Here, we describe a novel flow cytometric method for prospective isolation of early burst-forming unit erythroid (BFU-e) and colony-forming unit erythroid (CFU-e) progenitors directly from fresh mouse bone marrow and spleen. This protocol, developed based on single-cell transcriptomic data, is the first to isolate all the tissue’s erythroid progenitors with high purity.
Early erythroid progenitors were originally defined by their colony-forming potential in vitro and classified into burst-forming and colony-forming "units" known as BFU-e and CFU-e. Until recently, methods for the direct prospective and complete isolation of pure BFU-e and CFU-e progenitors from freshly isolated adult mouse bone marrow were not available. To address this gap, a single-cell RNA-seq (scRNAseq) dataset of mouse bone marrow was analyzed for the expression of genes coding for cell surface markers. This analysis was combined with cell fate assays, allowing the development of a novel flow cytometric approach that identifies and allows the isolation of complete and pure subsets of BFU-e and CFU-e progenitors in mouse bone marrow or spleen. This approach also identifies other progenitor subsets, including subsets enriched for basophil/mast cell and megakaryocytic potentials. The method consists of labeling fresh bone marrow or spleen cells with antibodies directed at Kit and CD55. Progenitors that express both these markers are then subdivided into five principal populations. Population 1 (P1 or CFU-e, Kit+ CD55+ CD49fmed/low CD105med/high CD71med/high) contains all of the CFU-e progenitors and may be further subdivided into P1-low (CD71med CD150high) and P1-hi (CD71high CD150low), corresponding to early and late CFU-e, respectively; Population 2 (P2 or BFU-e, Kit+ CD55+ CD49fmed/low CD105med/high CD71low CD150high) contains all of the BFU-e progenitors; Population P3 (P3, Kit+ CD55+ CD49fmed/high CD105med/low CD150low CD41low) is enriched for basophil/mast cell progenitors; Population 4 (P4, Kit+ CD55+ CD49fmed/high CD105med/low CD150high CD41+) is enriched for megakaryocytic progenitors; and Population 5 (P5, Kit+ CD55+ CD49fmed/high CD105med/low CD150high CD41–) contains progenitors with erythroid, basophil/mast cell, and megakaryocytic potential (EBMP) and erythroid/ megakaryocytic/ basophil-biased multipotential progenitors (MPPs). This novel approach allows greater precision when analyzing erythroid and other hematopoietic progenitors and also allows for reference to transcriptome information for each flow cytometrically defined population.
Erythropoiesis may be divided into two principal phases: early erythropoiesis and erythroid terminal differentiation (Figure 1)1,2,3. In early erythropoiesis, hematopoietic stem cells commit to the erythroid lineage and give rise to early erythroid progenitors, which were first identified in the 1970s based on their colony-forming potential in semi-solid medium4,5,6,7,8,9. Broadly, erythroid progenitors are divided into two categories: earlier progenitors that each give rise to a "burst" (a large aggregate of smaller erythroid cell clusters), named "burst-forming unit erythroid" or BFU-e4,5,6; and their progeny, which each form a single, small erythroid cell cluster or colony, named "colony-forming unit erythroid" or CFU-e7,8,9. BFU-e and CFU-e do not yet express terminal erythroid genes and are not morphologically recognizable. After a number of self-renewal or expansion cell divisions, the CFU-e undergoes a transcriptional switch in which erythroid genes such as globins are induced, thereby transitioning into erythroid terminal differentiation (ETD)1,10. During ETD, erythroblasts undergo three to five maturational cell divisions before enucleating to form reticulocytes, which mature into red cells.
Erythroblasts during terminal differentiation were originally classified based on their morphology into proerythroblasts, basophilic, polychromatic, and orthochromatic. The advent of flow cytometry allowed their prospective sorting and isolation based on cell size (measured by forward scatter, FSC) and two cell surface markers, CD71 and Ter11911,12,13 (Figure 1). This and similar flow cytometric approaches14 have revolutionized the investigation of the molecular and cellular aspects of ETD, allowing developmental stage-specific analysis of erythroblasts in vivo and in vitro10,15,16,17,18,19,20. The CD71/Ter119 approach is now used routinely in the analysis of erythroid precursors.
Until recently, a similar, accessible flow cytometric approach for direct, high-purity prospective isolation of CFU-e and BFU-e from mouse tissue has eluded investigators. Instead, investigators have used flow cytometric strategies that isolate only a fraction of these progenitors, often in the presence of non-erythroid cells that co-purify within the same flow cytometric subsets21. Consequently, the investigation of BFU-e and CFU-e was limited to in vitro differentiation systems that derive and amplify BFU-e and CFU-e from earlier bone marrow progenitors. It is then possible to apply flow cytometric strategies that distinguish CFU-e from BFU-e in these erythroid progenitor-enriched cultures22,23. An alternative approach makes use of fetal CFU-e and BFU-e, which are highly enriched in the Ter119-negative fraction of the mouse fetal liver at mid gestation10,24,25. Neither of these approaches, however, allow the investigation of adult BFU-e and CFU-e in their physiological state in vivo. The magnitude of the challenge may be appreciated when recalling that, based on colony formation assays, these cells are present in the adult bone marrow at a frequency of only 0.025% and 0.3%, respectively6.
The protocol described here is a novel flow cytometric approach based on single-cell transcriptomic analysis of freshly harvested Kit+ mouse bone marrow cells (Kit is expressed by all of the early progenitor populations of the bone marrow)1. Our approach contains some cell surface markers that were already in use by Pronk et al.21,26. Single-cell transcriptomes were used to determine combinations of cell surface markers that identify erythroid and other early hematopoietic progenitors (Figure 2). Specifically, the CD55+ fraction of lineage-negative (Lin–) Kit+ cells may be subdivided into five populations, three of which yield contiguous segments of the erythroid trajectory (Figure 2). The transcriptomic identities of each of these populations were confirmed by sorting, followed by scRNAseq and projection of the sorted single-cell transcriptomes back onto the original transcriptomic map (the gene expression in each of the five populations and the entire bone-marrow dataset can be explored in https://kleintools.hms.harvard.edu/paper_websites/tusi_et_al/index.html)1. The cell fate potential of each of the populations was confirmed using traditional colony formation assays (Figure 2), as well as a novel high-throughput single-cell fate assay1,27. These analyses show that the novel flow cytometric approach results in high-purity isolation of all the BFU-e and CFU-e progenitors of fresh adult bone marrow and spleen. Specifically, population 1 (P1) contains only CFU-e and no other hematopoietic progenitors, and population 2 (P2) contains all of the bone marrow's BFU-e progenitors and a small number of CFU-e but no other progenitors1. The detailed protocol below is further illustrated with an example experiment in mice that were injected with either saline or with the erythropoiesis-stimulating hormone erythropoietin (Epo).
All experiments were conducted in accordance with animal protocols A-1586 and 202200017 approved by the University of Massachusetts Chan Medical School Institutional Animal Care and Use Committee.
NOTE: Two protocols are detailed here: first, flow cytometric analysis (section 1), followed by protocol adjustments for flow cytometric sorting (section 2). The protocol below uses a flow cytometer/sorter with 10 channels. An example setup is provided in Table 1, referred to in step 1.14.5. It is also possible to run this protocol with only nine channels; see the legend in Table 2.
1. Flow cytometric analysis
2. Protocol adjustments for flow cytometric sorting
The protocol describes a flow cytometric approach to identify BFU-es and CFU-es in freshly harvested bone marrow and spleen cells. It starts with harvesting fresh BM and spleen from mice and immediately placing the tissue on ice. All procedures are conducted in the cold to preserve cell viability. Cells are labeled with a "lineage" antibody cocktail that allows the exclusion of all cells expressing markers of differentiated blood lineages (the FITC- Lin cocktail, Table 3, in the case of flow cytometric analysis; or magnetic bead enrichment, Table 4, for lineage-negative cells). Cells are also labeled with antibodies against markers that allow us to distinguish a number of early progenitor populations within the lineage-negative cell fraction. Labeling is followed by either flow cytometric sorting or analysis. The approach to flow cytometric sorting is similar to that of flow cytometric analysis but is preceded by magnetic-bead enrichment for the Lin– cell fraction to reduce the time and expense of sorting large numbers of cells. When sorting progenitors, BM is pooled from multiple mice, the number depending on the downstream application. Typically, BM from a total of 10 mice (approximately 1 x 109 cells) yields at least 50,000-100,000 cells for each of the P1-hi, P1-low, and P2 populations with a viability of approximately 70%.
The key difference between the spleen- and BM-derived populations identified by this protocol is the much lower frequency of the P6 population and its subpopulations, P4/P5/P3, in spleen-derived cells.
As an example of flow cytometric analysis, the effect of Epo administration in mice was examined. Epo binds to and activates the Epo receptor (EpoR), which is essential for both basal and accelerated erythropoiesis under stress conditions like hypoxia. Hypoxia stimulates an increase in blood Epo levels28,18, in turn promoting the expansion of CFU-e1,29 and erythroid precursors.
Epo (0.25 U/g body weight) or an equal volume of vehicle (saline) were injected subcutaneously into mice once every 24 h for 3 days. The BM and spleen were harvested at 72 h. Epo stimulation led to the expansion of the early and late CFU-e populations (P1-low and P1-hi) in both the BM and spleen (Figure 4 and Figure 5).
Figure 1: Two principal phases of definitive erythropoiesis. Erythropoiesis may be divided into an early phase, where progenitors are defined based on their colony-forming potential, and a later phase known as erythroid terminal differentiation (ETD), where the erythroblast stage is defined based on morphology. Flow cytometric techniques have permitted the prospective isolation of developmentally specific erythroblast stages using FSC, Ter119, and CD71. By contrast, there has not been an equivalent method for isolating BFU-e and CFU-e progenitors directly from tissue with high purity. Note that only a fraction of the BFU-e colony is shown. The scale bar applies to both the CFU-e and BFU-e colonies. Please click here to view a larger version of this figure.
Figure 2: A new flow cytometric strategy for isolating CFU-e and BFU-e directly from tissue. This manuscript describes a protocol for identifying five new flow cytometric populations, P1 to P5, based on an analysis of the scRNAseq data. (A) Force-directed graph projection ("Spring plot") of fresh Balb/C adult bone marrow single-cell transcriptomes. The colors indicate the predicted cell fate potential based on transcriptomic information and the population-balance analysis (PBA) algorithm1. The regions of the plot that correspond to each of the flow cytometric populations P1 to P5 are indicated. The regions delineated are hand-drawn approximations of the original data, which may be found in Tusi et al.1. The correspondence was confirmed by undertaking scRNAseq of each of the sorted P1 to P5 populations1. E = erythroid; Ba = basophilic or mast-cell progenitors; Meg = megakaryocytic progenitors; Ly = Lymphocytic progenitors; D = dendritic progenitors; M = monocytic progenitors; G = granulocytic progenitors. (B) Colony formation assays, replotted from Tusi et al.1. The P1 to P5 populations and CD55+ cells were sorted as described in this protocol, and each population was plated for the relevant colony formation assays. Erythroid colonies were scored on the indicated days. UF = unifocal; MF = multi-focal; CFU-MK = colony-forming unit megakaryocytic; CFU-GM = colony-forming unit granulocytic/monocytic; CFU-GEMM = colony-forming unit granulocytic, erythroid, monocytic, megakaryocytic. (C) Cell surface markers that define each of the five populations, as well as their cell fate potential. Cell fate potential is both the result of transcriptomic information and of the cell fate assays, both colony formation and single-cell fate assays1,27. Of note, the P1 population was split based on the expression of the CD71 marker into P1-hi and P1-low1. Based on data from Tusi et al.1. Please click here to view a larger version of this figure.
Figure 3: Flow cytometric gating strategy. The complete gating strategy for populations P1 to P5. Freshly harvested bone marrow (BM) or spleen (SP) from a Balb/C mouse subcutaneously injected with 100 µL of 0.9% sterile saline are first gated for the main population, discarding debris; this is followed by selecting singlets, discarding dead cells, and gating on cells that are Ter119-negative, lineage-negative, and express Kit. Further subdivision of Ter119–Lin–Kit+CD55+ into P6 and P7 is followed by the final subdivision into populations P1 to P5 (also see Figure 2). The Ter119 channel allows the analysis of erythroblasts on the same sample (using the Ter119/CD71 approach12). However, a separate Ter119 channel is not essential for this protocol, and the Ter119 antibody may be omitted from the master mix and included as part of the Lin-FITC cocktail instead. The exclusion step for Ter119 positive cells is not shown in this gating strategy. Numbers are percentages of each gate within the shown histogram. Please click here to view a larger version of this figure.
Figure 4: Analysis of bone marrow and spleen early erythroid progenitors following Epo stimulation. Representative flow cytometric plots of freshly harvested bone marrow (BM) or spleen (SP). Balb/C mice were injected subcutaneously with either 0.25 U/g body weight of Epo or with an equivalent volume of saline. Please click here to view a larger version of this figure.
Figure 5: The response of BM and spleen progenitors to Epo in vivo. Summary statistics showing changes in the absolute cell number in populations P1 to P5 following Epo injection in freshly harvested (A) bone marrow and (B) spleen cells; *p = 0.02. Experiment as described in Figure 4. n = 4 mice in each group. To calculate the absolute cell number, the frequency of cells in each population in either the BM or spleen (from Figure 4) was multiplied by the total number of BM or spleen cells harvested from each mouse and divided by the mouse weight in grams. Of note, only early CFU-e in the BM reached statistical significance (p = 0.02), likely a combination of a relatively low Epo dose and a small number of mice in each group. Please click here to view a larger version of this figure.
Detector Array | PMT | Dichroic mirror | Bandpass filter | Fluorophore detected |
Blue Laser (488 nm) | A | 685LP | 695/40 | |
B | 505LP | 530/30 | Lin-FITC (+Ter119-FITC) | |
C | 488/10 | |||
Violet Laser (405 nm) | A | 760LP | 800/80 | |
B | 630LP | 660/20 | CD150-BV650 | |
C | 595LP | 605/20 | CD41-BV605 | |
D | 475LP | 525/20 | ||
E | 450/50 | CD49f-BV421 | ||
Red Laser (640 nm) | A | 755LP | 785/50 | cKit-APC-Cy7 |
B | 700 LP | 720/30 | ||
C | 675/14 | CD55-AF 647 | ||
UV Laser (355 nm) | A | 505LP | 525/20 | |
B | 420LP | 450/50 | DAPI | |
C | 379/28 | Ter119-BUV 395 | ||
YG Laser (561 nm) | A | 755LP | 780/60 | CD71-PE-Cy7 |
B | 685LP | 710/50 | ||
C | 635LP | 660/20 | ||
D | 600LP | 610/20 | ||
E | 570LP | 580/20 | CD105-PE |
Table 1: Example of the channel layout in an LSRII cytometer. The panel of antibodies used in the described protocol and their arrangement in the LSRII flow cytometer to collect the data.
Reagent or Antibody | Conjugate | Stock conc. (mg/mL) | Dilution factor | Final conc. in cell suspension (µg/mL) | Clone |
CD71 | PE/Cy7 | 0.2 | 2000 | 0.1 | RI7217 |
TER-119 | BUV395 | 0.2 | 100 | 2 | TER-119 |
CD55 (DAF) | AF647 | 0.5 | 50 | 10 | RIKO-3 |
CD105 | PE | 0.2 | 50 | 4 | MJ7/18 |
CD150 (SLAM) | BV650 | 0.1 | 50 | 2 | TC15-12F12.2 |
CD49f | BV421 | 0.025 | 50 | 0.5 | GoH3 |
CD41 | BV605 | 0.2 | 100 | 2 | MWReg30 |
CD117 (cKit) | APC/Cy7 | 0.2 | 200 | 1 | 2B8 |
Lin cocktail | FITC | 0.08 (for each antibody in the cocktail) | 83 | 1 (for each antibody in the cocktail) | |
Rabbit IgG | 10 | 50 | 200 |
Table 2: Antibody master-mix components. Composition of the antibody master mix. The Ter119 channel allows the analysis of erythroblasts on the same sample (using the Ter119/CD71 approach12). However, a separate Ter119 channel is not essential for this protocol, and the Ter119 antibody may be omitted from the master mix and included as part of the Lin-FITC cocktail instead.
Reagent or Antibody | Conjugate | Stock conc. (mg/mL) | Clone |
Ly-6G and Ly-6C | FITC | 0.5 | RB6-8C5 |
CD11b | FITC | 0.5 | M1/70 |
CD19 | FITC | 0.5 | 1D3 |
CD4 | FITC | 0.5 | RM4-5 |
CD8a | FITC | 0.5 | 53-6.7 |
F4/80 | FITC | 0.5 | BM8 |
Table 3: Lin cocktail components. Composition of the Lin cocktail used as part of the antibody master mix (see details of the master mix in Table 2).
Reagent or Antibody | Conjugate | Stock conc. (mg/mL) | Dilution factor | Final conc. in cell suspension (µg/mL) | Clone |
Ly-6G and Ly-6C | Biotin | 0.5 | 200 | 2.5 | RB6-8C5 |
CD11b | Biotin | 0.5 | 200 | 2.5 | M1/70 |
CD19 | Biotin | 0.5 | 200 | 2.5 | 1D3 |
CD4 | Biotin | 0.5 | 200 | 2.5 | RM4-5 |
CD8a | Biotin | 0.5 | 200 | 2.5 | 53-6.7 |
F4/80 | Biotin | 0.5 | 200 | 2.5 | BM8 |
TER-119 | Biotin | 0.5 | 200 | 2.5 | TER-119 |
Normal rat serum | 50 |
Table 4: Antibodies for magnetic-bead enrichment. Antibodies used to enrich the Lin– fraction of either the BM or spleen prior to cell sorting.
The ability to prospectively isolate BFU-e and CFU-e progenitors directly from fresh tissue with high purity had previously eluded investigators. Our novel approach, validated using scRNAseq and cell fate assays1,27, now offers the tools to do this.
There are a number of key points for successfully executing both the sorting and the analytical protocols. First, the cells need to be spun at 900 x g to prevent the loss of low-density cells. Many cells at the CFU-e stage are large, low-density cells that may otherwise be lost. Second, it is necessary to use large orifice tips when pipetting up and down to help break the cell aggregates and prevent the loss of fragile P1 and P2 cells. This step is important since many CFU-e progenitors are associated with erythroblastic-island macrophages (EBIs) and would be lost if not successfully dissociated. Third, the cells are incubated with an EDTA-containing buffer ("SB5") prior to staining, which helps to dissociate the EBIs. Fourth, during flow cytometric data acquisition, the rate of acquisition should not exceed 7,000 events/s or the recommended rate for the flow cytometer instrument that is being used.
The P1 and P2 populations contain all of the CFU-e and BFU-e progenitors in the bone marrow, respectively, and contain no other progenitor cell types1. The P1 population is further divided into early CFU-e (P1-low) and late CFU-e (P1-hi) based on the expression of CD711. These highly pure flow cytometric populations provide a complete picture of the BFU-e and CFU-e progenitor populations in vivo.
One drawback is that the P2 population does contain a small fraction of early CFU-e progenitors. It does, however, contain all of the BFU-e cells in the BM and no other cell types. The P1 population contains all of the CFU-e in the BM, other than the small fraction that is present in P2. Unlike the P1 and P2 populations, P3 to P5 are highly enriched, but not pure, progenitor populations.
Applying this flow cytometric tool to mouse models of erythropoiesis would now allow the determination of cellular and molecular responses during physiological and pathological perturbations to erythropoiesis. In the shown example, mice were injected with the hormone Epo. This approach may be used in genetically modified mice, for example, in mouse models of hemoglobinopathies, for a more accurate molecular analysis of their altered erythroid progenitors in vivo.
DATA AVAILABILITY:
Additional data associated with the experiments in Figure 4 and Figure 5 are available upon request.
The authors have nothing to disclose.
This work is supported by NIH grants R01DK130498, R01DK120639, and R01HL141402
0.5 M EDTA, pH 8.0 | Life Technologies | 15575020 | |
1000 µL large orifice tips | USA sceintific | 1011-9000 | |
Alexa Fluor 647 anti-mouse CD55 (DAF) Antibody | BioLegend | 131806 | |
APC/Cyanine7 anti-mouse CD117 (c-kit) Antibody | BioLegend | 105826 | |
Biotin-CD11b | BD Biosciences | 557395 | M1/70 (clone) |
Biotin-CD19 | BD Biosciences | 553784 | 1D3 (clone) |
Biotin-CD4 | BD Biosciences | BDB553045 | RM4-5 (clone) |
Biotin-CD8a | BD Biosciences | BDB553029 | 53-6.7 (clone) |
Biotin-F4/80 | Biolegend | 123106 | BM8 (clone) |
Biotin-Ly-6G and Ly-6C | BD Biosciences | 553125 | RB6-8C5 (clone) |
Biotin-TER-119 | BD Biosciences | 553672 | TER-119 (clone) |
Bovine Serum Albumin | Sigma aldritch | A1470 | |
Brilliant Violet 421 anti-human/mouse CD49f Antibody | BioLegend | 313624 | |
Brilliant Violet 605 anti-mouse CD41 Antibody | BioLegend | 133921 | |
Brilliant Violet 650 anti-mouse CD150 (SLAM) Antibody | BioLegend | 115931 | |
BUV395 Rat Anti-Mouse TER-119/Erythroid Cells | BD Biosciences | 563827 | |
ChromPure Rabbit IgG, whole molecule | Jackson ImmunoResearch Laboratories | 011-000-003 | |
DAPI (4',6-Diamidino-2-Phenylindole, Dihydrochloride) | Life Technologies | D1306 | |
Digital DIVA hardware and software for LSR II | BD Biosciences | ||
FITC anti-mouse F4/80 Antibody | BioLegend | 123108 | |
FITC Rat Anti-CD11b | BD Biosciences | 557396 | |
FITC Rat Anti-Mouse CD19 | BD Biosciences | 553785 | |
FITC Rat Anti-Mouse CD4 | BD Biosciences | 553047 | |
FITC Rat Anti-Mouse CD8a | BD Biosciences | 553031 | |
FITC Rat Anti-Mouse Ly-6G and LY-6C | BD Biosciences | 553127 | |
FlowJo software | FlowJo | version 10 | Flow cytometer analysis software |
LSR II digital multiparameter flow cytometer analyzer | BD Biosciences | Flow cytometer | |
NewlineNY Stainless Steel Hand Masher & Bowl, Mortar and Pestle Set | Amazon | ||
Normal rat serum | Stem Cell Technologies | 13551 | |
PE anti-mouse CD105 Antibody | BioLegend | 120408 | |
PE/Cyanine7 anti-mouse CD71 Antibody | BioLegend | 113812 | |
Phosphate Buffered Saline, 10x Solution | Fisher scientific | BP3994 | |
Streptavidin Nanobeads | BioLegend | 480016 | Magnetic beads |