Menu

Close

Cholesterol and matrisome pathways dysregulated in human APOE ε4 glia

Julia TCW, Shuang A. Liang, Lu Qian, Nina H Pipalia, Michael J. Chao, Yang Shi, Sarah E. Bertelsen, Manav Kapoor, Edoardo Marcora, Elizabeth Sikora, David M. Holtzman, Frederick R. Maxfield, Bin Zhang, Minghui Wang, Wayne W. Poon, Alison M. Goate

Preprint posted on July 25, 2019 https://www.biorxiv.org/content/10.1101/713362v1

Back to the beginning in Alzheimer’s disease research: An extensive RNAseq study reveals that Apolipoprotein E isoforms differentially regulate lipid metabolism in human brain glia cells. Intriguing: discrepancies between human and mouse were found.

Selected by Theresa Pohlkamp

Background

The highest genetic risk factor for Alzheimer’s disease (AD) is Apolipoprotein E (ApoE) isoform ε4 (APOE 4). Three major ApoE isoforms exist in the human population: APOE 2, APOE 3, and APOE 4. Whereas APOE 2 has protective features, APOE 4 increases the risk for the disease. Age is the highest general risk factor to develop AD; each APOE 4 allele decreases the age of disease onset by 3-5 years. However, it is not well understood how APOE 4 contributes to AD pathology. Since its discovery as genetic risk factor for AD in the mid-90s scientists attributed toxic and/or defective functions to APOE 4.

The general function of ApoE is to nurture cells with lipids like fat and cholesterol, the major components of the cell membranes. In the brain, astrocytes provide the major source of ApoE. Unless stressed, other cells like microglia, the immune cells of the brain, and neurons only produce small amounts of ApoE. The difference between APOE 3 and APOE 4, the two isoforms this preprint is focusing on, is a substitution of only one amino acid. This substitution affects the structure and biochemical properties of APOE with regard to lipidation, degradation, toxicity, and intracellular trafficking. To study APOE 3 and APOE 4 in vivo, human APOE 3 and APOE 4 targeted replacement mice (APOE 2-TR, APOE 3-TR, APOE 4-TR), as well as Apoe-knockout (Apoe KO) mice have been created and are used intensely in the AD-field (reviewed in Balu et al., 2019). This preprint reveals detailed insight into how human APOE 4 compared to APOE 3 affects the transcriptome of diverse brain cell types of human and mouse.

 

What was done?

The authors analyzed the transcriptome of up to three homologous APOE genotypes (APOE 3/APOE 3 = 33, APOE 4/APOE 4 = 44, and KO/KO) in up to four brain cell types derived from human or mouse. Thirteen (7xAPOE 33, 6xAPOE 44) human induced pluripotent stem cell (hiPSC) lines were differentiated into microglia, astrocytes, mixed cortical cells (neurons and 5-25% astrocytes), and brain microvascular endothelial cells (BMEC). Isogenic glia (astrocytes and microglia) lines were obtained by mutating APOE 44 hiPSC lines to APOE 33 by using CRISPR/Cas9. Isogenic lines were used for experimental studies. Mouse primary glia were cultured from cortical tissue of APOE 33-TR, APOE 44-TR, and Apoe KO pups. RNA sequencing (RNAseq) was performed on human and mouse cell lines. Publicly available transcriptomic data of AD brain samples derived from postmortem patients (APOE 33 and APOE 44) were used to compute cell-line specific transcriptomes following a deconvolution algorithm.

 

What was found?

Lipid metabolism related pathways are dysregulated in human but not mouse APOE 4 glia

RNAseq analysis of hiPSC-derived cell lines showed APOE genotype-dependent gene expression discrepancies in astrocytes, microglia, and mixed cortical cultures, but not brain microvascular endothelial cells. The authors found that specific pathways were differentially regulated in APOE 4 versus APOE 3 glia, with greatest discrepancies in lipid metabolism pathways. Differences in gene expression predicted an increase in cholesterol synthesis and metabolism in APOE 4 glia. A reduced lipid/cholesterol efflux, lysosomal accumulation of cholesterol, and a decrease in lipid clearance and catabolism were predicted for microglia (Figure), only. The deconvoluted glia transcriptome of human AD brain tissue supported these findings.

Figure: Functional pathway analysis of hiPSC-derived microglia. Red and orange colors are upregulated, green and blue are downregulated genes/functions in APOE 4 compared to APOE 3. Reproduced with permission from Figure 2g of the preprint.

Intriguingly, the mouse primary cortical glia transcriptomes revealed different results. Whereas extracellular matrix and inflammation associated pathways were upregulated in mouse APOE 4 glia, no significant difference was observed in lipid metabolism pathways. In general, higher discrepancies were found in Apoe KO glia: most strikingly lipid metabolism pathways were upregulated, but also inflammation and extracellular matrix (ECM) related pathways were increased. Thus, dysregulations in Apoe KO are distinct from those linked to APOE 4.

APOE 4-mediated upregulation of matrisome pathways in glia requires the presence of neurons

The matrisome is an ensemble of ECM proteins and associated factors involved in cell-cell communication. Within the hiPSC-derived mixed cortical culture, matrisome related genes involved in chemotaxis, inflammation, and lipid-synthesis pathways were upregulated in APOE 4. Similar findings for APOE 4 compared to APOE 3 genotype were obtained in glia from cell-type deconvoluted RNAseq data of human AD brain tissue. The authors concluded that specific ApoE-isoform dependent communication pathways in glia are only activated in the presence of neurons. A transcriptome comparison across APOE 33 carriers, grouped by various criteria of AD related phenotypes, showed that matrisome pathways were enriched in AD. This enrichment pattern was independent of APOE genotype and also found in APOE 44 carriers. Hence, the matrisome seems to be upregulated by APOE 4 and other AD factors.

Experimental validation and mechanistic findings: decoupled lipid metabolism and increased inflammation in APOE 4 versus APOE 3 glia

To validate the RNAseq findings experimentally, hiPSC-derived, isogenic APOE 44 and APOE 33 glia were used to study cholesterol metabolism. APOE 4 astrocytes had increased free cholesterol levels but no changes in cholesteryl ester. After applying excess LDL to the cultures, cholesterol accumulated more intracellularly. This defective cholesterol accumulation was found in lysosomes of APOE 4 astrocytes. Intracellular and secreted APOE levels, as well as ABCA1 were much lower in APOE 4 than APOE 3 astrocytes. ABCA1 is the APOE lipidating transporter protein, also known as the cholesterol efflux regulatory protein, thus, these findings indicate a decreased cholesterol efflux. In total this suggests a decoupling of lipid synthesis and catabolism in APOE 4 astrocytes. In addition, APOE 4 astrocytes secreted higher amounts of proinflammatory chemokines, cytokines, and growth factors.

Overall, the findings suggest that restoring lipid homeostasis in glia by targeting the structure and/or function of APOE 4 provide a promising AD-treatment approach.

 

What I like about this paper

The study demonstrates that the involvement of APOE 4 in AD pathology likely falls back to the key role of ApoE in lipid metabolism and discovered a novel gene set/pathway called “matrisome” in APOE 4 and AD case. These findings are of tremendous importance to the field, since they make us researchers think why APOE-isoforms contribute to coronary artery disease, myocardial infarction, and AD in the same isoform specific stepwise pattern: APOE 4>APOE 3>APOE 2. Whereas this work demonstrates potential drawbacks of animal models for complex human diseases, it suggests that hiPSC-derived mixed cortical cultures provide a translatable in vitro model to study APOE 4-dysregulated pathways in AD.

 

Questions and future directions

(1) The Apoe gene promoter and the 5’UTR have not been replaced in APOE-TR mice, likely causing differences in gene regulation (Maloney et al., 2007) of human APOE isoforms in human versus mouse. Do you think this is causing the discrepancies in your study when you compare mouse and human glia?

(2) APOE expression was highest in BMEC, which was surprising to me. With regard to the high APOE-expression it is also surprising that these cells show the smallest changes in differentially expressed genes dependent on APOE isoforms, did you expect that?

(3) Did you compute the neuronal transcriptome? How does APOE affect transcription in neurons?

(4) Future studies could test whether (frequent) medium exchange between APOE 3 and APOE 4 glia can reverse the differences seen in lipid metabolism.

(5) Future studies could also investigate the effect of APOE-loss and APOE 2 in hiPSC-derived isogenic glia.

 

References

Balu D, Karstens AJ, Loukenas E, Maldonado Weng J, York JM, Valencia-Olvera AC, LaDu MJ. The role of APOE in transgenic mouse models of AD. Neurosci Lett. 2019 May 28;707:134285. Review.

Maloney B, Ge YW, Alley GM, Lahiri DK. Important differences between human and mouse APOE gene promoters: limitation of mouse APOE model in studying Alzheimer’s disease. J Neurochem. 2007 Nov;103(3):1237-57. Epub 2007 Sep 8. PubMed PMID: 17854398.

 

 

Tags: lipoprotein receptors, neurodegeneration, trem2

Posted on: 12th August 2019

Read preprint (No Ratings Yet)




  • Author's response to "Questions and future directions"

    Julia TCW shared

    (1) This is an important point. Taconic’s APOE targeted replacement mouse model is developed to identify the role of human APOE allelic functions (E4, E3 and E2) by replacing only the coding region (Exon 2-4) of the mouse endogenous Apoe with one of the human APOE alleles, see Fig. 1 in (Sullivan et al., 1997). Therefore, the APOE-TR mouse contains all normal mouse regulatory sequences (5′-UTR including non-coding exon one) together with the human protein coding exons. It is one possibility that the difference in regulation of Apoe in mouse compared to APOE in human results in difference in transcriptional levels in each case of our study. It is also possible that lipid regulatory gene sets (potential downstream genes influenced by APOE 4) in mouse vs human are differentially regulated in glia of each brain species.

    (2) I’m glad that you connected the APOE expression level with DEGs in BMECs. This is an interesting discovery. No, we did not expect that there were no difference by the genotype in BMECs. When the study was designed, the supporting evidence that we decided to use endothelial cells is due to the evidence that the Apoe-GFP reporter mouse shows Apoe expression in endothelial cells (Xu et al., 2006).

    (3) Yes, I did compute the neurons from the whole brain transcriptome, following the cell-type deconvolution. There are many interesting signals showing up, e.g. positively enriched gene sets including oxidative phosphorylation, TCA cycle and respiratory electron transport and negatively enriched gene sets including ribosome, neuroactive ligand receptor interaction in APOE 44 compared to APOE 33 in AD. Since we were focus on the most significantly contributed cell-type, which is glia, by APOE 4 followed by in vitro validation and deciphering the mechanism using hiPSC-derived isogenic APOE human glia, the study focus became glia in this preprint. With a field effort heading towards single cell RNAseq and multiple 2D and 3D neuronal models, this work will help us address the communication in neural network in which sub-type of neurons is affected by APOE 4 or AD glia in vitro.

    (4) In our preprint, the major defect found in APOE 44 glia is decreased lipid efflux due to the inhibition of LXR/RXR based on our analyses. If you administer conditioned medium (CM) from APOE 33 glia to APOE 44 glia, I would hypothesize that APOE 44 cells accumulate more lipids from higher level of effluxed lipids in CM from APOE 33 cells like the result of excessive LDL treatments (Fig. 5c) shown in the preprint. To reverse the phenotype of accumulated lipids in APOE 44 glia, we might want to increase lipid efflux by activation of LXR/RXR or enhance intracellular lipid digestion system as we observed differences in a lysosome pathway in mouse vs human. Additionally, it will be interesting to further investigate APOE-genotype dependent effects of glia to other cell types in brain, for example, APOE 3’ ability to promote neurite outgrowth in culture compared to APOE 4 (Fagan et al., 1996).

    (5) Considering our finding that Apoe knockout (KO) mouse microglia only present lipid dysregulation, it will be interesting to compare it to human APOE KO. Especially generating APOE 22 isogenic lines is critical, since it is rare in the population (allele frequency in general population <8% and in LOAD <2%). I’m currently engineering APOE 22 isogenic lines from these multiple isogenic lines to further investigate the role of human APOE 2 gene in glia of human vs mouse brain. This will be an up and coming story with additional resources that I’d like to contribute to the field.

     

    References

    Fagan, A.M., Bu, G., Sun, Y., Daugherty, A., and Holtzman, D.M. (1996). Apolipoprotein E-containing high density lipoprotein promotes neurite outgrowth and is a ligand for the low density lipoprotein receptor-related protein. The Journal of biological chemistry 271, 30121-30125.

    Sullivan, P.M., Mezdour, H., Aratani, Y., Knouff, C., Najib, J., Reddick, R.L., Quarfordt, S.H., and Maeda, N. (1997). Targeted replacement of the mouse apolipoprotein E gene with the common human APOE3 allele enhances diet-induced hypercholesterolemia and atherosclerosis. The Journal of biological chemistry 272, 17972-17980.

    Xu, Q., Bernardo, A., Walker, D., Kanegawa, T., Mahley, R.W., and Huang, Y. (2006). Profile and regulation of apolipoprotein E (ApoE) expression in the CNS in mice with targeting of green fluorescent protein gene to the ApoE locus. The Journal of neuroscience : the official journal of the Society for Neuroscience 26, 4985-4994.

     

    1 comment

    5 days

    Theresa Pohlkamp

    I am very excited for the authors! The paper is now available on the SneakPeek server of Cell:
    https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3435267
    It does contain additional data, especially in the last figure!

    1

    Have your say

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.

    Sign up to customise the site to your preferences and to receive alerts

    Register here
    Close