Abstract
Microglia maintain brain health and play important roles in disease and injury. Despite the known ability of microglia to proliferate, the precise nature of the population or populations capable of generating new microglia in the adult brain remains controversial. We identified Prominin-1 (Prom1; also known as CD133) as a putative cell surface marker of committed brain myeloid progenitor cells. We demonstrate that Prom1-expressing cells isolated from mixed cortical cultures will generate new microglia in vitro. To determine whether Prom1-expressing cells generate new microglia in vivo, we used tamoxifen inducible fate mapping in male and female mice. Induction of Cre recombinase activity at 10 weeks in Prom1-expressing cells leads to the expression of TdTomato in all Prom1-expressing progenitors and newly generated daughter cells. We observed a population of new TdTomato-expressing microglia at 6 months of age that increased in size at 9 months. When microglia proliferation was induced using a transient ischemia/reperfusion paradigm, little proliferation from the Prom1-expressing progenitors was observed with the majority of new microglia derived from Prom1-negative cells. Together, these findings reveal that Prom1-expressing myeloid progenitor cells contribute to the generation of new microglia both in vitro and in vivo. Furthermore, these findings demonstrate the existence of an undifferentiated myeloid progenitor population in the adult mouse brain that expresses Prom1. We conclude that Prom1-expressing myeloid progenitors contribute to new microglia genesis in the uninjured brain but not in response to ischemia/reperfusion.
SIGNIFICANCE STATEMENT Microglia, the innate immune cells of the CNS, can divide to slowly generate new microglia throughout life. Newly generated microglia may influence inflammatory responses to injury or neurodegeneration. However, the origins of the new microglia in the brain have been controversial. Our research demonstrates that some newly born microglia in a healthy brain are derived from cells that express the stem cell marker Prominin-1. This is the first time Prominin-1 cells are shown to generate microglia.
Introduction
Microglia, the innate immune cells of the CNS, play critical roles during homeostasis and in the setting of neuroinflammation caused by injury or disease (Jebelli et al., 2015; Hammond et al., 2018). Unlike most other CNS cells, microglia have the capacity to replicate throughout the lifespan (Ajami et al., 2007; Füger et al., 2017; Reu et al., 2017). The ability of microglia to multiply and quickly replace dead or damaged microglia is thought to contribute to the maintenance of the healthy brain. However, the overabundance of activated microglia populations may also contribute to excessive synaptic stripping and neurodegeneration (Salter and Stevens, 2017). Although microglia are capable of proliferation, the cellular source of new microglia in the adult brain remains unclear.
Several reports have generated competing hypotheses about the source of new microglia in the adult CNS. These studies use microglia depletion and subsequent repopulation to understand microglia population dynamics (Elmore et al., 2014, 2015; Bruttger et al., 2015; Huang et al., 2018; Zhan et al., 2019). In one set of studies, Elmore et al. (2014, 2015) used a CSF1R inhibitor to deplete microglia from the brains of adult mice. After removal of the CSF1R inhibitor, they observed a nestin-expressing putative progenitor that generated new microglia. An additional study using diphtheria toxin administration to deplete CX3CR1-expressing microglia demonstrated repopulation from multiple islands scattered throughout the brain suggesting that there may be a putative progenitor population capable of generating new microglia (Bruttger et al., 2015). In contrast, a study using a novel fate-mapping technique called “microfetti” demonstrated that individual mature microglia can generate new adult microglia in the naive adult mouse brain (Tay et al., 2017). Recent studies also used a CSF1R inhibitor to deplete microglia from the brain but did not demonstrate proliferation from nestin-expressing cells (Huang et al., 2018; Zhan et al., 2019). The authors concluded that the primary source of microglia proliferation after depletion was mature microglia that survived CSF1R inhibition (Huang et al., 2018; Zhan et al., 2019). These contrasting conclusions based on depletion models do not address how microglia are generated in a normal adult brain.
To determine whether undifferentiated myeloid progenitors exist in the CNS and contribute to replenishment or renewal of microglia, novel marker genes are needed. Specific markers of myeloid progenitors would enable the use of fluorescence-activated cell sorting (FACS) or fate-mapping approaches to study the potential contribution of a progenitor pool to the microglia population. We propose Prominin-1 (Prom1; also known as CD133) as a candidate marker to label an undifferentiated myeloid progenitor population in the adult brain. Prom1 expression is found on human hematopoetic stem and progenitor cells (Yin et al., 1997; Uchida et al., 2000). The cells that generate microglia in the developing CNS are hematopoetic in lineage, suggesting that Prom1 may also be expressed by undifferentiated myeloid progenitors in the brain (Alliot et al., 1999; Ginhoux et al., 2010; Schulz et al., 2012; Hashimoto et al., 2013; Kierdorf et al., 2013; Gomez Perdiguero et al., 2015; Hoeffel et al., 2015). Although the function and ligands of Prom1 are unknown, cells expressing Prom1 are known to have stem cell properties (Miraglia et al., 1997; Yin et al., 1997; Uchida et al., 2000; Shmelkov et al., 2005; Walker et al., 2013). Prom1-expressing cells generate neurons, astrocytes, and oligodendrocytes in the brain in vitro and in vivo (Lee et al., 2005; Corti et al., 2007; Coskun et al., 2008; Walker et al., 2013; Holmberg Olausson et al., 2014; Okazaki et al., 2018). Given the hematopoetic stem cell lineage of microglia and the presence of Prom1 expression on hematopoetic stem cells and the multipotent properties of Prom1 cells in the brain, we hypothesized that a subset of Prom1 cells might be capable of generating new microglia.
Materials and Methods
Animals
One male TdTomato mouse (stock #007914) and one female Prom1-Tamoxifen inducible Cre (Prom1Cre) mouse (stock #017743) were obtained from The Jackson Laboratory. These mice were back-crossed onto our in-house colony of C57Bl6/J mice for four generations to obtain the TdTomato and Prom1Cre lines used for breeding. Experimental male and female mice were obtained from our colonies in house. Animals were group housed on a 14:10 h light/dark schedule (lights on at 6:00 A.M.) with access to food and water ad libitum. ZsGreen mice (stock #007906) were obtained from The Jackson Laboratory. All ZsGreen and Prom1 animals were group housed on a 12:12 h light/dark schedule (lights on at 7:00 A.M.) with access to food and water ad libitum. The Institutional Animal Care and Use Committee of the University of Washington or of the University of North Carolina, Chapel Hill approved all protocols and procedures. All studies followed the Animal Research: Reporting of In Vivo Experiments guidelines (Kilkenny et al., 2010) and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
In vitro studies of Prom1 generation of microglia
Mixed glia cultures from which microglia can be harvested were generated according to previously published protocols (Jayadev et al., 2011). In brief, litters of C57Bl6/J mice were obtained from our in-house colony. At postnatal day 3 or 4, pups were killed and their brains extracted. After removal of the meninges, the cortices were dissected and dissociated using enzymatic and mechanical dissociation. Isolated cells were plated in D10C media (DMEM, high glucose, catalog #1965–092, Invitrogen), Ham's Nutrient Mixture F12 (catalog #51651C, Millipore Sigma), 10% heat-inactivated horse serum (catalog #N4888, Invitrogen), 5 mm glutamine (catalog #25030–081, Invitrogen), 10 mm HEPES (catalog #15630–080, Invitrogen), and 1% Penicillin/Streptomycin (catalog #15140-122, Invitrogen). Twenty-four h after plating, the media was replaced with D10C plus 20% L929 cell-conditioned media to encourage glia and microglia growth. Nine to 10 d after media change, microglia were harvested from the flask using agitation, and the adherent layer was trypsinized. Cells were stained with anti-CD11b (catalog #130–113-243, Miltenyi Biotec), anti-CD45 (catalog #552848, B D Pharmingen), anti-Prom1 (catalog #141208, BioLegend), and DAPI (catalog #D8417, Sigma-Aldrich). A BD Biosciences Aria III flow cytometer detected and isolated microglia (CD11b+CD45int) and Prom1-expressing cells from the mature microglia and adherent cell populations. Microglia from the mature population were lysed with Zymo RNA lysis buffer, and RNA was extracted using the Quick RNA Miniprep Kit (catalog #R1054, Zymo Research). Prom1-expressing cells from the adherent layer were replated in 60 mm Corning Primaria (catalog #25382-687, VWR) dishes in D10C media containing macrophage colony stimulating factor (MCSF, 1:1000,10 ng/ml; catalog #416-ML-050/CF, R&D Systems). Seven d after Prom1 cells were isolated and placed back in vitro, the cells were trypsinized, again stained with the same anti-CD11b, anti-CD45, and anti-Prom1 antibodies and DAPI. Using the BD Biosciences Aria III flow cytometer, CD11b+CD45int and Prom1-expressing cells were isolated. Cell populations were lysed with Zymo RNA lysis buffer, and RNA was extracted using the Quick RNA Miniprep Kit (catalog #R1054, Zymo Research). In one experiment, mature microglia harvested from the floating layer and Prom1-expressing cells isolated from the adherent layer by a BD Biosciences Aria III flow cytometer were plated on 4-well chamber slides and kept in vitro for 7 d. The cells were then fixed, and slides were immunostained for Iba-1 following the protocol described below in Immunohistochemistry.
Replication of the above experiments was performed using cultures derived from Prom1Cre+/+ and ZsGreen+/+ breeding pairs generating Prom1Cre+/−ZsGreen+/− pups. Procedures were identical to those above for generation of mixed glia cultures. On day 3 after plating, (z)−4-hydroxytamoxifen (1 µm; catalog #H7904, Sigma-Aldrich) was added along with the D10C plus 20% L929 cell-conditioned media to encourage glia and microglia growth and to activate the Cre-recombinase in Prom1-expressing cells. Ten to 12 d postmedia change, microglia were harvested from the flask using agitation, and the adherent layer was trypsinized. Cells were stained with anti-CD11b (catalog #130-113-802, Miltenyi Biotec), anti-CD45 (catalog #552848, B D Pharmingen), anti-Prom1 (catalog #141204, BioLegend), and DAPI (catalog #D8417, Sigma-Aldrich). A BD Biosciences Aria III flow cytometer detected and isolated microglia (CD11b+CD45int) and ZsGreen-positive Prom1-expressing cells from the mature microglia and adherent cell populations, respectively. Microglia from the mature population were lysed with Zymo RNA lysis buffer, and RNA was extracted using the Quick RNA Microprep Kit (catalog #R1050, Zymo Research). ZsGreen-positive Prom1-expressing cells from the adherent layer were replated in 35 mm Corning Primaria (catalog #25382-654, VWR Scientific) dishes in D10C media containing MCSF (1:1000, 10 ng/ml; catalog #416-ML-050/CF, R&D Systems). Seven d after ZsGreen-positive Prom1 cells were isolated and placed back in vitro, the cells were trypsinized, again stained with the same anti-CD11b, anti-CD45, and anti-Prom1 antibodies and DAPI. Using the BD Biosciences Aria III flow cytometer, CD11b+CD45int and Prom1-expressing cells were isolated. Cell populations were lysed with Zymo RNA lysis buffer, and RNA was extracted using the Quick RNA Microprep Kit (catalog #R1050, Zymo Research).
Isolation and identification of CD45+ Prom1-expressing cells
Six C57Bl6/J mice PND 84–112 (12–16 weeks old) from our colony were anesthetized fully with avertin injection (0.6 ml/kg, i.p.) and perfused with cold HBSS + 1 mm HEPES. The forebrain was dissected and dissociated using enzymatic and mechanical dissociation. Cells were strained through a 70 µm filter to reach a single-cell suspension. A 30% Percoll gradient isolated microglia and removed myelin and other debris (Kokiko-Cochran et al., 2018). Cells were stained with the same anti-CD11b, anti-CD45, anti-Prom1, and DAPI as previously. A BD Biosciences Aria III flow cytometer detected microglia (CD11b+/CD45int), CD11b−/CD45int nonmicroglia, and Prom1-expressing cells.
In vivo fate mapping of Prom1-expressing cells
We used a fate-mapping strategy to determine whether Prom1-expressing cells generate new microglia in vivo. Six male and six female heterozygous (Prom1Cre+/−TdTomato+/−) mice for each of two time points were administered tamoxifen (TM; 20 mg) via oral gavage at 10 weeks of age (Furrer et al., 2013) then were allowed to develop to 6 or 9 months of age. An additional four male and four female Prom1Cre+/−TdTomato+/− littermates treated with corn oil (vehicle) at 10 weeks of age were also allowed to age to 6 or 9 months. Six male and six female monogenic (either Prom1Cre+/−TdTomato−/− or Prom1Cre−/−TdTomato+/−) mice were also used as additional controls. At the appropriate age, mice were anesthetized fully with avertin injection (0.6 ml/kg, i.p.) and perfused intracardially with cold HBSS +1 mm HEPES. One half of the forebrain was dissected and dissociated using enzymatic and mechanical dissociation to reach a single-cell suspension. A 30% Percoll gradient isolated microglia and removed myelin and other debris (Kokiko-Cochran et al., 2018). Cells were stained with anti-CD11b (catalog #130-113-243, Miltenyi Biotec), anti-CD45 (catalog #552848, B D Pharmingen), and DAPI (catalog #D8417, Sigma-Aldrich). A BD Aria III flow cytometer detected TdTomato-negative microglia (CD11b+/CD45int) and TdTomato-expressing microglia (CD11b+/CD45int/TdTomato+). The other hemisphere of the brain was placed in 4% paraformaldehyde (PFA) overnight, moved to a 30% sucrose solution, and then frozen. Sections 30 µm thick representing the entire mouse brain were generated using a cryostat.
Induction of microglia proliferation using ischemia/reperfusion
To determine whether the Prom1-expressing cell population would contribute significantly to rapid microglia proliferation, we used our fate-mapping approach in a model of ischemic preconditioning shown to induce microglia proliferation (McDonough et al., 2020). Five male and five female heterozygous bigenic (Prom1Cre+/−TdTomato+/−) mice were administered TM (20 mg) via oral gavage at 10 weeks of age (Furrer et al., 2013). An additional male Prom1Cre+/− TdTomato−/− mouse was also administered TM with the same parameters. Middle cerebral artery occlusion and reperfusion followed established protocols (Longa et al., 1989; Stenzel-Poore et al., 2003; McDonough et al., 2020). In brief, at 16 weeks of age and under full anesthesia using isofluorane, a Doppler flowmeter was attached to the skull, and an incision was made in the neck to access the common carotid artery. An intraluminal filament was then inserted into the common carotid artery until it blocked 70% of blood flow to the middle cerebral artery as measured by the Doppler flow. Occlusion was maintained for 15 min. The filament was then withdrawn to allow for reperfusion. Animals were given saline and buprenorphine intraperitoneally and monitored for behavioral signs of stroke (Bederson et al., 1986). At 24 h intervals postsurgery mice were weighed and given BrdU injections (50 mg/kg, i.p.; catalog #19-160, Sigma-Aldrich). At 72 h postsurgery mice were anesthetized fully with an avertin injection (0.6 ml/kg, i.p.) and perfused intracardially with cold HBSS + 1 mm HEPES followed by 4% PFA. The brain was removed, placed in 4% PFA overnight, moved to a 30% sucrose solution, and then frozen. Sections 30 µm thick representing the entire mouse brain were generated using a cryostat.
Gene expression studies
Reverse-transcriptase PCR (RT-PCR) using the High Capacity cDNA Reverse Transcriptase Kit (catalog #4374966, Applied Biosystems) generated cDNA from the RNA samples obtained from lysed sorted samples described above. For the replication studies, extracted RNA was reverse transcribed to cDNA and amplified using the Ovation PicoSL WTA System according to the protocol provided by the manufacturer (catalog #3312-24, Tecan). Amplified cDNA was then extracted using the Microelute Cycle Pure Kit according to the protocol provided by the manufacturer (catalog #D6293-01, Omega Bio-Tek) and stored at −20°C.
Quantitative RT-PCR using the Roche Universal Probe Library identified gene expression levels for β-actin (Probe 71 with primers: Forward: GGAGGGGGTTGAGGTGTT, Reverse: GTGTGCACTTTTATTGGTCTCAA), GAPDH (Probe 9 with primers: Forward: AGCTTGTCATCAACGGGAAG, Reverse: TTTGATGTTAGTGGGGTCTCG), Prom1 (Probe 20 with primers: Forward: AGCAGCAGTGACTGTACCTCAG, Reverse: TCTATCCACTGATGGGAGCTG), Iba-1 (Probe 3 with primers: Forward: GGATTTGCAGGGAGGAAAAG, Reverse: TGGGATCATCGAGGAATTG), TMEM-119 (Probe 1 with primers: Forward: GCATGAAGAAGGCCTGGAC, Reverse: CTGGGTAGCAGCCAGAATGT), CSF1R (Probe 56 with primers: Forward: CCCTGATGTCAGAGCTGAAGA, Reverse: TACAGGCTCCCAAGAGGTTG), and ZsGreen (Probe 25 with primers: Forward: ATCTGCAACGCCGACATC, Reverse: CTTGGACTCGTGGTACATGC).
Immunohistochemistry
Five sections throughout the brain (one section from each of the following regions: prefrontal cortex, striatum, rostral hippocampus, intermediate hippocampus, and caudal hippocampus) ∼1800 µm apart were selected from each animal. Free-floating sections were washed three times in 1× Tris-buffered saline (TBS) then incubated in blocking solution (1× TBS with 0.4% Triton X-100 (catalog #9002-93-1, Sigma-Aldrich), 5% donkey serum (catalog #D9663, Sigma-Aldrich) at room temperature for 2 h on a shaker. Sections were incubated in half-dilute blocking solution with primary antibodies (1:500 Iba-1; catalog #ab5076, Abcam), 1:1000 RFP (catalog #600-401-397S, Rockland) for 18 h at 4°C on a shaker. Sections were washed three times in 1× TBS + 0.1% Tween 20 before secondary antibody incubation (1:1000 Alexa Fluor 448 donkey anti-goat, catalog #A-11055, Life Technologies; 1:1000 Alexa Fluor 594 donkey anti-rabbit, catalog #SA5-10040, Invitrogen) at room temperature for 3 h. DAPI (1:1000) was added during the last 30 min of secondary antibody incubation. Sections were mounted with Vectashield (catalog #H-1200, Vector Laboratories) and stored at 4°C in the dark until imaging.
To find Iba-1 and RFP double-positive cells, observers blind to the experimental conditions systematically viewed each section at 40× magnification. By switching between the FITC, and Cy3 channels, the observers confirmed that RFP-expressing cells were also immunolabeled by Iba1 antibody. Locations of the double-positive cells were recorded on a schematic brain map created by the observers. Only double-positive cells fully contained within a single section (visible DAPI) were recorded.
Statistical Analysis
FACS data were first analyzed in FlowJo version 10 to obtain mean fluorescence intensity values for the populations of interest. Statistics were run using GraphPad Prism version 8.03. For all fate-mapping analyses, 2-way ANOVAs were used. False discovery rate correction using the two-stage linear step-up procedure of Benjamini et al. (2006) was applied to all multiple comparisons to correct for the number of tests used. Gene expression analyses used 1-way ANOVA with Bonferroni's correction applied for multiple comparisons. An unpaired t test was used to compare the amount of Prom1 signal in young adult brain. All results are displayed using mean and SD.
Results
Prom1-expressing cells generate new microglia in vitro
Neonatal mixed glia cultures generate multiple harvests of mature microglia that float away from the adherent glial monolayer. Thus, we hypothesized that cells capable of generating new microglia are present in the adherent layer of neonatal mixed glia cultures. Mature microglia floating above, and intertwined with, the mixed glia monolayer were harvested from the media (Fig. 1A). The gating strategy for identifying Prom1-positive cells in the adherent layer is depicted in Figure 1B. The Prom1 cells were easily isolated from microglia, which are CD11b+ and are seen as a population above the Prom1-positive cells. Microglia were negative for Prom1 surface immunoreactivity but expressed the myeloid surface markers CD11b and CD45 in flow cytometric analysis (Fig. 1C). However, the adherent layer contained a population of Prom1-expressing cells that could be separated from CD11b+/CD45int microglia using FACS (Fig. 1D). When isolated using FACS and placed back in vitro for 7 d, these Prom1-expressing cells generate a new population of CD11b+/CD45int microglia separable from the Prom1-expressing population that remained in the adherent layer (Fig. 1E). These data indicate that Prom1-expressing cells will generate new Prom1-negative CD11b+/CD45int microglia in vitro.
Prom1-expressing cells are present in the adherent layer of the mixed glia culture and generate new microglia in vitro. A, Experimental design. The floating and adherent layers of a mixed glia culture 9 d after initial media change were isolated by agitation or trypsinization, and Prom1 cells or microglia (MG) were identified and isolated. Prom1 cells were plated for another 7 d. Populations were then isolated by fluorescence activated cell sorting (FACS) to determine the presence of microglia. B, The gating strategy for identifying Prom1+ cells as separable from microglia. Gates were drawn for all cells, and from these, single cells and then live cells were identified. Next, live cells that were Prom1+ could be clearly separated from the CD11b+ (microglia) population. Gates for Prom1+ cells from the adherent layer were drawn conservatively to isolate a pure population of Prom1+ cells that does not include microglia. C, Mature floating microglia (MG) from the mixed glia culture show typical CD11b and CD45int expression but do not show Prom1 expression. D, The adherent layer from the mixed glia culture system shows a small population of CD11b+/CD45int-expressing microglia (MG). There is an additional separable population of Prom1-expressing cells that are not CD11b+. The microglia population can be seen as the CD11b+ population in the graph of CD11b and Prom1, and are distinctly separate from the Prom1+ cells located in the Prom1 gate. E, When the Prom1 cells identified in C were isolated using FACS and placed back in vitro in D10C + MCSF (1:1000, 10 ng/ml), a new population of Prom1-negative but CD11b+/CD45int cells emerges after 7 d. F, Images of microglia (MG) and Prom1 cells after isolation and 7 d in vitro. Left, The bright field images demonstrate that microglia generate processes and remain as single cells spread apart from each other. In contrast, the Prom1-expressing cells generate a monolayer in vitro that appears to recapitulate the mixed glia layer seen in cortical cultures. Close-up images of Iba1 immunostaining shows that the mature microglia express the myeloid marker Iba1, whereas fewer Iba1-expressing cells are seen in the monolayer. Instead, small clusters of Iba1-positive cells can be observed in the representative image, which we hypothesize to be the newly generated microglia. Scale bars: 5 μm.
To further assess the identity of the Prom1 cells generated, we performed an additional experiment using immunostaining for Iba-1. The experiment replicated the above conditions except that after FACS isolation the microglia and Prom1 cells were plated separately in chamber slides for 7 d and then fixed and immunolabeled to detect Iba-1. A low-power brightfield image (10×) shows mature microglia isolated from mixed glia cultures express Iba-1 and maintain distance between cells (Fig. 1F). In contrast, plated Prom1 cells (+M-CSF) have generated a mixed-glia monolayer within which cells are completely confluent in the brightfield image (Fig. 1F). Despite their confluency, the higher power Iba-1 image shows a single cluster of small Iba-1-expressing cells. We hypothesize these clusters are the newly generated CD11b+/CD45int microglia.
To strengthen our hypothesis that microglia are derived from Prom1-expressing stem cells in vitro, we generated mixed glia cultures from Prom1-Cre+/− Floxed-ZsGreen+/− pups to fate map Prom1 cells. After treatment with (z)−4-hydroxytamoxifen, all Prom1-expressing cells and their daughters express ZsGreen (Fig. 2A). We sorted the Prom1-expressing ZsGreen-positive cells from the adherent layer, and replated them as before. After 7 d, we identified a population of ZsGreen-positive (Fig. 2B) CD11b+/CD45int microglia (Fig. 2C). These results replicate the findings observed in wild-type cultures, confirming that Prom1 cells generate microglia in vitro.
Prom1Cre+/−ZsGreen+/− cells generate new ZsGreen+ microglia in vitro. A, Experimental design for the replication of Figure 1 using Prom1+/−ZsGreen+/− cells in vitro. After treatment with tamoxifen (TM), many Prom1-expressing cells express ZsGreen, which is detectable by FACS. We isolated Prom1-expressing ZsGreen-positive cells to determine whether ZsGreen-positive microglia would be found after 7 d. B, The gating strategy for identifying ZsGreen-positive microglia. C, After 7 d in vitro, Prom1-expressing ZsGreen-positive cells generated a clear population of ZsGreen-positive CD11b+/CD45int microglia (MG).
To confirm that the CD11b+/CD45int cells generated by the Prom1-expressing population of the adherent layer were microglia, we investigated the expression of canonical microglia marker genes in these cells. Gene expression analysis demonstrated that the CD11b+/CD45int cells generated from Prom1-expressing cells have identical gene expression of canonical microglia-associated genes such as Iba-1 and CSF1R as mature microglia from the original mixed culture (Fig. 3A,C). The CD11b+/CD45int cells generated from Prom1-expressing cells also have higher expression of TMEM-119, a microglia-specific gene, than mature microglia from the original mixed glia culture (Fig. 3B). Additionally, neither the CD11b+/CD45int nor the mature floating microglia express Prom1, whereas the cells identified by FACS to be Prom1-positive do show Prom1 gene expression (Fig. 3D). Based on the similarity of gene expression between the cell populations, we demonstrate that Prom1-expressing cells isolated from mixed glia culture can generate new microglia in vitro.
Mature microglia isolated from mixed glia cultures and CD11b+/CD45int cells generated from isolated Prom1-expressing cells have similar gene expression of canonical microglia genes. A–D, CD11b+/CD45int cells generated from Prom1 cells do not differ in their Iba-1 (A) or CSF1R (C) expression from mature microglia (MG) in vitro. Interestingly, the CD11b+/CD45int cells generated from Prom1 cells demonstrate higher expression of TMEM-119 (B), a microglia-specific marker. The Prom1 cells that generated the CD11b+/CD45int cells do show expression of Prom1 (D) but do not show expression of canonical microglia genes (A, B, C). N.D. = not detected. Error bars are SD. *p < 0.05, **p < 0.01, ***p < 0.001.
To determine the proliferation rate of Prom1 cells that serve as microglia progenitors, we used the identical experimental design outlined in Figure 1A with the addition of carboxyfluorescein succinimidyl ester (CFSE) to the Prom1+ population after isolation and replating. Data from experiments using CFSE in our in vitro model suggest that 27.2 to 27.5% of the Prom1-positive population underwent cell division in the first 7 d after replating (data not shown). Each of the biological replicates had a proliferation index of 1.14 (data not shown). These data suggest that Prom1 cells divide to generate microglia, and as many as 27% of them may be microglia progenitors, although this remains to be replicated and demonstrated in vivo.
Mature microglia do not express Prom1, but some Prom1 cells express myeloid markers in vitro and in vivo
We hypothesized that a subpopulation of Prom1-positive cells that express other myeloid markers would generate new microglia. To determine whether there was a population of Prom1-expressing myeloid precursors in vitro, we performed flow cytometric analysis of the attached layer from mixed glia cultures. We observed a population of CD45-expressing cells separable from mature microglia by lack of CD11b expression (Fig. 4A). A high proportion of these CD45int/CD11b− cells also demonstrate Prom1 expression (Fig. 4B). The proportion of CD45int/CD11b− cells that express Prom1 is significantly larger than observed in CD11b+/CD45+ mature microglia (Fig. 4C). To determine whether this observation would be replicated by Prom1-expressing myeloid precursors in vivo, we performed flow cytometric analysis of the CD45-expressing cells in a young adult mouse brain. We observed an identifiable population of CD45-expressing cells separable from mature microglia by lack of CD11b expression (Fig. 4D). A high proportion of the CD45int/CD11b− cells also demonstrate Prom1 surface expression (Fig. 4E). Again, the population of CD45int/CD11b− cells that express Prom1 is significantly higher than in CD11b+/CD45int mature microglia (Fig. 4F). These findings confirm our in vitro observation that mature microglia do not express Prom1, whereas immature myeloid cells, similar to CD45+/CD11b− cells that enter the CNS during embryonic development (Kierdorf et al., 2013), demonstrate Prom1 surface expression. We additionally identified the proportion of the Prom1-expressing population that expresses CD45 both in vivo and in vitro. Using the same gating strategy shown in Figure 1B, we identified Prom1-expressing cells. Although the percentage of Prom1-expressing cells that are CD45 expressing was only 10% in vitro, it was 30% in vivo (Fig. 4G,H). As expected, the majority of Prom1-expressing cells are unlikely to be microglia progenitors, but a subset of cells expressing a stem cell marker in the CNS appear to coexpress an early myeloid lineage marker (CD45). Prom1-expressing cells from the young adult mouse brain that expressed CD45 by flow cytometric analysis may be a subset of Prom1-expressing cells capable of generating myeloid lineage cells including microglia.
A subset of CD45+ cells are Prom1 expressing indicating an undifferentiated microglia precursor like phenotype both in vitro and in vivo. A, In the adherent layer of the mixed glia culture of the neonatal mouse cortex, microglia are identified by their expression of both CD11b and CD45 (CD11b+/CD45int). In addition, there is a population of CD11b−/CD45int cells in the adherent layer. We examined both of these populations for expression of Prom1. B, The nonmicroglia CD11b−/CD45int cell population identified in A has a much higher expression of Prom1. C, The percentage of the CD11b+/CD45int and CD11b−/CD45int populations that are Prom1 expressing is much higher for CD11b−/CD45int cells. In contrast, the number of floating mature microglia that express Prom1 is low. D, In an ex vivo preparation, mature microglia are identified in the young adult mouse brain by expression of both CD11b and CD45 (CD11b+/CD45int). Similar to our findings in vitro, there is a large additional population of CD45-positive cells that are not CD11b expressing (CD11b−/CD45int). E, The CD11b−/CD45int cell population identified in D has much higher expression of Prom1. F, The percentage of the CD11b+/CD45int and CD11b−/CD45int populations that are Prom1 expressing is much higher for CD11b−/CD45int cells. In contrast, the number of mature microglia in the adult mouse brain that express Prom1 is low. G, We also identified the percentage of Prom1-expressing cells that are CD45+ in vivo. We demonstrate the Prom1 gating and then the percentage of CD45+ cells within that gate. This is ∼30% of the Prom1+ population, indicating that the majority of Prom1-expressing cells are unlikely to be microglia progenitors. H, We identified the percentage of Prom1-expressing cells that are CD45+ in vitro. This was ∼10% of the Prom1+ population, indicating that there are potentially even fewer Prom1+ microglia progenitors in the in vitro setting. Overall, these data confirm the in vitro findings from Figures 1 and 3 that mature microglia do not express Prom1 and further identify a subset of Prom1-expressing cells that also express CD45 as potential undifferentiated myeloid microglia progenitors. Error bars are SD. *p < 0.05.
Fate mapping reveals a population of microglia generated from Prom1 cells in vivo
As Prom1-expressing cells generate new microglia in vitro, we hypothesized they may also contribute new microglia in vivo. We tested this hypothesis using a fate-mapping approach in the adult mouse brain. Mice heterozygous for Prom1-CreERT2 and flx-stop-flx TdTomato were administered TM to induce TdTomato expression specifically in Prom1-expressing cells at 10 weeks of age. Using this approach, all cells expressing Prom1 at the time of TM treatment will continue to express TdTomato as well as all daughter cells generated from cells that expressed Prom1 at 10 weeks of age. Mice were then allowed to age until 6 or 9 months to determine whether the population of daughter cells expressing TdTomato include microglia. Previous studies have demonstrated that microglia turnover rates in the healthy adult brain are low (Ajami et al., 2007; Füger et al., 2017; Reu et al., 2017). As expected, at both 6 (data not shown) and 9 months of age, the majority of CD11b+/CD45int microglia are TdTomato negative (Fig. 5A). In addition, a small population of CD11b+/CD45int cells expressed TdTomato at both 6 and 9 months of age (Fig. 5B). The number of TdTomato-positive CD11b+/CD45int cells was significantly higher in the Prom1Cre+/−TdTomato+/− mice treated with TM than vehicle-treated controls overall (Fig. 5C; main effect of treatment, F(1,20) = 5.11, p = 0.035). The population of CD11b+/CD45int TdTomato-positive microglia significantly increased from 6 to 9 months of age (Fig. 5C; post hoc test, t(20) = 2.53, q = 0.021). These fate-mapping data indicate a slow increase in the number of new microglia generated during adulthood from Prom1-expressing cells over the lifetime of the mouse and in the absence of an inflammatory stimulus.
Fate mapping of Prom1-expressing cells after 10 weeks of age demonstrates the presence of a small but growing microglia daughter population in the healthy adult mouse brain. A, Flow cytometry analysis visualizes the canonical microglia population isolated from the adult mouse brain as CD11b+ and CD45int. The population of these cells that do not express TdTomato is large as displayed by the density of rings on the contour plot. This was anticipated as we hypothesized turnover rates would be low in the naive mouse brain. B, At both 6 and 9 months (mo) of age (∼3 and 6 months post-tamoxifen treatment to induce fate mapping), a small population of TdTomato-expressing CD11b+/CD45int cells in the same location as the non-TdTomato-expressing microglia population is seen using flow cytometry. These TdTomato-expressing microglia are daughters of Prom1-expressing cells. C, The population of microglia that are TdTomato positive is small (<1%). The population of TdTomato-positive microglia is greater in the tamoxifen (TM)-treated Prom1Cre+/−TdTomato+/− mice than the corn oil vehicle–treated animals (CO) at 9 months (mo) of age. The population of TdTomato+ microglia is also significantly higher at 9 months of age compared with 6 months of age, indicating that the population increases slowly over time. Error bars are SD. *p < 0.05.
We additionally investigated whether the TdTomato-positive microglia daughter cells of Prom1-expressing cells would be distributed throughout the brain similar to previous reports of microglia proliferation (Bruttger et al., 2015). Using immunohistochemistry to identify microglia (Iba-1) double-labeled with TdTomato (RFP), we were able to identify daughter cells spread throughout the brain parenchyma (Fig. 6A,B). Using a different marker of myeloid cells (PU.1), we identified several cells double labeled with PU.1 and TdTomato (Fig. 6C). These PU.1/TdTomato cells may be the progenitors themselves, or daughter microglia. We additionally identified a number of TdTomato-positive daughter cells that were not microglia (Fig. 6D), consistent with previously published literature regarding the stem cell nature of Prom1-expressing cells (Lee et al., 2005; Corti et al., 2007; Coskun et al., 2008; Walker et al., 2013; Codega et al., 2014; Okazaki et al., 2018). Finally, we confirmed that microglia express mRNA for the fluorescent marker protein, confirming that they produce this protein through gene expression rather than through endocytosis or other uptake measures (Fig. 6E). The combination of FACS and immunohistochemical analysis of fate mapping reveals a population of microglia generated by Prom1-expressing cells in the naive adult mouse brain.
Microglia daughters of Prom1-expressing cells can be identified using immunohistochemistry and are distributed throughout the extent of the mouse brain, and some TdTomato cells also express PU.1, but TdTomato is not distributed throughout microglia, although microglia do express fluorescent marker mRNA. A, Representative confocal images of double-positive Iba-1 (green) and TdTomato (red) microglia daughters of Prom1-expressing cells (arrows). The side views demonstrate the colocalization of TdTomato within the microglia cells. B, Microglia daughters of Prom1-expressing cells were identified throughout the extent of the brain with each black dot representing a confirmed double-positive Iba-1 and TdTomato daughter cell. C, Representative confocal image of a double-positive PU.1 (green) and TdTomato (red) cell (arrows). As PU.1 is a marker of early myeloid progenitors, cells that express both Prom1 and PU.1 may be microglia progenitor cells. D, Representative images of double-positive Iba-1 (green) and TdTomato (red) microglia daughters of Prom1-expressing cells (arrows) seen near other TdTomato-positive cells where the TdTomato fills the cell. The distribution of TdTomato inside microglia was punctate, making identification more difficult and may have caused us to underestimate how many daughter microglia were generated. E, Quantitative PCR demonstrates expression of fluorescent marker mRNA in microglia derived from Prom1+ cells. Although the level is lower than that seen in Prom1+ cells, fluorescent marker mRNA can be detected in microglia derived from Prom1+ cells, whereas WT Prom1+ cells and microglia treated with TM did not demonstrate detectable expression (N. D. = not detected). This suggests that although expression of TdTomato is punctate, it is likely to be real expression of fluorescent protein. Error bars are SD. Scale Bars: A, 10 µm; C, 10 µm; D 10 µm.
Following ischemia/reperfusion, most microglia proliferation is not derived from the Prom1-expressing progenitor population
Fate mapping demonstrates that Prom1-expressing myeloid progenitors can generate new microglia. We next asked whether Prom1-expressing cells are induced to generate new microglia in the setting of an inflammatory stimulus. To address this question we used a model of ischemia/reperfusion with documented induction of microglia proliferation (McDonough et al., 2020). We hypothesized that Prom1 cells would generate a significant portion of the new microglia following ischemia/reperfusion in this model. We again used our fate-mapping approach. Mice heterozygous for Prom1-CreERT2 and flx-stop-flx TdTomato were administered TM to induce TdTomato expression specifically in Prom1-expressing cells at 10 weeks of age (Fig. 7A). At 16–20 weeks of age, mice underwent a 15 min unilateral middle cerebral artery occlusion/reperfusion (MCAO/R; Fig. 7A). Prior studies demonstrate significant microglia proliferation in the cortex of mice ipsilateral to the site of injury at 72 h following reperfusion. We administered BrdU every 24 h beginning the day of surgery to mark newly divided cells and used immunohistochemistry and fluorescence microscopy to determine whether newly divided microglia were daughters of Prom1-expressing cells. We observed extensive TdTomato staining in the brain similar to our previous fate-mapping animals without MCAO/R (Fig. 7B). We believe the majority of these TdTomato+ cells to be astrocytes and endothelial cells based on morphology. Because MCAO/R is unilateral, a direct comparison of TdTomato can be made between the ipsilateral and contralateral cortices to determine whether MCAO/R changed the intensity of TdTomato labeling. No significant difference in TdTomato was observed between the ipsilateral and contralateral hemispheres (two-way ANOVA of TdTomato signal hemisphere by sex main effect of hemisphere, F(1,7) = 1.59, p = 0.248). As anticipated (McDonough et al., 2020), we observed increased numbers of Iba1-expressing microglia ipsilateral to ischemia/reperfusion (Fig. 7B) and increased BrdU+ Iba-1-expressing cells compared with the contralateral cortex (Fig. 7C). However, the majority of the BrdU+ Iba-1+ microglia were not TdTomato+, indicating they were not generated from Prom1-expressing cells (Fig. 7C,D). We observed rare triple-positive microglia daughter cells from Prom1-expressing progenitors (Fig. 7D, arrow), but these did not make up the majority of dividing microglia. We conclude that Prom1-expressing myeloid progenitors contribute to microglia renewal in the adult brain during homeostasis but do not appear to be involved in the rapid proliferation of Iba1+/BrdU+ populations under conditions of acute inflammation and injury induced by ischemia/reperfusion in this MCAO/R model.
Prom1-expressing microglia progenitors are not the primary contributors to microglia proliferation in the setting of acute injury and inflammation. A, Experimental design schematic. We again used our Prom1Cre+/−TdTomato+/− mice to fate map the Prom1-expressing cells. Mice were administered tamoxifen at 10 weeks of age and given a 15 min unilateral middle cerebral artery occlusion and reperfusion (MCAO/R) at 16–20 weeks of age. On the day of surgery and every 24 h following, mice were administered BrdU to identify dividing cells. Brains were perfused and isolated for immunohistochemistry 72 h after MCAO/R. B, A representative set of montages from a male mouse demonstrating increased Iba-1 (green) staining in the ipsilateral (left) hemisphere compared with the contralateral hemisphere that did not receive an insult. C, Representative 20× images from the ipsilateral and contralateral cortex stained with Iba-1 (green), RFP (red), and BrdU (white). The ipsilateral cortex demonstrates greater proliferation of Iba-1 cells with significantly more Iba-1 and BrdU+ cells that also display ameboid morphology. By contrast, microglia in the contralateral hemisphere demonstrate ramified morphology and very little BrdU signal. D, Representative 100× images from the ipsilateral and contralateral cortex stained with Iba-1 (green), RFP (red), and BrdU (white). We occasionally identified Iba-1, RFP, and BrdU triple-positive cells (arrow). However, the majority of Iba-1 and BrdU-positive cells identified in the ipsilateral cortex were not TdTomato positive, indicating they are daughters of a source other than Prom1-expressing cells. These close-up images again demonstrate the extreme changes in morphology and proliferation seen between the ipsilateral and contralateral cortices of the same mouse. Scale bars: C, 5 µm; D, 5 µm.
Discussion
Data from studies of microglia repopulation after experimental depletion have led to contradictory interpretations on the origins of new microglia (Elmore et al., 2014; Bruttger et al., 2015; Huang et al., 2018; Zhan et al., 2019). Although resolution of this controversy will yield important information about the source of new microglia during injury or disease, it is also important to understand how microglia populations are maintained in the intact, uninflamed brain. To our knowledge this is the first description of Prom1-expressing cells generating newly born microglia both in vitro and in vivo. We observed that the adherent layer of neonatal mixed glia culture contains Prom1-expressing cells that are capable of generating a new population of microglia. Fate mapping of Prom1 cells following TM administration in mice at 10 weeks of age identified a population of mature microglia generated from Prom1-expressing cells. Using a model of brain injury that induces microglia proliferation as well as inflammation, we determined that the Prom1-expressing population is not the primary contributor to rapid microglia proliferation after this specific acute injury model. Together, these data provide evidence that a subset of Prom1-expressing cells serve as microglia progenitor cells in the intact, uninjured mouse brain.
The repeated ability to harvest mature microglia from neonatal mixed glia cultures indicates there may be a progenitor population in the adherent layer of the flask (Witting and Möller, 2011; Tamashiro et al., 2012). Prom1-expressing cells isolated from the adherent layer generate a new population of microglia with subsequent time in vitro. Mature microglia do not express Prom1; therefore, mature microglia are unlikely to be the source of these new microglia. Although our data suggest that Prom1-expressing cells generate new microglia, we have not shown that all new microglia in mixed neonatal culture are born from asymmetric division of Prom1-expressing cells. We also have not fully identified the proliferation rate or proportion of Prom1 cells that give rise to microglia. Additional studies will be required to determine these factors and to extend these findings to in vivo studies. Whether it is through division or differentiation, our data are the first to establish that Prom1-expressing cells can generate microglia in vitro.
We used the inducible Cre-Lox system to fate map the progeny of Prom1-expressing cells in adult mice. When TM was administered at 10 weeks of age, a population of TdTomato-positive microglia daughter cells was observed in 6-month-old mice. Furthermore, the size of the TdTomato-positive microglia population increased from 6 to 9 months of age. The slow generation of these new microglia is expected given the low turnover rates previously reported for microglia in the intact brain (Ajami et al., 2007; Füger et al., 2017; Reu et al., 2017). One limitation of fate-mapping studies is Cre-mediated recombination without induction by TM (Álvarez-Aznar et al., 2020). However, other studies using the same strain of Prom1-Cre mice have noted little to no recombination without TM induction (Snippert et al., 2009; Zhu et al., 2009). In our studies we did note a small increase in TdTomato fluorescence from mice treated with vehicle by FACS but not immunolabeling. This fluorescence did not increase with time, whereas the number of TdTomato-positive microglia increased from 6 to 9 months (Fig. 5C). We believe this small amount of fluorescence to likely be an artifact of a permissive gating strategy to identify TdTomato because we did not identify TdTomato protein by immunolabel in the same vehicle-treated brains. Based on these findings, we believe that our data supports the slow generation of microglia from Prom1-expressing cells in the adult brain, however, additional studies are needed to confirm and replicate our findings.
Although we confirmed that mature microglia generate fluorescent marker mRNA indicating that they did arise from a Prom1-expressing population (Fig. 6E), an additional limitation in our studies is the distribution of TdTomato in fate-mapped mature microglia (Fig. 6D). Although the cytoplasm is filled with TdTomato protein in nonmyeloid cell types, microglia tended to package TdTomato into punctate vacuoles. This subcellular localization made the identification of the double-positive cells more difficult and may have affected our ability to identify TdTomato-positive cells via FACS. Therefore, our estimates of the number of microglia derived from Prom1-expressing cells either by immunohistochemistry or by FACS may underestimate the total population. The majority of other studies that mark microglia with fluorescent proteins use GFP rather than RFP, and the fluorescence is more readily identifiable throughout the cell (Elmore et al., 2014; Huang et al., 2018). Future fate-mapping studies may be advised to use GFP to better detect their populations of interest. Despite this confound, we were able to identify a population of TdTomato-positive microglia that increased in the healthy mouse brain. These data reveal that Prom1-expressing cells are capable of slowly generating new microglia in vivo.
The literature demonstrates that Prom1 is expressed by cells with stem cell properties capable of generating many cell types, yet our data are the first to demonstrate that Prom1 cells are capable of generating microglia. Several studies have identified Prom1-expressing cells as neural progenitors in the brain (Lee et al., 2005; Corti et al., 2007; Coskun et al., 2008; Walker et al., 2013; Codega et al., 2014; Okazaki et al., 2018). In addition to the ability to generate neurons, multiple studies have shown that Prom1-expressing cells can additionally generate astrocytes and oligodendrocytes both in vitro and in vivo (Lee et al., 2005; Corti et al., 2007; Okazaki et al., 2018). These data all support a multipotent role for Prom1-expressing cells as being capable of generating neurons, astrocytes, oligodendrocytes, and microglia. It remains to be determined whether there are subpopulations of Prom1-expressing cells that generate each of these different cell populations. There is evidence to support multiple populations of Prom1-expressing cells (Codega et al., 2014; Holmberg Olausson et al., 2014) and our hypothesis is that only a subset of this larger stem cell population is likely to be committed myeloid progenitors that are able to generate microglia.
Microglia are derived from hematopoetic progenitor cells that migrate into the CNS early in development (Alliot et al., 1999; Ginhoux et al., 2010; Schulz et al., 2012; Kierdorf et al., 2013; Gomez Perdiguero et al., 2015; Hoeffel et al., 2015; Sheng et al., 2015). Prom1 is expressed by hematopoetic stem cells (Miraglia et al., 1997; Yin et al., 1997), supporting the hypothesis that Prom1-expressing cells in the CNS have the potential to generate microglia. In addition, one proven approach to generate induced pluripotent stem cell (iPSC)–derived microglia involves differentiation into a hematopoetic progenitor state before the induction of microglia differentiation (Abud et al., 2017; McQuade et al., 2018). Although the expression of Prom1 has not been reported either in CNS-invading erythromyeloid precursors or during iPSC to microglia differentiation protocols, the data that Prom1 is expressed by hematopoetic progenitor and stem cells in both murine and human bone marrow suggests that myeloid progenitors expressing Prom1 in the brain may generate microglia. An additional study has identified early erythromyeloid progenitors as endothelial cell progenitors as well (Plein et al., 2018). This link between endothelial and microglia progenitors may explain the prevalence of TdTomato-positive endothelial cells in our immunostaining results. Although we hypothesize that the endothelial progenitors (CSF1R+; Plein et al., 2018) and microglia progenitors (Prom1+/CSF1R−/CD45+) are likely different populations of cells, further studies are needed to identify these potential subpopulations and their origins. Our data identify CD45, a myeloid population marker, as a potential secondary marker of Prom1-expressing microglia progenitors. Although Prom1 cells were CD11b negative, a large population of CD45+ cells also expressed Prom1. Our data indicate that somewhere between 10 and 30% of Prom1-expressing cells express CD45, a small proportion of the overall Prom1 population. Although not all of these may be microglia progenitors, we hypothesize that Prom1 cells that are microglia progenitors will express myeloid markers like CD45. Further studies are needed to determine whether the hematopoetic progenitors used in iPSC generation of microglia or the erythromyeloid precursors that migrate into the brain during early development express Prom1.
We determined that Prom1-expressing microglia progenitors are not the primary contributors to rapid microglia proliferation in a transient ischemia/reperfusion injury model. We observed extensive TdTomato staining in the brain and believe these cells to be primarily endothelial cells and astrocytes based on morphology. There was no difference between ipsilateral and contralateral hemispheres in total TdTomato staining, indicating that transient ischemia/reperfusion injury does not affect the number of TdTomato-positive cells in this model. As expected, our data demonstrated significant proliferation of Iba-1-positive cells following a 15 min middle cerebral artery occlusion and reperfusion (McDonough et al., 2020). We did find a few newly divided microglia that were daughters of Prom1-expressing cells (BrdU+/Iba-1+/TdTomato+). However, the majority of newly divided microglia were not generated from Prom1-expressing progenitors, indicating that another source, potentially the division of adult microglia (e.g., Tay et al., 2017; Huang et al., 2018; Zhan et al., 2019), may be the likely cellular origin of this rapidly dividing population. It is important to note that this is a single model of acute CNS injury and associated microglia proliferation. The results described may not be generalizable to other forms of CNS disease or injury. Further studies using additional models are needed to determine exactly what the cellular origin of new microglia is under different circumstances of disease, injury, depletion, and in the intact, uninjured, healthy brain.
Conclusion
Prom1 (CD133)-expressing cells are capable of generating new microglia both in vivo and in vitro. This new population of committed brain resident myeloid progenitors may be identified by their coexpression of Prom1 and CD45. Prom1-expressing progenitor cells fate mapped via Prom1-driven Cre-recombinase expression are spread throughout the brain and generate new microglia slowly over time in the healthy adult mouse. Inflammation caused by exposure to transient ischemia/reperfusion injury leads to microglia proliferation; however, the majority of new microglia in this setting did not arise from Prom1-expressing myeloid progenitors. The generation of new microglia by a subset of Prom1-expressing cells deserves further study as this population of progenitors may influence microglia population dynamics in other settings.
Footnotes
This work was supported by National Institutes of Health Grants R21-NS096334-01A1 and R01-AG051437-01. K.E.P. was supported by National Institutes of Health Grant 5T32-AG052354-02, and M.S.A. was supported by the Howard Hughes Medical Institute Gilliam Fellowship for Advanced Study. We thank Dr. Fang Kuang along with Kevin Green, Lewis Luo, Allen Do, Angel Cheung, Jayven Manibusen, and Anna Huang for technical support and discussions.
The authors declare no competing financial interests.
- Correspondence should be addressed to Gwenn A. Garden at gagarden{at}email.unc.edu