Cholesterol and matrisome pathways dysregulated in astrocytes and microglia

In the paper

“While APOE is upregulated in human microglia in AD, the mouse DAM phenotype is absent or only partially overlaps, highlighting species differences between human and mouse and the importance of studying APOE4 in human systems (Galatro et al., 2017; Li et al., 2021; Srinivasan et al., 2020; Zhou et al., 2020).”

What the sources say

• Partly supported

  The Discussion (p. 9) reports noteworthy differences between human and mouse microglial expression, and the age-associated changes section (p. 9) finds limited overlap in their aging signatures. These passages support the general point about differences, but not the AD-specific claims: our copy does not report APOE upregulation in human AD microglia or test whether the mouse DAM phenotype is absent or partially overlapping (searched: APOE, disease-associated microglia, and AD). (LLM, conf=0.82)

  cites Galatro et al., 2017 sourceExperimental — accuracy not yet measured
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Li et al., 2021
• Supported

  The Results section reports that APOE was significantly increased in AD patient microglia, while the mouse-derived DAM module showed only a slight increase and most of its other genes showed no clear trend. The Discussion describes the human AD microglial profile as almost entirely distinct from mouse DAM and notes a relative lack of DAM induction in AD microglia. Page numbers are not provided in our copy. (LLM, conf=0.93)

  cites Srinivasan et al., 2020 source

In the paper

“APOE4 also contributes to decreased synaptic plasticity and blood-brain barrier dysfunction (Chen et al., 2010; Montagne et al., 2020).”

What the sources say

• Partly supported

  The Abstract and Discussion report that ApoE4 impairs Reelin-enhanced synaptic plasticity and hippocampal LTP. The Discussion mentions isoform-dependent Aβ clearance across the blood–brain barrier, but does not establish that ApoE4 causes blood–brain barrier dysfunction. Because both citations attach to the combined sentence without assigning its two claims separately, this source supports only the synaptic-plasticity part. (LLM, conf=0.88)

  cites Chen et al., 2010 sourceExperimental — accuracy not yet measured
• Partly supported

  The Introductory Paragraph reports blood–brain barrier breakdown in APOE4 carriers, supporting the barrier-dysfunction part of the sentence. It also discusses possible synaptic dysfunction, but does not establish decreased synaptic plasticity. Because the two citations appear together without assigning a separate part to each, that part of the attributed sentence remains unsupported by this paper. (LLM, conf=0.85)

  cites Montagne et al., 2020 sourceExperimental — accuracy not yet measured

In the paper

“Notably, the APOE4 allele is the strongest genetic AD risk factor; APOE ɛ4/ɛ4 (APOE 44) increases AD risk by 15-fold in Europeans (Genin et al., 2011).”

What the sources say

• Supported

  The Introduction reports an odds ratio of 14.9 for APOE 44 versus APOE 33 in Caucasian subjects, supporting the approximate 15-fold figure. The figure refers specifically to odds relative to APOE 33, not an absolute lifetime risk. (LLM, conf=0.90)

  cites Genin et al., 2011 source

In the paper

“Africans exhibit lower APOE4-associated AD risk while East Asians show higher APOE4 risk than Europeans (Liu et al., 2014; Tang et al., 1998).”

What the sources say

• ?Could not verify

  The Discussion reports APOE ε4-associated AD risk in the Chinese population and says susceptibility can vary by ethnicity, but it does not compare the risk in Africans, East Asians, and Europeans. Its regional comparison concerns ε4 genotype prevalence among AD patients, not the strength of ε4-associated risk. Searches for African–European and East Asian–European risk comparisons found no passage establishing the claimed directions. (LLM, conf=0.75)

  cites Liu et al., 2014 sourceExperimental — accuracy not yet measured
• Partly supported

  The Results report a lower APOE-ε4-associated relative risk among African Americans than among whites (RR 1.0 versus 2.5), supporting that comparison for the populations studied. The paper does not establish the broader comparison of Africans with Europeans, and its Comment says the association is strong among both whites and Asians, not that East Asians have a higher APOE-ε4-associated risk than Europeans. (LLM, conf=0.88)

  cites Tang et al., 1998 sourceExperimental — accuracy not yet measured

In the paper

“Upstream regulators, transcription factors regulating cholesterol synthesis, SREBF2 and SCAP, are activated, but POR (Cytochrome P450 oxidoreductase) is inhibited in both APOE4 microglia and astrocytes (Figure 2H)(Brown and Goldstein, 2009; Schmidt et al., 2009).”

What the sources say

• ?Not checked — no full text of the sourcecites Brown and Goldstein, 2009 source
• Partly supported

  The Microarray analysis section reports increased SREBP2 expression and upregulation of cholesterol-biosynthesis genes after POR deletion in mouse limb buds. It does not report SCAP activation or examine APOE4 microglia or astrocytes; the citing sentence points to its own Figure 2H for those cell-specific findings. (LLM, conf=0.86)

  cites Schmidt et al., 2009 sourceExperimental — accuracy not yet measured

In the paper

“Astrocyte-specific APOE4 deletion rescues tau-mediated neurodegeneration (Wang et al., 2021).”

What the sources say

• Partly supported

  The Abstract reports that astrocytic APOE4 removal markedly reduced tau-mediated neurodegeneration in P301S Tau mice, and the Discussion describes strong attenuation after deletion. The claim’s unqualified “rescues” may imply complete reversal, which these passages do not establish. (LLM, conf=0.82)

  cites Wang et al., 2021 sourceExperimental — accuracy not yet measured

In the paper

“In mice, Apoe is mainly produced by astrocytes with low microglial expression (Xu et al., 2006), similarly in humans, APOE is primarily expressed in astrocytes (Zhang et al., 2016).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Xu et al., 2006
• ?Could not verify

  The Results section on transcriptomes of purified human brain cell types describes comparisons among astrocytes, microglia, and other cells, but does not state which cell type primarily expresses APOE. Searches for “APOE” and “apolipoprotein E” found no passage in our copy that establishes the human-expression claim attributed to Zhang et al.; the mouse clause is cited to a different source. (LLM, conf=0.70)

  cites Zhang et al., 2016 sourceExperimental — accuracy not yet measured

In the paper

“In aging or neurodegeneration, mouse microglia adopt Apoe and Trem2-dependent disease-associated or neurodegenerative microglia phenotype (DAM, MGnD), upregulating Apoe (Keren-Shaul et al., 2017; Krasemann et al., 2017).”

What the sources say

• Partly supported

  The Results and Discussion describe DAM in mouse models of neurodegeneration, with Apoe upregulated during an initial Trem2-independent stage and full DAM activation requiring Trem2. This paper does not establish that DAM activation depends on Apoe or demonstrate that aging alone induces this phenotype; it calls the phenotype DAM rather than MGnD. No page numbers are available in our copy. (LLM, conf=0.87)

  cites Keren-Shaul et al., 2017 sourceExperimental — accuracy not yet measured
• Partly supported

  The Results sections on phagocytic microglia and Trem2 deletion report Apoe upregulation in the MGnD-like phenotype and show that disrupting Trem2 suppresses the APOE pathway and restores homeostatic genes in neurodegenerative mouse models. The Discussion includes aging in its MGnD transcriptome profiling, but the paper does not establish that the phenotype in aging itself is Trem2-dependent. No page numbers are available in our copy. (LLM, conf=0.82)

  cites Krasemann et al., 2017 sourceExperimental — accuracy not yet measured

In the paper

“NCKAP1L (a.k.a. hematopoietic cell-specific actin regulatory protein (HEM-1)) interacts with ABI3, signals downstream of TREM2 and PLCG2 and functions in actin cytoskeleton reorganization during chemotaxis and efferocytosis (Andreone et al., 2020; Park et al., 2010).”

What the sources say

• ?Not checked — no full text of the sourcecites Andreone et al., 2020 source
• Partly supported

  The Abstract identifies Hem-1 as NCKAP1L, and sections 5 and 7 link it to actin polymerization and chemotaxis. Section 9 describes Hem-1 interacting with Abi1 or Abi2 and signaling downstream of Rac, not ABI3 or TREM2 and PLCG2. Section 7 reports phagocytosis of E. coli, not efferocytosis. Page numbers are unavailable in our copy. (LLM, conf=0.90)

  cites Park et al., 2010 sourceExperimental — accuracy not yet measured

In the paper

“While APOE mRNA levels were unchanged in mouse (Gong et al., 2002)”

What the sources say

• ?Could not verify

  The Results report similar amounts of apoE protein released by astrocytes from apoE3 and apoE4 knock-in mice, but protein release does not establish unchanged APOE mRNA levels. Searches for mRNA and gene-expression measurements found no passage addressing the claim in our copy; the cited Results passages have no page number available. (LLM, conf=0.75)

  cites Gong et al., 2002 sourceExperimental — accuracy not yet measured

In the paper

“Further, intracellular free cholesterol levels measured with filipin (Pipalia et al., 2011) was significantly increased in APOE4 astrocytes, but donor fibroblasts did not exhibit genotype-dependent cholesterol differences (whole cells and lysosomes), potentially due to the lack of APOE (Figures 6B, 6C and S6A).”

What the sources say

• Supported

  The citation is attached to the filipin measurement method, not to the citing authors’ findings about APOE4 astrocytes or donor fibroblasts. The Results describe quantifying free cholesterol in lysosomal storage organelles using filipin staining. (LLM, conf=0.94)

  cites Pipalia et al., 2011 source

In the paper

“Previous studies identified APOE-isoform specific effects on cholesterol flux and APOE levels in rodent brain cells and human periphery (Gong et al., 2002; Huang et al., 1995; Michikawa et al., 2000).”

What the sources say

• Partly supported

  The Results report greater cholesterol release from apoE3- than apoE4-expressing mouse astrocytes, supporting an isoform-specific effect on cholesterol flux. But the Results also report comparable total amounts of apoE released by the two cell types. This paper studies mouse brain cells, not human peripheral samples; the other citations may address that part of the sentence. (LLM, conf=0.90)

  cites Gong et al., 2002 sourceExperimental — accuracy not yet measured
• Supported

  The Results report that human apoE2/2 and apoE4/4 plasmas promoted less cholesterol efflux from labeled fibroblasts than apoE3/3 plasma. Results Table II also shows apoE levels by phenotype. This supports the human-periphery portion of the sentence; the rodent-brain-cell portion is not attributed to Huang et al. alone. (LLM, conf=0.91)

  cites Huang et al., 1995 source
• ?Not checked — no full text of the sourcecites Michikawa et al., 2000 source

In the paper

“25-hydroxycholesterol (25HC), reported to decrease cholesterol synthesis and increase ABCA1 and APOE (Waltl et al., 2013)”

What the sources say

• Supported

  In Results and discussion, the paper reports that 25-HC significantly inhibited cholesterol biosynthesis in murine CATH.a cells. The section on transcriptional regulation reports increased ABCA1 and apoE mRNA in those cells, supporting both parts of the attributed claim. (LLM, conf=0.93)

  cites Waltl et al., 2013 source

In the paper

“Levels of LAMP1/2, which are involved in NPC2 to NPC1 cholesterol export pathway from lysosome were decreased in APOE4 astrocytes and microglia (Figures S6E and S6F) in which LAMP1 deficiency causes defective lysosome-ER cholesterol transport and lysosomal cholesterol accumulation (Li and Pfeffer, 2016).”

What the sources say

• Partly supported

  The Results and discussion section reports lysosomal cholesterol accumulation and impaired cholesterol export to the endoplasmic reticulum in mouse embryonic fibroblasts lacking both LAMP1 and LAMP2; LAMP2 expression rescued export. It does not establish that LAMP1 deficiency alone causes these defects, or demonstrate them in APOE4 astrocytes or microglia. (LLM, conf=0.88)

  cites Li and Pfeffer, 2016 sourceExperimental — accuracy not yet measured

In the paper

“NPCs (CD271−/CD133+) were enriched by magnetic activated cell sorting (Miltenyi Biotec) (Bowles et al., 2019) and validated immunocytochemically using SOX2 (Cell signaling, 3579S), PAX6 (Abcam, ab5790), FOXP2 (Abcam, ab16046) and NESTIN (Abcam, ab22035).”

What the sources say

• Supported

  The citation is attached to the MACS-enrichment clause, before the citing authors’ description of their immunocytochemical validation. Bowles et al.’s Materials & methods (Magnetic cell sorting) describe CD271 depletion followed by CD133 selection using Miltenyi equipment, and the Results report enrichment of CD271−/CD133+ NPCs. (LLM, conf=0.91)

  cites Bowles et al., 2019 source

In the paper

“APOE4 has a strong odds ratio (OR) in non-Hispanic Europeans (OR=14.9), whereas APOE4 risk varies substantially across populations (African Americans and Caribbean Hispanics, OR=2.2–5.7 and East Asians, OR=11.8–33.1)(Liu et al., 2014; Tang et al., 1998).”

What the sources say

• Partly supported

  The Discussion says APOE ε4 is associated with Alzheimer’s disease risk and that genotype prevalence varies by region. But this meta-analysis examines the Chinese population; it does not report the claimed odds ratios for non-Hispanic Europeans, African Americans, Caribbean Hispanics, or East Asians. Its reported odds ratios are 3.93 for Chinese ε4 carriers and 11.76 for the Chinese ε4/ε4 genotype (Abstract and Discussion; page numbers unavailable). (LLM, conf=0.88)

  cites Liu et al., 2014 sourceExperimental — accuracy not yet measured
• Partly supported

  The Results report a stronger APOE-ε4 association with Alzheimer disease in whites than in African Americans or Hispanics, and the Comment describes a strong association in whites and Asians but a weaker, less consistent one in African Americans and Hispanics. The paper reports relative risks, not the cited odds ratios: it gives 2.5 for whites, 1.0 for African Americans, and 1.1 for Hispanics, and does not provide the quoted East Asian range. Page numbers are unavailable in our copy. (LLM, conf=0.87)

  cites Tang et al., 1998 sourceExperimental — accuracy not yet measured

In the paper

“Recent hiPSC genetic modeling efforts have utilized either a pooled patient iPSC strategy to uncover disease phenotypes in the context of patient genetic heterogeneity or a comprehensive isogenic platform to assess the effect of disease mutations by limiting genetic background variation (Lim et al., 2020; Mitchell et al., 2020; Ramos et al., 2021).”

What the sources say

• Supported

  The Results (page unavailable) describe generating cerebral organoids from iPSCs of 25 individuals, including patients and controls, and pooling individual organoids for analysis. The Results also compare patient-derived organoids with isogenic iPSC-derived models. The paper therefore exemplifies the modeling approaches described, although it pooled organoids rather than the iPSCs themselves. (LLM, conf=0.87)

  cites Lim et al., 2020 source
• Supported

  The 'Cellular phenotyping and genetic analysis in villages' section describes a pooled PSC village containing iPSCs from a spinal muscular atrophy patient; comparing donor representation revealed the patient's low-SMN phenotype. The same section describes a larger village of 113 iPSC lines used to analyze genetic variation in that phenotype. This supports Mitchell et al. as an example of the pooled-iPSC strategy, not the alternative isogenic platform. (LLM, conf=0.90)

  cites Mitchell et al., 2020 source
• Partly supported

  The 'Advancements and challenges' section explains that isogenic iPSC lines help separate mutation effects from differences in genetic background. The 'What is iNDI? > Phase 1' section describes a comprehensive isogenic platform, but says its lines will be generated; it does not establish that this platform had already been utilized. The pooled-patient strategy is not the part of the multi-citation sentence for which Ramos et al. appears to be cited. (LLM, conf=0.82)

  cites Ramos et al., 2021 sourceExperimental — accuracy not yet measured

In the paper

“Using technical replicates from one isogenic pair, Lin et al. reported intracellular free cholesterol accumulation in APOE4 astrocytes, consistent with our data, but identified downregulated cholesterol biosynthesis, opposing our data.”

What the sources say

• ?Could not verify

  The indexed copy contains a correction notice (p. 1294) about a Figure 6A recording trace, not the study’s substantive results. Searches for free cholesterol, cholesterol biosynthesis, and technical replicates found no passages addressing the attributed findings; the relevant article text appears to be missing from our copy. (LLM, conf=0.76)

  cites Lin et al., 2018 sourceExperimental — accuracy not yet measured

In the paper

“Neuropathologic diagnosis of AD was in accordance with current NIA-AA guidelines (Hyman et al., 2012; Montine et al., 2012).”

What the sources say

• Supported

  The Summary identifies this paper as updated consensus criteria for the neuropathologic assessment of AD and recommends the ABC staging protocol. The AD Neuropathologic Change > REPORTING section specifies how to report findings in tissue specimens. The citation identifies the guidelines the citing authors say they followed; it does not, by itself, verify their application to particular samples. (LLM, conf=0.92)

  cites Hyman et al., 2012 source
• Supported

  The Introduction says this paper summarizes the revised consensus guidelines for neuropathologic evaluation of AD and provides direction for applying them. The AD Neuropathologic Change > Classification section specifies the ABC scoring approach. It therefore supports the citation as an identification of the guidelines used, though it cannot independently verify how the citing study assessed its samples. (LLM, conf=0.91)

  cites Montine et al., 2012 source

In the paper

“in which adding microglia (Abreu et al., 2018; Lin et al., 2018) to such models with increasing cellular complexity may further mimic the brain environment.”

What the sources say

• Supported

  The Introduction says many 3D brain models lack microglia, a cell population involved in brain development and inflammatory responses. The Discussion reports that adding microglia to BrainSpheres changed their response to viral infection and describes the resulting model as physiologically relevant. This supports the hedged claim that adding microglia may better mimic the brain environment; page numbers are not provided in our copy. (LLM, conf=0.91)

  cites Abreu et al., 2018 source
• ?Could not verify

  The indexed copy supplies a correction notice (p. 1294) and publication metadata, but no passage about adding microglia to models or whether doing so better mimics the brain environment. Searches for “microglia” and for iPSC-derived cell co-culture found no relevant paper text, so the claim cannot be assessed from this copy. (LLM, conf=0.72)

  cites Lin et al., 2018 sourceExperimental — accuracy not yet measured

In the paper

“For multiple test adjustment, the false discovery rate (FDR) of the differential expression test was estimated using the Benjamini–Hochberg method(Reiner et al., 2003).”

What the sources say

• Supported

  The ‘Linear step-up procedure (BH)’ section describes using the Benjamini–Hochberg method to control FDR in multiple testing. The ‘Multiplicity adjusted p-values’ section gives the BH-adjusted p-value as the lowest FDR level at which a hypothesis is rejected. The paper applies this approach to differential-expression analysis; it does not report the citing authors’ own analysis. (LLM, conf=0.92)

  cites Reiner et al., 2003 source

In the paper

“Examining two isogenic pairs, Sienski et al. showed that APOE4 astrocytes were choline deficient, causing increased unsaturated fatty acids and lipid droplet accumulation in which choline supplementation restored the cellular lipidome to its basal state.”

What the sources say

• Partly supported

  The Results sections on astrocyte lipid dysregulation and choline supplementation report two isogenic pairs, increased unsaturated triacylglycerides and lipid droplets in APOE4 astrocytes, and improvement in those defects after CDP-choline treatment. But the supplementation Results report increased intracellular choline and glycerophosphocholine—not choline deficiency—and describe increased demand as a suggestion, not an established cause. They show improvement in measured lipid defects, not restoration of the entire cellular lipidome to a basal state. (LLM, conf=0.90)

  cites Sienski et al., 2021 sourceExperimental — accuracy not yet measured

In the paper

“Aligned and sorted bam files were loaded into IGV to verify APOE genotype (Robinson et al., 2011).”

What the sources say

• Supported

  The cited paper identifies IGV as a tool for loading and viewing aligned sequence reads (To the Editor). It also describes visual inspection of reads to validate variant calls. The APOE-specific verification describes the citing authors’ use of IGV, not a result attributed to Robinson et al. (LLM, conf=0.91)

  cites Robinson et al., 2011 source

In the paper

“The ERCC spike-in control (Jiang et al., 2011) was used to adjust for sequencing batch effect.”

What the sources say

• Supported

  The citation follows “ERCC spike-in control” and identifies the control used by the citing authors. The ERCC control RNA pools section (p. 8) explicitly names these materials “ERCC spike-in control mixes”; the statement about adjusting for sequencing batch effect describes the citing authors’ use, not a result attributed to Jiang et al. (LLM, conf=0.92)

  cites Jiang et al., 2011 source

In the paper

“We used a consensus top ranked cell type-specific marker set from a recent meta-analysis of multiple cell type-specific and single cell RNAseq datasets (McKenzie et al., 2018).”

What the sources say

• Supported

  The Results section, “Identification and comparison of cell type-associated expression across multiple data sets,” describes a meta-analysis of RNA-seq data from five studies. The Discussion says the resulting consensus cell type-specific marker genes can be used for deconvolution of bulk gene expression data, matching the cited marker set’s stated purpose. (LLM, conf=0.94)

  cites McKenzie et al., 2018 source

In the paper

“Dehydroergosterol (DHE) is a naturally-occurring sterol that closely mimics the behavior of cholesterol in cells and membranes (Hao et al., 2002; Maxfield and Wustner, 2012).”

What the sources say

• ?Not checked — no full text of the sourcecites Hao et al., 2002 source
• Partly supported

  The Properties of fluorescent sterols section identifies DHE as naturally occurring and reports cholesterol-like membrane phase behavior and intracellular trafficking. It also calls DHE the ideal mimic of ergosterol, not cholesterol, and describes differences from cholesterol in membrane ordering. The Metabolism of DHE in cells section says DHE mimics many aspects of cholesterol in cells but is not recognized by two cholesterol-regulatory proteins. The copy does not provide page numbers for these passages. (LLM, conf=0.86)

  cites Maxfield and Wustner, 2012 sourceExperimental — accuracy not yet measured

In the paper

“Sample ancestry was confirmed by multidimensional scaling (MDS) using PLINK 1.9 (Chang et al., 2015); MDS values for each subject were compared to 1000 Genome Project (Phase 3) sample values (Genomes Project et al., 2015).”

What the sources say

• Supported

  Chang et al. is cited to identify PLINK 1.9, the software the citing authors say they used for MDS. The paper’s Findings > Performance comparisons section explicitly discusses PLINK 1.9, so it supports that software-identification citation. The authors’ ancestry confirmation is their own analysis; the Phase 3 comparison has a separate citation. (LLM, conf=0.88)

  cites Chang et al., 2015 source
• Supported

  The citation is attached to the Phase 3 sample reference, not to the citing authors’ MDS comparison or their use of PLINK. The Main section identifies the final-phase samples from 26 populations and describes their genotype data; the Discussion describes the project’s samples and data as a broadly shareable benchmark. (LLM, conf=0.90)

  cites Genomes Project et al., 2015 source

In the paper

“Apoe WT and Apoe KO cluster independently and distinctly from APOE3 and APOE4 which were indistinguishable for each cell type (Figures 5B and 5C), suggesting mouse Apoe and human APOE functionally diverge.”

In the paper

“After cell type proportion correction, matrisome (MEyellow, MElightcyan) were upregulated, while lipid metabolism (MEgreen), transcription and translation (ME orange, MEdarkgoldenrod, MEdakrslatblue) were downregulated in APOE4 mixed cortical cultures (inconsistent in Indiv3) (Figures 3G).”

In the paper

“DEGs are in astrocytes > microglia > mixed cortical cultures, but significant DEGs in BMECs were not observed, supporting no difference in tight junction formation assessed by transepithelial electrical resistance by APOE genotype (Figures 2A, 2B and S2B).”

In the paper

“Isogenic microglia also clustered with the donor and retain individuality (Figure S3A).”

In the paper

“APOE4 occurred on 7 unique haplotypes, while APOE3 occurred on 12 different haplotypes (Figures 1C, and S1C); haplotypes were not shared between APOE3 and APOE4 carriers.”

In the paper

“Further, cell type deconvoluted APOE4 neurons revealed positively enriched TCA cycle and respiratory electron transport and Oxidative phosphorylation, but negatively enriched Ribosome in all brain regions (Figures S4M–S4Q), predicting mitochondrial and translational dysregulation.”

In the paper

“Each cell type clustered with their corresponding primary human brain cell type also unaffected by APOE genotype (Figures S1G–S1J).”

In the paper

“APOE expression was lower in APOE4 microglia and astrocytes compared to APOE3 for three individuals sharing common modules but higher in APOE4 for the outlier (Figure S3F), suggesting APOE expression underlies phenotypic variability.”

In the paper

“Microglia exclusively clustered by individual but not by either sex or APOE genotype (Figures 3C and 3D).”

In the paper

“DEGs of APOE4 vs. APOE3 in the top dysregulated pathways, lipids and focal adhesion/endocytosis astrocytes showed similar patterns, while exhibiting individual variability (Figure S3B).”

In the paper

“In APOE3 brains matrisome was the most significantly upregulated pathway in AD vs. control with lower significance compared to APOE4 (Figure S4C), confirming APOE4-driven matrisome enrichment.”

In the paper

“Principal component analysis (PCA) and Spearman correlation of 52 transcriptomes (4 cell types, 13 lines) obtained by RNA sequencing showed samples clustered by cell type but were indistinguishable by APOE genotype (Figures 1F, 1G and S1F), suggesting robust differentiation methods independent of APOE status.”

In the paper

“Deconvolved APOE4 astrocytes and microglia from the AD brain are positively enriched for Cholesterol/Steroid biosynthesis and other lipid metabolism, while negatively enriched for HDL-mediated lipid transport and Lysosome (Figures 4C–4J and S4E–S4L, S4R), consistent with the enrichment in APOE4 hiPSC-astrocytes and microglia.”

In the paper

“Matrisome associated gene sets are a Matrisome subset in hiPSC-mixed cortical cultures and AD brain, and genes within the same category highly overlapped between different brain regions (~65%) (Figure S4B).”

In the paper

“Interestingly, the pan-APOE antibody detected differences between intracellular and extracellular APOE in astrocytes and between microglia and astrocytes (Figures S6J and S6N), indicating post-translational modifications (glycosylation) may influence cell type-specific APOE function.”

In the paper

“Although LDL significantly bound more cell surface receptors (e.g. LDLRs) in APOE4 astrocytes, highlighting increased cholesterol uptake potential (Figure 6E), LDL supplementation did not increase intracellular free cholesterol in APOE4 compared to APOE3 over baseline (Figures 6B and 6C), suggesting cholesterol uptake is not the mechanism for increased cholesterol.”

In the paper

“Lipid metabolic dysregulation in Apoe KO is less significant when compared to Apoe WT rather than APOE3 or APOE4, and matrisome is not enriched in Apoe KO astrocytes but in microglia (Figures S5H and S5I), suggesting mouse Apoe LOF does not mimic APOE4 in either species.”

In the paper

“Despite the variability due to genetic heterogeneity, cell type-based transcriptomic analyses of hiPSC models, post-mortem AD brain and APOE-TR mice identified human-specific dysregulation of lipid metabolism in APOE4 astrocytes and microglia and matrisome in APOE4 mixed cortical cultures and AD brain derived from astrocyte-neuron communication.”

In the paper

“Apoe KO microglia displayed enriched Cholesterol biosynthesis and other lipid pathways, predicting upregulated lipid synthesis (Figures 5G and 5J).”

In the paper

“APOE4-TR mouse neurons and astrocytes exhibit decreased cholesterol efflux compared to APOE3, similar to our data.”

In the paper

“Cleaved SREBP2 (p<0.001), SCAP (p<0.05) and HMGCR (p<0.05) were indeed elevated (Figures 6D, S6B and S6C), indicating increased de novo cholesterol biosynthesis despite elevated intracellular cholesterol in APOE4 astrocytes.”

In the paper

“Maximal efflux (MBCD) was significantly decreased in APOE4 microglia, although HDL-mediated efflux was not impaired (Figure 6M).”

In the paper

“Apoe KO astrocytes showed enriched matrisome, enhancing chemotaxis and inflammation (Figures 5H and 5K).”

In the paper

“We observed less than 1% bleaching in fluorescence intensity per exposure (data not shown).”

In the paper

“Our population and isogenic studies identified common APOE genotype-phenotypes across individuals with opposing effects in certain individuals, which highlight the complexity of the interaction between APOE4 and genetic background in different cell types that cannot be examined with a single isogenic pair.”

In the paper

“The outline of cells using the selected values were comparable to cell outlines in transmitted light images.”

In the paper

“In summary, we demonstrate human-specific and AD brain cell type-dependent transcriptional and cellular phenotypes of APOE4, uncovering major deficits in lipid homeostasis and glial activation.”

In the paper

“Although PCA-based BRETIGIA appropriately measured cell type proportions, thus we used it as a reference, the vector value of the algorithm could not be applied to deconvolute gene expression.”

In the paper

“Despite performing GRS analysis to select hiPSCs and further incorporating this genetic variation as a covariate in analyses to minimize the contribution of other AD risk loci, pathway divergences still emerged in which one individual showed discordance but in different cell types (Figures 3E–3G).”

In the paper

“DSA (marker size 5) performed best and therefore was used in all subsequent deconvolution analyses.”

In the paper

“We uncovered altered matrisome, ECM and immune pathways by APOE4 in hiPSC-mixed cortical cultures and AD brain, identifying astrocytes as a key driver of this signal through cell type deconvolution analyses likely arising from astrocyte-neuron communication in APOE4 brain, stressing human astrocyte-specific non-cell autonomous dysregulation.”

In the paper

“Downregulated DEGs in Lysosome (e.g., LAMP1/2, NPC1/2, SMPD1 and LIPA) predicts enhanced cholesterol accumulation, decreased lipid clearance and catabolism.”

In the paper

“APOE KO was confirmed by sequencing and APOE protein expression in astrocytes (Figure S3G).”

In the paper

“Principal component analysis (PCA) and Spearman correlation of 52 transcriptomes (4 cell types, 13 lines) obtained by RNA sequencing showed samples clustered by cell type but were indistinguishable by APOE genotype (Figures 1F, 1G and S1F), suggesting robust differentiation methods independent of APOE status.”

In the paper

“Potential gRNA off-targets were negative and local APOE haplotypes remained intact; isogenic iPSCs were karyotypically normal and genetically matched the donor fibroblasts (Tables S4 and S5).”

In the paper

“Suppressed lysosomal genes in APOE4 microglia predict cholesterol/lipid accumulation, decreased lipid catabolism and cholesterol transport (Figure 4O).”

In the paper

“Together, APOE4 hiPSC-mixed cortical cultures resemble human APOE4 AD brain, and the enriched common matrisome is associated with support of glia migration, inflammation and lipid synthesis, derived from APOE4 astrocytes potentially via astrocyte-neuron crosstalk.”

In the paper

“Thus, global transcriptomic analyses reveal that APOE4 is associated with higher cholesterol synthesis and lower catabolism/efflux, predicting altered intracellular cholesterol levels in APOE4 microglia and astrocytes.”

In the paper

“The most negatively enriched, Generic transcription pathway in APOE4 microglia represents decreased FXR/RXR inhibition caused by decreased low-density lipoprotein (LDL) (Figures 2C, 2J, S2C and S2D), suppressing lipid efflux potentially due to lower intracellular cholesterol.”

In the paper

“Indiv1 and 2 astrocytes clustered and were most similar (Figure S3H), suggesting APOE local haplotype dictates gene expression and is a major determinant in particular cell types.”

In the paper

“Together, these data support a mechanism of dysregulated cholesterol metabolism due to lysosomal sequestration of free cholesterol away from ER that misinforms APOE4 cells to respond as if intracellular cholesterol levels are low, leading to upregulated de novo cholesterol synthesis.”

In the paper

“increased cleaved SREBP2 upregulates cholesterol biosynthesis and cell surface lipid receptors to facilitate extracellular cholesterol uptake, however, uptake was dampened by actin cytoskeletal defects.”

In the paper

“Exogenous APOE4 compared to APOE3 impaired cholesterol efflux in rat neurons but not astrocytes (Michikawa et al., 2000).”

What the sources say

• ?Not checked — no full text of the sourcecites Michikawa et al., 2000 source

In the paper

“We confirmed enzymes (heat-inactivated serum (HI-S)) or other factors (knockout serum replacement (KOSR), cell density) were not driving the phenotype (Figures 7A, 7B and S7B–S7D).”

In the paper

“Therefore, LXR agonists rescued decoupled lipid/cholesterol metabolism in APOE4 astrocytes, restoring efflux to APOE3 levels.”

In the paper

“Under homeostatic conditions, APOE4 astrocytes exhibit actin cytoskeletal defects independent of exogenous lipid exposure.”

In the paper

“Proinflammatory matrisome dysregulation is in part induced by APOE4 astrocyte-neuron communication and worsens lipid dysregulation including cholesterol biosynthesis and efflux.”

In the paper

“Together, APOE4 drives lysosomal cholesterol sequestration in astrocytes which leads to lipid dysregulation by increasing cholesterol biosynthesis and decreasing efflux.”

In the paper

“All animal procedures and experiments were performed under guidelines approved by the animal studies committee at Washington University School of Medicine (D.M.H. protocol: 20180139).”

In the paper

“For each BMEC line, experiments were carried out in triplicate.”

In the paper

“Multiple biological and technical replicates (6 images per slide from N = 3 independent lines per genotype) of images were taken using a 20x magnification objective from mixed cortical cultures, pure astrocytes and microglia.”

In the paper

“which incorporates choline into phosphocholine in phosphatidylcholine biosynthesis (Wu et al., 2008).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Wu et al., 2008

In the paper

“We employed both population and isogenic cell type-based models and factored in GRS and APOE local haplotype to address the genetic complexity underlying human APOE4 risk to model AD.”

In the paper

“Experiments were repeated at least twice per pair.”

In the paper

“hiPSCs lines balanced for sex (Table S1) were generated by Icahn School of Medicine at Mount Sinai, UCI ADRC, WashU ADRC Induced Pluripotent Stem Cell Core from subject fibroblasts with approved Institutional Review Boards (IRB) and human Stem Cell Research Oversight (hSCRO) committee protocols.”

In the paper

“All tissue sections were processed within a single experiment to reduce variability.”

In the paper

“The correct APOE sgRNA sequence orientation was confirmed by Sanger sequencing and CRISPR/Cas9-APOE sgRNA plasmid cleavage efficiency was determined using the Surveyor mutation detection kit in 293T cells.”

In the paper

“HPCs were validated by FACS for CD43+(> 98%) and CD41+/CD235a+/CD45− (Biolegend) while microglia were validated by immunocytochemistry with microglia-specific markers.”

In the paper

“To ensure high quality genotype data, samples passed the quality control (QC) metrics (SNP call rate > 95%, minor allele frequency (MAF) > 5%, Hardy-Weinberg equilibrium (HWE) p-value > 1E-6, sample call rate > 95%) were analyzed.”

In the paper

“Experiment was repeated 3 times.”

In the paper

“Together these observations highlight the value of co-culture and organoid systems to extract the impact of APOE4 in complex cellular environments.”

In the paper

“thus, modulating gene expression may be an alternative to study disease-associated polymorphisms.”

In the paper

“Cells were continually passaged at 95% confluency and harvested as astrocytes at 30 DIV, validated immunocytochemically and/or by FACS for the astrocyte-specific markers and used for subsequent experiments.”

In the paper

“The APOE4 mixed cortical model recapitulates some aspects of the AD brain environment and implicates a reactive astrocyte state that upregulates ECM, cytokines and growth factors.”

In the paper

“Related samples were determined based on pair-wise identity-by-descent (IBD) estimation using PLINK. Duplicated samples with PI-HAT values between 0.99 to 1 were identified (see Table S1 and S4).”

In the paper

“Total protein concentration was determined by BCA method (ThermoFisher) and equal amounts of total protein were loaded onto 8% or 4–12% Bis-Tris Plus Gels (ThermoFisher).”

In the paper

“A recent 3D organoid model exhibits increased AD-related Aβ and APOE4-dependent tau pathology, (Zhao et al., 2020a)”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Zhao et al., 2020a

In the paper

“Camera and microscope settings were adjusted at the start of the experiment and maintained for uniformity.”

In the paper

“Enriched matrisome includes increased chemotaxis, inflammation and lipid synthesis (like pure astrocytes), and the proinflammatory state promotes cholesterol biosynthesis but impedes efflux, suggesting APOE4 alters matrisome signaling to exacerbate lipid dyshomeostasis.”

In the paper

“Upregulated ECM caused by astrogliosis is associated with amyloid plaques in AD and glial scar formation after CNS injury, creating a mechanical barrier for neurite outgrowth (Bovolenta et al., 1997; Pike et al., 1995).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Bovolenta et al., 1997
• ?Not checked — no full text of the sourcecites Pike et al., 1995 source

In the paper

“Mouse cortices balanced for sex (52% female, sex per each brain is labeled in the GEO metadata in the key resource table) were dissected in Hanks’ Balanced Salt solution (HBSS without Ca2+, Mg2+).”

In the paper

“Since altered matrisome pathways in astrocytes from AD brain were recapitulated in APOE4 hiPSC-mixed cortical cultures, incorporating microglia in 2D may more faithfully model the in vivo context.”

In the paper

“However, APOE LOF is limited to particular individuals, suggesting that individual donor genetics dominate over single risk allele conversion or KO in isogenics, which results in discordant outcomes among studies.”

In the paper

“The scored subjects are independent from the AD GWAS study used to generate the original weighted score.”

In the paper

“These studies suggest that therapeutic approaches aimed at restoring glial lipid homeostasis and inflammation may be beneficial for AD, particularly in APOE4 carriers.”

In the paper

“Variation in APOE levels may be due to differences in APOE local haplotype for each individual and differing gene regulation in different cell types, leading to cell type-specific individual variance.”

In the paper

“The discordant phenotypes due to APOE4 among studies may result from introducing APOE ε3 or ε4 into different haplotypic backgrounds, while clonal variability < individual variability.”

In the paper

“however, our hiPSC study could only model APOE genotype but not AD neuropathological phenotypes or age.”

In the paper

“Using the same cortical neuron protocol as our study, the neurons studied are likely a mixed neuron/astrocyte population, although RNA sequencing was not performed.”

In the paper

“Based on the consistent results achieved using both BRETIGIA and DSA, we therefore chose DSA for deconvolution following cell type proportion determination.”

In the paper

“HPCs were validated by FACS for CD43+(> 98%) and CD41+/CD235a+/CD45− (Biolegend) while microglia were validated by immunocytochemistry with microglia-specific markers.”

In the paper

“All images were captured and analyzed by a blind observer using coded slides.”

In the paper

“Sex effects were corrected by linear regression.”

In the paper

“Other disease relevant genetic factors (GRSnoAPOE, APOE local haplotype) were considered in our model, however aging or epigenetics cannot be recapitulated in hiPSC models yet.”

In the paper

“Filipin images were taken within 24 hours of staining to limit bleaching in fluorescence intensity per exposure.”

In the paper

“confirmed >90% purity by FACS using an antibody for myelin basic protein (Milteny Biotec, REA1154).”

In the paper

“Wang et al. characterized APOE4 neuronal phenotypes in 6 subjects and a single clone of each genotype from one isogenic pair.”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Wang et al., 2018

In the paper

“To test if known gene ontology (GO) and pathways are enriched for DEGs, fast preranked gene set enrichment analysis (fGSEA) was performed using GO annotations and canonical pathways (Biocarta, KEGG and Reactome) available from the Molecular Signatures Database (MSigDB) with FDR<0.1 as a significance threshold (Subramanian et al., 2005), following 1 million gene set permutations.”

What the sources say

• Not checked yetcites Subramanian et al., 2005 source

In the paper

“One challenge was collecting an equal number of cases vs. controls for each APOE genotype to confirm that differences were attributed to APOE4 but not unknown effects of disease status.”

In the paper

“Cell number was determined in order to load equal total protein.”

In the paper

“For example, enriched Matrisome is not only driven by APOE4 genotype but also by AD pathological phenotypes in the context of APOE3”

In the paper

“The correct APOE sgRNA sequence orientation was confirmed by Sanger sequencing and CRISPR/Cas9-APOE sgRNA plasmid cleavage efficiency was determined using the Surveyor mutation detection kit in 293T cells.”

In the paper

“As a control we treated the cells with 2.5-fold excess unlabeled LDL in presence of Alexa546-LDL to ensure that the Alexa546 signal was not non-specific.”

In the paper

“After clonal expansion, isolated genomic DNA was used as a template to amplify an APOE fragment by PCR to screen for loss of the NotI restriction site. DNA from clones lacking a NotI restriction site were subjected to Taqman SNP genotyping and gene editing validated by Sanger sequencing to identify homozygous gene-edited hiPSCs.”

In the paper

“Informed consent was received by each of the participants who donated fibroblasts.”

In the paper

“In order to avoid bias, different pairs of APOE3 and APOE4 clones were used.”

In the paper

“Potential off-targets (quality score≥0.5) for the designed gRNA were confirmed negative.”

In the paper

“The sva v3.34.0 “ComBat” function was used to adjust for the following covariates: sex, RIN, batch, cell type proportion for mixed cortical cultures, and GRSnoAPOE (GRS-APOE locus) for population samples and individual for isogenic samples.”

In the paper

“Before cell type proportion inference, we corrected for: postmortem interval (PMI), age of death (AOD), RIN, exonic rate, race, rRNA rate, sex and batch.”

In the paper

“Only modules with p<0.1 were considered to be differentially expressed.”