Sorting of growth plate chondrocytes allows the isolation and characterization of cells of a defined differentiation status

Abstract

Axial growth of long bones occurs through a coordinated process of growth plate chondrocyte proliferation and differentiation. This maturation of chondrocytes is reflected in a zonal change in gene expression and cell morphology from resting to proliferative, prehypertrophic, and hypertrophic chondrocytes of the growth plate followed by ossification. A major experimental limitation in understanding growth plate biology and pathophysiology is the lack of a robust technique to isolate cells from the different zones, particularly from small animals. Here, we report on a new strategy for separating distinct chondrocyte populations from mouse growth plates. By transcriptome profiling of microdissected zones of growth plates, we identified novel, zone-specific cell surface markers and used these for flow cytometry and immunomagnetic cell separation to quantify, enrich, and characterize chondrocytes populations with respect to their differentiation status. This approach provides a novel platform to study cartilage development and characterize mouse growth plate chondrocytes to reveal unique cellular phenotypes of the distinct subpopulations within the growth plate. © 2010 American Society for Bone and Mineral Research

Introduction

The mammalian growth plate represents a road map of cartilage development built up by chondrocytes that are arranged in zones of distinct differentiation stages. The zones, composed of resting/proliferative, prehypertrophic, and hypertrophic chondrocytes, can be defined according to their specific morphology and gene expression pattern. Chondroblasts within the resting/proliferative zone are committed to enter a continuous differentiation and maturation process that terminates in hypertrophic cartilage, which becomes calcified, resorbed by osteoclasts, and then replaced by primary bone trabeculae in a process termed endochondral ossification.1 Defects in growth plate development and its homeostasis caused by abnormal proliferation and maturation of chondrocytes may result in skeletal chondrodysplasias, dwarfism, and skeletal malformations.2 Considerable insight into the pathogenesis of human chondrodysplasias has been gained from studies of mouse models,3, 4 which also provide important information on the cell biology of normal cartilage development and maturation. The analysis of spontaneous and transgenic mouse disease models, however, has been hampered by difficulty in isolating distinct populations of growth plate chondrocytes suited for cell biologic studies. The separation and enrichment of cell populations from growth plate cartilage therefore would greatly facilitate studies of the molecular mechanisms of cartilage development and hereditary disorders of the skeletal system.

The isolation of chondrocyte populations from cartilage of larger species, including bovine and chicken, has been achieved by using specific cell characteristics, such as different cell sedimentation rates,5–8 but the low recovery of purified cells renders this approach unsuitable for mouse cartilage.9 Fluorescence-activated cell sorting (FACS) combined with immunomagnetic cell separation appears as an ideal alternative method to sort viable cell populations9, 10 but could not be applied to growth plate chondrocytes owing to the lack of suitable cell surface markers and corresponding antibodies. Many known growth plate markers are transcription factors (eg, members of the Sox and Runx families), secreted growth factors and signaling molecules (eg, Ihh and PTHrP), or extracellular molecules (eg, collagens, glycoproteins, and matrix metalloproteinases) not appropriate for cell sorting.11–13 Some specific cell surface markers such as Fgfrs or Igfrs offer possible approaches, but no antibodies suitable for cell separation are available up to now. Expression of green fluorescent protein under control of the collagen 2 promotor in transgenic mice was used previously to purify all chondrocytes from rib cages but did not allow the detection, quantification, and fractionation of growth plate chondrocytes.14

To identify new antigens that allow cell sorting, we conducted whole-genome transcriptome analysis of dissected distinct zones of juvenile mouse growth plate and thereby discovered several differentially expressed genes coding for cell surface proteins. A subset of these proteins was selected to set up a multiparametric cell-sorting approach that allows the qualitative and quantitative assessment of growth plate composition and the preparative separation of all distinct chondrocyte subpopulations. We also established a simple two-step magnetic bulk-sorting approach that enabled us to isolate viable subpopulations of growth plate chondrocytes and characterize their apoptosis rate and differentiation status.

Material and Methods

Mice

Experiments were performed with C57BL/6 wild-type BAC-Col10a1-LacZ transgenic15 or Col9a1-deficient mice,16 in accordance with the animal ethics guidelines of the Murdoch Children's Research Institute and German law.

Whole-genome expression profiling

Mouse growth plate cartilage was microdissected and RNA-extracted, and microarray experiments were performed as described previously.17 Here, one 14-day-old mouse femoral growth plate was completely cryosectioned, and the individual maturation zones of all sections were isolated by microdissection. RNA was isolated and linearly amplified using the MessageAmp aRNA Kit (Ambion, Austin, TX), cRNA was labeled with Cy3 and Cy5 dyes (Amersham Biosciences, Piscataway, NJ) and hybridized to mouse 44K whole-genome microarrays (G4122A, Agilent Technologies, Santa Clara, CA). Arrays were scanned on an Axon4000B scanner, the features acquired with GenePix Pro 4.1 (Axon Instruments, Sunnyvale, CA), and the raw data were processed using a loess normalization in LimmaGUI (WEHI, Australia).18 Candidate genes coding for cell surface antigens expressed within the growth plate were selected according to their differential expression, their relative expression level, and their B-statistic values (Bayes-approach).19 Candidate genes for cell sorting were prioritized by the availability of commercial antibodies suitable for cell sorting.

Immunohistochemistry

Immunohistochemical analyses were performed as described previously.20, 21 Primary antibodies against CD24a (Biolegend, San Diego, CA), CD81, CD44, CD45 (Pharmingen, Franklin Lakes, NJ), TrkB (R&D, Minneapolis, MN), CD200 (Serotec, Düsseldorf, Germany), Igf2r, Fgfr3 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), or Pthr1 (Abcam, Cambridge, UK) were incubated on sections and detected by secondary antibodies conjugated with Alexa or Cy-dyes (Molecular Probes, Jackson, West Grove, PA, USA). Incubations of sections with secondary antibodies only were used as negative controls. Sections of Col10a1-LacZ transgenic mice were stained for β-galactosidase activity, as described previously.23

Isolation of growth plate chondrocytes

Distal femurs and proximal tibias of 12- or 13-day-old wild-type (C57BL/6), Col9a1-deficient or BAC-Col10a1-LacZ mice15 were isolated and predigested with 0.2% collagenase P (Roche, Berlin, Germany) in DMEM and 10% fetal calf serum (FCS) for 30 minutes at 37°C. Growth plate cartilage was prepared and digested with 0.2% collagenase II (Worthington) in DMEM and 10% FCS for 12 hours at 37°C. Subsequently, cell suspensions were filtered through a 70-µm nylon mesh (Becton Dickinson, Franklin Lakes, NJ), washed with PBS, and resuspended in PBS containing 5% FCS. Approximately 7.5 × 10⁴ cells were recovered from the pooled growth plates of a single mouse.

Immunomagnetic cell selection

Cell suspensions from growth plate cartilage of 12-day-old mice were fractionated by immunomagnetic cell selection according to the manufacturer's recommendations (Stem Cell Technologies, Vancouver, BC). Briefly, cell suspensions were incubated in the magnetic field prior to immunostaining to remove any bound particles. The unbound supernatant containing approximately 4 × 10⁵ cells was centrifuged, resuspended in 50 µL of PBS–5% FCS, and stained with Fc-block reagent as well as with primary conjugated antibodies CD200 phycoerythrin PE (Serotec, Düsseldorf, Germany) and CD24a fluorescein isothiocyanate (FITC; Biolegend) for 15 minutes at room temperature. Subsequently, cells were incubated with 10 µL of PE-selection cocktail in a total volume of 110 µL for 15 minutes at room temperature, followed by an addition of 5 to 10 µL of magnetic nanoparticles and incubation for another 10 minutes at room temperature. PE⁺ cells were recovered as bound cells in a magnetic separation step, whereas negatively selected cells in the supernatant were further incubated with the FITC-selection cocktail and nanoparticles as described earlier. In a second magnetic separation step, FITC⁺ cells were obtained. All cell fractions were costained with 7AAD followed by FACS analysis (FACSCalibur, Becton Dickinson, WEHI, Australia) or used in cell culture experiments.

Fluorescence-activated cell sorting

Up to 3 × 10⁵ cells from growth plate cartilage of 12- or 13-day-old mice were incubated with primary conjugated antibodies specific for CD24a, CD81, CD44, CD45, TrkB, or CD200 in 50 µL of PBS–5% FCS for 15 minutes on ice. Alternatively, cells were incubated with unconjugated antibodies specific for Fgfr3, Igf2r, Pthr1, or collagen X,26 washed, and labeled with corresponding conjugated secondary antibodies (Molecular Probes, Carlsbad, CA) in 50 µL PBS–5% FCS on ice for 15 minutes. For detection of intracellular collagen X, cells were fixed in 1% paraformaldehyde (PFA)-PBS for 10 minutes and permeabilized in 0.2% Triton-X-100–PBS for 5 minutes on ice prior to staining. β-Galactosidase activity was detected by fluorescein-di-β-D-galactopyranoside staining as described previously.23 Cells were subsequently washed and costained with 7AAD for dead-cell detection, followed by flow cytometry (FACSCalibur, FACSCantoII, Becton Dickinson) or cell sorting (FACSVantageSE, Becton Dickinson). For viability and cell-cycle experiments, cells were additionally stained with AnxA5-Alexa647 or Vybrant dye cycle green (Invitrogen) as described previously.21 Sorted cells were snap frozen, and RNA was purified (see below). All antibodies were tested on positive controls by flow cytometry, and corresponding isotypes were used as negative controls.

RNA isolation, amplification, and quantification by quantitative PCR (qPCR)

RNA of the sorted cell populations was extracted using the RNAeasy Micro Kit (Qiagen, Hilden, Germany). Quality of total RNA samples was assessed by capillary electrophoresis using a RNA6000-Pico Kit according to the manufacturer's specifications (Agilent). RNA samples with an RNA integrity number above 8 were considered as nondegraded. Each sample was linearly amplified (MessageAmpII Kit, Ambion).27 Amplified RNA was reversely transcribed into cDNA using the SuperscriptIII 1°-Strand Synthesis Kit (Invitrogen), and 10 ng of cDNA was used for each qPCR assay. Gene-specific primers and prevalidated probes of the Roche Universal Probe Library were used for Gdf10, osteomodulin, Spp1, Col10a1, Runx2, Ihh, and Sox9, whereas a preexisting primer-probe pair was employed for Mmp10 and Mmp13.28 qPCR reactions were performed using the qPCR SuperMix-UDG Kit (Invitrogen) on the DNAengine-Opticon2 System (Bio-Rad, Hercules, CA). For each primer-probe pair, the efficiency was determined and the relative expression levels were calculated using the ΔΔC_T method.29, 30Mapk7 was used for normalization.

In vitro culture

Purified cells were plated on 96-well plates (5 × 10³ cells per well) in DMEM-F12 [10% FCS, 1 µg/mL of amphotericin B (Lonza, Switzerland) and 100 µM of 2-phospho-ascorbate (Fluka, Milwaukee, WI)] for up to 5 days and then subjected to FACS analysis. Alternatively, sorted cells were cultured in hanging drops (5 × 10³ cells/50 µL) attached to the lid of a 6-cm dish for 3 days. The resulting cell clusters were fixed with icecold methanol and stained with 0.03% alcian blue or 0.5% alizarin red S (pH9). Attachment and migration of freshly isolated growth plate cells also were monitored at 37°C, 5% CO₂, and 60% humidity using an IX81 microscope (Olympus) and the corresponding analysis software OBS Cell R (Olympus).

Results

Identification of novel growth plate chondrocyte cell surface markers

RNA was extracted from proliferative, prehypertrophic, and hypertrophic zones microdissected from serial cryosections of whole femoral growth plates from 14-day-old mice (Fig. 1A). The mRNA was amplified and subjected to microarray analysis using the Agilent 44K whole-genome array platform. By analyzing the transcriptome, we identified genes for several cell surface antigens differentially expressed in discrete maturation zones of the growth plate. In addition to known chondrocyte surface proteins, including Fgfr3, Igf2r, and Pthr1, we selected five putative cell surface antigens previously not known to be expressed in the growth plate: TrkB, a receptor for neurotropins, and four “clusters of differentiation” antigens (CDs),31 CD24a, CD81, CD200, and CD44. CD24a and CD81 were highly expressed (Fig. 1B) in all three zones (Fig. 1C). TrkB, CD200, and CD44 were expressed at lower levels (Fig. 1B) but show a differential expression pattern (Fig. 1C). TrkB was preferentially expressed within the proliferative zone, whereas CD200 was upregulated in the prehypertrophic and hypertrophic zones and CD44 in the hypertrophic zone.

Figure 1. Identification of cell surface markers expressed within the growth plate. (A) A hematoxylin and eosin–stained cryosection of growth plate cartilage from a 14-day-old mouse after microdissection of the proliferative (P), prehypertrophic (PH), and hypertrophic (H) zones for RNA extraction and microarray expression analysis.17 (B) The average intensity of expression is given for each gene. (C) The relative expression ratio of selected candidate genes between the proliferative (P), prehypertrophic (PH), or hypertrophic (H) zone is shown. (D-O) Expression of CD24a (E, H, K, N) and CD200 (F, I, L, O) at different developmental stages in consecutive sections of femoral tissue is detected by immunofluorescence microscopy. Bars = 100 µm (D–I, M–O), 500 µm (J–L).

Analytical detection of cell surface antigens in situ and on isolated cells

To analyze the expression of the novel markers on the chondrocyte surfaces, immunofluorescence studies were performed on different developmental stages of murine growth plates. Consistent with the mRNA expression (Fig. 1B, C), in sections, CD24a protein was localized in cartilaginous tissues. Chondrogenic condensations in the hind limbs of E13.5 embryos stained positive for CD24a protein (Fig. 1E). With the progression of chondrogenesis, a moderate staining of prehypertrophic/hypertrophic chondrocytes, as well as a stronger staining of the more distant proliferative chondrocytes, was detectable in the femur of E15.5 embryos (Fig. 1H). In 6- and 12-day-old mice, CD24a protein was distributed ubiquitously throughout the growth plate of the femur and in the area of the developing secondary ossification center but not in the adjacent resting zone (Fig. 1K, N). Additionally, a strong staining of hematopoietic cells is found in bony areas owing to the presence of CD24a on most hematopoietic cell lineages32 (Fig. 1K). CD200 protein was localized within the area of prehypertrophic and hypertrophic chondrocytes throughout all developmental stages, but not in the resting and proliferating zone (Fig. 1F, I, L, O). No clear signal was detectable for Fgfr3, Igf2r, CD44, and TrkB (not shown). Therefore, CD24a was chosen as a suitable marker for identifying cells of growth plate origin, except for cells of the resting zone. Discrimination between the proliferative and prehypertrophic/hypertrophic cells could be achieved by CD200. In order to independently identify hypertrophic chondrocytes, we used growth plates derived from a mouse line expressing the reporter gene LacZ under the control of a BAC-Col10a1 promoter (BAC-Col10a1-LacZ).15 X-Gal staining of sections from such growth plates confirmed the exclusive expression of Col10a1-LacZ within the hypertrophic zone (not shown).

These markers then were applied in analytical cell sorting experiments. Growth plate cartilage from femurs and tibias of 12-day-old BAC-Col10a1-LacZ mice were prepared, and any bone marrow tissue was removed by accurate dissection (Fig. 2A, B). Subsequently, growth plates were digested with collagenase to obtain a single-cell suspension that contains large hypertrophic and smaller nonhypertrophic chondrocytes (Fig. 2C). Viable cell populations (Fig. 2D, E) were stained in parallel for the individual cell markers CD200, CD24a, and Col10a1-LacZ and analyzed by analytical flow cytometry. Most cells (90%; Fig. 2G) show a strong signal above isotype control (Fig. 2F) for the growth plate marker CD24a, whereas CD200 protein was present in about 25% of the cells (Fig. 2H). All CD200⁺ cells express CD24a, as highlighted in the dot plot for CD24a (Fig. 2G, red dots) and consequently represent a population of prehypertrophic/hypertrophic cells. To isolate viable hypertrophic chondrocytes, growth plate cell suspensions were stained in parallel with the β-galactosidase substrate fluorescein (FIC)–di-β-D-galactopyranoside.23 Using this analytical method, we identified a population of Col10a1-LacZ-FIC⁺ cells (9%, Fig. 2I) that also expresses the cell surface marker CD200 (Fig. 2H, purple dots) and thus corresponds to hypertrophic chondrocytes. None of the other markers, that is, Fgfr3, Igf2r, Pthr1, CD44, and TrkB, gave detectable signals (not shown).

Figure 2. Detection of cell marker expression in chondrocytes by flow cytometric analysis. (A–C) Preparation of single-cell suspensions. Femurs and tibias from 12-day-old mice (A) were dissected, and isolated growth plate cartilage (B) was further digested to obtain single-cell suspensions. (C-1) Digestion of growth plate cartilage releases the chondrocytes from the surrounding matrix. (C-2) Single-cell suspensions of digested growth plates contain large hypertrophic and small nonhypertrophic cells. (C-3) In culture, some of these isolated chondrocytes rearrange in a growth plate–like columnar alignment. (D–F) For flow cytometric analysis, only viable cells within the selected polygons (D, E) were considered for further analysis, and isotype staining was used to determine the threshold for each antibody staining (F). (G–L) Expression of cell surface antigens CD24a, CD200, and the Col10a1-LacZ reporter gene was analyzed in flow cytometry after (G–I) overnight digestion or (J–L) expansion of growth plate cells for 5 days in culture. Subpopulations were positive for CD24a (G, J), CD200 (H, K), or Col10a1-LacZ (I, L). CD200⁺ cells are marked by a red box (H, K) and highlighted as red dots in the CD24a/FSC dot plots (G, J). Col10a1-LacZ-FIC⁺ cells are marked by a pink box (I, L) and highlighted as pink dots in the CD200/FSC dot plot (H, K). FSC = forward scatter; SSC = side scatter; LIVE/DEAD blue = fluorescent reactive dye detecting dead cells; APC = allophycocyanin; PE = phycoerythrin; FIC = fluorescein. Bars = 1 mm (A), 0.5 mm (B), 50 µm (C1-3).

To analyze changes in marker gene expression in vitro over time, collagenase-digested growth plate cell suspensions of 12-day-old BAC-Col10a1-LacZ mice were kept in culture on plastic dishes for a period of 5 days and then tested for CD24a, CD200, and Col10a1-LacZ expression in analytical flow cytometry. Most cells maintained their growth plate character, as indicated by the expression of CD24a, whereas the relative number of CD200⁺ prehypertrophic cells increased (37%, Fig. 2K) and the number of Col10a1-LacZ-FIC⁺ hypertrophic cells decreased (3%, Fig. 2L).

Preparative separation of growth plate chondrocytes using magnetic bead techniques

To establish a simple but still robust method for isolating viable cells from mouse growth plate cartilage on the basis of this marker gene expression, we developed an immunomagnetic cell separation strategy (Fig. 3A). Single-cell suspensions of 12-day-old growth plate chondrocytes consisting of a heterogeneous population ranging from small, proliferating chondrocytes to large, hypertrophic cells (Fig. 3E) were stained with fluorescein isothiocyanate-labeled CD24a antibodies (CD24a FITC) and phycoerythrin-labeled CD200 antibodies (CD200 PE). Analytical flow cytometric analysis showed that a subfraction of all CD24a⁺ cells (Fig. 3B, UL + UR) also stained for CD200 (Fig. 3B, UR), whereas 7% of the total cells were CD24a⁻ and therefore not of growth plate origin (Fig. 3B, LL + LR). After preparative separation with PE-selective antibodies and PE antibody–selective magnetic nanoparticles, a CD200⁺ population was purified in a first magnetic separation step (Fig. 3A). Most of these separated cells were CD200⁺ and CD24a⁺ (62%, Fig. 3C, UR), indicating an approximately twofold enrichment of prehypertrophic/hypertrophic cells and a threefold decrease in proliferative cells compared with the original cell suspension (Fig. 3B). Dot-plot analysis of the forward and sideward scatter parameters revealed that the separated population was enriched in larger and more granular hypertrophic cells but still contained smaller prehypertrophic cells (Fig. 3F). The CD200⁻ population retained after removing the CD200⁺ population in the first separation step was incubated subsequently with an anti-FITC antibody and magnetic nanoparticles to purify the CD200⁻/CD24a FITC⁺ cells in a second magnetic separation step (Fig. 3A). Almost 85% of the enriched cells correspond to proliferative CD200⁻/CD24a⁺ cells (Fig. 3D, UL), and the average cell size (Fig. 3G) was smaller and less granular than in the unsorted (Fig. 3E) or purified prehypertrophic/hypertrophic population (Fig. 3F) when analyzing the corresponding dot plots for the forward and sideward scatter. This demonstrates that efficient enrichment of proliferative or prehypertrophic/hypertrophic populations can be achieved by an easy-to-use immunomagnetic cell sorting using the novel cell surface antigens.

Figure 3. Enrichment of chondrocytes from distinct maturation zones by immunomagnetic cell selection. (A) The scheme illustrates the subsequent purification steps from growth plate single-cell suspensions to enriched populations of prehypertrophic/hypertrophic or proliferative zone cells. (B–D) Dot plots of flow cytometric analysis showing the marker gene expression (CD24a FITC versus CD200 PE) from (B) unpurified, purified (C) prehypertrophic/hypertrophic or (D) proliferative cells and the corresponding size and granularity of the cell suspensions (E–G). Detected antigens and labels are indicated, and the proportion of cells within a certain region is given. FSC = forward scatter; SSC = side scatter; PE = phycoerythrin; FITC = fluorescein isothiocyanate.

Preparative separation of growth plate chondrocytes by multiparametric fluorescence-activated cell sorting

To establish a preparative separation system independent of the immunomagnetic bead system that allows a greater resolution in cell fractionation, we established an analytical multiparametric (five-color) flow cytometry protocol using a FACSVantage cell sorter. Cell suspensions derived from femurs and tibias of 12-day-old BAC-Col10a1-LacZ transgenic mice were stained with 7-aminoactinomycin D (7AAD) to identify dead cells, fluorescein (FIC)–di-β-D-galactopyranoside for detecting Col10a1-LacZ expression as well as with fluorochrome-labeled antibodies directed against CD24a, CD45, and CD200. Simultaneous staining and use of a gating hierarchy enabled the discrimination of all cell subpopulations from growth plate cartilage (Fig. 4A). First, dead cells and hematopoietic cells were excluded in flow cytometry by selecting for 7AAD⁻/CD45⁻ cells. An average starting material of 7.5 × 10⁴ viable (Fig. 4C, R1) CD45⁻ cells (Fig. 4D, not R2) were isolated per mouse, and within this population, only CD24a⁺ cells (85%, Fig. 4E, R3) were considered to be of growth plate origin. Of all viable growth plate–derived cells, 57% corresponded to CD24a⁺/CD200⁻ proliferative cells (Fig. 4G, LL). Approximately 40% of all viable growth plate–derived cells represent CD24a⁺/CD200⁺ prehypertrophic/hypertrophic cells (Fig. 4G, UL + UR) that show a detectable signal for CD200 above isotype control stainings (Fig. 4F). Within this population, 30% are related to CD24a⁺/CD200⁺/Col10a1-LacZ-FIC⁻ prehypertrophic cells (Fig. 4G, UL), whereas 10% represent CD24a⁺/CD200⁺/Col10a1-LacZ-FIC⁺ hypertrophic cells (Fig. 4G, UR). The reproducibility of separation was confirmed in two independent experiments. In parallel, proliferative, prehypertrophic, and hypertrophic cell populations were separated subsequently in a cell sorter to analyze their gene expression profiles.

Figure 4. Fractionation of growth plate chondrocyte populations by multiparametric cell sorting. (A) The gating scheme illustrates the subsequent purification of the chondrocyte population during cell sorting. (B–G) Typical dot plots of flow cytometric experiments are shown. The gated regions (R) and the proportion of cells within a region of the total viable cells are indicated as well as the markers used. (B) Forward scatter (FSC) and sideward scatter (SSC) detection indicate the heterogeneity of the cell suspension. (C) Dead cells were excluded by 7AAD staining, and only viable cells were used for further analysis (R1). (D) R2 corresponds to hematopoietic cells (CD45⁺), (E) R3 to growth plate chondrocytes (CD24a⁺), (G) LL to proliferative chondrocytes (CD200⁻), UL to prehypertrophic chondrocytes (CD200⁺), and UR to hypertrophic chondrocytes (Col10a1-LacZ⁺). The corresponding isotype control for wild-type chondrocytes (WT Col10a1) is shown (F). FSC = forward scatter; SSC = side scatter; 7AAD = 7-aminoactinomycin D fluorescent reactive dye-detecting dead cells; APC/Cy7 = allophycocyanin/Cy7; APC = allophycocyanin; PE = phycoerythrin; FIC = fluorescein.

Validation of the maturation zone–specific origin of sorted cell populations

The mRNAs of separated cell populations from flow cytometry and immunomagnetic sorting experiments were purified, amplified in one round of amplification,27 and assayed by qPCR for expression of marker genes that are characteristical for the different zones of the growth plate. From the previous microarray experiment (Fig. 5A), we identified Gdf1033 as a marker for proliferative cells. Osteomodulin was found predominantly in prehypertrophic cells and to a lesser extent in proliferative cells. Col10a115, 33 was upregulated within the prehypertrophic to hypertrophic cells, whereas Mmp13,34Mmp10, and Spp1 (osteopontin)35 were expressed mainly within hypertrophic cells. This expression pattern resembled the marker-gene expression in the purified chondrocyte populations isolated by multiparametric cell sorting, although minor variations were found owing to the different isolation and detection methods (Fig. 5B). The highest expression of Gdf10 was restricted to the sorted proliferative populations, whereas osteomodulin was expressed mainly in the separated prehypertrophic cells. As expected, Col10a1 was found predominantly in the prehypertrophic/hypertrophic cell population, whereas the highest levels of Mmp10 and Spp1 were found in the hypertrophic population and lower levels in the prehypertrophic cells. Similar results were obtained for the cell populations isolated by immunomagnetic cell sorting (Fig. 5C). The origin of the sorted cell populations also was confirmed by analyzing the expression of genes that are characteristically expressed within the distinct zones of the growth plate and regulate the differentiation of chondrocytes. Ihh and Runx2 are normally upregulated in prehypertrophic to hypertrophic chondrocytes, whereas the highest expression of Sox9 is found in proliferative cells.12 In microarray experiments and in the sorted-cell population, a similar expression pattern of these marker genes was detected (Fig. 5D). Ihh and Runx2 were expressed predominantly in isolated prehypertrophic to hypertrophic cells, whereas Sox9 expression was upregulated in the separated proliferative cell population. Therefore, these expression data confirmed that cells from the distinct maturation zones of the growth plate can be fractionated by two independent methods using either the simple immunomagnetic separation approach or the multiparametric cell-sorting protocol.

Figure 5. Expression of growth plate zone-specific markers in separated cell populations. (A) Relative expression levels of growth plate zone-specific marker genes (Gdf10, osteomodulin, Col10a1, Mmp13, Mmp10, and Spp1) determined by microarray expression analysis of microdissected proliferative (P), prehypertrophic (PH), or hypertrophic zones (H). (B, C) Relative qPCR of zone-specific marker-gene expression by the distinct populations of proliferative, prehypertrophic, or hypertrophic zone cells recovered by multiparametric cell sorting (B) and by the proliferative and prehypertrophic/hypertrophic (PH-H) cells obtained by immunomagnetic cell separation (C). (D) Relative expression levels of Ihh, Runx2, and Sox9 in microdissected or cell-sorted chondrocyte populations. All qPCR results are normalized to Mapk7 expression using the ΔΔC_T method with consideration of the efficiency according to Pfaffl.30 The standard deviations (n = 3) are shown. The relative expression levels between the distinct zones in both microarray and qPCR assays were determined with the zone of lowest expression set at 1.0.

Assessment of the growth plate composition during development

The establishment of cell-sorting methods to fractionate chondrocytes in different maturation zones offered us the opportunity of using this approach to establish a new quantitative analytical method to analyze growth plate development. We assessed the cell surface marker distribution of CD24a and CD200 in growth plate cell suspensions of 12- and 13-day-old mice, a time period when the secondary ossification centers are expanded, and the growth plate undergoes major morphologic changes.36 In a proof-of-principle experiment, pooled isolated cell suspensions of the dissected femoral and tibial growth plates of three 12- or 13-day-old mice were isolated. Single-cell suspensions of the dissected growth plates were analyzed in flow cytometry for the expression of both markers CD24a and CD200 (Fig. 6A), and the ratio between CD24a⁺/CD200⁻ proliferative and CD24a⁺/CD200⁺ prehypertrophic/hypertrophic cells was determined (Fig. 6C). The growth plates of 12-day-old mice are comprised of four times as much proliferative than prehypertrophic/hypertrophic cells (ratio [P/PH-H] = 3 to 4), whereas in growth plates of 13-day-old mice approximately twice as many proliferative cells are found in growth plate cell suspensions (ratio [P/PH-H] = 2) (Fig. 6C). We then used this approach to assess the growth plate composition of Col9a1-deficient mice that display severe growth plate abnormalities with large hypocellular areas and a widening of the circumferences of the bone.25, 37 Quantitative flow cytometry analysis of growth plate cell suspensions from three 12- or 13-day-old mutant mice (Fig. 6B) showed that these mice contain a significantly higher proportion of proliferative cells (ratio 12 days: [P/PH-H] = 7 to 8; 13 days: [P/PH-H] = 4 to 5) compared with wild-type mice (Fig. 6C). This is consistent with the finding that the hypertrophic area appears to be reduced37 (Fig. 6D, E). Therefore, FACS analysis reflects the changes in growth plate composition during development and can be used to quantify changes in mutant mice.

Figure 6. Characterization of growth plate maturation and cell death in chondrocyte populations. (A, B) Flow cytometric detection of CD200 and CD24a cell surface antigens in cell suspensions of 13-day-old growth plates in wild-type (A) and Col9a1-deficient mice (B). (C) The ratio of proliferative versus prehypertrophic/hypertrophic cell numbers for 12- and 13-day-old mice is given with standard deviation (n ≥ 3). Results were considered statistically highly significant for p < .01**. (D, E) Detection of CD200 in 13-day-old wild-type (D) and Col9a1-deficient mice (E). (F, G) Detection of intracellular collagen X in 13-day-old wild-type (F) and Col9a1-deficient mice (G). (H) The percentage of positive cells in P1 is given with standard deviation (n ≥ 3, p < .05**). (I) CD24a⁺/CD200⁻ proliferative, (J) CD24a⁺/CD200⁺ prehypertrophic/hypertrophic, and (K) Col10a1-LacZ-FIC⁺ hypertrophic cells were stained with annexin A5-Alexa647 (AnxA5-Alexa647) and 7AAD. Viable (R7, lower left), apoptotic (R8, lower right), and dead cells (R6, upper right) were quantified by FACS analysis. Bar = 100 µm (D, E).

A major drawback originates from the fact that we cannot separately detect hypertrophic chondrocytes. Collagen X represents the most promising marker for identifying hypertrophic cells, but owing to the collagenase digestion during chondrocyte isolation, no extracellular collagen X can be detected by FACS analysis on the cell surface (not shown). For this reason, we established a new staining method to detect intracellular collagen X. Once growth plate–derived cells were fixed and permeabilized, a monoclonal antibody directed against collagen X26 was used to quantify collagen X–containing hypertrophic chondrocytes in three 13-day-old wild-type and Col9a1-deficient mice (Fig. 6F–H). For both genotypes, a population of collagen X⁺ hypertrophic chondrocytes was detected. In Col9a1-deficient mice, the number of hypertrophic chondrocytes was significantly reduced (WT: 16%; Col9a1^−/−: 8%; Fig. 6H), confirming the previous results of the CD24a/CD200 marker expression analysis. Hence this approach now allows the quantitative assessment of hypertrophic cell content in the growth plate during development and in mutant mice.

Phenotypic characterization of growth plate–derived chondrocyte subpopulations

To determine to what extent the separated chondrocyte populations of the growth plate can be used for in vitro culturing, we first studied their cell-cycle status and viability. For cell-cycle analysis, collagenase-digested growth plate cell suspensions of 12-day-old mice were stained with CD24a-specific antibodies and the cell-cycle stain Vybrant green. Subsequent FACS experiments demonstrated that CD24a⁺ growth plate–derived cells still proliferated in suspension because 6% of the cells were found in the S-phase and 3.2% of the cells in the G₂ phase of the cell cycle (not shown). The viability of the CD24a⁺/CD200⁻ proliferative (Fig. 6I), CD24a⁺/CD200⁺ prehypertrophic/hypertrophic (Fig. 6J), and Col10a1-LacZ-FIC⁺ hypertrophic cells (Fig. 6K) then was determined by flow cytometry after staining with annexin A5-Alexa647 and 7AAD. Prehypertrophic cells only could not be displayed owing to the incompatibility of the dyes used. All cell populations irrespectively of their origin showed signs of cell death after overnight collagenase digestion. Moreover, 11% of the proliferative, 21% of the prehypertrophic/hypertrophic, and 18% of the hypertrophic cell population were annexin A5-Alexa647⁺, indicating the onset of apoptosis in these cells, and nearly equal amounts of necrotic cells (3.5% to 4.2%) are found in all subpopulations of the growth plate. The general shift in the 7AAD⁻ population (Fig. 6K) detected in the hypertrophic cell population is caused by the staining procedure for the detection of the β-galactosidase. This does not indicate an increased cell death, as demonstrated by the similar percentage of apoptotic and necrotic cells. Hence most of the isolated and separated cells were viable and could be used in downstream experiments.

In order to study growth plate–derived chondrocytes in vitro, we analyzed the attachment and migration of isolated whole-growth-plate cells derived from 12-day-old C57BL/6 wild-type mice (Fig. 7). The individual cells are numbered to visualize their fate in culture over time. Time-lapse data indicated that chondrocytes attach to the substrate and maintain a rounded phenotype for up to 3 days in culture irrespective of their size (Fig. 7A). Within this period, hardly any dividing or dying cells are detected. Then the cells flatten, start to migrate, and transform into a fibroblast-like cell type. These cells, which originally differ in size, adopt a similar morphology, and we speculated that they also express a similar marker signature. Consequently, we studied the differentiation status of FACS-sorted chondrocyte subpopulations isolated from 12-day-old C57BL/6 wild-type mice using CD24a- and CD200-specific antibodies (Fig. 7B–D). Unfractionated growth plate cells and proliferative and prehypertrophic/hypertrophic chondrocytes were analyzed for the expression of the differentiation markers CD24a and CD200 on days 3 and 5 after cultivation on a plastic substrate. These populations show a similar but slightly increased signal intensity for CD24a at both time points. In contrast, changes were seen in the expression of CD200. Directly after sorting, one-third of the whole-growth-plate cell population is stained positive for CD200 (27.8%, Fig. 7B, UR). After 3 days, approximately two-thirds of the unfractionated growth plate chondrocytes express CD200 (76%, Fig. 7B, UR) and maintain this level of expression after 5 days in culture (70%, Fig. 7B, UR). Prehypertrophic/hypertrophic cells preserve a high level of CD200 expression after 3 days of culture (92%, Fig. 7C, UR). After 5 days, a decrease in the CD200⁺ populations is detected (79%, Fig. 7C, UR). Half the proliferative chondrocytes originally negative for the CD200 start to express this marker after 3 days in culture (40%, Fig. 7D, UR) and increase the level of expression up to 58% (Fig. 7D, UR). Unfractionated growth plate cells and proliferative and prehypertrophic/hypertrophic chondrocytes, which initially differ in their CD24a/CD200 expression profile, show a similar marker signature (CD24a⁺/CD200⁺) when cultured for 5 days on a plastic substrate (Fig. 7B–D). We also could demonstrate that in long-term cultures (>7days), immunomagnetically enriched proliferative or prehypertrophic/hypertrophic chondrocytes of 12-day-old mice show only minor differences in extracellular matrix protein synthesis and produce a calcified matrix in vitro (not shown). Hence chondrocytes that originally differ in size and marker expression adopt a common cellular phenotype after long-term culture on a plastic substrate at subconfluent conditions.

Figure 7. Plasticity of isolated chondrocyte populations. Growth plate cells (GP), prehypertrophic/hypertrophic (PH-H) cells, or proliferative (P) cells were cultured in DMEM/F12 medium supplemented with ascorbate to maintain the chondrogenic phenotype. (A) Attachment and migration of growth plate cells on plastic substrate were analyzed for a period of 5 days. Individual cells are marked by consecutive numbers to follow up attachment and migration in culture. (B–D) Dot plots of FACS analysis showing the marker gene expression (CD24a FITC versus CD200 PE) from (B) purified growth plate cells (CD24a⁺), (C) prehypertrophic/hypertrophic cells (CD24a⁺/CD200⁺), or (D) proliferative cells (CD24a⁺/CD200⁻) in culture for a period of 5 days. (E, F) Unpurified growth plate cells, immunomagnetic sorted prehypertrophic/hypertrophic cells, or proliferative cells were cultured for 3 days in hanging drops and then stained by (E) alizarin red S to detect calcium deposits or (F) alcian blue to determine the proteoglycan content of the cell clusters. APC = allophycocyanin; PE = phycoerythrin. Bar = 100 µm (A), 200 µm (E).

Consequently, we studied chondrocytes shortly after cell sorting. Immunomagnetically separated cell populations were cultured for 3 days in hanging drops, and the cell clusters obtained were analyzed for mineralization and proteoglycan deposition using alizarin red S (Fig. 7E) and alcian blue (Fig. 7F) staining. Enriched prehypertrophic/hypertrophic chondrocyte populations stain most intensively for mineral and proteoglycan deposits. A moderate staining for mineral deposits is detected for whole-growth-plate chondrocytes, whereas for proliferative cells hardly any deposits are found. Chondrocytes prior to cell sorting and isolated proliferative cells also show a less intense but similar staining for proteoglycans. Hence immunomagnetically purified prehypertrophic/hypertrophic chondrocytes have a greater capacity than whole-growth-plate and proliferative cell populations to calcify and produce a proteoglycan-enriched matrix when cultivated in hanging-drop cultures directly after isolation. These differences are nearly abolished when these cells are cultured in monolayers on a plastic substrate.

Discussion

Fluorescence-activated cell sorting (FACS) and immunomagnetic cell separation are methods of choice to sort viable cell populations from blood or cell suspensions.9, 10 They have not been applied to growth plate chondrocytes mainly owing to a lack of appropriate cell surface markers. To identify such markers, we used whole-genome expression profiling of microdissected zones of mouse growth plate cartilage. The established gene-expression data set well reflects the in situ gene expression previously determined by other methods for, for example, Col10a1,15Mmp13,34 and Spp1,35 and in addition to these already well-characterized genes involved in chondrocyte maturation, many novel genes were found to be expressed in growth plate cartilage.

We now identified a number of putative cell surface markers including two novel “clusters of differentiation” antigens31 that are either found throughout the whole growth plate (CD24a) or within the prehypertrophic/hypertrophic zone (CD200). Both proteins are normally involved in hematopoiesis. Interestingly, CD200 is an immunomodulatory molecule potentially suppressing inflammation and autoimmune destruction, as well as osteoclast differentiation.38–40 The juvenile growth plate is a continuously remodeling tissue in which hypertrophic cells secrete not only matrix metalloproteinases for degradation of the cartilage matrix34, 41 but also vascular endothelial growth factor that induces invasion of bone marrow sprouts.42 While a portion of the hypertrophic chondrocytes undergoes apoptosis, endochondral bone is deposited on the scaffold of residual spicules of hypertrophic cartilage. In this process, numerous antigens are potentially released in close vicinity to hematopoietic cells of the bone marrow. Therefore, CD200 expression may attenuate the immune response in the growth plate to prevent local damage owing to the release of potentially proinflammatory signals and autoantigens during endochondral ossification. CD24a is a regulator of tissue homeostasis within self-renewing adult tissues43 such as the epidermis or bone marrow. Consequently, lack of CD24a induces epidermal hyperproliferation and affects B cell development. The restricted expression of CD24a in growth plate cartilage, but not in the adjacent epiphyseal cartilage, therefore may point to a regulatory function in growth plate homeostasis, although impaired skeletal development was not yet reported for CD24a-deficient mice.

None of the other identified cell surface markers (Fgfr3,44 Pthr1,45 or TrkB) was suitable for cell-sorting experiments. No signals were detected in immunofluorescence or flow cytometric studies. This indicates that the antigens are either not accessible to antibodies or that some of the extracellular cell surface antigens may be degraded by proteolytic cleavage46 so that these antigens cannot be recognized. Moreover, mRNA expression sometimes does not reflect steady-state protein levels correctly.47

The use of CD200 and CD24a now provides adequate cell signatures of the different maturation zones for growth plate sorting, as shown by two independent separation methods. The purity of the isolated cell populations is reflected by the change in expression levels of characteristic marker genes, as demonstrated by qPCR analysis for selected genes (Fig. 5). Moreover, qPCR analysis clearly showed that the sorted cell populations correspond to their predicted origin within the in situ growth plate. Although we detected some variations in qPCR analysis of the relative expression levels of maturation zone–specific marker genes compared with microarray data generated from microdissected cartilage, there was good overall consistency. Discrepancies were potentially caused by the different sources of the RNA. Microarray data were generated from individual manually microdissected regions of growth plate cartilage and thus are subjected to operator bias, whereas RNA of the cell populations was isolated from cell suspensions sorted by the use of signature-cell surface marker. Overnight digestion also may influence the expression profile of the distinct chondrocyte populations and explain the increased Mmp13 expression in the prehypertrophic population of fluorescence-activated cell-sorted chondrocytes. Nevertheless, the overall expression pattern of typical maturation markers such as the transcription factors Runx2 and Sox9 and the morphogenetic protein Ihh resembles the in situ situation and confirms the predicted origin of the sorted population.

For detailed experimental analysis of cartilage maturation, the CD200 and CD24a signatures can be used in FACS analysis to define the relative distribution of the subpopulations in growth plate cartilage and to preparatively separate chondrocytes for in vitro experiments. Thus the method presented here can be used to supplement in situ hybridization or immunohistologic analysis of the growth plate. The established separation techniques allowed us to study the cell biology of distinct subpopulations of the murine growth plate as isolated chondrocytes, irrespectively of their origin within the growth plate, that display no strongly increased signs of senescence or apoptosis. In monolayer cultures, the distinct cell populations of the growth plate adopt a similar phenotype, whereas in hanging-drop cultures a more pronounced proteoglycan and mineral depositions is found for chondrocytes of the prehypertrophic/hypertrophic chondrocytes compared with proliferative cells. Hence, shortly after cell sorting, all distinct subpopulations of the growth plate display a unique phenotype in vitro.

In conclusion, we applied transcriptome profiling to growth plate cartilage and translated the overwhelming amount of data into a simple strategy to discriminate, enumerate, and purify cell populations of the growth plate. Cells of the distinct zones from the murine growth plate now can be separated by conventional cell-sorting methods using newly identified cell surface markers to study the unique properties of growth plate chondrocytes in culture. The marker signature of chondrocytes can be employed for assessing the differentiation status of chondrocyte populations ex vivo. Assessment of the composition of the growth plate by qualitative and quantitative flow cytometry and the availability of cells to study gene and protein expression, as well as their cell biology, will be of importance in further unraveling cartilage development, endochondral bone growth, and the molecular basis of inherited cartilage disorders.

Disclosures

All the authors state that they have no conflicts of interest.

Acknowledgements

We gratefully acknowledge the scientific advice of Ernst Poeschl (University of East Anglia, Norwich, United Kingdom) and would like to thank Reinhard Fässler (Max Planck Institute of Biochemistry, Munich, Germany) for providing the Col9a1-deficient mice. This project was supported by grants from the Deutsche Forschungsgemeinschaft (BR2304/4-1 and SFB829-B6).