mdoTcta(t(gpsoflssIubGciiaeatacctdme
dc3
Experimental Cell Research 250, 485–498 (1999)Article ID excr.1999.4528, available online at http://www.idealibrary.com on
A Recombinant Human TGF-b1 Fusion Protein with Collagen-BindingDomain Promotes Migration, Growth, and Differentiation
of Bone Marrow Mesenchymal CellsJose A. Andrades,*,† Bo Han,* Jose Becerra,*,†,1 Nino Sorgente,*,‡ Frederick L. Hall,* and Marcel E. Nimni*
*Surgical Research Laboratories, School of Medicine, University of Southern California, 1335 San Pablo Street, DOH-104, Los Angeles,California 90033; †Department of Cell Biology and Genetics, Faculty of Sciences, Campus Universitario de Teatinos, University of
Malaga, 29071-Malaga, Spain; and ‡Vide Pharmaceuticals, Los Angeles, California 90026
mscrmtpsWpcltriatgtwn
T
ppbcoc(Attpis
A continuous source of osteoblasts for normal boneaintenance, as well as remodeling and regeneration
uring fracture repair, is ensured by the mesenchymalsteoprogenitor stem cells of the bone marrow (BM).he differentiation and maturation of osteoprogenitorells into osteoblasts are thought to be modulated byransforming growth factors-b (TGF-b1 and TGF-b2)nd TGF-b-related bone morphogenetic proteinsBMPs). To define the responses of mesenchymal os-eoprogenitor stem cells to several growth factorsGFs), we cultured Fischer 344 rat BM cells in a colla-en gel medium containing 0.5% fetal bovine serum forrolonged periods of time. Under these conditions,urvival of BM mesenchymal stem cells was dependentn the addition of GFs. Recombinant hTGF-b1-F2, ausion protein engineered to contain an auxiliary col-agen binding domain, demonstrated the ability toupport survival colony formation and growth of theurviving cells, whereas commercial hTGF-b1 did not.nitially, cells were selected from a whole BM cell pop-lation and captured inside a collagen network, on theasis of their survival response to added exogenousFs. After the 10-day selection period, the survivingells in the rhTGF-b1-F2 test groups proliferated rap-dly in response to serum factors (10% FBS), and max-mal DNA synthesis levels were observed. Upon theddition of osteoinductive factors, osteogenic differ-ntiation in vitro was evaluated by the induction oflkaline phosphatase (ALP) expression, the produc-ion of osteocalcin (OC), and the formation of miner-lized matrix. Concomitant with a down-regulation ofell proliferation, osteoinduction is marked by in-reased ALP expression and the formation of colonieshat are competent for mineralization. During the in-uction period, when cells organize into nodules andineralize, the expression of OC was significantly el-
vated along with the onset of extracellular matrix
1 To whom correspondence and reprint requests should be ad-ressed at Dpto. de Biologıa Celular y Genetica, Facultad de Cien-ias, Campus Universitario de Teatinos, 29071 - Malaga, Spain. Fax:
t4-952 13 20 00. E-mail: [email protected].
485
ineralization. Differentiation of BM mesenchymaltem cells into putative bone cells as shown by in-reased ALP, OC synthesis, and in vitro mineralizationequired the presence of specific GFs, as well as dexa-ethasone (dex) and b-glycerophosphate (b-GP). Al-
hough rhTGF-b1-F2-selected cells exhibited the ca-acity to mineralize, maximal ALP activity and OCynthesis were observed in the presence of rhBMPs.e further report that a novel rhTGF-b1-F2 fusion
rotein, containing a von Willebrand’s factor-derivedollagen binding domain combined with a type I col-age matrix, is able to capture, amplify, and stimulatehe differentiation of a population of cells present inat BM. When these cells are subsequently implantedn inactivated demineralized bone matrix (iDBM)nd/or diffusion chambers into older rats they are ableo produce bone and cartilage. The population of pro-enitor cells captured by rhTGF-b1-F2 is distinct fromhe committed progenitor cells captured by rhBMPs,hich exhibit a considerably more differentiated phe-otype. © 1999 Academic Press
Key Words: osteoprogenitor cells; collagen matrix;GF-bs; BMPs; ectopic ossification.
INTRODUCTION
Bone, a dynamic tissue that provides structural sup-ort for soft tissues as well as stores of calcium andhosphate, depends for its normal function on a criticalalance between formation and resorption. Since osteo-ytes and osteoclasts do not replicate, the maintenancef a healthy bone must rely on an exogenous source ofells which appear to be derived from the bone marrowBM) mesenchymal stem precursor cells (MSCs) [1].lthough the biochemical characteristics of a pluripo-
ent MSC which can give rise to a variety of differen-iated cells has not been adequately defined, this hy-othesis provides a useful conceptual basis fornvestigating the developmental potential of marrowtromal tissue. Friedenstein [2] was the first to show
he osteogenic potential of such cells in the mouse.0014-4827/99 $30.00Copyright © 1999 by Academic Press
All rights of reproduction in any form reserved.
ScmfsilotiocfpaccgidcNtg
eosatlTtar(
rTcsrtAbiceatstMoBi
uieat[
uthl(fWtibacTa
siiipsT
citgcrw(v
AtaaTtc[rhoodbp
486 ANDRADES ET AL.
ubsequent in vitro studies have shown the presence ofells with the phenotype of bone cells in both rat andouse BM when cultured under conditions permissive
or osteogenic development (L-ascorbate, dexametha-one, dex, b-glycerophosphate, b-GP, and demineral-zed bone matrix, DBM) [3–6]. A cloned murine celline from marrow stroma was established that displayssteogenic characteristics, providing direct evidencehat a single stromal progenitor cell can differentiatento osteoblast-like cells [7]. More recently, on the basisf the evidence that adult human bone marrow-derivedolony-forming units are capable of differentiating intounctional osteoblasts [8], human clonogenic osteoblastrogenitors have been isolated and characterized [9],nd they display features of immature osteoprogenitorells which can differentiate into mature osteogenicells by cell–cell interactions or inducing agents. Theeneration of homogeneous populations of immortal-zed human marrow stromal cells by retroviral trans-uction demonstrates the capacity of a human clonalell line to differentiate in osteogenic direction [10, 11].evertheless, the biochemical signals that modulate
he differentiation of BM stem cells along the osteo-enic lineage are not well known.TGF-bs comprise a family of growth factors (GFs)
ncoded by closely related genes expressed in numer-us tissues and species. Bone was one of the first tis-ues in which it locally produced TGF-b-like factor thatppeared to regulate normal cellular function [12]. Al-hough the most concentrated source of TGF-b is plate-ets, bone represents the most abundant deposits ofGF-b in the body (200 mg/kg tissue), 100-fold greaterhan soft tissues such as placenta and kidney [13]. Inddition to TGF-b, bone contains a number of otherelated proteins, identified as osteogenic proteinsOPs) or bone morphogenetic proteins (BMPs).
Addition of TGF-b to osteoblastic-like cells has givenise to conflicting results. Whereas in some studies,GF-b has been shown in vitro to stimulate type Iollagen, osteocalcin (OC), osteonectin, osteopontinynthesis, and alkaline phosphatase (ALP) as well aseplication in osteoblast-like cells [14], other investiga-ions have shown that proliferation and expression ofLP and OC are inhibited [15–17]. TGF-b was found toe nonmitogenic and to cause a decrease in ALP activ-ty in osteoblasts derived from enzymatic digests of ratalvaria [18], to be highly mitogenic, and to have noffect on ALP [19]. These conflicting results may bettributed to a possible biphasic effect of TGF-b [12] oro differences in cell lineage characteristics, cell den-ity, the presence or absence of serum, culture condi-ions, and whether or not other GFs are present [20].oreover, Ballock et al. [21] have recently speculated
n the existence of different target cells for TGF-bs andMPs: the first acting on committed stem cells already
n the osteogenic pathway, while BMPs may influence t
ncommitted stem cells. TGF-b effects on chondrocytesn vitro remain controversial [22]: whereas they influ-nce cell morphology and synthesis of type II collagennd proteoglycans [18, 21] and increase cell prolifera-ion [23], TGF-bs seem to inhibit cell differentiation24].
It appears that MSCs are capable of self-renewal andndergo expansion in the presence of TGF-b1 [25]. Inhis study we have used a genetically engineeredTGF-b1 fusion protein, bearing an auxiliary von Wil-
ebrand’s factor-derived collagen binding domainrhTGF-b1-F2) [26]. Such a structure bestows on theactor the ability to bind to collagen type I specifically.
e hypothesized that the collagen binding domain ofhe rhTGF-b1-F2 fusion protein allows a slow releas-ng of the GF from the collagen matrix to which it isound and therefore a longer half-life and better avail-bility to the target cells. Thus, under the appropriateell culture conditions, we would be able to select forGF-b-responsive stem cells and to direct them alongn osteogenic lineage.With this in mind, we developed an in vitro culture
ystem for BM mesenchymal progenitor cells, whichncludes a mesenchymal cell compatible collagen gelmpregnated with the rhTGF-b1-F2. The cultures werenitiated under low-serum conditions, operating on therinciple that under these serum-restricted circum-tances the hematopoietic cells would die, but not theGF-b-responsive osteoprogenitors.To further evaluate the ability of this system to
apture and expand cells with osteogenic potential, wenvestigated the capacity of these in vitro selected cellso undergo differentiation through the chondro/osteo-enic lineage, after in vitro reimplantation. In theourse of these studies, we compared the effects ofecombinant hTGF-bs with those osteogenic factorshich have been widely documented, such as OP-1
BMP-7) and BMP-2, as well as bFGF, which is in-olved in early morphogenetic events.
MATERIALS AND METHODS
Genetic engineering of a recombinant hTGF-b1-F2 fusion protein.prokaryotic expression vector was engineered to produce a tripar-
ite fusion protein (Fig. 1) consisting of a 6xHis purification tag, anuxiliary vWF-derived collagen binding domain (WREPSFMALS),nd the carboxy-terminal 112-amino-acid sequence of humanGF-b1 active fragment (TGF-b1-F2). This was labeled TGF-b1-F2o distinguish it from an earlier prototype which did not contain theollagen binding decapeptide and which was designated TGF-b1-F126, 27]. The recombinant fusion protein was isolated from Esche-ichia coli inclusion bodies, solubilized with 8 M urea, purified toomogeneity using nickel chelate chromatography, and renatured byxidative refolding under optimized redox conditions. Stock solutionsf rhTGF-b1-F2 were prepared in 0.1% BSA/0.01 N HCl and stan-ardized for protein concentration by the Bio-Rad procedure. Theiological activity of this construct was evaluated by in vitro cellroliferation assays, using purified commercial hTGF-b1 (R&D Sys-
ems, Minneapolis, MN) as a standardized control.
Cguwebt7OGc
rftca1Tt
ButttwB2tv3mstfi(BCmlal
2pcahrw
aM(
dimco
htdmp
487TYPE I COLLAGEN AND rhTGF-b1-F2 IN OSTEOGENESIS
Isolation and culture of BM stromal cells inside collagen gels.ollagen matrices were prepared using rat tail tendon type I colla-en isolated and purified as described previously [28]. The solutionsed to prepare the collagen substrate for the experiments was madeith purified collagen plus 0.1 M NaOH and 53 alpha-minimumssential medium (a-MEM), containing 0.5% heat-inactivated fetalovine serum (FBS; GIBCO Lot No. 31K8041, Gaithersburg, MD) inhe ratio 4:1 for a final concentration of 0.35 mg/ml of collagen, pH.4. Forty-eight-well plates (Falcon, Div. Becton–Dickinson and Co.,xnard, CA) were coated with 100 ml of collagen without cells andFs and placed for 30 min at 37°C in order for it to solidify into a thin
ollagen gel matrix.BM stromal cells were isolated from 2-month-old male Fischer 344
ats by the method of Maniatopoulos et al. [29]. Briefly, tibias andemurs were excised aseptically and cleaned of adhering soft tissues,he epiphysis was cut, and the marrow flushed out with a stream ofomplete medium. A single cell suspension was obtained by gentlyspirating the disrupted marrow several times sequentially through8-, 20-, and 22-gauge needles and filtered through a sterile 20-mmeflon sheet (Cell Strainer, Falcon, Lincoln Park, NJ) to exclude anyissue debris or cells clumps.
FIG. 1. Schematic representation of the genetically engineeredTGF-b1-F2 fusion construct. The expressed protein contains a his-idine purification tag, a protease site, an auxiliary collagen bindingomain, and the carboxy-terminal 112-amino-acid sequence of theature human TGF-b protein. Adapted from Gordon et al. with
ermission.
FIG. 2. Schematic diagram of the experimental proce
To standardize the biological activities of the various GFs for ourM cell cultures, different concentrations of the GFs were evaluatedsing a chemotactic response assay. For each experimental condi-ion, 150 ml of collagen (0.35 mg/ml) was mixed with GF at concen-rations previously selected as those that induced maximal chemo-actic activity. Briefly, confluent cultures of primary rat BM cellsere dissociated and cells placed in the upper compartments ofoyden chambers (Neuroprobe, Rockville, MD) at a concentration of3 105 cells/ml of a-MEM. The lower compartment of the chemo-
axis chamber contained medium (controls) or medium to whicharious concentrations of GFs were added. After 4 h of incubation at7°C in 5% CO2, the filters were removed from the chambers, fixed inethanol, washed, stained, and mounted upside down on glass
lides. Chemotaxis was quantified by counting the number of cellshat migrated in 20 light microscope high-power fields (3400) perlter. The concentrations of GFs selected were hTGF-b1 0.5 ng/ml
R&D Systems), rhTGF-b1-F2 1 ng/ml, rhOP-1 40 ng/ml (Creativeiomolecules, Boston, MA), rhBMP-2 50 ng/ml (Genetics Institute,ambridge, MA), and bFGF 2.5 mg/ml (R&D Systems). We alsoeasured the chemotactic activity of BM cells in low serum (simu-
ating the selection period and in 10% FBS media (simulating themplification period—see Fig. 2), either with or without type I col-agen matrices with added GFs at the selected concentrations.
BM cells were mixed with the collagen–GF solution at a density of3 106 cells/150 ml collagen/well in 48-well plates. The culture
lates, 6 plates per experiment, were left 15 min at 37°C to allow theollagen to gel. Then, 150 ml/well of a-MEM containing the GF wasdded on top of the collagen gel, and cultures were maintained in aumidified atmosphere of 95% air and 5% CO2 at 37°C and 100%elative humidity. Every 3 days the medium was changed and GFsere added. Control cultures did not contain exogenous GFs.The culture medium consisted of a-MEM, containing FBS and
ntibiotics [100 mg/ml penicillin G (Sigma Chemical Co., St. Louis,O), 50 mg/ml gentamicin sulfate (Sigma), and 0.3 mg/ml fungizone
Flow)].Figure 2 shows the schematic diagram of the experimental proce-
ure. For the initial 10 days, cultures were carried out in 0.5% FBSn order to eliminate hematopoietic cells (selection period, S). The
edium was then changed to 10% FBS (amplification period, A) andultured for an additional 4 days. To help with the induction ofsteogenesis, cells were cultured in the presence of 1028 M dex
dure. See Materials and Methods for explanation.
(tc
1HpD
tucsaAwofaTDCA7cuw
wpaNwd(mlync
eaw(As
gtft2d
wdintfwtpabi
Ct
yeeTN4sraphTt
fdmwC
tfitSddatfcd
tStwicbapa
I
(2Byssaieg(
488 ANDRADES ET AL.
Sigma) and 2 mM b-GP (Sigma) (induction period, I) for an addi-ional 7 days. These inducers were added each time the medium washanged.Biochemical analysis. DNA synthesis was determined at days 10,
4, 18, and 21, by a semiautomated microfluorimetric method usingoechst 33258 dye, a specific vital stain for DNA, as describedreviously [30]. Results were expressed as mean 6SD micrograms ofNA/well.Cell numbers were estimated in quadruplicate at the above men-
ioned times, using a hemocytometer. ALP activity was determinedsing a modification of a method described previously [31]. Briefly,ells were washed and homogenized in ice-cold 0.15 NaCl, 3 mModium biocarbonate buffer, pH 7.4. Homogenates were centrifugednd supernatants assayed for ALP with p-nitrophenol as a substrate.bsorbance was measured at 495 nm in an ELISA reader. All assaysere done in triplicate. Results were expressed as mean 6 SD unitsf ALP per micrograms of DNA/well. OC synthesis was extractedrom residues of ALP assays, with 10% formic acid for 16 h at 4°Cnd determined by radioimmunoassay using 125I rat OC (Biomedicalechnologies, Inc., Stoughton, MA), as described previously [31].ata were expressed as nanograms of OC per microgram of DNA.alcium (Ca) content [31] was measured using residues from theLP assay, weighed, and washed in porcelain crucibles overnight at00°C. The assay for Ca was carried out by comparison to standardalcium solutions using atomic absorption spectrophotometry. Val-es were expressed as micrograms of calcium per milligram of dryeight of tissue.Cell culture studies. Cell morphology and colony appearanceere documented at various times during culture, using an invertedhase-contrast Photomicroscope III (Zeiss, Oberkochen, Germany),nd photographed using Kodak (Eastman Kodak Co., Rochester,Y). Technical PAN 400 ASA black and white film. Mineralizationas examined in situ with the von Kossa stain. Colony size wasetermined by measuring the diameter (mm) of colonies at days 10end S period), 14 (A period), 18, and 21 (I period), using a gradedonocular eyepiece and a standard graticule. The number of colonies
arger than 1 mm2 was quantified by computer-assisted image anal-sis (Olympus CIA-102, Olympus Optical Co., Tokyo) by transilumi-ating plates on a light box and images were captured on a televisionamera (Olympus Vanox-S).When necessary, the cultures were fixed in 10% phosphate-buff-
red neutral Formalin or Bouin’s fixative, followed by dehydration inlcohol, embedding in paraffin, and sectioned at 5 mm. The sectionsere stained with hematoxylin–eosin (HE), picrosirius–hematoxilin
PSH) for detection of collagen under polarized light [32], and 0.5%lcian blue (AB; Fluka) in 0.1 N HCl, pH 1.0 [33], for detection ofulfated glycosaminoglycans (GAGs).In vivo studies. After 16 days of culture, collagen gels were di-
ested with collagenase as described, and cells dissociated by diges-ion with 0.05% trypsin–0.02% EDTA at 37°C. Cells were centri-uged, washed, and resuspended in serum-free MEM, counted, andransferred to inactivated DBM (iDBM) and diffusion chambers (Fig.). Inactivated DBM chambers prepared from compact bone [31], andiffusion chambers [34], were constructed as described previously.Seven-month-old male isogeneic (inbred strain) Fischer 344 ratsere anesthetized by a subcutaneous injection of pentobarbital so-ium 3.5 mg/100 g body weight (Abbot, North Chicago, IL). A 2-cmncision was made over the midline of the upper chest and subcuta-eous tissue plains, was dissected laterally to form pouches in whichhe chambers were implanted. Each rat received duplicate implantsor each experimental condition and one empty chamber. Woundsere closed, and washed with Betadine, and the animals returned to
heir cages after recovering from anesthesia. Four weeks later (im-lantation period, IM) the host rats were sacrificed by an overdose ofnesthetic, and the chambers harvested and immediately analyzedy soft roentogenogram. The tissues were further analyzed biochem-
cally and processed for light microscopy after appropriate fixation. shambers containing erythrocytes or showing other evidence of rup-ure were eliminated from the experiment.
Biochemical analysis of chamber contents. The biochemical anal-sis was carried out as described previously [31] in one chamber fromach pair. To measure ALP activity, iDBM chamber implants werexcised, weighed, and cut over ice into approximately 3-mm3 pieces.issues were then homogenized in a cold buffer solution (0.15 MaCl and 3 mM NaHCO3, pH 7.4) and centrifuged at 900g for 1 h at°C. Supernatants were assayed for ALP with p-nitrophenol as aubstrate. Activity was expressed as micromoles of p-nitrophenoleleased per gram of wet weight of tissue after 30 min of incubationt 37°C. OC was extracted from freeze-dried iDBM chamber im-lants, or residues after ALP assay, with 0.5 M EDTA in ammoniumydroxide and determined by radioimmunoassay as described above.he Ca content of iDBM implants was determined from freeze-driedissues or residues from ALP assay.
The contents of each diffusion chamber were carefully removedrom between the membrane filters and homogenized in 1 ml doubleistilled water at 4°C using an Ultra Turrax homogenizer (Brink-ann Instruments Inc., CA). The homogenates were transferred toeighed containers and analyzed for ALP activity, OC synthesis, anda content following the protocols described above.Histological studies of implants. Immediately after removal from
he host animals, one chamber of each pair was immersed in Bouin’sxative or fixed in 10% phosphate-buffered neutral Formalin. Inac-ivated DBM chambers were partially decalcified with Decalcifyingolution (Stephens Scientific, Riverdale, NJ), and the plastic rings ofiffusion chambers were dissolved in acetone. Chambers were dehy-rated with three changes of absolute ethanol, embedded in paraffin,nd cut in serial sections 5 mm thick sagitally at several levelshrough the entire chamber. Every 10th section was stained with HEor detection of cartilage and bone formation, PSH for detection ofollagen under polarized light, and 0.5% AB in 0.1 N HCl, pH 1.0, foretection of GAGs.Quantitative analysis and statistics. The areas of bone and car-
ilage were quantified using a computerized image analysis program.lides of the all histologic sections stained with picrosirius–hema-oxilin (or hematoxilin–eosin) from each experimental conditionere scanned on an Apple digital scanner (n 5 20). The computerized
mages were analyzed for cross-sectional area on a Macintosh IIxomputer using software (version 6.02) from BrainPower Inc. (Cala-asas, CA). Areas of cartilage and bone were measured in triplicatend averaged for each histological image. A Student’s t test waserformed after using the F test to determine the equality of vari-nces.
RESULTS
n vitro Studies
The concentrations of 0.5 ng/ml (hTGF-b1), 1 ng/mlrhTGF-b1-F2), 40 ng/ml (rhOP-1), 50 ng/ml (rhBMP-), and 2.5 mg/ml (bFGF) were considered the best forM cell migration. Zigmond-Hirsch checkerboard anal-sis shows that all GFs at the concentrations selectedtimulated chemotaxis and not chemokinesis (data nothown). Under conditions described under Materialsnd Methods, maximal chemotactic activity was exhib-ted in the presence of serum by all GFs, being mark-dly diminished when these were entrapped in colla-en gels, and rhTGF-b1-F2 showed the highest valuesTable 1).
Cell replication (Fig. 3) and cell number (data not
hown) were measured at different days of culture.
Cctlc1vntondsGdb
mcil4bOptrcTro(tgd
lmpOlfcOHi(tdp
tloratrtdsssfh
lptr
cowso
ta
hrrrb
ubmFa
489TYPE I COLLAGEN AND rhTGF-b1-F2 IN OSTEOGENESIS
ells cultured in type I collagen gels under low-serumonditions diminished drastically in number. Duringhe first 10 days of culture (S period), the lowest rep-ication and proliferation rates were detected in controlultures. Treatment of cultures with both hTGF-bs for0 days promoted cell survival, resulting in higheralues of DNA synthesis and moderate increases in cellumber. Maximal cell growth stimulation was ob-ained using rhTGF-b1-F2 at any time. In the presencef defined GFs, maximal increases in DNA and cellumber occurred at day 18, when normal serum con-itions were reestablished. Recombinant hTGF-b1-F2howed the highest effects. In cultures treated withFs, the presence of dex and b-GP for osteogenic in-uction, from days 10 to 16, resulted in a modest inhi-ition of cell proliferation (data not shown).Biochemical analysis. ALP is a well-known earlyarker of bone cell differentiation. Because of the low
ell number during the selection period (S period, thenitial 10 days of culture), values of ALP activity wereow but detectable in the presence of all GFs used (Fig.); however, cells cultured in the presence of rhTGF-1-F2 exhibited twofold higher levels of ALP activity.nce normal serum conditions were reestablished (am-lification period, A period, from days 11 to 14 of cul-ure), the ALP activity increased notably. Values forhTGF-b1-F2 were threefold greater than those forontrols and onefold greater than those for hTGF-b1.he highest levels of ALP (twofold greater thanhTGF-b1-F2 and fivefold greater than controls) werebtained after inducers (dex and b-GP) were addedinduction period, 1 period, from days 15 to 21 of cul-ure) to cells initially selected with rhBMPs. Basic FGFave somewhat higher values than controls, especiallyuring the A period (P , 0.002).
TABLE 1
Chemotactic Activity (Cells Migrated/20 OIF) of Rat BM inhe Presence of GFs, under Experimental Conditions Equiv-lent to S and A Periods
Media (% FBS) Collagen gel (% FBS)
0.5 10 0.5 10
TGF-b 22.2 6 5.0 28.0 6 5.0 6.0 6 1.0 8.0 6 1.7hTGF-b1-F2 23.5 6 4.0 30.0 6 5.0 11.0 6 3.0 14.0 6 4.0hOP-1 16.0 6 2.0 20.0 6 2.0 8.0 6 1.8 10.0 6 2.6hBMP-2 26.0 6 4.0 32.0 6 3.0 7.0 6 1.5 7.0 6 1.4FGF 35.0 6 4.0 40.0 6 3.0 4.0 6 0.3 6.0 6 1.0
Note. OIF, light microscope high-power fields (3400). GFs weresed at the following concentrations: hTGF-b1 (0.5 ng/ml), rhTGF-1-F2 (1 ng/ml), rhOP-1 (40 ng/ml), rhBMP-2 (50 ng/ml), bFGF (2.5g/ml). Bold characters highlight the maximum effect of rhTGF-b1-2. P , 0.05 compared with untreated conditions. Values are givens means 6SD for four filters.
OC, an osteoblast-specific protein and a well-estab- E
ished marker of the mature osteoblast phenotype, waseasured in the cell layer (Table 2) or as a secreted
rotein in the medium (data not shown). The amount ofC secreted into the culture media was consistently
ower than OC in cell layers. No detectable OC wasound in controls or in cells cultured with bFGF. BMells treated with hTGF-bs showed very low levels ofC during the early I period (dex and b-GP added).owever, cells cultured with rhBMPs synthesized OC
n cell layers and secreted it into the culture mediacells treated with rhBMP-2 in greater amounts thanhose treated with rhOP-1), exhibiting a slight increaseuring the early A and a steep increasing during the Ieriod.The precipitated Ca analyzed in cell cultures reflects
he ordered accumulation of mineral in the extracellu-ar matrix. No Ca was detected in controls or hTGF-b1-r bFGF-treated cells (Table 3). BM cells treated withhTGF-b1-F2 and rhBMPs did not show detectable Caccumulation prior to the I period. Maximum Ca con-ent was reached on day 21 in cultures treated withhBMP-2 (threefold greater than that in rhTGF-b1-F2-reated cultures). Ca accumulation was found to beependent on the amount of inducers added and wasignificantly diminished in the absence of dex (data nothown). Mineral deposition occurs during the late Itage, considered the period of mineralization. Valuesor Ca are in agreement with light microscopy andistochemical analysis described below.Morphological studies. BM cells cultured under
ow-serum conditions in the absence (controls) and theresence of GFs showed different morphological fea-ures. Small mononuclear round cells, presumably de-ived from the hematopoietic component of the BM,
FIG. 3. Influence of GFs on cell replication. Treatment of BMells in the presence of low serum (S period) decreases the synthesisf DNA even in the presence of GFs. When normal serum conditionsere reestablished (A period), the presence of GFs always resulted in
timulation of DNA synthesis. Maximal cell growth stimulation wasbtained using rhTGF-b1-F2, independently of the culture period.
ach bar represents the average 6SD of four samples.
dt5bRcliWtttc5csb
rami5s
puaiAohwtb
atgdr
V
490 ANDRADES ET AL.
ecrease dramatically within the first 10 days of cul-ure (S period). Untreated cultures remain sparse (Fig.a), while cultures treated with either TGF-b1 or TGF-1-F2 show a greater number of cells (Figs. 5b and 5c).ecombinant hTGF-b1-F2 was able to induce not only
ell aggregates but also well-defined colonies in theater S period. Some fibroblast-like cells were visible bynterference microscopy at different levels of the gel.
hen cultures were returned to 10% FBS (A period),he number of round blastoid cells increased in cul-ures treated with hTGF-b1 (Fig. 5e). In culturesreated with rhTGF-b1-F2, abundant well-separatedolonies of round blastoid cells were also evident (Fig.f). During the I period, mineralization took place. Inultures treated with hTGF-b1 no colonies were ob-erved (Fig. 5h), whereas in those treated with rhTGF-1-F2 prominent colonies were formed with small
FIG. 4. Effects of culture conditions on ALP activity. ALP exprell culture conditions. Once normal serum was reestablished (A periohe I period (day 18) for control and hTGF-b1-, rhTGF-b1-F2-, and bFreater than those for controls, and onefold greater than for hTGF-bifferentiation stage (I period, day 21), were very similar for rhOP-1 aepresents the average 6SD of four samples.
TAB
Effects of Culture Conditions
0 10
hTGF-b1 ND NDrhTGF-b1-F2 ND NDrhOP-1 ND NDrhBMP-2 ND ND
Note. ND, no OC detected. Control and bFGF cultures did not show
alues are given as means 6SD for four samples.ound cells. With the addition of the osteoinductivegents (see Materials and Methods), the extracellularatrix of these colonies subsequently became mineral-
zed, as indicated by Ca deposition into the matrix (Fig.i). In control cultures (Fig. 5g) there was evidence ofubstantial cell death throughout experimental period.The number of colonies was quantified as they ap-
eared (Fig. 6a). Only rhTGF-b1-F2 was able to stim-late colony formation during the early S period. Twices many colonies were formed at the end of this periodn the TGF-b group as in those treated with rhBMPs.fter 4 days of amplification (14 days in culture), col-ny numbers doubled in the presence of rhTGF-b1-F2;owever, only a slight increase over the initial numberas observed in the presence of rhBMPs. About 25% of
he colonies formed in cultures treated with rhTGF-1-F2 became mineralized in the A period. Approxi-
n decreased to minimum levels during the S period (day 10) underALP activity increased progressively, reaching the maximum during-treated cultures. The values of ALP for rhTGF-b1-F2 were threefoldMaximum values of ALP for rhBMPs, determined at the end of therhBMP-2 but twofold greater than those for rhTGF-b1-F2. Each bar
2
OC Synthesis (ng/mg DNA)
Days in culture
14 18 21
ND 0.11 6 0.02 NDND 0.14 6 0.01 ND
30 6 0.01 0.90 6 0.02 1.20 6 0.0241 6 0.02 1.22 6 0.04 1.30 6 0.03
synthesis at any time. P , 0.01 compared with untreated conditions.
ssiod),GF1.nd
LE
on
0.0.
OC
mtbptcmiUiceccalg
6mdsrtrgei
I
Aht0
Ar(Oc
ftawBwc
tcDwtwtvrabc
lovptfidloc(g
abfiPl
c
491TYPE I COLLAGEN AND rhTGF-b1-F2 IN OSTEOGENESIS
ately half as many colonies showed Ca deposition inhe presence of rhOP-1 as in the presence of rhTGF-1-F2. No mineralized colonies were detected in theresence of rhBMP-2 in the A period. Nevertheless, athe end of the experiment (day 21), the number ofolonies mineralized in the presence of rhBMP-2 wasuch greater (28 colonies/well) than of those mineral-
zed in the presence of rhTGF-b1-F2 (8 colonies/well).nlike cultures treated with rhTGF-b1-F2, where no
ncrease in the number of mineralized colonies oc-urred over time, almost all the colonies present min-ralized in cultures treated with the rhBMPs. In allases, the relationship between the number of capturedells (determined by DNA content and cell proliferationt the end of S period) and the number of colonies wasinear, thus suggesting that one captured cell generallyave rise to one colony.Colony sizes were found to be heterogeneous (Fig.
b). During the S period all colonies were less than 1m2 in size, and no differences were observed with
ifferent GFs treatments. The major increase in colonyize occurred during the A period: colony sizes in thehTGF-b1-F2 treated cultures were four times largerhan those at the S period and three times larger forhBMPs. Recombinant hTGF-b1-F2-treated culturesave rise to the largest colonies at the end of thexperiment (6 mm2), while rhBMPs colonies did notncrease much in size (approximately 3 mm2).
n Vivo Studies
Biochemical analysis. Figure 7a shows values forLP, OC, and Ca in iDBM-explanted chambers. BothTGF-bs behaved similarly; ALP was 2.75-fold greaterhan control (P , 0.005), OC was 4-fold greater (P ,.01), and Ca was 8-fold greater (P , 0.05).In diffusion chambers (Fig. 7b), maximum values forLP were found in hTGF-b-treated cells, particularlyhTGF-b1-F2: 5.5-fold greater than those for controlsP , 0.001) and double those of rhBMPs. Conversely,C levels were highest in chambers implanted with
TAB
Effects of Culture Conditions on
0 10
hTGF-b1 ND NDrhTGF-b1-F2 ND NDrhOP-1 ND NDrhBMP-2 ND ND
Note. ND, no Ca detected. Control, hTGF-b1, and bFGF cultures donditions. Values are given as means 6SD for four samples.
ells treated with rhBMP-2, 9-fold greater than those d
or controls (P , 0.01), and were significantly lower inhe hTGF-bs-treated cells. Values of Ca were very lownd similar for both hTGF-bs. Maximum Ca contentas detected in chambers with cells treated with rh-MP-2. OC or Ca was detected in chambers implantedith cells treated with bFGF. No Ca was detected in
ontrols.Histological studies. Four weeks after implanta-
ion, observations were made on either two or threehambers for each experimental condition. InactivatedBM chamber controls (Fig. 8a) were completely filledith loose fibrous connective tissue. Chambers con-
aining cells treated with hTGF-b1 (Fig. 8b) were filledith cartilage but no bone; chrondocytes appeared ma-
ure, with a loose cytoplasm and big lacunae. Inacti-ated DBM chambers containing cells treated withhTGF-b1-F2 (Fig. 8c) showed fibrous tissue and largemounts of cartilage, which in some cases appeared toe replaced by new bone adjacent to the walls of thehamber.Diffusion chamber controls were partially filled with
oose fibrous connective tissue, tightly attached to eachf the membrane filters, and containing collagen fibrilsisualized by PSH stain (Fig. 8d) and observed witholarized light (Fig. 8e). Chambers containing cellsreated with hTGF-b1 (Fig. 8f) produced a mixture ofbrous connective tissue, cartilage with mature chon-roblasts, and perichondral tissue surrounding carti-age. Large areas at the center of the chambers wereccupied by a dense fluid. Similar results were seen inhambers containing cells treated with rhTGF-b1-F2Fig. 8g), although the amount of tissue formed wasreater.Morphometrical results. Histological sections were
nalyzed quantitatively for cartilage and endochondralone formation, and the results showed statistical dif-erences for the various experimental conditions. InDBM chambers (data not shown), both hTGF-b1 (60%,
, 0.001) and rhTGF-b1-F2 (45%, P , 0.001) stimu-ated cartilage formation, but only rhTGF-b1-F2 in-
3
a Content (mg/mg Dry Weight)
Days in culture
14 18 21
ND ND NDND 8.0 6 2.0 12.3 6 3.18ND 19.2 6 5.17 32.0 6 8.10ND 22.0 6 3.96 37.2 6 7.10
ot show Ca content at any time. P , 0.05 compared with untreated
LE
C
id n
uced significant bone formation (5%, P , 0.01). Com-
FIG
.5.
Mor
phol
ogy
ofcu
ltu
res
atdi
ffer
ent
stag
esof
sele
ctio
n,p
roli
fera
tion
,an
ddi
ffer
enti
atio
n.A
fter
the
Spe
riod
,on
lycu
ltu
res
trea
ted
wit
hrh
TG
F-b
1-F
2(c
)sh
owed
suffi
cien
tce
lln
um
bers
tofo
rmcl
earl
yor
gan
ized
colo
nie
sdu
rin
gpr
olif
erat
ion
(Ape
riod
)st
age
(f).
Col
onie
sco
nta
ined
rou
nd-
shap
edce
lls,
and
they
min
eral
ized
(Ipe
riod
)st
arti
ng
from
the
cen
ter
ofth
eco
lon
y(i
).C
ontr
olco
ndi
tion
s(a
)n
ever
gave
rise
toli
vece
lls
atth
een
dof
expe
rim
ent
(g).
Bar
,18
0m
m.
492 ANDRADES ET AL.
prt
twcb
wlmcdr
cmdbhh
aipfebfndcmbdcsngm
493TYPE I COLLAGEN AND rhTGF-b1-F2 IN OSTEOGENESIS
aring rhBMPs, chambers filled with cells treated withhBMP-2 contained the largest amount of bone (80% of
FIG. 6. Effect of experimental conditions on colony numbers (a)nd sizes (b). Recombinant hTGF-b1-F2 was the only GF able tonduce cell colonies during the early S period. At the end of thiseriod, colonies appeared in the presence of rhBMPs, although five-old less in number than with rhTGF-b1-F2. No colonies were min-ralized at this time. During the A period, cells treated with rhTGF-1-F2 formed abundant colonies which began to mineralize. Coloniesormed in the presence of rhOP-1, even though they didn’t increase inumber, and began to mineralize. No colony mineralization wasetected for rhBMP-2-treated cells during this period. The number ofolonies increased markedly during the I period. The number ofineralized colonies was also much higher than of colonies formed
y rhTGF-b1-F2 treatment, which remained constant in size. (b) Noifferences were observed in colony size during the S period inolonies formed under different treatments. Once the normal serumituation was reestablished (A period), the colony size increasedotably. While no significant differences were seen between rhBMPsroups after the A–I period, rhTGF-b1-F2 showed the largest incre-ent in size. Values are given as means 6SD for four samples.
he total tissue, P , 0.001), double the amount seen in c
he presence of rhOP-1, but no significant differencesere found in percentages of cartilage formation. No
artilage or bone was measured in controls or in cham-ers containing cells treated with bFGF.Diffusion chambers (Fig. 9) containing cells treatedith rhTGF-b1-F2 formed the same quantity of carti-
age as iDBM chambers (45%, P , 0.001), but treat-ent with hTGF-b1 resulted in a lesser amount of
artilage (25%, P , 0.001), and no bone formation wasetected. Chambers containing cells treated withhOP-1 formed amounts of cartilage similar to those of
FIG. 7. (a) Levels of ALP, OC, and Ca corresponding to iDBMhambers implanted with cells cultured under the different experi-ental conditions indicated in abscissa. (b) Values corresponding to
iffusion chambers. No OC or Ca were detected in the presence ofFGF. No significant difference was seen between commercialTGF-b1 and rhTGF-b1-F2 in any type of chamber. RecombinantBMPs always induce the highest values of OC expression. E, empty
hambers. Each bar represents the average 6SD of 15 implants.
cafib
494 ANDRADES ET AL.
FIG. 8. Histological paraffin-embedded sagittal sections of iDBM (a–c) and diffusion chambers (d–g), explanted after 28 days, containingells treated under different experimental conditions (a, d, e, control; b, f, hTGF-b1; c, g, rhTGF-b1-F2). Section seen under conventional (d)nd polarized (e) light which shows birefringence due to the presence of collagen fibrils. Bar, 500 mm. dot, chamber walls (DBM, Milliporelters); asterisk, fibrous tissue; c, cartilage; small arrows, bone; arrows, perichondral tissue surrounded cartilage. No cartilage or bone can
e observed under control conditions.
hwctcc
asamktd
tioFtWwtreesci
I
acotmfitwmttat
cesr1Ftacabhaomppilu
fAmtpcdeldcaSvt3
hiRaE
495TYPE I COLLAGEN AND rhTGF-b1-F2 IN OSTEOGENESIS
TGF-b1, but bone was found only in chambers treatedith rhOP-1 (30%, P , 0.002). Chambers containing
ells treated with rhBMP-2 were filled with both car-ilage (60%, P , 0.001) and bone (40%, P , 0.001). Noartilage or bone formed from controls or bFGF-treatedells.
DISCUSSION
For bone growth and repair, enough progenitor cellsre needed. A rich source of such stem cells is thetroma of the BM. Amplification and differentiationre controlled by factors such as TGF-bs and BMPs,embers of the same molecular superfamily, each well
nown for osteoinductive capacity. The specific roles ofhese osteoinductive factors in stem cell growth andifferentiation remain to be defined.To test the effect of TGF-bs on putative osteoprogeni-
or cells, freshly isolated BM cells were incorporatednto a collagen matrix containing commercial hTGF-b1r rhTGF-b1-F2, in the presence of low serum (0.5%BS). These conditions resulted in the selection (i.e.,he survival) of a population of round blastoid cells.
hen normal serum culture conditions (10% FBS)ere reestablished, this TGF-b-responsive cell popula-
ion increased proliferating rates in the presence ofhTGF-b1-F2 and then differentiated into cells thatxpressed markers associated with the osteoblast lin-age when stimulated by appropriate modulators. Sub-equently, large amounts of cartilage, and in someircumstances bone, appeared when those cells were
FIG. 9. Diffusion chambers filled with cells treated withTGF-bs and rhBMPs produced cartilage, but bone was formed only
n chambers containing cells treated with rhOP-1 and rhBMP-2.ecombinant hBMP-2 formed the highest amount of bone. Controlnd bFGF-treated cells never form cartilage or bone in any chamber.ach bar represents the average 6SD of 15 implants.
mplanted in chambers subcutaneously in rats. B
n Vitro Cell Behavior
Migration is the first critical event in the life cycle ofBM osteoprogenitor cell. In our study we considered
ell migration the basic criterion for standardizing theptimal concentrations of GFs. However, the concen-ration of GF that gives maximal chemotactic activityay not, in fact, be the concentration that is optimal
or other activities of a GF. In these comparative stud-es, recombinant hTGF-b1-F2 was much more potenthan hTGF-b1 in stimulating chemotactic activityhen these were entrapped into the collagen gel, whichay be a consequence of hTGF-b1-F2 slow release from
he collagen network to which it binds, suggesting thathe collagen binding domain may impart greaternd/or prolonged biological activity to the fusion pro-ein.
In the cultures exposed to rhTGF-b1-F2, slow butontinuous cell proliferation through the S period wasvident, indicating both survival and proliferative re-ponse. Cells cultured with hTGF-b1 showed a lesseresponse. Cell proliferation was enhanced notably by0% serum (much more in the presence of rhTGF-b1-2), which resulted in the amplification (A period) ofhe selected cells(s) populations(s). Although TGF-blone has been described to induce fibroblasts to formolonies in soft agar, in our study rhTGF-b1-F2 wasble to generate colonies during the S and A periods,ehavior that was not observed with commercialTGF-b1. This effect of rhTGF-b1-F2 on cell growthppears to be restricted to the early proliferative phasef the culture (A period), before the cells expressed theature osteoblastic phenotype, when cultured in the
resence of osteogenic inducers such as dex and b-GP (Ieriod). These facts speak in favor of a better availabil-ty of the rhTGF-b1-F2 to the cells, probably due to aonger half-life time of the ECM-targeted growth factornder these culture conditions.Although at present there is not a definitive marker
or cells of the osteogenic lineage, high expression ofLP, bone cell-specific protein OC, and production of aineralized matrix are three widely accepted charac-
eristics of bone cells. In our experiments, during the Ieriod, such markers for cytodifferentiation as well ashondrogenesis (measured by Alcian blue staining,ata not shown) confirmed the chrondro/osteogenic lin-age of the selected, amplified, and induced cell popu-ation. A number of studies have shown that, althoughex can induce terminal differentiation of osteogenicells in cultures, the presence of this steroid is not anbsolute requirement for in vitro osteogenesis [35].ynthesis of a mineralized matrix in culture has pre-iously been shown to be dependent on supplementa-ion of the growth medium with organic phosphate [36,7]. Chick embryo periosteum-derived cells [38], rat
M cells on collagen matrices containing rhTGF-b1-F2
aiwAshik[bmtbFtir
htrnTraGrs
rvpactl
filsprBafpc
I
ceoivdpboirctiCcwish
uid
496 ANDRADES ET AL.
t concentrations as low as 1 ng/ml, and osteogenicnducers exhibited a greater capacity to proliferatehich decreased concomitantly with the increase ofLP activity, as cells reached more differentiatedtages (I period): 1-fold higher ALP activity (2.5-foldigher than that of controls) than that of cells selected
n the presence of hTGF-b1. At present, OC is the onlynown bone-specific protein produced by osteoblasts39]. It was reported to appear at late stages of osteo-last differentiation. At variance, ALP appears in im-ature osteoblasts [39]. In our experiments, low syn-
hesis of OC occurred in the I period in the presence ofoth hTGF-bs. As far as Ca is concerned, rhTGF-b1-2, but not hTGF-b1, was able to induce Ca precipita-ion during the I period, consistent with the mineral-zation of colonies formed in the presence of theecombinant growth factor.
In comparison recombinant hBMPs generated theighest initial values of ALP, OC, matrix mineraliza-ion, and chondrogenesis. However, the number of cellsecruited in the presence of rhTGF-b1-F2 as well as theumber and size of colonies were considerably greater.he prompt mineralization of colonies formed underhBMPs treatment could limit their growth and maylso explain the decrease of Alcian blue staining ofAGs found during the A period. Alternatively, the
hBMP-selected cells may represent a more committedtem cell.Taken together (Fig. 10), these facts suggest that
hTGF-b1-F2 applied to a bovine collagen matrix asehicle and delivery system could be of advantage inromoting the survival, proliferation, differentiation,nd colony mineralization of the osteogenic precursorell population. Differences between rhTGF-b1-F2 andhe commercial hTGF-b1 may be due to the slow re-
FIG. 10. Schematic representation of chondro/osteogenic markersed in vitro. In each case, maximum values correspond with the
ntermediate values of osteogenic markers compared to the commerata.
ease of the collagen binding factor from the collagen t
brillar network to which it is bound, resulting in aonger half-life. The rhBMPs appear to promote theurvival and differentiation of committed osteogenicrecursors into more mature osteoblasts. In fact, theesults reported here suggest that rhOP-1 and rh-MP-2 select as well as stimulate the differentiationnd maturation of committed osteoprogenitors arisingrom BM, while rhTGF-b1-F2 is selective for a moreroliferative, less differentiated mesenchymal stemell.
n Vivo Cell Response
TGF-b stimulates the chondrogenic response whenoimplanted with partially purified fractions of bonextracts that have no osteogenic activity alone [40]. Inur study, when the selected BM cells, amplified andnduced in the presence of hTGF-bs, are implanted inivo, they were able to form cartilage in both iDBM andiffusion chambers. Moreover, when the cells were ex-osed to rhTGF-b1-F2, not only cartilage but also someone was formed in the iDBM chambers. The formationf bone by rhTGF-b1-F2-treated cells when implantedn iDBM chambers may be due to the presence ofesidual traces of BMPs in the bone matrix [6], to ahemical attraction effect exerted by the DBM matrixoward circulating BMPs, or to the presence of invad-ng vasculature, or a combination of all these variables.omparing hTGF-bs with rhBMPs, we observed thatells cultured in the presence of rhOP-1 and rhBMP-2ere able to form large amounts of cartilage and bone,
n either iDBM or diffusion chambers. Values of OCynthesis and Ca content are in concordance with theistology. However, ALP activity was much less than
ression and colony number and size, under the influence of the GFskest bars. As can be observed, rhTGF-b1-F2-treated cells expresshTGF-b1 and rhBMPs, whereas they are the highest in the colony
expthiccial
hat obtained with hTGF-bs, probably due to the ear-
ls
atpatewptoBpt
optdhev
ICooBmSa
1
1
1
1
1
1
1
1
1
1
2
2
2
497TYPE I COLLAGEN AND rhTGF-b1-F2 IN OSTEOGENESIS
ier expression of this enzyme at the initiation of thekeletogenic pathway.In contrast to the sequential model of TGF-b/BMP
ction presented by Ballock et al. [21], the results ofhese studies suggest a more complex role of TGF-b inromoting bone formation. While the effects of TGF-bsnd BMPs on bone formation are clearly synergistic,he instructional role of TGF-b in the recruitment andxpansion of undifferentiated mesenchymal stem cellshich could be induced to differentiate into osteogenicrecursors was not appreciated. Our results suggesthat TGF-b1 plays a crucial role in early stages ofsteogenic commitment and differentiation, while rh-MPs promote the survival and maturation of a lessroliferative and more differentiated population of os-eoprogenitor cells.
In summary, these results support the concept thatsteogenic precursor cells can be selected from a mixedopulation of BM mesenchymal stem cells by virtue ofheir distinctive survival responses in the presence ofefined GFs and further, that these selected cells ex-ibit unique properties in the chondroosteogenic lin-age that can ultimately be utilized to therapeutic ad-antage.
The authors are grateful for the financial support of the Nationalnstitutes of Health (AGO2577), of the Ministerio de Educacion yultura (PB95-1134, SAF99-0133, Spain), which supported the visitf Dr. Jose A. Andrades to the University of Southern California, andf the Fundacion MAPFRE Medicina (Spain). We also thank Dr. Paulenya and Dr. Pedro Fernandez-Llebrez for their valuable com-ents on the manuscript and Mrs. Delia C. Ertl, Mrs. Robin T.impkins, and Mrs. Silvia Hernandez for their excellent technicalssistances.
REFERENCES
1. Owen, M. E., and Friedenstein, A. J. (1988). Stromal stem cells:Marrow-derived osteogenic precursors. In “Cell and MolecularBiology of Vertebrate Hard Tissues” (D. Evered and S. Harnett,Eds.), pp. 42–60, Wiley, New York.
2. Friedenstein, A. J., Kuralesova, A. I., and Frolova, G. F. (1969).Heterotopic transplantation of bone marrow. Analysis of pre-cursor cells for osteogenic and hematopoietic tissues. Trans-plantation 6, 230–247.
3. Leboy, P. S., Beresford, J. N., Devlin, C., and Owen, M. E.(1991). Dexamethasone induction of osteoblast mRNAs in ratmarrow stromal cell cultures. J. Cell Physiol. 146, 370–378.
4. Satomura, K., Hiraiwa, K., and Nagayama, M. (1991). Miner-alized nodule formation in rat bone marrow stromal cell culturewithout beta-glycerophosphate. Bone Miner. 14, 41–54.
5. Herbertson, A., and Aubin, J. E. (1995). Dexamethasone altersthe subpopulation make-up of rat bone marrow stromal cellcultures. J. Bone Miner. Res. 10, 285–294.
6. Becerra, J., Andrades, J. A., Ertl, D. C., Sorgente, N., andNimni, M. E. (1996). Demineralized bone matrix mediates dif-ferentiation of bone marrow stromal cells in vitro: effect of ageof cell donor. J. Bone Miner. Res. 11, 1703–1714.
7. Ghosh-Choudhury, N., Windle, J. J., Koop, B. A., Harris, M. A.,
Guerrero, D. L., Wozney, J. M., Mundy, G. R., and Harris, S. E.(1996). Immortalized murine osteoblasts derived from BMP 2-Tantigen expressing transgenic mice. Endocrinology 137, 331–339.
8. Gronthos, S., Graves, S. E., Ohta, S., and Simmons, P. J. (1994).The STRO-11 fraction of adult human bone marrow containsthe osteogenic precursors. Blood 84, 4146–4173.
9. Oyajobi, B. O., Lomri, A., Hott, M., and Marie, P. J. (1999).Isolation and characterization of human clonogenic osteoblastprogenitors immunoselected from fetal bone marrow stromausing STRO-1 monoclonal antibody. J. Bone Miner. Res. 14,351–361.
0. Houghton, A., Oyajobi, B. O., Foster, G. A., Russell, R. G., andStringer, B. M. (1998). Immortalization of human marrow stro-mal cells by retroviral transduction with a temperature sensi-tive oncogene: Identification of bipotential precursor cells capa-ble of directed differentiation to either an osteoblast oradipocyte phenotype. Bone 22, 7–16.
1. Hicok, K. C., Thomas, T., Gori, F., Rickard, D. J., Spelsberg,T. C., and Riggs, B. L. (1998). Development and characteriza-tion of conditionally immortalized osteoblast precursor celllines from human bone marrow stroma. J. Bone Miner. Res. 13,205–217.
2. Centrella, M., McCarthy, T. L., Canalis, E., and Connecticut, H.(1991). Transforming growth factor-beta and remodeling ofbone. J. Bone Joint Surg. 73-A, 1418–1428.
3. Roberts, A. B., Anzano, M. A., Meyers, C. A., Wideman, J.,Blacher, R., Pan, Y. C. E., Stein, S., Lehrman, R., Smith, J. M.,Lamb, L. C., and Sporn, M. B. (1983). Purification and proper-ties of type b transforming growth factor from bovine kidney.Biochemistry 22, 5692–5698.
4. Lian, J. B., and Gundberg, C. M. (1988). Osteocalcin: biochem-ical considerations and clinical applications. Clin. Orthop. 226,267–291.
5. Robey, P. G., Young, M., Flanders, K. C., Roche, N. S., Kon-daiah, P., Reddi, A. H., Termine, J. D., Sporn, M. B., andRoberts, A. B. (1987). Osteoblasts synthesize and respond totransforming growth factor-type b (TGF-beta) in vitro. J. CellBiol. 105, 457–463.
6. Noda, M., Yoon, K., Prince, C. W., Butler, W. T., and Rodan,G. A. (1988). Transcriptional regulation of osteopontin produc-tion in rat osteosarcoma cells by type beta transforming growthfactor. J. Biol. Chem. 263, 13916–13921.
7. Noda, M. (1989). Transcriptional regulation of osteocalcin pro-duction by transforming growth factor-beta in rat osteoblast-like cells. Endocrinology 124, 612–617.
8. Rosen, D. M., Stempien, S. A., Thompson, A. Y., and Seyedin,S. M. (1988). Transforming growth factor-beta modulates theexpression of osteoblast and chondroblast phenotypes in vitro.J. Cell Physiol. 134, 337–346.
9. Centrella, M., McCarthy, T. L., and Canalis, E. (1987). Trans-forming growth factor beta is a bifunctional regulator of repli-cation and collagen synthesis in osteoblast-enriched cell cul-tures from fetal rat bone. J. Biol. Chem. 262, 2869–2874.
0. Bonewald, L. F., and Mundy, R. (1990). Role of transforminggrowth factor-beta in bone remodeling. Clin. Orthop. Relat. Res.250, 261–276.
1. Ballock, R. T., Heydemann, A., Izumi, T., and Reddi, A. H.(1997). Regulation of the expression of the type-II collagen genein periosteum-derived cells by three members of the transform-ing growth factor-b superfamily. J. Orthop. Res. 15, 463–467.
2. Seyedin, S. M., Thomas, T. C., Thompson, A. Y., Rosen, D. M.,and Piez, K. A. (1985). Purification and characterization of twocartilage-inducing factors from bovine demineralized bone.
Proc. Natl. Acad. Sci. USA 82, 2267–2271.
