Systematic comparison of culture media uncovers phenotypic shift of primary human microglia defined by reduced reliance to CSF1R signaling

Efforts to understand microglia function in health and diseases have been hindered by the lack of culture models that recapitulate in situ cellular properties. In recent years, the use of serum‐free media with brain‐derived growth factors (colony stimulating factor 1 receptor [CSF1R] ligands and TGF‐β1/2) have been favored for the maintenance of rodent microglia as they promote morphological features observed in situ. Here we study the functional and transcriptomic impacts of such media on human microglia (hMGL). Media formulation had little impact on microglia transcriptome assessed by RNA sequencing which was sufficient to significantly alter microglia capacity to phagocytose myelin debris and to elicit an inflammatory response to lipopolysaccharide. When compared to immediately ex vivo microglia from the same donors, the addition of fetal bovine serum to culture media, but not growth factors, was found to aid in the maintenance of key signature genes including those involved in phagocytic processes. A phenotypic shift characterized by CSF1R downregulation in culture correlated with a lack of reliance on CSF1R signaling for survival. Consequently, no improvement in cell survival was observed following culture supplementation with CSF1R ligands. Our study provides better understanding of hMGL in culture, with observations that diverge from those previously made in rodent microglia.

neurons that express C-X3-C motif chemokine ligand 1 (CX3CL1) (Harrison et al., 1998). The purinergic receptor P2RY12 aids in the detection and migration of microglia to damaged sites as first responders to injury (Haynes et al., 2006). In diseased contexts, upregulation of human leukocyte antigen (HLA) complex and inflammatory mediator expression suggests they orchestrate the dynamic process of neuroinflammation (Butovsky & Weiner, 2018).
The establishment of a unique homeostatic microglial signature is achieved through a stepwise and tissue factor-dependent differentiation process (Butovsky & Weiner, 2018). Microglia precursors arise from yolk sac hematopoiesis and colonize the brain during embryonic development, a process that is highly dependent on colony stimulating factor 1 receptor (CSF1R) (Ginhoux et al., 2010;Oosterhof et al., 2019;Rojo et al., 2019).
Ligands of this receptor include the growth factors colony stimulating factor 1 (CSF1) and interleukin-34 (IL-34), which are differentially produced in the brain: CSF1 is primarily expressed by glial cells including microglia, whereas IL-34 is mainly expressed by neurons (Kana et al., 2019). The latter is an example of tissue-specific cues driving microglia differentiation, as Il-34 LacZ/LacZ mice show selective absence of microglia and Langerhans cells, but not of other myeloid populations (Wang et al., 2012). Another signaling pathway important for microglia identity acquisition is that of transforming growth factor-beta (TGF-β) (Butovsky et al., 2014;Gosselin et al., 2017), without which microglia fail to develop properly in mouse brain (Butovsky et al., 2014). The importance of tissue-specific cues in microglia development has been further supported by studies that showed partial acquisition of microglia characteristics by bone marrow-derived monocytes upon transplantation into brain (Bennett et al., 2018;Shemer et al., 2018).
The in situ identity of microglia is rapidly lost when transferred to a cell culture environment, with downregulation of homeostatic markers such as CX3CR1 and P2RY12 (Gosselin et al., 2017). Loss of exposure to brain-derived cues upon isolation likely contributes to this phenomenon, since the transcriptomic alteration can be partly prevented by supplementing the media with CSF1/IL-34 and TGF-β1 (Butovsky et al., 2014;Gosselin et al., 2017). From these observations derive successful generation of microglia-like cells from induced pluripotent stem cells (iPSCs) using CSF1, IL-34 and TGF-β1 (Abud et al., 2017;McQuade et al., 2018). Co-culture of primary microglia or iPSC-derived microglia (iMGL) with other brain cells has been proven to be beneficial in further reproducing in situ microglia characteristics (Abud et al., 2017;Grubman et al., 2020;Timmerman et al., 2022).
An additional variable in the current methods for the culture of primary microglia is the widespread use of animal-derived serum.
Since its introduction in the 1950's (Puck et al., 1958), fetal bovine serum (FBS) has been used to promote cell survival and growth in vitro, but contains bioactive molecules that might not be present in the brain (Subbiahanadar Chelladurai et al., 2021). A serum-free alternative medium for microglia has been proposed by Bohlen et al., who discovered that the combination of CSF1R ligand, TGF-β2 and cholesterol can sustain rodent microglia survival in serum-free conditions and promote a ramified morphology (Bohlen et al., 2017). The impact of such serum-replacement strategy on primary human microglia (hMGL) had yet to be studied in a systematic manner. The present study aims to clarify the effects of serum and growth factors supplementation of culture media on hMGL survival, immune functions and signature gene expression.

| Cell culture reagents
A list of cell culture reagents used in this study is provided as supporting information (Supplementary Table S1).

| Microglia isolation and culture
Isolation of hMGL was carried out as previously described (Durafourt, et al., 2013) with slight modifications. Briefly, human brain tissues were obtained from non-malignant cases of temporal lobe epilepsy, at sites distant from suspected primary epileptic foci. Samples were digested by trypsin (Thermo Fisher Scientific) and DNase (Roche) and passed through a nylon mesh filter. Tissue homogenate was then subjected to Percoll (Sigma-Aldrich) gradient centrifugation. Further purification of microglia was performed through magnetic-activated bead sorting of CD11b + cells (Miltenyi Biotec). This isolation method results in a culture of $97% PU.1 + , $1% O4 + , and 0% glial fibrillary acidic protein-positive (GFAP + ) cells (data not shown). Cells were maintained at 37 C under a 5% CO 2 atmosphere. Unless otherwise specified, cells were cultured for 6 days prior to downstream experiments to allow cell adhesion to cultureware and stabilization of morphologies, a duration similar to those of other studies (Bohlen et al., 2017;Butovsky et al., 2014;Gosselin et al., 2017;Timmerman et al., 2022). Use of human cells was approved by the McGill University Health Centre Research Ethics Board. Microglia from at least one female donor and one male donor ranging between 2 and 68 years old were included in each experiment.

| Primary microglia
Cells were stained in phosphate buffered saline (PBS) containing 1 μg/ml propidium iodide (PI) and 5 μg/ml Hoechst 33342 (Thermo Fisher Scientific). The average number of PI-live cells per condition was determined using a CellInsight CX7 high content screening platform (Thermo Fisher Scientific). All conditions were assessed in triplicate.

| iPSC-derived microglia
Due to the loose adherence of iMGL to cell culture plates (Abud et al., 2017;McQuade et al., 2018), the quantification of PI-live cells was carried out using flow cytometry (Attune™ Nxt Flow Cytometer; Thermo Fisher Scientific). Doublets and debris were excluded from the analysis using appropriate forward/side scatter profiles.

| Immunocytochemistry
Cells were fixed in 4% formaldehyde and permeabilized/blocked using PBS with 3% goat serum and 0.2% triton X-100. Cells were incubated at 4 C overnight with primary antibodies ( (Bourgey et al., 2019) was used to align the raw files and quantify the read counts. Briefly, raw fastq files were aligned to the GRCh38 genome reference using STAR aligner (Dobin et al., 2012) with default parameters and raw reads were quantified using HTseq count (Anders et al., 2015). Transcript per million (TPM) values are provided as supporting information (Supplementary Table S2).

| Differential expression gene analysis
Read counts were used for differential expression gene (DEG) analysis with the edgeR package (Robinson et al., 2009). DEGs were identified using an adjusted p-value cutoff of .05.

| Enrichment analysis
Gene ontology (GO) enrichment analyses were performed using the web-based tool offered by the Gene Ontology Consortium. The Enrichr web tool (Chen et al., 2013) was used for pathway enrichment analysis using Molecular Signatures Database (MSigDB) 2020 hallmark gene sets.

| Data visualization
Heatmaps were generated using the Python data visualization library Seaborn. Scatter plots and volcano plots were generated using the Python data visualization library Matplotlib. Principal component analysis (PCA) was carried out using the Python module sklearn and visualized using Matplotlib. Venn diagrams were drawn using Venny 2.1.0 (BioinfoGP). Histograms were generated using GraphPad Prism 8.0 software.

| Conversion of mouse genes to human orthologs
Lists of mouse genes were converted to lists of human ortholog genes using the HUGO Gene Nomenclature Committee Comparison of Orthology Predictions search tool (Eyre et al., 2007).

| Phagocytosis assay
Human myelin debris (Healy et al., 2016), recombinant α-synuclein fibrils (Del Cid Pellitero et al., 2019;Maneca et al., 2022) and immunoglobulin G (IgG)-opsonized human red blood cells  were labeled with pHrodo Green™ STP ester (Thermo Fisher Scientific) and were used at the following respective concentrations: 1 μM, 15 μg/ml and 500,000 cells/ml. Bioparticles of pHrodo Green™labeled Escherichia coli (E. coli) were purchased from Thermo Fisher Scientific and used at a concentration of 25 μg/ml. After a two-hour incubation with the labeled phagocytosis targets, cells were counterstained with Hoechst 33342 (5 μg/ml) and total green fluorescence intensity per cell was quantified using a CellInsight CX7 high content screening platform. All conditions were assessed in triplicate.
Resulting iMGL (day 28 and older) showed a ramified morphology

| Fetal astrocyte isolation and culture
Human astrocytes were isolated from second trimester fetal tissue (17-23 weeks of gestation) obtained from the University of Washington Birth Defects Research Laboratory and cultured as previously described (Kieran et al., 2022). Supernatants of cells maintained in normoxic condition or in 1% atmospheric oxygen in the presence of IL-1β (Thermo Fisher Scientific) for 24 hours were collected as described in (Kieran et al., 2022).

| Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
RNA was extracted using a RNeasy mini kit (Qiagen) and reverse transcription was performed using Moloney murine leukemia virus reverse transcriptase (Thermo Fisher Scientific). Real-time polymerase chain reaction (PCR) was performed using TaqMan assays (Thermo Fisher Scientific) on a QuantStudio™ 5 real-time PCR system (Thermo Fisher Scientific).

| Statistical analyses
Statistical analyses were performed using GraphPad Prism 8.0 software. Primary cells obtained from independent donors were considered biological replicates. iMGL generated at different points in time were considered biological replicates. A t-test was used to compare the mean of two groups of data. A one-way analysis of variance (ANOVA) was used to compare the mean of three or more groups of data. When the assumptions of a t-test (homoscedasticity and normal distribution) were not met, a Mann Whitney test was employed instead. When the assumptions of a one-way ANOVA (homoscedasticity and normal distribution) were not met, a Kruskal-Wallis test was used instead. p-values were adjusted using Sidak's post hoc test and Dunn's post hoc test following a one-way ANOVA and a Kruskal-Wallis test, respectively. A mixed-effects analysis followed by Tukey's post hoc test was used for the comparison of multiple groups of data  (Abud et al., 2017;Durafourt et al., 2013;Healy et al., 2016). Based on findings related to the culture of rodent microglia (Bohlen et al., 2017), we first sought to replace FBS using a combination of TGF-β1, IL-34, CSF1, CD200, CX3CL1, and cholesterol, which we abbreviate to TIC 4 . In addition to the growth factors and cholesterol, both CD200 and CX3CL1 were added as brain-derived cues that are known to modulate microglia function (Abud et al., 2017;Harrison et al., 1998;Hoek et al., 2000). When hMGL survival was assessed on the sixth day of culture using PI staining, a significant reduction in survival was observed in MEM compared to MEM + 5% FBS (Figure 1a,b).
We hypothesized that the serum-free media used for the generation and maintenance of iMGL might be able to support the survival of

| FBS enhances the phagocytic activities of human microglia
Because culture media had differential effect on the expression of MERTK, PROS1, and GAS6 with known importance in microglia phagocytic processes (Healy et al., 2016), the impact of media formulation on myelin phagocytosis by hMGL was next investigated. It was observed that hMGL have greater phagocytic capacity in Abud medium +5% FBS, compared to any other media formulations (Figure 4a,b). Relative to Abud medium, the addition of FBS increased myelin uptake $7.5-fold (Figure 4a,b). This effect of FBS on enhancing myelin internalization was maintained even when FBS was removed immediately prior to the phagocytosis assay (Figure 4c), indicating that this effect is not due to the presence of opsonizing molecules in FBS.
F I G U R E 3 Differential effect of media formulation on microglia key signature genes expression. hMGL were cultured for 6 days in different media or not (ex vivo) following isolation From brain tissues. hMGL from four donors were used for (a, c-e). (a) Heatmap showing the expression of microglia signature genes identified by Butovsky et al. (Butovsky et al., 2014). Kruskal-Wallis tests were performed, followed by Dunn's post hoc test. *p < .05, **p < .01 compared to ex vivo hMGL. (b) Flow cytometry assessment of MerTK and P2RY12 cell surface expression on hMGL cultured in Abud medium supplemented or not with TIC 4 or 5% FBS (n = 1 donor). (c) Expression of select microglia markers in hMGL cultured in Abud medium supplemented with TIC 4 or 5% FBS compared to Abud medium. One-way ANOVA were performed, followed by Dunnett's post hoc tests. Mean +/À SEM. *p < .05. (c) Pie charts presenting the proportion of DEGs in Abud medium supplemented with TIC 4 or 5% FBS, compared to Abud medium. Results are presented as percentages of the total number of genes. (d) Volcano plot of the transcriptomic change induced by 5% FBS supplementation of Abud medium. Light gray and dark gray dots depict genes for which p > .05 and p < .05 by DEG analysis, respectively. Fuchsia dots depict microglia signature genes that were identified by Patir et al. (Patir et al., 2019). (f ) qRT-PCR assessment of key microglia marker expression in hMGL culture in Abud medium with or without 5% FBS. Data are presented as row z-score of 2 ÀΔCt values obtained using GAPDH and YWHAZ as internal controls. Mean of n = 3 donors. DEG, differential expression gene; FBS, fetal bovine serum; hMGL, human microglia; qRT-PCR, quantitative reverse transcription polymerase chain reaction When the expression of genes related to phagocytosis was evaluated, FBS was observed to significantly increase the expression of genes encoding MerTK (Healy et al., 2016) and CD36 (Grajchen et al., 2020) with known roles in myelin phagocytosis (Figure 4d). This promoting effect of FBS on myelin uptake could also be confirmed using iMGL (Supplementary Figure S6).
Similar to its effect on myelin uptake, FBS was also observed to increase the phagocytosis of α-synuclein fibrils, which was the highest in Abud medium +5% FBS. FBS also increased E. coli phagocytosis by hMGL, but E. coli uptake was the highest in MEM + 5% FBS ( Figure 4e). In contrast, phagocytosis of IgG-opsonized red blood cells was unaffected by FBS ( Figure 4e). Altogether, data suggest an F I G U R E 4 Effect of culture media on the phagocytic activity of hMGL. (a-d) hMGL were cultured for 6 days in various culture media and exposed or not for 2 hours to pHrodo Green™-labeled myelin debris. (a) Fluorescent images of internalized myelin (green) in cells stained with PI (red) and Hoechst 33342 (blue). Scale bar = 1 mm. (b) Quantification of myelin uptake normalized to the number of viable cells. A one-way ANOVA with Sidak's post hoc test was performed. Mean +/À SEM of n = 5 donors, *p < .05, **p < .01. (c) Quantification of myelin uptake in cells exposed or not to FBS during myelin challenge. A Kruskal-Wallis test with Dunn's post hoc test was performed. Mean +/À SEM of n = 4 donors, ns = non-significant, *p < .05. (d) Transcriptional expression of membrane proteins involved in phagocytosis in hMGL cultured in Abud medium +5% FBS compared to Abud medium (mean of n = 4 donors). Red bar denotes genes for which p < .05 in DEG analysis. (E) hMGL were cultured for 6 days in various culture media and exposed or not for 2 hours to pHrodo Green™-labeled α-synuclein fibrils, E. coli or IgG-opsonized red blood cells (IgG-RBC). Green fluorescence intensity per live cell was quantified. One-way ANOVA with Sidak's post hoc test was performed. Mean +/À SEM of n = 5 donors, *p < .05. DEG, differential expression gene; FBS, fetal bovine serum; hMGL, human microglia Effect of culture media on the inflammatory activity of microglia. (a and b) hMGL were cultured for 6 days in various culture media and exposed for 24 hours to vehicle or 100 ng/ml LPS. (a) Heatmap presenting the LPS-induced secretion of cytokines relative to vehicle treatment. A mixed-effects analysis followed by Tukey's post hoc test was used. Mean of n = 5 donors, *p < .05, **p < .01, ***p < .001. (b) Quantification of cytokine concentrations in cell supernatants. A Kruskal-Wallis test followed by Dunn's post hoc test was performed. Mean +/À SEM of n = 5 donors, *p < .05, **p < .01. (c) iMGL were exposed for 24 hours to vehicle, 100 ng/ml LPS, 250 ng/ml R-FSL-1 or 100 ng/ml Pam 3 CSK 4 and cytokine concentrations in cell supernatants were measured. A Kruskal-Wallis test followed by Dunn's post hoc test was performed. Mean +/À SEM of n = 6 biological replicates, *p < .05, **p < .01, ***p < .001 vs vehicle. (d) iMGL were exposed to vehicle or 100 ng/ml LPS for 24 hours, with (w/) or without (w/o) the presence of B27. A Kruskal-Wallis test followed by Dunn's post hoc test was performed. Mean +/À SEM of n = 4 biological replicates obtained from three iPSC lines (DYR0100, GW25256 and 3450), *p < .05. (E) iMGL were cultured with or without the presence of 5% FBS for 48 hours, and exposed to vehicle or 100 ng/ml LPS for 24 hours. Cytokine concentrations in cell supernatants were measured. A Kruskal-Wallis test followed by Dunn's post hoc test was performed. Mean +/À SEM of n = 6 biological replicates obtained from three iPSC lines (DYR0100, GW25256 and 3450), *p < .05, **p < .01. FBS, fetal bovine serum; hMGL, human microglia; iGML, iPSC-derived microglia F I G U R E 6 CSF1R-dependence of hMGL survival in vitro. hMGL were cultured for 6 days in different media or not (ex vivo) following isolation from brain tissues. (a) RNAseq assessment of CSF1R, CSF1, and IL-34 expression in hMGL cultured for 6 days in Abud medium compared to ex vivo hMGL obtained from the same donors. Paired t-tests were performed. n = 4 donors, *p < .05, ***p < .001. (b) Measurement of CSF1 and IL-34 in the supernatants of hMGL cultured for 6 days in Abud medium compared to unconditioned media (mean +/À SEM of n = 3 donors). (c) Immunoblotting of CSF1R and GAPDH in hMGL and iMGL cell lysates. (d) Quantification of CSF1R relative to GAPDH band intensity. Mean +/À SEM of n = 3 biological replicates. (e) Viability of hMGL cultured in Abud medium and iMGL treated every other day with PLX3397 for 6 days. Mean +/À SEM of n = 3 donors. Dash lines represent regression curves. (f) Viability of hMGL cultured in Abud medium and iMGL treated every other day with PLX3397 or cycloheximide for 6 days. A one-way ANOVA followed by Dunnett's post hoc test was performed. Mean +/À SEM of n = 3 biological replicates, ***p < .001. (g) Heatmaps showing the expression of ortholog genes identified by Zhan et al. (Zhan et al., 2020) to be high (left heatmap) or low (right heatmap) in PLX res compared to homeostatic mouse microglia (p < .01, jlog(fold change)j > 1). Genes for which log 2 (TPM + 1) < 2 in hMGL were excluded from the analysis. CSF1R, colony stimulating factor 1 receptor; hMGL, human microglia; iGML, iPSC-derived microglia enhancing effect of FBS on phagocytosis, with the exception of Fc-gamma receptor-mediated uptake.

| Abud medium dampens the responsiveness of microglia to LPS
Another important function of microglia is to elicit an inflammatory response to pathogen and danger-associated molecular patterns. The secretion of TNF-α, IL-6, IL-10, and IL-1β was measured following hMGL treatment with the widely used inflammatory stimulus LPS.
Secretion of these cytokines was observed to be dampened in Abud medium with or without TIC 4 /FBS compared to MEM + 5% FBS, with the lowest secretory activities observed in Abud medium +5% FBS (Figure 5a,b). Accordingly, GO term analysis suggested a difference in the expression of genes involved in immune system processes and response to stimuli in hMGL cultured in MEM + 5% FBS compared to all other media (Supplementary Table S4). Altogether, these data suggest that Abud medium has an immunosuppressive effect.
Because Abud medium was initially developed for the generation of iMGL (Abud et al., 2017), the inflammatory response of iMGL (generated and maintained in Abud medium) to LPS was next studied.
Assessment of TNF-α, IL-6, IL-10, and IL-1β secretion revealed a lack of response of iMGL to the TLR4 agonist LPS (Figure 5c). iMGL responded to the TLR2/6 agonist R-FSL-1 and the TLR2/1 agonist Pam 3 CSK 4 by secreting TNF-α, IL-6, and IL-10 ( Figure 5c). This suggests the immunosuppressive effect of Abud medium is restricted to the TLR4 pathway. We suspected B27 supplement, which contains corticosterone, in Abud medium to be responsible for this effect.
Treatment of iMGL with LPS in the absence of B27 supplementation resulted in significantly higher secretion of TNF-α, IL-6, and IL-1β ( Figure 5d). Intriguingly, FBS supplementation of culture media also enhanced the response of iMGL to LPS, resulting in significantly higher secretion of TNF-α, IL-6, and IL-10 ( Figure 5e).

| In vitro hMGL undergo a phenotypic shift characterized by resistance to CSF1R inhibition
The maintenance of microglia population in the mature brain depends on CSF1R signaling, as both pharmacological inhibition of CSF1R (Elmore et al., 2014;Zhan et al., 2020) Figure S7A). In comparison, human astrocytes subjected to stress (hypoxia and IL-1β treatment) secreted $263 pg/ml CSF1 (Supplementary Figure S7B). In order to verify whether the endogenous level of CSF1 is sufficient for in vitro hMGL to sustain their survival in an auto/paracrine manner, cells were treated with the CSF1R inhibitor PLX3397. hMGL and iMGL both expressed CSF1R at the protein level (Figure 6c,d), however, only iMGL were susceptible to PLX3397-induced cell death upon a six-day chronic treatment (Figure 6f). Both hMGL and iMGL were susceptible to the toxicity of the protein synthesis inhibitor cycloheximide, which significantly decreased viability (Figure 6f). These data indicate that hMGL survival in serum-free culture is independent of CSF1R signaling.
In 2020, Zhan et al. identified a population of microglia in mouse brain with high expression of Lgals3 (encoding the protein galectin-3, also called MAC2) and low expression of Csf1r which are resistant to the CSF1R inhibitor PLX5622 (termed PLX res ). Gene set enrichment analysis revealed that this population differs in "TNF-α signaling via nuclear factor-kappa B," "oxidative phosphorylation," and "mTORC1 signaling" compared to other microglia populations (Zhan et al., 2020), similarly to what is observed over the ex vivo to in vitro transition of

| DISCUSSION
Microglia are involved in a plethora of processes in both health and disease. The secretion of trophic factors, as well as phagocytic clearance of apoptotic cells and redundant synapses are critical for proper brain network formation (Schafer & Stevens, 2015). Immunohistochemical staining and brain imaging of patients' brains have evidenced an inflammatory activation of microglia in several neuroinflammatory and neurodegenerative diseases (Butovsky & Weiner, 2018). In addition, genetic associations have been established between genes enriched in microglia and the risk of developing diseases such as Alzheimer's disease (Lambert et al., 2013). It is imperative that the in situ microglia transcriptomic signature is faithfully replicated in vitro to facilitate the accurate modeling of microglia function, to expand our fundamental understanding of these cells, and to accelerate the therapeutic targeting of microglia for the treatment of neurologic diseases. In this study we investigated culture media formulation as a variable in replicating hMGL in situ characteristics, with a focus on how FBS and brain-derived cues could impact hMGL transcriptomics and immune activities.
By comparing the transcriptome of in vitro hMGL to ex vivo hMGL from the same donors, an unexpected beneficial effect of FBS as a culture supplement was revealed in this study. Not only did FBS supplementation improve cell survival and enhance phagocytic capacity as previously observed in rodent microglia (Bohlen et al., 2017;Fourgeaud et al., 2016), it also helped maintain expression of microglia signature genes in vitro. Microglia pool in the postnatal brain is maintained through a balance between cell death and self-renewal with a slow turnover (Askew et al., 2017;Réu et al., 2017). Studies using pharmacological inhibitors (PLX3397 or PLX5622) demonstrated that this maintenance is dependent on CSF1R signaling in mice (Elmore et al., 2014;Zhan et al., 2020). Here it was observed that hMGL in culture are resistant to pharmacological inhibition of CSF1R. Cells showed decreased expression of CSF1R compared to the ex vivo state and shared transcriptomic similarities with a PLX5622-resistant microglia population identified in the mouse brain (Zhan et al., 2020). It is unlikely that selective survival of a pre-existing CSF1R-independent population of hMGL is expanding in culture, as addition of TIC 4 did not increase cell yield nor did it prevent a decrease in CSF1R expression. Rather, it is likely that hMGL in culture adopt a cellular phenotype characterized by a reduced reliance on signaling through the CSF1R. This has profound implications in the study of CSF1R as a therapeutic target.
Adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP) is a fatal disease caused by inactivating mutations of CSF1R and is characterized by low number, dysmorphic microglia in the brain. Aside from its pro-survival and differentiation effect, the mitogenic effect of CSF1R signaling has been linked to exacerbated inflammation in neurological diseases such as Alzheimer's disease (Gomez-Nicola & Perry, 2016;Hu et al., 2021) and multiple sclerosis (Hagan et al., 2020). Titrated modulation of microglia density through CSF1R targeting has therefore been tested in animal models as a promising therapeutic strategy (Green et al., 2020).  (Bohlen et al., 2017). On the contrary, hMGL could be successfully cultured in serum-free medium.
Addition of TIC 4 did not further increase cell viability whereas FBS did. The discrepancy in our observations could stem from interspecies differences and/or differences in basal media used. Several crucial interspecies differences have so far been detected between rodent and primate microglia. Differences not only relate to their transcriptome, but also in the way cells react to insults, such as morphological alterations and nitric oxide synthesis in response to interferon/LPS treatment (Geirsdottir et al., 2019;Gosselin et al., 2017;Healy et al., 2019;Owen et al., 2017). Survival requirements in growth factors, hormones and/or metabolites might represent yet another distinctive difference between microglia of various species. Supporting this notion, Bohlen et al.'s medium + TIC fails to sustain the initial adhesion and survival of rhesus monkey microglia in culture (Timmerman et al., 2022).
When the transcriptomic difference of in vitro versus ex vivo murine microglia was first described, it was shown that TGF-β1 partly prevents the culture-induced loss of microglia signature gene expression (Butovsky et al., 2014). Robust increases in Gpr34, P2ry12, Olfml3, and Sall1, among other genes, was observed in TGF-β1-supplemented culture (Butovsky et al., 2014). The effect of TGF-β1 supplementation on microglia transcriptome was observed to be much more modest in studies using hMGL (Gosselin et al., 2017) and rhesus monkey microglia, with the latter even showing a TGF-β1-induced decrease in OLFML3 expression (Timmerman et al., 2022). Similarly, TIC 4 was not observed to have any significant effect on the transcriptome of hMGL cultured in Abud medium. FBS was surprisingly better than TIC 4 at maintaining key gene expression in hMGL, such as that of MERTK and P2RY12.
With over 1800 proteins and 4000 metabolites (Subbiahanadar Chelladurai et al., 2021), determining which component(s) of FBS is responsible for the maintenance of microglia core genes would unfortunately require tremendous time and resources. However, our data suggest small, non-lipophilic and heat-resistant molecule(s) such as small proteins, amino acids, nucleotides, minerals, or salts could be mediating the effect of FBS on key microglia marker genes. We found no advantage of using serum-free media over the conventional MEM + 5% FBS, and no media formulation tested in this study fully recapitulated the ex vivo transcriptome of hMGL.
Specific media formulations did have a profound effect on functional readouts commonly used when assessing microglia activity, raising the question as to whether the selection of culture media should be tailored depending on the specific biological processes under investigation.
Recent efforts to more accurately model human brain tissue has led to the optimization of co-culture and higher dimensional culture systems, circumventing the problem of environment deprivationinduced changes in microglia transcriptome (Abud et al., 2017;Grubman et al., 2020;Timmerman et al., 2022). Nevertheless, microglia monoculture remains essential for the study of cell autonomous effects of genetic manipulations and the testing of potential therapeutics. This study revealed the importance of carefully choosing appropriate culture media when evaluating key functional aspects of microglia. This work also highlights the usefulness of complementary models such as iMGL to study certain molecular pathways (e.g., CSF1R signaling pathway) that are inactive/absent in cultured hMGL.