What this is
- This research investigates how different () isoforms affect the () and (Aβ) interactions.
- Using isogenic induced pluripotent stem cells (iPSCs), the study differentiates these into brain microvascular endothelial cell-like cells (BMEC-like) and pericyte-like cells.
- Findings indicate that APOE4 impairs Aβ clearance, while APOE2 enhances it, providing insights into the roles of these isoforms in Alzheimer's Disease.
Essence
- APOE4 exacerbates amyloid deposition and clearance issues at the , while APOE2 shows a protective effect by enhancing amyloid transport. This study underscores the importance of isoforms in Alzheimer's Disease pathology.
Key takeaways
- APOE4 significantly reduces the transport capabilities of BMEC-like cells for Aβ40, suggesting a role in decreased amyloid clearance. In contrast, APOE2 enhances Aβ42 transport, indicating a potentially protective role.
- Pericyte-like cells expressing APOE4 show increased extracellular amyloid deposition, while those expressing APOE2 exhibit the least deposition. This aligns with known vascular pathologies in Alzheimer's patients.
- iPSC-derived models effectively capture -related amyloid pathologies, offering a platform for future Alzheimer's Disease research and drug development.
Caveats
- The study does not establish direct causation between isoforms and amyloid pathology but rather correlates their effects. Further research is necessary to clarify mechanisms.
- The iPSC-derived models may not fully replicate in vivo conditions, limiting the generalizability of findings to human Alzheimer's Disease pathology.
Definitions
- Blood-brain barrier (BBB): A selective permeability barrier that separates the bloodstream from the central nervous system, regulating substance entry and maintaining homeostasis.
- Apolipoprotein E (APOE): A protein involved in lipid metabolism, with three common isoforms (APOE2, APOE3, APOE4) that influence Alzheimer's Disease risk.
- Amyloid beta (Aβ): A peptide that aggregates to form plaques in the brains of Alzheimer's Disease patients, contributing to neurodegeneration.
AI simplified
Background
The blood-brain barrier (BBB) is a dynamic interface that separates the bloodstream and the central nervous system (CNS), helping to maintain CNS homeostasis. BBB properties are largely characteristics of brain endothelial cells, that are in turn modulated by other cellular constituents of the neurovascular unit (NVU), such as pericytes, astrocytes, neurons, and perivascular macrophages [1, 2]. In healthy individuals, the BBB safeguards the CNS from deleterious substances by providing both a passive barrier and a host of transporters that combine to regulate the entry and efflux of many small molecules, proteins, and cells [3]. However, in many CNS diseases, the BBB not only provides a significant barrier for the delivery of many neurotherapeutics, but also contributes to disease pathogenesis. For example, increased BBB permeability has also been observed in several neurodegenerative diseases, including Alzheimer’s Disease (AD) and Parkinson’s Disease (PD), but the role of the BBB in the pathophysiology of these diseases is not yet clear [4, 5].
AD is the leading cause of dementia, affecting more than 30 million people worldwide [6, 7]. At the cellular level, synaptic dysfunction, neurodegeneration, inflammation, and vascular dysfunction are common pathological features of AD [8–10]. In recent years, genetic risk factors for AD have been identified, among which alleles of the APOE gene are the strongest and most prevalent [7, 11, 12]. APOE encodes for apolipoprotein E, a 34 kDa glycoprotein that plays a crucial role in lipid metabolism by binding to cholesterol and phospholipids and facilitating their transport throughout the body [7, 13]. In the periphery, the liver is the predominant producer of APOE [14]. In the CNS, astrocytes, microglia and pericytes have been shown to secrete APOE [15, 16]. There are three dominant isoforms of APOE, namely APOE2, APOE3 and APOE4, which differ by only two single nucleotide polymorphisms (SNPs) [15]. Among the three isoforms, APOE3 is the most prevalent in the human population [17]. APOE4 allele carriers have significantly increased AD risk [18] and the APOE2 allele is considered a protective isoform against AD [19]. While the exact mechanism by which different APOE isoforms contribute to AD pathophysiology remains to be elucidated, human and animal studies have consistently shown that APOE4 is linked to increased deposition of amyloid beta (Aβ) plaques in the brain [18, 20–25], impaired intracellular lysosomal degradation of Aβ [26–29], and increased levels of neuroinflammation [30–33].
The BBB plays a central role in Aβ trafficking and is thus heavily involved in AD pathogenesis [34–37]. It is estimated that the BBB clears more than 85% of all Aβ produced in the CNS through various pathways [38, 39]. Many studies have demonstrated different aspects of BBB breakdown and dysregulation in AD patients, such as increased hippocampal BBB permeability [40, 41], impaired Aβ efflux from the CNS [42, 43], reduced pericyte coverage of CNS microvessels [44, 45], infiltration of peripheral immune cells into the CNS [46, 47], and reduced tight junction gene expression and basement membrane alterations [48–51]. These BBB pathologies are more pronounced in patients that are APOE4 carriers [52], including more severe BBB breakdown [21, 53, 54], accelerated degeneration of brain pericytes [55–57], increased amyloid deposition [24], and worsened cognitive decline [58, 59]. Moreover, APOE4 carriers have lower levels of peripheral blood Aβ and cerebral spinal fluid (CSF) Aβ [60–62], consistent with impaired clearance. Despite the prevalence of vascular pathologies in AD patients, the causal relationship between BBB damage and AD remains a subject of research [63].
To investigate the role of different APOE isoforms on BBB function and amyloid deposition and transport, we used established protocols to differentiate isogenic human induced pluripotent stem cells (iPSCs) expressing different APOE alleles to brain microvascular endothelial cell-like cells (BMEC-like cells) and brain pericyte-like cells [64–67]. We used commercially available, isogenic iPSC lines that were created with CRISPR/Cas9 genome editing of both APOE alleles to produce APOE KO/KO, APOE2/APOE2, APOE3/APOE3 and APOE4/APOE4 lines to minimize the confounding effect of genetic background. We found that iPSC-derived BMEC-like cells and pericyte-like cells with different APOE isoforms are similar in terms of BBB-related gene expression and key barrier and transporter phenotypes. However, APOE2 BMEC-like cells demonstrated improved Aβ42 clearance compared to APOE3 and APOE4 counterparts while recombinant APOE4 protein led to decreased Aβ42 clearance compared with APOE2 and APOE3 proteins. In pericyte-like cells, APOE4 pericyte-like cells had increased extracellular Aβ42 deposition and APOE2 pericyte-like cells had reduced deposition compared to APOE3 pericyte-like cells. Collectively, our findings are consistent with clinical observations that APOE4 leads to higher amyloid load while APOE2 leads to a lower amyloid load.
Methods
Isogenic iPSC differentiation to BMEC-like cells and brain-like Pericytes
Isogenic iPSC lines with different APOE genotypes iPS16 (APOE4/E4, Alstem, abbreviated as APOE4 in this manuscript), iPS26 (APOE3/E3, Alstem, abbreviated as APOE3 in this manuscript), iPS36 (APOE KO/KO, Alstem, abbreviated as APOE KO in this manuscript), iPS46 (APOE2/E2, Alstem, abbreviated as APOE2 in this manuscript) were obtained. iPSCs were cultured between passages 20–50 on Matrigel (Corning, 354277) with daily mTESR1 (StemCell Technologies, 85850) medium changes.
BMEC-like cells were differentiated as previously described [64, 65, 68–70]. In brief, iPSCs were singularized with Accutase (Life Technologies, 00-4555-56) and expanded to 30,000 cells/cm2 prior to the initiation of the differentiation. Unconditioned medium (UM), which is prepared by mixing 100 mL Knock-out serum replacement (Life Technologies, 10828028), 5 mL MEM non-essential amino acids solution (Life Technologies, 11140050), 2.5 mL of GlutaMAX (Life Technologies, 35050061), 392.5 mL of DMEM/F12 (Life Technologies, 11320033) and 3.5 µL of β-mercapto-ethanol (Sigma, 444203), was changed daily for 6 days. UM was changed to hECSR medium, which is human Endothelial Serum-Free medium (Life Technologies, 11111044), 20 ng/mL basic fibroblast growth factor (Peprotech, 100-18B) and 2% B-27 (Life Technologies, 17504044), for 48 h supplemented with 10 µM retinoic acid (StemCell Technologies, 72264).
Pericytes were derived in a two-step process as described previously [67, 70]. Initially iPSCs were differentiated in Essential 6 medium (Life Technologies, A1516401) supplemented with 1 μm dorsomorphin (StemCell Technologies, 72102), 10 μm SB431542 (Tocris, 1614), 1 μm CHIR99021 (Tocris, 4953), 10 ug/L basic fibroblast growth factor (Peprotech, 100-18B), and 22.5 mg/L heparin (Sigma, H3393) to a neural-crest stem cell population. CD271 + cells were enriched with neural crest stem cell microbeads (Miltenyi, 130-097-127) and seeded at 10,000 cells/cm2 onto tissue culture-treated plates. Following 6 days of expansion in Essential 6 medium (Life Technologies, A1516401) supplemented with 10% FBS (Life Technologies, A5256701), pericytes were harvested for analysis.
Aβ transcytosis assays
Transwells (Corning, 3460) pre-coated with 200 uL 0.5 mg/mL collagen IV (Sigma, C5533) overnight at 37 °C. iPSC-derived BMEC-like cells were seeded at 1 million/cm2 on Transwells (Corning, 3460) in hECSR media. Recombinant human Aβ40 (Anaspec, AS-24235) or recombinant human Aβ42 (Anaspec, AS20276) were prepared in 8 M urea 100 mM glycine buffer at pH = 10 to a concentration of 12 mg/mL and stored in aliquots at -80 °C. This buffer helps to maintain Aβ in monomeric and denatured state. Stocks were then diluted to 500 nM in hECSR medium immediately before experiments. For experiments conditions when recombinant APOE is added, recombinant APOE2 (Peprotech, 350 − 12), APOE3 (Peprotech, 350-02) or APOE4 (Peprotech, 350-04) were added at concentration of 500 nM. In transcytosis assay, 1.5 mL of hECSR medium with Aβ and/or recombinant APOE was added to the basolateral chamber. 0.5 mL of hECSR media that is Aβ-free and APOE-free was added to the apical chamber. The Transwells were then incubated at 37 °C on a 300 rpm orbital shaker for 3 h. Media samples from the apical chamber were taken and analyzed. To quantify Aβ concentrations in the media samples, ELISA kits for Aβ40 (Life Technologies, KHB3481) or Aβ42 (Life Technologies, KHB3441) were used.
Trans-endothelial electrical resistance (TEER) measurements
EVOM 2 with STX4 electrodes (World Precision Instruments) were used to measure the TEER value in Ω after BMEC differentiation and seeding on Transwells on Day 10. The values are normalized by subtracting the TEER reading from a collagen IV-coated blank well and then multiplied by the surface area of the Transwell insert and reported as Ω∙cm2.
P-glycoprotein efflux transporter assay
BMEC-like cells were seeded on collagen IV-coated plates at 100,000 cells/cm2. Half of the wells of each experimental group were treated with 10 μm Cyclosporin A (CsA, Tocris, 1101) for 30 min in hECSR medium, with the other half treated with DMSO (Sigma, D2650) in hECSR. After incubation, a solution of 10 μm rhodamine 123 (Rh123, Life Technologies, R302) in hECSR with and without CsA, was added to corresponding wells. The plates were kept at 37 °C for one hour at 300 rpm on an orbital shaker. After incubation, cells were washed three times with ice cold DPBS (Life Technologies, 14040117) and then lysed with RIPA buffer (Life Technologies, 89901). We used a fluorescent plate reader to measure the fluorescence of Rh123 accumulated inside the cells. We used a BCA assay kit (Life Technologies, 23225) to quantify total protein from cell lysates. The fluorescence readings were normalized to total protein concentrations of the lysates.
Immunocytochemistry
Cells were fixed with − 20 °C methanol (Sigma, 67-56-1) or 4% paraformaldehyde (PFA; Electron Microscopy Sciences, 15700) in DPBS for 15 min at room temperature. Following three washes in DPBS the cells were blocked in 10% goat serum (Sigma, G9023) in PBS for 30 min. Cells were then incubated with primary antibodies at indicated dilution ratios at 4 °C overnight (Table 1). Cells were then washed with PBS three times. Cells were then incubated with secondary antibodies at the indicated dilution ratios (Table 1) and 20 μm Hoechst 33,342 (Thermo Scientific, 62249) for 1 h at room temperature. Cells were then washed three times. For widefield fluorescence microscopy, we used a Nikon Eclipse Ti2 microscope. For confocal microscopy, we treated samples with ProLong Gold Antifade (Life Technologies, P36941↗) and imaged using a Nikon A1R microscope. For quantification of immunocytochemistry images, background signals from a secondary antibody only control were subtracted from all conditions. Area fraction index was calculated as previously reported [71].
| Antibody | Source | Dilution ratios |
|---|---|---|
| Anti-claudin-5 (mouse monoclonal IgG1, clone 4C3C2) | Invitrogen 35-2500 | 1:100 (ICC) |
| Anti-OCLN (mouse monoclonal IgG1, clone OC-3F10) | Invitrogen 33-1500 | 1:100 (ICC) |
| Anti-LRP1 (rabbit monoclonal IgG, clone SA0290) | Invitrogen MA5-31959 | 1:100 (ICC) |
| Anti-CDH5 (mouse monoclonal IgG2a, clone BV9) | Santa Cruz Biotechnology sc-52,751 | 1:100 (ICC) |
| Anti-NG2 (mouse monoclonal IgG2a, clone 9.2.27) | Millipore MAB2029 | 1:50 (Flow Cytometry) |
| Anti-PDGFRβ (mouse monoclonal IgG2a, clone 28D4) | BD Biosciences, 558,820 | 1:100 (Flow Cytometry) |
| Anti-Aβ (mouse monoclonal IgG2a, clone B-4) | Santa Cruz Biotechnology sc-28,365 FITC | 1:100 (Flow Cytometry) |
| Alexa Fluor 488 goat anti-mouse IgG (goat polyclonal) | Invitrogen A-28,175 | 1:200 (ICC), 1:200 (Flow Cytometry) |
| Alexa Fluor 647 goat anti-rabbit IgG (goat polyclonal) | Invitrogen A-21,245 | 1:200 (ICC), 1:200 (Flow Cytometry) |
Reverse transcription real-time PCR analysis
RNA was extracted using Direct-zol RNA Miniprep kit (Zymo, R2050). Reverse transcription was performed to obtain cDNA using GoScript Reverse Transcriptase with Oligo(dT) kit (Promega, A2791). Real-time gene expression analysis was conducted using 25 µl reactions containing SYBR Green PCR Master Mix (Life Technologies, 4309155) along with primers specific for genes of interest (Table 2). PCR was run according to manufacturer protocols on an Agilent AriaMX Real-Time PCR system.
| Gene | Forward primer (5’-3’) | Reverse primer (5’-3’) |
|---|---|---|
| CLDN5 | TGACCTTCTCCTGCCACTA | AAGCGAAATCCTCAGTCTGAC |
| OCLN | ATGGCAAAGTGAATGACAAGC | AGGCGAAGTTAATGGAAGCTC |
| TJP1 | CGCGTCTCTCCACATACATTC | GCTGGCTTATTCTGAGATGGA |
| SLC2A1 | GTGCCATACTCATGACCATCG | GGCCACAAAGCCAAAGATG |
| ABCB1 | ACTCACTTCAGGAAGCAACC | GATTGACTGAATGCTGATTCCTC |
| ABCG2 | CTCAGATCATTGTCACAGTCGT | GTCGTCAGGAAGAAGAGAACC |
| LRP1 | TCCAGTACAGATTGTCTCCCA | ATCTACTTTGCCGACACCAC |
| AGER | ATTCTGCCTCTGAACTCACG | TCCTTCACAGATACTCCCTTCT |
| APOE | GGGTCGCTTTTGGGATTACCTG | CAACTCCTTCATGGTCTCGTCC |
Flow cytometry-based Aβ deposition analysis
After differentiation, pericyte-like cells were treated with Essential 6 medium (Life Technologies, A1516401) with or without 500nM Aβ42 (Anaspec, AS20276). Cells were incubated for 18 h at 37 °C on a 300 rpm orbital shaker. The extended incubation time allowed Aβ42 to interact with pericyte-like cell-secreted APOE, aggregate, and form extracellular amyloid deposits. After incubation, pericyte-like cells were treated with Accutase (Life Technologies, 00-4555-56) for 7 min to singularize the cells. Cells were removed from the plate, washed three times with cold PBS, and fixed with 4% PFA (Electron Microscopy Sciences, 15700). Fixed pericyte-like cells were then incubated with Alexa Fluor 488-conjugated anti-Aβ antibody (Biolegend, 6E10) at 1:100 dilution ratio in eBioscience Flow Cytometry Staining Buffer (Life Technologies, 00-4222-26) for 1 h. Pericyte-like cells were then washed three times with cold PBS, and analyzed on a BD Accuri C6 flow cytometer.
Statistics
Biological replicates in this manuscript refer to individual wells of cultured cells that underwent identical experimental treatments. The authors of the study ensured that all key experiments were repeated using multiple independent iPSC differentiations from the isogenic iPSC lines. The replication strategy for each experiment is stated in figure legends. To compare means of two experimental groups, Student’s t test was used. For experiments with three or more experimental groups, one-way analysis of variance (ANOVA) was used for comparison of means. For experiments with two factors of experimental conditions, two-way ANOVA was used. Following ANOVA, Dunnett’s post hoc test was used for comparison of multiple treatments to a single control, or Tukey’s honest significant difference test was used for multiple pairwise comparisons. Statistical tests were performed in GraphPad Prism. Descriptions of the statistical tests used are provided in figure legends.
Results
iPSC-derived BMEC-like cells with differentisoforms have similar barrier and transporter properties APOE
Established protocols [64, 65, 68] were used to differentiate isogenic iPSC cell lines with homozygous APOE2, APOE3, APOE4 or APOE Knockout (APOE KO) isoforms into BMEC-like cells. Briefly, the iPSCs were differentiated in unconditioned medium (UM) for six days, maintained in endothelial cell (EC) medium for two days and the mixed differentiating cell populations were plated on collagen IV matrix to purify the BMEC-like cells (Fig. 1A). In addition, flow cytometry analysis of endothelial markers claudin-5 and VE-cadherin revealed that isogenic BMEC-like cells, regardless of APOE genotypes, are homogenous cell populations (Ext. Figure 1D). The resulting cells possess BMEC-like properties as previously described, including tight junction protein expression (Fig. 1B and E, Ext. Figure 1A), tight barrier function (Fig. 1D, Ext. Figure 1E), and efflux transporter gene expression (Fig. 1G). Immunocytochemistry was used to visualize and quantify the expression and localization of junctional proteins occludin and claudin-5, and we observed no significant differences in junctional area fraction index or mean fluorescence intensity of these proteins at junctions as a function of APOE genotype (Fig. 1B, C, E and F). The iPSC-derived BMEC-like cells were also seeded onto Transwells for quantification of trans-endothelial electrical resistance (TEER). We found that BMEC-like cells with different APOE genotypes all developed an indistinguishably tight barrier with TEER greater than 3000 Ω∙cm2 (Fig. 1D). Monolayers formed by isogenic BMEC-like cells also demonstrate similar permeability to sodium fluorescein (Ext. Figure 1E). By RT-qPCR, we found that BMEC-like cells with different APOE genotypes expressed similar levels of APOE transcript (Ext. Figure 1B), and the concentration of APOE in cell culture media conditioned by isogenic BMEC-like cells for 24 h was below the limit of detection of ELISA (Ext. Figure 1C), indicating that the APOE protein was not substantially synthesized in BMEC-like cells. Since the APOE KO line was generated by introducing a 100 bp homozygous deletion in the exon 2 of APOE gene, which is not at the region of our qPCR primers, we were still able to detect mRNA transcripts for APOE by qPCR, but APOE protein expression was absent by ELISA. (Ext. Figure 1B and C). We assessed the transcript expression of several canonical BMEC tight junction (CLDN5, OCLN, CDH5, TJP1), transporter (SLC2A1, ABCB1, ABCG2) and Aβ-related transporter (LRP1, RAGE) [72, 73] genes by RT-qPCR and observed no significant differences in BMEC-like cells derived from different APOE genotypes (Fig. 1G). Collectively, these data indicate that isogenic iPSC-derived BMEC-like cells with different APOE genotypes have similar barrier properties and indistinguishable expression levels of key BMEC genes.
iPSC-derived BMEC-like cells expressing different APOE isoforms have similar barrier and transporter properties. () Isogenic iPSCs carrying differentgenotypes were differentiated to BMEC-like cells via the UM differentiation method. () Immunocytochemistry analysis of claudin-5 (CLDN5) expression and nuclear staining by DAPI in Day 10 isogenic iPSC-derived BMEC-like cells. Scale bars: 100 μm. () Quantification of claudin-5 expression and localization in () by area fraction index of immunocytochemistry images, which analyzes the fraction of pixels that belong to a junction and the mean fluorescence intensity of the junctions.= 3 independent differentiations for each condition. Data are reported as mean ± standard deviation. ns: p-value > 0.05 in one-way ANOVA analysis. () Comparison of Day 10 trans-endothelial electrical resistance (TEER) among BMEC-like cells expressing different APOE alleles. Data are reported as mean ± standard deviation.= 8 independent differentiations for each cell line. ns: p-value > 0.05 in one-way ANOVA analysis. () Immunocytochemistry analysis of occludin (OCLN) expression and nuclear staining by DAPI in Day 10 isogenic iPSC-derived BMEC-like cells. Scale bars: 100 μm. () Quantification of occludin expression and localization in () by area fraction index of immunocytochemistry images, which analyzes the fraction of pixels that belong to a junction and the mean fluorescence intensity of the junctions.= 3 independent differentiations for each condition. Data are reported as mean ± standard deviation. ns: p-value > 0.05 in one-way ANOVA analysis. () RT-qPCR analysis of expression of BBB junction (,,,) and transporter (,,,,) genes in Day 10 isogenic BMEC-like cells (= 3 independent differentiations each). Data are normalized to the APOE3 condition for each transcript. Data are reported as mean ± standard deviation. ns: p-value > 0.05 in one-way ANOVA analysis A B C B D E F E G APOE n n n CLDN5 OCLN CDH5 TJP1 SLC2A1 ABCB1 ABCG2 LRP1 RAGE n
LRP1 and P-glycoprotein-mediated pathways are involved in Aβ transport by iPSC-derived BMEC-like cells
Previous research has shown that APOE can bind to Aβ to form APOE-Aβ complexes [23, 74]. Both Aβ monomer and APOE-Aβ complexes can be cleared at the BBB [23, 38, 75–78] (Fig. 2A). BMEC-like cells are polarized, with LRP1 localized on the basolateral (brain) side and p-glycoprotein localized on the apical (blood) side [79–81]. Both LRP1 and p-glycoprotein have been implicated in brain-to-blood Aβ trafficking. Specifically, the APOE-Aβ complex binds to LRP1 and is then internalized into the cells. The complex can then be effluxed from BMEC-like cells by p-glycoprotein (Fig. 2A). Here, we assessed whether Aβ trafficking across the BMEC-like cell model had similar dependences on LRP1 and p-glycoprotein. We used immunocytochemistry to validate expression of LRP1 in APOE3 iPSC-derived BMEC-like cells (Fig. 2B). To assess the effect of LRP1 in Aβ clearance, we developed an in vitro Aβ transcytosis assay to quantify the effect of APOE isoforms on brain-to-blood Aβ trafficking (Fig. 2C). APOE3 iPSC-derived BMEC-like cells were seeded on a Transwell and allowed to form a confluent monolayer. Aβ was dosed into the basolateral chamber in the presence or absence of recombinant human APOE3 (rhAPOE3) and the LRP1 inhibitor RAP. RAP competes with APOE for LRP1 binding with an efficacy similar to anti-LRP1 antibodies [82, 83]. Using ELISA to quantify apical Aβ concentrations, we found that rhAPOE3 facilitated Aβ40 transcytosis and that inhibiting LRP1 with RAP reduced Aβ transcytosis (Fig. 2D). In addition, inhibiting LRP1 with RAP reduced but did not completely inhibit Aβ transcytosis, suggesting that alternative pathways are also likely involved in Aβ40 transport in BMEC-like cells. These data demonstrate that LRP1 is involved in basolateral-to-apical transport of Aβ in iPSC-derived BMEC like cells. We then used a p-glycoprotein activity assay to validate the function of p-glycoprotein, where Rh123 accumulation was increased in the presence of p-glycoprotein inhibitor cyclosporin A (CsA) as previously described [84] (Fig. 2E). To determine if p-glycoprotein plays a role in mediating Ab efflux in iPSC-derived BMEC-like cells, we incubated BMEC-like cells in medium containing fluorescently labeled Aβ40-Hylite488, with p-glycoprotein inhibitor CsA or DMSO negative control. Confocal microscopy was used to quantify the intracellular accumulation of Aβ40-Hylite488. CsA-treated BMEC-like cells had significantly higher intracellular accumulation of Aβ40-Hylite488 than DMSO control (Fig. 2F and G, Ext. Fig. 2). Cell lysates of BMEC-like cells treated with media containing Aβ40-Hylite488, with CsA or DMSO, were also collected for quantitative analysis by a fluorescence plate reader. Lysates of cells treated with CsA also contained significantly higher levels of Aβ40-Hylite488 than DMSO control, corroborating the microscopic evidence (Fig. 2H). Collectively, these data demonstrate that p-glycoprotein is capable of Aβ efflux in the iPSC-derived BMEC-like cell model.
LRP1 and P-glycoprotein-mediated pathways are involved in Aβ transport by iPSC-derived BMEC-like cells. () Schematic depicting the involvement of LRP1 and p-glycoprotein in Aβ trafficking in BMECs. () Immunocytochemistry analysis of LRP1 expression, along with isotype control antibody and nuclear staining by Hoechst 33342, in Day 10iPSC-derived BMEC-like cells Scale bars: 100 μm. () Schematic showing Transwell experimental design for analyzing brain-to-blood Aβ trafficking in iPSC-derived BMEC-like cells. () Basolateral-to-apical transport of Aβ40 with and without RAP treatment. Aβ40 was dosed into the basolateral chamber of the Transwell, and apical concentration of Aβ40 was determined by ELISA after 3 h. (= 8 independent wells of BMEC-like cells for each condition) *:< 0.05, ****:< 0.0001 by one-way ANOVA followed by Tukey’s test. Data are reported as mean ± standard deviation. () Intracellular Rhodamine 123 (Rh123) accumulation with and without CsA treatment across three differentiations ofiPSC-derived BMEC-like cells. BMEC-like cells were incubated with Rh123 pretreated with DMSO or CsA. Fluorescence of Rh123 from cell lysates was quantified by plate reader. (= 3 independent biological replicates) Data are normalized to the control DMSO condition for each pair. **:< 0.01 by paired Student’s t-test. () Representative confocal microscopy images showing accumulation of intracellular Aβ40-Hylite488 insideiPSC-derived BMEC-like cells with and without P-gp inhibition by CsA. Confocal microscopy images were taken at the z-plane where DAPI nucleus staining was visible. Intracellular Aβ40-Hylite488 fluorescence was then quantified in this same z-plane image. GLUT1 can be found both at the cell surface and junctions in BMEC-like cells and its immunolabeling was included to help visualization of the cell junctions at the same z-plane. Scale bars: 10 μm. () Quantification of the mean fluorescence intensity of Aβ40-Hylite488 accumulated insideiPSC-derived BMEC-like cells by confocal microscopy. (= 3 independent differentiations for each condition). *:< 0.05 by Student’s t-test. Data are reported as mean ± standard deviation. () Relative fluorescence units of Aβ40-Hylite488 accumulated insideiPSC-derived BMEC-like cells treated with DMSO or CsA. BMEC-like cells pretreated with DMSO or CsA were incubated with Aβ40-Hylite488. BMEC-like cells were washed and lysed. Fluorescence signals from lysates were quantified by plate reader. (= 3 independent differentiations for each condition) *:< 0.05 by Student’s t-test. Data are reported as mean ± standard deviation A B C D E F G H APOE3 n p p APOE3 n p APOE3 APOE3 n p APOE3 n p
APOE isoforms affect amyloid clearance in iPSC-derived BMEC-like cells
Using the in vitro Aβ transcytosis assay described above, the roles of different APOE genotypes on Aβ transport were quantitatively compared. Since Aβ monomer has been reported to be a ligand for LRP1-mediated transcytosis [23, 85–87], the two most common Aβ monomers, Aβ40 and Aβ42, were dosed into the basolateral chamber. After incubation, media in the apical chamber was collected. We then quantified the concentration of Aβ in the collected media using ELISA (Fig. 3A). While BMEC-like cells having different APOE genotypes transported Aβ40 at a similar rate, APOE2 BMEC-like cells cleared more Aβ42 than isogenic BMEC-like cells from other APOE genotypes (Fig. 3B). Comparing Aβ40 and Aβ42 trafficking, we also noticed approximately two orders of magnitude greater Aβ40 clearance than Aβ42 (Fig. 3B and D, Ext. Fig. 3A). This may be a result of Aβ42 being more prone to aggregation and deposition compared to Aβ40 [88, 89], thus contributing to its reduced apparent clearance.
We next assessed the effect of APOE protein isoform on Aβ trafficking. To eliminate any background from endogenously produced APOE protein, APOE KO BMEC-like cells were seeded on the Transwell and allowed to form a confluent monolayer. Since the BMEC-like cells expressed very little (< 1.5 ng/mL, < 44 pM) APOE protein, we added 500 nM recombinant human APOE (rhAPOE), a physiological concentration [90, 91], to the basolateral chamber in media also containing Aβ40 or Aβ42, and ELISA was used to quantify Aβ concentrations in the apical chamber after incubation (Fig. 3C). We found that rhAPOE3 significantly increased the trafficking of Aβ40, and that supplementation of rhAPOE2 significantly increased the trafficking of Aβ42 (Fig. 3D). These data also suggest that rhAPOE4 does not provide any increase in Aβ40 or Aβ42 clearance compared to the no APOE control condition (Fig. 3D). Together, these data suggest that APOE2 mediated enhanced clearance of both free Aβ42 (Fig. 3B) and APOE-Aβ42 (Fig. 3D) complexes, while APOE3 enhanced clearance of APOE-Aβ40 complexes. Performing the transcytosis assay on APOE KO BMEC-like cells at 4 °C and 37 °C, we observed minimal transcytosis of Aβ40 at 4 °C, indicating that Aβ40 trafficking in this assay is largely transcellular (Ext. Figure 3B). These results are consistent with human clinical data where APOE4 carriers exhibit significantly higher amyloid plaque deposition and APOE2 carriers exhibit significantly lower amyloid deposition than AD patients with APOE3/APOE3 genotypes [19, 92].
APOE isoforms alter the level of amyloid clearance in iPSC-derived BMEC-like cells. () Schematic showing experimental design analyzing the effect of differentgenotypes on BMEC-like cells trafficking of Aβ. Isogenic BMEC-like cells with differentgenotypes were seeded on a Transwell. 500nM Aβ40 or Aβ42 was dosed in the basolateral chamber. After a 3-hour incubation, Aβ concentrations in the apical chamber were quantified by ELISA. () Differential transport of Aβ40 or Aβ42 by isogenic BMEC-like cells to the apical chamber after 3 h of incubation quantified by ELISA. (= 4 independent differentiations for each condition. Mean of three biological replicates in each differentiation is plotted.) *:< 0.05, **:< 0.01 in one-way ANOVA followed by Tukey’s test. Data are reported as mean ± standard deviation. () Schematic showing experimental design analyzing the effect of different APOE protein isoforms on BMEC-like cells trafficking of Aβ.BMEC-like cells were seeded on a Transwell. 500nM Aβ40 or Aβ42 along with different isoforms of 500nM recombinant human APOE (rhAPOE) was dosed in the basolateral chamber. After a 3-hour incubation, Aβ40 or Aβ42 concentrations in the apical chamber were quantified by ELISA. () Differential transport of Aβ40 or Aβ42 byBMEC-like cells with different recombinant APOE protein isoforms was measured in the apical chamber after 3 h of incubation quantified by ELISA. (= 12 independent differentiations for each condition. Mean of three biological replicates in each differentiation is plotted.) ****:< 0.0001 in one-way ANOVA followed by Tukey’s test. Data are reported as mean ± standard deviation A B C D APOE APOE n p p APOE KO APOE KO n p
iPSC-derived brain pericyte-like cells of differentgenotypes have similar pericyte marker expression and APOE protein secretion profiles APOE
To determine how APOE genotype affects pericyte interactions with Aβ, established protocols were used to differentiate the isogenic, homozygous APOE iPSC lines into pericyte-like cells [67, 70] (Fig. 4A). After differentiation, pericyte-like cells derived from iPSCs with different APOE genotypes all expressed similar levels of pericyte markers NG2 and PDGFRβ as assessed flow cytometry (Fig. 4B and C). Since brain pericytes are one of the CNS cell types that secrete APOE [83], we quantified the expression level of APOE by RT-qPCR, and found that pericytes with different APOE genotypes expressed similar levels of APOE (Fig. 4D). We also quantified the concentrations of APOE protein secreted by pericyte-like cells by ELISA, and found no significant differences in total APOE protein among the APOE2, APOE3, and APOE4 pericytes (Fig. 4E). We also found no significant differences in LRP1 transcript expression between pericyte-like cells having different APOE alleles (Fig. 4D). These data suggest that the different APOE genotypes do not significantly affect canonical pericyte marker expression or APOE expression and APOE secretion levels.
iPSC-derived brain pericyte-like cells with differentgenotypes possess similar properties. () Isogenic iPSCs carrying differentgenotypes were differentiated to brain pericyte-like cells via a neural crest cell progenitor as previously described []. () Flow cytometry analysis of NG2 and PDGFRβ expression in Day 24 isogenic pericyte-like cells. FSC: forward scatter. Gating is according to isotype control performed on a pooled population of Day 24 isogenic pericyte-like cells with differentgenotypes. () Histograms of flow cytometry analysis in (). Samples shown in histograms are normalized to mode. () RT-qPCR analysis ofandexpression in Day 24 isogenic pericyte-like cells (= 4 wells of independent differentiations for each condition). Bars indicate mean values. Data are normalized to thecondition. ns: p-value > 0.05 in one-way ANOVA analysis. Data are reported as mean ± standard deviation. () Concentration of APOE protein in cell culture media conditioned by isogenic pericyte-like cells for 24 h. APOE concentration was quantified by ELISA. (= 3 wells of independent differentiations for each condition). ****: p-value < 0.0001 in one-way ANOVA analysis followed by Tukey’s test. Data are reported as mean ± standard deviation APOE APOE APOE APOE LRP1 n APOE3 n A B C B D E [67]
APOE isoforms affect amyloid deposition and internalization in iPSC-derived pericyte-like cells
Patients with AD commonly have amyloid deposits associated with their brain vasculature, contributing to a pathological condition known as cerebral amyloid angiopathy (CAA) [93]. Recently, a role of pericytes in CAA has been reported [94]. We quantified deposition of Aβ42, the Aβ isoform more prone to aggregation, for pericyte-like cells expressing different APOE isoforms. We incubated pericyte-like cells differentiated from different APOE iPSCs with Aβ42 for 18 h. The pericyte-like cells were then singularized, washed and fixed, but not permeabilized, for cell surface immunolabeling with an anti-Aβ antibody. Flow cytometry was subsequently used to quantify extracellular deposits of Aβ42 (Fig. 5A). The APOE KO pericyte-like cells exhibited the least extracellular deposition of Aβ42 (Fig. 5B and C), indicating that APOE expression, regardless of isoform, exacerbates Aβ42 aggregation and deposition. A comparison of isogenic APOE2, APOE3 and APOE4 pericyte-like cells revealed that APOE4 pericyte-like cells had the highest level of extracellular Aβ42 deposition while APOE2 pericyte-like cells had the least amount of Aβ42 deposition (Fig. 5B and C). This is consistent with clinical data that in patients with CAA, APOE4 carriers consistently demonstrate elevated perivascular amyloid in the brain [20, 59]. Since it has been reported that pericytes utilize LRP1 to internalize APOE-Aβ complexes [22, 76], we probed whether the iPSC-derived pericyte-like cells also utilize LRP1 to internalize Aβ42. Using confocal microscopy, we found that iPSC-derived APOE3 pericyte-like cells internalized fluorescently labeled Aβ42-HyLite488 (Fig. 5D). However, when pericyte-like cells were treated with the LRP1 inhibitor RAP, internalization was diminished (Fig. 5D and E). In the same way, we then examined the impact of APOE genotype on pericyte-like cell internalization of Aβ42. APOE4 pericyte-like cells internalized less Aβ42 than APOE3 pericyte-like cells, while APOE2 pericyte-like cells internalized the most Aβ42 (Fig. 5F and G). Overall, our data suggests that APOE4 pericyte-like cells have higher levels of extracellular Aβ42 deposition but lower levels of internalization while APOE2 pericyte-like cells have less extracellular deposition and more internalization of Aβ42.
Isogenic iPSC-derived pericyte-like cells demonstrate differential Aβ42 deposition and internalization. () Schematic showing the experimental design analyzing the effect of different APOE isoforms on extracellular Aβ42 deposition. 500nM Aβ42 was incubated with pericyte-like cells for 18 h to allow deposition of extracellular Aβ aggregates. Pericyte-like cells were then singularized, washed and labeled with anti-Aβ-FITC antibody for flow cytometry quantification of cell surface-associated Aβ42. () Flow cytometry analysis of extracellular deposition of Aβ42 by isogenic pericyte-like cells after 18 h of incubation. FSC: forward scatter. Histogram is normalized to mode. () Quantification of level of Aβ42 deposition on pericytes by flow cytometry (= 3 wells of biological replicates per condition). *:< 0.05 in one-way ANOVA analysis followed by Tukey’s test. Data are reported as mean ± standard deviation. () Representative confocal microscopy images showing accumulation of intracellular Aβ42-Hylite488 with or without LRP1 inhibition by RAP inpericyte-like cells. Confocal microscopy images were taken at the z-plane where DAPI nucleus staining was visible to distinguish intracellular Aβ42-Hylite488. Images were then captured for the Aβ42-Hylite488 channel on the same z-plane. The last column contains the merged images. Scale bars: 10 μm. () Quantification of mean fluorescence intensity (MFI) of Aβ42-Hylite488 accumulated insidepericyte-like cells by confocal microscopy. (= 4 wells of biological replicates per condition) **:< 0.01 in Student’s t-test. Data are reported as mean ± standard deviation. () Maximum intensity projection of confocal microscopy images showing accumulation of intracellular Aβ42-Hylite488 by isogenic pericyte-like cells. Confocal microscopy images were taken at 5 μm z-slice intervals. All z-slices with Hoechst 33342 nuclear staining were aggregated for maximum intensity projection analysis. Scale bars: 10 μm. () Quantification of mean fluorescence intensity (MFI) in maximum intensity projection confocal microscopy images of Aβ42-Hylite488 accumulated inside iPSC-derived pericytes by confocal microscopy. (= 5 slides of pericyte-like cells treated independently by Aβ42-Hylite488 per genotype). **:< 0.01, ****:< 0.0001 in one-way ANOVA followed by Tukey’s test A B C D E F G n p APOE3 APOE3 n p n p p
Discussion
AD is a progressive neurodegenerative disorder that leads to memory loss, cognitive decline, and eventual loss of ability to perform everyday activities [95]. Recently, many genetic risk factors have been identified for AD, including mutations in APP, PSEN1, PSEN2 for familial AD [96], and mutations in APOE, CLU, PICALM for sporadic AD [97, 98]. Among these genetic risk factors, APOE has proven to be the strongest risk factor [7] with APOE4 being detrimental and APOE2 being protective. While neurons, astrocytes and microglia have been investigated as cellular targets for the differential effects of the APOE allele [99–102], the vascular contribution has more recently been a target of study given its direct role in Aβ trafficking and clearance. Clinical studies have linked APOE4 with elevated leakage at the BBB, reduced pericyte coverage, and increased vascular amyloid deposition [21, 41, 54, 94, 103] compared to APOE3 patient controls. These data have sparked interest in clarifying the role of different APOE isoforms at the BBB, given that BBB is thought to be the main site of clearance of Aβ [37]. One lingering question that is unanswered by clinical studies is whether the enhanced BBB dysfunction in APOE4 carriers is a result of intrinsic deficiencies in BMECs or pericytes, or a result of chronic exposure to APOE4-producing astrocytes, enhanced neuroinflammation, or other pathologies associated with APOE4 carriers [2, 4].
Here, by employing in vitro human iPSC-derived BBB models, we isolated two individual BBB cell types from chronic influences to better understand the roles of different APOE isoforms in BMEC-like cells and pericyte-like cells. By using isogenic iPSC lines with different APOE genotypes, we also minimized the effect of background genetic variation to further target the specific effects elicited by APOE isoforms. Overall, our work illustrated the multifaceted role that different APOE isoforms could play at the BBB. While different APOE isoforms do not directly affect standard properties of BMEC-like cells and pericyte-like cells, we observed differential Aβ clearance and deposition profiles associated with APOE isoforms. Our results collectively predict that APOE4 would lead to more Aβ load in the CNS, while APOE2 would reduce the Aβ load compared to the most common APOE3 isoform. Furthermore, our results indicate that in BMEC-like cells, APOE isoforms do not influence intrinsic BMEC differentiation, and lead to similar BBB gene expression of efflux transporters, glucose transporter, and tight junction components in addition to indistinguishable paracellular barriers. Similarly, in pericyte-like cells, APOE isoforms do not affect pericyte differentiation, and do not influence the level of APOE protein secretion. These results highlight that simply possessing different APOE alleles does not directly affect BMEC-like cell and pericyte-like cell differentiation and function. These results are consistent with the premise that elevated BBB damage observed clinically in APOE4 carriers, is at least in part the result of chronic pathological exposure from other APOE4 CNS cell types such as neurons, astrocytes or microglia. The observed similarities in BMEC-like cell properties amongst isogenic lines in this study are in contrast with the differences observed in BMEC-like cells differentiated from iPSC lines derived from different AD patients, where differences in genetic background could potentially confound results [104].
We next validated the usage of the BMEC-like cells as an in vitro BBB model for Aβ trafficking studies by showing that the LRP1 and p-glycoprotein pathways are expressed and functional in Aβ trafficking. We then evaluated isogenic BMEC-like cells for differential Aβ trafficking effects associated with APOE allele expression. We showed that APOE2 BMEC-like cells exhibited greater transcytosis of monomeric Aβ42 than APOE3 and APOE4 BMEC-like cells. For the transcytosis of APOE-Aβ complexes, APOE4 protein resulted in the lowest combined clearance of Aβ40 and Aβ42, while APOE3 protein increased the transport rate of Aβ40 and APOE2 protein increased the transport rate of Aβ42. These collective data are consistent with the putative biochemical mechanisms that drive altered Aβ trafficking by different APOE isoforms. For instance, monomeric Aβ, and APOE-Aβ complexes can be internalized through cell surface receptor LRP1, which has been suggested as the main brain-side Aβ clearance pathway utilized by brain endothelial cells [22, 37, 72, 75, 83]. Since monomeric Aβ has been reported to oligomerize into insoluble fibrils in the presence of APOE protein in a APOE4 > APOE3 > APOE2 fashion [7, 23], and oligomerized Aβ are less likely to be cleared by LRP1- based receptor-mediated transcytosis [76, 85, 105], differential Aβ oligomerization levels caused by different APOE isoforms could in part explain our observations. Our data also agree with clinical observations where Aβ brain depositions in AD patients are ranked in a APOE4 > APOE3 > APOE2 fashion, suggesting differential clearance of Aβ with different APOE isoforms [23, 106–108]. One caveat of the iPSC-derived BMEC-like cells used in this manuscript is the existence of a mixed endothelial-epithelial phenotype [109, 110] within the homogenous cell population. While other protocols generate iPSC-derived BBB-like cells without the epithelial character, the resultant cells lack tight enough paracellular barriers [109, 111, 112] to discern differences in transcellular Aβ passage. By contrast, the iPSC-derived BMEC-like cells used here possess requisite tightness along with functional LRP1 and p-glycoprotein mediated Aβ transport pathways, indicating their relevance for Aβ transport modeling.
Vascular Aβ deposition has been shown to be related to pericytes in AD and cerebral amyloid angiopathy (CAA), and vascular Aβ deposition is dependent on APOE isoform, with an APOE4 > APOE3 > APOE2 trend [94, 113, 114]. Thus, we also assessed the role of APOE alleles in pericyte-like cells in contributing to Aβ deposition and uptake. For extracellular Aβ42 deposition on pericyte-like cells, a pattern of APOE4 > APOE3 > APOE2 was observed. For Aβ42 internalization, we found that APOE4 pericyte-like cells exhibited the least Aβ42 uptake compared to APOE2 and APOE3 pericyte-like cells. Since Aβ can be degraded by intracellular lysosome-dependent pathways [26–29], a reduced internalization could also contribute to increased amyloid load in the brain. These combined data suggest a potential detrimental role for APOE4 in increasing extracellular Aβ42 deposition while also decreasing Aβ42 internalization for potential degradation. In contrast, APOE2 improves the Aβ42 deposition profile. Our results also confirm a previously reported pathway that pericytes can internalize Aβ via LRP1 receptors [83], while providing additional insights into the effect of pericytes APOE isoforms on Aβ42 internalization. Differential Aβ42 association with pericyte-like cells was observed despite indistinguishable APOE secretion, which could be a result of the differential biochemical association of Aβ42 with the different APOE isoforms [23]. Previous reports have suggested that isogenic APOE4 iPSC-derived pericyte-like cells secrete more APOE compared to APOE3 controls94. This difference with our work could potentially be attributed to the use of different pericyte differentiation protocols. In this work, pericyte-like cells were differentiated through neural crest stem cells [67], analogous to the developmental profile of forebrain pericytes [115–118], while pericyte-like cells in the contrasting work were derived from a mesodermal lineage [94, 119].
Finally, this work highlights the potential of using isogenic iPSC lines for modeling AD effects on the BBB. We demonstrated that isogenic iPSC-derived in vitro BBB models successfully captured aspects of AD-related amyloid pathologies by illustrating the effects of different APOE isoforms on Aβ trafficking and deposition. These in vitro BBB models could prove helpful in studying AD, and in the future, they could be used to isolate cell type-specific effects and help identify mechanisms by which the disease alters the BBB. For instance, previous studies using different stem cell-based in vitro BBB models have demonstrated applications in the modeling of BBB permeability after exposure to AD neural stem cell-derived neurons [120], and amyloid deposition modeling using vascular assembloids with pluripotent stem cell-derived AD mural cells [94]. Specifically, through the use of gene-edited isogenic iPSC lines carrying pathogenic mutations, researchers can explore potential disease influencing impacts with minimal influence from the genetic background [121]. Such efforts have been applied to study the effect of pathogenic mutations in Down syndrome [122], Parkinson’s Disease [123] and AD [94, 124, 125]. We envision that similar strategies could illustrate the role of other pathogenic mutations at the BBB in neurological diseases such as AD.
Conclusions
iPSC-derived BBB models can capture aspects of amyloid pathologies at the BBB. Different APOE genotypes did not directly affect general BMEC-like cell or pericyte-like cell properties in the iPSC-derived cells. However, recombinant APOE4 (BMEC-like cells) and APOE4 genotype (pericyte-like cells) exacerbated amyloid clearance and deposition at the BBB. In addition, APOE2 demonstrated a protective role against amyloid pathologies using the models. These results are consistent with clinical observations on the effect of different APOE genotypes on CNS amyloid load.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
We thank Dr. Benjamin D. Gastfriend and Professor Richard Daneman for their valuable suggestions for this work. We thank Dr. Chandler B. Est and Professor Regina Murphy for providing protocols for handling recombinant Aβ. Illustrations in Fig. 1A, 2A, 2C, 3A, 3C, 4A, 5A were created using Biorender.com and licensed under agreement number PS26SC45KZ.
Abbreviations
Author contributions
YD, SPP, EVS conceived and designed the experiments. YD performed the experiments and analyzed the data. YD, SPP, and EVS wrote and corrected the manuscript. SPP and EVS supervised the study. All authors read and approved the final manuscript.
Funding
This work was supported in party by the National Institutes of Health Grants HL154254, NS107461 and NS109486.
Data availability
All data supporting the findings of this study are available within the paper and its Supplementary Information.
Declarations
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The authors have multiple issued patents and patent applications in the field of BBB modeling.
Footnotes
Contributor Information
Sean P. Palecek, Email: sppalecek@wisc.edu
Eric V. Shusta, Email: eshusta@wisc.edu
References
Associated Data
Supplementary Materials
Data Availability Statement
All data supporting the findings of this study are available within the paper and its Supplementary Information.