The human bone marrow harbors a CD45− CD11B+ cell progenitor permitting rapid microglia‐like cell derivative approaches

Abstract Microglia, the immune sentinel of the central nervous system (CNS), are generated from yolk sac erythromyeloid progenitors that populate the developing CNS. Interestingly, a specific type of bone marrow‐derived monocyte is able to express a yolk sac microglial signature and populate CNS in disease. Here we have examined human bone marrow (hBM) in an attempt to identify novel cell sources for generating microglia‐like cells to use in cell‐based therapies and in vitro modeling. We demonstrate that hBM stroma harbors a progenitor cell that we name stromal microglial progenitor (STR‐MP). STR‐MP single‐cell gene analysis revealed the expression of the consensus genetic microglial signature and microglial‐specific genes present in development and CNS pathologies. STR‐MPs can be expanded and generate microglia‐like cells in vitro, which we name stromal microglia (STR‐M). STR‐M cells show phagocytic ability, classically activate, and survive and phagocyte in human brain tissue. Thus, our results reveal that hBM harbors a source of microglia‐like precursors that can be used in patient‐centered fast derivative approaches.

controlling immunomodulation of the CNS environment 7 and also possesses a potential in cell therapy applications as microglia exhibit neuroprotection in ischemia 8 and are scavengers for amyloid β (Aβ) in Alzheimer's disease. 9 Presently, the potential therapeutic application of human microglia-like cells is limited by the lack of CNS microglial sources and fast and safe stem cell approaches for deriving microglia.
Current methods of generating induced pluripotent stem cells (iPSCs) that subsequently can be differentiated to microglia-like cells are timeconsuming (≥4 months), have low efficiency, and are potentially associated with tumor risks when used for transplantation purposes. 10,11 The use of embryonic stem cells suffers also from ethical concerns and limited access, 12 whereas approaches using microglia-like cells derived from bone marrow monocytes bring functional differences that are associated to their unique roles in lesion formation in the curse of CNS degeneration. [13][14][15] Interestingly, a specific type of murine bone marrow-derived microglia-like cells has been reported to be able to populate CNS in disease and to express a yolk sac microglial signature. 16 In the present work, we have examined human bone marrow (hBM) in an attempt to identify the presence of a cell source for generating human microglia-like cells in fast and reliable derivative approaches.

| RNA sequencing extraction and single-cell gene analysis
Single HLA-DR − CD14 − CD19 − CD34 − CD45 − CD11b + cells were sorted into 4 μL lysis buffer. 19 Preamplification was performed using TaqMan primers and Taq/SSIII reaction mix (Invitrogen, Thermo Fisher Scientific). Linearity control and negative controls were included in each plate. Preamplification was performed according to a

Significance statement
The authors believe that the discovery of an intermediate specified myeloid microglia progenitor will open the way for developing novel patient-centered approaches and in vitro modeling. The use of stromal microglial progenitor as the starting point in derivative microglia-like cell approaches, instead of a somatic cell to first generate human pluripotent stem cells, reduces the risk of teratoma formation and the overall waiting time of current microglia. Thus, these results will be of broad general interest to a number of high-profile areas of research, including stem cells and in vitro modeling, neurodegeneration, neuropsychiatry, brain tumors, bone marrow stromal cells, hematopoietic system, and human ontogeny.

| Data analysis
Quantitative PCR data were analyzed in the BiomarkHD analysis software (quality threshold 0.60, automatic global cycle threshold, and linear derivative baseline correction). After excluding controls, data were preprocessed in SCExV 20 where cycle threshold values were inverted, normalized to median expression of each cell, and z-score normalized.
Unsupervised clustering analysis was performed using random forest clustering 21 and principal component analysis, and correlations were measured using the Spearman rank method.

| hBM MSCs and nonhematopoietic CD11b + cell expansion in serum-containing and serum-free conditions
For serum-containing experiments: When they reached $90% confluence in T75 plastic flasks (Thermo Fisher Scientific), cells were kept in their StemMACS MSC Expansion Media (MiltenyBiotec) supplemented with Primocin (1:1000, InvivoGen) for 6 days, mimicking the serum-free medium experiments (described in the paragraph below). Next, preconditioned cells were seeded in 24 multiwells onto laminin (Sigma-Aldrich) and polyornithine (Advanced BioMatrix, Carlsbad, California) coated coverslips at a density of 80 000 to 100 000 cells per well. Next day after the seeding, cells were exposed to (a) only this expansion medium, BM; (b) BM supplemented with brainderived neurotrophic factor (BDNF) and human neutrophin 3 (NT3) (both from PeproTech, Rocky Hill, NJ; final concentration of 10 ng/mL) added fresh every second day for as long as required by the time points (2 or 4 weeks; Figure S1B); or (c) BM supplemented with macrophage colony-stimulating factor (M-CSF; 25 ng/mL; ThermoFisher Scientific), interleukin (IL)-34 (100 ng/mL; PeproTech), and transforming growth factor (TGF) β-1 (50 ng/mL; Militenyi) added fresh every second day for as long as required by the time points (2 or 4 weeks; Figure S1B).

| Counting
Quantification analysis was done in an automated image analysis using the software Thermo Scientific HCS Studio: Cellomics Scan, version 6.6.
A manual counting was added to exclude possible artifacts. The threshold levels were set to an internal control that contained only secondary antibodies. The high intensity levels were calculated per well as equal to or higher than the mean for the intensities plus twice the SD.
Quantification of cells derived from hBM MSCs was done manually after taking images at ×20 magnification using the filter for

| Microglial phagocytosis assay
Fluorescence latex beads of 1 μm diameter (Sigma-Aldrich) were first preopsinized for 1 hour at 37 C at a ratio of 1:5 with fetal bovine serum (Gibco). Preopsinized beads were subsequently diluted in the different serum-containing and serum-free culture conditions for a final concentration of latex beads of 0.01% ( Figure S2A). Cells were in contact with medium and preopsonized latex beads for 1 hour at 37 C and subsequently fixed with cold 4% PFA solution for a posterior immunocytochemistry study for CD11b and Iba1 expression. nonhematopoietic CD11b + cells were seeded in a 96 multiwell, at a density of 50 000 cells per well, and, after being exposed to either NM + CK or NM + NT for 6 days, media containing the Aβ 42 oligomers and cells were incubated for 1 hour at 37 C, respectively, and fixed in cold 4% PFA for posterior analysis.

| LPS induced activation
Presorted nonhematopoietic CD11b + cells were seeded at a density of 80 to 100 × 10 3 cells per well in 24 multiwell plates and treated as depicted in the scheme ( Figure S1B). Cell cultures were incubated for 24 hours with 100 ng/mL of LPS from Escherichia coli O111:B4 (Sigma-Aldrich) added to the different serum-free culture conditions ( Figure S2B). After 24-hour incubation, supernatants were collected, frozen in dry ice, and stored at −80 C for a subsequent Meso Scale analysis.

| Meso Scale
The release of proinflammatory CKs in culture media upon LPS activation was measured using Meso Scale (Meso Scale Diagnostics, Rockville, Maryland) plates with the proinflammatory panels for interferon-γ, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12, and tumor necrosis factor-α. The plates were developed using the 4× reading buffer diluted to a factor of 1× with distilled water, and the plates were read using the QuickPlex Q120 reader from Meso Scale. The detection ranges of the different CKs measured were as follows: IL-1β Forty-eight hours after infection, cultures were treated with puromycin at a final concentration of 10 μg/mL for 5 days.

| Coculture of human nonhematopoietic CD11b + cells, stromal cells, and adult human cortex organotypic slice cultures
Adult human cortical tissue from three patients with glioma was obtained by informed consent from the patients according to guidelines approved by the Lund-Malmö Ethical Committee. All three patients were male and aged 52 to 66 years, and tumors were located in the frontotemporal region in these cases. Tumor pathology varied: isocitrate dehydrogenase (IDH) wild-type primary glioblastoma, IDH wild-type secondary glioblastoma, and IDH-mutated diffuse astrocytoma. In all cases, tissue for culturing was collected from magnetic res- thick tissue slices were cut. The sliced tissue was transferred to prepared six-well plates with equilibrated culture medium and cultured at 37 C in a humidified atmosphere of 5% CO 2 . Organotypic tissue slices were cultured on Alvetex Strata six-well membrane inserts (AMS Biotechnology, Milton, UK) that had been presoaked in organotypic culture medium (neurobasal, 2% B27 without vitamin A, 0.5% Glutamax, and 10 μg/mL gentamicin; all from Thermo Fisher Scientific) in six-well plates, for a total of 3 weeks, and every second day the inserts were transferred to a new plate with 6 mL fresh and equilibrated medium in similar fashion as previously reported by Miskinyte et al. 23 At day 7, one well per medium was selected and grafted with GFP + human predifferentiated nonhematopoietic CD11b + as part of the hBM MSC cultures. Briefly, 10 μL of a suspension mix (1 000 000 GFP + hBM MSCs) was collected into a cold glass capillary and injected as small drops stabbing the semidry slices at various sites in a randomized fashion trying to cover most of the surface. Slices were transplanted with 10 μL of suspension leading to approximately 300 000 cells per slice. Additional medium was added 30 minutes later to fully immerse the organotypic culture. The medium was changed every other day during the week, and coculture was maintained for 3 weeks before fixation. Brain tissue was fixed in 4% PFA solution, washed three times with PBS, and submerged in sucrose at an increase of 10% each day for 3 days, with the final day being 30%, and subsequently stored in antifreeze solution (30% ethylene glycol and 30% glycerol, both from VWR International, West Chester, Pennsylvania) in 0.012 M NaH 2 PO 4 ÁH 2 O and 0.031 M Na 2 HPO 4 Á2H 2 O (both from Sigma-Aldrich) at −20 C ( Figure S2C).

| iDISCO
Immunolabeling-enabled three-dimensional imaging of solvent-cleared organs (iDISCO) was performed using the readily available protocol at https://idisco.info/idisco-protocol/ with slight modifications. In brief, samples were fixed with 4% PFA and prepared using the methanol pretreatment; samples were then incubated in permeabilization solution

| Functional assays in the cocultures and histology
For the functional assays in the cocultures, adult human cortical tissue from one patient with glioma was obtained, sliced, and brought on the inserts as described in the previous section. Three to 5 days after slicing, a total of 30 000 STR-MP cells per slice previously sorted were labeled with the fluorescent vital dye CellTracker Red CMTPX (Thermo Fisher Scientific) according to the manufacturer's instructions and were grafted the same day in similar fashion as described above.
Cocultures were kept for 6 days in NM + CK (prepared as described above), and medium was changed every 3 days to a previously equili-

| Statistical analysis
All statistical analysis was performed using Graph Pad Prism version 7.0c. Data were assumed to fit a normal distribution, and statistical comparisons were made using unpaired, two-tailed t tests for comparisons of two means between untreated groups and/or for comparisons between different time points within a condition; onetailed unpaired t test when comparing two groups exposed to a treatment, that is, LPS 24 ; and, depending on the number of variables, one-or two-way analysis of variance (ANOVA) followed by Tukey's test when comparing multiple groups. A value of P < .05 was considered statistically significant. Data are presented as means ± SEM.

| Identifying STR-MP cells
We hypothesized that, because the adult hematopoietic system derives from the hematopoietic stem cells originated during the definitive hematopoiesis, 25,26 if EMPs or EMP-like cells are present in the adult hBM, they are likely to be part of the stromal fraction. In order to target the stromal fraction and avoid genetic drift, selection bias, and enrichment of specific populations in the samples, fresh isolated hBM was analyzed using flow cytometry ( Figure 1A), initially to negatively select monocyte/macrophages. Within the remaining fraction, CD14, CD19, CD34, and HLA-DR cell surface markers were used to remove hematopoietic and endothelial cells in the bone marrow extractions ( Figure 1B). The absence of the above-mentioned markers together with CD45 was used to help to identify the hBM stromal compartment 27 from which CD11b + cells were selected, with HLA- 0.04% ± 0.01% of hBM cells (n = 5; Figure 1B). Of the STR-MP cells, 97.9% ± 1.1% were negative for CD73 ( Figure 1C), 99.8% ± 0.1% were negative for CD90 ( Figure 1D), 97.7% ± 1.1% were negative for CD105 ( Figure 1E), and 82.7% ± 9.2% were negative for HLA-ABC ( Figure 1F), indicating the presence of an expected CD45 − CD11b + human microglia-like/myeloid progenitor profile 28,29 in the stromal fraction that does not share the mesenchymal molecular signature. 27 Sorted CD45 − CD11b + stromal cells were subjected to single-cell analysis for essential, disease-involved, and developmental microglialspecific genes, as well as for the consensus microglial signature. The (CSF1R), which are essential for regulation of microglial cell development, 1,30 and the microglia-enriched protein TMEM119 and beta-hexosaminidase subunit beta (HEXB), indicating an early commitment toward microglial fate. 31,32 The STR-MP single-cell gene profile   showed a canonical microglial gene expression pattern, including   TREM2, P2RY12, CD33, GPR34, GPR56, C1Q, CABLES1, BHLHE41,   TMEM119, TGFBR1, ENTPD1, ITGB2, ITGAM, AIF, IRF8, ADORA3, and PPARD. 30,31,[33][34][35] Because the vast majority of the microglial core gene signature is shared between immature progenitors and more mature microglial cells, 36 our data suggest that a myeloid/microglialike progenitor cell compartment is present in the hBM stromal compartment, which we refer to as the STR-MP compartment. Two distinct populations defined by their Iba1 immunoreactivity and nuclear area size were observed under the different serum-containing conditions ( Figure S1D and Table S1A). Iba1 + CD11b + cell nuclei area was restricted to 100 to 400 squared pixels   Figure 2E).

| Culturing STR-MP cells
We next addressed the influence of in vitro long-term culturing on the STR-MP cellular characteristics, as functionality and differentiation of cultured cells can be affected and are important to consider for stem cell-based therapies. 39 Our results show that the use of NT in serum-containing medium significantly increased Iba1 + CD11b + cell numbers when compared with the CK-supplemented medium at 3 weeks ( Figure 2F and one-tailed Student's t test, P ≤ .05; n = 2 donors; passage numbers 7-8) but were significantly reduced to zero at 5 weeks in all conditions ( Figure 2G and Table S2A- Long-term cultured Iba1 + CD11b-high cells were also able to survive under serum-free condition ( Figure 2H-K and  Figure 2H and Table S2H) F I G U R E 2 STR-MP CD11b + cells grow as part of mesenchymal stromal cultures, which act as a feeder cell layer. A, Twelve hours after sorting, sorted cells were seen attaching to plastic (a) showing a cubic cellular shape (denoted by an asterisk; scale bar, 20 μm; magnification in the upper frame to the right (a 0 ); scale bar, 10 μm). In order to force a morphological change, cells were exposed to a serum-free medium supplemented with neurotrophins to promote survival of human bone marrow-derived mesenchymal stromal cells. After 2 days of serum-free conditions, different ramified morphologies could be observed (b) (asterisk; scale bars, 20 μm; n = 2 donors, two repetitions). B, Cells were able to survive for 5 days on plastic and in contact with stromal cells. After 1 week of serum-free condition exposure, cells were fixed in 4% and analyzed for Iba1 and TMEM119 expression. The arrow denotes a small GFP − CD11b + cell denoted by Hoechst and immunoreactive toward Iba1 and TMEM119. The asterisk shows a small GFP − CD11b + cell immunoreactive to Iba1 and negative to TMEM119 expression (scale bar, 20 μm; n = 2 donors, two repetitions). C, Analysis of the small nuclei compartment. The condition BM + CK showed a significant increase on Iba1 and CD11b immunoreactive cells (n = 2 donors; two-way ANOVA; *P ≤ .05; **P ≤ .005; ***P ≤ .001). D, Expression of microglial markers Iba1, TMEM119, and CX3CR1 were included to determine if cells maintain their expression under the different culturing conditions. HLA-DR was included as a negative control (Table S1). E, Fluorescence-activated cell sorting analysis comparison of the STR-MP compartment between high passage number (after passage 6) and fresh bone marrow sample shows that the STR-MP compartment increases with passage number (0.04% ± 0.01% of STR-MP in fresh sample vs 5.74% ± 3.05% of STR-MP in cell cultures after passage 6, unpaired two-tailed Student's t test, n = 5, P < .05). F, Stromal microglial CD11b + cells were detected based on CD11b and Iba1 coexpression. The graph shows the percentage of these cells compared across the different medium conditions and two time points, 3 and 5 weeks. A higher percentage of Iba1 CD11b immunoreactive cells was observed at 3 weeks under the serum medium supplemented with neurotrophins (n = 2 donors; one-way ANOVA; *P ≤ .05). G, The graph shows the percentage of cells in high passage cell cultures expressing the microglial markers Iba1, TMEM119, and CX3CR1 across culture serumcontaining conditions. HLA-DR was included as a negative control (Table S2). H, The graph shows the percentage of cells in high passage cell cultures under serum-free conditions for microglial markers Iba1, TMEM119, and CX3CR1. HLA-DR was included as a negative control (Table S2).   Figure 3H and Table S5A,B; unpaired, two-tailed Student's t test, P ≤ .05; n = 2 donors). The small percentage of TMEM119 + and HLA-DR + cells present in the cultures could phagocyte beads ( Figure 3I,J and Table S5A,B; unpaired, two-tailed Student's t test, P > .05; n = 2 donors) and may be an indirect indication of the antigen-presenting nature of these cells.

| Phagocytosis and classic activation in CD11b + cells
To determine if CD11b + cells can be "classically activated," cultures were exposed to bacterial LPS. In order to work with a pure population of STR-Ms, hBM MSCs were expanded, and STR-MPs were sorted and seeded in 96 multiwell plates at a confluence of 30 000 cells per well. The experimental design was restricted to the different serum-free conditions in high passage cultures, taking advantage of the higher STR-M content and avoiding the presence of serum, which can alter the experimental outcome. 45 Our results show that, after being exposed to NT and CK conditions for only 5 days, STR-Ms can be classically activated ( Figure 3L,M). In the case of being showing the number of engulfed beads per HLA-DR + TMEM119 + cell under the different culture conditions at 3 weeks, as a representative example of the engulfment activity of HLA-DR + TMEM119 + cells (n = 2 donors; one-way ANOVA; P > .05). K, Graph representing the percentage of sorted stromal microglial cells engulfing Aβ fibrils after being exposed to NM + NT and NM + CK conditions for 6 days (unpaired, two-tailed Student's t test, P > .05; n = 3 repetitions). L, Classic activation of stromal microglia after 5 days of exposure to neutrophins in serum-free conditions. An increase of the proinflammatory cytokine secretion of IL-6, IL-10, and IL-12p70NT was detected upon 24-hour exposure to LPS (n = 3 donors; paired, one-tailed Student's t test, *P < .05; **P < .01). M, Classic activation of stromal microglia after 5 days of exposure to CK in serum-free conditions. An increased secretion of the proinflammatory cytokines secretion of IL-4, IL-6, IL-10, and IL-13 was detected (n = 3 donors; paired, one-tailed Student's t test, *P < .05; **P < . showed retained phagocytic abilities when challenged for phagocytic abilities by adding GFP + latex beads in the medium ( Figure 3N).

| STR-M in a human ex vivo brain model
To determine if derivatives of the STR-MP CD11b + cells can integrate and survive in a human brain environment, GFP + stromal cultures consisting of, on average, 0.236% ± 0.109% GFP + nonhematopoietic CD11b + cells were grafted into ex vivo human brain tissue containing glioblastoma. The ex vivo model was obtained from glioblastoma resections sliced for organotypic culturing. Human brain slices engrafted with the GFP + cells were kept in culture for 2 weeks; slices were fixed and treated for iDISCO immunostaining to identify CD11b and TMEM119 immunoreactive cells among the engrafted GFP + cells. The two cell morphologies-ameboid like and those with soma projections-found in our GFP + engrafted cells were also seen in the slices ( Figure 4A,B). In the human tissue, host cells that were either CD11b immunoreactive or TMEM119 immunoreactive were observed ( Figure 4C). Likewise, host cells immunoreactive to both markers could be detected, which had a more microglia-like morphology ( Figure 4D). Some of the GFP + engrafted cells with a bone marrow stromal cell-like morphology showed low immunoreactivity to TMEM119 in comparison with host brain cells ( Figure 4E). Some of the ameboid GFP + engrafted cells were not reactive to CD11b + , whereas host cells were immunoreactive to both CD11b and TMEM119 ( Figure 4F). Some other ameboid GFP + cells were immunoreactive to CD11b ( Figure 4G) and even appeared to be integrated into cell groups immunoreactive to CD11b ( Figure 4H).

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

ETHICAL STATEMENT
All procedures performed were in accordance with the ethical standards of the institutional and/or national research committee. The ethical permit for hBM extraction was under permit 2009/12 and for use of human brain tissue resections, 212/2007.