genes in LECs from the male E4/E4 PLX group were Gadd45b, Igfbp3, and Plaur (Figure 2J).
The high number of DEGs observed in BECs from the E4/E4 groups in response to PLX (Figure S7A), and the anatomical proximity between blood and lymphatic vessels in the meningeal dura, prompted us to perform another NicheNet analysis, but now using the transcriptomes of BECs as “target genes” and of LECs, B cells, T & NK cells, and macrophages as “ligand genes” (Figures S8A–S8D). Contrarily to the NicheNet analysis where the LECs’ transcriptomes were used as “target genes” (Figures 2G and 2H), the BECs from the female E4/E4 Ctrl group established a limited crosstalk with macrophages (Figure S8A), whereas the opposite was observed in the male E4/E4 Ctrl group (Figure S8B). The BECs from the female E4/E4 Ctrl group seemed to establish a stronger crosstalk with LECs instead, via the genes Tnfsf10, Tgfb2, and Cxcl12 (Figure S8A), further substantiating the strong crosstalk between LECs and BECs observed before in this group (Figure 2G). The partial innate immune depletion by PLX resulted in a rescue of the crosstalk between BECs and macrophages in E4/E4 females (Figure S8C), in detriment of the signaling from LECs (Figure S8A). Of note, similar to the male E4/E4 Ctrl group, the macrophage ligands Igf1, Il15, and Tslp, appeared amongst the main drivers of downstream transcriptomic changes in BECs in E4/E4 females upon exposure to PLX. Increased Ifng signaling from T & NK cells to BECs also became prominent in the male E4/E4 PLX group (Figure S8D). However, exposure of E4/E4 males to PLX had little impact on the crosstalk between dural BECs and the remainder macrophages, still mediated by Igf1, Il15, and Hgf (Figure S8D).
The sex-specific ligand gene-to-LEC target gene interactions observed in E4/E4 mice led us to investigate the impact of CSF1R inhibition by PLX on meningeal lymphatic vessel morphology. As we have shown before, the 4-week PLX exposure led to a reduction in MHC-II+ innate immune cells in the dura, namely at the confluence of sinuses (COS) and the transverse sinus (TS), of mice from all groups (Figures 3A, 3B, S9A, and S9B). All the groups of middle-aged female mice showed similar LYVE-1+ lymphatic vessel lengths at the COS and TS even after exposure to the PLX diet for 4 weeks (Figures 3A and 3C). Conversely, in males, the significant differences between the E3/E3 and E4/E4 groups in terms of lymphatic vessel length at the COS were abrogated by inhibition of CSF1R signaling (Figures 3B and 3D). Moreover, the group of male E4/E4 PLX mice showed a reduction in lymphatic vessel length at the TS when compared with the male E4/E4 Ctrl and male E3/E3 PLX groups (Figures 3B and 3D). Next, we evaluated the effect of innate immune depletion by PLX on lymphatic drainage of CSF to the deep or superficial cervical LNs in the groups of middle-aged mice (Figures 3E–3J). Despite showing similar lymphatic morphology at the dorsal meningeal dura, the groups of E3/E3 and E4/E4 females exposed to PLX presented reduced CSF drainage to the deep cervical LNs, when compared to their sex-matched Ctrl groups (Figures 3E and 3G). The levels of CSF drainage into the superficial cervical LNs remained similar across all female groups, which resulted in reduced deep/superficial ratio of net CSF lymphatic drainage in the PLX groups (Figures 3E, 3G, and 3I). Conversely, E4/E4 males showed reduced CSF drainage to the superficial cervical LNs, when compared to E3/E3 males, independently of PLX exposure (Figures 3F and 3H). Yet this effect had no impact on the deep/superficial ratio of CSF lymphatic drainage (Figure 3J). In sum, we show that a suppression of the innate immune response via CSF1R inhibition leads to distinct cellular inflammatory profiles in the meningeal dura of female and male E4/E4 mice, and different outcomes in terms of CSF lymphatic drainage to the cervical LNs. Of note, overall, depletion of innate immune cells in female E4/E4 mice (female E4/E4 PLX group) resulted in profiles of meningeal cellular crosstalk and CSF lymphatic drainage to the cervical LNs that resembled the profiles observed in the male E4/E4 Ctrl group. In middle-aged male E4/E4 mice, CSF1R inhibition led to increased signaling by interferon (IFN)-γ (encoded by the gene Ifng) in meningeal LECs and a concomitant loss of lymphatic vasculature, but no further reductions in CSF lymphatic drainage to the cervical LNs.
Figure 3. Modulation of innate immunity by CSF1R inhibition has divergent effects on meningeal lymphatic morphology and drainage in middle-aged female and male E4/E4 mice.
(A and B) Confluence of sinuses (COS) and transverse sinus (TS) of meningeal dural whole mounts stained for LYVE-1 (green) and MHC-II (blue). Females in (A) and males in (B).
(C and D) Graphs showing the quantifications of LYVE-1+ lymphatic vessel length per area of ROI. Females in (C) and males in (D).
(E and F) Representative images of the deep and superficial cervical LNs depicting tissue autofluorescence (green) and OVA-AF555 (red) drained from the CSF. Females in (E) and males in (F).
(G and H) Graphs showing the quantifications of OVA-AF555 fluorescence intensity in the deep and superficial cervical LNs. Females in (G) and males in (H).
(I and J) Ratios between OVA-AF555 in the deep and superficial cervical LNs. Females in (I) and males in (J).
Data in (C), (D), and (G–J) are presented as mean ± SEM; n = 6–12 mice per group in (C); n = 7–14 mice per group in (D); n = 5–6 mice per group in (G–J); two-way ANOVA with Fisher’s LSD multiple comparisons test.
See also Figure S9.
APOE4 and CSF1R inhibition affect the immune responses at the cervical LNs in a sex-dependent manner
To further understand how E4 expression differently influences neuroimmune responses in females and males, we analyzed the frequencies of immune cells and inflammatory cytokines and chemokines in the brain-draining cervical LNs (Figures 4A–4D, S9C, and S9D, and Supplemental Table S1). Expression of E4 or exposure to PLX had no effect on the frequencies of type I conventional DCs (cDCs), type II cDCs, B cells, or conventional TCRβ+ T cells in the deep cervical LNs of middle-aged female or male mice (Figures S9C and S9D). Female groups also showed similar frequencies of CD8+, CD4+FOXP3–, and CD4+FOXP3+ T cells within the total pool of conventional TCRβ+ T cells (Figure 4A). However, male E4/E4 mice showed differences in the balance between CD8+ and CD4+FOXP3– T cells in the deep cervical LNs, with a significant increase in CD4+FOXP3– T cells in detriment of decreased CD8+ T cells (Figure 4B). Innate immune depletion by PLX further deepened the reduction in CD8+ T cells, and increased the frequencies of CD4+FOXP3+ T cells in the deep cervical LNs, particularly in the E4/E4 male group (Figure 4B).
Figure 4. APOE4 expression and modulation of innate immunity induce distinct inflammatory profiles in the cervical LNs of middle-aged females and males.
(A and B) Frequencies of CD8+ T cells, CD4+FOXP3− T cells, and CD4+FOXP3+ T cells in the deep cervical LNs. Females in (A) and males in (B).
(C and D) Heatmaps showing the levels of inflammatory proteins in the cervical LNs isolated from females, in (C), or males, in (D), of each group (scale bar shows the fold change relative to each sex-matched E3/E3 Ctrl group). Asterisks highlight proteins altered between the E4/E4 PLX and E4/E4 Ctrl groups in either females or males. Interleukin (IL), colony-stimulating factor (CSF), C-C motif chemokine ligand (CCL), C-X-C motif chemokine ligand (CXCL), hepatocyte growth factor (HGF).
Data in (A and B) are presented as mean ± SEM; n = 4–5 mice per group; two-way ANOVA with Bonferroni’s multiple comparisons test. Data in (C and D) are presented as fold change relative to the E3/E3 Ctrl group; n = 5 mice per group; two-way ANOVA with Bonferroni’s multiple comparisons test performed between the E3/E3 Ctrl and E4/E4 Ctrl groups, and the E4/E4 Ctrl and E4/E4 PLX groups for each sex.
See also Figure S9 and Supplemental Table S1.
Expression of E4 resulted in different levels of inflammatory cytokines and chemokines in the cervical LNs of middle-aged females and males (Figures 4C and 4D). Female E4/E4 mice showed higher levels of interleukin (IL)-1β, IL-10, IL-17A, and IL-17F in the cervical LNs, when compared to female E3/E3 (Figure 4C). On the other hand, the cervical LNs of male E4/E4 mice showed increased levels of C-C motif chemokine ligand (CCL)5, CCL11, CCL17, C-X-C motif chemokine ligand (CXCL)12, and hepatocyte growth factor (HGF), and only trends for higher levels of IL-1β, IL-6, and IL-10 (Figure 4D). Inhibition of CSF1R signaling by PLX resulted, once again, in a sexually dimorphic inflammatory response in the cervical LNs of E4/E4 females and males. The cervical LNs of the female E4/E4 PLX group presented reduced levels of IL-1β, IL-10, IL-17F, CCL12, and CXCL9, yet increased levels of colony-stimulating factor (CSF)1, CSF3, and CCL2, when compared to the female E4/E4 Ctrl group (Figure 4C). The male E4/E4 PLX group showed an overall trend for reduced levels of cytokines and chemokines when compared to the male E4/E4 Ctrl group. Still, statistically significant reductions were only detected for IL-6, CCL11, CCL12, and CXCL2 (Figure 4D).
These data disclosed the distinct APOE4-related immune responses at the brain-draining cervical LNs that accompanied the changes in meningeal lymphatic morphology and drainage in females and males.
APOE4 and CSF1R inhibition interact to shape the brain lipidome and oligodendrocyte responses
Inhibition of CSF1R signaling by PLX for 2 weeks led to a partial sex- and genotype-independent depletion of brain leptomeningeal macrophages and parenchymal microglia in middle-aged mice (Figures S10A–S10L). In females, the innate immune cell depletion was accompanied by a nominal, yet statistically significant, reduction in APOE protein in the forebrain (Figures S10M and S10N). However, exposure of male mice to PLX did not affect the levels of APOE protein in the forebrain, pointing once again to a sex-specific effect (Figure S10O). Subsequently, we sought to investigate the effects of APOE genotype, sex, and the PLX-induced partial depletion of innate immune cells on the brain lipidome of middle-aged mice (Figures 5A–5D, S11A, and S11B, and Supplemental Table S2). Principal component analysis of the levels of all the detected forebrain lipid species revealed a clear separation of the samples by APOE genotype. Of note, expression of E4 led to a further segregation of the forebrain lipid profile between females and males, a phenomenon that was not as evident in females and males expressing E3 (Figure 5B). We then identified all the significantly altered lipid species between the E4/E4 Ctrl and E3/E3 Ctrl groups and the E4/E4 PLX and E4/E4 Ctrl groups and plotted them as heatmaps segregated by sex (normalized to their sex-matched E3/E3 groups; Figures 5C and 5D). Expression of two E4 alleles led to higher levels of sulfatide (Sulf), phosphatidylinositol (PI), phosphatidylglycerol (PG), phosphatidylethanolamine (PE), N-acyl phosphatidylserine (NAPS), bis(monoacylglycerol)phosphate (BMP), and acylcarnitine (AC) and lower levels of phosphatidylcholine-ether (PCe), PC, and N-acyl-serine (NSer) lipid species in both females and males. Yet, for certain lipid species, such as Sulf and BMP lipid species, the increases were more evident in E4/E4 males than E4/E4 females (Figures 5C and 5D). We have also observed sex-specific lipid alterations in the E4/E4 groups, when compared with their respective sex-matched control E3/E3 groups. Female E4/E4 mice showed unique decreases in lysophosphatidylcholine (LPC) and diacylglycerol (DG) and increases in lactosylceramide (LacCer) and dihydrosphingomyelin (dhSM) lipid species (Figure 5C). Conversely, increased TG, sphingomyelin (SM), N-acyl-phosphatidylethanolamine (NAPE), mono-hexosylceramide (MhCer), monoacylglycerol (MG), globotriaosylceramide (GB3), dihydroceramide (dhCer), cholesteryl ester (CE), and acylphosphatidylglycerol (AcylPG) and decreased LacCer lipid species were evident in male E4/E4 mice only (Figure 5D). Innate immune cell depletion by PLX led to a reduction in DG lipid species in female E4/E4 mice only, when compared with the female E4/E4 Ctrl group (Figure 5C). On the other hand, the male E4/E4 PLX group showed sex-specific reductions in specific sphingolipid species, such as SM, MhCer, and GB3, and increased AC lipid species, when compared with their respective sex-matched E4/E4 Ctrl group (Figure 5D). Of note, when compared with their genotype-matched Ctrl groups, females from the E3/E3 PLX group presented higher levels of LacCer and dhSM lipid species (Figure S11A), whereas males from the E3/E3 PLX group presented higher levels of BMP lipids (Figure S11B).
Figure 5. Partial innate immune suppression affects the brain lipidome and cellular lipid droplet content in middle-aged E4/E4 mice.
(A) Whole forebrains were collected and processed for targeted lipidomics by liquid chromatography–mass spectrometry (LCMS).
(B) Principal component (PC) analysis graph showing the distribution of the samples based on the profile of lipid species.
(C and D) Heatmaps depicting the levels of all lipid species altered in the forebrains of females in (C), or males in (D). Asterisks highlight lipids altered between the E4/E4 PLX and E4/E4 Ctrl groups in either females or males. Triacylglycerol (TG), sulfatide (Sulf), sphingomyelin (SM), phosphatidylinositol (PI), phosphatidylglycerol (PG), phosphatidylethanolamine (PE), phosphatidylcholine-ether (PCe), phosphatidylcholine (PC), N-acyl-serine (NSer), N-acyl-phosphatidylserine (NAPS), N-acyl-phosphatidylethanolamine (NAPE), mono-hexosylceramide (MhCer), monoacylglycerol (MG), lysophosphatidylcholine (LPC), lactosylceramide (LacCer), globotriaosylceramide (GB3), dihydrosphingomyelin (dhSM), dihydroceramide (dhCer), diacylglycerol (DG), cholesteryl ester (CE), bis(monoacylglycerol)phosphate (BMP), acylphosphatidylglycerol (AcylPG), and acylcarnitine (AC).
(E–J) Images of the brain cortex in (E) and corpus callosum (highlighted with dashed lines) in (H) stained for ionized calcium-binding adaptor molecule 1 (IBA1, green) and perilipin 3 (PLIN3, red). Quantifications of the total volume of PLIN3+ cells and of the levels of PLIN3 in IBA1+ cells in the cortex of females in (F) and males in (G), and in the corpus callosum of females in (I) and males in (J).
Data in (C and D) are presented as log2(fold change) relative to their sex-specific E3/E3 Ctrl groups; n = 5 mice per group; two-tailed unpaired Student’s t-test or Wilcoxon rank sum test were used according to data normality as assessed using the Shapiro-Wilk test in (C and D). Data in (F, G, I and J) are presented as mean ± SEM; n = 5–6 mice per group; two-way ANOVA with Fisher’s LSD multiple comparisons test.
See also Figures S10, S11, S12, and Supplemental Table S2.
We next evaluated the changes in immune myeloid and non-immune myeloid lipid droplet content via staining for perilipin 3 (PLIN3) in the brain grey and white matter of the mice from different groups (Figures 5E–5J). The volumetric quantifications of PLIN3 within cells expressing ionized calcium-binding adaptor molecule 1 (IBA1) showed that innate immune myeloid cells are not the main source of PLIN3 in either the cortex or corpus callosum of mice from the Ctrl groups (Figures 5E–5J). Partial depletion of innate immune cells by PLX led to increased PLIN3 content in IBA1+ cells from the corpus callosum, but not the cortex, of female E4/E4 mice (Figure 5F and 5I). In male E4/E4 mice, the innate immune myeloid turnover induced by PLX resulted in increased PLIN3 levels in cortical IBA1+ cells, which became the main source of PLIN3 in that brain region (Figure 5G).
The core contribution of oligodendrocytes and their myelin sheaths to alterations in brain lipid composition10,39,40 led us to inquire about the effects of E4 expression and PLX-induced innate immune depletion on the densities of oligodendrocytes and oligodendrocyte precursor cells (OPCs) in grey or white matter regions (Figures S12A–S12F). All groups of female and male mice presented similar densities of cortical OPCs (Figures S12A–S12C), and of oligodendrocytes and OPCs in the corpus callosum, regardless of APOE genotype or exposure to PLX (Figures S12D–S12F). However, and somewhat unexpectedly, partial depletion of innate immune cells by PLX resulted in increased densities of oligodendrocytes in the cortex of E4/E4 mice, but not of E3/E3 mice, irrespectively of sex (Figures S12A–S12C).
Innate immune suppression incites sex-specific effects on neuroinflammation and cognitive function of E4/E4 mice
We next sought to evaluate the impact of E4 expression on the levels of inflammatory cytokines and chemokines in the forebrain of middle-aged females and males (Figures 6A and 6B, and Supplemental Table S3). Relative to E3/E3 females, E4/E4 females showed a generalized increase in the levels of inflammatory cytokines and chemokines, namely in IL-1α, IL-1β, IL-6, IL-16, IL-22, IL-33, CCL2, CCL4, CCL5, CCL11, CCL17, CCL22, CXCL1, CXCL2, and CXCL9 (Figure 6A). Conversely, overall, E3/E3 and E4/E4 males presented a similar inflammatory profile in the forebrain, despite minor increases in the levels of CXCL1 and CXCL2, and lower levels of CXCL11 and HGF in the male E4/E4 group (Figure 6B). Exposure to PLX was linked to reduced levels of IL-1α, IL-16, and CCL12, and increased levels of IL-22, regardless of sex or APOE genotype (Figures 6A and 6B). Besides the changes in the afore-mentioned cytokines, we also detected lower levels of programmed death-ligand 2 in the female E4/E4 PLX group (Figure 6A), whereas higher levels of CSF1, CCL2, CCL4, HGF, and FGF21 were detected in the male E4/E4 PLX group (Figure 6B), when compared to the respective sex-matched E4/E4 Ctrl groups.
Figure 6. Neuroinflammation and cognitive function are ameliorated in E4/E4 females, yet worsened in E4/E4 males, upon innate immune suppression.
(A and B) Heatmaps showing the levels of inflammatory proteins in the cervical LNs isolated from females in (A), or males in (B), of each group (scale bar shows the fold change relative to each sex-matched E3/E3 Ctrl group). Asterisks highlight proteins altered between the E4/E4 PLX and E4/E4 Ctrl groups in either females or males. Programmed death-ligand 2 (PD-L2), fibroblast growth factor 21 (FGF21).
(C) Image of the brain hippocampal Cornu ammonis 1 (CA1) stained for CD31 (green) and glial fibrillary acidic protein (GFAP, red). 3D renderings depict the perivascular GFAP+ astrocytes (upper panel) and the non-perivascular GFAP+ astrocytes (lower panel).
(D and E) Graphs showing the perivascular GFAP+ astrocytic volumes in (D), and the non-perivascular GFAP+ astrocytic volumes in (E).
(F–M) Graphs showing the total distance in the open field (in meters, m), percentage of distance in the center of the open field arena, and percentage of time freezing in the context or cued trials of the fear conditioning test in females in (F–I), or males in (J–M).
Data in (A and B) are presented as fold change relative to the E3/E3 Ctrl group; n = 5 mice per group; two-way ANOVA with Bonferroni’s multiple comparisons test performed between the E3/E3 Ctrl and E4/E4 Ctrl groups, and the E4/E4 Ctrl and E4/E4 PLX groups for each sex. Data in (D and E) are presented as mean ± SEM; n = 5–6 mice per group; two-way ANOVA with Fisher’s LSD multiple comparisons test. Data in (F–M) are presented as mean ± SEM; n = 15 mice per group; two-way ANOVA with Bonferroni’s multiple comparisons test.
See also Supplemental Table S3.
Expression of E4 did not affect the morphology and localization of glial fibrillary acidic protein (GFAP)+ cells in the hippocampal Cornu ammonis 1 of middle-aged females or males (Figures 6C–6E), which points to similar levels of astrogliosis despite the differences in inflammatory mediators. However, after innate immune depletion, male E4/E4 mice presented a lower volume of perivascular GFAP, whereas female E4/E4 mice presented a higher volume of non-perivascular GFAP (Figures 6D and 6E). We interpreted these data as a lower engagement of GFAP+ astrocytes towards the vasculature in male E4/E4 mice, and as an activation of non-perivascular GFAP+ astrocytes in female E4/E4 mice, after depletion of BAMs and microglia.
We next evaluated the cognitive function of middle-aged E3/E3 and E4/E4 mice, exposed to either Ctrl or PLX diets, by subjecting the mice to the open field and fear conditioning behavior tests (Figures 6F–6M). While the female groups showed similar values in total distance traveled in the open field, a significant reduction in the distance traveled in the center of the open field arena was observed between E4/E4 Ctrl and E4/E4 PLX females (Figures 6F and 6G). We detected no differences between the groups of male mice in the open field test (Figures 6J and 6K). Females from the E4/E4 Ctrl group also showed a lower freezing time in the cued trial of the fear conditioning test, denoting a worse cognitive function when compared with their E3/E3 Ctrl group counterparts (Figures 6H and 6I). Of note, exposure to PLX led to a worse performance of female E3/E3 in the fear conditioning context trial (Figure 6H), and a better performance of female E4/E4 mice in the fear conditioning cued trial (Figure 6I). On the other hand, male E4/E4 mice, which showed similar freezing times as male E3/E3 mice in both the context and cued trials, presented reduced freezing in both trials of the fear conditioning test (indicative of a worse cognitive performance) after innate immune cell depletion by PLX (Figures 6L and 6M).
Collectively, the data presented so far support a strong effect of sex on the immune responses triggered by E4 expression at the brain meninges and cervical LNs of middle-aged mice. Suppressing the distinct sex-specific innate immune responses is beneficial for female E4/E4 cognition, but deleterious for male E4/E4 cognition.
APOE4 expression affects human brain leukocyte activation in a sex-dependent manner
We hypothesized that, as observed in the humanized E4/E4 mouse lines, human patients carrying at least one E4 allele would also show sexually dimorphic brain leukocyte responses. To test this, we started by collecting and integrating 6 publicly available single-nucleus RNA-seq (snRNA-seq) datasets of brain cells from AD and non-AD patients for a sufficient sample size to reliably identify brain leukocytes and focus on gene expression changes linked to sex, expression of APOE4, or both (Figure 7A).10,41–45 The integrated datasets included nuclei gene expression profiles from the prefrontal cortex, the whole temporal cortex, or just the hippocampus or entorhinal cortex (Figures S13A and S13B). The integrated brain nuclei split into 47 clusters, among which 4 clusters presented elevated expression of the leukocyte-specific gene protein tyrosine phosphatase receptor type C (PTPRC). The brain leukocyte nuclei–composed of clusters 1, 19, 28, and 35–were selected and reclustered for further analysis (Figures S13C–S13E). Within brain leukocyte nuclei, we could distinguish clusters corresponding to peripheral leukocytes, BAMs, and microglia 1–6 (Figure 7B). Brain peripheral leukocytes expressed the highest levels of PTPRC, CD3E, CD2, IL7R, FCN1, S100A8, and S100A9, suggesting that these encompassed a mixture of recruited adaptive immune cells (including T cells) and granulocytes (Figure S13F). Both brain BAMs and microglia depicted similar levels of PTPRC and ITGAM expression. However, brain BAMs expressed the highest levels of the characteristic genes MRC1, LYVE1, and CD163, yet low levels of P2RY12, SPP1, and CX3CR1. Conversely, the microglia clusters showed a generalized overexpression of P2RY12, SPP1, and CX3CR1, and lower expression of MRC1, LYVE1, and CD163, when compared with BAMs (Figure S13F). The clusters of peripheral leukocytes, BAMs, and microglia 1–3 and 5 were composed of nuclei originating from all datasets. The cluster of microglia 4, however, contained nuclei originating from only 4 out of 6 datasets, and microglia 6 was solely composed of nuclei from the “Morabito et al., 2021” dataset (Figure S13G). Based on these observations, we decided to include all leukocyte clusters for further analyses, except for microglia 6, which was a low frequency cluster derived from a single dataset. All the considered leukocyte clusters were composed of nuclei from the brains of patients of both sexes, and from both E4 carriers (expressing at least one E4 allele) and E4 non-carriers (expressing E2 and/or E3 alleles; Figures S14A and S14B). Moreover, all brain leukocyte clusters were represented in all four groups at comparable relative frequencies, namely the female E4 non-carriers, female E4 carriers, male E4 non-carriers, and male E4 carriers (Figure S14C).
Figure 7. Human brain leukocytes show a sexually dimorphic activation profile in response to E4 expression.
(A) Human brain single-nucleus RNA-sequencing (snRNA-seq) datasets were integrated and the leukocyte nuclei identified and reclustered based on the expression of the protein tyrosine phosphatase receptor type C (PTPRC) gene.
(B) UMAP representation of the 8 clusters of brain leukocyte nuclei and respective cluster annotations. Border-associated macrophages (BAMs).
(C and D) Total numbers of up- and down-regulated genes in the clusters of peripheral leukocytes, BAMs, and microglia 1–5 in female E4 carriers in (C), or male E4 carriers in (D).
(E and F) Top 10 gene set enrichment pathways altered in peripheral leukocytes in (E) and in BAMs in (F) of E4 carrier females compared to E4 non-carrier females.
(G) Heatmaps showing the scaled average expression levels of human leukocyte antigen (HLA)-DP, HLA-DQ, HLA-DR, HLA-DM, and HLA-DO genes, in the clusters of brain leukocytes.
See also Figures S13 and S14.
Differential gene expression analyses showed that sex or APOE4 are each associated with transcriptomic changes among the brain leukocyte populations (Figures S14D and S14E). However, examining the two covariates independently can mask the sex-specific effects of E4 expression on brain leukocyte transcriptomes. By comparing E4 carriers and E4 non-carriers within each sex, we found that the clusters of microglia 1 and 3 in female E4 carriers, and of microglia 1 and 4 in male E4 carriers showed the highest number of DEGs (Figures 7C and 7D, and Supplemental Table S4). Similarly to observations in the innate immune cells of E4 mice, gene set enrichment analysis using the DEGs found in microglia 1 showed that gene programs enriched in female versus male E4 carriers were distinct (Figures S14F and S14G). Expression of E4 induced inflammatory pathways in female microglia 1, such as “GO:0002697: regulation of immune effector process” and “R-HSA-6785807: interleukin-4 and interleukin-13 signaling”, whereas male microglia 1 presented alterations in pathways linked to neuronal support, such as “GO:0050804: modulation of chemical synaptic transmission” and “GO:0031175: neuron projection development” (Figures S14F and S14G). Curiously, the only functional pathway similarly affected by E4 expression in the two sexes was, once again, the “R-HSA-6798695: neutrophil degranulation” (as seen in the transcriptomic analyses involving the E4/E4 mice).
Of note, expression of E4 impacted the transcriptomes of the peripheral leukocytes, BAMs, and microglia 5 clusters in females, but not males (Figures 7E and 7F, and Supplemental Table S4). The DEGs in the peripheral leukocytes from female E4 carriers were involved in pathways like “GO:0045321: Leukocyte activation” and “R-HSA-198933: Immunoregulatory interactions between a lymphoid and a non-lymphoid cell” which denote increased activation of recruited peripheral leukocytes and signaling to brain cells (Figure 7E). In female E4 carrier BAMs, pathways like “R-HSA-3371568: Attenuation phase”, “R-HSA-3000171: Non-integrin membrane-ECM interactions”, and “R-HSA-2173782: Binding and uptake of ligands by scavenger receptors”, pointed to changes in the heat shock response, extracellular matrix remodeling, and phagocytosis (Figure 7F). A closer look at the list of DEGs in female E4 carrier BAMs showed that the human leukocyte antigen (HLA)-DRB1 was the most upregulated when compared to female E4 non-carriers (Supplemental Table S4), which is line with previous reports showing higher HLA expression in the brains of E4 carrier AD patients.45–47 Motivated by this, we evaluated the expression profiles of other classical (HLA-DP, HLA-DQ, and HLA-DR) and non-classical (HLA-DM and HLA-DO) HLA class II genes in all groups (Figure 7G). Expression of E4 led to a generalized upregulation of HLA class II genes in female BAMs, whereas in male brain BAMs the opposite effect was observed (Figure 7G). In the remaining male brain leukocytes, E4 expression led to modest changes in HLA class II gene expression. Higher levels of classical HLA class II genes were observed in the peripheral leukocytes of male E4 carriers, whereas microglia 2 nuclei showed an overall downregulation of gene expression similar to BAMs (Figure 7G). Contrarily, the female brain peripheral leukocytes and microglia 2–4 clusters showed higher expression levels of most HLA class II genes in E4 non-carriers, yet lower expression levels in E4 carriers (Figure 7G).
These data show that each leukocyte population from the human brain responds differently to E4 allele expression in females and males. Moreover, the often-contrasting HLA class II gene expression profiles observed in peripheral leukocytes, BAMs, and microglia from either female or male E4 carriers further corroborates the important interaction between sex and APOE4 in the context of antigen presentation and immune activation in the human brain.
DISCUSSION
To advance our understanding about the sex-specific effects of APOE4, we focused on the cellular responses at the brain–meningeal border, a neuroimmune interface that integrates stimuli arriving either from the brain, blood, or skull bone marrow.48–51
In agreement with previous reports,16,52,53 we show that macrophages express the highest levels of APOE mRNA transcripts in the meningeal dura and that females present more dural CD206+MHC-II+ macrophages than males, irrespectively of APOE genotype. However, the minor, despite statistically significant, decrease in dural APOE protein levels upon partial depletion of BAMs and microglia by PLX points to a limited contribution by CSF1R-expressing innate immune cells. Future experiments should test whether the pool of dural APOE is maintained by brain parenchymal astrocytes31,32,54, which secrete APOE into the CSF that can then be transported into the dura via arachnoid cuff exit points,51,55 or by hepatocytes,22,29 which secrete APOE into the bloodstream that might then be transported into the dura via fenestrated blood vessels.49,51 Regulation of APOE levels in the meningeal dura might represent a previously unappreciated therapeutic target in neurodegenerative diseases like AD.
Middle-aged male mice showed an expanded meningeal lymphatic vasculature in the presence of APOE4 that was correlated with reduced drainage of CSF into the cervical LNs. Similar outcomes have been observed before in male mice in response to inflammation triggered by neurotropic infections, where a supraphysiological meningeal lymphangiogenesis reduced the vessels’ capacity to drain CSF.35 Future studies should test if the LEC target genes Edn1, Fas, Ifit3b, and Myc, whose encoded proteins have been previously associated with processes like apoptosis, LEC proliferation, and lymphatic vessel contractility,56–60 are underlying the observed lymphatic vascular alterations in male E4/E4 mice. Except for the skin, we did not evaluate the lymphatics in other peripheral organs. The lymphatics at the extracranial nasal mucosa can have access to CSF that might cross the cribriform plate under certain abnormal conditions, namely higher intracranial pressure.61,62 The putative impact of sex or APOE4 on the nasal lymphatics should be evaluated in the future. Altogether, the data in this manuscript indicate that the effects of APOE4 on CNS immunity go beyond the brain parenchyma, spread all the way into the brain–meningeal border, and affect brain lymphatic drainage differently in middle-aged males and females. Yet it remains unclear whether the meningeal lymphangiogenesis and reduced lymphatic CSF outflow occurring in male E4/E4 mice is part of a protective compensatory mechanism that is absent in E4/E4 females.
Our results suggest that CSF lymphatic drainage can decrease even without morphological changes to lymphatic vessels (as seen in middle-aged E4/E4 females), and that both expansion and regression of lymphatic vessels at the dorsal dura can be linked to less CSF lymphatic drainage to the cervical LNs (as seen in middle-aged E4/E4 males before and after PLX). The effect of PLX on E4/E4 females suggests that other cell types, like SMCs at the extracranial lymphatic collectors,63 or changes in meningeal blood flow,16,25 might be underlying the reduction in CSF lymphatic outflow, and should be explored in future studies. The proteins encoded by Gadd45b, Igfbp3, and Plaur have been previously linked to changes in cycle arrest, apoptosis, and defects in lymphatic vascular valves, all processes that might be motivating the lymphatic vessel regression observed at the dorsal meningeal dura of middle-aged male E4/E4 after PLX.64–68 Evaluating the structure and functionality of valves at the basal collecting lymphatic vessels afferent to the cervical LNs will also be important, as changes in lymphatic valves might contribute to the discrepant rates of CSF lymphatic drainage shown by males and females in the presence of APOE4.63 Innate immune suppression was accompanied by a shift from tumor necrosis factor (encoded by the gene Tnf) to IFN-γ signaling in the meningeal LECs and BECs of male E4/E4 mice. Exacerbated IFN-γ signaling arising from T cells has been previously described in the aged meninges and linked to reduced lymphatic drainage of CSF into the cervical LNs.69–71 Future studies should aim at testing the relationship between heightened IFN-γ signaling and meningeal lymphatic vessel regression in the presence of APOE4.
Expression of APOE4 led to discrepant sex-specific immune activation profiles in the brain-draining cervical LNs. In fact, the cytokine and chemokine profiles detected in the cervical LNs and brains suggest that males, unlike females, are more resilient to APOE4-induced neuroinflammation. The increase in IL-17F in the cervical LNs of middle-aged E4/E4 females is particularly curious, as heightened IL-17 signaling has been previously linked to cognitive decline in mice.72,73 The effects of PLX on brain inflammation were modest in female E4/E4 mice, but still able to normalize the levels of IL-1α, a proinflammatory cytokine whose signaling can impair cognition.74,75 However, PLX had the opposite effect in male E4/E4 mice, reducing the chemokine levels in the cervical LNs, and elevating the concentrations of proinflammatory chemokines in the brain to levels comparable to female E4/E4 mice.
In line with a previous study,20 we show that middle-aged male E4/E4 mice are resilient to cognitive decline, while middle-aged female E4/E4 mice present early cognitive deficits. Cognitive resilience in E4/E4 males seems to be, at least in part, mediated by protective CNS innate immune surveillance, as partial depletion of BAMs and microglia by PLX led to increased neuroinflammation, impaired gliovascular crosstalk, and the appearance of memory impairment.
The presence of APOE4 in the brain led to sex-independent increases in BMP and Sulf lipids, enriched in endosomal/lysosomal membranes and myelin sheaths, and the cell membrane-associated glycerophospholipids PI, PG, and PE.27,30 Yet, our data suggest that the well-described effect of APOE4 on MhCer, AcylPG, and the lipid droplet-related BMP, TG, GB3, and CE neutral lipids27,30,76–80 is more pronounced in males, rather than females. The reduced levels of neutral lipid species in the brains of male E4/E4 mice after PLX may also point to deficits in myelination that, along with an abnormal accumulation of lipid droplets in cortical microglia and a reduced gliovascular engagement, might no longer favor cognitive resilience.10,13,27,30,40,76,78,79,81 Suppression of innate immunity in female E4/E4 mice led to better cognitive function and was accompanied by a normalization of Sulf and LacCer lipid species, and drops in MG, DG, and CE lipids to somewhat infraphysiological levels, which might imply lessened cellular intracellular signaling and endolysosomal activity.27,30,76,82
Leukocytes from the female and male human brain respond differently to APOE4. Previous publications have reported higher numbers of effector memory CD4+ T cells and proinflammatory cytokines like IFN-γ in the blood of E4 carriers.83,84 However, these studies did not explore putative sex differences. Our data indicates that APOE4 has a more prominent effect on the activation of brain BAMs and peripheral leukocytes in females than in males. However, we were underpowered to efficiently profile individual subpopulations of peripheral leukocytes. The sex-specific effects of APOE4 on innate immune HLA class II gene expression suggest that adaptive immune cells in the human brain, namely helper T cells, whose activation can be dependent on antigen presentation via HLA class II, might present distinct activation profiles in females and males. It will be important to contemplate the interaction between sex and APOE in future studies focusing on the roles of brain innate and adaptive immunity in AD.
Collectively, the data presented herein strengthen the need to consider the meningeal dura as a relevant neuroimmune niche when it comes to aging-related neurodegenerative diseases whose risk varies according to age, sex and E4 genotype, like AD and Lewy body dementia.1,5,85,86 Determining the nature and roles of the exact innate immune mechanisms regulating meningeal lymphatic vessel plasticity and cognition in females and males will be essential to develop tailored immunotherapeutic approaches that can effectively fend off cognitive decline due to APOE4 in each sex.
RESOURCE AVAILABILITY
Lead contact
All information and requests for further resources and reagents should be directed to and will be fulfilled by the lead contact: Sandro Da Mesquita (damesquita@mayo.edu).
Materials availability
This study did not generate new unique reagents.
Data and code availability
The scRNA-seq data reported in this manuscript have been deposited at GEO and are publicly available as of the date of publication. The accession number is listed in the key resources table.
Analyzed mouse meningeal dural scRNA-seq data are also accessible for consultation through shinyapp at https://fryerlab.shinyapps.io/meningeal_dura_scRNAseq/
Original code is reported at the Github repository at: https://github.com/kenneditodd/Da_Mesquita (mouse scRNA-seq analysis code); DOI: https://doi.org/10.5281/zenodo.18700908.
https://github.com/amoghc94/Da_Mesquita_APOE4 (human snRNA-seq analysis code); DOI: https://doi.org/10.5281/zenodo.18687961.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
KEY RESOURCES TABLE
| REAGENT or RESOURCE | SOURCE | IDENTIFIER |
|---|---|---|
| Antibodies | ||
| Rat anti-mouse LYVE-1-eF660, clone ALY7 | Thermo Fisher Scientific | Cat#50-0443-82 |
| Rat anti-mouse I-A/I-E-BV421, clone M5/114.15.2 | BioLegend | Cat#107631 |
| Rabbit anti-mouse LYVE-1 | AngioBio | Cat#11-034 |
| Goat anti-mouse CD31 | R&D Systems | Cat#AF3628 |
| Goat anti-mouse Podocalyxin | R&D Systems | Cat#AF1556 |
| Rat anti-mouse CD206, clone MR5D3 | Bio-Rad Laboratories | Cat#MCA2235 |
| Rabbit anti-mouse hAPOE | Cell Signaling Technology | Cat#13366S |
| Goat anti-mouse IBA1 | Abcam | Cat#ab5076 |
| Rat anti-mouse IBA1 | Abcam | Cat#ab5076 |
| Guinea pig anti-mouse PLIN3 | Progen | Cat#GP37 |
| Armenian hamster anti-mouse CD31 | Sigma-Aldrich | Cat#MAB1398Z |
| Chicken anti-mouse GFAP | Abcam | Cat#ab4674 |
| Mouse anti-mouse Quaking 7, clone CC1 | Sigma-Aldrich | Cat#OP80 |
| Goat anti-mouse PDGFRα | R&D Systems | Cat#AF1062 |
| Donkey anti-goat AF488 | Thermo Fisher Scientific | Cat#A-11055 |
| Donkey anti-rabbit AF488 | Thermo Fisher Scientific | Cat#A-21206 |
| Donkey anti-chicken AF488 | Thermo Fisher Scientific | Cat#A78948 |
| Donkey anti-goat AF594 | Thermo Fisher Scientific | Cat#A-11058 |
| Donkey anti-mouse AF594 | Thermo Fisher Scientific | Cat#A-21203 |
| Donkey anti-rabbit AF594 | Thermo Fisher Scientific | Cat#A-21207 |
| Donkey anti-rabbit AF594 | Jackson Immunoresearch | Cat#711-585-152 |
| Donkey anti-rat AF647 | Thermo Fisher Scientific | Cat#A-48272 |
| Donkey anti-goat AF647 | Thermo Fisher Scientific | Cat#A-21447 |
| Donkey anti-rabbit AF647 | Thermo Fisher Scientific | Cat#A-31573 |
| Donkey anti-goat AF647 | Jackson Immunoresearch | Cat#705-605-147 |
| Rat anti-mouse CD16/32, clone 93 | BioLegend | Cat#101302 |
| Zombie Aqua Fixable Viability Dye | BioLegend | Cat#423102 |
| Rat anti-mouse CD45-PerCP-Cy5.5, clone 30-F11 | BD Biosciences | Cat#550994 |
| Armenian hamster anti-mouse CD11c-BV605 | BioLegend | Cat#117334 |
| Rat anti-mouse CD11b-PE-Cy7 | BD Biosciences | Cat#552850 |
| Rat anti-mouse CD206-PE-Dazzle594 | BioLegend | Cat#141732 |
| Rat anti-mouse MHC-II-AF647 | BD Biosciences | Cat#562367 |
| Rat anti-mouse FOXP3-PE | Thermo Fisher Scientific | Cat#12-5773-82 |
| Rat anti-mouse Ly6C-PE | BioLegend | Cat#128008 |
| Rat anti-mouse F4/80-FITC | BioLegend | Cat#123108 |
| Rat anti-mouse CD19-APC-Cy7, clone 6D5 | BioLegend | Cat#115530 |
| Hamster anti-mouse TCRb-BV711, clone H57-597 | BD Biosciences | Cat#563135 |
| Rat anti-mouse CD4-FITC | BioLegend | Cat#100406 |
| Rat anti-mouse CD8-PB | BD Bioscience | Cat#558106 |
| Rat anti-mouse CD31-FITC | Thermo Fisher Scientific | Cat#11-0311-82 |
| Experimental models: Organisms/strains | ||
| C57BL/6 | Taconic Biosciences | Taconic: B6 |
| B6.129P2-Apoetm1Unc N11 | Taconic Biosciences | Taconic: APOE |
| C57BL/6NTac-Apoe<tm4206.1(APOE*C130,*R176)Tac> | Taconic Biosciences | Taconic: CureAlz huAPOE3 |
| C57BL/6NTac-Apoe<tm4207.1(APOE*R130,*R176)Tac> | Taconic Biosciences | Taconic: CureAlz huAPOE4 |
| Chemicals | ||
| PLX5622 | Chemgood | Cat#C-1521 |
| DirectPCR (tail) | Viagen | Cat#102-T |
| Ultrapure water | Apex Bioresearch Products | Cat#20-102 |
| Phosphate buffered saline (PBS) 10×, pH 7.4 | Fisher BioReagents | Cat#BP-399 |
| Heparin | Fisher BioReagents | Cat#BP2425 |
| 4% PFA in 1× PBS | Boster Bio | Cat#AR1068 |
| Sucrose | Cargill | Cat#62-112 |
| Tissue-Plus O.C.T. Compound | Fisher HealthCare | Cat#4585 |
| Gelatin | Sigma-Aldrich | Cat#G1890 |
| Sodium Azide 1% | G-Biosciences | Cat#786-750 |
| Triton X-100 | Sigma-Aldrich | Cat#X100 |
| Bovine serum albumin | Genesee Scientific | Cat#25-529 |
| Citrate buffer 10× | Sigma-Aldrich | Cat#C9999 |
| 4,6-diamidino-2-phenylindole (DAPI) | Thermo Fisher Scientific | Cat#62248 |
| Epredia Immu-Mount™ | Thermo Fisher Scientific | Cat#9990402 |
| Halt protease inhibitor cocktail | Thermo Fisher Scientific | Cat#78430 |
| Phosphatase inhibitor cocktail | Cell Signaling Technology | Cat#5870 |
| Phenylmethylsulfonyl fluoride | Cell Signaling Technology | Cat#8553 |
| Ethanol | DeconLabs | Cat#2701 |
| Pierce RIPA Buffer | Thermo Fisher Scientific | Cat#89901 |
| RPMI 1640 | Genesee Scientific | Cat#25-506 |
| DNase I | Sigma-Aldrich | Cat#11284932001 |
| Collagenase VIII | Sigma-Aldrich | Cat#C2139 |
| Collagenase D | Sigma-Aldrich | Cat#11088866001 |
| Fetal bovine serum (FBS, heat inactivated) | Corning | Cat#35-011-CV |
| RNAscope Target Probe APOE-C1 | Advanced Cell Diagnostics | Cat#433091-C1 |
| RNAscope Target Probe APOE-C3 | Advanced Cell Diagnostics | Cat#313271-C3 |
| RNAscope Target Probe Itgam-C2 | Advanced Cell Diagnostics | Cat#311491-C2 |
| RNAscope Target Probe Pecam1-C1 | Advanced Cell Diagnostics | Cat#316721-C1 |
| RNAscope Target Probe Pecam1-C3 | Advanced Cell Diagnostics | Cat#316721-C3 |
| Opal 520 reagent | Akoya Biosciences | Cat#FP1487001KT |
| Opal 620 reagent | Akoya Biosciences | Cat#FP1495001KT |
| Opal 690 reagent | Akoya Biosciences | Cat#FP1497001KT |
| RNAscope® Multiplex TSA buffer | Advanced Cell Diagnostics | Cat#322809 |
| Non-acetylated BSA | Thermo Fisher Scientific | Cat#AM2618 |
| Trypan blue | Gibco | Cat#15250061 |
| Actinomycin-D | Sigma-Aldrich | Cat#A1410 |
| Dimethyl sulfoxide (DMSO) | Sigma-Aldrich | Cat#276855 |
| UltraPure ethylenediaminetetraacetic acid (EDTA) 0.5 M | Thermo Fisher Scientific | Cat#15575020 |
| Commercial Assays | ||
| Anti-FITC Microbeads kit | Miltenyi Biotec | Cat#130-048-701 |
| Chromium Next GEM Single Cell 3’ GEM Kit v3.1 | 10× Genomics | Cat#PN-1000123 |
| Library Construction Kit | 10× Genomics | Cat#PN-1000190 |
| Chromium Next GEM Single Cell 3’ Gel Bead Kit v3.1 | 10× Genomics | Cat#PN-1000122 |
| Chromium Next GEM Chip G Single Cell Kit | 10× Genomics | Cat#PN-1000127 |
| Dual Index Kit TT Set A | 10× Genomics | Cat#PN-1000215 |
| RNAscope® Multiplex Fluorescent Reagent Kit v2 | Advanced Cell Diagnostics | Cat#323100 |
| Micro BCA Protein Assay Kit | Thermo Fisher Scientific | Cat#23235 |
| Pierce BCA Protein Assay Kit | Thermo Fisher Scientific | Cat#23227 |
| Human APOE ELISA Kit | Abcam | Cat#ab108813 |
| Foxp3/Transcription factor staining buffer set | Thermo Fisher Scientific | Cat#00-5523-00 |
| ZymoPURE Plasmid Miniprep Kit | Zymo Research | Cat#D4209 |
| ZymoPURE II Plasmid Maxiprep Kit | Zymo Research | Cat#D4203 |
| Deposited Data | ||
| scRNA-seq (mouse meningeal dura) | This paper | GEO: GSE295612 |
| snRNA-seq (human brain) | Mathys et al., 20191 | Synapse: Syn18485175 |
| snRNA-seq (human brain) | Morabito et al., 20212 | Synapse: Syn26670419 |
| snRNA-seq (human brain) | Blanchard et al., 20223 | Synapse: Syn38120890 |
| snRNA-seq (human brain) | Fujita et al., 20244 | Synapse: Syn31512863 |
| scRNA-seq (human brain) | Mathys et al., 20245 | Synapse: Syn52293417 |
| snRNA-seq (human brain) | Li et al., 20256 | GEO: GSE237718 |
| Software and Algorithms | ||
| FIJI version 2.3.0 | Schindelin et al.7 | https://imagej.net/software/fiji/downloads |
| Simple Neurite Tracer (FIJI plugin) version 1.53q | Arshadi et al.8 | https://imagej.net/plugins/snt |
| R Statistical Software versions 4.1.2 and 4.4.2 | R Foundation for Statistical Computing | https://www.r-project.org |
| ImSpector Pro microscope controller version 7 | Miltenyi/LaVision Biotec | https://www.miltenyibiotec.com/US-en/about-us/miltenyi-biotec-companies/lavision-biotec-gmbh |
| Imaris versions ×64 10.0.1 and 10.2.0 | Oxford Instruments, Bitplane | https://imaris.oxinst.com/versions/10 |
| Freezeframe version 4.104 | Actimetrics | https://actimetrics.com/products/freezeframe |
| AnyMaze version 7.8 | Stoelting | https://www.any-maze.com |
| FlowJo version 10.10.0 | BD Biosciences | https://www.flowjo.com/flowjo10/overview |
| Prism version 10.4.1 | GraphPad software | https://www.graphpad.com/updates/prism-10-4-1-release-notes |
| 10× Genomics Cell Ranger version 7.1.0 | 10× Genomics | https://www.10xgenomics.com/software |
| Seurat v5 versions 5.0.2 and 5.3.0 | R toolkit | https://github.com/satijalab/seurat/releases |
| MAST version 1.26.0 | Finak et al.9 | https://www.bioconductor.org/packages/devel/bioc/html/MAST |
| Nichenetr (R package) version 2.0.4 | GitHub, Browaeys et al.10 | https://github.com/saeyslab/nichenetr |
| RStudio versions 1.4.1103 and 4.3.0 | Posit | https://forum.posit.co/t/rstudio-1-4-1103-desktop/116542 |
| Harmony version 1.2.3 | Korsunsky et al.11 | https://cran.r-project.org/web/packages/harmony |
| biomaRt version 2.62.1 | Durinck et al.12 | https://bioconductor.org/packages/release/bioc/html/biomaRt |
| dplyr version 1.1.4 | dplyr.tidyverse.org | https://cran.r-project.org/web/packages/dplyr |
| ggplot2 version 3.5.2 | ggplot2.tidyverse.org | https://cran.r-project.org/web/packages/ggplot2 |
| Metascape versions 3.5.20250101 and 3.5.20250701 | Zhou et al.13 | https://metascape.org |
| Custom code used for mouse scRNA-seq data analysis | This paper | DOI: https://doi.org/10.5281/zenodo.18700908 |
| Custom code used for human snRNA-seq data analysis | This paper | DOI: https://doi.org/10.5281/zenodo.18687961 |
| Other | ||
| PLX5622 supplemented diet (600 p.p.m.) with blue dye | Research diets | Cat#D21102810 |
| PicoLab Rodent Diet 20 | LabDiet | Cat#5053 |
STAR METHODS
EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS
Mice of the Apoe+/+ (C57BL/6) and Apoe−/− (B6.129P2-Apoetm1Unc N11) lines were purchased from Taconic Biosciences. Humanized mice expressing two alleles of either APOE3 (C57BL/6NTac-Apoe<tm4206.1(APOE*C130,*R176)Tac>; “CureAlz huAPOE3” model) or APOE4 (C57BL/6NTac-Apoe<tm4207.1(APOE*R130,*R176)Tac>; “CureAlz huAPOE4” model) instead of the murine Apoe alleles were generously provided by the Cure Alzheimer’s Fund via Taconic Biosciences. All mice used in the experiments described in this manuscript were bred and aged in-house at the Mayo Clinic Jacksonville vivarium facilities. Littermate mice from different cages were randomly assigned to different experimental groups to prevent bias, and maintained under standard housing conditions, in ventilated cages, on a 12-hour light/dark cycle (lights on at 6 a.m.), in a temperature- and humidity-controlled environment. Mice were fed with standard diet (regular rodent chow, PicoLab Rodent Diet 20), unless stated otherwise, and water ad libitum. Both female and male mice were used in experiments at ages ranging between 2–4 (young adult) and 11–13 (middle-aged) months. The sexes, ages, and total numbers of mice used in each experimental group are stated in the figure legends. All animal procedures were approved by the Mayo Clinic Institutional Animal Care and Use Committee and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
METHOD DETAILS
Administration of PLX5622
Female and male mice were fed a diet supplemented with the CSF1R inhibitor PLX5622 (Chemgood) at 600 ppm (Research Diets), for either two or four weeks. Sex- and age-matched mice from the control groups were kept on a standard diet with a nutritionally similar composition but without PLX5622. According to the objective of each experiment, mice were kept under standard (control) or PLX5622 diets for a total of 2 or 4 weeks. Behavior tests (open field test followed by fear conditioning test) were performed at week 3 of exposure to the PLX5622 diet (or standard diet as control) and mice were kept on the same diets until the experimental endpoint at 4 weeks.
Open field test
Each mouse underwent a period of habituation to the handling by the experimenter for 3 consecutive days before the behavioral test, for a duration of approximately 10 minutes (min) each day. All mice were habituated to the open field behavior room, including the background white noise, for at least 1 hour before the beginning of the test. Each mouse was placed in the open field arena (40 cm in width × 40 cm in length × 30 cm in height) for a total of 15 min in standard room-lighting conditions. Activity in the open field was monitored by an overhead camera to track movement with AnyMaze software (Stoelting). Multiple parameters were assessed, including total distance traveled (in meters) and the percentage of distance traveled in the center of the arena (as % of total distance). Mice were returned to their home cage immediately following testing. The open field arena was cleaned with a 10% ethanol aqueous solution and allowed to air dry before the start of the next trial. The cohorts of female and male mice were tested on separate days.
Fear conditioning test
Mice were always subjected to the fear conditioning test at least one day after the open field test. All mice were habituated to the fear conditioning behavior room, including the background white noise, for at least 1 hour before the beginning of the test. The fear conditioning test was conducted in a sound-attenuated chamber with an electrified grid floor. On day 1 (the training day), mice were placed in the conditioning chamber and allowed to explore it for 3 min, immediately followed by three consecutive pairs of a 30-second 80 dB tone and a 2-second mild foot shock (0.5 mA). Each pair of tone plus shock was separated by 2 min. After the last tone-shock pair, mice were placed back into their respective home cages. On day 2, mice were first subjected to the context trial. Each mouse was placed in the conditioning chamber (under the exact same conditions as day 1) and allowed to explore it for 3 min, in the absence of tone or shock. After this time each mouse was returned to its respective home cage. Two hours later, the mice were subjected to the cued trial. The mice were first allowed to explore a conditioning chamber with a novel smell (vanilla extract), lighting, floor texture and color, and walls’ textures and colors, for a total of 3 min. Immediately after this habituation period, the mice received a continuous tone stimulus for a total of 3 min. After the cued trial (a total of 6 min), each mouse was returned to its respective home cage. Mouse behavior was recorded by a digital video camera mounted above each conditioning chamber and freezing was measured with the use of FreezeFrame software (Actimetrics). The individual performances in the fear conditioning test were denoted by the percentages of time freezing in the context trial (for the trial’s total duration of 3 min) and in the cued trial (for the trial’s last 3 min).
Intra-cisterna magna injections and in vivo or ex vivo imaging of cervical LNs
Mice were anesthetized via intraperitoneal (i.p.) injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). Under deep anaesthesia and before any surgical incision, mice were placed on a heating pad to maintain their body temperature at 37°C (monitored with the help of a rectal probe) and received a subcutaneous injection of carprofen (5 mg/kg). The skin was shaved at the front and back neck regions, as well as at the caudal dorsal skull, and cleaned 3 times with povidone-iodine (7.5%) and 70% ethanol. A drop of ophthalmic solution (Optixcare, CLC Medica) was placed on both eyes to prevent drying, and the mouse’s head was secured in a small stereotaxic frame, using non-traumatic ear bars, palate bar and nose clamp. After making a small skin incision (5–7 mm in length, longitudinally), the muscle layers were retracted and the atlantooccipital membrane of the cisterna magna was exposed. Using a Hamilton syringe (coupled to a 30-gauge needle), 5 μL of artificial CSF (Harvard Apparatus) containing OVA-AF555 (0.5 mg/mL; Thermo Fisher Scientific) were injected, by means of an automated injection pump, into the CSF-filled cisterna magna compartment at a rate of 2.5 μL per minute. After injecting, the syringe was left in place for 5 min to prevent backflow of the injected solution and CSF. After carefully withdrawing the syringe, the neck muscle and skin were sutured, and the mice were further prepared for in vivo imaging of the deep cervical LNs, or kept on heating pad for further ex vivo assessment of OVA-AF555 drainage into the deep or superficial cervical LNs.
For in vivo imaging, the mice were turned into supine position and secured with medical tape on top of a heating pad prior to the surgical procedure to expose the deep cervical LNs. A midline incision on the neck skin of approximately 2 cm was made to expose the salivary glands. The salivary glands were carefully retracted laterally with the help of sterile surgical hooks, followed by the omohyoid and sternomastoid muscles, exposing the deep cervical lymph nodes for imaging. We consistently imaged the deep cervical LN on the left side of the mouse. Alternatively, the deep cervical LN on the right side of the mouse was imaged if we were unable to locate the left deep cervical LN. Imaging under the Leica stereo microscope (M205 FCA with THUNDER Imager; Leica Microsystems) began consistently between 10–15 minutes after i.c.m. injection of the fluorescent OVA-AF555 tracer. Images of the deep cervical LNs were acquired every 20 seconds for a total of 1800 seconds. All mice were euthanized by i.p. injection of pentobarbital (150 mg/kg) followed by transcardial perfusion (see the next section for more details) immediately after in vivo imaging for tissue collection. The rates of OVA-AF555 drainage into the deep cervical LN (gain in OVA-AF555 mean fluorescent intensity) in the different mice were quantified using the FIJI image processing software. Briefly, the drainage rates were obtained by subtracting the mean fluorescence intensity (at a hand-drawn ROI around the LN) of the first acquired image (time point 0) to the mean fluorescence intensities obtained for all the remainder acquisition timepoints. To avoid including mice that showed low drainage rates due to putative technical artifacts associated with i.c.m. tracer injections, mice that showed an average gain in OVA-AF555 mean fluorescent intensity of less than 2500 for the total of the 90 acquired images were excluded from the analysis. For ex vivo imaging of the deep and mandibular superficial cervical LNs, the mice were maintained in prone position on a heating pad for 2 hours, and then euthanized and transcardially perfused (see the next section for more details). The LNs were immediately isolated and fixed in 4% PFA at RT for 1 hour prior to imaging under the Leica stereo microscope. The mean fluorescent intensities of OVA-AF555 in the images of the cervical LNs from mice of the different groups were quantified using the FIJI image processing software and used to calculate the ratios between OVA-AF555 drained into the deep or superficial cervical LNs.
Tissue collection, processing, and storage
Mice received an i.p. injection of pentobarbital (150 mg/kg) followed by transcardial perfusion. Under deep anesthesia, a ~2 cm incision was made along the midline to open the abdominal cavity, and the heart was exposed after lateral incisions on the diaphragm and ribs (~2.5 cm cuts on each side). A small incision (2–4 mm) was made in the heart’s right atrium, immediately followed by a controlled injection of approximately 20 mL of ice-cold 1x PBS (pH 7.4, prepared from 10x PBS; Fisher Bioreagents) containing heparin (10 U/mL; Sigma-Aldrich) into the left ventricle, using a 20-gauge needle. Tissues were collected immediately after euthanasia and processed and stored accordingly. Tissue samples collected for protein or lipid extractions were immediately collected, snap-frozen in dry ice and kept at −80°C. For immunofluorescence stainings and imaging, tissues were collected and placed in 4% paraformaldehyde (PFA) in 1x PBS (Boster Bio) for either 12–18 hours (dorsal skull caps with attached meningeal dura), or approximately 48 hours (in the case of ear and brain tissues). After fixation, the ear dermis and epidermis were carefully dissected with fine scissors and Dumont #7 forceps (Fine Science Tools) and transferred to a 24-well plate containing 1x PBS. The skull caps were further transferred to a 24-well plate containing 1x PBS and the meningeal dura was carefully peeled within 2 days using fine scissors and Dumont #5 and Dumont #7 forceps. Dural whole mounts and ear tissues were kept in 1x PBS at 4°C if used within 1 week, or in 1x PBS containing 0.001% sodium azide (G-Biosciences) at 4°C if stored for longer than 1 week. Brain tissue samples were placed in a preserving solution of 30% sucrose (Cargill) in 1x PBS and, after equilibrated (after dropping into the bottom of the tube), frozen in Tissue-Plus O.C.T. Compound (Fisher HealthCare). Brain sections were cut at a thickness of 50 μm in a Leica CM 3050 S cryostat (Leica Microsystems) and collected into a 24-well plate containing ice-cold 1x PBS. For long-term storage, brain slices were stored in 1x PBS containing 0.001% sodium azide (G-Biosciences) at 4°C.
Meningeal dura scRNA-seq
Each meningeal dural sample was peeled from the skull cap immediately after transcardial perfusion of the mouse with ice-cold heparinized 1x PBS supplemented with actinomycin-D (5 μM; Sigma-Aldrich). The dural tissue samples were collected into ice-cold RPMI 1640 (Genesee Scientific) containing actinomycin-D (15 μM, Sigma-Aldrich) and then further supplemented with collagenase VIII (1 mg/mL; Sigma-Aldrich), collagenase D (1 mg/mL; Sigma-Aldrich), and DNase I (50 U/mL; Sigma-Aldrich). After digestion for 25 min at 37°C, an equal volume of RPMI 1640 with 10% fetal bovine serum was added to the cell suspension, which was then filtered through a 45 μm cell strainer (Fisher Scientific). Ice-cold RPMI 1640 (15 mL) was flushed through the strainers into the filtered cellular suspensions, which were then centrifuged for 10 min at 320 × g at 4°C. Single-cell suspensions were then submitted to a sorting step to obtain higher frequencies of endothelial cells (namely LECs), in detriment of fibroblasts and immune cells. For that, the cell pellets from 3 samples of the same group were pooled and subjected to endothelial cell enrichment using MACS® Cell Separation technology and equipment (Miltenyi Biotec). Briefly, cells from each sample were incubated with anti-mouse CD31-FITC antibodies (1:100; Thermo Fisher Scientific), followed by incubation with Miltenyi anti-FITC Microbeads, and positive selection using LS columns and Quadromacs Separators with MACS® MultiStands, as per the manufacturer’s instructions. Single-cells were resuspended in sterile 1x PBS supplemented with 2% non-acetylated BSA and transferred into LoBind™ tubes (Eppendorf). After cell counting, approximately 5000 sorted cells per sample were loaded onto the Chromium Single Cell A Chip and run on a 10x Genomics Chromium Controller. Sequencing libraries were generated using the Chromium Next GEM Single Cell 3’ GEM Kit v3.1, the Library Construction Kit, and Dual Index Kit TT Set A (10x Genomics). After a cDNA library quality control step performed on an Agilent DNA 7500 system, libraries were sequenced on the Illumina NovaSeq 6000, using a NovaDeq S4 PE100 flow cell (paired-end sequencing). The 10x Genomics Cell Ranger v7.1.0 and mouse reference refdata-gex-mm10–2020A were used to process FASTQ files.93 After Cell Ranger processing, the filtered_feature_bc_matrix.h5 output file for every sample was read into R. The scRNA-seq analysis was performed using Seurat version 5.0.2.94 Cells were filtered out of the analysis if they had a total UMI count less than 500 or greater than 25000, a unique gene count less than 100, greater than 20% reads mapping protein coding mitochondrial genes, greater than 3% reads mapping to hemoglobin genes, and a cell complexity score log2(unique genes / total UMI) less than 0.8. Genes were excluded from the analysis if they were expressed in less than 10 cells. Protein coding mitochondrial genes were also excluded to enhance clustering. Cells were normalized for sequencing depth and variable genes were identified using SCTranform. Principal component analysis and UMAP dimensional reductions were performed. UMAP used the first 15 PCs as input. Clusters were identified using a resolution of 0.2. Cluster identities were assigned and merged based on expression of the following canonical murine genes: Pdgrfb, Acta2, Vwf, Cldn5, Pecam1, Flt4, Prox1, Lyve1, Ptprc, Cd19, Ms4a1, Ighd, Igha, Sdc1, Cd3e, Trbc2, Il7r, Nkg7, Klrb1b, Gata3, Rora, Itgax, H2-Eb1, Ccr2, Ly6c2, Lyz2, Ly6g, Itgam, Mrc1, Csf1r, Cd38, Mki67, Mcpt4, Ms4a2, Col1a2, and Plp1. Differential gene expression analysis was performed using MAST89 and P values were corrected with Bonferroni correction. DEGs with a q-value less than 0.05 were identified in volcano plots and compared in UpSet plots. Gene set enrichment analysis using the DEGs from each cluster was performed using Metascape.92 Ligand gene–target gene analysis was done using the R package Nichenetr.38 Genes in LECs with an average log2(fold change) > 2 were considered to generate gene sets for differential expression comparisons using NicheNet. Genes in BECs with an average log2(fold change) > 0.5 were considered to generate gene sets for differential expression comparisons using NicheNet. For the NicheNet analyses depicted as circos plots, the cluster of LECs (or alternatively of BECs) was used to determine the target genes and the clusters of BECs (or alternatively of LECs), B cells, T & NK cells, and macrophages were used to determine the main ligand genes in each group.
Tissue immunofluorescence staining and confocal microscopy
Free-floating fixed meningeal dural whole mounts, brain sections, or ear skin tissues were first incubated in a solution of 0.5% Triton X-100 (Sigma-Aldrich) in 1x PBS (PBS-T) for 30 min at room temperature (RT) in 24-well plates with constant agitation. The tissues were then incubated overnight (at 4°C with agitation) with a solution of PBS-T containing 1% bovine serum albumin (Prometheus Protein Biology Products) and different combinations of primary antibodies at the following specific concentrations: anti-mouse LYVE-1-eFluor660 (1:150; Thermo Fisher Scientific), anti-mouse I-A/I-E-BV421 (1:200; BioLegend), anti-mouse IBA1 (1:200; Abcam), anti-mouse CD206 (1:200; Bio-Rad Laboratories), anti-human APOE (1:200; Cell Signaling Technology), anti-mouse GFAP (1:500; Abcam), anti-mouse CD31 (1:50; Sigma-Aldrich), anti-mouse AQP4 (1:200; Sigma-Aldrich), anti-mouse PLIN3 (1:200; Progen), anti-mouse Quaking 7 (1:200; clone CC1, Sigma-Aldrich), anti-mouse PDGFRα (1:200; R&D Systems). After 3 consecutive 10-minute washes in PBS-T the tissues were incubated with the appropriate species-specific secondary antibodies (see key resources table for more details) in PBS-T for 1–2 hours at RT with agitation. Tissues were then washed once in PBS-T, whenever necessary, incubated with DAPI in 1x PBS (1 μg/mL, Thermo Fisher Scientific) for 10–20 min at RT, and finally washed twice in 1x PBS at RT (10 min each). The tissue samples were mounted onto Superfrost Plus slides (Fisher Scientific) with Epredia Immu-Mount™ (Thermo Fisher Scientific) and glass coverslips, dried at RT for at least 30 min and kept at 4°C in the dark until imaged.
All images were acquired using the Zeiss LSM 880 confocal microscope and ZenBlack software, or the Zeiss LSM 900 confocal microscope and ZenBlue software (Zeiss, Germany). Identical acquisition parameters and tile scans (with equal z steps) were used for all samples in all groups. Image analysis was performed using FIJI or Imaris (v10.2.0, Oxford Instruments). Cell densities were calculated after determining cell numbers (using the Cell Counter plug-in in FIJI) in the established regions of interest (ROI). Lymphatic vessel length in the meningeal dura or ear skin whole mounts was measured within a fixed ROI, using the FIJI plugin Simple Neurite Tracer, and further used to calculate the length of lymphatic vessels per area. The quantifications of CC1+PDGFRα− and PDGFRα+ cell densities were performed in FIJI. Independent thresholds were defined for the CC1, PDGFRα, and DAPI channels. The PDGFRα mask was then subtracted from the CC1 mask using the Image Calculator plug-in to isolate the CC1+PDGFRα− signal and then added to the DAPI+ signal. The resulting DAPI+CC1+PDGFRα− image was then processed with a watershed step to separate individual cells, followed by automated cell counting with the Analyze Particles plug-in (size: ≥9 μm2; circularity: 0–1). The PDGFRα+ cells were counted manually using FIJI on the same images used to quantify the CC1+PDGFRα− cells. Cell densities were then calculated and plotted.
Three-dimensional image analyses of IBA1+, PLIN3+, GFAP+, and CD31+ vessels in the mouse brain tissue were performed using Imaris (v10.2.0, Oxford Instruments). For each individual channel, surfaces were generated after applying smoothing and background subtraction to reduce noise and improve object (cell and/or vessel) detection. Intensity thresholds were set independently for each image while fully blinded to experimental group, using image-appropriate thresholds to ensure consistent detection across variable signal intensities. All smoothing, object-size filters, and analysis settings were kept constant across samples. Perivascular and non-perivascular astrogliosis were quantified by generating GFAP and CD31 surfaces and extracting the Imaris statistical parameter “Shortest Distance to Surface” between GFAP and CD31 objects. GFAP surfaces with a shortest-distance value of ≤ 5 μm from CD31 were classified as perivascular. The volume of PLIN3 within IBA1+ cells was measured by generating IBA1+ cell surfaces, masking the PLIN3 signal within each IBA1+ surface, and extracting the resulting PLIN3 volume associated with each cell. The total volume of PLIN3 was also measured and plotted. 3D renderings of representative images were generated to depict perivascular or non-perivascular GFAP+ astrocytes.
Tissue clearing and light sheet fluorescence microscopy
Mice heads were placed in 10% ethylenediaminetetraacetic acid (EDTA) and 4% PFA at RT for two weeks for decalcification (the decalcifying solution was replaced twice a week). Decalcified heads were cut along the horizontal axis with a razor blade and the bottom portion of head was cleared following the iDISCO+ protocol.95,96 Considering the size of the samples, the incubation times were increased as follows: 3 hours per methanol wash, 4 days in permeabilization and blocking solutions, 1 month in the solution with primary antibodies, and 1 month in the solution of secondary antibodies. The primary antibodies against murine proteins, and respective secondary antibodies, used for staining were the following: anti-mouse LYVE1 (1:800; AngioBio), anti-mouse CD31 (1:1000; R&D Systems); anti-mouse Podocalyxin (1:1000; R&D Systems), donkey anti-rabbit Alexa Fluor 594 (1:1000; Jackson ImmunoResearch), donkey anti-goat Alexa Fluor 647 (1:1000; Jackson ImmunoResearch). Each head sample was imaged in horizontal orientation with a light sheet fluorescence microscope (UltraMicroscope Blaze™, Miltenyi Biotec), using a 4×/0.6 objective lens, and the ImSpector Pro microscope controller software (version 7, Miltenyi/LaVision Biotec). The microscope chamber was filled with dibenzyl ether. A two-sided 3-sheet illumination configuration was used, with a 3-step dynamic focusing. The light sheet was generated by LED lasers tuned to 561 nm and 640 nm. The light-sheet numerical aperture was set to 0.06, and the sheet width to 50%. The following emission filters were used: 595/40 for Alexa Fluor 594, and 680/30 for Alexa Fluor 647. Stacks were acquired using 3-μm z steps and a 106.7-ms exposure time per step. Mosaic acquisitions were done with a 5% overlap on the full frame. Images acquired with the ImSpector acquisition software in TIFF format were converted to IMS files using the Imaris (Oxford Instruments) File Converter. Mosaics were reconstructed with Imaris Stitcher and the Imaris software (v. x64 10.0.1, Bitplane) was used to create the 3D renderings of the samples. After segmenting by hand, the volume of the lymphatic vessels in each ROI (at the cavernous, sigmoid, and petrosquamosal sinuses) was determined with the Surface tool of Imaris, using a surface detail of 5.42 μm and selecting background subtraction for thresholding (diameter of the largest sphere fitting into the object, 30 μm). Segmentation errors were manually corrected by deleting or fragmenting incorrectly segmented surfaces.
RNAscope on meningeal dural whole mounts
The day after fixation the meningeal dural tissues were mounted onto gelatin-coated (Sigma-Aldrich) SuperFrost slides and air-dried overnight at RT. The following morning, staining was performed using the target probes hAPOE-C3 or hAPOE-C1), Itgam-C2, and Pecam1-C2 or Pecam1-C3 and the RNAscope® Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics), according to the manufacturer’s instructions, except for the target epitope retrieval step, which was not performed. Slides were imaged using a Zeiss LSM 880 confocal microscope and ZenBlack software. Using the FIJI software, the number of cells positive for each target probe were determined at ROIs established in the vicinity of the dural superior sagittal sinus and further used to calculate cell densities.
Tissue protein extraction for hAPOE ELISA
Frozen samples were thawed and homogenized using an extraction buffer composed of RIPA lysis buffer (Thermo Fisher Scientific), Halt Protease Inhibitor Cocktail (1mM; Thermo Fisher Scientific), phosphatase inhibitors (1mM; Cell Signaling Technology), and Phenylmethylsulfonyl fluoride (1mM; Cell Signaling Technology). A total of 80 μL of extraction buffer was used for the meningeal dural samples, whereas 2.5 mL were used for hemi-forebrains. Samples were homogenized using handle pestles followed by ultrasonic homogenization on ice using a Sonifier™ SFX150 Cell Disruptor (Branson Ultrasonics). Samples were centrifuged for 20 min at 15,000 × g at 4°C, the supernatants were transferred to clean 1.5 mL Eppendorf tubes and stored at −80°C. The total protein concentrations were assessed using the Micro BCA Protein Assay Kit (Thermo Fisher Scientific) for each supernatant of meningeal dura, and the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) for each supernatant of forebrain, following the respective manufacturers’ instructions.
The concentrations of hAPOE in the whole protein extracts obtained from either meningeal dura or forebrain tissues were determined using the Human Apolipoprotein E ELISA Kit (Abcam), following the manufacturer’s instructions. The absorbances relative to each sample were measured at 450 nm using a BioTek Synergy microplate reader (Agilent Technologies) and the concentrations were calculated using standard curves for hAPOE.
Measurements of inflammatory proteins in tissues
One hemi-forebrain per mouse was dissected (immediately after transcardial perfusion), snap-frozen in dry-ice, and kept at −80°C until protein extraction. All the deep and mandibular superficial cervical LNs were dissected (immediately after transcardial perfusion), pooled, snapfrozen in dry-ice, and kept at −80°C until protein extraction. Protein extractions and quantifications were performed using the methodology described in the section above. All samples were diluted to the same final concentration in lysis buffer before shipment. Cytokine and chemokine protein levels were measured using the proximity extension immunoassay technology – Olink® Target 48 Mouse Cytokine Panel – via the services of Vanderbilt University Medical Center High-Throughput Biomarker Core. Data were received as concentration values in pg/mL which were used to calculate the fold change relative to the respective sex-matched E3/E3 Ctrl group average. The value of zero was attributed to samples for which a concentration was below the detection limit. Proteins were excluded from the analysis if two or more samples presented values below the detection limit in all groups. In Figure 4, the inflammatory proteins depicted in the heatmaps were significantly altered either in the E4/E4 Ctrl groups (when compared to their sex-matched E3/E3 Ctrl groups) or in the E4/E4 PLX groups (when compared to their sex-matched E4/E4 Ctrl groups). Asterisks point to inflammatory proteins whose levels were significantly different between the E4/E4 PLX and E4/E4 Ctrl groups in either females or males. In Figure 6, the inflammatory proteins depicted in the heatmaps were significantly altered either in the E4/E4 Ctrl groups (when compared to their sex-matched E3/E3 Ctrl groups) or in the E4/E4 PLX groups (when compared to their sex-matched E4/E4 Ctrl groups). Asterisks point to inflammatory proteins whose levels were significantly different between the E4/E4 PLX and E4/E4 Ctrl groups in either females or males. The raw data are included in Tables S1 and S3.
Flow cytometry
After perfusing mice with ice-cold 1x PBS with heparin, the deep cervical LNs and skull caps were immediately collected, and the meningeal dura samples dissected in ice-cold RPMI 1640 (Genesee Scientific). The tissues were digested for 25 min at 37°C in 1 mL of RPMI 1640 containing collagenase VIII (1 mg/mL; Sigma-Aldrich), collagenase D (1 mg/mL; Sigma-Aldrich), and DNase I (50 U/mL; Sigma-Aldrich). An equal volume of RPMI 1640 containing 10% fetal bovine serum (Corning Inc.) was added to the digested tissues to stop the enzymatic digestion, and the suspensions were filtered through a 70 μm cell strainer (Thermo Fisher Scientific). Ice-cold fluorescence-activated cell sorting (FACS) buffer (1x PBS with 1 mM EDTA and 1% BSA; pH 7.4) was flushed through the strainers into the filtered cellular suspensions, which were then centrifuged for 10 min at 320 × g at 4°C. The cell pellets were resuspended in fresh FACS buffer and transferred into U-shaped-bottom 96-well plates prior to staining for extracellular markers. Cells were incubated with anti-CD16/32 (1:200; BioLegend) in FACS buffer for 10 min at 4°C, immediately followed by an incubation step for 20 min at 4°C with the following antibodies recognizing murine surface proteins: anti-CD45-PerCP-Cy5.5 (1:200; BD Biosciences), anti-CD11c-BV605 (1:200; BioLegend), anti-CD11b-PE-Cy7 (1:200; BD Biosciences), anti-MHC-II-AF647 (1:200; BD Biosciences), anti-CD206-PE-Dazzle594 (1:200; BioLegend), anti-Ly6C-PE (1:200; BioLegend), anti-F4/80-FITC (1:200; BioLegend), anti-TCRb-BV711 (1:100; BD Biosciences), anti-CD19-APC-Cy7 (1:100; BioLegend), anti-CD4-FITC (1:200; BioLegend), and anti-CD8-PB (1:200; BD Biosciences). Cell viability was determined by using the Zombie Aqua Fixable Viability dye (BioLegend) following the manufacturer’s instructions. After staining for surface proteins, single-cell suspensions were fixed with 2% PFA and kept in FACS buffer at 4°C and protected from light until acquisition. Alternatively, if required, intracellular staining with anti-FOXP3-PE (1:100; Thermo Fisher Scientific) was performed immediately after the surface protein staining. Briefly, single-cells were fixed and permeabilized using the FOXP3/Transcription factor staining kit (Thermo Fisher Scientific), following the manufacturer’s instructions, and kept in FACS buffer at 4°C and protected from light until acquisition. Fluorescently labeled single-cell suspensions were acquired in an Attune NxT Flow Cytometer (Thermo Fisher Scientific). Data analysis was performed using the FlowJo software (BD Biosciences) and the different subpopulations of live leukocytes were gated as indicated in Figures S1 and S5.
Brain tissue lipidomics
Each sample consisted of one hemi-forebrain previously snap-frozen in dry-ice and kept at −80°C. Samples were shipped in dry ice to the Biomarkers Core Laboratory of the Irving Institute for Clinical and Translational Research, Columbia University Irving Medical Center, for lipid extraction and targeted lipidomics by liquid chromatography–mass spectrometry, using a LC-MS/MS platform comprised of Agilent 6490 Triple Quadrupole MS integrated with an Agilent 1260 Infinity LC system, according to the core’s established service protocols. From the raw data provided by the Biomarkers Core Laboratory, absolute lipid species levels (nmol/mg) were rounded to three decimal places and undetected lipid species were counted for each group. Specific lipid species that were below detection limit (not detected) in more than 30% of the group’s samples were excluded from the analysis. Absolute lipid class levels were recalculated based on the remaining species and all lipids were converted to percentage of the new total lipid level, which resulted in a total of 32 lipid classes and 466 lipid species. The percentage levels of the lipid species were then log2-transformed, scaled and centered for the principal component analysis. The first and second principal component were used for plotting and assessing group separation with an ellipsoid based on the 95% confidence interval for each group. For group comparisons, multiple pairwise Welch’s t-tests or Wilcoxon’s tests were performed, depending on normality assumption tested using Shapiro-Wilk’s test (considered non-normal distribution for P value ≤ 0.05). False-discovery rate was used to control for group multiple comparisons with α set to ≤ 0.05. Heatmaps are presented as fold-change of mean or median values depending on the usage of parametric or non-parametric tests for each comparison. In Figure 5, heatmaps depict the levels of all lipid species altered in the forebrains for the comparisons between the E4/E4 Ctrl and E3/E3 Ctrl groups, and between the E4/E4 PLX and E4/E4 Ctrl groups. Levels shown in the E3/E3 PLX, E4/E4 Ctrl, and E4/E4 PLX groups are presented as the log2(fold change) relative to their sex-specific E3/E3 Ctrl groups. Asterisks point to lipid species whose levels were significantly different between the E4/E4 PLX and E4/E4 Ctrl groups in either females or males. All brain lipidomics’ analyses were performed using RStudio Version 1.4.1103.
Integration and analyses of human brain snRNA-seq data
Datasets and study design
This study integrated snRNA-seq data from six independent datasets of postmortem human brain tissue, selected based on the availability of APOE genotype annotations and sex metadata to enable stratified analyses of myeloid cell populations. Five transcriptomic datasets originated from the Religious Orders Study and Rush Memory and Aging Project (ROSMAP) cohort, with one additional temporal cortex transcriptomic dataset generated at the Mayo Clinic.10,41–45
The first dataset (Mathys et al., 2019)41 consisted of single-nucleus profiles from prefrontal cortex of 48 individuals from ROSMAP. The cohort included 24 individuals with high levels of brain pathology and 24 individuals with no or very low brain pathological burden. A total of 67,341 nuclei were profiled after quality control. The second dataset (Morabito et al., 2021)42 comprised single-nucleus profiles from prefrontal cortex of individuals with late-stage AD (n=11). A total of 61,472 nuclei were retained after quality-control filtering. The third dataset (Blanchard et al., 2022)10 included single-nucleus profiles from prefrontal cortex tissue from 32 ROSMAP individuals balanced by sex, APOE genotype, and AD pathological diagnosis. A total of 172,659 nuclei were retained after quality control. The fourth dataset (Fujita et al., 2024)43 consisted of dorsolateral prefrontal cortex samples from 424 older participants in ROSMAP. Participants had a median of 3,824 nuclei each. We used the microglia.h5seurat object for our integration, containing a total of 86,610 nuclei after quality control. The fifth dataset (Mathys et al., 2024)44 represented a comprehensive snRNA-seq transcriptomic atlas covering six distinct anatomical brain regions from 48 ROSMAP participants (26 with pathologic diagnosis of AD, 22 without). A total of 283 post-mortem brain samples were profiled across multiple regions, though for our study we integrated only the entorhinal cortex and hippocampus datasets which totaled 385,691 nuclei after quality control. The sixth dataset (Li et al., 2025)45 included snRNA-seq data from temporal cortex samples from 56 cases retained after quality control, including 29 neuropathologically confirmed AD cases and 27 age, sex, and APOE-matched non-AD cases from the Mayo Clinic brain bank. A total of 425,963 nuclei were recovered after excluding low-quality nuclei and doublets. Datasets were prioritized for inclusion based on two primary criteria: first, the availability of APOE genotype information, for participant stratification to enable genotype-specific analyses; and second, the availability of sex to permit sex-stratified analyses of cell populations.
Data processing and integration using Harmony
Each dataset was initially processed using quality control parameters specific to the dataset characteristics and original study protocols. Quality control thresholds varied across datasets to account for differences in tissue processing, sequencing depth, and biological characteristics of the samples. For the Mathys et al., 201941 ROSMAP dataset, nuclei were retained if they contained between 200 and 4,000 genes with mitochondrial gene percentage below 10%. For the Morabito et al., 202142 dataset, nuclei were retained if they contained a minimum of 200 genes, and genes were required to be detected in a minimum of 3 cells. For the Blanchard et al., 202210 dataset, nuclei were retained if they contained between 200 and 6,000 genes with mitochondrial gene percentage below 10%. For the Fujita et al., 202443 dataset, nuclei were retained if they contained between 200 and 6,000 genes with mitochondrial gene percentage below 25%. For the Mathys et al., 202444 extended ROSMAP dataset, nuclei were retained if they contained between 200 and 6,000 genes with mitochondrial gene percentage below 10%. Finally, for the temporal cortex dataset from Li et al., 202545, nuclei were retained if they contained between 500 and 7,000 genes and had mitochondrial gene percentage below 20%.
To integrate the six datasets while preserving biological variation and enabling cross-dataset comparisons, we employed Harmony batch correction,90 which corrects for technical batch effects while maintaining biological heterogeneity. The integration workflow proceeded through six sequential steps. In the first step, each dataset was assigned a unique identifier in the metadata through a “dataset_name” column to track dataset origin after merging. In the second step, all six datasets were merged using Seurat’s merge function employing a union approach for gene handling. This method retained all genes detected across any dataset, with genes absent in specific datasets represented as zeros in the sparse matrix format. Cell identifiers were made unique by adding dataset names as prefixes to cell barcodes. This union-based approach was chosen to preserve all genes detected across datasets, maintaining maximum biological information rather than restricting to commonly detected genes across all studies. Unified brain region metadata were created based on dataset-specific annotations. In the third step, the merged object underwent standard Seurat preprocessing. Normalization was performed using the LogNormalize method with a scale factor of 10,000. Variable features were identified using variance-stabilizing transformation (vst), selecting 3,000 features for downstream analysis. Scaling was performed on the variable features only, and principal component analysis was computed using 50 principal components. In the fourth step, Harmony batch correction was applied to correct for dataset-specific batch effects. The batch variable was set to “dataset_name” to correct for dataset-specific technical variation, with correction applied in PCA space using the first 30 principal components. Convergence plotting was enabled to verify successful batch correction. Harmony iteratively adjusts cell positions in PCA space to mix cells from different datasets while maintaining biological structure, with the convergence plot confirming successful batch correction through decreasing objective function values across iterations. In the fifth step, downstream analyses were performed on the batch-corrected embeddings. UMAP visualization was computed using the Harmony-corrected embeddings based on dimensions 1 through 30. A nearest neighbor graph was constructed on the Harmony embeddings using the same dimensional range, and clustering was performed at multiple resolutions (0.4, 0.6, 0.8, and 1.0) to identify nuclei populations at different granularities.
Clinical metadata integration
Clinical and pathological metadata from ROSMAP were integrated with the snRNA-seq and scRNA-seq data to enable demographic- and genotype-stratified analyses. ROSMAP clinical data representing 3,584 participants were matched using the projid variable as the primary participant identifier. For datasets with complex barcode structures, dataset-specific mapping files were used to link cell barcodes to ROSMAP identifiers. A total of 38,208 nuclei/cells had overlapping transcriptomic and clinical data in the integrated object. Sex information was extracted from the ROSMAP msex variable, where values of 1 indicated male and values of 0 indicated female. Nuclei without ROSMAP linkage were labeled as “Unknown” for sex and excluded from sex-stratified analyses. APOE genotype was extracted from the apoe_genotype variable and categorized by E4 allele carrier status for stratified analyses. “E4 carriers” were defined as individuals expressing one or two E4 alleles. “E4 non-carriers” were defined as individuals without E4 alleles and expressing either one or two E3 or E2 alleles. This binary classification enabled comparison of APOE4-related transcriptional signatures across cell types and demographic groups. Different datasets required specialized approaches for clinical data integration due to variations in identifier systems and data structures. The Mathys et al., 201941 dataset employed direct projid matching to link cells to participant metadata. The Morabito et al., 202142 dataset, necessitated participant matching. The Fujita et al., 202443 immune cell-enriched samples required matching via the “individualID” variable from cell annotation files to ROSMAP identifiers. The Mathys et al., 202444 dataset required parsing of barcode suffix patterns, with suffixes −1 through −48 corresponding to specific sample identifiers. The Li et al., 202545 dataset required metadata extraction from the GEO series matrix file (GSE237718) with cross-referencing to ROSMAP data whenever possible.
Leukocyte nuclei identification, cluster annotation, and differential gene expression analysis
Brain leukocyte nuclei were isolated from the integrated dataset based on protein tyrosine phosphatase receptor type C (PTPRC) expression patterns observed at the cluster level using resolution 0.6. Cluster-level PTPRC screening was performed by visualizing its expression across all clusters using FeaturePlot, which identified the clusters 1, 19, 28, and 35 as having elevated PTPRC expression characteristic of leukocytes. Leukocyte nuclei were then extracted from the aforementioned clusters at resolution 0.6, reclustered, and further classified into distinct subtypes through manual annotation based on the expression of the CD3E, CD2, IL7R, ITGAM, FCN1, S100A8, S100A9, MRC1, LYVE1, CD163, P2RY12, SPP1, and CX3CR1 genes. Gene expression patterns were visualized using FeaturePlot and VlnPlot to confirm immune and myeloid identities. The identified brain leukocyte nuclei underwent independent clustering using the Harmony-corrected embeddings and dimensions 1 through 30, which identified 8 distinct subclusters annotated as peripheral leukocytes, BAMs, and microglia 1–6.
Differential gene expression analysis was performed to identify genes altered by sex (female versus female), APOE4 genotype alone (E4 carriers versus E4 non-carriers), and APOE4 genotype within each sex (E4 carriers versus E4 non-carriers within each sex), analyzed separately for each leukocyte cluster. The comparison structure involved within-sex contrasts between E4 carriers and E4 non-carriers, specifically comparing male E4 carriers versus male E4 non-carriers and female E4 carriers versus female E4 non-carriers. Cell type resolution was maintained by performing separate analyses for peripheral leukocytes, BAMs, and each of the six microglia clusters. Differential gene expression was tested using the Wilcoxon rank-sum test as implemented in Seurat’s FindMarkers function. A log2(fold change) > 0.25 was applied to focus on biologically meaningful differences. An expression frequency threshold required genes to be detected in at least 10% of cells in at least one comparison group to ensure sufficient expression for reliable detection. Multiple testing correction employed the Benjamini-Hochberg false discovery rate (FDR) adjustment to control the expected proportion of false discoveries among all discoveries, with correction applied separately within each cell type analysis. Genes were considered differentially expressed if they met all the following criteria: adjusted P value (FDR) ≤ 0.05, absolute log2(fold change) > 0.25, and detection in more than 10% of cells in at least one group (either E4 carriers or E4 non-carriers). These criteria ensured biological relevance through sufficient effect size, statistical significance through controlled FDR, and sufficient expression levels for reliable detection and interpretation. Results were visualized using complementary approaches to capture different aspects of the data structure and findings. Dot plot graphs illustrated the expression patterns of the top 5 differentially expressed genes per leukocyte cluster, with dot size representing the percentage of cells expressing each gene and dot color representing scaled average expression levels, stratified by cell cluster, sex, and APOE4 expression. Bar plots show the up-regulated and down-regulated gene counts stratified by leukocyte cluster, sex, and APOE4 expression, facilitating comparison of effect magnitudes across conditions. Leukocyte cluster-specific heatmaps display the scaled average expression of human leukocyte antigen (HLA) genes, namely HLA-DP, HLA-DQ, HLA-DR, HLA-DM, and HLA-DO genes, arranged in rows and the groups (E4 carriers and E4 non-carriers, stratified by sex) arranged in columns, using z-scored expression values for the color scale.
QUANTIFICATIONS AND STATISTICAL ANALYSIS
Sample sizes for each experiment were appropriately chosen based on standard power calculations (with α = 0.05 and power of 0.8) performed for similar experiments that were previously published by the authors. All quantifications and analyses were performed by a blinded experimenter. Microsoft Excel was used to record values and perform calculations in each experiment and all the statistical analyses were performed using Prism 10 (GraphPad Software). The Kolmogorov-Smirnov or Shapiro-Wilk tests were used to evaluate data distribution. Data in graphs are presented as mean ± SEM. The exact number of biological replicates per group (n) used in the experiments and are indicated in the figure legends. Two-tailed unpaired Student’s T test was used to compare two groups. Two-way ANOVA with Bonferroni’s or Fisher’s LSD multiple comparisons tests were used to analyze data involving multiple groups and two independent variables. Statistical tests used for in vivo CSF drainage, mouse scRNA-seq, human snRNA-seq, and lipidomics analyses are specified in the respective methodological sections. Statistical significance was considered for P values ≤ 0.05. The exact P values ≤ 0.05 are discriminated in all the graphs included in the figures.
Supplementary Material
Table S1. Concentrations of cytokines and chemokines measured in the cervical LNs of all the groups of female and male mice, related to Figure 4.
Table S2. Female and male forebrain lipidomics data and respective statistical analyses, related to Figures 5 and S11.
Table S3. Concentrations of cytokines and chemokines measured in the forebrains of all the groups of female and male mice, related to Figure 6.
Table S4. Lists of DEGs between E4 carriers and E4 non-carriers in each human brain leukocyte cluster, related to Figure 7.
Supplemental information can be found online at https://doi.org/
Document S1. Figures S1–S14.
Highlights.
APOE4 is linked with altered meningeal lymphatic function in females and males.
APOE4 elicits sex-specific immune responses in the brain, dura, and cervical LNs.
Suppression of innate immunity has distinct effects in E4/E4 females and E4/E4 males.
APOE4 expression leads to sexually dimorphic leukocyte activation in the human brain.
ACKNOWLEDGEMENTS
We are grateful to Dr. Laura J. Lewis-Tuffin (while she was heading the Cytometry and Cell Imaging Laboratory, Mayo Clinic Florida) for the support during image acquisition, and Aishe I. Kurti (Animal Behavior Core, Mayo Clinic Florida) for the support with murine behavior test equipment. We acknowledge Dr. Renu Nandakumar (Biomarkers Core Laboratory, Columbia University Irving Medical Center) for the lipidomic analysis services. We acknowledge Dr. Seiko Ikezu and Dr. Tsuneya Ikezu for their generosity in sharing access to the Imaris image analysis software with the members of the Da Mesquita lab during the revision experiments. We thank all members of the Department of Neuroscience at Mayo Clinic for valuable comments during discussions of this work. Schemes shown in figures were created in BioRender.com. This work was supported by grants from the BrightFocus Foundation (A2021025S), Cure Alzheimer’s Fund, NIH/NIA/Mayo Clinic Alzheimer’s Disease Research Center (P30 AG062677), Glaucoma Research Foundation and the Melza M. and Frank Theodore Barr Foundation (Catalyst For A Cure Initiative to Prevent and Cure Neurodegeneration), and NIH/NIA (1RF1AG080556-01A1), attributed to S.D.M.. This work was also supported by grants from the Venolymphatic (BBT.1300), attributed to M.S., and the NIH/NINDS (1R01NS130057-01), attributed to J.L.T..
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
DECLARATION OF INTERESTS
G.B. consults for SciNeuro Pharmaceuticals. S.D.M. is listed as an inventor in patent applications concerning modulating lymphatic vessels in neurological disease (University of Virginia Licensing & Ventures Group, and PureTech Ventures LLC). The authors have no other conflicts of interest to report.
REFERENCES
- 1.Yamazaki Y, Zhao N, Caulfield TR, Liu CC, and Bu G (2019). Apolipoprotein E and Alzheimer disease: pathobiology and targeting strategies. Nat Rev Neurol 15, 501–518. 10.1038/s41582-019-0228-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Yang LG, March ZM, Stephenson RA, and Narayan PS (2023). Apolipoprotein E in lipid metabolism and neurodegenerative disease. Trends Endocrinol Metab 34, 430–445. 10.1016/j.tem.2023.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chin AL, Negash S, and Hamilton R (2011). Diversity and disparity in dementia: the impact of ethnoracial differences in Alzheimer disease. Alzheimer Dis Assoc Disord 25, 187–195. 10.1097/WAD.0b013e318211c6c9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Farrer LA, Cupples LA, Haines JL, Hyman B, Kukull WA, Mayeux R, Myers RH, Pericak-Vance MA, Risch N, and van Duijn CM (1997). Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. JAMA 278, 1349–1356. [PubMed] [Google Scholar]
- 5.Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, Roses AD, Haines JL, and Pericak-Vance MA (1993). Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261, 921–923. [DOI] [PubMed] [Google Scholar]
- 6.Neu SC, Pa J, Kukull W, Beekly D, Kuzma A, Gangadharan P, Wang LS, Romero K, Arneric SP, Redolfi A, et al. (2017). Apolipoprotein E Genotype and Sex Risk Factors for Alzheimer Disease: A Meta-analysis. JAMA Neurol 74, 1178–1189. 10.1001/jamaneurol.2017.2188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kodama L, Guzman E, Etchegaray JI, Li Y, Sayed FA, Zhou L, Zhou Y, Zhan L, Le D, Udeochu JC, et al. (2020). Microglial microRNAs mediate sex-specific responses to tau pathology. Nat Neurosci 23, 167–171. 10.1038/s41593-019-0560-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yamazaki Y, Liu CC, Yamazaki A, Shue F, Martens YA, Chen Y, Qiao W, Kurti A, Oue H, Ren Y, et al. (2021). Vascular ApoE4 Impairs Behavior by Modulating Gliovascular Function. Neuron 109, 438–447 e436. 10.1016/j.neuron.2020.11.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Moser VA, Workman MJ, Hurwitz SJ, Lipman RM, Pike CJ, and Svendsen CN (2021). Microglial transcription profiles in mouse and human are driven by APOE4 and sex. iScience 24, 103238. 10.1016/j.isci.2021.103238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Blanchard JW, Akay LA, Davila-Velderrain J, von Maydell D, Mathys H, Davidson SM, Effenberger A, Chen CY, Maner-Smith K, Hajjar I, et al. (2022). APOE4 impairs myelination via cholesterol dysregulation in oligodendrocytes. Nature 611, 769–779. 10.1038/s41586-022-05439-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yin Z, Rosenzweig N, Kleemann KL, Zhang X, Brandao W, Margeta MA, Schroeder C, Sivanathan KN, Silveira S, Gauthier C, et al. (2023). APOE4 impairs the microglial response in Alzheimer’s disease by inducing TGFbeta-mediated checkpoints. Nat Immunol 24, 1839–1853. 10.1038/s41590-023-01627-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Liu CC, Wang N, Chen Y, Inoue Y, Shue F, Ren Y, Wang M, Qiao W, Ikezu TC, Li Z, et al. (2023). Cell-autonomous effects of APOE4 in restricting microglial response in brain homeostasis and Alzheimer’s disease. Nat Immunol 24, 1854–1866. 10.1038/s41590-023-01640-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Lee S, Devanney NA, Golden LR, Smith CT, Schwartz JL, Walsh AE, Clarke HA, Goulding DS, Allenger EJ, Morillo-Segovia G, et al. (2023). APOE modulates microglial immunometabolism in response to age, amyloid pathology, and inflammatory challenge. Cell Rep 42, 112196. 10.1016/j.celrep.2023.112196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chen X, Firulyova M, Manis M, Herz J, Smirnov I, Aladyeva E, Wang C, Bao X, Finn MB, Hu H, et al. (2023). Microglia-mediated T cell infiltration drives neurodegeneration in tauopathy. Nature. 10.1038/s41586-023-05788-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lopez-Lee C, Torres ERS, Carling G, and Gan L (2024). Mechanisms of sex differences in Alzheimer’s disease. Neuron 112, 1208–1221. 10.1016/j.neuron.2024.01.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Anfray A, Schaeffer S, Hattori Y, Santisteban MM, Casey N, Wang G, Strickland M, Zhou P, Holtzman DM, Anrather J, et al. (2024). A cell-autonomous role for border-associated macrophages in ApoE4 neurovascular dysfunction and susceptibility to white matter injury. Nat Neurosci 27, 2138–2151. 10.1038/s41593-024-01757-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lopez-Lee C, Kodama L, Fan L, Zhu D, Zhu J, Wong MY, Ye P, Norman K, Foxe NR, Ijaz L, et al. (2024). Tlr7 drives sex differences in age- and Alzheimer’s disease-related demyelination. Science 386, eadk7844. 10.1126/science.adk7844. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rosenzweig N, Kleemann KL, Rust T, Carpenter M, Grucci M, Aronchik M, Brouwer N, Valenbreder I, Cooper-Hohn J, Iyer M, et al. (2024). Sex-dependent APOE4 neutrophil-microglia interactions drive cognitive impairment in Alzheimer’s disease. Nat Med 30, 2990–3003. 10.1038/s41591-024-03122-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee BH, Cevizci M, Lieblich SE, and Galea LAM (2025). Sex-specific influences of APOEepsilon4 genotype on hippocampal neurogenesis and progenitor cells in middle-aged rats. Biol Sex Differ 16, 10. 10.1186/s13293-025-00694-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bour A, Grootendorst J, Vogel E, Kelche C, Dodart JC, Bales K, Moreau PH, Sullivan PM, and Mathis C (2008). Middle-aged human apoE4 targeted-replacement mice show retention deficits on a wide range of spatial memory tasks. Behav Brain Res 193, 174–182. 10.1016/j.bbr.2008.05.008. [DOI] [PubMed] [Google Scholar]
- 21.Riddell DR, Zhou H, Atchison K, Warwick HK, Atkinson PJ, Jefferson J, Xu L, Aschmies S, Kirksey Y, Hu Y, et al. (2008). Impact of apolipoprotein E (ApoE) polymorphism on brain ApoE levels. J Neurosci 28, 11445–11453. 10.1523/JNEUROSCI.1972-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Huynh TV, Wang C, Tran AC, Tabor GT, Mahan TE, Francis CM, Finn MB, Spellman R, Manis M, Tanzi RE, et al. (2019). Lack of hepatic apoE does not influence early Abeta deposition: observations from a new APOE knock-in model. Mol Neurodegener 14, 37. 10.1186/s13024-019-0337-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Shi Y, Yamada K, Liddelow SA, Smith ST, Zhao L, Luo W, Tsai RM, Spina S, Grinberg LT, Rojas JC, et al. (2017). ApoE4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature 549, 523–527. 10.1038/nature24016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Liu CC, Zhao N, Fu Y, Wang N, Linares C, Tsai CW, and Bu G (2017). ApoE4 Accelerates Early Seeding of Amyloid Pathology. Neuron 96, 1024–1032 e1023. 10.1016/j.neuron.2017.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Koizumi K, Hattori Y, Ahn SJ, Buendia I, Ciacciarelli A, Uekawa K, Wang G, Hiller A, Zhao L, Voss HU, et al. (2018). Apoepsilon4 disrupts neurovascular regulation and undermines white matter integrity and cognitive function. Nat Commun 9, 3816. 10.1038/s41467-018-06301-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Shi Y, Manis M, Long J, Wang K, Sullivan PM, Remolina Serrano J, Hoyle R, and Holtzman DM (2019). Microglia drive APOE-dependent neurodegeneration in a tauopathy mouse model. J Exp Med 216, 2546–2561. 10.1084/jem.20190980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Miranda AM, Ashok A, Chan RB, Zhou B, Xu Y, McIntire LB, Area-Gomez E, Di Paolo G, Duff KE, Oliveira TG, and Nuriel T (2022). Effects of APOE4 allelic dosage on lipidomic signatures in the entorhinal cortex of aged mice. Transl Psychiatry 12, 129. 10.1038/s41398-022-01881-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zhao N, Ren Y, Yamazaki Y, Qiao W, Li F, Felton LM, Mahmoudiandehkordi S, Kueider-Paisley A, Sonoustoun B, Arnold M, et al. (2020). Alzheimer’s Risk Factors Age, APOE Genotype, and Sex Drive Distinct Molecular Pathways. Neuron 106, 727–742 e726. 10.1016/j.neuron.2020.02.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Liu CC, Zhao J, Fu Y, Inoue Y, Ren Y, Chen Y, Doss SV, Shue F, Jeevaratnam S, Bastea L, et al. (2022). Peripheral apoE4 enhances Alzheimer’s pathology and impairs cognition by compromising cerebrovascular function. Nat Neurosci 25, 1020–1033. 10.1038/s41593-022-01127-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Litvinchuk A, Suh JH, Guo JL, Lin K, Davis SS, Bien-Ly N, Tycksen E, Tabor GT, Serrano JR, Manis M, et al. (2024). Amelioration of Tau and ApoE4-linked glial lipid accumulation and neurodegeneration with an LXR agonist. Neuron 112, 2079. 10.1016/j.neuron.2024.05.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Wang C, Xiong M, Gratuze M, Bao X, Shi Y, Andhey PS, Manis M, Schroeder C, Yin Z, Madore C, et al. (2021). Selective removal of astrocytic APOE4 strongly protects against tau-mediated neurodegeneration and decreases synaptic phagocytosis by microglia. Neuron 109, 1657–1674 e1657. 10.1016/j.neuron.2021.03.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Jackson RJ, Meltzer JC, Nguyen H, Commins C, Bennett RE, Hudry E, and Hyman BT (2022). APOE4 derived from astrocytes leads to blood-brain barrier impairment. Brain 145, 3582–3593. 10.1093/brain/awab478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bell RD, Sagare AP, Friedman AE, Bedi GS, Holtzman DM, Deane R, and Zlokovic BV (2007). Transport pathways for clearance of human Alzheimer’s amyloid beta-peptide and apolipoproteins E and J in the mouse central nervous system. J Cereb Blood Flow Metab 27, 909–918. 10.1038/sj.jcbfm.9600419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Bell RD, Winkler EA, Singh I, Sagare AP, Deane R, Wu Z, Holtzman DM, Betsholtz C, Armulik A, Sallstrom J, et al. (2012). Apolipoprotein E controls cerebrovascular integrity via cyclophilin A. Nature 485, 512–516. 10.1038/nature11087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Li X, Qi L, Yang D, Hao S, Zhang F, Zhu X, Sun Y, Chen C, Ye J, Yang J, et al. (2022). Meningeal lymphatic vessels mediate neurotropic viral drainage from the central nervous system. Nat Neurosci 25, 577–587. 10.1038/s41593-022-01063-z. [DOI] [PubMed] [Google Scholar]
- 36.Spangenberg E, Severson PL, Hohsfield LA, Crapser J, Zhang J, Burton EA, Zhang Y, Spevak W, Lin J, Phan NY, et al. (2019). Sustained microglial depletion with CSF1R inhibitor impairs parenchymal plaque development in an Alzheimer’s disease model. Nat Commun 10, 3758. 10.1038/s41467-019-11674-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Profaci CP, Harvey SS, Bajc K, Zhang TZ, Jeffrey DA, Zhang AZ, Nemec KM, Davtyan H, O’Brien CA, McKinsey GL, et al. (2024). Microglia are not necessary for maintenance of blood-brain barrier properties in health, but PLX5622 alters brain endothelial cholesterol metabolism. Neuron 112, 2910–2921 e2917. 10.1016/j.neuron.2024.07.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Browaeys R, Saelens W, and Saeys Y (2020). NicheNet: modeling intercellular communication by linking ligands to target genes. Nat Methods 17, 159–162. 10.1038/s41592-019-0667-5. [DOI] [PubMed] [Google Scholar]
- 39.das Neves SP, Delivanoglou N, Ren Y, Cucuzza CS, Makuch M, Almeida F, Sanchez G, Barber MJ, Rego S, Schrader R, et al. (2024). Meningeal lymphatic function promotes oligodendrocyte survival and brain myelination. Immunity 57, 2328–2343 e2328. 10.1016/j.immuni.2024.08.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wang N, Wang M, Jeevaratnam S, Rosenberg C, Ikezu TC, Shue F, Doss SV, Alnobani A, Martens YA, Wren M, et al. (2022). Opposing effects of apoE2 and apoE4 on microglial activation and lipid metabolism in response to demyelination. Mol Neurodegener 17, 75. 10.1186/s13024-022-00577-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Mathys H, Davila-Velderrain J, Peng Z, Gao F, Mohammadi S, Young JZ, Menon M, He L, Abdurrob F, Jiang X, et al. (2019). Single-cell transcriptomic analysis of Alzheimer’s disease. Nature 570, 332–337. 10.1038/s41586-019-1195-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Morabito S, Miyoshi E, Michael N, Shahin S, Martini AC, Head E, Silva J, Leavy K, Perez-Rosendahl M, and Swarup V (2021). Single-nucleus chromatin accessibility and transcriptomic characterization of Alzheimer’s disease. Nat Genet 53, 1143–1155. 10.1038/s41588-021-00894-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Fujita M, Gao Z, Zeng L, McCabe C, White CC, Ng B, Green GS, Rozenblatt-Rosen O, Phillips D, Amir-Zilberstein L, et al. (2024). Cell subtype-specific effects of genetic variation in the Alzheimer’s disease brain. Nat Genet 56, 605–614. 10.1038/s41588-024-01685-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Mathys H, Boix CA, Akay LA, Xia Z, Davila-Velderrain J, Ng AP, Jiang X, Abdelhady G, Galani K, Mantero J, et al. (2024). Single-cell multiregion dissection of Alzheimer’s disease. Nature 632, 858–868. 10.1038/s41586-024-07606-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Li Z, Martens YA, Ren Y, Jin Y, Sekiya H, Doss SV, Kouri N, Castanedes-Casey M, Christensen TA, Miller Nevalainen LB, et al. (2025). APOE genotype determines cell-type-specific pathological landscape of Alzheimer’s disease. Neuron 113, 1380–1397 e1387. 10.1016/j.neuron.2025.02.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Egensperger R, Kosel S, von Eitzen U, and Graeber MB (1998). Microglial activation in Alzheimer disease: Association with APOE genotype. Brain Pathol 8, 439–447. 10.1111/j.1750-3639.1998.tb00166.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Minett T, Classey J, Matthews FE, Fahrenhold M, Taga M, Brayne C, Ince PG, Nicoll JA, Boche D, and Mrc C (2016). Microglial immunophenotype in dementia with Alzheimer’s pathology. J Neuroinflammation 13, 135. 10.1186/s12974-016-0601-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Rego S, Sanchez G, and Da Mesquita S (2023). Current views on meningeal lymphatics and immunity in aging and Alzheimer’s disease. Mol Neurodegener 18, 55. 10.1186/s13024-023-00645-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Smyth LCD, and Kipnis J (2025). Redefining CNS immune privilege. Nat Rev Immunol. 10.1038/s41577-025-01175-0. [DOI] [PubMed] [Google Scholar]
- 50.Da Mesquita S, and Rua R (2024). Brain border-associated macrophages: common denominators in infection, aging, and Alzheimer’s disease? Trends Immunol 45, 346–357. 10.1016/j.it.2024.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Pinho-Correia LM, McCullough SJC, Ghanizada H, Nedergaard M, Rustenhoven J, and Da Mesquita S (2025). CSF transport at the brain-meningeal border: effects on neurological health and disease. Lancet Neurol 24, 535–547. 10.1016/S1474-4422(25)00115-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Van Hove H, Martens L, Scheyltjens I, De Vlaminck K, Pombo Antunes AR, De Prijck S, Vandamme N, De Schepper S, Van Isterdael G, Scott CL, et al. (2019). A single-cell atlas of mouse brain macrophages reveals unique transcriptional identities shaped by ontogeny and tissue environment. Nat Neurosci 22, 1021–1035. 10.1038/s41593-019-0393-4. [DOI] [PubMed] [Google Scholar]
- 53.Golden LR, Siano DS, Stephens IO, MacLean SM, Saito K, Nolt GL, Funnell JL, Pallerla AV, Lee S, Smith C, et al. (2025). APOE4 to APOE2 allelic switching in mice improves Alzheimer’s disease-related metabolic signatures, neuropathology and cognition. Nat Neurosci 28, 2461–2475. 10.1038/s41593-025-02094-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Boyles JK, Pitas RE, Wilson E, Mahley RW, and Taylor JM (1985). Apolipoprotein E associated with astrocytic glia of the central nervous system and with nonmyelinating glia of the peripheral nervous system. J Clin Invest 76, 1501–1513. 10.1172/JCI112130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Smyth LCD, Xu D, Okar SV, Dykstra T, Rustenhoven J, Papadopoulos Z, Bhasiin K, Kim MW, Drieu A, Mamuladze T, et al. (2024). Identification of direct connections between the dura and the brain. Nature 627, 165–173. 10.1038/s41586-023-06993-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Sakai H, Ikomi F, and Ohhashi T (1999). Effects of endothelin on spontaneous contractions in lymph vessels. Am J Physiol 277, H459–466. 10.1152/ajpheart.1999.277.2.H459. [DOI] [PubMed] [Google Scholar]
- 57.Spinella F, Garrafa E, Di Castro V, Rosano L, Nicotra MR, Caruso A, Natali PG, and Bagnato A (2009). Endothelin-1 stimulates lymphatic endothelial cells and lymphatic vessels to grow and invade. Cancer Res 69, 2669–2676. 10.1158/0008-5472.CAN-08-1879. [DOI] [PubMed] [Google Scholar]
- 58.Risso V, Lafont E, and Le Gallo M (2022). Therapeutic approaches targeting CD95L/CD95 signaling in cancer and autoimmune diseases. Cell Death Dis 13, 248. 10.1038/s41419-022-04688-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Dieterich LC, Ducoli L, Shin JW, and Detmar M (2017). Distinct transcriptional responses of lymphatic endothelial cells to VEGFR-3 and VEGFR-2 stimulation. Sci Data 4, 170106. 10.1038/sdata.2017.106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Alderfer L, Russo E, Archilla A, Coe B, and Hanjaya-Putra D (2021). Matrix stiffness primes lymphatic tube formation directed by vascular endothelial growth factor-C. FASEB J 35, e21498. 10.1096/fj.202002426RR. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Yoon JH, Jin H, Kim HJ, Hong SP, Yang MJ, Ahn JH, Kim YC, Seo J, Lee Y, McDonald DM, et al. (2024). Nasopharyngeal lymphatic plexus is a hub for cerebrospinal fluid drainage. Nature 625, 768–777. 10.1038/s41586-023-06899-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Papadopoulos Z, Smyth LCD, Smirnov I, Gibson DA, Herz J, and Kipnis J (2025). Differential impact of lymphatic outflow pathways on cerebrospinal fluid homeostasis. J Exp Med 222. 10.1084/jem.20241752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Du T, Raghunandan A, Mestre H, Pla V, Liu G, Ladron-de-Guevara A, Newbold E, Tobin P, Gahn-Martinez D, Pattanayak S, et al. (2024). Restoration of cervical lymphatic vessel function in aging rescues cerebrospinal fluid drainage. Nat Aging 4, 1418–1431. 10.1038/s43587-024-00691-3. [DOI] [PubMed] [Google Scholar]
- 64.Chen Z, Wan X, Hou Q, Shi S, Wang L, Chen P, Zhu X, Zeng C, Qin W, Zhou W, and Liu Z (2016). GADD45B mediates podocyte injury in zebrafish by activating the ROS-GADD45B-p38 pathway. Cell Death Dis 7, e2068. 10.1038/cddis.2015.300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Hoffman B, and Liebermann DA (2009). Gadd45 modulation of intrinsic and extrinsic stress responses in myeloid cells. J Cell Physiol 218, 26–31. 10.1002/jcp.21582. [DOI] [PubMed] [Google Scholar]
- 66.Franklin SL, Ferry RJ Jr., and Cohen P (2003). Rapid insulin-like growth factor (IGF)-independent effects of IGF binding protein-3 on endothelial cell survival. J Clin Endocrinol Metab 88, 900–907. 10.1210/jc.2002-020472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Granata R, Trovato L, Garbarino G, Taliano M, Ponti R, Sala G, Ghidoni R, and Ghigo E (2004). Dual effects of IGFBP-3 on endothelial cell apoptosis and survival: involvement of the sphingolipid signaling pathways. FASEB J 18, 1456–1458. 10.1096/fj.04-1618fje. [DOI] [PubMed] [Google Scholar]
- 68.Crosswhite PL, Podsiadlowska JJ, Curtis CD, Gao S, Xia L, Srinivasan RS, and Griffin CT (2016). CHD4-regulated plasmin activation impacts lymphovenous hemostasis and hepatic vascular integrity. J Clin Invest 126, 2254–2266. 10.1172/JCI84652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 69.Rustenhoven J, Pavlou G, Storck SE, Dykstra T, Du S, Wan Z, Quintero D, Scallan JP, Smirnov I, Kamm RD, and Kipnis J (2023). Age-related alterations in meningeal immunity drive impaired CNS lymphatic drainage. J Exp Med 220. 10.1084/jem.20221929. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Rustenhoven J, Drieu A, Mamuladze T, de Lima KA, Dykstra T, Wall M, Papadopoulos Z, Kanamori M, Salvador AF, Baker W, et al. (2021). Functional characterization of the dural sinuses as a neuroimmune interface. Cell 184, 1000–1016 e1027. 10.1016/j.cell.2020.12.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Da Mesquita S, Herz J, Wall M, Dykstra T, de Lima KA, Norris GT, Dabhi N, Kennedy T, Baker W, and Kipnis J (2021). Aging-associated deficit in CCR7 is linked to worsened glymphatic function, cognition, neuroinflammation, and beta-amyloid pathology. Sci Adv 7. 10.1126/sciadv.abe4601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.Faraco G, Hochrainer K, Segarra SG, Schaeffer S, Santisteban MM, Menon A, Jiang H, Holtzman DM, Anrather J, and Iadecola C (2019). Dietary salt promotes cognitive impairment through tau phosphorylation. Nature 574, 686–690. 10.1038/s41586-019-1688-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Santisteban MM, Schaeffer S, Anfray A, Faraco G, Brea D, Wang G, Sobanko MJ, Sciortino R, Racchumi G, Waisman A, et al. (2024). Meningeal interleukin-17-producing T cells mediate cognitive impairment in a mouse model of salt-sensitive hypertension. Nat Neurosci 27, 63–77. 10.1038/s41593-023-01497-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 74.Rogers JT, Morganti JM, Bachstetter AD, Hudson CE, Peters MM, Grimmig BA, Weeber EJ, Bickford PC, and Gemma C (2011). CX3CR1 deficiency leads to impairment of hippocampal cognitive function and synaptic plasticity. J Neurosci 31, 16241–16250. 10.1523/JNEUROSCI.3667-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75.Newell EA, Todd BP, Mahoney J, Pieper AA, Ferguson PJ, and Bassuk AG (2018). Combined Blockade of Interleukin-1alpha and −1beta Signaling Protects Mice from Cognitive Dysfunction after Traumatic Brain Injury. eNeuro 5. 10.1523/ENEURO.0385-17.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 76.Guo JL, Braun D, Fitzgerald GA, Hsieh YT, Rouge L, Litvinchuk A, Steffek M, Propson NE, Heffner CM, Discenza C, et al. (2025). Decreased lipidated ApoE-receptor interactions confer protection against pathogenicity of ApoE and its lipid cargoes in lysosomes. Cell 188, 187–206 e126. 10.1016/j.cell.2024.10.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 77.Qi G, Mi Y, Shi X, Gu H, Brinton RD, and Yin F (2021). ApoE4 Impairs Neuron-Astrocyte Coupling of Fatty Acid Metabolism. Cell Rep 34, 108572. 10.1016/j.celrep.2020.108572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 78.Farmer BC, Walsh AE, Kluemper JC, and Johnson LA (2020). Lipid Droplets in Neurodegenerative Disorders. Front Neurosci 14, 742. 10.3389/fnins.2020.00742. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 79.Ding J, Ji J, Rabow Z, Shen T, Folz J, Brydges CR, Fan S, Lu X, Mehta S, Showalter MR, et al. (2021). A metabolome atlas of the aging mouse brain. Nat Commun 12, 6021. 10.1038/s41467-021-26310-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 80.Almeida FC, Patra K, Giannisis A, Niesnerova A, Nandakumar R, Ellis E, Oliveira TG, and Nielsen HM (2024). APOE genotype dictates lipidomic signatures in primary human hepatocytes. J Lipid Res 65, 100498. 10.1016/j.jlr.2024.100498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 81.Kinoshita M, Kyo T, and Matsumori N (2020). Assembly formation of minor dihydrosphingomyelin in sphingomyelin-rich ordered membrane domains. Sci Rep 10, 11794. 10.1038/s41598-020-68688-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 82.Chan RB, Oliveira TG, Cortes EP, Honig LS, Duff KE, Small SA, Wenk MR, Shui G, and Di Paolo G (2012). Comparative lipidomic analysis of mouse and human brain with Alzheimer disease. J Biol Chem 287, 2678–2688. 10.1074/jbc.M111.274142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 83.Bonacina F, Coe D, Wang G, Longhi MP, Baragetti A, Moregola A, Garlaschelli K, Uboldi P, Pellegatta F, Grigore L, et al. (2018). Myeloid apolipoprotein E controls dendritic cell antigen presentation and T cell activation. Nat Commun 9, 3083. 10.1038/s41467-018-05322-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 84.Zhang YJ, Cheng Y, Tang HL, Yue Q, Cai XY, Lu ZJ, Hao YX, Dai AX, Hou T, Liu HX, et al. (2024). APOE epsilon4-associated downregulation of the IL-7/IL-7R pathway in effector memory T cells: Implications for Alzheimer’s disease. Alzheimers Dement 20, 6441–6455. 10.1002/alz.14173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 85.Zhao N, Attrebi ON, Ren Y, Qiao W, Sonustun B, Martens YA, Meneses AD, Li F, Shue F, Zheng J, et al. (2020). APOE4 exacerbates alpha-synuclein pathology and related toxicity independent of amyloid. Sci Transl Med 12. 10.1126/scitranslmed.aay1809. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 86.Tsuang D, Leverenz JB, Lopez OL, Hamilton RL, Bennett DA, Schneider JA, Buchman AS, Larson EB, Crane PK, Kaye JA, et al. (2013). APOE epsilon4 increases risk for dementia in pure synucleinopathies. JAMA Neurol 70, 223–228. 10.1001/jamaneurol.2013.600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 87.Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, et al. (2012). Fiji: an open-source platform for biological-image analysis. Nat Methods 9, 676–682. 10.1038/nmeth.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 88.Arshadi C, Gunther U, Eddison M, Harrington KIS, and Ferreira TA (2021). SNT: a unifying toolbox for quantification of neuronal anatomy. Nat Methods 18, 374–377. 10.1038/s41592-021-01105-7. [DOI] [PubMed] [Google Scholar]
- 89.Finak G, McDavid A, Yajima M, Deng J, Gersuk V, Shalek AK, Slichter CK, Miller HW, McElrath MJ, Prlic M, et al. (2015). MAST: a flexible statistical framework for assessing transcriptional changes and characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol 16, 278. 10.1186/s13059-015-0844-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 90.Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh PR, and Raychaudhuri S (2019). Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 16, 1289–1296. 10.1038/s41592-019-0619-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 91.Durinck S, Spellman PT, Birney E, and Huber W (2009). Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. Nat Protoc 4, 1184–1191. 10.1038/nprot.2009.97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 92.Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, Benner C, and Chanda SK (2019). Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun 10, 1523. 10.1038/s41467-019-09234-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 93.Zheng GX, Terry JM, Belgrader P, Ryvkin P, Bent ZW, Wilson R, Ziraldo SB, Wheeler TD, McDermott GP, Zhu J, et al. (2017). Massively parallel digital transcriptional profiling of single cells. Nat Commun 8, 14049. 10.1038/ncomms14049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 94.Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, and Satija R (2024). Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol 42, 293–304. 10.1038/s41587-023-01767-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 95.Renier N, Wu Z, Simon DJ, Yang J, Ariel P, and Tessier-Lavigne M (2014). iDISCO: a simple, rapid method to immunolabel large tissue samples for volume imaging. Cell 159, 896–910. 10.1016/j.cell.2014.10.010. [DOI] [PubMed] [Google Scholar]
- 96.Jacob L, Boisserand LSB, Geraldo LHM, de Brito Neto J, Mathivet T, Antila S, Barka B, Xu Y, Thomas JM, Pestel J, et al. (2019). Anatomy and function of the vertebral column lymphatic network in mice. Nat Commun 10, 4594. 10.1038/s41467-019-12568-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Concentrations of cytokines and chemokines measured in the cervical LNs of all the groups of female and male mice, related to Figure 4.
Table S2. Female and male forebrain lipidomics data and respective statistical analyses, related to Figures 5 and S11.
Table S3. Concentrations of cytokines and chemokines measured in the forebrains of all the groups of female and male mice, related to Figure 6.
Table S4. Lists of DEGs between E4 carriers and E4 non-carriers in each human brain leukocyte cluster, related to Figure 7.
Data Availability Statement
The scRNA-seq data reported in this manuscript have been deposited at GEO and are publicly available as of the date of publication. The accession number is listed in the key resources table.
Analyzed mouse meningeal dural scRNA-seq data are also accessible for consultation through shinyapp at https://fryerlab.shinyapps.io/meningeal_dura_scRNAseq/
Original code is reported at the Github repository at: https://github.com/kenneditodd/Da_Mesquita (mouse scRNA-seq analysis code); DOI: https://doi.org/10.5281/zenodo.18700908.
https://github.com/amoghc94/Da_Mesquita_APOE4 (human snRNA-seq analysis code); DOI: https://doi.org/10.5281/zenodo.18687961.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
KEY RESOURCES TABLE
| REAGENT or RESOURCE | SOURCE | IDENTIFIER |
|---|---|---|
| Antibodies | ||
| Rat anti-mouse LYVE-1-eF660, clone ALY7 | Thermo Fisher Scientific | Cat#50-0443-82 |
| Rat anti-mouse I-A/I-E-BV421, clone M5/114.15.2 | BioLegend | Cat#107631 |
| Rabbit anti-mouse LYVE-1 | AngioBio | Cat#11-034 |
| Goat anti-mouse CD31 | R&D Systems | Cat#AF3628 |
| Goat anti-mouse Podocalyxin | R&D Systems | Cat#AF1556 |
| Rat anti-mouse CD206, clone MR5D3 | Bio-Rad Laboratories | Cat#MCA2235 |
| Rabbit anti-mouse hAPOE | Cell Signaling Technology | Cat#13366S |
| Goat anti-mouse IBA1 | Abcam | Cat#ab5076 |
| Rat anti-mouse IBA1 | Abcam | Cat#ab5076 |
| Guinea pig anti-mouse PLIN3 | Progen | Cat#GP37 |
| Armenian hamster anti-mouse CD31 | Sigma-Aldrich | Cat#MAB1398Z |
| Chicken anti-mouse GFAP | Abcam | Cat#ab4674 |
| Mouse anti-mouse Quaking 7, clone CC1 | Sigma-Aldrich | Cat#OP80 |
| Goat anti-mouse PDGFRα | R&D Systems | Cat#AF1062 |
| Donkey anti-goat AF488 | Thermo Fisher Scientific | Cat#A-11055 |
| Donkey anti-rabbit AF488 | Thermo Fisher Scientific | Cat#A-21206 |
| Donkey anti-chicken AF488 | Thermo Fisher Scientific | Cat#A78948 |
| Donkey anti-goat AF594 | Thermo Fisher Scientific | Cat#A-11058 |
| Donkey anti-mouse AF594 | Thermo Fisher Scientific | Cat#A-21203 |
| Donkey anti-rabbit AF594 | Thermo Fisher Scientific | Cat#A-21207 |
| Donkey anti-rabbit AF594 | Jackson Immunoresearch | Cat#711-585-152 |
| Donkey anti-rat AF647 | Thermo Fisher Scientific | Cat#A-48272 |
| Donkey anti-goat AF647 | Thermo Fisher Scientific | Cat#A-21447 |
| Donkey anti-rabbit AF647 | Thermo Fisher Scientific | Cat#A-31573 |
| Donkey anti-goat AF647 | Jackson Immunoresearch | Cat#705-605-147 |
| Rat anti-mouse CD16/32, clone 93 | BioLegend | Cat#101302 |
| Zombie Aqua Fixable Viability Dye | BioLegend | Cat#423102 |
| Rat anti-mouse CD45-PerCP-Cy5.5, clone 30-F11 | BD Biosciences | Cat#550994 |
| Armenian hamster anti-mouse CD11c-BV605 | BioLegend | Cat#117334 |
| Rat anti-mouse CD11b-PE-Cy7 | BD Biosciences | Cat#552850 |
| Rat anti-mouse CD206-PE-Dazzle594 | BioLegend | Cat#141732 |
| Rat anti-mouse MHC-II-AF647 | BD Biosciences | Cat#562367 |
| Rat anti-mouse FOXP3-PE | Thermo Fisher Scientific | Cat#12-5773-82 |
| Rat anti-mouse Ly6C-PE | BioLegend | Cat#128008 |
| Rat anti-mouse F4/80-FITC | BioLegend | Cat#123108 |
| Rat anti-mouse CD19-APC-Cy7, clone 6D5 | BioLegend | Cat#115530 |
| Hamster anti-mouse TCRb-BV711, clone H57-597 | BD Biosciences | Cat#563135 |
| Rat anti-mouse CD4-FITC | BioLegend | Cat#100406 |
| Rat anti-mouse CD8-PB | BD Bioscience | Cat#558106 |
| Rat anti-mouse CD31-FITC | Thermo Fisher Scientific | Cat#11-0311-82 |
| Experimental models: Organisms/strains | ||
| C57BL/6 | Taconic Biosciences | Taconic: B6 |
| B6.129P2-Apoetm1Unc N11 | Taconic Biosciences | Taconic: APOE |
| C57BL/6NTac-Apoe<tm4206.1(APOE*C130,*R176)Tac> | Taconic Biosciences | Taconic: CureAlz huAPOE3 |
| C57BL/6NTac-Apoe<tm4207.1(APOE*R130,*R176)Tac> | Taconic Biosciences | Taconic: CureAlz huAPOE4 |
| Chemicals | ||
| PLX5622 | Chemgood | Cat#C-1521 |
| DirectPCR (tail) | Viagen | Cat#102-T |
| Ultrapure water | Apex Bioresearch Products | Cat#20-102 |
| Phosphate buffered saline (PBS) 10×, pH 7.4 | Fisher BioReagents | Cat#BP-399 |
| Heparin | Fisher BioReagents | Cat#BP2425 |
| 4% PFA in 1× PBS | Boster Bio | Cat#AR1068 |
| Sucrose | Cargill | Cat#62-112 |
| Tissue-Plus O.C.T. Compound | Fisher HealthCare | Cat#4585 |
| Gelatin | Sigma-Aldrich | Cat#G1890 |
| Sodium Azide 1% | G-Biosciences | Cat#786-750 |
| Triton X-100 | Sigma-Aldrich | Cat#X100 |
| Bovine serum albumin | Genesee Scientific | Cat#25-529 |
| Citrate buffer 10× | Sigma-Aldrich | Cat#C9999 |
| 4,6-diamidino-2-phenylindole (DAPI) | Thermo Fisher Scientific | Cat#62248 |
| Epredia Immu-Mount™ | Thermo Fisher Scientific | Cat#9990402 |
| Halt protease inhibitor cocktail | Thermo Fisher Scientific | Cat#78430 |
| Phosphatase inhibitor cocktail | Cell Signaling Technology | Cat#5870 |
| Phenylmethylsulfonyl fluoride | Cell Signaling Technology | Cat#8553 |
| Ethanol | DeconLabs | Cat#2701 |
| Pierce RIPA Buffer | Thermo Fisher Scientific | Cat#89901 |
| RPMI 1640 | Genesee Scientific | Cat#25-506 |
| DNase I | Sigma-Aldrich | Cat#11284932001 |
| Collagenase VIII | Sigma-Aldrich | Cat#C2139 |
| Collagenase D | Sigma-Aldrich | Cat#11088866001 |
| Fetal bovine serum (FBS, heat inactivated) | Corning | Cat#35-011-CV |
| RNAscope Target Probe APOE-C1 | Advanced Cell Diagnostics | Cat#433091-C1 |
| RNAscope Target Probe APOE-C3 | Advanced Cell Diagnostics | Cat#313271-C3 |
| RNAscope Target Probe Itgam-C2 | Advanced Cell Diagnostics | Cat#311491-C2 |
| RNAscope Target Probe Pecam1-C1 | Advanced Cell Diagnostics | Cat#316721-C1 |
| RNAscope Target Probe Pecam1-C3 | Advanced Cell Diagnostics | Cat#316721-C3 |
| Opal 520 reagent | Akoya Biosciences | Cat#FP1487001KT |
| Opal 620 reagent | Akoya Biosciences | Cat#FP1495001KT |
| Opal 690 reagent | Akoya Biosciences | Cat#FP1497001KT |
| RNAscope® Multiplex TSA buffer | Advanced Cell Diagnostics | Cat#322809 |
| Non-acetylated BSA | Thermo Fisher Scientific | Cat#AM2618 |
| Trypan blue | Gibco | Cat#15250061 |
| Actinomycin-D | Sigma-Aldrich | Cat#A1410 |
| Dimethyl sulfoxide (DMSO) | Sigma-Aldrich | Cat#276855 |
| UltraPure ethylenediaminetetraacetic acid (EDTA) 0.5 M | Thermo Fisher Scientific | Cat#15575020 |
| Commercial Assays | ||
| Anti-FITC Microbeads kit | Miltenyi Biotec | Cat#130-048-701 |
| Chromium Next GEM Single Cell 3’ GEM Kit v3.1 | 10× Genomics | Cat#PN-1000123 |
| Library Construction Kit | 10× Genomics | Cat#PN-1000190 |
| Chromium Next GEM Single Cell 3’ Gel Bead Kit v3.1 | 10× Genomics | Cat#PN-1000122 |
| Chromium Next GEM Chip G Single Cell Kit | 10× Genomics | Cat#PN-1000127 |
| Dual Index Kit TT Set A | 10× Genomics | Cat#PN-1000215 |
| RNAscope® Multiplex Fluorescent Reagent Kit v2 | Advanced Cell Diagnostics | Cat#323100 |
| Micro BCA Protein Assay Kit | Thermo Fisher Scientific | Cat#23235 |
| Pierce BCA Protein Assay Kit | Thermo Fisher Scientific | Cat#23227 |
| Human APOE ELISA Kit | Abcam | Cat#ab108813 |
| Foxp3/Transcription factor staining buffer set | Thermo Fisher Scientific | Cat#00-5523-00 |
| ZymoPURE Plasmid Miniprep Kit | Zymo Research | Cat#D4209 |
| ZymoPURE II Plasmid Maxiprep Kit | Zymo Research | Cat#D4203 |
| Deposited Data | ||
| scRNA-seq (mouse meningeal dura) | This paper | GEO: GSE295612 |
| snRNA-seq (human brain) | Mathys et al., 20191 | Synapse: Syn18485175 |
| snRNA-seq (human brain) | Morabito et al., 20212 | Synapse: Syn26670419 |
| snRNA-seq (human brain) | Blanchard et al., 20223 | Synapse: Syn38120890 |
| snRNA-seq (human brain) | Fujita et al., 20244 | Synapse: Syn31512863 |
| scRNA-seq (human brain) | Mathys et al., 20245 | Synapse: Syn52293417 |
| snRNA-seq (human brain) | Li et al., 20256 | GEO: GSE237718 |
| Software and Algorithms | ||
| FIJI version 2.3.0 | Schindelin et al.7 | https://imagej.net/software/fiji/downloads |
| Simple Neurite Tracer (FIJI plugin) version 1.53q | Arshadi et al.8 | https://imagej.net/plugins/snt |
| R Statistical Software versions 4.1.2 and 4.4.2 | R Foundation for Statistical Computing | https://www.r-project.org |
| ImSpector Pro microscope controller version 7 | Miltenyi/LaVision Biotec | https://www.miltenyibiotec.com/US-en/about-us/miltenyi-biotec-companies/lavision-biotec-gmbh |
| Imaris versions ×64 10.0.1 and 10.2.0 | Oxford Instruments, Bitplane | https://imaris.oxinst.com/versions/10 |
| Freezeframe version 4.104 | Actimetrics | https://actimetrics.com/products/freezeframe |
| AnyMaze version 7.8 | Stoelting | https://www.any-maze.com |
| FlowJo version 10.10.0 | BD Biosciences | https://www.flowjo.com/flowjo10/overview |
| Prism version 10.4.1 | GraphPad software | https://www.graphpad.com/updates/prism-10-4-1-release-notes |
| 10× Genomics Cell Ranger version 7.1.0 | 10× Genomics | https://www.10xgenomics.com/software |
| Seurat v5 versions 5.0.2 and 5.3.0 | R toolkit | https://github.com/satijalab/seurat/releases |
| MAST version 1.26.0 | Finak et al.9 | https://www.bioconductor.org/packages/devel/bioc/html/MAST |
| Nichenetr (R package) version 2.0.4 | GitHub, Browaeys et al.10 | https://github.com/saeyslab/nichenetr |
| RStudio versions 1.4.1103 and 4.3.0 | Posit | https://forum.posit.co/t/rstudio-1-4-1103-desktop/116542 |
| Harmony version 1.2.3 | Korsunsky et al.11 | https://cran.r-project.org/web/packages/harmony |
| biomaRt version 2.62.1 | Durinck et al.12 | https://bioconductor.org/packages/release/bioc/html/biomaRt |
| dplyr version 1.1.4 | dplyr.tidyverse.org | https://cran.r-project.org/web/packages/dplyr |
| ggplot2 version 3.5.2 | ggplot2.tidyverse.org | https://cran.r-project.org/web/packages/ggplot2 |
| Metascape versions 3.5.20250101 and 3.5.20250701 | Zhou et al.13 | https://metascape.org |
| Custom code used for mouse scRNA-seq data analysis | This paper | DOI: https://doi.org/10.5281/zenodo.18700908 |
| Custom code used for human snRNA-seq data analysis | This paper | DOI: https://doi.org/10.5281/zenodo.18687961 |
| Other | ||
| PLX5622 supplemented diet (600 p.p.m.) with blue dye | Research diets | Cat#D21102810 |
| PicoLab Rodent Diet 20 | LabDiet | Cat#5053 |