Immuno-regulation of Brain Region-specific Organoids Containing Isogenic Microglia-like Cells

Abstract

Most brain organoids derived from human induced pluripotent stem cells (iPSCs) lack microglia and thus immune function. Microglia-like cells (MGCs) can be differentiated from iPSCs, while the characteristics of isogenic MGC-containing brain organoids in modeling neurodegeneration and cell-cell communications have not been well investigated. In this study, iPSC-derived MGCs were co-cultured with isogenic forebrain cortical organoids (iFCo), which were stimulated with extracellular vesicles (EVs) of brain organoids differentiated from Alzheimer’s disease (AD) patient-derived iPSCs (APOE ε4/ε4 and presenilin 1). The AD EV-stimulated co-culture organoids were treated with EVs from healthy MGCs or co-culture. MGCs expressed IBA1 and P2RY12 and responded to Aβ42 and dexamethasone by regulating carbon metabolism and inflammation. Differential responses of the co-cultured organoids and the MGCs to AD EVs were demonstrated. The co-cultured organoids mitigated pro-inflammatory genes IL-6, IL-12β, iNOS, and TNFα. EVs from healthy MGCs or co-culture reduced the expression of IL-12β, iNOS, TREM2, and CASS4, which are associated with neural inflammation and degeneration, as well as showed regulation on genes involved in microglial activation and carbon metabolism. AD EV cargo analysis by proteomics and microRNA-sequencing revealed APOE and APP proteins and microRNAs regulated pathways such as mitophagy. This study paves the way for understanding the role of microglia and brain organoids in modeling neural degeneration and the development of EV-based cell-free therapeutics for AD treatment.

1. Introduction

In 2013, Lancaster et al. developed the first three-dimensional (3D) human brain organoid model differentiated from induced pluripotent stem cells (iPSCs) [1]. However, this method did not include growth factors and chemicals to induce lineage-specific differentiation; therefore, those organoids have low reproducibility and high variability [2]. Other groups reported protocols that required fine-tuning small molecules and growth factors [3], which permitted the induction of specific regions of the human brain, such as the forebrain, hindbrain, choroid plexus, etc. [4–6]. Current brain organoids still face several challenges, such as a lack of microglia and vascularization, and the immature status of differentiated neuronal cells [7, 8]. Among these challenges, microglia have a significant impact on the formation of the early cortex and the progression of neurological disorders [9].

Microglia are the resident innate immune cells of the central nervous system (CNS), including the brain and spinal cord. Besides their role as immune cells, microglia also participate in regulating the CNS, such as neurogenesis, synapse maturation and elimination, and neural network formation [10]. Moreover, microglia were found to be involved in Alzheimer’s disease (AD) pathology [11, 12]. Microglia cells in the brain express many risk genes of AD, such as the TREM2 gene that encodes a membrane protein selectively expressed in myeloid cells, including microglia. Growing evidence shows that microglia provide protection (resting phenotype) against the occurrence of AD, because dysfunctional microglia and decreased microglial responses to β-amyloid are associated with increased AD risk [13, 14]. However, there is also a wealth of evidence suggesting that activated microglia (pro-inflammator or disease-associated phenotype) may be detrimental to neurons [15]. Unlike other CNS cells that originate from neuroectoderm, microglia are generated from the mesoderm lineage [16]. Since the common differentiation protocols for brain organoids use dual SMAD inhibition to induce neuroectoderm formation, microglia usually do not innately populate brain organoids [17]. Impaired interactions between microglia and CNS cells, such as neurons and astrocytes, are increasingly linked to neurodegenerative and neurodevelopmental disorders such as AD [18]. Consequently, it is necessary to include microglia in cerebral organoids to mimic the complicated combination of cells derived from different germ layers that are present in the brain in vivo. In the past few years, several studies derived human microglia-like cells from iPSCs [19–21]. These findings showed that iPSC-derived microglia-like cells (MGCs) are comparable to human microglia, indicating that it is feasible to differentiate microglia from iPSCs and integrate them with human brain organoids.

Among the cerebral models that include human microglia, some studies used microglia from primary brain tissue or an immortalized microglia cell line [22–24]. However, human tissue-sourced microglia are limited, while the phenotypes and transcriptional patterns of the immortalized microglia cell line may be altered after immortalization [25]. Following Muffat et al.’s first protocol to generate MGCs from human iPSCs [19], brain organoids co-cultured with iPSC-derived MGCs emerged as attractive model systems for drug discovery. The procedure also allows brain organoids and MGCs to be isogenic, which better mimics a human brain. Studies showed that the co-culture of brain organoids with MGCs promoted MGCs to be mature, ramified, and respond to injury [20]. Enhanced cell proliferation and reduced reactive oxygen species were observed in co-cultures compared to MGCs only [7]. Additionally, MGCs increased the synchronization and frequency of oscillatory bursts in the brain organoids that benefited maturation of neural networks [24], highlighting the critical role that microglia play in brain development.

Extracellular vesicles (EVs) are lipid membrane-bounded vesicles that can be secreted by basically all kinds of cells, including stem cells [26–28]. Their rich cargo, such as nucleic acids, proteins, lipids, etc., make them a promising class of therapeutic candidates [29–31]. For example, brain spheroid- and organoid-derived EVs facilitated in vitro stroke recovery after oxygen and glucose deprivation by reducing apoptosis, enhancing reactive oxygen species scavenging, and elevating anti-inflammatory ability, as shown in our previous studies [32, 33]. Microglia-derived EVs carrying miR-711 were shown to alleviate neurodegeneration in a murine AD model. EVs containing miR-711 inhibited 1,4,5-trisphosphate 3-kinase B (Itpkb) and further repressed Tau phosphorylation and increased the ratio of anti-/pro-inflammatory microglia, leading to reduced scores of neurological deficits and improved cognitive function in mild traumatic brain injury mice [34]. On the other hand, EVs from cells with disease phenotypes may aggravate pathology, thus showing their double-sided effects [35]. For instance, Aβ and tau propagation were observed in EVs from different studies, which suggested new strategies for AD treatment, such as microglia depletion to stop aberrant protein spread and EV surface coating to inhibit extracellular EV motion [36, 37].

The goal of this study was to construct an MGC-brain organoid neural degeneration model by exposure to EVs from AD-associated iPSCs and evaluate the therapeutic effects of co-culture EVs and MGC EVs on inflammatory genes, AD risk genes, and carbon metabolism. This study hypothesized that the MGCs and MGC-containing brain organoids can be induced to a neurodegenerative phenotype by AD EVs, and co-culture EVs and MGC EVs can ameliorate neural degeneration. In this study, MGCs were generated from human iPSCs and characterized by immunocytochemistry, flow cytometry, and changes in gene expression in response to stimulation. Then, MGCs were co-cultured with isogenic forebrain cortical organoids (iFCo). The co-cultures were characterized by immunocytochemistry and gene expressions related to immune function. An in vitro neural degeneration model based on co-cultures was induced by EVs from brain organoids derived from AD patient-derived iPSCs with sporadic APOE ε4/ε4 (APOE4) genotype and a familial presenilin 1 (PS1) genotype, which were characterized by proteomics and microRNA (miRNA)-sequencing for the EV cargo. Further, an IL-4 stimulation was applied to MGCs to induce the M2 phenotype. Media from M0 and M2 MGC, as well as co-cultures, were collected for EV isolation, and these EVs were evaluated in the neurodegeneration model. This study paves the way for understanding the role of microglia and brain organoids in modeling neural degeneration and the development of EV-based cell-free therapeutic interventions.

2. Materials and Methods

2.1. Human iPSC culture

Human foreskin fibroblasts were transfected with plasmid DNA encoding reprogramming factors octamer-binding transcription factor 4 (OCT4), NANOG, SRY-box transcription factor 2 (SOX2), and LIN28 to produce human iPSK3 cells [38, 39]. The human iPSCs were maintained in mTeSR Plus serum free medium (StemCell Technologies, Inc., Vancouver, Canada) on growth factor-reduced Matrigel-coated surface (Corning, Inc., Corning, NY, USA). The cells were passaged every five to seven days using Accutase and seeded at 1×10⁶ cells per well of a six-well plate in the presence of rho-associated protein kinase (ROCK) inhibitor Y27632 (10 μM, Sigma-Aldrich, St. Louis, MO, USA) for the first 24 hours.

Another human iPSC line, Pluristyx, was thawed from a vial of the Matched Research Grade iPSC Working Cell Bank (WCB) generated under current good manufacturing practice (cGMP) from Lonza (Walkersville, MD, USA). Pluristyx human iPSCs were reprogrammed from human CD34+ umbilical cord blood cells using a non-integrating episomal vector reprogramming method (Certificate of Analysis and Lutheran Hospital Institutional Review Board Approval available upon request). Pluristyx human iPSCs cryogenically preserved in CryoStor CS10 (StemCell Technologies Inc.) were thawed and plated into growth factor-reduced Matrigel (Corning, Inc.)-coated CellBIND T-25 cm² and T-75 cm² Rectangular Canted Neck Cell Culture Flasks with Vent Cap (Corning Inc.). Cells were seeded at a density of 15,000 cells/cm² and grown for 3 days as P+1 and 5,000 cells/cm² for 4 days as P+2 in mTeSR Plus media (StemCell Technologies Inc.). ROCK inhibitor Y27632 (10 μM) was added for the first 24 hours of each passage.

2.2. Human iPSC differentiation into induced forebrain cortical organoids (iFCo)

Undifferentiated human iPSC cells were seeded into Ultra-Low Attachment (ULA) 96-well plates (Corning Inc.) at 1.5×10⁴ cells/well in a differentiation medium composed of DMEM/F-12 plus 2% B27 serum-free supplement (Life Technologies, Carlsbad, CA), in the presence of Y27632 (10 μM). After 24 hours, Y27632 was removed, and the formed embryoid bodies (EB) were treated with dual SMAD signaling inhibitors of 10 μM SB431542 (Sigma) and 100 nM LDN193189 (Sigma) over 7 days. Then on day 8, the spheroids were treated with fibroblast growth factor (FGF)-2 (10 ng/mL, PeproTech, Inc., Cranbury, NJ, USA) for cortical differentiation for over 30 days [40].

This study also differentiated human iPSCs from a sporadic AD patient (MC0020, female, 71.3 years old, APOE ε4/ε4) and a familial presenilin 1 (PS1) AD patient into forebrain cortical organoids using the STEMdiff™ Cerebral Organoid Kit from StemCell Technologies, as shown in our previous publication [41, 42]. After 40 days of differentiation, the organoids were cultured on an orbital shaker in neuron medium composed of DMEM/F12 and Neurobasal (1:1) , 2% B27 supplement with vitamin A, 1% N2 supplement, 1% non-essential amnio acid (NEAA), L-glutamine (2 mM), ascorbic acid (200 ng/mL), dibutyryl cyclic adenosine monophosphate (dbCAMP, 500 ng/mL), brain-derived neurotrophic factor (BDNF, 10 ng/mL), glial cell line-derived neurotrophic factor (GDNF, 10 ng/mL), amphotericin B (0.25 μg/mL), and 1% penicillin-Streptomycin. The neural medium was replaced every 3–4 days for another month. The collected conditioned media were collected for EV isolation.

Our previous studies have characterized iFCo and shown that healthy control (HC) organoids are mature around 30 days [43–45]. However, AD organoids are challenging in differentiation, so the STEMdiff™ Cerebral Organoid Kit was used to enhance the differentiation, a 40-day period is required by the kit for mature organoids.

2.3. Human iPSC differentiation into induced hindbrain cerebellar organoids (iHCo)

Cerebellar differentiation and characterization was performed as reported in our previous study [5, 46]. Human iPSCs were seeded at 1.5×10⁴ cells/well into ULA 96-well plates (Corning Inc.) in differentiation medium composed of DMEM/F-12 with 2% B-27 serum-free supplements (Life Technologies). Y27632 (10 μM) and SB431542 (10 μM) were added for the first week of culture. On day 2, FGF-2 (50 ng/mL) was added until the end of the first week. During days 7–14, both retinoic acid (RA) (1.0 μM, Sigma) and Wnt activator CHIR99021 (CHIR, 10 μM, Sigma) were added to the culture. During the third week of culture, caudalization of neural rosettes in the derived aggregates was induced using FGF-19 (100 ng/mL, Peprotech). Nothing was added in the fourth week. During days 28–35, the treatment of purmorphamine (2 μM, Sigma) and stromal cell-derived factor 1-α (SDF1A, 50 ng/mL, Peprotech) was used to activate the SHH pathway and promote ventralization, as well as to induce organoid self-organization into the molecular layer, the Purkinje cell layer, and the granule cell layer.

2.4. Human iPSC differentiation into microglia-like cells

MGCs were induced from human iPSCs in mTeSR serum-free medium on a 24-well plate for four days, as shown in our previous study [7]. The medium was changed to fresh mTeSR Plus on day −1. On day 0, cells were induced with 30 ng/mL vascular endothelial growth factor (VEGF, Peprotech), 30 ng/mL bone morphogenetic protein 4 (BMP4, Peprotech), 40 ng/mL stem cell factor (SCF, Peprotech), and 50 ng/mL Activin A (Invitrogen, Carlsbad, CA, USA) in RPMI medium (Life Technologies) plus 2% B27 serum-free supplement. Four days later, the medium was changed to RPMI plus 2% B27 supplemented with 50 ng/mL SCF, 50 ng/mL Flt3L (Peprotech), 10 ng/mL IL-3 (Peprotech), 50 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF, Peprotech), and 25 ng/mL BMP4. On day 7 and day 10, half of the spent media were replaced with fresh media, and the cells were grown for another 3–5 days. Cells were harvested on day 15. Cells from one well of a 24-well plate were split and plated into two wells of a new 24-well plate in DMEM, 10% defined fetal bovine serum (FBS), 1% penicillin/streptomycin (all from Life Technologies), and 20 ng/mL of GM-CSF. After 2–5 days, 20 ng/mL of IL-3 was added. The hematopoietic progenitors were further cultured for 1–2 weeks for differentiation to MGCs. The MGCs (M0) were induced to a pro-inflammatory phenotype (M1) by amyloid beta (Aβ) 42 oligomers (0.5 µM, Bachem) and an anti-inflammatory phenotype (M2) by dexamethasone (1 µM, Sigma) and IL-4 (20 ng/mL, Peprotech) [7]. MGCs in different activation statuses were then characterized. The conditioned media of the M0 and M2 groups were collected for EV isolation.

2.5. MGC co-culture with region-specific brain organoids

MGCs were harvested using Accutase. The cells were resuspended gently in pre-warmed Cell Tracker Red working solution (10 μM, Invitrogen) and were incubated for 30 min. Then the cells were washed with PBS and resuspended in the media. Brain organoids (forebrain or hindbrain) at days 25–35 were co-cultured with isogenic MGC at a 6:1 ratio (2×10⁵ – 4×10⁵ cells per organoid) in 50% DMEM plus 10% FBS and 50% DMEM/F12 plus 2% B27. The co-cultured organoids were observed using a fluorescent microscope (Olympus IX70, Melville, NY, USA) to demonstrate the integration of MGCs with the brain organoids. The co-culture forebrain organoids were characterized by immunocytochemistry and gene expression. The conditioned media were collected for EV isolation.

2.6. Stimulation of MGCs and co-cultured brain organoids for neural degeneration

MGCs and co-cultured brain organoids were treated with brain organoid EVs differentiated from AD patient (APOE4 and PS1)-derived human iPSCs at a dose of 500 particles/cell for 72 h to mimic an in vitro AD environment, along with PBS control. Changes at the gene level were assessed by reverse transcription quantitative polymerase chain reaction (RT-qPCR). The unstimulated MGCs were considered as M0 phenotype. An M2 phenotype was induced by 48-hour treatment of IL-4 (20 ng/mL, Peprotech). EVs were isolated from the spent media of M0 MGC and M2 MGC, as well as the unstimulated co-culture spent media. These EVs were evaluated in APOE4 AD EV-treated (72 h) unstimulated co-culture. The EV doses were 1000 and 5000 particles/cell and treatment duration was 72 h. Gene expression and cytokine secretion were assessed by RT-qPCR and ELISA, respectively, to examine the alleviation of neural degeneration.

2.7. Extracellular vesicles isolation

Differential ultracentrifugation combined with an inexpensive polyethylene glycol (PEG)-based method was used to isolate the EVs from the conditioned media [47, 48]. Briefly, the conditioned media were centrifuged at 500 g for 5 min (Eppendorf, Centrifuge 5810 R, Germany). The supernatants were centrifuged again at 2,000 g for 10 min. The collected supernatants were then centrifuged at 10,000 g for 30 min. Then, supernatants were mixed with PEG solution (24% wt/vol in 1.5 M NaCl) at a 2:1 volume and incubated at 4ºC overnight. The solutions were centrifuged at 3,214 g for 60 min. The crude EV pellets were resuspended in 1 mL of filtered phosphate buffered saline (PBS) and then ultracentrifuged (Beckman Coulter, Optima MAX-XP Ultracentrifuge, CA, USA) at 100,000 g for 70 min. Purified EV pellets were suspended in 100 µL PBS and shaken (Eppendorf, ThermoMixer C, Germany) for 30 min under 1500 rpm. The isolated EVs were used for various characterizations and cell culture experiments.

2.8. Decellularization and matrix-bound nanovesicle (MBV) isolation

The replated organoids were washed with PBS and decellularized using 0.5% Triton X-100 for 5 min [49–52] The samples were rinsed once with Hanks’ balanced salt solution and once with ultrapure water [53]. The cultures were then treated with extracellular matrix (ECM) harvesting solution (50 mM tris [pH 7.5–8], 5 mM CaCl₂, and 200 mM NaCl) containing enzymes (proteinase K at 0.01 mg/mL) at 37°C for 1 h to release MBVs from ECM [54]. The collected solution was used for MBV isolation. Briefly, samples were sequentially centrifuged at 500 g, 2,000 g, and 10,000 g, then were passed through a 0.22 μm filter. After that, the MBV isolation procedure was the same as the EV isolation procedure. In the comparison of MBVs from ECM and EVs from spent media, the latter were referred to as supernatant EVs (SuEVs).

2.9. Immunocytochemistry and flow cytometry

The cells were fixed using 4% paraformaldehyde (PFA) for 60 min and permeabilized using 0.2% Triton-X 100 for 10 min. The samples were blocked with 5% FBS in PBS for 30 min and stained with the primary antibodies (Supplementary Table S1) at 4ºC overnight, followed by incubation with Alexa Fluor 488 goat anti-mouse IgG1 or Alexa Fluor 488 goat anti-Rabbit IgG (Life Technologies) for 60 min at room temperature. Both primary and secondary antibodies were diluted in a staining buffer (2% FBS in PBS) based on the manufacturer’s recommendations. Then the nuclei were counterstained with Hoechst 33342 (blue), and images were taken under a fluorescent microscope (Olympus IX70). For marker quantification, the flow cytometry method was used. Briefly, 1×10⁶ cells were fixed with 4% PFA and washed with staining buffer (2% FBS in PBS). For intracellular markers, the cells were permeabilized with 100% cold methanol for 15 min, while no permeabilization was performed for surface markers. The samples were blocked with blocking buffer (5% FBS in PBS) and then incubated with different primary antibodies (Supplementary Table S1), followed by the corresponding secondary antibody. The cells were acquired using a BD FACSCanto II flow cytometer (Becton Dickinson) and analyzed against isotype controls using FlowJo v10 software.

2.10. Nanoparticle tracking analysis (NTA)

NTA was performed on the isolated EVs to determine particle size distribution and concentration. The NanoSight LM10-HS instrument (Malvern Instruments, Malvern, UK) was used. It is configured with a blue laser (488 nm) and sCMOS camera. The samples were diluted as 1:1000 in filtered PBS. Three videos of 60 s were captured with the camera shutter speed fixed at 30.00 ms. The camera level was set to 12, and the detection threshold was set to 5. Between each sample reading, the laser chamber was cleaned thoroughly with particle-free milliQ H₂O. The collected videos were analyzed using NTA3.4 software to obtain the mode and mean size distributions, as well as the concentration of particles. Compared to the mean size, the mode size is usually a more accurate representation because the vesicle aggregates may affect the value of the mean size.

2.11. Western blot

The samples were lysed in radio-immunoprecipitation assay (RIPA) buffer (150 mM sodium chloride, 1.0% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 50 mM Tris, pH 8) with 1% of proteinase inhibitor cocktails (2 µg/mL Aprotinin, 5 µg/mL Leupeptin, 5 µg/mL Antipain, 1 mM PMSF, Invitrogen). The samples were then digested for 20 min on ice, and spun down at 14,000 rpm for 20 min. The supernatant was collected, and the protein concentration was determined by a Bradford assay. Protein lysate concentrations were normalized, and 20 µg of each sample was denatured at 95°C in 2x Laemmli Sample Buffer. Proteins were loaded into 12% Bis-Tris-SDS gels and transferred onto a nitrocellulose membrane (Bio-Rad). The membranes were then blocked against non-specific binding for 1 h in 5% skim milk (w/v) in Tris-buffered saline (10 mM Tris-HCl, pH 7.5, and 150 mM NaCl) with 0.1% Tween 20 (v/v) (TBST). Membranes were incubated overnight in the presence of the primary antibodies (Supplementary Table S1) diluted in the blocking buffer at 4°C. Afterward, the membranes were washed four times with TBST for 10 min each time and then incubated with an IR secondary (LI-COR, Lincoln, NE) at 1:5,000 for 90 min at room temperature. The blots were then washed four more times with TBST for 10 min each time before being processed with the LI-COR Odyssey (LI-COR Biosciences).

2.12. Transmission electron microscopy (TEM)

Electron microscopy imaging was used to confirm the morphology and size of EVs, as shown in our previous studies [55, 56]. Briefly, EV isolates were resuspended in 30 μL of filtered PBS, and intact EVs (15 µL) were dropped onto Parafilm. A carbon-coated 400 Hex Mesh Copper grid (Electron Microscopy Sciences, EMS) was positioned using forceps with the coating side down on top of each drop for 1 h. Grids were rinsed three times with 30 µL filtered PBS before being fixed in 2% PFA for 10 minutes (EMS, EM Grade). The grids were then transferred on top of a 20 µL drop of 2.5% glutaraldehyde (EMS, EM Grade) and incubated for 10 min. Samples were stained for 10 min with 2% uranyl acetate (EMS grade). Then the samples were embedded for 10 min in a mixture of 0.13% methyl cellulose and 0.4% uranyl acetate. The coated side of the grids was left to dry before imaging on the Transmission Electron Microscope HT7800 (Hitachi, Japan).

2.13. mRNA quantitative analysis by RT-qPCR

Total RNA was isolated from different cell samples using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol. The isolated RNA samples were further treated with the DNA-Free RNA Kit (Zymo, Irvine, CA, USA) to remove genomic DNA contamination. Reverse transcription was carried out according to the manufacturer’s instructions using 2 μg of total RNA, anchored oligo-dT primers (Operon, Huntsville, AL), and Superscript III (Invitrogen, Carlsbad, CA, USA). The software Oligo Explorer 1.2 Primers (Genelink, Hawthorne, NY, USA) was used to design the real-time PCR primers specific for target genes (Supplementary Table S2). For normalization of expression levels, β-actin was used as an endogenous control. Using SYBR1 Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA), real-time PCR reactions were performed on an ABI7500 instrument (Applied Biosystems). The amplification reactions were performed as follows: 2 min at 50 °C, 10 min at 95 °C, and 40 cycles of 95 °C for 15 sec and 55 °C for 30 sec, and 68 °C for 30 sec, followed by a melt curve analysis. The Ct values of the target genes were first normalized to the Ct values of the endogenous control β-actin. The corrected Ct values were then compared to the experimental control. Fold changes in gene expression were calculated using the comparative Ct method: $2^{{-(∆C}_{ttreatment}-{∆C}_{tcontrol})}$ to obtain the relative expression levels.

2.14. Enzyme-linked immunosorbent assay (ELISA)

The ELISA assay was performed to determine the secreted cytokines BDNF, IL-1β, and IL-6. MGC culture supernatants were collected at 72 h after adding stimulations. Concentrations of cytokines were measured according to the manufacturers’ instructions (Invitrogen). Briefly, the standards and samples were incubated in 96-well microplates overnight at 4°C or at room temperature for 2–2.5 h. Then biotin conjugate was added, and the plate was incubated for one hour at room temperature. Next, the avidin-horseradish peroxidase (HRP) solution was added for 30–45 min, followed by 3,3’,5,5’-tetramethylbenzidine (TMB) substrate solution for 30 min in the dark. Then the reaction was stopped by the stop solution. The absorbance was measured using a microplate reader (Bio-Rad, Richmond, CA) at a wavelength of 450 nm. All cytokine samples were run in triplicate.

2.15. AD EV protein cargo analysis by Liquid chromatography-tandem mass spectrometry (LC-MS/MS) based proteomics

The EV protein extraction and analysis were described in our previous studies [32, 57]. Up to 30 µg proteins were isolated on S-trap micro column (Protifi, K02-micro). The isolated proteins were alkylated and digested on column based on manufacturer’s instructions. Then all the samples (triplicate for each group) were vacuumed dried and submitted to FSU Translational Science Laboratory. The samples were analyzed on the Thermo Q Exactive HF as previously described [58, 59]. Briefly, resulting raw files were searched with Proteome Discoverer 2.4 using SequestHT, Mascot and Amanda as search engines. Scaffold (version 5.0) was used to validate the protein and peptide identity. Peptide identity was accepted if Scaffold Local false discovery rate (FDR) algorithm demonstrated a probability greater than 99.0%. Likewise, protein identity was accepted if the probability level was greater than 99.0% and contained a minimum of two recognized peptides. Gene Ontology (GO) annotation was performed by g:Profiler.

2.16. AD EV miRNA cargo analysis by miRNA-sequencing

EV-associated miRs were isolated and sequenced in triplicate. EV samples were treated with RNase (ThermoFisher, AM2294) to final concentration of 50 ng/mL, at room temperature for 30 mins. RNase inhibitor (NEB, M0314) and PCR grade water were added to EV samples to make a total volume of 200 μL. miRs were isolated by adding 600 μL Trizol LS (ThermoFisher, 10296010) according to manufacturer’s instruction. To increase the yield of small RNAs, three volumes of 100% ethanol and linear acrylamide (VWR, 97063–560) were used instead of isopropyl alcohol and incubation time was also increased to overnight at −20°C. The isolated RNAs were quantified by Qubit microRNA assay kit (ThermoFisher, Q32880). Small RNA libraries were generated with NEBNext Multiplex Small RNA Library Prep Set for Illumina (NEB; E7300). To increase yield and prevent primer/adaptor dimer, 3’ SR primer was diluted to 1:5 and ligation time was increased to overnight at 16°C. Similar to mRNA-seq library preparation, HS DNA chip and KAPA library quantification kit were used before submitting to sequencing by Illumina NovaSeq 6000 in Florida State University College of Medicine Translational lab.

2.17. Statistical analysis

Experimental results were expressed as means ± standard deviation (SD). Statistical comparisons were performed by one-way ANOVA and Tukey’s post hoc test for multiple comparisons, and significance was accepted at p<0.05. For comparisons of two conditions, the student’s t-test was performed for the statistical analysis.

3. Results

3.1. Differentiation of iPSC-derived MGCs and response characterization to stimulation

Differentiation of MGCs from human iPSCs included several phases through stage-wise induction by different cocktails of growth factors: hemogenic endothelial cells, hematopoietic progenitors, myeloid progenitors, microglia precursors, and finally MGCs (Figure 1A). Immunocytochemistry demonstrated that on Day 15, the marker of hematopoietic progenitor cells, CD31, was expressed. On Day 35, the derived MGCs expressed markers such as CD11b, IBA1, and P2RY12 (Figure 1B). Flow cytometry results showed that CD11b was moderately expressed (29.7%) on day 35, while IBA1 and P2RY12 were highly expressed at 84.0% and 74.3%, respectively (Figure 1C). These results demonstrate that the derived cells had a microglia-like phenotype.

Figure 1. Differentiate of microglia-like cells (MGCs) from human iPSCs.

(A) MGC differentiation timeline. (B) MGC markers examined by immunocytochemistry. Scale bar: 100 μm. (C) MGC marker expression, examined by flow cytometry. Red color: isotype control; blue color: marker of interest.

The functional characteristics of the MGCs were examined by exposure to Aβ42 oligomers for pro-inflammatory response (M1) and dexamethasone for anti-inflammatory response (M2). When stimulated with Aβ42 oligomers, RT-qPCR results showed that the expression of pro-inflammatory markers TNFα, IL-6, and IL-12β was significantly increased by 1.3–2.3-fold compared to the untreated control. The anti-inflammatory markers IL-10, TGF-β, and CD163 were also elevated by 1.3–2.8-fold compared to the control (Figure 2A). When stimulated with dexamethasone, the microglia were directed to an anti-inflammatory state. IL-10 and CD163 were significantly upregulated by 1.8–2.2-fold compared to the control. The expression of pro-inflammatory markers TNFα and IL-12β was also slightly increased (1.3–1.8-fold). The immunophenotype of the microglia has been observed to be linked with the metabolic pathways [60, 61]. A shift between oxidative phosphorylation (OXPHOS) and glycolysis after exposure to proinflammatory stimuli could be induced. In this study, PDK1 expression was increased (~1.4-fold) after exposure to Aβ42 oligomers, while HK2, LDHA, and PKM2 remained similar to the control (Figure 2B). These four genes in the glycolytic pathway were enhanced (1.2–1.4-fold) under the dexamethasone treatment. The MGC exposure to Aβ42 oligomers showed the increased gene expression (1.2–1.4-fold) in the pentose phosphate pathway, including 6PGD, TKTL1, and TALDO1 (Figure 2C). But the exposure to dexamethasone did not affect the genes in the pentose phosphate pathway. These results demonstrated the functional characteristics of the MGC in immunometabolism.

Figure 2. Characterization of MGC metabolism in response to the stimulation of Aβ42 oligomers and dexamethasone.

mRNA expression was determined by RT-qPCR. (A) Pro- (M1) and anti- (M2) inflammation gene expression. N=3. (B) Glycolytic pathway and (C) pentose phosphate pathway gene expression. N=3. * indicates p<0.05.

In addition, MGCs were stimulated by Aβ42 oligomers with the addition of supernatant EVs (SuEVs, isolated from spent media of brain organoids) and matrix-bound nanovesicles (MBVs, isolated from ECM of the brain organoids) from iFCo to examine the influence of the EVs on MGC immunophenotype. SuEVs and MBVs were added at different doses (i.e., 0.2 μg, 1 μg, and 5 μg), based on their protein contents (μg). The SuEV 0.2 condition showed slightly reduced mRNA expression of the M1 marker IL-6, and all three doses led to decreased M2 markers CD163 and ARG1 (Supplementary Figure S1A). For the MBV group, the MBV 1 condition enhanced the expression of M2 markers IL-10 and CD163 (Supplementary Figure S1B). The ELISA of IL-1β secretion by MGC did not show differences among all the treatments, but the SuEV 0.2 condition had the lowest IL-6 secretion compared to the SuEV 1 and SuEV 5 conditions (Supplementary Figure S1C, S1D), which was consistent with the IL-6 gene expression from the RT-qPCR results. The MBV group showed a dose-dependent pattern for IL-6 secretion. When more MBVs (0.2, 1, and 5 μg based on protein content) were used, lower IL-6 concentration was observed. These results demonstrated that the MGC immunophenotype could be affected by exposure to the brain organoid-derived EVs.

3.2. Co-culture of iPSC-derived forebrain cortical organoids (iFCo) and MGCs

Forebrain [43–45] and hindbrain [5, 46] organoid characterizations have been reported in our previous studies. After the MGC derivation, the functionalization of brain organoids with MGCs was investigated. MGCs were labeled with Cell Tracker Red for 30 min and then mixed with iFCo at a ratio of 1:6 based on cell numbers. The migration of MGCs to iFCo was observed for 7 days (Figure 3A). On D0, iFCo were surrounded by MGCs; on D1, MGCs started migrating into iFCo, especially at the location where MGCs were more concentrated on D0. On D4 and D7, the red color kept spreading into iFCo, which indicated continuous integration of MGCs into iFCo. After 7 days, the co-cultured organoids were plated on a Matrigel-coated surface for immunocytochemistry. Neuronal markers β-tubulin III, glutamatergic and GABAergic neuron markers GLU and GABA, as well as pre-synaptic and post-synaptic markers SYN1 and PSD95, were detected in the co-cultured cells (Figure 3B). RT-qPCR showed that the microglial marker P2RY12 in the co-culture was much lower than MGCs, which was reasonable considering MGCs only accounted for 14% of cells in the co-culture (Figure 3C). However, IBA1 was 1.6–1.7-fold higher in the co-culture compared to the MGC group, suggesting that the brain organoid microenvironment promoted MGC maturation. The forebrain marker FOXG1 was 4-fold higher in the co-culture than in the iFCo condition, showing the contribution of MGCs to the forebrain identity of iFCo (Figure 3D). But the expression of cortical layer markers CTIP2, SATB2, and TBR1 did not change with the integration of MGCs. A co-culture consisting of MGCs and induced hindbrain cerebellar organoids (iHCo) was also performed (Supplementary Figure S2). Consistent with MGC+iFCo, the MGC+iHCo also promoted IBA1 expression, and the hindbrain microenvironment was more favorable for P2RY12 expression than the forebrain niche. However, MGCs did not influence the expression of any tested hindbrain markers (EN1, MATH1, NEPH3, and OLIG2).

Figure 3. Characterization of MGCs and iFCo co-culture.

MGCs were labeled with Cell Tracker Red and added to the day 25 iFCo. (A) Images of the co-culture showing migration of Cell Tracker Red-labeled MGCs into iFCo. Scale bar: 200 μm. (B) Neuronal and synaptic marker expression in the co-culture revealed by immunocytochemistry. Scale bar for β-Tubulin III, Glutamate, GABA, and SYN1: 200 μm; scale bar for PSD95: 10 μm. mRNA expression of (C) microglia and (D) forebrain cortical markers in the co-culture. N=3. * indicates p<0.05. iFCo: induced forebrain cortical organoids.

3.3. Comparison of co-cultured brain organoids and MGCs for AD-associated gene expression and immunometabolism

To induce neural degeneration, two types of EVs from the AD patient-derived brain organoids, with APOE4 mutation for sporadic AD and PS1 mutation for familial AD, respectively, were isolated and used to treat co-cultured brain organoids and MGCs at a dose of 500 EVs per cell. (Figure 4A, Supplementary Figure S3). After 72 h, gene expressions were examined, including amyloid β and Tau pathology regulating genes CASS4, UNC5C, CD2AP, and BIN1; oxidative stress and neuroinflammation regulating genes, MEF2C and MINK1; microglia activation regulating gene TREM2; and mitochondrial function regulating gene TOMM40. CASS4 expression was enhanced by APOE4 EV treatment in the co-culture group (~1.4-fold) and the MGC group (~2.5-fold) (Figure 4B). UNC5C was increased by APOE4 EVs in the co-culture group, but it was increased by PS1 EVs in the MGC group (Figure 4C). CD2AP had an elevated level in MGCs by APOE4 EV treatment but was decreased (~0.5-fold) in co-culture by both APOE4 and PS1 EV treatment (Figure 4D). BIN1 expression was similar for all the groups (Figure 4E), as well as MEF2C (Figure 4F). MINK1 expression was enhanced by APOE4 EV treatment in the co-culture group (~1.3-fold) and the MGC group (~1.2-fold) (Figure 4G). PS1 EVs induced higher TREM2 expression (~1.5-fold) in the MGC group but inhibited TREM2 (~0.5-fold) in the co-culture group (Figure 4H). TOMM40 expression was similar for all the groups (Figure 4I). These results indicate the differential responses of the co-cultured organoids and MGCs to the stimuli of the AD phenotype.

Figure 4. Exposure of MGCs and the co-culture to EVs secreted by brain organoids differentiated from AD (APOE4 and PS1)-associated iPSCs to induce neural degeneration.

(A) Illustration of exposure of MGCs and co-culture to AD EVs. mRNA expression of neural degeneration markers by RT-qPCR. Aβ and Tau pathology regulating genes: (B) CASS4, (C) UNC5C, (D) CD2AP, and (E) BIN1, respectively. (F) Oxidative stress and (G) neuroinflammation regulating genes, MEF2C and MINK1. (H) Microglia activation regulating gene TREM2. (I) Mitochondrial function regulating gene TOMM40. N=3. * indicates p<0.05. N.S.: not significant.

Genes in the glycolytic pathway and pentose phosphate pathway were investigated. EV treatment did not have much impact on the glycolytic pathway, except for enhancing PDK1 expression in both co-culture and MGCs by APOE4 EVs and slightly decreasing LDHA in MGCs by PS1 EVs (Figure 5A, 5B). Genes in pentose phosphate pathway, including 6PGP, G6PD, TALDO1, and TKTL1, were all reduced (~0.5–0.75-fold) by PS1 EVs in the co-culture group, but not affected by APOE4 EVs. 6PGD and TALDO1 were slightly inhibited (~0.8-fold) in the MGC group by PS1 EVs, but interestingly, TKTL1 was enhanced (2–3-fold) in the MGC group by both APEO4 and PS1 EVs (Figure 5C, 5D). APOE4 EVs induced strong pro-inflammatory gene expression in the MGC group, including IL-6 (~2.5-fold), IL-12β (~5-fold), iNOS (~1.5-fold), and TNFα (~2-fold), but these genes remained at a constant level in the co-culture group except for the reduced iNOS (Figure 5E, 5F). PS1 EVs upregulated IL-6 (~2-fold) and IL-12β (~7-fold) in the MGC group, but reduced iNOS and TNFα (~ 0.5-fold) in co-culture. The two interleukin genes were not significantly affected. These results indicate that the neural microenvironment in the co-cultured brain organoids reduced the pro-inflammatory response induced by the EVs from the AD patient-derived brain organoids compared to the MGC alone.

Figure 5. Metabolic and inflammatory alterations of MGCs and the co-culture in response to EVs secreted by brain organoids from AD-associated iPSCs.

mRNA expression of carbon metabolism and inflammation markers was determined by RT-qPCR. (A) Glycolytic pathway genes for co-culture. (B) Glycolytic pathway genes for MGCs. (C) Pentose phosphate pathway genes for co-culture. (D) Pentose phosphate pathway genes for MGCs. (E) Pro-inflammatory genes for co-culture. (F) Pro-inflammatory genes for MGCs. N=3. * indicates p<0.05.

3.4. Evaluation of therapeutic MGC and co-culture EVs in AD-stimulated brain organoids

To induce neural inflammation and degeneration, EVs from the AD patient-derived brain organoids with APOE4 mutation were added to the MGC and co-culture. Cytokine release of the EV-treated cells was measured by ELISA. For MGCs, APOE4 EVs did not significantly change the concentrations of BDNF (a potential diagnostic biomarker and a therapeutic molecule for AD [62]) and IL-6 compared to PBS vehicle control (Figure 6A, 6B). But in co-culture, APOE4 EVs slightly decreased BDNF secretion and increased IL-6 concentration (Figure 6C, 6D), suggesting a stronger response to inflammation.

Figure 6. Evaluation of therapeutic MGC and co-culture EVs on the cytokine secretion of brain organoids exposed to EVs from AD-associated brain organoids (AD EVs).

ELISA results for concentrations of (A) BDNF and (B) IL-6 secreted by AD EV-treated MGCs. Concentrations of (C) BDNF and (D) IL-6 secreted by AD EV-treated co-culture (CO). (E) Illustration of therapeutic EV treatment on the brain organoids exposed to AD EVs. EVs from brain organoids derived from iPSCs of AD patient (APOE4) cells were added to co-culture. Then, the EVs from healthy M0 MGC, M2 MGC, and co-culture were added to the stimulated organoids. Scale bar of the plated co-culture brain organoids: 100 μm. N=3. Concentrations of (F) BDNF and (G) IL-6 secreted by organoids that were treated by various types of HC EVs. * indicates p<0.05. N.S.: not significant. M0 and M2: unstimulated and IL-4-stimulated MGCs. HC: healthy control.

Next, the EVs from M0 MGC, M2 MGC, and the co-cultured brain organoids (CO) derived from healthy human iPSCs were evaluated for the ability to reduce neural inflammation and degeneration (Figure 6E). Size distributions, exosomal markers, and morphology of different types of EVs were characterized (Supplementary Figure S4). These EVs had mode size of 150–200 nm based on NTA analysis (Supplementary Figure S4A-S4C). The EVs expressed positive exosomal markers HSC70 and CD81 and showed no expression of the negative exosomal marker Calnexin (Supplementary Figure S4D). The TEM images demonstrated the cup-shaped morphology typical for exosome for the three EV groups (Supplementary Figure S4E).

For the healthy M0 MGC EV, M2 MGC EV, and CO EV treatments of APOE4 AD EV-stimulated co-culture, ELISA results showed that different EVs did not induce significant changes in BDNF secretion (Figure 6F). However, M2 MGC EVs and CO EVs significantly lowered IL-6 secretion compared to the M0 MGC EVs, showing the ability to reduce neural inflammation (Figure 6G). The EV dose-dependent (i.e., 5000 and 1000 EV particles/cell) effects were observed, with IL-6 increased by the M2 5000 and CO 5000 groups compared to the M2 1000 and CO 1000 groups. These results indicated that 5000 EVs/cell may be overdosed, so the dose of 1000 EVs/cell was used in the following experiments.

Then the effects of the three types of healthy control EVs (HC EVs, M0 MGC, M2 MGC, and CO EV) on the gene expression of the APOE4 AD EV-included co-culture were determined, including pro-inflammatory genes IL-6, TNFα, IL-12β, and iNOS; brain-derived neurotrophic factor gene BDNF; microglia activation regulating gene TREM2; and Aβ and Tau pathology regulating genes CASS4, CD2AP, and UNC5C. RT-qPCR results demonstrated that these HC EVs did not affect IL-6 and TNFα expression compared to PBS vehicle control (Figure 7A, 7B). M0 and CO groups reduced IL-12β levels (~0.6-fold and ~0.2-fold, respectively) (Figure 7C). All three types of EVs significantly reduced iNOS level (~0.4–0.6-fold) (Figure 7D). BDNF was significantly decreased (~0.4-fold) by the CO group (Figure 7E). All three types of EVs inhibited TREM2 by approximately 0.5–0.7-fold (Figure 7F). CASS4 was inhibited (~0.5-fold) by the M0 and CO groups (Figure 7G). However, the EV treatments did not affect the expression of CD2AP, UNC5C, and MINK1 (Figure 7H, 7I, Supplementary Figure S5A). Aβ and Tau expression levels were examined by Western blot, but AD EVs could not induce detectable Aβ and Tau accumulation in MGCs (Supplementary Figure S5B, S5C). Taken together, these data indicate that M0 MGC EVs, M2 MGC EVs, and CO EVs differently reduced the expression of pro-inflammatory genes and affected some microglia function and Aβ and Tau pathology-regulating genes in the cells of APOE4 AD EV-included co-culture.

Figure 7. Effects of therapeutic MGC and co-culture EVs on APOE4 AD EV-stimulated brain organoids for inflammation and neural degeneration markers.

mRNA expression was performed by RT-qPCR. Expression of pro-inflammatory genes (A) IL-6, (B) TNFα, (C) IL-12β, and (D) iNOS. (E) Brain-derived neurotrophic factor (BDNF) gene. (F) Microglia activation regulating gene TREM2. Expression of Aβ and Tau pathology regulating genes (G) CASS4, (H) CD2AP, and (I) UNC5C. N=3. * indicates p<0.05. N.S.: not significant.

3.5. AD EV cargo analysis by proteomics and miRNA-sequencing

The protein cargo of EVs from organoids of healthy controls (HC), APOE4, and PS1 genotypes were analyzed by LC-MS/MS based proteomics. Principal component analysis (PCA) plot showed distinct clusters of HC EV, APOE4 EV, and PS1 EV groups (Figure 8A), which indicates the selective protein packing under the impact of mutated genes. Heatmap analysis showed that protein profiles of the three groups were different from each other, supporting the PCA results (Figure 8B). Venn diagram showed that there were 594 total proteins, and 375 of them were detected in all three EV groups (Figure 8C). 157 proteins were shared by HC EV and PS1 EV, 16 proteins were shared by HC EV and APOE4 EV, and only 9 proteins were shared by APOE4 EV and PS1 EV. The number of unique proteins in HC EV, APOE4 EV, and PS1 EV were 50, 0, and 5, respectively. Several selected proteins were plotted in Figure 8D. The APOE4 EV carried the highest amount of APOE and APP proteins, and HC EV contained the least amount of APOE and APP. EV marker CD63 was comparable among the three groups, but TSG101 demonstrated higher expression in PS1 EV.

Figure 8. Proteomic analysis of APOE4 EV and PS1 EV protein cargo.

(A) Principal component analysis (PCA) plot to show the clusters of APOE4 EVs, PS1 EVs, and HC EVs; (B) heatmap illustrations of differentially expressed proteins (DEPs) among the three EV groups; (C) Venn diagram of DEPs for the three EV groups; (D) Selected protein expression of EV markers and function related proteins. HC: Healthy control.

miRNA-sequencing was performed to analyze miRNA cargo in the EVs. PCA plot showed that HC EV had a tight cluster, while APOE4 EV condition was mainly clustered on PC2, and PS1 EV condition scattered along both PC1 and PC2 (Figure 9A), indicating a high variability within the group. Since APOE4 EVs were used in MGC and co-culture stimulation, these EVs were further analyzed for the top 20 miRs, including miR-128, 92a, 92b, 148a, 1246, 320a, 99a, 92b etc. (Figure 9B and Supplementary Table S3). There were 70 targeted genes of the 20 miRs, and the target analysis identified 9 highlighted genes to be central hubs, including BCL2, BCL2L11, BACE1, COX1, COX2, CCND2, CDKN1A, MTOR, and SOX4. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of these target genes revealed 9 pathways, such as mitophagy - animal pathway, cell cycle, and neuroactive ligand–receptor interaction pathway (Figure 9C). Taken together, the EV cargo analysis indicates the proteins and miRNAs in the AD EV groups may contribute to neural degeneration and the healthy EVs may ameliorate this process and exhibit therapeutic cargo.

Figure 9. MicroRNA sequencing of APOE4 EV and PS1 EV miRNA cargo.

(A) PCA plot to show the cluster of APOE4 EVs, PS1 EVs, and HC EVs. (B) miR target analysis of top 20 miRs in the APOE4 EV group. (C) The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of the targeted genes for the top miRNAs in the APOE4 EV group.

4. Discussion

In this study, human iPSC-derived MGCs were evaluated for the response to various stimulants. For the treatment with dexamethasone for 48 h, both pro- and anti-inflammatory genes showed increased expression compared to the non-treated control. Dexamethasone is known to perform an anti-inflammatory role. In a short-term treatment (24 h), it attenuates lipopolysaccharide-induced neuroinflammation in BV2 microglial cells, including reducing secretion of pro-inflammatory factor nitric oxide and increasing secretion of anti-inflammatory cytokine IL-10 [63]. In a longer-term treatment (72 h), primary rat microglial cells showed a reduction of phagocytosis function and cell proliferation, as well as pro- and anti-inflammatory cytokine production [64]. Aβ42 stimulation was expected to induce pro-inflammatory behavior of microglia [65], but in this study the anti-inflammatory genes were enhanced too. This may reflect a transition state from M1 to M2 as the changes of other factors such as TREM2, IL-4, and translocator protein 18 KDa, which needs to be further verified [66–68]. On the other hand, this may also be attributed to the immaturity of iPSC-derived MGCs. The validation of MGCs as a microglia surrogate needs further investigations related to development, health, and disease progression [20]. Moreover, pro- and anti-inflammatory cytokine levels in the conditioned media may provide more definitive results. In addition, microglia have a sequence of activation states indicated by early M1-like deleterious activities followed by an M2-like neuroregenerative state [69]. The shift is dynamic, and what was observed in this study may be due to the co-existence of M1 and M2 states during the shift, including the upregulated pro- and anti-inflammatory genes by dexamethasone and Aβ42 respectively.

In this study, a 7-day co-culture showed the increased expression of the microglial marker IBA1 and the forebrain marker FOXG1, which indicated the mutual benefit of iFCo and MGCs in co-culture. In Fagerlund et al.’s study, erythromyeloid progenitors (on day 8 of differentiation from iPSC) could migrate to brain organoids and develop into MGCs [17]. The 90-day co-culture showed that MGC within the organoids promoted neuronal network maturation and recapitulated some aspects of microglia-neuron co-development in vivo. In another study, iPSC-derived primitive-like macrophages and brain organoids, both on their 26th day were co-cultured for 15–19 days, and the former could differentiate into MGCs in the organoids [70]. MGCs modulated neuronal progenitor differentiation by limiting proliferation and promoting axonogenesis. The mechanism was revealed: MGCs contained high levels of PLIN2+ lipid droplets that exported cholesterol and its esters, which were taken up by neuronal progenitor cells in the organoids. Longer co-culture time should be performed in future studies to investigate the interactions between iFCo and MGCs.

TREM2 is a triggering receptor expressed on myeloid cells 2, which was found to be related to an increased risk of late-onset AD [71]. TREM2 is essential for microglial activation, proliferation, and clustering around Aβ plaques, thus maintaining cellular function. However, depending on timing, amplitude, and duration, microglial activation may have either beneficial or detrimental effects on AD development [72]. The role of TREM2 in developed Tau neurofibrillary tangles remains unclear: The TREM2 deficiency was shown to reduce pathology in one study and correlate with exacerbated pathology in another study [73, 74]. In this study, TREM2 gene expression was reduced in co-culture but was enhanced in MGCs. EVs from MGCs and co-culture also showed TREM2 regulatory role in vitro. Whether cells or EVs play a more protective role could be further investigated in in vivo models.

CD2AP, a CD2-associated protein, has been reported to be associated with sporadic AD [75]. CD2AP plays critical roles in AD by affecting the amyloid precursor protein (APP) sorting process in early and late endosome [76], modulating Tau-mediated neurotoxicity, including preventing neural loss, supporting synaptic strength, regulating plasticity [77, 78], and maintaining the blood-brain barrier integrity [79]. Its elevation in the brain could be a promising treatment, which makes CD2AP a target for AD therapy. In this study, the CD2AP gene was decreased in co-culture but was increased in MGCs, while EVs from MGCs and co-culture did not change their level. MGCs showed its potential in promoting both expressions of TREM2 and CD2AP genes. CASS4, a CAS scaffolding protein family member 4, is another late-onset AD risk factor [80]. It is involved in cell adhesion, AD progression, neurofibrillary tangles and neuritic plaque burden, and Tau toxicity [81]. In this study, CASS4 gene expression was promoted in both MGCs and co-culture cells but was inhibited by their EVs, which indicates a different CASS4-regulating mechanism between the cells and EVs.

UNC5C, the UNC-5 netrin receptor C, is another late-onset AD risk factor [82]. Its rare mutation T835M has been shown to increase neuronal cell death in vulnerable regions of the AD brain [83]. UNC5C could be cleaved by δ-secretase, and the overexpression of δ-secretase cleaved UNC5C-promoted AD pathologies in APP/PS1 mice, whereas reducing its level alleviated AD pathologies [84]. MINK, misshapen/NIKs-related kinase 1, is also considered an AD risk gene [85]. MINK1 has been implicated in many signaling pathways, such as Wnt signaling, JNK signaling, Hippo signaling, etc. It also plays important roles in the regulation of reactive oxygen species, inflammation, and synaptic morphology [86–90]. Inhibition of MINK1 blocks the downstream Aβ accumulation and AD-like pathology in murine models [91]. In this study, both UNC5C and MINK1 genes were enhanced in MGCs and co-culture cells when exposed to AD stimuli but did not change with treatment of HC EVs from MGC and co-culture. The findings demonstrated that HC EVs may not be able to reverse the expression of these two genes under AD stimuli.

The glycolytic pathway is critical for cellular energy production, transforming glucose into pyruvate while generating ATP and NADH, essential for various cellular functions. In the context of neurodegenerative diseases such as AD, alterations in the glycolytic pathway can have profound implications. Research indicates that the pyruvate dehydrogenase kinase 1 (PDK1) plays a significant role in the metabolism of neurons affected by AD [92]. It is reported that PDK1 was elevated in the brains diagnosed with AD [93]. In this study, PDK1 was upregulated in the MGCs and co-culture under AD stimuli. Inhibition of PDK1 is favorable in the mouse model due to the reduced motor impairment and memory impairment [93, 94]. However, in diabetic mice, enhanced PDK1 has demonstrated a protective effect against neuronal injury and memory loss [95]. This complex interplay highlights the importance of PDK1 regulation in the pathophysiology of AD, emphasizing the need for further research in this area.

The pentose phosphate pathway (PPP) serves as a vital metabolic pathway that diverges from glycolysis, primarily facilitating the generation of NADPH and ribose 5-phosphate, crucial for biosynthetic processes and antioxidant defense mechanisms. In AD, where oxidative stress and inflammation contribute to neuronal death, the compromised functionality of the PPP can exacerbate these pathological processes [96]. In this study, 4 genes in PPP were decreased in co-culture after AD-EV stimulation, and only 2 genes were decreased in MGC-only group. But the TKTL1 (transketolase-like 1) gene was enhanced in MGCs. TKTL1 is proposed to encode an enzyme that contributes to the non-oxidative phase of the PPP, facilitating the interconversion of sugar phosphates [97]. As a reversible rate-limiting enzyme, TKTL1 regulates metabolic flux to produce either NADPH or pentoses according to cellular requirements [98]. The different TKTL1 levels in MGCs and co-culture may be correlated to different NADPH modulation, which needs further investigation.

Pro-inflammatory responses play a pivotal role in the AD pathology, significantly influencing its onset and progression. In AD, the buildup of Aβ plaques and neurofibrillary tangles triggers an inflammatory cascade characterized by the activation of microglia and astrocytes [99]. Once activated, these immune cells release a range of pro-inflammatory cytokines, such as IL-1β, IL-6, IL-12β, inducible nitric oxide synthase (iNOS), and TNF-α [100, 101], which were also demonstrated in this study. This chronic neuroinflammation can exacerbate neuronal dysfunction and contribute to synaptic loss and cognitive decline as the hallmarks of AD [102]. Studies indicate that interventions aimed at modulating inflammatory pathways may hold therapeutic potential, as reducing pro-inflammatory signaling can mitigate neurotoxic effects and promote neuronal survival [103–105]. In this study, the co-cultured cells, and EVs from MGCs and co-culture exhibited the ability in alleviating pro-inflammatory genes, serving as promising therapeutic strategies targeting neuroinflammatory processes in AD.

In this study, co-culture MGC-brain organoids were examined as neural degeneration models by exposure to AD-EVs to evaluate AD-related genes and risk genes. Of interest, co-culture and MGCs demonstrated different gene regulation on two of the risk genes, TREM2 and CD2AP. It has been demonstrated that TREM2 deficiency attenuated neuroinflammation and protected against neurodegeneration in the PS19 mice model; however, TREM2 haploinsufficiency increased tau pathology and showed a greater impairment in their injury response [73, 106]. Likewise, contradictory results were reported for the role of CD2AP in AD: on the one hand, CD2AP deficiency in mice neurons aggravated AD phenotypes and pathology; on the other hand, CD2AP was found to be elevated in microglia of AD patients, and microglial CD2AP deficiency exerted protection in an AD model of amyloidosis [107, 108]. Therefore, whether it is beneficial or detrimental when these two genes are up- or down-regulated in MGC or co-culture needs more evidence. Though glycolytic pathway genes tended to remain unchanged, PDK1 expression increased. It has been reported that PDK1 was elevated in AD mice, and such elevation not only shifted cells toward an anerobic metabolism but also affected the PDK1/Akt axis and aggravates AD pathology [92, 109]. While AD-EVs, especially PS1-EVs, downregulated the PPP pathway in co-culture, their impact was mainly on TKTL1 in MGCs. As a branch pathway from glycolysis, PPP produces NADPH and synthesizes ribose-5-phosphate, and it tends to be enhanced in AD alongside the glycolytic pathway [110–112]. However, the mechanism underlying decreased PPP of co-culture upon AD-EV treatment and different responses in co-culture and MGCs for reducing equivalents, nucleotide precursors, or more complex requirements needs to be further investigated.

AD-EVs increased pro-inflammatory genes in microglia as expected [113, 114]. But it was interesting to observe that these genes were reduced in co-culture. Our study hypothesized that both co-culture and MGCs had the ability to protect themselves from AD-EVs’ negative impact. Our study further hypothesized that EVs from co-culture and MGCs may possess therapeutic potential to prevent neural degeneration. While the dose of 5000 particles/cell was considered an overdose, 1000 EVs/cell did not induce an increase in IL-6 cytokine concentration. At the mRNA level, EVs from co-culture and MGCs significantly attenuated pro-inflammatory gene expression, and co-culture EVs showed stronger regulation in reducing these gene levels. AD risk genes TREM2 and CASS4 were decreased by co-culture EVs and MGC EVs. Though CASS4 may be involved in AD through the NEDD9-CASS4-PTK2B axis, which has been widely studied as a cancer signaling pathway [115], reports about the individual CASS4 gene in AD were barely seen. Interestingly, M0 EVs tended to be more responsive than M2 EVs in mitigating the neural degenerative gene expression, which could be attributed to EV dose. Taken together, co-culture and MGCs displayed differences in expression of AD risk and pro-inflammatory genes upon AD-EV treatment, which requests further investigation to reveal the mechanism, including changes in protein or signaling cascades. More importantly, EVs from both co-culture and MGCs demonstrated the potential of downregulating pro-inflammatory effects in neural degeneration.

In AD EV protein cargo analysis, APOE4 EVs contained less overlapped proteins with HC EVs than the PS1 EV vs. HC EV pair. Therefore, APOE4 EVs were chosen as the stimulant to evaluate HC EV therapeutic potential. Among the differentially expressed proteins, APOE and APP were found to be highly enriched in APOE4 EVs. This is consistent with previous reports that EVs spread APOE and APP proteins in AD models [116, 117]. APOE and APP contribute to AD pathology by inducing disrupting lipid transfer, impairing mitochondrial metabolism, regulating innate immune pathways, and modulating neural cell adhesion molecule 1-dependent neurite outgrowth and synaptogenesis [118, 119]. APOE and APP also interact with each other, increasing each other’s abundance [116, 120]. miRNA cargo analysis of APOE4 EVs revealed several pathways targeted by top 20 miRNAs. Among these pathways, impaired mitophagy was reported to affect AD due to the accumulation of dysfunctional mitochondria [121]. Specifically, compared to APOEε2 and APOEε3, APOEε4 genotype demonstrated enrichment of mitochondrial proteins in glioblastoma T98G cells, and higher mitochondria marker level (TOMM40) [122–124]. In this study, APOE4 EVs did not significantly change the TOMM40 level in MGC and co-culture, but affected inflammatory responses and other AD risk genes, which may be related to the APOE and APP protein cargo and the top miRNAs.

5. Conclusions

In this study, human iPSC-derived MGCs were co-cultured with isogenic forebrain organoids. MGCs expressed microglial markers and were responsive to Aβ42 and dexamethasone stimulation. The co-cultured brain organoids expressed neuronal markers and synaptic markers, indicating a functional signal transduction. The presence of brain organoids and MGCs mutually promoted the expression of neural markers and microglia markers in the co-culture. In the neural degeneration model that was induced by EVs associated with AD patients, the co-culture organoids and the EVs from the healthy MGCs and the co-culture showed regulation of genes involved in pro-inflammation, Aβ and Tau pathology, microglial activation, and carbon metabolism, which provide multiple targets for potential AD therapies. This study paves the way for understanding the role of microglia and brain organoids in modeling neural degeneration and the development of EV-based therapeutic interventions.

Data and materials availability:

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.