Isolation of the stromal-vascular fraction of mouse bone marrow markedly enhances the yield of clonogenic stromal progenitors

Abstract

The low incidence of CFU-F significantly complicates the isolation of homogeneous populations of mouse bone marrow stromal cells (BMSCs), a common problem being contamination with hematopoietic cells. Taking advantage of burgeoning evidence demonstrating the perivascular location of stromal cell stem/progenitors, we hypothesized that a potential reason for the low yield of mouse BMSCs is the flushing of the marrow used to remove single-cell suspensions and the consequent destruction of the marrow vasculature, which may adversely affect recovery of BMSCs physically associated with the abluminal surface of blood vessels. Herein, we describe a simple methodology based on preparation and enzymatic disaggregation of intact marrow plugs, which yields distinct populations of both stromal and endothelial cells. The recovery of CFU-F obtained by pooling the product of each digestion (1631.8 + 199) reproducibly exceeds that obtained using the standard BM flushing technique (14.32 + 1.9) by at least 2 orders of magnitude (P < .001; N = 8) with an accompanying 113.95-fold enrichment of CFU-F frequency when plated at low oxygen (5%). Purified BMSC populations devoid of hematopoietic contamination are readily obtained by FACS at P0 and from freshly prepared single-cell suspensions. Furthermore, this population demonstrates robust multilineage differentiation using standard in vivo and in vitro bioassays.

Figure 1

BM plug isolation and histologic assessment of intact vascular structures in BM plugs. (A) Representative images of denuded bones, removal of metaphysis, and isolated intact BM plug. (B-D) Resin-embedded sections of BM plugs (B,D) and remaining bone tissue (mid-diaphysis; C) after removal of marrow plug were sectioned as 5-μm-thick longitudinal sections and stained with hematoxylin and eosin, demonstrating intact vascular structures. (E) Whole-mount image of BM plug stained with a combination of the endothelial cell-reactive antibodies MECA32 and VE-Cadherin reveals a well-organized vascular reticulum throughout the marrow. BM plugs were stained with DRAQ5 to provide a nuclear counterstain and then immersed in prolong gold anti-fade mounting medium (Invitrogen). After applying a glass coverslip and sealing with nail hardener, specimens were inverted and allowed to cure overnight in the dark at room temperature before confocal imaging. Images were collected using 63× oil immersion objective of a Leica TCS SP5 confocal microscope and processed with the Leica LAS AF lite Version 2.6.7266 software. Z-stacked images were collected in 0.2-μm slices at depths of 15 to 25 μm with a pinhole of 1 (×63).

Figure 2

Evaluation of clonogenic stromal progenitor cells (CFU-F) recovered from sequential enzymatic disaggregation of BM plugs. (A) Average mononucleated cell yields obtained from either standard flushing methods or from each successive digestion (n = 4). (B) Incidence of CFU-F obtained from either standard flushing technique (5 × 10⁶ mononuclear cells/well) or from each fraction of digested marrow plugs (2.5 × 10⁵ mononuclear cells/well) plated in triplicate. (C) Recovery of CFU-F from flushed BM and each fraction of DBM calculated per femur-tibia pair. (D) Incidence of CFU-F obtained from flushed BM versus the pool of DBM fractions (1-3; n = 8). (E) Recovery of CFU-F from flushed BM and the pool of DBM fractions (1-3) calculated per femur-tibia pair. Only colonies containing more than 50 stromal cells are scored. CFU-F data are presented both as incidence of clonogenic cells (CFU-F/1 × 10⁶ mononuclear cells) and as the total number of CFU-F recovered from a given number of bones (CFU-F per total nucleated cells). (F) CFU-F incidence quantified by limit dilution analysis. Limit dilution assays were performed by plating BM mononuclear cells at various doses (500, 1000, 2500, 5000, and 10 000 cells/well in 24-well plates) with 24 replicates per dilution (n = 3). Plates were scored and negative wells enumerated. Data were analyzed with L-Calc 1.1.1 software (StemCell Technologies) and plotted as a negative linear relationship to identify the frequency of colony-forming cells. Data are mean ± SD. Statistical analysis of CFU-F incidence was performed with SigmaStat Version 3.5.

Figure 3

Immunostaining of P0 cultures and isolation and characterization of BM vascular endothelial cells. (Ai-vi) In situ staining of P0 cultures plated on fibronectin-coated chamber slides (LabTek; Nunc), cultured in EGM-2MV for 5 to 7 days. Vascular endothelial cells were identified by staining with a combination of VE-Cadherin–Alexa 488 and MECA32–Alexa 488 antibodies, and stromal cells were stained with rat anti–mouse PDGFRα/β (purified) antibodies and revealed with donkey anti–rat Cy3 and counterstained with DAPI. IgG2a and IgG1 isotypes were used for controls (Aiv-v). (B) Gating strategy for FACS purification of vascular endothelial cells from P0 cultures plated on fibronectin-coated 10-cm² dishes at 1 × 10⁶ mononuclear cells/cm² and cultured in EGM-2MV for 5 to 7 days and stained as described in “In situ staining.” (C) Phase-contrast images of Lin⁻CD105^BRIGHTPDGFRαβ⁻ cells at passage 3 and functional analysis of DiI-Ac-LDL uptake. (D) In situ staining of Lin⁻CD105^BRIGHTPDGFRαβ⁻ cells at passage 3 for endothelial markers, including VEGFR2 (Di), VE-Cadherin (Dii), CD31 and eNOS (Diii), MECA32 (Div), CD105 (Dv), and isotype controls (Dvi-vii). Nuclei were counterstained DAPI. Imaging was performed on an inverted fluorescence microscope (Olympus BX51) at both ×10 and ×40 (original magnification) and captured with an Olympus DP71 camera.

Figure 4

Isolation and phenotypic analysis of long-term cultured Lin⁻PDGFRαβ⁺ BMSCs. (Ai-ii) Representative gating strategy of viable cells for FACS isolation of Lin⁻PDGFRαβ⁺ cells from P0 cultures. (Aii-iii) Phase-contrast image and PDGFRβ immunostaining at passage 3. (B) FACS analysis of MSC markers in cultures of passage 3 Lin⁻PDGFRαβ⁺ cells (n = 3). FACS analysis demonstrating phenotypic differences between flushed BM (C) and DBM cells (D). FACS data were collected on BD LSR II, and postacquisition analysis was performed with BD FACS Diva Version 6.1.3. Data are mean ± SD.

Figure 5

Multilineage differentiation capacity of Lin⁻PDGFRαβ⁺ BMSCs. (Ai-iv) Histology of subcutaneous transplants of either empty scaffolds (Ai) or Lin⁻PDGFRαβ⁺ MSC (Aii-iv). Gelfoam scaffolds loaded with Lin⁻PDGFRαβ⁺ MSCs were either decalcified and embedded in paraffin for hematoxylin and eosin staining (Aii-iii) or nondecalcified and embedded in methylmethacrylate resin for Von Kossa staining (Aiv). (B) Three-dimensional pellet cultures of Lin⁻PDGFRαβ⁺ MSCs embedded in paraffin and stained with Toluidine blue (0.1% weight/volume; Bi), Alcian blue (Bii), collagen type II (Biv), and mouse IgG1 isotpe (Biii). (C) Oil red O staining of Lin⁻PDGFRαβ⁺ MSCs after adipogenic differentiation for 14 days.

Figure 6

Prospective isolation of Lin⁻PDGFRαβ⁺ clonogenic progenitors from DBM. (A) Gating strategy (left panel) and FACS analysis of whole BM from C57Bl/6 and BALB/c mice demonstrating the frequency of Lin⁻PDGFRαβ⁺ BMSCs obtained from either standard flushing or sequential enzymatic disaggregation of BM plugs. (B) Gating strategy (left panel) for prospective isolation of Lin⁻PDGFRαβ⁺ population from C57Bl/6 and BALB/c inbred mouse strains. (C-D) Incidence of CFU-F from prospective isolation of Lin⁻PDGFRαβ⁺ BMSCs from C57Bl/6 and BALB/c mice. Colonies > 50 stromal cells; clusters represent 10 to 49 stromal cells.

Figure 7

PDGFRαβ⁺ stromal cells are localized to perivascular and intersinusoidal regions in vivo. Whole-mount staining of BM plugs. Vascular endothelial cells were identified with VE-Cadherin–Alexa 488 and MECA32–Alexa 488 antibodies (A), and stromal cells were identified with PDGFRα/β antibodies and revealed with donkey anti–rat Cy3 (C). Nuclei were counterstained with DRAQ5. Z-stack merged image (B) and single-step merged image identifying perivascular (white asterisk) and intersinusoidal (white arrow) localization. Images were collected using a 63× oil immersion objective on zoom factor of 3 with a Leica TCS SP5 confocal microscope and processed with the Leica LAS AF lite Version 2.6.7266 software. Z-stacked images were collected in 0.2-μm slices at depths of 15 to 25 μm with a pinhole of 1.