Development and validation of a simplified method to generate human microglia from pluripotent stem cells

Our new differentiation protocol still mimics in vivo microglia ontogeny by first differentiating iPSCs into hematopoietic progenitor cells (HPCs), followed by passage into microglial differentiation medium, and concludes with a final maturation step by adding neural and astrocytic factors, thereby educating the microglia in a brain-like environment while maintaining a pure, homeostatic population of microglia (Fig. 1).

In order to test whether or not an extensively simplified and commercially available hematopoietic stem cell differentiation protocol could be used to generate equivalent homeostatic microglia, we differentiated four independent iPSC lines, and one ESC line (H9) in parallel using our previously published protocol [8] (iPS-microglia) and the simplified iPS-microglia 2.0 protocol described here. Our prior protocol required FACS to isolate CD43⁺ hematopoietic progenitors before transitioning to microglial differentiation medium containing IL-34, TGFβ-1, and M-CSF [14‐16]. To determine whether FACS sorting was also necessary for our newer approach, we compared three different sorting methods on the same iPSC background. On day 11 when HPCs were ready to be transitioned into microglial medium, we isolated CD43⁺ HPCs using either FACS, magnetic-activated cell sorting (MACS), or by simply collecting all non-adherent cells. After analyzing transcriptome changes through RNA-sequencing on these samples, we performed unbiased clustering and found that FACS- and MACS-sorted iPS-microglia 2.0 intercluster with unsorted samples demonstrating that these isolation procedures are not necessary for this updated protocol (Additional file 1: Figure S1). Notably, one of the first results we observed with this simplified protocol was that the number of CD43⁺ HPCs produced using this novel commercially available method was substantially increased, while still maintaining the high degree of purity (90–94%) for the HPC marker CD43 (Additional file 2: Figure S2). From one million starting iPS cells, 125 million CD43⁺ cells can be produced, representing an approximately 60-fold increase over our prior method. Following transition to microglial medium, the four lines of unsorted HPCs were further differentiated and matured. At the final day of microglial maturation, iPS-microglia or iPS-microglia 2.0 were harvested for RNA isolation and analyzed via RNA sequencing.

At a transcriptome level, our new protocol produced homeostatic human microglia that were virtually identical to microglia generated using our prior methods. Principal component analysis of the full transcriptome explained 73% of variation in all samples within PC1 (44%) and PC2 (29%) and revealed that our new microglia (iPS-microglia 2.0) closely cluster and are interspersed with microglia differentiated using our previously published protocol, yet are highly distinct from human CD14⁺ or CD14/CD16⁺ blood monocytes and dendritic cells (Fig. 2a, b, Fig. 3, Additional file 3: Table S1). Consistent with our previous findings, both iPS-microglia and iPS-microglia 2.0 exhibit very similar gene expression profiles to brain-derived cultured human microglia, although some differences between these groups remain. In order to highlight important microglial and monocyte enriched genes, we performed a secondary principal component analysis with a previously-identified microglial/monocyte focused gene set which revealed a developmental component (PC1, 48% of variance) and again shows interclustering of iPS-microglia 2.0 with microglia generated using our previously published approach (Fig. 2b, d). Importantly, this developmental trajectory remains quite distinct from monocytes and dendritic cells (Fig. 2b).

Interestingly, RNA sequencing analysis of HPCs generated using our previously published approach versus HPC 2.0 revealed some intriguing differences between these two populations. Based on the developmental trajectory shown in Fig. 2b, these data suggest that HPC 2.0 samples are closer to microglia on the primary principle component showing the developmental trajectory than HPCs from our previously described protocol. Indeed HPC 2.0 populations include lower percentages of cells expressing the primitive HPC marker CD235a, although both populations express equivalently high levels of another primitive HPC marker; CD43 which is typically absent in definitive HPCs [17] (Additional file 2: Figure S2, Additional file 4: Table S3). Importantly, although some differences do exist in gene expression between these two HPC populations, very few differences in gene expression persist once HPCs are matured into microglia (Fig. 1, Fig. 2c, d, Additional file 5: Table S2). For example, expression analysis comparing iPS-microglia and iPS-microglia 2.0 revealed only 55 differentially expressed genes. To determine whether these differences occur in a uniform pathway or are indicative of important functional effects that need to be considered, we next used Reactome 2016 gene ontology analysis to examine these 55 genes. This analysis revealed only three significant gene ontology pathways (Platelet activation, signaling and aggregation_Homo sapiens_R-HAS-76002, FDR = 0.03665; Chemokine receptors bind chemokines_Homo sapiens_R0HSA-380,108, FDR = 0.03665; Tryptophan catabolism_Homo sapiens_HSA-72,140, FDR = 0.03665). The chemokine receptor pathway is of course important for microglial function, although this pathway was implicated by only 3 differentially expressed genes: CXCL10, CCL5, and PF4. Thus, we expect that microglial functional activity will be largely equivalent between iPS-microglia and iPS-microglia 2.0.

To determine whether the functional activity of iPS-microglia 2.0 is indeed equivalent to microglia generated using our prior approach, we next compared phagocytic activity of cells generated using both methods [8]. Since the ability of microglia to clear pathogens and extracellular aggregates via phagocytosis is an important aspect of microglial function, we exposed microglia to several different substrates and measured the percentage of cells which phagocytose each substrate using an Amnis Imagestream which combines flow cytometry and high throughput immunofluorescence (Fig. 4). As expected, the levels of phagocytosis differed between the three varying substrates with beta-amyloid fibrils producing the highest response. S. aureus bioparticles, a TLR 1,2,6 agonist produced an intermediate degree of phagocytosis and Zymosan, a TLR 2/Dectin 1 agonist from S. cerevisia induced the lowest level of phagocytosis (Fig. 4). Importantly, regardless of the differential response to these three phagocytic substrates, iPS-microglia and iPS-microglia 2.0 exhibited identical rates of phagocytosis for each of the substrates, demonstrating that this simplified differentiation method does not alter this important microglial function (Fig. 4).

We previously demonstrated that iPS-microglia can engraft and ramify, fulfilling characteristic microglia morphology and marker expression in the brains of xenotransplantation-compatible MITRG (Knock-out: Rag2; Il2rg; Knock-in: M-CSFh; IL-3/GM-CSFh; TPOh) mice [8]. Thus, we aimed to further validate the identity of our iPS-microglia 2.0 through intracranial transplantation of iPS-microglia 2.0 into MITRG mice, and to compare this engraftment to equivalently transplanted iPS-microglia that were generated using our previously described differentiation method. In each case, fully mature microglia were transplanted into the hippocampus and overlaying cortex of adult mice which were sacrificed after 2 months for histological examination of morphology and key marker expression. Both iPS-microglia and iPS-microglia 2.0 can be identified within the mouse brain via expression of the human-specific nuclear marker, Ku80 (Fig. 5, green). Importantly, regardless of the differentiation method, transplanted human microglia display typical microglial morphology, extending complex branching processes. Both iPS-microglia and iPS-microglia 2.0 also express the microglial/monocyte marker Iba1 (Fig. 5, Overlay images C, G, K, & O, red) and the homeostatic microglial marker P2RY12 (Fig. 5 Overlay images, D, H, L, & P, red) in both cortex and hippocampus, indicating that these cells engraft well and remain homeostatic. Transplanted iPS-microglia 2.0 also exhibit the tiling and distinct niches typical of in vivo microglia, and can be seen interspersed with the endogenous population of mouse microglia (Fig. 5, arrows indicate Iba1⁺/Ku80⁻ mouse cells). Taken together, these findings further demonstrate that iPS-microglia 2.0 are equivalent to microglia generated using our previously published protocol and can be readily transplanted into MITRG mice to enable in vivo studies of human microglia.

TGFβ1 is a crucial astrocyte-derived cytokine that promotes microglial homeostasis [15, 18]. TGFβ1 signaling results in phosphorylation of smad2/3 and ultimately up-regulates expression of CX3CR1, an important receptor for microglial function and survival [19, 20]. Indeed, removal of either TGFβ1 or CX3CR1 greatly decreases microglia populations in murine models [15, 20]. As we have previously shown, removal of TGFβ1 from iPS-microglia for even 24 h also results in dramatic changes in the microglial transcriptome, including down-regulation of homeostatic signatures [8].

In order to increase cost-efficiency during iPS-microglia 2.0 differentiation, we attempted to replace recombinant TGFβ1 with Inducer of Definitive Endoderm 1 or 2 (IDE1, IDE2). As its name suggests, IDE has been used to differentiate iPS cells into definitive endoderm through activation of TGFβ signaling [21]. More specifically, IDE1/2 have been shown to induce phosphorylation of the downstream TGFβ signaling molecule smad2 [21]. This led us to hypothesize that IDE1 or IDE2 could induce expression of microglial genes in an equivalent fashion to recombinant TGFβ if added after mesoderm formation and hematopoiesis.

To determine whether IDE1 or IDE2 could replace recombinant TGFβ in our differentiation protocol, HPCs harvested on day 10 were placed into microglia differentiation media and varying concentrations of IDE1 or IDE2 (1 μM, 10 μM, 100 μM, 1000 μM) were added in place of TGFβ (Fig. 6a). During the first 4 days of microglia differentiation we used an Incucyte live imaging system to examine the growth kinetics of each group to provide an initial assessment of the effects of varying IDE concentrations. Surprisingly, this analysis demonstrated that IDE2, regardless of concentration, impaired normal microglial proliferation and thus was not studied further. In contrast, IDE1 was able to mimic the typical growth kinetics observed in control cells differentiated in parallel and maintained in normal TGFβ-containing medium. Because the control TGFβ microglial growth curve fell between the 10 μM and 1 μM IDE1 curves, we next adjusted the IDE1 concentrations to include a 5 μM dose. In addition, two higher concentrations of IDE1 (50 μM, and 500 μM) were also included as the growth kinetic measurements suggested that these concentrations might further increase the yield of microglia.

All groups were differentiated in parallel for the complete 38 day paradigm before RNA-sequencing was performed. To compare these new samples with our iPS-microglia, fetal and adult microglia, and other cell types, we generated a correlation matrix (Fig. 6b) which demonstrated that IDE1 iPS-microglia remain distinct from monocytes and dendritic cells and cluster closely with our other iPS-microglia. IDE-treated cells also exhibited strong expression of key microglial genes including CSF1R, P2YR12, TREM2, OLFML3, HEXB, and C1Q (Additional file 6: Table S4). Based on the hierarchical clustering within our correlation matrix (Fig. 6b), we find that microglia differentiated in IDE1 have transcriptomic profiles that are more similar to primary cultured microglia (fetal and adult microglia). One possible explanation for this finding is that IDE1 may have greater stability within the culture media than TGFβ1 and thus provide a more uniform and consistent activation of TGFβ signaling pathways. Alternatively, IDE1 could potentially provide increased specificity by only targeting smad2 signaling. To further understand the comparative effects of IDE1 versus TGFβ1 we further compared these groups of iPS-microglia using DEseq2 to highlight any differentially expressed genes (Additional file 6: Table S4). Interestingly, gene ontology analysis of these differentially expressed genes revealed only five pathways that were significantly enriched between these two populations of microglia (Fig. 6c), again showing their strong similarity. Taken together these data suggest that IDE1 can indeed be used in place of TGFβ1, although researchers should also consider these differentially altered pathways and genes in there decision whether or not to use this further modified approach.

Improved and simplified differentiation of iPSCs to CD43⁺ primitive hematopoietic progenitor cells (HPCs) is achieved using Stem Cell Technologies STEMdiff™ Hematopoietic Kit (Catalog # 05310). On day − 1, feeder-free iPSCs that have been expanded in TeSR-E8 media are passaged with ReLeaSR (STEMCELL technologies) into mTeSR E8 medium with 0.5 μM Thiazovivin onto matrigel coated (1 mg/mL) 6-well plates (Corning Costar). Small aggregates of ~ 100 cells each are plated at 10–20 aggregates per cm². The initial plating density is critical as higher density impairs mesoderm differentiation and lower density decreases yield. Thus one can plate iPSCs at 2–3 different densities and select the wells on day 0 that have optimal density to proceed with. When approximately two 100 cell colonies per cm² have been achieved, replace TeSR-E8 medium with medium A (Basal medium plus Supplement A at 1:200 dilution, 2 mL per well of a 6-well). On day 2 (48 h after original media change), do not fully change media, but rather replace 50% medium A, 1 mL per well of a 6-well. On day 3, carefully remove all media by tilting the plate to one side and aspirating from the edge. Then add 2 mL/well medium B (Basal medium plus supplement B at 1:200). Without removing media, supplement with 1 mL/well of medium B on days 5, 7, 9. On day 10 and again on day and 12, non-adherent cells may be collected. To maintain purity, do not wash cells off the well, but merely remove medium with non-adherent cells carefully and centrifuge 300 x G 5 min. After centrifugation, replace conditioned medium back to each well and add 1 mL fresh medium B if further collection on day 12 will be completed.

FACS analysis has confirmed that these non-adherent cells represent highly pure populations (> 93%) of CD43⁺ hematopoietic progenitor cells (Additional file 2: Figure S2). Importantly, simply collecting the floating cells is all that is required to isolate large numbers of highly purified CD43⁺ cells. No FACS or MACS isolation is required as identical microglia are produced using any of these three methods (Additional file 1: Figure S1). However, because the cells are not being sorted for purity, the collection of non-adherent cells must be carefully completed. Do not spray medium over adherent cells to wash as this will loosen cells which are not CD43⁺ and decrease culture purity.

At this point, HPCs may be frozen at 2–4 million cells per mL in BamBanker (Wako). If frozen, HPCs should be thawed directly into microglial differentiation medium with cytokines (below) and plated onto Matrigel-coated plates at 10,000 cells per cm². We typically find that viability post-thaw is between 70-95%, with improved viability when greater densities of HPCs are thawed together.

Volumes specified for 35 mm well (1 well of a 6-well plate).

On day 0 of iPS-microglia differentiation, plate HPCs at 10,000 cells per cm² onto 1 mg/mL Matrigel-coated plates (100,000 per 35 mm well). Plate cells into iPS-microglia medium at 2 mL per 35 mm well: DMEM/F12, 2X insulin-transferrin-selenite, 2X B27, 0.5X N2, 1X glutamax, 1X non-essential amino acids, 400 μM monothioglycerol, 5 μg/mL insulin. Immediately before use, microglial medium should be supplemented with 100 ng/mL IL-34, 50 ng/mL TGFβ1, and 25 ng/mL M-CSF (Peprotech) taken from single-use frozen aliquots (important: do not freeze/thaw these cytokines as it will significantly impair differentiation and yield as well as induce activation. It is crucial to thaw cytokines immediately before use). Throughout the differentiation of HPCs to microglia, these cells will predominantly grow non-adherently. On days 2, 4, 6, 8, and 10, add 1 mL fresh media plus freshly thawed tri-cytokine cocktail. Cytokines are diluted to the concentrations listed above before adding to conditioned medium. Do not fully remove media during the microglial differentiation as the cells secrete paracrine cytokine signals and will not properly differentiate upon removal of those. On day 12, collect 6 mL media from each 35 mm well leaving 1 mL conditioned medium on the plate. Centrifuge non-adherent cells in removed medium for 5 min at 300 x G. Aspirate medium and resuspend non-adherent cells in 1 mL fresh medium plus tri-cytokine cocktail per 35 mm well and add back to the same well which contains the 1 mL conditioned medium. Continue to supplement media (1 mL) on days 14, 16, 18, 20, 22, and 24. On day 25, centrifuge cells leaving 1 mL conditioned media per 35 mm well as on day 12. On day 25, cells should be resuspended in microglia media plus 100 ng/mL IL-34, 50 ng/mL TGFβ1, 25 ng/mL M-CSF, 100 ng/mL CD200 and 100 ng/mL CX3CL1 to further mature microglia and ensure homeostasis. On day 27, feed cells with microglia media with five cytokine cocktail (1 mL per well). On day 28 cells collected for RNA sequencing or use for transplantation or functional assays. If necessary, cells can be maintained for 1–2 additional weeks via media supplementation as above, although longer-term culture is not advised.

Total RNA was isolated from cells using RNeasy Mini kit (Qiagen). Approximately 1 million iPS-microglia cells were lysed in RLT buffer and RNA was isolated per manufacturer’s instructions with DNAse treatment (10 min) and increased spin times to maximize yield (16,000 x G for 1.5 min). RNA integrity was measured using the Bioanalyzer Agilent 2100. All libraries were prepared from samples with RNA integrity values ≥9.7. 500 ng RNA per sample was used to create RNA-seq libraries through the Illumina TruSeq mRNA stranded protocol. Each sample was then sequenced in the Illumina HiSeq 4000 platform.

RNA sequencing read integrity was verified using FastQC. BBDuk was used to trim adapters and filter out poor quality reads [31]. Reads were aligned to the GRCh.38.12 human transcriptome using Kallisto [32]. Lowly expressed genes (expression count summed over all samples < 10) were removed before differential expression analysis. Differential Expression of TPM was calculated using DESeq2 [33]. An FDR cutoff of 0.001 and fold change of at least 2 was used to determine differentially expressed genes (Additional file 3: Table S1, Additional file 4: Table S2, Additional file 5: Table S3, Additional file 6: Table S4). Visualizations were constructed in part using R in addition to the Genialis visual informatics platform (app.genialis.​com) [34]. Gene ontology analysis was performed using EnrichR.

All animal procedures were conducted in accordance with the guidelines set forth by the National Institutes of Health and the University of California, Irvine Institutional Animal Care and Use Committee. The MITRG mouse was purchased from Jackson Laboratories (stock #017711); briefly, this strain was developed on a BALB/c background containing two knockouts alleles: Rag2⁻ (Rag2^tm1.1Flv) and il2γc⁻ (Il2rg^tm1.1Flv); and three humanized knock-in alleles: hCSF-1 (Csf1^tm1(CSF1)Flv), h-IL-3/GM-CSF (Csf2/Il3^{tm1.1(CSF2,IL3)Flv}), and hTPO (Thpo^{tm1.1(TPO)Flv}). All mice were age and sex matched and group housed on a 12 h/12 h light/dark cycle with food and water ad libitum.

All mouse surgeries and use were performed in strict accordance with approved NIH and AALAC-certified institutional guidelines. Direct intracranial injections of iPS-microglia into the cortex and hippocampus were performed on adult MITRG mice. Briefly, adult mice (2–3 months old) were anesthetized under continuous isoflurane and secured to a stereotaxic frame (Kopf). Using a 30-guage needle affixed to a 10 μL Hamilton syringe, mice received 2 μL of mature iPS-microglia suspended in sterile 1X DPBS at 50,000 cells/μL at each injection site. Transplantation was conducted bilaterally in the cortex and hippocampus at the following coordinates relative to bregma: anteroposterior, − 2.06 mm; dorsoventral, − 1.75 mm (hippocampus), − 0.95 mm; mediolateral, ±1.75 mm. Cells were injected at a rate of 50,000/30s with 4 min in between injections. The needle was cleaned with consecutive washes of PBS, 70% (vol/vol) ethanol, and PBS in between hemispheres and animals. Animals were allowed to recover on heating pads before being placed in their home cages and received 2 mg/mL Acetaminophen (Mapap) diluted in water for five days. Animals were perfused 2 months following surgery with 1X PBS followed by 4% paraformaldehyde, entire brains were removed for immunohistochemistry and confocal microscopy.

Fixed half brains were first cryoprotected in a 30% sucrose and 0.05% NaN₃ solution in 1X PBS for 72 h. Tissue was then sectioned into 40 μm thick slices on a freezing microtome (Leica SM 2010R), and stored in 0.05% NaN₃ solution in 1X PBS as free floating wells. For staining, tissue was blocked for 1 h in 1X PBS, 0.2% Triton X-100, and 10% goat serum. Immediately following blocking, sections were placed in primary antibodies diluted in 1X PBS and 1% goat serum and incubated overnight on a shaker at 4 °C. Sections were labeled with combinations of anti-Ku80 (1:250; Abcam ab79220), anti-Iba1 (1:200; Wako 019–19,741), anti-P2RY12 (1:200; Sigma HPA014518) and mounted with DAPI Fluoromount (SouthernBiotech). Immunofluorescent sections were then visualized and captured using an Olympus FX1200 confocal microscope. Images represent confocal Z-stack taken with equivalent laser and detection settings.