Identification and specification of the mouse skeletal stem cell

Abstract

How are skeletal tissues derived from skeletal stem cells? Here, we map bone, cartilage, and stromal development from a population of highly pure, postnatal skeletal stem cells (mouse skeletal stem cells, mSSCs) to their downstream progenitors of bone, cartilage, and stromal tissue. We then investigated the transcriptome of the stem/progenitor cells for unique gene-expression patterns that would indicate potential regulators of mSSC lineage commitment. We demonstrate that mSSC niche factors can be potent inducers of osteogenesis, and several specific combinations of recombinant mSSC niche factors can activate mSSC genetic programs in situ, even in nonskeletal tissues, resulting in de novo formation of cartilage or bone and bone marrow stroma. Inducing mSSC formation with soluble factors and subsequently regulating the mSSC niche to specify its differentiation toward bone, cartilage, or stromal cells could represent a paradigm shift in the therapeutic regeneration of skeletal tissues.

Fig 1. Bone and cartilage are derived from clonal, lineage-restricted progenitors

(A) Micrographs: 6-week old Rainbow Actin-Cre-ERT mouse femur, following TMX-induction at P3, shows clonal expansion at the growth plate. Fluorescent microscopy (left), pentachrome stain (middle), and dissection microscope (right). Scale bar: 500µM. Representative of 10 replicates. (B) FACS plots: cells isolated from three different parts of the femur illustrate that [AlphaV+] is most prevalent in the growth plate (uppermost horizontal panel in the middle) (p<0.001, ANOVA, n=3). DN = double negative, negative for Thy and 6C3 surface expression. (C) Scheme of experiment: Actin-Cre-ERT transgenic mouse was crossed with rainbow reporter gene mouse. Cre recombination of offspring was induced by TMX induction on E15, P3 and postnatal week 6. Bones were harvested 6 weeks post induction. (D) Graphic representation illustrating different numbers of clones present, which span the bone and cartilage. Representative of sections with 10 mice. See also Figure S1. (E) FACS gating strategy for isolation of eight distinct skeletal tissue subpopulations obtained from the [AlphaV+] subset. a=BCSP, b=BLSP, c=6C3, d=HEC, e=mSSC, f=pre-BCSP, g=PCP, h=Thy. Representative of 50 replicates. (F) P3 Femur stained with Movat’s pentachrome (top). Sections stained with pentachrome of tissue grafts following cell transplant beneath the renal capsule (bottom). Populations e (mSSC), f (pre-BCSP) and a (BCSP) can reconstruct entire bone, consisting of bone, cartilage and a functional marrow cavity. Populations b (BLSP), c (6C3), d (HEC), and h (Thy) formed bone only. Population g (PCP) formed cartilage with a minimal of bone. Scale bar: 200µM. Representative of 3–20 experiments. (G) Graph depicting the percentage tissue composition [bone (yellow), marrow (red), and cartilage (blue)] of each of the explanted grafts a to h. Representative of 3–20 experiments. (H) Scheme of experiment: 20,000 cells of each subpopulation of [AlphaV+] were isolated from the long bones of GFP-labeled P3 mice using FACS. Purified GFP+ cells were then transplanted beneath the kidney capsules of recipient mice. One-month later, the grafts were explanted. See also Figures S1, S3, S4, S6.

Fig 2. Identification of the mSSC (mouse Skeletal Stem Cell)

(A) Scheme of experiment: CD200+TN [CD45⁻ Ter119⁻ Tie2⁻ AlphaV⁺ Thy⁻ 6C3⁻ CD105⁻ CD200⁺ ] cells were isolated from femora of GFP+ mice at P3. (i) Purified mSSCs were seeded and harvested on days 0, 11, 25 for FACS analysis. On day 25 following re-fractionation of the cells by FACS, 20,000 cells of each subset (mSSC, BCSP, Thy, 6C3) were transplanted beneath the kidney capsules of recipient mice. The grafts were explanted 1 month later. (ii) Purified GFP-labeled mSSCs were also directly transplanted beneath the kidney capsules of recipient mice. 1 month post transplantation, the grafts were explanted. (B) FACS analysis of cultured mSSC on days 0, 11 and 25 in culture (i, ii). Transplanted mSSC (red box) and BCSP (green box) formed bone, cartilage and a marrow cavity. Thy (blue box) and 6C3 (orange box) formed bone only, without a marrow cavity (iii). Scale bar: 500 µm (iii, upper panel), 200 µm (iii, lower panel). Representative of 3 replicates/subpopulation. (C) (i, ii) FACS analysis of explanted kidney capsule grafts, in which highly purified populations of GFP-labeled[CD45⁻ Ter119⁻ Tie2⁻ AlphaV⁺ Thy⁻ 6C3⁻ CD105⁻ CD200⁺ ](mSSC) cells were transplanted beneath the kidney capsule: graft consisted of 7 downstream subpopulations (blue, red and orange boxes). Brightfield micrograph of pentachrome-stained explant (iii, far lower left) demonstrates that mSSCs are capable of generating bone, cartilage and marrow. Immunostained explant (purple box, zoomed) shows that mSSCs are capable of generating cells that express Thy and 6C3 (iii, middle; Thy = red, 6C3 = white). Merged image (iii, extreme right). Fluorescent image of GFP+ graft (iii far upper left). Scale bar: 500 µm (iii, upper panel), 100 µm (iii, lower panel), 50 µm (iii). Representative of 3 replicates per transplanted subpopulation. (D) Scheme of experiment: Unsorted cells from the long bones of RFP+ P3 mice served as feeder cells. Cells from the long bones of P3 GFP+ mice were isolated following mechanical and enzymatic dissociation and mSSCs were obtained following FACS. (i) A single GFP-labeled mSSC was co-transplanted with 5,000 RFP(+) feeder cells beneath the renal capsule of immunodeficient mice. (ii) A single purified GFP-labeled mSSC was plated per well of a 96-well culture dish. Following 14 days, formed colonies were counted, harvested, re-sorted using FACS and a single purified GFP-labeled mSSC was again plated per well of a 96-well culture dish and the assay repeated. (E) In vitro, colony formation assays were performed by plating a single mSSC in each well of a 96-well culture dish. (i) Representative micrograph of a primary colony depicted at 14 days post plating. (ii) Passaging of primary colonies resulted in the formation of secondary colonies with similar morphology to primary colonies. (iii) Primary colonies stained positive for anti-collagen 2 (purple), anti-osteocalcin (green), and DAPI (blue) following immunofluorescent staining. (iv) Vertical panel on right depicts FACS analysis of initial mSSC cells isolated (top), and subsequent primary colony (middle), and secondary colony cell (bottom). Scale bar: 500 µm (i, ii), 100 µm (iii). Representative of 10 assays. (F) Microscopy of explanted grafts (as per Figure 2D (i)). The extent of the in vivo colony formation from GFP-labeled mSSCs is outlined by a yellow broken line (i, iii, v) or black broken line (ii, iv, vi). Clonally expanded, GFP-labeled mSSCs are seen as green cells (fluorescent image, i, iii, v). Corresponding brightfield micrographs (ii, iv, vi). Fluorescent imaging of transverse sections of grafts (iii, v). Corresponding brightfield micrographs of pentachrome-stained sections demonstrates clonally expanded cells are fated into bone (yellow) and cartilage (blue) (iv, vi). Graph (vii) is representative of the contribution by GFP or RFP cells (per cell transplanted) to bone or cartilage formation (mean SEM). Scale bar: 200 µm (i-vi). Representative of 5 transplants of each assay. (G) Schematic representation of the skeletal stem cell lineage tree. The mSSC occupies the apex of this hierarchal tree and is multipotent, capable of self-renewal, and differentiation into more lineage restricted progenitor cells (pre-BCSP and BCSP). The mSSC, pre-BCSP and BCSP are capable of giving rise to bone, cartilage and hematopoietic supportive stroma. The immunophenotype of each cell is shown.

Fig 3. The mSSC niche is composed of other skeletal-lineage cells

(A) Scheme of experiment: Long bones of P3 mice were harvested as previously described. The cell suspension was sorted by FACS to obtain mSSC; BCSP; Thy(+), which encompasses the PCP, Thy and BLSP subsets; and 6C3(+), which encompasses 6C3 and HEC. Cell subpopulations were prepared for single cell RNA sequencing. (B) Hierarchical clustering of single cell RNA sequencing data demonstrate four molecularly distinct patterns of single cell transcriptional expression between mSSC, BCSP, Thy(+) and 6C3(+). (C) Percentage transcriptional expression of morphogen (left circle in each Venn diagram), receptor (right circle in each Venn diagram) or both (overlapping central portion of the Venn diagram) on single cell RNA sequencing. Note the percentage denotes the percentage of cells within each subset (mSSC, BCSP, Thy(+) and 6C3(+)), which express the relevant gene sequence. (Bottom) The patterns show the potential for paracrine (right) and autocrine (left) signaling. (D) Gene expression levels of Wnt-associated genes in skeletal populations as determined using the Gene Expression Commons analysis platform. The range of transcriptional expression is illustrated by a color change as depicted on the extreme right of the figure (i.e. dark purple correlates to high expression, whereas dark blue correlates to low expression). Heat map shows high expression of both Wnt ligands (Wnt3a, Wnt4, Wnt5a) and receptors (Fzd5-9), demonstrating that Wnt signaling may be actively involved in skeletal progenitor function. Fzd = frizzled receptor. These data were compiled in triplicate, with 10,000 cells of each subset analyzed in each sample. (E) Ligand-receptor interaction maps demonstrating that the skeletal stem/progenitor cells can act as their own niche and signal through each other to promote skeletogenesis. Gene expression analysis of microarray data extracted from skeletal subsets illustrating ligands in the left column and cognate receptors in the right column. The connecting arrows indicate possible ligand-receptor interaction pathways. Note: GDF= growth and differentiation factor, VCAM-1= vascular cell adhesion molecule-1. Data were compiled from triplicate samples, with 10,000 cells of each subset analyzed in each sample. (F) Diagram illustrating potential signaling pathways influencing activity of skeletal stem / progenitor cells. Paracrine and/or autocrine signaling may occur in the skeletal stem cell niche and regulate cell activity and maintenance. (G) Fluorescent micrographs illustrate colony morphology of mSSCs post culture with morphogen (BMP2 / TGFβ /TNFα) supplementation/control (right). Graph shows the number of mSSC present following culture for 14 days under the different conditions (left). rhBMP-2 supplementation was associated with significant amplification of the mSSC populations in vitro (bottom left) in comparison to control, non-supplemented media (top left) (mean SEM, p<0.01, t-test, n=3). Supplementation with TGFβ or TNFα resulted in altered colony morphology (upper and lower right). These results show that niche signaling can influence mSSC proliferation. Scale bar: 200µm. (H) Graph illustrates the effect of rhBMP2 titration on mSSC proliferation in culture (left). The effect of BMP-2 was significantly greater than control at each of the following concentrations: 50ng/mL (mean SEM, p<0.001, ANOVA), 500ng/mL (mean SEM, p<0.01, ANOVA). Phase images illustrate the colonies of cells derived from the mSSC (indicated by arrowhead in control, 50pg/mL, 500pg/mL, 5ng/mL and by a broken line in 50ng/mL and 500ng/mL) (right). Scale bar: 500µM, n=3. (I) Ligand-receptor interaction maps show gene expression levels of BMP antagonists gremlin-2 and Noggin (right top and bottom respectively). There is increased expression of the receptors for the systemic hormones leptin and thyroid stimulating hormone (TSH) on the same subpopulations of downstream progenitors, Thy and BLSP (left top and bottom), which express BMP antagonists. Viewing the left and right panels in unison, the ligand-cognate receptor interaction graph shows that signaling through systemic hormones leptin and TSH receptors may produce inhibitory signals for mSSC expansion by BMP2 antagonism (via gremlin 2, noggin) and subsequent osteogenesis in skeletal stromal populations. Arrows illustrate the potential receptor-ligand interactions. Data were compiled from triplicate samples, with 10,000 cells of each subset analyzed in each sample. See also Figures S2, S5.

Fig 4. Shifting fates: cartilage to bone

(A) Scheme of experiment: BCSPs were isolated from femora of GFP+ mice at P3 and PCPs (pro-chondrogenic progenitor) were isolated from ears/sternum of RFP+ adult mice following mechanical and enzymatic digestion, and subsequent FACS fractionation. Purified GFP+ BCSP and RFP+ PCP were co-transplanted beneath the kidney capsules of immunodeficient recipient mice. RFP+ PCP transplant served as a control. 1 month after transplantation, all grafts were removed for analysis. (B) Microscopy of explanted grafts 1 month following transplantation of 20,000 RFP+ PCP cells (see Figure 4A). The white dotted line outlines the extent of the graft formed. Fluorescence micrograph of explanted graft demonstrating the presence of an RFP+ graft (upper left). Corresponding brightfield micrograph is shown in upper right panel. Transverse section stained with Movat’s Pentachrome demonstrates that RFP-labeled PCP cells form cartilage (blue stain) (lower right). Scale bar: 500µm (upper panel), 200µm (lower panel). Representative of 4 replicates. (C) Microscopy of explanted grafts following co-transplantation of 20,000 RFP-labeled PCP cells with 20,000 GFP-labeled BCSPs (as detailed in Figure 4A). The grafts, which formed, are outlined by a white dotted line (indicating RFP+ portion) and a yellow solid line (indicating GFP+ portion) (upper left). A corresponding brightfield micrograph is shown in the upper right panel. A transverse section stained with Movat’s pentachrome shows that both RFP-labeled PCPs and GFP-labeled BCSPs form bone (yellow stain) (lower right). Scale bar: 500µm (upper panel), 200µm (lower panel). Representative of 5 replicates.

Fig 5. Shifting fates: bone to cartilage

(A) Scheme of experiment: mSSCs were isolated from P3 mice as described previously. Either intact pre-osteogenic femora isolated from E14.5 mice or 20,000 mSSCs were then transplanted beneath the kidney capsule of recipient mice with / without systemic inhibition of VEGF signaling. For inhibition of VEGF signaling, adenoviral vectors encoding soluble VEGFR1 ectodomain (Ad sVEGFR1) were delivered intravenously to the recipient mice 24 hours prior to cell transplantation, leading to systemic release of this potent antagonist of VEGF signaling. Ad Fc encoding an immunoglobulin Fc fragment was served as a negative control. Grafts were explanted 3 weeks later. (B) Explanted grafts are shown in the first and third row from the top of the panel (control, Ad Fc in left panel and Ad sVEGFR1 in the right panel). Representative sections stained with Movat’s pentachrome are shown in the second and fourth rows from the top of the panel. (top two rows: intact fetal bone, bottom two rows: mSSC) VEGF signaling inhibition resulted in the formation of cartilage (blue stain) (right), with the Ad Fc group forming bone (yellow stain) (left). Scale bar: 500µm (Intact fetal bone, top), 100µm (Intact fetal bone, bottom), 500 µm (mSSC, top), 200 µm (mSSC, bottom). Representative of 5 replicates per assay. (C) Graphic representation of the fate of the mSSC/fetal bone transplants in the presence of Ad sVEGFR1 or Ad Fc control. Cartilaginous fate is promoted in the presence of systemic VEGF antagonism. (n=5, ANOVA, p<0.001).

Fig 6. Manipulation of the BMP pathway can induce de novo formation of the mSSC in extraskeletal regions

(A) Collagen sponges containing 3ug of lyophilized rhBMP2 were placed into extraskeletal sites in C57BL6 wild-type mice. 1 month later, the graft was explanted for analysis. Brightfield images of explants, with renal capsule transplant shown above and subcutaneous transplant shown below (left). Transverse sections stained with Movat’s Pentachrome demonstrate that induced osseous osteoids formed a marrow cavity (red stain) (right). Scale bar: 500µm (left), 200 µm (right). Representative of 5 replicates. (B) FACS analysis of cells within the induced osteoid marrow reveals that circulating SlamF1 positive HSC engraftment occurs in the osteoids (bottom row) similar to that which occurs naturally in “normal” adult femurs (top row). Representative of 3 replicates. (C) (Left) Following explantation of rhBMP2-laced collagen sponges at day 10 post extraskeletal placement, FACS analysis of constituent cell populations present within the graft revealed that mSSC (red box on FACS plot) and BCSP (green box on FACS plot) are readily detectable in the rhBMP2 treated explants. (Right) In contrast, FACS-analysis of adipose tissue in the absence of BMP2 does not detect either mSSC (red box on FACS plot) or BCSP (green box on FACS plot). Scale bar: 500 µm. Representative of 3 replicates. (D) A parabiosis model of GFP+ and non-GFP mouse shows that circulating skeletal progenitor cells did not contribute to BMP2-induced ectopic bones. A GFP+ mouse was parabiosed to a non-GFPmouse. Two weeks later, a collagen sponge containing 3ug of lyophilized rhBMP2 was transplanted into the inguinal fat pad of the non-GFP mouse. Ten days later, the tissue was explanted and isolated the constituent cell populations of the ectopic bone tissue as described previously. The contribution of the GFP-labeled cells to ectopic bone formation in the non-GFP mouse was analyzed by FACS (broken red line; GFP+ = circulating cells, and non-fluorescent = local cells). GFP-labeled cells contributing to the graft were solely CD45(+) hematopoietic cells (extreme left panel, broken purple line), and not consistent of the skeletal progenitor population (horizontal upper panel, mSSC shown in red box on FACS plot). Representative of 3 replicates. (E) (Left) Diagram of reporter gene mouse model shows that Tie2 expression leads to GFP expression. Tie2+ cells turn green but Tie2− cells remain red. (right) Scheme of experiment: In order to determine the cell types, which could undergo BMP-2 mediated reprogramming to mSSC in extraskeletal sites, we implemented a Tie2Cre x MTMG reporter mouse and placed a collagen sponge containing rhBMP2 into the subcutaneous inguinal fat pad. The ossicle was explanted 1 month later for histological analysis. (F) Fluorescent micrographs: BMP-2 derived ossicles (yellow broken line) clearly incorporate both GFP+ Tie2+ derived osteocytes with visible canaliculi and Tie2− negative RFP-labeled osteocytes. Area denoted by white oval is shown at higher magnification in the box on the extreme right, showing the presence of (GFP+) Tie2+ canaliculi in the presence of (RFP+) Tie2−cells. Scale bar: 500µM (left), 50um (right). Representative of 3 replicates.

Fig 7. Co-delivery of BMP2 and VEGF inhibitor is sufficient to induce de novo formation of cartilage in adipose tissue

(A) Scheme of experiment: co-delivery of BMP2 and soluble VEGFR1 in adipose tissue. BMP2 and inhibition of VEGF signaling leads to cartilage formation. BMP2 alone leads to bone formation. (B) Fate of subcutaneous collagen sponge implants containing BMP2 without (2 panels on left) or with VEGF blockade (2 panels right). 1 month after placement of collagen sponges, the grafts were explanted for analysis. Brightfield images of grafts (left) are shown alongside representative sections stained with Movat’s pentachrome (right). The co-delivery of BMP2 and sVEGFR1 resulted in the formation of blue staining cartilage (extreme right panels, top and bottom). The result following systemic VEGF blockade is shown in the top series on the right, while the result of local VEGF blockade is shown in the bottom series on the right. Scale bar: (panels moving from right to left): 1mm, 200µm, 1mm, 200 µm. Representative of 4 replicates. (C) FACS plot analysis of the constituent cells of the induced cartilage (top horizontal panel) (experimental scheme as shown in Figure 7A) versus those of freshly isolated cells from ear cartilage of age-matched mice (bottom horizontal panel) demonstrates that PCPs (pro-chondrogenic progenitor) are found in similar frequency in the induced and natural cartilage tissue. Representative of the assay performed in triplicate.