13
REGULAR ARTICLE Human and rodent bone marrow mesenchymal stem cells that express primitive stem cell markers can be directly enriched by using the CD49a molecule F. Gindraux & Z. Selmani & L. Obert & S. Davani & P. Tiberghien & P. Hervé & F. Deschaseaux Received: 14 February 2006 / Accepted: 20 June 2006 / Published online: 16 November 2006 # Springer-Verlag 2006 Abstract Bone marrow (BM) from human and rodent species contains a population of multipotential cells referred to as mesenchymal stem cells (MSCs). Currently, MSCs are isolated indirectly by using a culture step and then the generation of fibroblast colony-forming units (CFU-fs). Unprocessed or native BM MSCs have not yet been fully characterised. We have previously developed a direct enrichment method for the isolation of MSCs from human BM by using the CD49a protein (α1-integrin subunit). As the CD49a gene is highly conserved in mammals, we have evaluated whether this direct enrich- ment can be employed for BM cells from rodent strains (rat and mouse). We have also studied the native phenotype by using both immunodetection and immuno- magnetic methods and have compared MSCs from mouse, rat and human BM. As is the case for human BM, we have demonstrated that all rodent multipotential CFU-fs are contained within the CD49a-positive cell population. However, in the mouse, the number of CFU-fs is strain- dependent. Interestingly, all rat and mouse Sca-1-positive cells are concentrated within the CD49a-positive fraction and also contain all CFU-fs. In human, the colonies have been detected in the CD49a/CD133 double-positive population. Thus, the CD49a protein is a conserved marker that permits the direct enrichment of BM MSCs from various mammalian species; these cells have been phenotyped as true BM stem cells. Keywords Bone marrow . Mesenchymal stem cells . Direct enrichment . Characterisation . Fibroblast colony-forming units . CD49a protein (α1-integrin subunit) . Mouse . Rat . Human Introduction Adult mammalian bone marrow (BM) contains two types of stem cells: mesenchymal stem cells (MSCs) and haemato- poietic stem cells (HSCs). MSCs have the ability to differentiate, both in vivo and in vitro, into osteo-chondro- blastic, adipocytic, stromacytic or neural lineages (Pittenger et al. 1999; Conget and Minguell 1999; Bianco et al. 2001; Gao et al. 2001). Usually, they are isolated indirectly by their ability to adhere to the plastic surface of tissue culture flasks and they are characterised by their capacity to form individual colonies arising from a single precursor termed the fibroblast colony-forming unit (CFU-f; Pittenger et al. 1999). This indirect selection method yields a phenotypi- Cell Tissue Res (2007) 327:471483 DOI 10.1007/s00441-006-0292-3 Z.S. was in receipt of a grant from the Conseil Scientifique de lEtablissement Français du Sang. This work was supported by grants from the Ligue Contre le Cancer (Doubs) and the Fondation pour la Transplantation (ET-040615). F. Gindraux : Z. Selmani : L. Obert : S. Davani : P. Tiberghien : P. Hervé : F. Deschaseaux Inserm U645, IFR 133, Establissement Français du Sang Bourgogne-Franche-Comté, Université de Besançon, Besançon, France L. Obert Departement dOrthopédie, CHU Jean Minjoz Besançon, Besançon, France S. Davani Laboratoire de Pharmacologie, CHU Jean Minjoz Besançon, Besançon 25020, France F. Deschaseaux (*) IBCT IFR 133, 240 Route de Dole, Besançon 25020, France e-mail: [email protected]

Human and rodent bone marrow mesenchymal stem cells that express primitive stem cell markers can be directly enriched by using the CD49a molecule

Embed Size (px)

Citation preview

REGULAR ARTICLE

Human and rodent bone marrow mesenchymal stem cellsthat express primitive stem cell markers can be directlyenriched by using the CD49a molecule

F. Gindraux & Z. Selmani & L. Obert & S. Davani &P. Tiberghien & P. Hervé & F. Deschaseaux

Received: 14 February 2006 /Accepted: 20 June 2006 / Published online: 16 November 2006# Springer-Verlag 2006

Abstract Bone marrow (BM) from human and rodentspecies contains a population of multipotential cellsreferred to as mesenchymal stem cells (MSCs). Currently,MSCs are isolated indirectly by using a culture step andthen the generation of fibroblast colony-forming units(CFU-fs). Unprocessed or native BM MSCs have not yetbeen fully characterised. We have previously developed adirect enrichment method for the isolation of MSCs fromhuman BM by using the CD49a protein (α1-integrinsubunit). As the CD49a gene is highly conserved inmammals, we have evaluated whether this direct enrich-ment can be employed for BM cells from rodent strains(rat and mouse). We have also studied the nativephenotype by using both immunodetection and immuno-

magnetic methods and have compared MSCs from mouse,rat and human BM. As is the case for human BM, wehave demonstrated that all rodent multipotential CFU-fsare contained within the CD49a-positive cell population.However, in the mouse, the number of CFU-fs is strain-dependent. Interestingly, all rat and mouse Sca-1-positivecells are concentrated within the CD49a-positive fractionand also contain all CFU-fs. In human, the colonies havebeen detected in the CD49a/CD133 double-positivepopulation. Thus, the CD49a protein is a conservedmarker that permits the direct enrichment of BM MSCsfrom various mammalian species; these cells have beenphenotyped as true BM stem cells.

Keywords Bone marrow .Mesenchymal stem cells .

Direct enrichment . Characterisation .

Fibroblast colony-forming units .

CD49a protein (α1-integrin subunit) . Mouse . Rat . Human

Introduction

Adult mammalian bone marrow (BM) contains two types ofstem cells: mesenchymal stem cells (MSCs) and haemato-poietic stem cells (HSCs). MSCs have the ability todifferentiate, both in vivo and in vitro, into osteo-chondro-blastic, adipocytic, stromacytic or neural lineages (Pittengeret al. 1999; Conget and Minguell 1999; Bianco et al. 2001;Gao et al. 2001). Usually, they are isolated indirectly bytheir ability to adhere to the plastic surface of tissue cultureflasks and they are characterised by their capacity to formindividual colonies arising from a single precursor termedthe fibroblast colony-forming unit (CFU-f; Pittenger et al.1999). This indirect selection method yields a phenotypi-

Cell Tissue Res (2007) 327:471–483DOI 10.1007/s00441-006-0292-3

Z.S. was in receipt of a grant from the Conseil Scientifique del’Etablissement Français du Sang. This work was supported by grantsfrom the Ligue Contre le Cancer (Doubs) and the Fondation pour laTransplantation (ET-040615).

F. Gindraux : Z. Selmani : L. Obert : S. Davani : P. Tiberghien :P. Hervé : F. DeschaseauxInserm U645, IFR 133, Establissement Français du SangBourgogne-Franche-Comté, Université de Besançon,Besançon, France

L. ObertDepartement d’Orthopédie, CHU Jean Minjoz Besançon,Besançon, France

S. DavaniLaboratoire de Pharmacologie, CHU Jean Minjoz Besançon,Besançon 25020, France

F. Deschaseaux (*)IBCT IFR 133,240 Route de Dole,Besançon 25020, Francee-mail: [email protected]

cally and functionally heterogeneous fibroblastoid cellpopulation (Kuznetsov et al. 1997; Muraglia et al. 2000)that may not reflect the phenotype of native MSCs.

Many stem cell experimental studies are based on theuse of rodents as a convenient model because of their sizeand short breeding cycles (Phinney et al. 1999; Javazon etal. 2001; Tropel et al. 2004). However, murine MSCs(mMSCs) have been far more difficult to isolate and culturefrom BM than rat MSCs (rMSCs) or human MSCs(hMSCs). This is largely attributable to the presence ofnumerous haematopoietic progenitors that contaminatemouse CFU-fs (mCFU-fs; Phinney et al. 1999). Variousprotocols have been developed to try to circumvent thisproblem including: (1) antibody depletion of haemato-poietic cells by using HSC-specific markers before MSCculture (Baddoo et al. 2003), (2) preferential stimulation ofMSC growth by the addition of basic fibroblast growthfactor (bFGF) to the culture (Kuznetsov et al. 1997;Phinney et al. 1999; Sun et al. 2003), (3) variation in theplating density of the culture (Sun et al. 2003; MeirellesLda and Nardi 2003; Peister et al. 2004), or (4) theemployment of the retroviral selection of cycling adherentcells (Kitano et al. 2000; Tropel et al. 2004). However, theuse of additional steps adds to the difficulty of selectingmMSCs and complicates the interpretation of data fromdifferent laboratories with regard to the true nature of thesestem cells. Changes induced by the indirect selection maskthe true phenotype of native MSCs.

Previously, we have reported a method to enrich hMSCsdirectly by using antibodies that recognise the α1-integrinsubunit or CD49a. This system has the advantage that itenables us to isolate all hMSCs from a human BM aspiratedirectly by using immunomagnetic beads. These hMSC retaintheir in vitro differentiation potential and we have also beenable accurately to assess the phenotype by identifyingmolecules whose expression is modulated by culture onplastic (Deschaseaux et al. 2003). In addition, the CD49agene is highly conserved between human and rat species(Gardner et al. 1996) and antibodies made against the humanprotein also cross-react in the rabbit (Makihira et al. 1999).

Here, we have assessed our direct enrichment method forthe isolation of native rodent BM MSCs b using anti-CD49a monoclonal antibodies (mAbs). We demonstratethat all of the CFU-f potential present in a BM sample,independent of the species, is contained within the discreteCD49a-positive (CD49a+) cell fraction. Importantly, afterdirect enrichment, cultures of rodent CFU-fs are multipo-tential and free from adherent haematopoietic cells. Flowcytometric analyses performed on human and rodentspecies has demonstrated that the CD49a+ fraction concen-trates cells with an immature stem cell phenotype: CD133+

cells in humans and, most interestingly, Sca-1 positivity inthe both rat and mouse.

Materials and methods

Human bone marrow

BM samples were collected with informed consent,following the ethical guidelines of the Jean Minjoz Hospitalin Besançon, France, from patients undergoing thoracicsurgery. Aspirates were centrifuged on Hypaque-Ficolldensity gradients. The interface mononuclear cells (MNCs)were recovered and washed twice with phosphate-bufferedsaline (PBS; Invitrogen, Paisley, UK) containing 0.1%human serum albumin (HSA; Laboratoire Français deFractionnement et de Biotechnologies, Courtaboeuf,France). Cells from this step were designated as the initialfraction. Cells were then cultured in optimal culturemedium as described below and then phenotyped by flowcytometry or subjected to CD49a cell sorting.

Rodent bone marrow

Male mice (4–17 weeks old; strains BALB/cByJ H-2d,C57BL/6J, FVB/N) and 10-week-old male rats (strainLEW/CRL ICO) were purchased from IFFA Credo (SaintGermain sur l’Arbresle, France). They were housed understandard conditions for 1 week prior to use. All animalsreceived chow and acidified water ad libitum. Rodents werekilled with carbon dioxide. To collect BM, the femurs andtibiae were dissected away from attached muscle andconnective tissue. The ends of the bones were cut awayand a 25-gauge 5/8 needle was inserted into the shaft of thebone. Marrow was extruded by flushing with α-MEMmedium (Invitrogen) supplemented with 20% fetal bovineserum (FBS; StemCells Technologies, Vancouver, Canada).Any marrow plugs were dispersed by repeated passagesthrough the same gauge needle. BM was then washed twicewith PBS/0.1% HSA. Cells at this stage were termed theinitial fraction. Cells were cultured at the appropriateplating densities in optimal culture medium as describedbelow and then phenotyped by flow cytometry or subjectedto CD49a cell sorting.

Cell separation

Human initial fractions were incubated with anti-humanCD49a mAbs (clone SR84, Pharmingen, San Diego, Calif.;clone TS2/7, T-Cell Science, Cambridge, Mass.) at aconcentration of 1 μg per 107 MNCs. Rodent initialfractions were incubated separately with a mouse anti-ratCD49a mAb (clone 3A3, Serotec, Oxford, UK), mouseanti-rat CD49a/CD29 mAb (clone RA1.1, ChemiconInternational, Temecula, Calif.) or mouse anti-humanCD49a mAb (clone SR84) at the same concentration. Allcell suspensions were gentle agitated for 30 min at 4°C.

472 Cell Tissue Res (2007) 327:471–483

After two washes with PBS/HSA, cells were incubated withbeads coated with either human anti-mouse IgG microbeads(Cellection Pan Mouse IgG Kit Dynal, Oslo, Norway) orgoat anti-mouse IgG microbeads (mini-Macs, MiltenyiBiotec, Bergisch Gladbach, Germany) according to themanufacturer’s instructions. Cell suspensions were separat-ed by using a magnetic particle concentrator (MCP-1,Dynal) or a magnetic separation column (Miltenyi Biotec).The uncoated cells (designated as the CD49a– fraction)were removed by several washes with PBS/HSA. Theretained cells were termed the CD49a+ fraction. For eachspecies, a sample of CD49a+ and CD49a– cells wereanalysed by flow cytometry.

In some experiments, murine and human CD49a+ cellswere separated from the beads when using the DynalCellection kit; Sca-1+ cells, for mouse, and CD133+, forhuman, were isolated from these selected CD49a+ fractionsby using a magnetic separation column and microbeadscoated with either mouse anti-Sca-1 or anti-CD133 mAbs(Miltenyi Biotec). Sca-1-positive and -negative cells andCD133-positive and -negative cells were then cultured todetermine their cloning efficiency (CE) as described below.

Culture conditions

Human and rat BM cells were plated in α-MEM supple-mented with 10% FBS (StemCell Technologies) at aconcentration of 10×103 MNCs/cm2, whereas we deter-mined the optimal growth conditions for initial fractions ofmMSCs. These occurred when mouse cells were cultured inα-MEM supplemented with 20% FBS and 1 ng/ml bFGF(R&D Systems, Minneapolis, Minn.). Mouse cells wereseeded from 3.2×103 to 24×103 MNCs/cm2. Fractions ofCD49a− cells were cultured at a density of 1×105 MNCs/cm2. When using the Dynal kit, the detached beads werealso placed into culture to ensure that all recognised cellswere correctly detached. After 96 h culture, the non-adherent cell population was removed and re-plated in a12.5-cm2 culture flask. The number of CFU-f, comprisingmore than 10 adherent elongated cells, was counted after8–14 days of culture for initial, positive and negativefractions. The CE was calculated as the ratio between thenumbers of detected CFU-fs for 105 MNCs seeded. Theenrichment factor was determined by the ratio betweenthe CE of the CD49a+ fraction and the CE of the initialfraction. The cells were cultured for a further 4–6 weeks.

Flow cytometric studies

Analysis of cells was performed by using a FACSortapparatus and CellQuest software (Becton Dickinson,Mountain View, Calif.). Initially, CD49a+ and CD49a−

fractions from each species were incubated with the

primary mAb or isotype-matched control mAb (Pharmin-gen) in the dark at 4°C for 30 min. Mouse anti-humanCD133 (AC 133−2) mAb conjugated to phycoerythrin (PE;Miltenyi Biotec) was employed for human cells. The mAbsused for rat cells were: mouse anti-rat Pan-T (OX-52), CD4(OX-35), CD8a (OX-8) and CD90 (OX-7) mAbs conjugat-ed to fluorescein isothiocyanate (FITC) and mouse anti-ratCD45 (OX-1) mAb conjugated to phycocyanin-5 (PC-5;Pharmingen). The mAbs used against murine cells were: ratanti-mouse Ly-6A/E (Sca-1; clone D7) and CD34 (cloneRAM34) mAbs conjugated to FITC, rat anti-mouse CD117(c-Kit; clone 2B8) mAb conjugated to PE, mouse anti-ratCD49a (clone 3A3) mAb used with a PE-conjugated goatanti-mouse IgG (Beckman-Coulter Immunotech, Marseille,France) and rat anti-mouse CD45 (clone 30-F11) mAbconjugated to PC-5 (Pharmingen). After three washes inPBS/0.1% HSA, cells were resuspended in PBS and passedthrough the FACSort. Ten thousand events for each samplewere recorded. The percentage of positive cells wasdetermined by using an appropriate gate on a forwardversus side scatter dot plot (FSC vs SSC). The fluorescencehistogram for each mAb was displayed alongside thecontrol antibody thereby allowing, by subtraction, anevaluation of the percentage of positive cells.

Flow cytometric cell sorting

BM MNCs were stained by using the mAbs anti-CD49a(TS2/7) and anti-human CD133-PE for human cells or byusing the mAbs anti-CD49a (3A3) and anti-Sca-1-PE formouse cells. After three washes, cells were passed througha Beckmann Coulter cell sorter Epics Altra (with Expo32software) with PBS in the sheath fluid. For sorting, twotypes of gates were used: the first was assigned to rejectdebris and the second (named R1) was determined byCD133 (or Sca-1) vs CD49a fluorescence (Figs. 1 and 5).Sorted cells, whose purity ranged from 95%–99%, werecultured into expansion medium.

In vitro differentiation of cells from CFU-Fs

The potential of BM MSCs to differentiate, via the CFU-Fs,into osteogenic, chondrogenic and adipogenic lineages wasverified after the first passage. The cells of each well ordish was maintained in expansion medium until confluencewas reached. Cultures were stimulated with the appropriatedifferentiating medium as previously described (Jaiswal etal. 1997; Banfi et al. 2000). Briefly, osteogenic andadipogenic media comprised of expansion medium supple-mented with 100×10−9 M dexamethasone, 0.05×10−3 Mascorbic acid and 10×10−3 M β-glycerophosphate forosteoinduction, whereas 1×10−6 M dexamethasone and100 μg isobutylmethylxanthin (IBMX; Sigma, St. Louis,

Cell Tissue Res (2007) 327:471–483 473

Mo.) were added for adipogenic culture conditions.Chondrogenic culture conditions were set up by employingchondrogenic medium (Cambrex, Walkersville, Md.) with10 ng/ml human recombinant transforming growth factorβ3 (TGF-β3). As a control, cultures were maintained inexpansion medium. After 2 weeks of culture, cells wereexamined for the expression of lineage-specific molecules.

Immunofluorescence in situ studies were performed aspreviously described (Deschaseaux et al. 2003). Briefly,confluent adherent layers of cells were grown on 8-wellchamber slides and then fixed for 30 min at 4°C. Variousfixations were used: 3.7% (w/v) formaldehyde with 0.5%(w/v) glutaraldehyde for extracellular matrix (ECM)antigens and absolute methanol for cytoskeleton antigens.We used the following antibodies: mAb anti-alkalinphosphatase (clone8B6; Sigma), goat anti-osteopontin(R&D Systems), goat anti-osteocalcin (M-15; Santa CruzBiotechnology, Calif.) and rabbit anti-CO I and rabbitanti-CO II (Chemicon). All of these antibodies cross-reacted between human, mouse and rat proteins. Primary

antibodies were added and stained by goat anti-mousepolyvalent Ig conjugated to FITC (Sigma) for mAbs, bygoat anti-rabbit IgG conjugated to tetra-methylrhodamine-B-isothiocyanate (TRITC; Sigma) or bovine anti-goat Igconjugated to FITC for polyclonal antibodies. Lipiddroplets within adipocytic cells were stained with Nile-red-oil solution (Sigma). Slides were examined by using amicroscope equipped for fluorescence (Leitz Aristoplan,Leica, Weltzar, Germany).

Histochemical staining by Alizarin red and Alcian bluewere performed on cells fixed in using 3.7% (w/v)formaldehyde.

Statistical analysis

In the text, values are given as means±SEM. Comparisonsbetween means were made by using either a Student’s t-testor a non-parametric Mann and Witney test. Differencesbetween means were considered significant when P-valueswere less than 0.05.

Fig. 1 Human and mouse BMcells were stained by anti-CD49a Mabs and anti-Sca-1 formouse cells or anti-CD133Mabs for human cells. a and bIsotype controls. b and d Gatesfor sorting labelled cells

474 Cell Tissue Res (2007) 327:471–483

Results

Cell separation

After enrichment, cells from initial, CD49a− and CD49a+

fractions were cultured in expansion medium. At day 10,the CFU-fs were quantified. CFU-fs were detected entirelyin CD49a+ fractions for all species, independent of the anti-CD49a mAbs.

The percentages of CD49a+ cells obtained from humanBM (Table 1) were significantly lower (P<0.01) whenusing the Miltenyi kit (n=18) than the Dynal kit (n=50),independent of the mAbs used (clone SR84 or clone TS2/7). The CE for 105 CD49a+ seeded human cells, isolated byusing either the Miltenyi kit or Dynal kit, and theenrichment factors were also significantly different(P<0.01). Therefore, the number of CD49a+ cells seededin order to obtain one CFU-f after only one separation stepwas approximately 500 cells for the Miltenyi kit and 3,333CD49a+ cells for the Dynal kit. Direct enrichment ofBM hCFU-fs was clearly more efficient when using theMiltenyi kit. For rat BM, the percentage recovery ofthe CD49a+ fraction was 1.1±0.1% (n=4) when using theMiltenyi kit with the mouse anti-rat CD49a (clone 3A3)mAb. The CE was 200 for 105 cells seeded, the enrichmentfactor was 59 and one CFU-f was found in approximately1,030 CD49a+ cells. We were unable to perform selectionby using Dynal beads in this case (data not shown).

For the separation of mouse cells, an anti-murine CD49amAb is currently not available. However, the CD49amolecule is highly conserved between mouse and rat(94% molecular identity calculated from the web site:http://www.ncbi.nlm.nih.gov/HomoloGene/homol.Cgi). Wetherefore assessed two types of mAbs: the mouse anti-ratCD49a (clone 3A3) mAb and the mouse anti-rat CD49a/CD29 (clone RA1.1) mAb whole heterodimer. We wereunable to isolate mCFU-fs by using the Miltenyi kit witheither mouse anti-rat or mouse anti-human CD49a mAbs todate. Thus, separations were performed by using either

antibody in combination with the Dynal kit and resulted inmCFU-fs only being detected in the CD49a+ fraction. Thepercentage of CD49a+ cells isolated with the anti-ratCD49a/CD29 (clone RA1.1) mAb represented 22±3% forthe BALB/c strain and 13±6% for the FVB strain. The CEwas significantly different (P<0.01) between the strains,being 0.7±0.1 and 0.28±0.05 for 105 cells seeded, respec-tively. The CE for the BALB/c CD49a+ fraction wassignificantly higher (P<0.05) than that found for the CEfor the BALB/c initial fraction, with an enrichment factor ofapproximately 2. The CE for FVB CD49a+ cells was lowerthan that of the initial fraction (0.48±0.04), whereas noCFU-f was detected in the CD49a− fraction. We thereforedecided to perform further studies on BALB/c BM to de-termine the optimal conditions for the selection of mMSCs.Indeed, by using clone 3A3 mAb, the percentage ofselected cells decreased to 10±2.7% (n=15) with a CE of18.8±7 for 105 seeded cells compared with the results foundabove with the clone RA1.1 mAb. This CE was significant-ly higher (P<0.05) than the CE of the initial fraction (3±0.7) giving an enrichment factor of 6. By using the anti-human CD49a (clone SR84) mAb, the percentage ofCD49a+ cells was 40±0.5% with a CE of 3±0.6 for 105

seeded cells. As the selections of cells were significantlymore effective by using the 3A3 clone, we decided to use itfor the following studies. Having obtained a workableselection rate, we proceeded to use the 3A3 antibody on theC57BL mouse strain generally used for performing stemcell biology phenotypic studies. The percentage of CD49a+

cells for the C57BL strain was 5.8±2.4% (n=8), beingsignificantly lower (P<0.05) than the percentage of CD49a+

cells recovered from the BALB/c strain (10±2.7%). The CEof C57BL CD49a+ cells was 5.1±1 for 105 seeded cells, avalue much higher than the CE of the C57BL initialfraction (1.8±0.8; P<0.05) but still significantly lower thanthe CE found for BALB/c CD49a+ cells (18.8±7). Theenrichment factor was 3 for the C57BL strain comparedwith 6 for the BALB/c mouse. In addition, for all experi-ments, no CFU-f was found in non-adherent fractions

Table 1 Selection of human bone marrow CFU-fs

Cell fraction (type of separation kit used) Recovered fraction Cloning efficiency Enrichment factorc

(%±SEM)a (×10−5)b

CD49a+(Dynal) 3.6±0.4 30±3 21CD49a+(Miltenyi) 0.4±0.06 200±1 143CD49a–(Dynal) 96±10 0 0CD49a–(Miltenyi) 99±14 0 0CD49a+(Dynal)/CD133+(Miltenyi) 0.3±0.04 150±1 100CD49a+(Dynal)/CD133–(Miltenyi) 0.09±0.02 0 0

a Percentage of CD49a+ cells recovered after selectionb Number of CFU-fs found in a fraction per MNC seededc Ratio between the cloning efficiency (CE) found for a fraction and the CE found for unseparated MNC (total bone marrow)

Cell Tissue Res (2007) 327:471–483 475

recovered after the first medium change. Thus, there was asignificant difference in the efficiency of directly enrichingthe native mCFU-fs from the BM of different strains ofmice. The BALB/c strain yielded the most CFU-fs.

Phenotypic studies

Since CFU-fs were present in a discrete population of cells,we then assessed the expression of proteins specific to stem

cells. We chose the CD133 molecule for human cells as itrepresents a highly specific marker for stem cells from boththe BM (Yin et al. 1997; Peichev et al. 2000) and thecentral nervous system (Uchida et al. 2000). We hadpreviously reported that all BM native hMSCs were locatedexclusively in the CD49a+ cell fraction that expressed thehaematopoietic marker CD45 at low to moderate levels(CD45med,low; Deschaseaux et al. 2003). Here, the flowcytometric studies showed that 72±5% human CD49a+

cells were positive for the primitive stem cell markerCD133 and all of the CD49a+/CD45med,low cells expressedthe CD133 marker (Fig. 2). In order to confirm this, twodifferent protocols were performed. By using a serialimmunomagnetic selection, the CFU-fs were detected onlyin the CD133+/CD49a+ fraction (Table 1). The CE for 105

seeded cells was 150 (n=2). In addition, these cells werealso sorted by using a flow cytometer (Figs. 1 and 5). After10 days of culture, the CFU-fs were observed in flasksconfirming CD133 expression by native human CFU-fs.However, we were not able to detect any CD133 moleculeson cells in cultures (not shown).

In the rodent model, Sca-1, CD34 and c-kit (stem cellfactor receptor) markers are known to be present on BMstem cells (Krause et al. 2001; Bonyadi et al. 2003;Benveniste et al. 2003). CD90 is another BM stem cellmarker (Craig et al. 1993) also expressed by lymphocytesand BM stromal cells. We have also used mAbs against ratT-cell markers. Therefore, the expression of Sca-1, CD34,

Fig. 2 Phenotypic characterisation of human CD49a+ cells by usingflow cytometry. Following sorting, CD49a+ cells were labelled byboth PC-5-conjugated anti-CD45 and PE-conjugated anti-CD133.Compared with the negative control (a), all CD49a+/CD45med,low

cells appeared to express the CD133 marker (b)

Fig. 3 Phenotypic characterisation of rat CD49a+ cells by using flowcytometry. Following sorting, the CD49a+ cells were evaluated for theexpression of molecules specific for either BM stem cells or maturehaematopoietic cells. The expression of CD45, CD90, c-kit, Sca-1,CD34 and T-cell markers was assessed on total BM (filled bars),CD49a+ (heavily stippled bars) and CD49a– (lightly stippled bars)cells (NT not tested)

476 Cell Tissue Res (2007) 327:471–483

c-kit and CD45 antigens were analysed on the totalpopulation of cells from both CD49a+ and CD49a−

fractions. Flow cytometric studies performed on rat BMshowed that the CD49a+ population contained numerouscells phenotypically defined as stem cells or progenitorcells (Fig. 3). This population consisted mainly of cellspositive for CD45, CD90 and c-kit molecules. Interestingly,over 20% of CD49a+ cells were also positive for the bothCD34 and c-kit molecules (Fig. 3) and Sca-1 antigen wasonly detected in the CD49a+ fraction (Figs. 3, 4a).However, all cells were negative for T-lymphocyte-specificmarkers (pan T, CD4, CD8). We detected no CD45−/CD90+

cells in this fraction (not shown).We also performed phenotypic studies on C57BL cells.

We chose this mouse strain because of its frequently use forstem cell research. For C57BL BM, the percentage ofCD49a+ cells in whole BM represented 6±2% as assessed

by flow cytometry and was similar to the 5.8±2.4% ofCD49a selected cells, confirming the specificity of theimmunomagnetic cell sorting. After being sorted, cellsexpressing Sca-1, CD34, CD45 and c-kit were found to beenriched in the CD49a+ fraction (Fig. 1 and 5) comparedwith the initial population but the difference was significantonly for Sca-1 (25%±9 vs 8%±3, P<0.05) and CD45 (92±5vs 56±8, P<0.01). Interestingly, we did not detect Sca-1+ inthe CD49a− fraction (Figs. 1, 4c and 5) as found above forrat model. Moreover, cells from the initial fraction with thehighly primitive stem cell phenotype (Sca-1+/c-kit+: 0.8±0.2%) were also concentrated in the CD49a+ selectedpopulation (3±1%).

The stem-cell-specific marker Sca-1 has previously beendescribed as a marker for native MSCs or a mesenchymalprogenitor-like population (Bonyadi et al. 2003). Otherprevious studies have also shown that cultured mMSCs

Fig. 4 Assessment of expres-sion of Sca-1 molecule by bothrat (a) and mouse (b) BM cells.CD49a+ and CD49a− fractionswere analysed by using flowcytometry. The CD49a+ fractionfrom the rat and mouse BMcontained all Sca-1+ cells

Cell Tissue Res (2007) 327:471–483 477

express this protein (Jiang et al. 2002; Peister et al. 2004).In order to confirm that all Sca-1+ cells expressed theCD49a molecule, as demonstrated above by flow cytometryfor C57BL mouse strain, we performed Sca-1 cell selectionon the C57BL CD49a+ fraction. Serial immunomagneticselections showed that double-positive cells (Sca-1+/CD49a+) represented 1.1±0.5% (n=3) of the CD49a+

fraction, hence 0.11% of the initial fraction. No CFU-fswere observed in the Sca-1−/CD49a+ population, whereasall the mCFU-f potential was detected in the Sca-1+/CD49a+ fraction. The CE for Sca-1+/CD49a+ cells repre-sented 48±10 for 105 seeded cells and was significantlyhigher than the CE of both CD49a+ and initial fractions.Furthermore, flow cytometry cell sorting was performed(Figs. 1 and 5). As observed above, we also detected CFU-fs in the Sca-1+/CD49a+ population of cells. These resultsconfirmed that: (1) in the rodent, all BM Sca-1+ cellsexpressed the CD49a molecule, as shown by flow cytom-etry and (2) all native mCFU-fs were contained in the Sca-1+/CD49a+ population. Thus, native mCFU-fs were presentin the Sca-1+/CD49a+ fraction. Unfortunately, we wereunable to achieve the same double-selection in rat BMcells.

In conclusion, for either human or rodents, directenrichment of cells by CD49a concentrated CFU-fs intoan enriched population expressing a primitive stem cellphenotype: CD133+ in human and Sca-1+ in rat and mouse.

CFU-f culture conditions

Cultures of CD49a+ and initial fractions at day 15 weremonitored by phase-contrast microscopy. The emergence ofhuman and rat colonies is shown in Fig. 6a,b, respectively.In these species, detectable CFU-fs appeared after 7–10 days of culture and comprised spindle-like cells. Themorphology of rat and human cultures was identical with acomplete absence of adherent refringent haematopoietic-like cells. Rat cells strongly proliferated still we stopped theculture (at week 14). Furthermore, some clones of rCFU-fswere easily expanded from limiting dilutions, unlike hCFU-fs. These clones underwent 25 divisions without showing adecrease in their proliferation rate (not shown).

In the initial fraction cultures of the mouse, adherentspindle-shaped cell clusters made up of a few cells could beseen as early as day 5. Between day 5 and day 8, thenumber of mCFU-fs increased but numerous round adher-ent haematopoietic-like cells also emerged (Fig. 6c). Incontrast, in cultures of the CD49a+ fraction, mCFU-fs weredetected only on day 8 (Fig. 6d) and adherent haemato-poietic refringent cells failed to materialise. This wasconfirmed by the lack of the CD45 staining on culturedadherent cells as measured by flow cytometry (data notshown). Moreover, this also allowed us to determine theoptimum plating densities for the initiation of murinecultures: 14×103 and 125×103 cells/cm2 for both CD49a+

and initial fractions, respectively.

Mesenchymal stem cell multipotential studies

After expansion, human, rat and mouse cells from the CFU-fs were cultured in various inducing media: osteogenic,chondrogenic and adipogenic. Cells from the same BMsample were cultured in both differentiation medium andexpansion medium (as a negative control of differentiation).After 3 weeks of induction, cells were tested for theexpression of proteins specific to a mesenchymal lineage.

As is well known, osteoblastic cells can be characterisedin vitro by the specific expression of osteocalcin, by thestrong secretion of CO I and osteopontin and by calciumdeposition. High alkaline phosphatase expression has alsobeen described (Bruder et al. 1997). In contrast, chondro-blastic cells express specific glycoaminoglycanes (GAG)and CO II at the end of differentiation process (Barry et al.2001), whereas lipid droplets, stained with Nile red oil, areeasily detected within adipocytic cells.

Irrespective of the species, we observed strong expres-sion of CO I, alkaline phosphatase and osteopontin andnotable calcium deposition labelled by Alizarin red underosteogenic conditions in comparison with cells cultured inexpansion medium (Fig. 7). Moreover, the fluorescencedepicting osteocalcin was weak but undetectable in the

Fig. 5 Phenotypic characterisation of mouse CD49a+ cells using flowcytometry. Following the sorting, the CD49a+ cells were evaluated forthe expression of some molecules specific either to the BM stem cellsor mature hematopoietic cells. The expression of CD45, CD90, c-kit,Sca-1, CD34 and T-cell markers was assesed on total BM(■), CD49a+

(▦) and CD49a+ (□) cells

478 Cell Tissue Res (2007) 327:471–483

control. Under chondrogenic conditions, the cells aggregat-ed themselves, whereas such a phenomenon was notobserved when cells were cultured under other conditions(Fig. 7). CO II and Alcian blue staining were solelydetected in chondrogenic cultures. A large number of lipiddroplets, specifically stained by Nile red, were also detectedonly when cells were cultured in adipogenic medium(Fig. 7).

In conclusion, human, rat and mouse CFU-fs weremultipotential confirming their MSCs origin. Thus, the directenrichment of BM MSCs by using anti-CD49a mAbs waseffective for human and rats and to a lesser extent for mice. Inaddition, CFU-fs were only detected in the selected CD49a+

fractions and this step eliminated contaminating adherenthaematopoietic-like cells, even from murine cultures.

Discussion

Mammalian BM MSCs are normally isolated indirectly bytheir ability to adhere to plastic tissue culture flasks

(Pittenger et al. 1999; Phinney et al. 1999; Javazon et al.2001) and are characterised specifically by their CFU-fproducts. However, this method has severe drawbacks.Primarily, the prolonged culture of cells, independent ofspecies and cell type, alters the phenotype from thatdisplayed by the native progenitors.

In this study, we have therefore used CD49a directenrichment (1) to optimise selection protocols for theisolation of native MSCs from several species, (2) toextend our characterisation of native hMSCs and themorphology and phenotypes of rodent native MSCs and(3) to ascertain whether direct enrichment of mMSCs fromvarious strains can be used to provide a model to studyhuman stem cell biology. Here, we demonstrate, for the firsttime that, independently of the percentage of recoveredcells and the type of mAb used, all CFU-fs are presentexclusively in the CD49a+ fractions for all species tested. Inour initial experiments on human cells, we had utilised theDynal kit to isolate the CD49a+ fraction (Deschaseaux et al.2003). This report shows that the use of the Miltenyi kit ismore efficient for the enrichment of human CFU-fs than is

Fig. 6 Phase contrast micro-graphs of human (a), rat (b)and mouse (c, d) CFU-fs fromprimary culture of CD49a+

(a, b, d) and initial (c) fractions.×40. At day 10 of culture,adherent spindle-shaped cellsformed clusters in both frac-tions. No adherent refringenthaematopoietic cells weredetected in rCFU-f and hCFU-fcultures from initial and CD49a+

fractions, although some werefound in murine cultures of theinitial population of cells(arrowheads)

Cell Tissue Res (2007) 327:471–483 479

the Dynal kit. This efficient selection has allowed us todetect the CD133 stem-cell-specific marker (Yin et al.1997; Uchida et al. 2000) in our population. We havepreviously demonstrated that directly enriched hMSCs areCD45med,low and have found, in this study, that all cellsfrom this population were also CD133+. Furthermore, whenusing both immunomagnetic selection and flow cytometrycell sorting, we have observed human CFU-fs within theCD49a+/CD133+ fraction. Together, these results confirmthe stem or progenitor cell origin of these enriched cells.However, cultured MSCs become CD133– (not shown)without apparent lost of differentiation potential, as de-scribed in the literature (Yin et al. 1997; Peichev et al.

2000; Uchida et al. 2000). Thus, as for other markers, e.g.CD45, CD34 and CD49d (Deschaseaux et al. 2003), theCD133 molecule is down-regulated by the expansionprocess.

Direct sorting of BM hMSCs has been carried out byusing three types of mAbs: anti-nerve growth factorreceptor (NGFR or CD271) mAb (Quirici et al. 2002),Stro-1 mAb (Peichev et al. 2000) and anti-CD49a mAb(Deschaseaux et al. 2003), with a large population ofCD34+ and CD133+ cells having been detected within thehuman BM NGFR+ fraction. These results are thus inaccordance with our report. However, the enrichment factorfor NGFR selection has been found to be 45 (Quirici et al.

Fig. 7 Mesenchymal differenti-ation potential assessment. Hu-man, mouse and rat cells fromCFU-fs were cultured in variousmedia in order to induce osteo-blastic, chondroblastic and adi-pocytic lineages. After 3 weeks,the cultures were tested by usingin situ immunofluorescence forthe expression of moleculesspecific to osteoblasts (CO I,osteopontin), chondroblasts (COII) and adipocytes (intracyto-plasmic lipid droplets) or byhistochemical staining (Alizarinred for calcium deposition spe-cific to osteoblasts; Alcian bluefor GAG specific to cartilagi-nous tissue). Negative controls(MSCs without induction ofdifferentiation) were used toevaluate the expression of thetested molecules and showedeither no or low fluorescence(inserts). Under osteogenic con-ditions, as depicted by phasecontrast (first row), the culturescontained small aggregates re-sembling mineral accumulation.×50. This was confirmed byAlizarin red staining (secondrow). Fluorescence for osteo-pontin was bright and foundboth intra- and extracellularlyfor all species (third row). ×600.CO I protein was observedessentially extracellularly in ratand human cultures and intra-cellularly for mouse cells (fourthrow). ×600. Cells cultured inchondrogenic-inducing mediumgathered together and formedsmall pellets positive for CO II(fifth row). ×600. The GAG inthese cells was stained byAlcian blue (sixth row). Intra-cellular lipid droplets were eas-ily detected after using Nile redoil staining (bottom row). ×600

480 Cell Tissue Res (2007) 327:471–483

2002) compared with 140 for the CD49a isolation; hMSCsthus seem to be more highly enriched by using the latterprotocol. In order to confirm this, both types of selectionshould be studied with the same BM samples. Previousstudies have shown that Stro-1 is non-reactive withhaematopoietic progenitors and all detectable clonogenicCFU-fs are essentially included within the Stro-1+ popula-tion (Simmons and Torok-Storb 1991). Moreover, otherdata also strongly suggest that primitive mesenchymalprecursors including native MSCs with the capacity todifferentiate into multiple lineages are restricted to the Stro-1+ fraction in adult human BM (Dennis et al. 2002).However, the Stro-1 antibody is specific to human cells andits target is as yet uncharacterised. Recently, in a compar-ison of the efficiency of mAbs recognising a series ofstromal antigens (Stro-1 antigen, HOP-26, CD49a and SB-10/CD166) to enrich CFU-fs prior to culture, Stewart et al.(2003) have found that essentially all CD49a+ cells co-express Stro-1 and that a one-pass immunoselection withanti-CD49a mAb enriches the number of CFU-fs obtainedfrom human BM MNCs 18-fold; their report confirms ourprevious conclusion that the direct selection of hMSCs canbe achieved by using antibodies against the CD49amolecule (Deschaseaux et al. 2003). However, our enrich-ment factor here is 8-fold higher (enrichment factor: 143).This discrepancy could be attributable to the protocol used,because 24% of cells are recovered compared with our70%, possibly as a result of a fraction of BM MSCs beinglost in the former separation.

The CD49a molecule is known to be highly conserved inmammals. The homology is over 90% between humans andmice, whereas a comparison of the rat and murinemolecules shows a 94% similarity. As rodent species arecrucial for research into stem cell biology, we have testedthe direct enrichment of native MSCs by using anti-CD49amAb. As found for humans, all MSCs have been detectedin the CD49a+ fraction. In the rat model, we have shownthat the CD49a+ fraction represents 1.1% of MNCs with aCE corresponding to 97 for 105 seeded cells. Rat CD49a+

cells highly express the CD90 stem cell antigen and wehave failed to detect CD45–/CD90+ cells. Moreover, theCD90+/CD49a+ cells are either CD45high or CD45med,low.Thus, three hypotheses can be proposed: native rMSCs are(1) CD90+/CD45+, (2) CD90–/CD45+ or (3) CD90–/CD45–. Future research should confirm which of these iscorrect. We have also tested the expression of the Sca-1molecule. Sca-1 is a GPI-linked cell-surface moleculefound on haematopoietic, mammary, gland and cardiaccells and BM MSCs. Sca-1 has been shown to be necessaryfor normal HSC activity (Badfute et al. 2005). Moreover,Sca-1 has also indirectly been described as a marker ofmMSCs influencing the self renewal and cell-fate decisionsof mesenchymal progenitors and other stem cells, since

Sca-1 knockout mice have decreased numbers of cells intheir stem cell/progenitors pools (Benveniste et al. 2003).Here, we have shown that the selection of CD49a+ BMcells also enriches Sca-1+ cells. Moreover, all of the mCFU-fs are contained in the CD49a+/Sca-1+ population. Thisfinding is in agreement with other reports that havedemonstrated that cultured mMSCs express the haemato-poietic stem cell specific marker Sca-1 (Jiang et al. 2002;Benveniste et al. 2003; Peister et al. 2004). However, atpresent, we are unable to perform Sca-1 selection on ratCD49a+ cells, because of the failure of the Dynal Cellectionkit to isolate rat CD49a+ cells. The human homologue ofSca-1 is the CD87 protein, viz. the urokinase-like plasmin-ogen activator receptor (uPAR; Blasi and Carmeliet 2002).We have therefore tested CD87 expression on CD49a+ cellsbut have failed to detect it (not shown). This resultdistinguishes human from rodent BM MSCs. Interestingly,CD87 is a ligand for some integrins (CD49c, CD49d andCD49f) and for an integrin-associated protein (Tarui et al.2001). These complexes can cis- or trans-activate theoutside-in signals. Thus, Sca-1-like proteins might beassociated with the CD49a molecule to induce an unknownrole in MSC biology. CD87 has not been observed inhuman native BM MSCs, which can however cleave uPAR.Other reports have shown that some strains MSCs are alsoSca-1– (Peister et al. 2004). Moreover, Solberg et al. (2001)have reported that uPAR is also highly expressed in tissuesactively undergoing remodelling. As MSCs are harvestedfrom BM and as the remodelling of bone is low (comparingwith that of haematopoietic cells), uPAR can be down-regulated. In addition, the involvement of this protein mightbe either species- or strain-dependent.

We have not directly compared the proliferation potentialof rMSCs but we have obtained more than 38×106

expansions for one clone (the equivalent of 25 divisions)without a decrease in its growth rate (not shown). Thisvalue is higher than one previously reported for hMSCs(Deschaseaux et al. 2003). Collectively, our data show thatrat and human cultures are broadly the same. Nevertheless,we have found two exceptions: (1) directly enriched ratCD49a+ cells do not appear to contain T-cells in contrast totheir human counterparts and (2) rMSCs are more prolif-erative than hMSCs, as observed elsewhere (Javazon et al.2001). These exceptions probably reflect species differ-ences (Deschaseaux et al. 2003).

Murine cultures are however somewhat different. Wehave used our direct enrichment method on BM fromvarious murine strains (FVB, BALB/c, C57BL) by usingdiverse anti-CD49a mAbs. The percentage of directlyenriched CD49a+ cells is always higher (and lower for theenrichment factor) than that recovered from human or ratCD49a+ preparations. This is probably a reflection ofspecies differences or, perhaps the use of beads from the

Cell Tissue Res (2007) 327:471–483 481

Dynal kit is less stringent in purifying human or rat cells.The contamination by CD45+ cells might be also explainedby the mouse origin of the mAb used (clone 3A3)increasing the unspecific binding in mouse BM. Inaddition, when using indirect isolation methods, mCFU-fshave haematopoetic cells adhering to them and this resultsin the production of mixed cultures (Phinney et al. 1999).This contamination has been suggested to be attributable tothe presence of mature haematopoietic cells that arecontained in the centre of the bone and that are collectedin murine BM harvests after bone-flushing (Sun et al.2003). However, rat BM is collected in the same fashion.Haematopoietic cell growth in culture is also partlyattributable to the capacity of murine haematopoietic cellsto adhere directly to the tissue culture plastic and tofibroblastoid cells (Baddoo et al. 2003). By using a culturedCD49a+ fraction, this problem has been eliminated, despitethe enrichment of CD45+ cells as shown by flow cytometricstudies. Thus, direct enrichment also enables the deletedadherent haematopoietic cells to be characterised. TheseCD45+ deleted cells might therefore be the cells that persistin cultures of indirectly selected mCFU-fs.

As calculated for human and rat species, we haveestimated that the number of mMSCs in BALB/c BMcontained in the initial fraction is 1 in 33,000, changing to avalue of 1 in 5,320 for CD49a+-sorted cells. These figuresare similar to those found in the literature (Meirelles andNardi 2003). In setting up a mouse model for MSCcultures, the choice of an appropriate strain may beimportant, since our study has also revealed intra-specificdifferences. For example, the CE of the initial fraction issignificantly higher for BALB/c than for both C57BL andFVB. Other investigators using different murine strainshave similarly reported this phenomenon (Phinney et al.1999; Tropel et al. 2004).

The capacity to enrich unprocessed mMSCs directly, asfor rat and human cells, has allowed us to phenotype thecells more accurately. In the CD49a+ fraction, we have beenable to show enrichment of markers historically thought tobe exclusive to the haematopoietic lineage, such as Sca-1,CD34, CD133 and c-kit. The detection of CFU-fs in theCD49a+/Sca-1+ and CD49a+/CD133+ fractions purified byboth immunomagnetic and flow cytometer support theproposal that rodent MSCs are CD49a+/Sca-1+ and hMSCsare CD49a+/CD133+.

In conclusion, BM MSCs can readily be directlyenriched by one separation step via the CD49a marker.This removes the differences previously found betweencultures of various species. The conserved expression ofthis integrin subunit by BM native MSCs between differentspecies might signify a conserved “niche” for this type ofBM stem cell. Confirmation of this hypothesis would thusbe of interest. In addition, the general applicability of the

method allows the creation of a more realistic murinemodel for hMSCs, with CD49a specificity permitting stemcells to be phenotyped accurately. This in turn may providean opportunity to examine the MSCs compartment furtherand to gain insights into this aspect of stem cell biology.

References

Baddoo M, Hill K, Wilkinson R, Gaupp D, Hughes C, Kopen GC,Phinney DG (2003) Characterization of mesenchymal stem cellsisolated from murine bone marrow by negative selection. J CellBiochem 89:1235–1249

Badfute SB, Graubert TA, Goodell MA (2005) Roles of Sca-1 inhematopoietic stem/progenitor cell function. Exp Hematol33:836–843

Banfi A, Muraglia A, Dozin B, MastrogiacomoM, Cancedda R, QuartoR (2000) Proliferation kinetics and differentiation potential of exvivo expanded human bone marrow stromal cells: implications fortheir use in cell therapy. Exp Hematol 28:707–715

Barry F, Boynton RE, Liu B, Murphy JM (2001) Chondrogenicdifferentiation of mesenchymal stem cells from bone marrow:differentiation-dependent gene expression of matrix components.Exp Cell Res 268:189–200

Benveniste P, Cantin C, Hyam D, Iscove NN (2003) Hematopoieticstem cells engraft in mice with absolute efficiency. Nat Immunol4:708–713

Bianco P, Riminucci M, Gronthos S, Robey PG (2001) Bone marrowstromal stem cells: nature, biology, and potential applications.Stem Cells 19:180–192

Blasi F, Carmeliet P (2002) uPAR: a reversible signaling orchestrator.Nat Rev 3:932–943

Bonyadi M, Waldman SD, Liu D, Aubin JE, Grynpas MD, StanfordWL (2003) Mesenchymal progenitor self-renewal deficiencyleads to age-dependent osteoporosis in Sca-1/Ly-6A null mice.Proc Natl Acad Sci USA 100:5840–5845

Bruder SP, Jaiswal N, Haynesworth SE (1997) Growth kinetics, self-renewal, and the osteogenic potential of purified humanmesenchymal stem cells during extensive subcultivation andfollowing cryopreservation. J Cell Biochem 64:278–294

Conget PA, Minguell JJ (1999) Phenotypical and functional propertiesof human bone marrow mesenchymal progenitor cells. J CellPhysiol 181:67–73

Craig W, Kay R, Culter RL, Lansdorp PM (1993) Expression of Thy-1on human hematopoietic progenitor cells. J Exp Med 177:1331–1342

Dennis JE, Carbillet JP, Caplan AI, Charbord P (2002). The STRO-1+

marrow cell population is multipotential. Cells Tissues Organs170:73–82

Deschaseaux F, Gindraux F, Saadi R, Obert L, Chalmers D, Herve P(2003) Direct selection of human bone marrow mesenchymal stemcells using an anti-CD49a antibody reveals their CD45med,low

phenotype. Br J Haematol 122:506–517Gao J, Dennis JE, Muzic RF, Lundberg M, Caplan AI (2001) The

dynamic in vivo distribution of bone marrow-derived mesenchy-mal stem cells after infusion. Cells Tissues Organs 169:12–20

Gardner H, Kreidberg J, Koteliansky V, Jaenisch R (1996) Deletion ofintegrin α1 by homologous recombination permits normalmurine development but gives rise to a specific deficit in celladhesion. Dev Biol 175:301–313

Jaiswal N, Haynesworth SE, Caplan AI, Bruder SP (1997) Osteogenicdifferentiation of purified, culture-expanded human mesenchymalstem cells in vitro. J Cell Biochem 64:295–312

482 Cell Tissue Res (2007) 327:471–483

Javazon EH, Colter DC, Schwarz EJ, Prockop DJ (2001) Rat marrowstromal cells are more sensitive to plating density and expandmore rapidly from single-cell-derived colonies than humanmarrow stromal cells. Stem Cells 19:219–225

Jiang Y, Jahagirdar BN, Reinhardt RL, Schwartz RE, Keene CD,Ortiz-Gonzalez XR, Reyes M, Lenvik T, Lund T, Blackstad M,Du J, Aldrich S, Lisberg A, Low WC, Largaespada DA,Verfaillie CM (2002) Pluripotency of mesenchymal stem cellsderived from adult marrow. Nature 418:41–49

Kitano Y, Radu A, Shaaban A, Flake AW (2000) Selection,enrichment, and culture expansion of murine mesenchymalprogenitor cells by retroviral transduction of cycling adherentbone marrow cells. Exp Hematol 28:1460–1469

Krause DS, Theise ND, Collector MI, Henegariu O, Hwang S,Gardner R, Neutzel S, Sharkis SJ (2001) Multi-organ, multi-lineage engraftment by a single bone marrow-derived stem cell.Cell 105:369–377

Kuznetsov SA, Friedenstein AJ, Robey PG (1997) Factors requiredfor bone marrow stromal fibroblast colony formation in vitro. BrJ Haematol 97:561–570

Makihira S, Yan W, Ohno S, Kawamoto T, Fujimoto K, Okimura A,Yoshida E, Noshiro M, Hamada T, Kato Y (1999) Enhancementof cell adhesion and spreading by a cartilage-specific non-collagenous protein, cartilage matrix protein (CMP/Matrilin-1),via integrin a1b1. J Biol Chem 274:11417–11423

Meirelles Lda S, Nardi NB (2003) Murine marrow-derived mesen-chymal stem cell: isolation, in vitro expansion, and characteriza-tion. Br J Haematol 123:702–711

Muraglia A, Cancedda R, Quarto R (2000) Clonal mesenchymalprogenitors from human bone marrow differentiate in vitroaccording to a hierarchical model. J Cell Sci 113:1161– 1166

Peichev M, Naiyer AJ, Pereira D, Zhu Z, Lane WJ, Williams M, OzMC, Hicklin DJ, Witte L, Moore MA, Rafii S (2000) Expressionof VEGFR-2 and AC133 by circulating human CD34+ cellsidentifies a population of functional endothelial precursors.Blood 95:952–958

Peister A, Mellad JA, Larson BL, Hall BM, Gibson LF, Prockop DJ(2004) Adult stem cells from bone marrow (MSCs) isolatedfrom different strains of inbred mice vary in surface epitopes,rates of proliferation, and differentiation potential. Blood103:1662–1668

Phinney DG, Kopen G, Isaacson RL, Prockop DJ (1999) Plasticadherent stromal cells from the bone marrow of commonly usedstrains of inbred mice: variations in yield, growth, and differen-tiation. J Cell Biochem 72:570–585

Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, MoscaJD, Moorman MA, Simonetti DW, Craig S, Marshak DR (1999)Multilineage potential of adult human mesenchymal stem cells.Science 284:143–147

Quirici N, Soligo D, Bossolasco P, Servida F, Lumini C, LambertenghiDeliliers G (2002) Isolation of bone marrow mesenchymal stemcells by anti-nerve growth factor receptor antibodies. Exp Hematol30:783–791

Simmons PJ, Torok-Storb B (1991) Identification of stromal cellprecursors in human bone marrow by a novel monoclonalantibody, STRO-1. Blood 78:55–62

Solberg H, Ploug M, Hoyer-Hansen G, Nielsen BS, Lund LR (2001)The murine receptor for urokinase-type plasminogen activator isprimarily expressed in tissues actively undergoing remodelling. JHistochem Cytochem 49:237–246

Stewart K, Monk P, Walsh S, Jefferiss CM, Letchford J, Beresford JN(2003) STRO-1, HOP-26 (CD63), CD49a and SB-10 (CD166) asmarkers of primitive human marrow stromal cells and their moredifferentiated progeny: a comparative investigation in vitro. CellTissue Res 313:281–290

Sun S, Guo Z, Xiao X, Liu B, Liu X, Tang PH, Mao N (2003)Isolation of mouse marrow mesenchymal progenitors by a noveland reliable method. Stem Cells 21:527–535

Tarui T, Mazar AP, Cines DB, Takada Y (2001) Urokinase-typeplasminogen activator receptor (CD87) is a ligand for integrinsand mediates cell-cell interaction. J Biol Chem 276:3983–3990

Tropel P, Noel D, Platet N, Legrand P, Benabid AL, Berger F (2004)Isolation and characterisation of mesenchymal stem cells fromadult mouse bone marrow. Exp Cell Res 295:395–406

Uchida N, Buck DW, He D, Reitsma MJ, Masek M, Phan TV,Tsukamoto AS, Gage FH, Weissman IL (2000) Direct isolation ofhuman nervous system stem cells. Proc Natl Acad Sci USA97:14720–14725

Yin AH, Miraglia S, Zanjani ED, Almeida-Porada G, Ogawa M,Leary AG, Olweus J, Kearney J, Buck DW (1997) AC133, anovel marker for human hematopoietic stem and progenitor cells.Blood 90:5002–5012

Cell Tissue Res (2007) 327:471–483 483