Identification of a Siglec-F+ granulocyte-macrophage progenitor

In recent years multi-parameter flow cytometry has enabled identification of cells at major stages in myeloid development; from pluripotent hematopoietic stem cells, through populations with increasingly limited developmental potential (common myeloid progenitors and granulocyte-macrophage progenitors), to terminally differentiated mature cells. Myeloid progenitors are heterogeneous, and the surface markers that define transition states from progenitors to mature cells are poorly characterized. Siglec-F is a surface glycoprotein frequently used in combination with IL-5 receptor alpha (IL5Rα) for the identification of murine eosinophils. Here, we describe a CD11b+ Siglec-F+ IL5Rα− myeloid population in the bone marrow of C57BL/6 mice. The CD11b+ Siglec-F+ IL5Rα− cells are retained in eosinophil deficient PHIL mice, and are not expanded upon overexpression of IL-5, indicating that they are upstream or independent of the eosinophil lineage. We show these cells to have GMP-like developmental potential in vitro and in vivo, and to be transcriptionally distinct from the classically described GMP population. The CD11b+ Siglec-F+ IL5Rα− population expands in the bone marrow of Myb mutant mice, which is potentially due to negative transcriptional regulation of Siglec-F by Myb. Lastly, we show that the role of Siglec-F may be, at least in part, to regulate GMP viability.

mM EDTA, 1 µg/mL propidium iodide (Sigma) to enable identification and exclusion of dead cells. Stained cells were analyzed on a BD LSR Fortessa X-20 flow cytometer (BD Biosciences) or sorted on a BD FACS Aria III (BD Biosciences).
For cell sorting of CMP, GMP, EoP and Eos for RNAseq, red cells and other mature blood cells were removed by positive selection as follows: total bone marrow was incubated with a cocktail of antibodies against mature cell markers (including Ter119) and selected with Biomag goat anti-rat IgG magnetic beads (Qiagen). Flow-through cells (negative cells) were subsequently stained with fluorescently-conjugated antibodies against surface markers and sorted as described below. Prior to sorting of the Siglec-F+IL5Rα-population, Siglec-F+ cells were enriched through staining with a PE-conjugated anti-Siglec-F antibody, and positive selection with anti-PE microbeads (Miltenyi Biotec). Due to the fluorophore combinations used in this sort, we were unable to include an antibody against IL5Rα. The Siglec-F+IL5Rα-fraction was sorted using additional gates to exclude known IL5Rα+ populations on the basis of other markers, such as excluding Eos and EoPs on the basis of scatter. A terminal FSC Int SSC Lo gate was also applied to positively sort Siglec-F+IL5Rα-cells.

Subsequent statistical tests (ANOVA, corrections for multiple comparisons using
Sidak's method and unpaired Student's t tests with Welch's correction) and graphs were generated with Prism (GraphPad Software). Student's t test p-values were adjusted for multiple testing according the method of Bonferroni using the p.adjust function in R.

Annexin-V staining:
Cells were prepared and stained for flow cytometry as described above. After the antibody washes, cells were washed once with 1x Annexin-V binding buffer (10mM Hepes pH 7.4, 140mM NaCl, 2.5mM CaCl 2 ) and incubated in 1x Annexin-V binding buffer containing 1/100 Annexin-V-FITC (WEHI) and 1 µg/mL propidium iodide (Sigma) for 10 minutes at room temperature. Stained cells were analyzed by flow cytometry on an LSR Fortessa X-20 flow cytometer as described above.

In vivo anti-Siglec-F antibody injection:
Eight-to-twelve-week-old C57BL/6 mice were injected intraperitoneally every second day with 20 µg anti-Siglec-F (clone 9C7, a gift from Dr. James Paulson, The Scripps Research Institute, La Jolla, CA, USA) or rat IgG2b isotype control antibody (clone LTF-2, Tonbo Biosciences) a total of 4 times. Tissues were harvested 24 hours after the final administration. Bone marrow, blood and spleens were processed and stained for flow cytometry. Surface and intracellular anti-Siglec-F staining was performed with clone E50-2440 or IgG2a isotype control, and cell viability was assessed using Fixable Viability Dye eFluor 506 (Thermo Fisher Scientific). Following surface staining, cells were fixed and permeabilized with BD Cytofix/Cytoperm (BD Biosciences). Cells were then washed and stained for intracellular Siglec-F in BD Perm/Wash buffer. Cells were analyzed on a BD LSR II flow cytometer (BD Biosciences).

Antibodies:
Cells were stained with the following rat anti-mouse monoclonal antibodies prior to

Cytocentrifuge preparations and May Grünwald Giemsa stains:
Sorted cells were cytocentrifuged onto glass slides using a Shandon Cytospin 3 cytocentrifuge (Thermo Fisher Scientific) for 5 min at 500 rpm. Slides were air dried, fixed with 100% methanol for 10 min and stained with May Grünwald's stain (Merck) for 5 min. Slides were immediately transferred into 5% Giemsa solution (in pH 6.8 buffered water, Merck) for 20 min, washed twice for 30 s in pH 6.8 buffered water, washed for 1 min in dH 2 O, then air-dried. Slides were coverslipped with DPX neutral mounting medium (Thermo Fisher Scientific). Cells were imaged on a Nikon 90i microscope fitted with a DXM1200C camera, at 1000x magnification.

In vitro developmental potential (colony forming) assays:
GMPs, EoPs and CD11b+Siglec-F+IL5Rα-cells were sorted from the bone marrow (following red cell lysis) of C57BL/6 mice, using the stains and surface markers as described above. Colony assays were performed in triplicate as described by [5].

In vivo developmental potential assays:
Total bone marrow was flushed from the hips, femurs and tibiae of three 14 week old UBC-GFP mice in PBS/2% FCS, triturated using a 23 gauge needle, and overlaid onto 60% Percoll in PBS. Cells were centrifuged at room temperature at 400 g for 25 min. Cells at the Percoll interface were collected, washed twice with PBS/2% FCS,

RNA-sequencing:
Siglec-F+IL5Rα-cells, CMPs, GMPs EoPs and Eosinophils were sorted from the BM of 6-10 week old C57BL/6 and 6 week old Myb Plt4/Plt4 mice as described above, Reads were aligned to the Mus musculus genome (Ens84, GRCm38) using the Rsubread package [6] and assigned to genes by the featureCounts function [7] using the Ensembl annotation. Filtering and normalization used the edgeR package [8].
Genes with a count per million (CPM) of at least 1 in 2 or more samples were retained for further analysis. Compositional differences between libraries were normalized using the trimmed mean of M-values (TMM) method [9]. Subsequent differential expression analysis was performed using the limma package [10]. Counts were transformed to log2-CPM values (with an offset of 0.5) with associated observational and sample-specific weights obtained from the voomWithQualityWeights method [11] assuming a linear model [12] with effects for cell type. p-values were corrected for multiple testing using the method of Benjamini and Hochberg [13].
Contrasts between the different cell-types were estimated and differential expression was tested relative to a fold-change of 1.5 using TREAT [14] and a false discovery rate (FDR) cut-off of 0.05. Heatmaps of log 2 counts per million (CPM) were generated for various sets of genes (100 most variable across all samples, 100 most differentially expressed genes based on FDR) using the heatmap.2 function from the gplots R package. Multidimensional scaling of the counts using the 500 most variable genes between each pair of samples was used to explore the relationships between samples.
This dataset has been submitted to the Gene Expression Omnibus under the accession number GSE107495.

Gene ontology analyses:
Gene ontology analyses were performed on lists of differentially expressed genes using the MSigDB tab of the Broad Institute's online GSEA software, selecting for analyses on C5 GO_Gene_Sets [15,16]. The curated list of transcription factors (TFs) published in [17] were used in the identification of differentially expressed transcription factors.