Bone marrow stroma in childhood myelodysplastic syndrome: composition, ability to sustain hematopoiesis in vitro, and altered gene expression

Leukemia Research 28 (2004) 831–844

Bone marrow stroma in childhood myelodysplastic syndrome:composition, ability to sustain hematopoiesis in vitro,

and altered gene expression

Radovan Borojevica, Rosimeire A. Roelab,1, Renato S. Rodartea,1, Leandro S. Thiagoa,Fátima S. Pasinib, Fabiana M. Contia, Maria Isabel D. Rossia,

Luiz F.L. Reisc, Luiz F. Lopesd, M. Mitzi Brentanib,∗a Hospital Universitário Clementino Fraga Filho and Departamento de Histologia e Embriologia, Instituto de Ciencias Biomédicas,

Universidade Federal do Rio de Janeiro, Brazilb Disciplina de Oncologia, Departamento de Radiologia, Faculdade de Medicina da Universidade de Sao Paulo (LIM-24),

Av. Dr Arnaldo, 455-sala 4112, CEP 01246-903 Sao Paulo, Brazilc Instituto Ludwig, São Paulo, Brazil

d Departamento de Pediatria do Hospital do Cancer A.C. Camargo, Brazil

Received 8 September 2003; accepted 24 November 2003

Abstract

We studied bone marrow stromal cell cultures from patients with childhood myelodysplastic syndromes (MDS, refractory anemia withexcess of blasts, RAEB) and from matched normal donors. Stromal cell monolayers were characterized as myofibroblasts by the expressionof smooth muscle�-actin, collagen IV, laminin and fibronectin. When normal cord blood cells were plated onto myelodysplastic stromas,a pathologic cell differentiation was observed, indicating altered myelosupportive properties. cDNA array analysis showed that patientstromas expressed increased levels of thrombospondin-1, collagen-I�2-chain, osteoblast-specific factor-2 and osteonectin, indicating thepresence of increased osteoblast content, as confirmed by enhanced alkaline phosphatase synthesis. Alterations in the myelodysplasticstroma environment might contribute to abnormal hematopoiesis in this pathology.© 2004 Elsevier Ltd. All rights reserved.

Keywords: Childhood myelodysplastic syndrome; Bone marrow stromal cells; cDNA array; Gene expression; Environment; Hematopoiesis

1. Introduction

Myelodysplastic syndromes (MDS) are disorders thatcomprise a distinct albeit heterogeneous group of patholo-gies, in which a hypercellular bone marrow frequentlycoexists with an impaired maturation of blood cells result-ing in peripheral deficiency of one or several cell lineages,or with pancytopenia. While MDS has been extensivelystudied in adults, its poor prognosis and frequently rapidevolution to leukemia in childhood are not fully understood.The clonal nature of MDS has been extensively proven, andthe most frequently found genomic changes are deletions,thought to be related to genomic instability or deficient

∗ Corresponding author. Tel.:+55-11-30667488;fax: +55-11-30826580.

E-mail address: [email protected] (M.M. Brentani).1 The authors have contributed equally to this study.

DNA repair [1,2]. Whereas undetectable or single chromo-some abnormalities occur frequently in the early stages ofMDS, multiple genomic pathologies are observed duringthe disease progression and its conversion to leukemia. Thisis a frequent outcome of MDS that characterizes it as apotential “preleukemia” condition[3].

The pathogenesis of bone marrow failure in MDS is com-plex, since both blood cell progenitors and the hematopoi-etic bone marrow environment can be involved in theestablishment and evolution of the disease. The preciserole of the stromal environment is not clear. MDS-derivedstromas have a decreased capacity to support both normaland MDS blood progenitors. They are also deficient in sus-taining hematopoiesis in vitro, but great discrepancies wereobserved among individual patients, probably reflectingthe heterogeneity of MDS[4–6]. In contrast, some authorssuggested that MDS adherent cells supportive functionnormal [7]. The reported functional abnormalities include

0145-2126/$ – see front matter © 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.leukres.2003.11.019

832 R. Borojevic et al. / Leukemia Research 28 (2004) 831–844

unregulated cytokine production and increased induction ofpro-apoptotic factors considered to be potentially derivedfrom stroma macrophages and fibroblasts ([8–10], and ref-erences therein). The disturbed cell distribution in the bonemarrow environment is clearly indicated by the observationof abnormally localized immature precursors (ALIPs), typ-ical of MDS, which are indicators of a poor prognosis forthe evolution of the disease[11].

The role of the bone marrow stroma in the establishmentand/or evolution of disorders of the myeloid compartment isrelevant, since transformed hematopoietic progenitors, liketheir normal counterparts, are largely dependent upon thestromal environment[12]. This role can be essentially sum-marized into three situations. (i) The stroma can be modi-fied by the presence of malignant infiltrating cells that caninduce an altered production of adhesion and extracellularmatrix molecules, as well as cytokines. (ii) The stroma canbe permanently modified by the presence of resident cellsderived from malignant hematopoietic clones, such as bonemarrow macrophages, mast cells, or lymphocytes. (iii) Theintrinsically modified stroma can generate conditions thatare favorable to the development of preleukemic and subse-quently leukemic states, being the primary cause of the de-velopment of a malignant myeloproliferative disease. Here,we have focused our attention on the third proposal. Wehave addressed the question of whether the connective tissuestromas derived from bone marrow tissue of patients withMDS have an intrinsically different pattern of biologicalcharacteristics, which can be maintained in vitro during cul-ture under standard conditions, and whether this pattern canbe correlated to the pathologic profile of the patient’s dis-ease. Since altered myelosupportive capacity of MDS stro-mas was observed in cultures, expression of a selected set ofgenes was analyzed by differential display and cDNA arrayanalysis comparing normal stromas to those obtained frompatients.

In order to avoid the heterogeneity of MDS, we havechosen to study a selected group of pediatric MDS patientsincluded in the FAB category as having refractory anemiawith excess of blasts (RAEB). This category of patientshave high numbers of ALIPs indicating a disturbed bonemarrow organization, a relatively poor prognosis particu-larly when associated with chromosomal aberrations, and ahigh incidence of transformation to acute myeloid leukemia(AML) [13–16]. We consider these characteristics to iden-tify this group of patients as a good model for the studyof preleukemic disorders associated with overt pathologicmodifications of the bone marrow environment.

In the present study, we obtained morphologically ho-mogenous populations of bone marrow stromal cells fromnormal donors and MDS patients. We monitored theirgrowth pattern in vitro, their capacity to sustain and to con-trol growth and differentiation of hematopoietic progenitorsof the same bone marrow sample or of normal nucleatedcells harvested from cord blood, and we analyzed their geneexpression profiles.

2. Materials and methods

2.1. Samples

Seven MDS patients (six males and one female rang-ing in age from 10 months to 10 years) were attended atthe Cancer Treatment and Research Center, A.C. CamargoCancer Hospital (São Paulo), and Hematology Departmentof the University Hospital Clementino Fraga Filho, FederalUniversity of Rio de Janeiro. The study was conducted ac-cording to the guidelines for human studies of the institu-tions involved. The diagnoses were established according tothe FAB classification criteria elaborated by the pathologyCommittee of the Brazilian Cooperative Group on PediatricPatients with Myelodysplastic Syndrome (BCG-PED-MDS)[17]. Age-matched controls included leftover material fromsamples used for the characterization of bone marrow aspi-rates of normal transplant donors. Umbilical cord cells wereobtained from the Obstetric Unit of the Federal Universityof Rio de Janeiro, RJ, Brazil. Not all assays were performedin all cases because of the limited availability of materialcollected from patients.

2.2. Cell lines

Primary human osteoblasts and murine S17 cells that sus-tain human and murine hematopoiesis were obtained fromthe Rio de Janeiro Cell Bank—PABCAM, Federal Univer-sity of Rio de Janeiro, RJ, Brazil, and were used after ob-taining the authorization of Dr. Dorshkind[18].

2.3. Primary bone marrow cell cultures

Bone marrow aspirates were obtained from the poste-rior iliac crest. Leukocytes were separated and plated ontoIscove’s medium (Sigma, St Louis, MO, USA) supple-mented with 20% fetal bovine serum (FBS—Laborclin,Pinhais, PR, Brazil), kanamycin 100�g/ml and tylosin10�g/ml (both from Sigma). One-half of the supernatantwith non-adherent cells was removed weekly and replacedwith fresh medium. Cytosmears were prepared using acyto-centrifuge, fixed in methanol and stained with standardMay-Grünwald and Giemsa solutions. When monolayerswere established (ca. 20–30 days for normal bone mar-rows), cells were trypsinized and plated under the sameconditions. Since macrophages are trypsin-resistant, aftertwo replatings a homogeneous cell population was obtained,designated hereafter as “stroma”.

Proliferation of stromal cells was monitored in 24-wellculture plates. Cells were plated (2× 104 cells per well),incubated as described above for the desired period of time,and quantified by Coomassie Brilliant Blue (CBB-R250)staining as previously described, with slight modifications[19]. Briefly, cells were washed and fixed with 4% formalinin phosphate-buffered saline (PBS), and stained for 1 h withCBB-R250. They were then washed with distilled water,

R. Borojevic et al. / Leukemia Research 28 (2004) 831–844 833

and the retained dye was eluted overnight with 1 ml 1%sodium dodecyl sulfate solution. The eluates were harvestedand 100�l aliquots were transferred to 96-well plates andanalyzed with an ELISA reader with a 570 nm filter.

For immunofluorescence characterization of cytoskele-ton and extracellular matrix, cells were plated onto mul-tiwell dishes provided with glass coverslips, fixed inparaformaldehyde, permeabilized with saponin, and re-acted with rhodamine-labeled polyclonal antibodies againstcollagen IV, laminin or smooth muscle�-actin (all fromSigma).

2.4. Co-cultures

Umbilical cord blood cells were processed as describedabove for bone marrow aspirates, and leukocytes were platedonto the stroma monolayers. Alternatively, non-adherentcells harvested from the patients’ bone marrow primarycultures were harvested after 1–2 weeks of culture. Thesecell populations were co-cultured with the selected stromas.In order to stimulate the long-term growth of blood pro-genitors, they were maintained thereafter in Iscove mediumsupplemented with 12.5% FBS, 12.5% horse serum and10−6 M hydrocortisone (Sigma). We plated the non-adherentcells onto S17 murine bone marrow stroma cells or onto areference “normal human bone marrow stroma” establishedfrom a pool of the leftover material used for bone mar-row transplantation. Non-adherent cells were subsequentlyharvested and studied as described above.

CD34+ cells were obtained from umbilical cord bloodusing the Dynal CD34 Progenitor Cell Selection System(Dynal Biotech, Lake Success, NY, USA) following themanufacturer’s protocol. Stroma feeder layers were preparedin 24-well culture plates as described above (2× 104 cellsper well). Non-adherent cells were harvested weekly, and theexpression of surface antigens was monitored by flow cy-tometry using a FACS Calibur apparatus (Becton-Dickinson,San Jose, CA, USA) and analyzed by the CellQuest soft-ware. We used the following directly conjugated antibod-ies: phycoerythrin (PE) mouse anti-human CD34 (HPCAII,Becton-Dickinson), FITC mouse anti-human CD13(SJ1D1),FITC mouse anti-human CD33 (D3HL60.251), both fromImmunotech, (Beckman Coulter, Marseille, FR), and FITCmouse anti-CD14 (TüK4, Caltag Lab., Burlingame, CA).

2.5. Alkaline phosphatase assays

For fluorescence microscopy identification of alkalinephosphatase, bone marrow stromal cells were plated onto24-well dishes provided with glass coverslips. The cellswere maintained under standard conditions or induced to os-teogenic differentiation with 10−7 M 1,25-dihydroxyvitaminD and dexamethasone. After 24, 48, and 72 h, the cellswere washed with PBS fixed with 4% paraformaldehydein sucrose-supplemented 0.1 M PBS, pH 7.4, at roomtemperature for 20 min, and reacted with the Vector Red

Alkaline Phosphatase Substrate Kit I (Vector Laborato-ries, Burlingame, CA, USA) following the manufacturer’sprotocol.

For quantitative analysis, bone marrow stromal cells wereplated onto 24 well dishes and maintained under standardculture conditions. When reaching confluence, they weretreated with Triton X-100 (Sigma) in Tris–HCl, pH 7.5. Afterhomogenization, the samples were cleared by centrifugationand reacted with the alkaline phosphatase kinetic reagentfollowing the manufacturer’s protocol (Biobrás, Sao Paulo,SP, Brazil). Total protein content was monitored simultane-ously by the BCA microassay (Pierce Chemical Company,Rockford, IL, USA). A primary culture of human osteoblastswas used as an internal control.

2.6. RNA extraction

Total RNA was extracted from stroma cells using the TRI-zol Reagent (Life Technologies, Grand Island, NY) follow-ing the procedure recommended by the manufacturer. RNAintegrity was assessed by fractionation through a denaturingagarose gel and by staining with ethidium bromide. Onlysamples with an 28S/18S ratio equal to 1 or higher werefurther processed.

2.7. Differential display RT-PCR (DDRT-PCR)

Differential-display experiments, isolation of bands of in-terest and re-amplification of recovered fragments were per-formed as previously described[20]. After re-amplification,fragments were cloned into pUC18 using the Sureclone lig-ation kit (Amersham Pharmacia Biotech) or into PGEMusing the PGEM-T Easy system (Promega). At least twoindependent clones from each band were sequenced (ABIPrisma 377, PE Applied Biosystems) and incorporated intothe cDNA Array.

2.8. cDNA array

Using a collection of 49 clones isolated from DDRT-PCRplus 4471 cDNA ORESTES fragments derived from theFAPESP/LICR Human Cancer Genome Project[21], weconstructed a nylon-based cDNA array having 4512 spots.As positive control for labeling and hybridization a cDNAcorresponding to a fragment of the lambda phage Q genewas also spotted. Membranes were printed with the aid ofthe Flexys Robot (Genomic Solutions, USA).

For RNA labeling, 30�g of total RNA extracted fromstroma cells were spiked with a defined concentration ofsynthetic, polyadenylated RNA corresponding to the lambdaphage Q gene. Labeling was done by RTase in the presence0.25 mM each dATP, dGTP and dTTP, 1.66�M dCTP and30�Ci of [�-33P]dCTP (3000 Cimmol; Amersham, UK)using 2.0�g of oligo(dT)15 and Superscript II reverse tran-scriptase (Life Technologies, Inc.). Probes were purified bygel chromatography (Sephadex G50) and then cooled on


ice. Arrays were pre-hybridized, hybridized, and washedas described elsewhere[22] and images were acquiredwith a phosphorimager (molecular Dynamics Storm Im-ager, Molecular Dynamics, USA). For each cDNA sample,three identical nylon membranes were hybridized simulta-neously.

2.9. Data acquisition, normalization, and analysis

Data acquisition was performed with the aid of ArrayVision software (Amershan, UK), using gel files. The back-ground from a given array was subtracted from all 4512 spotintensity values and only spots with positive background-corrected values across all arrays were considered foranalysis. Data were then normalized in all arrays using totalenergy. As reference sample, we used the average valuesfrom the triplicates corresponding to stromal cells from ahealthy child donor. Data from all triplicates and referencesample were then loaded to GeneSpring software (SiliconGenetics, USA). A two-fold ratio against the referencesample was used as cut-off to identify genes differentiallyexpressed to bone marrow stromal cells from childhoodMDS.

2.10. RT-PCR

The first-strand cDNA was generated using Superscript(Gibco BRL). PCR was performed with 1�g of total RNA,200�M of dATP, dGTP, dTTP and dCTP and 1 pM ofprimers and 1 U Taq polymerase (Gibco BRL). Primers forsemi-quantitative PCR were designed with the “primer3Output” software (http://www.genome.wi.mit.edu/egi-bin/primer/primer3) mentioned below and performed by cDNAamplification carried out with 21 cycles of initial meltingat 95◦C for 5 min and cyclic melting at 95◦C for 1 min,annealing at 62◦C for 1 min, and extension at 72◦C for1 min, and a final extension at 72◦C for 10 min. The primersfor the detection of alternative splicing of fibronectinmRNA were constructed as described by Kumazaki et al.[23].

As control for cDNA loading we amplified frag-ments from glyceraldehyde-3-phosphate dehydrogenase(GAPDH). Amplicons were fractionated on 2% agarosegel and visualized with ethidium bromide. Images wereacquired with the ImageMaster (Amersham, UK) and bandintensity was determined with the ImageMaster VDS Soft-ware.

Sense Antisense

GAPDH 5′ctgccaacgtgtcagtggtg3′ 5′cagtgtggtgggggactgag3′OSF-2 5′aacgcagcgctattctgacg3′ 5′cccatggatgattcgagcac3′SPARC 5′catcgggccttgcaaataca3′ 5′gtacccgtcaatggggtgct3′COL1A2 5′tgctcctggtgaagctggtc3′ 5′agacctctgggccccttttc3′THBS1 5′cctcaatgaacgggacaact3′ 5′gtcatcatcgtggtcacagg3′

Fig. 1. Bone marrow derived stroma cultures. Representative immunola-beling for laminin (A), collagen IV (B) and smooth muscle�-actin (C).Original magnification: (A and B) 400×; (C) 600×.

http://www.genome.wi.mit.edu/egi-bin/primer/primer3

http://www.genome.wi.mit.edu/egi-bin/primer/primer3


3. Results

3.1. Characteristics and growth pattern of bone marrowstromal cells

All the stromas were composed of cells displaying fibrob-lastoid morphology. Immunofluorescence analyses showedthat all cells expressed laminin at high levels (Fig. 1A).RT-PCR analyses showed that the stroma expressed lamininchains�1,�2,�4 and�1 (data not shown) in agreement withpreviously reported data[24]. A variable quantity of colla-gen IV was present in all cells, either barely detectable orconspicuously present in the perinuclear region, in secretoryvesicles and accumulated at focal adhesion sites (Fig. 1B).Smooth muscle�-actin was observed in all cells at variableintensity, being detected as part of an extensive meshwork

Fig. 2. RT-PCR analysis of alternative splicing of fibronectin mRNA.Agarose gel analysis of PCR products. (A) Analysis of the EDA site.EDA(+) represents the 420 bp band and EDA(−) represents the 150 bpband. (B) Analysis of the EDB site. EDB(+) represents the 462 bp bandand EDB(−) represents the 189 bp band. (C) Analysis of the IIICS site.The letters a–d represent 484 bp, 409/391 bp, 316 bp and 124 bp bands,respectively. (D) Analysis of GAPDH corresponds to the 395 bp band. CT,control stromal cells; myelodysplastic syndrome (MDS1–MDS4) stromalcells.

of stress fibers, and in the submembrane cytoskeleton ofruffling membranes (Fig. 1C). Expression of smooth muscle�-actin is down-regulated in culture, but it can be inducedby pro-inflammatory mediators, and we did not monitorthis issue in the present study. The stromal cells did notexpress CD13, confirming that the adherent macrophageswere not present among myofibroblasts. Three sites of al-ternative splicing of fibronectin mRNA (EDA, EDB, orIIICS) were evaluated by RT-PCR and no differences inthe profile of transcripts to EDA(+)/EDA(−), EDB(+) andIIICS variants were identified in stromal cells from patientsor normal donor (Fig. 2). Taken together, these data indicatethat stromal cells could be classified as myofibroblasts, inagreement with other studies of bone marrow stroma[25].Admixture of hematopoietic lineages potentially derivedfrom malignant clones was not observed, since no cellsgrew in the non-adherent fraction, and no monocyte-derivedCD13+ cells were detected in the adherent fraction.

In long-term cultures, the stromas of normal donors werecontact-inhibited, while the stromas from the patients hada low contact inhibition, overgrowing in several layers andretracting after reaching confluence. The overall rate of pro-liferation of the stromal cells from patients was significantlyhigher compared to the normal controls, indicating a mod-ified behavior of patient stromas that was retained in vitrothrough several replatings (Fig. 3).

3.2. Primary culture of bone marrow cells

Our conditions for primary culture of bone marrowcells sustained mainly proliferation and/or differentiationof myeloid lineages, and erythropoiesis or lymphopoiesiswas not monitored. While the stroma was progressivelyestablished during the first 2 weeks of culture, the overalloutput of cells essentially reflected the properties of the har-vested bone marrow cell population. The quantitative and

Fig. 3. Proliferation of bone marrow stromal cells in vitro from MDSpatients (–) and normal donors (�). Data represent the mean values(quadruplicate) of the optical density units (Coomassie Blue assay). Thedifferences between the normal and MDS stromas were significant fromthe fifth day on (P < 0.005, Studentt-test).


qualitative aspects of non-adherent cells of these culturesare shown inFigs. 4 and 5.

All the cell cultures of normal pediatric donors followed acomparable pattern of evolution. A typical normal culture isshown inFig. 4. During the first week, the predominant cellsin the non-adherent cell fraction were small blasts with highnucleo-cytoplasmic ratio. They were replaced by monocytesand myeloid progenitors during the second week, followedby the terminal differentiation of the neutrophil cell lineage

Fig. 4. Differential counts of non-adherent cells in primary cultures (left columns), and of umbilical cord blood cells plated onto the pre-established MDSstromas (right columns). Cord blood cell growth corresponds to the general pattern of the MDS cell growth from which the stroma was established. Notethe capacity to sustain long-term survival and/or proliferation of blasts, with the exception of MDS1 in which blasts are induced to differentiate alongthe mono-macrophagic lineage. MDS stromas could not induce a full differentiation of the granulocyte lineage in primary cultures, and differentiatedcells were not observed in cord blood cell cultures.

in the third week. Foamy macrophages were present duringthe second week and decreased thereafter (Fig. 5A).

In contrast, cultures of bone marrow cells harvestedfrom different MDS patients followed a specific growthpattern. In accordance with the finding of a delayed matu-ration of leukocytes in the patients’ bone marrow aspirates,all MDS cultures showed decreased relative numbers ofmyeloid progenitors and of differentiated neutrophil granu-locytes (Fig. 4). The initial amount of blasts was variable


Fig. 5. Non-adherent cell fraction in primary cultures of bone marrow from normal donors and MDS patients. (A) Normal bone marrow. Typical blasts(arrow), several normal terminally differentiated myeloid cells, and a foamy macrophage (original magnification: 400×). (B) MDS bone marrow primaryculture, 2 days: typical blasts (original magnification: 400×). (C) MDS bone marrow primary culture (10 days): several basophilic myelo-monocyticprogenitors, myeloid progenitors, a single differentiated hypogranular neutrophil granulocyte, and a macrophage (original magnification: 400×). (D)MDS bone marrow primary culture, 25 days: hypergranular myeloblasts (original magnification: 400×). (E) Primary MDS cell-culture non-adherentcells plated onto the normal stroma, 4 days: normal polymorphonuclear granulocytes. (F) Primary MDS cell-culture non-adherent cells plated onto S17cells, 6 days: normal granulocytes and megakaryocytes (×400). (G) Normal umbilical cord blood cells plated onto the pre-established MDS stroma, 14days: basophilic myelomonocytes with heterogeneous abnormal nuclear morphology (original magnification: 400×). (H) Normal umbilical cord bloodcells plated onto the pre-established MDS stroma, 23 days: hypergranular myeloid progenitors, mitosis and three apoptotic cells (original magnification:400×). (I) Normal umbilical cord blood cells plated onto the pre-established MDS stroma, 25 days: hypergranular myeloid progenitors and vacuolatedbasophilic myelomonocyte (original magnification: 400×).

in patient bone marrow cultures, and they all decreasedduring the first weeks in the absence of support by thepre-established stroma (Figs. 4 and 5B). Abnormal myeloidprogenitors were observed during the first week. They hadan eccentric small nucleus and granular cytoplasm, withouta developed Golgi system, intermediate between type IIblasts and promyelocytes, in agreement with Bennett et al.[26].

Two patients who exhibited abnormal monocytoid nucleiin the bone marrow aspirates, without absolute peripheralmonocytosis, showed a strong increase of monocytoid cellsduring the first weeks in culture (Fig. 4, MDS1 and MDS2).

A part of these cells differentiated into macrophages duringthe third week (Fig. 5C) while another fraction generatednumerous and large granules resembling the myeloblastsobserved in acute promyelocytic leukemia (Fig. 5C and D).Although most of these large granules were basophilic,some were acidophilic, and by their size resembled thoseof eosinophil granulocytes, but no crystalloid body typicalof eosinophil granules could be observed by electron mi-croscopy (data not shown). A high rate of apoptosis wasobserved in this cell fraction. The remaining patients alsoshowed a decrease in type I blasts during the first week,but no myelo-monocyte proliferation. Both displayed a


deficient terminal differentiation of the myeloid lineage(Fig. 4, MDS3 and MDS4).

3.3. Co-culture of MDS blood cells with normalstromas

When non-adherent cells harvested for primary culturesfrom MDS patients were plated onto the pre-established nor-mal bone marrow stroma they lost their pathologic char-acteristics. While the atypical granular myeloblasts becameapoptotic, the normal myeloblasts differentiated into neu-trophil granulocytes (Fig. 5E). The same patients (MDS1,MDS2) had marked thrombocytopenia and did not presentmature megakaryocytes in bone marrow aspirates or in pri-mary cultures. Preliminary studies showed that normal hu-man stroma without the appropriate addition of cytokinesdid not sustain megakaryocytopoiesis, but the murine bonemarrow stroma cell line S17 that sustains both human andmurine hematopoiesis did. When the same cells were platedonto the murine S17 cell line, megakaryocytes appeared inlarge numbers, and numerous myeloid cells differentiatedinto neutrophil granulocytes (Fig. 5F). Taken together, thesedata show that the patient’s stroma was inefficient in sup-porting the normal growth and differentiation of myeloidprogenitors, which, however, were able to follow this dif-ferentiation pathway in the presence of a normal myelosup-portive stroma.

3.4. Co-cultures of umbilical cord blood cells with thestromas

When normal cord blood mononuclear cells were platedonto the pre-established MDS stromas, the non-adherentfraction contained roughly an equal amount of mature neu-trophils and blasts during the first days. In the cultures ofcord blood cells over normal stromas, the progressive de-crease of blasts corresponded to the differentiation of themyeloid cell lineage and macrophages (Fig. 4). This is sim-ilar to the growth of normal adult hematopoietic cells overthe pre-established bone marrow stromas, the only differ-ence being the increased proportion of granulocytes in cordblood cells cultures (data not shown). All the stromas ofMDS patients were unable to sustain the cord blood myeloidprogenitors, but they maintained viable blasts throughoutthe study, i.e., they were able to sustain the growth of im-mature progenitors but unable to sustain an efficient differ-entiation of myeloid cell lineages (Fig. 5G–I). Accordingly,the number of neutrophil granulocytes decreased sharply atthe end of the first week, when myeloid progenitors fromcord blood exhausted their proliferation capacity and via-bility. Conversely, the normal stroma sustained neutrophilsfor more than 3 weeks at the expense of newly generatedmyeloid cells. The only exception was one of the patientswho had abnormal monocytoid nuclei in the bone marrow(Fig. 4, MDS1). The stroma of this patient elicited a sharpincrease of monocytoid cells during the third week of cord

blood cell culture, followed by the presence of abnormal hy-pergranular myeloblasts that entered apoptosis, correspond-ing exactly to the pattern of the patient’s primary bone mar-row cultures (Fig. 5H and I).

In order to confirm the capacity of MDS stroma to sus-tain immature progenitors but not their differentiation, wepurified CD34+ cells from cord blood and plated them ontothe pre-established stroma of MDS patients and controls.Purification with magnetic beads yielded 90% CD34+ cellswhose proliferation and phenotype were monitored after 7,14, and 21 days.Fig. 6A depicts a representative experimentwith MDS1. The total number of cells increased slightlyafter the first week, as expected for stem cells that have alow replication rate. During the second week, the number ofcells maintained in co-culture with normal stroma increasedsix-fold, indicating an accelerated expansion of intermedi-ate myeloid progenitors (committed progenitors), which cannormally respond to peripheral demand by rapid prolifera-tion. Conversely, in the culture of CD34+ cells over MDSstromas, there was a slow increase in cell number during thesecond week, reflecting an inefficient support for the pro-duction of intermediate myeloid progenitors. Despite thisfact, the MDS stroma, but not the stroma from a normaldonor, sustained the proliferation of stem cells that main-tained the expression of CD34, as shown by an absolute in-crease of their numbers after 7 days of culture. The propor-tion of CD34+ cells decreased in both cultures during thethird week, as expected for an in vitro culture system thatcannot sustain a long-term proliferation of CD34+ cells.

CD14 is a marker of monocyte differentiation. Theanalysis of the expression of CD34 and CD14 in cordblood-derived CD34+ cells showed that, as expected, thenormal bone marrow stroma did not sustain CD34+ cellsurvival for more than 7 days, and a population of immaturemyeloid cells with low CD14 expression was present fromthe second week on (Fig. 6B). The MDS1 stroma that hada high capacity to sustain monocyte differentiation both inprimary culture and in co-culture of the total mononuclearfraction of cord bloods cells (Fig. 4), induced large num-bers of CD14+ cells divided into two fractions with highand low CD14 expression, respectively. The MDS2 stroma,which had a lower capacity to sustain monocyte differenti-ation as compared to MDS1 but higher than normal stroma(Fig. 4), showed the corresponding pattern of CD14 induc-tion (see graphs in detail inFig. 6B). Confirming the resultsof the primary bone marrow cultures and the cord bloodcell cultures, the MDS2 stroma sustained survival and/orproliferation of CD34+ cells through 21 days of culture.

In view of the rapid induction of a pathologic pattern ofcord blood cells growth and of published reports on the pres-ence of viral particles in MDS stromas[27], we addressedthis question by electron microscopy. We could not identifyany viral particles in the studied stromas (data not shown).However, molecular studies on viral gene expression werenot done, and a possible role of viral infection in the studiedphenomenon cannot be excluded[28].


Table 1Comparative fold increase analysis between cDNA array and semi-quantitative PCR

OSF-2 SPARC COL1A2 THBS1

RT-PCR cDNA array RT-PCR cDNA array RT-PCR cDNA array RT-PCR cDNA array

Control 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0MDS1 1.5 6.4 2.0 2.8 1.9 6.6 2.5 4.5MDS2 – 2.6 – 1.9 – 2.8 – 2.7MDS3 0.6 4.3 1.1 1.4 1.7 2.3 2.1 1.6MDS4 1.9 2.5 1.6 2.3 0.8 7.2 2.2 11.3MDS5 1.1 – 1.4 – 1.9 – 2.0 –MDS6 0.7 – 1.1 – 1.9 – 2.6 –MDS7 1.3 – 1.6 – 2.5 – 2.8 –

PCR amplification of OSF-2, SPARC, THBS1 and COL1A2. The linear ranges of amplification for all substances were determined with 21 cycles.Amplified products were fractionated on agarose gel. GAPDH cDNA was amplified as an internal control. Primers were designed using Primer 3 software.

3.5. Altered gene expression

We compared the genes differentially expressed by bonemarrow stromal cells from childhood MDS with those froma healthy donor by a cDNA array containing 4500 cDNAfragments including 49 sequences obtained from differen-tial display, in which we compared stromal cells from threeMDS patients with those from a healthy donor. The cDNAarray indicated that 343 cDNAs were overexpressed two-foldor more in bone marrow stromal cell from childhood MDS.Among the fragments which were overexpressed, we identi-fied mainly genes encoding proteins involved in metabolism(17 genes) and cell motility or extracellular matrix (8 genes),which potentially contribute to the proliferation and differ-

Fig. 6. Flow cytometry analysis during a period of 3 weeks of CD34+ cells harvested from cord blood, plated onto bone marrow stroma from a normaldonor and stromas from two MDS patients. (A) Representative data of proliferation and monocytic differentiation of CD34+ cells over one MDS stromasample as compared with healthy children. The percentage of CD34 (II) or CD14 (III) expression was used to calculate the number of cells expressingeach surface marker (IV and V) related to the total number (I). (B) Expression of CD34 vs. CD14 was weekly observed. The intensity of CD14 expressionis shown graphically in the detail, disclosing the CD14-high population in CD34+ cells cultured over MDS stromas. CT: control stromal cells.

entiation of hematopoietic precursors and may be importantfor the pathophysiology of MDS.

To confirm the preferential expression of these extra-cellular matrix and related genes, we analyzed in bonemarrow stromal cells from six MDS patients and from achild donor by semi-quantitative PCR the following genes:osteoblast-specific factor-2 (OSF-2), secreted protein acidicand rich in cysteine (SPARC), collagen type I alpha 2 chain(COL1A2) and thrombospodin 1 (THBS1). Genes wereconsidered to show increased expression if presenting lev-els two-fold higher than those displayed by the child donorAfter GAPDH normalization, the ratio between MDS andcontrol values was calculated (Table 1). PCR analysis con-firmed the cDNA array results for THBS1 for all patients,


Fig. 6. (Continued ).

whereas COL1A2, OSF-2 and SPARC were, respectively,overexpressed in five, two and four of six patients.

3.6. Alkaline phosphatase activity

In view of the increased expression of osteoblast differ-entiation markers in stroma cultures, we monitored the pres-ence and activity of alkaline phosphatase. The results werenot homogeneous for the MDS patients, with one of them(MDS4) being similar to controls. After 7 days in cultureunder standard conditions, the total phosphatase activity wassignificantly higher in stromas of three out of four MDS pa-tients as compared to controls (P < 0.05, Mann–WhitneyU-test) (Fig. 7A). Primary culture of normal human os-

teoblasts yielded 10-fold higher values than those found inpatients. The number of cells that expressed a detectablelevel of phosphatase activity was monitored by immunoflu-orescence in the stromal cells for the first 3 days of cultureafter replating (Fig. 7B, Table 2). A low reactivity was ob-served in a fraction of normal stromal cells from the sec-ond day on, as a faint fluorescence in the cytoplasm. In thestromas of MDS1 and MDS2 patients, an intense accumula-tion of phosphatase was observed in the trans-Golgi region.The capacity of stroma to respond to osteoblast induction bydexamethasone and Vitamin D was comparably increased inMDS stromas and was already detectable after 48 h culture(MDS1, MDS2, MDS3). In stromas with a rapid responseto induction (MDS1, MDS2), the number of induced cells


Fig. 7. Alkaline phosphatase activity in bone marrow stroma from the MDS patients studied (1–4) and in two normal stromas from age-matched subjects(CT). (A) Results are expressed as arbitrary densitometric units and represent the mean values and standard deviation of two independent experimentsdone in duplicate. The differences between MDS patients 1, 2 and 3 and normal stromas were significant (P < 0.05, Mann–WhitneyU-test). (B)Alkaline phosphatase detection by fluorescence microscopy in normal bone marrow stroma (I), in stromas from the patients studied (MDS1–MDS4, II–V,respectively), and in a primary culture of normal human osteoblasts (VI). Original magnification: 200×.

Table 2Number of alkaline phosphatase-positive cells (%), detected by immunofluorescence

CT MDS1 MDS2 MDS3 MDS4

SC Ind SC Ind SC Ind SC Ind SC Ind

24 h 0 0 1 2 2 4 0 0 0 048 h 2 5 6 70 80 85.7 8 22 0 3.272 h 10 19.9 54.5 75.2 88 97 10.6 79.4 3.2 6

Data represent the mean values of three independent experiments in which 100–200 cells were counted. SC: cells maintained under standard cultureconditions; Ind: cells induced to osteoblast differentiation with hydrocortisone and Vitamin D; CT: control stromal cells; MDS: myelodysplastic patients.


tended to stabilize after 48 h, although immunofluorescencestill increased thereafter. The low alkaline phosphatase ac-tivity in patient MDS4 remains to be clarified.

4. Discussion

The present study has shown that bone marrow stromalcells from pediatric MDS patients have an intrinsic tendencyto maintain blood cell progenitors in the undifferentiatedstate, sustaining the proliferation and delaying the terminaldifferentiation of myeloid lineages. This characteristic wasmaintained during a long-term culture of stromal cells invitro and was associated with an increased expression ofosteoblast markers identified by the study of gene expressionand of alkaline phosphatase activity.

Taken together, these observations indicate that the stromamay be one of the causes of the myelodysplastic state of bonemarrow. The hematopoietic progenitors of the MDS patientsdisclosed a full capacity to follow the normal differentiationprogram when plated onto a normal stroma. Conversely,normal hematopoietic progenitors harvested from umbilicalcord blood expressed the myelodysplatic phenotype aftera short-term culture in contact with the patients’ stromas.The in vitro pattern of myelodysplasia reflected that of thepatients’ bone marrow.

The stroma maintained the pathologic hemosupportiveproperties in vitro after a long-term growth under standardconditions. This may indicate either that (i) the stroma wasalready genetically modified in vivo, (ii) that it establishedan autocrine loop which sustained the defective hemosup-portive function in vitro, or (iii) that it had an altered com-position in terms of different stroma cell subpopulations.

The first hypothesis was not thoroughly studied and, toour knowledge, extensive clonal modifications of the stromahave not been reported. Several alterations of the cytokineprofile in the bone marrow environment have been describedin adult and in one case of pediatric MDS, and may resultin the establishment of autocrine loops, in agreement withthe second hypothesis[9,29–31].

Our results are consistent with the third proposal. Thepresent study was not an exhaustive screening analysis andpotential genes are certainly missing but we were able toidentify by cDNA array four genes which were overex-pressed in MDS as compared to the levels found in thestromal cells of an normal child donor: OSF-2, SPARC,collagen type I and thrombospondin, the first three beingosteoblast markers. SPARC was previously identified inchildren bone marrow stromal cells[32]. The overexpres-sion of osteoblast makers suggest that the composition ofthe MDS bone marrow sample might be enriched with os-teoblasts and this enrichment was maintained during the invitro culture, as confirmed by the increased activity of alka-line phosphatase in stroma cultures. Alternatively, a set ofgenes typical of osteoblasts was permanently up-regulatedin the stroma sample, and this property was maintained in

culture. Osteoblasts are known to sustain preferentially theearly blood cell progenitors, maintaining their proliferationand delaying their differentiation[33]. Accordingly, condi-tional ablation of osteoblasts is associated with a decreasein bone marrow elements adjacent to the endosteal sur-face[34]. The increased presence of cells with osteoblasticmarkers in our culture system may thus correspond to theincreased proliferation of immature progenitors and to areduced support for their differentiation. This behavior cor-responds to the in vivo hematopoiesis occurring in MDS,whereby proliferation of progenitors and their inefficientdifferentiation result in the presence of a hypercellular bonemarrow and peripheral cytopenia. In normal bone marrow,immature myeloid progenitors are located in the suben-dosteal compartment, whereas in MDS they are locatedinside the hematopoietic environment, identified as ALIPS.We may infer that this abnormal location of immatureprogenitors may be due to the presence of osteoblast-likestroma generating the microenvironment that is mimetic ofthe endosteal space, and suggest that ALIPS are a conse-quence of an alteration in the spatial organization of stromalcells.

The in vitro study indicated that the myelodysplastic pat-tern of hematopoiesis was apparently due to the pathologicinteraction between the stroma and the hematopoietic cells,and was not an intrinsic property of an abnormal blood cellclone. Recent studies in vivo using engraftment of MDSand normal human bone marrow mononuclear cells inNOD/SCID mice demonstrated that only normal blood cellscould be retrieved from MDS-engrafted mice, suggestinga requirement for a specific support for the expansion ofMDS clones, absent in normal stroma of NOD/SCID mice[35]. The same argument can be applied to normal differ-entiation of MDS blood cell progenitors when plated onto anormal stroma. However, a typical MDS pattern of growthof normal cord blood cells plated onto the MDS stromacannot be explained by the generation and expansion of anabnormal clone during such a short period of co-culture.

Although a pathological alteration of the stroma can bea primary cause of MDS, the fact that MDS progresses toa typical clonal disease with a potential evolution to acuteleukemia deserves consideration. The long-term support forthe proliferation of undifferentiated progenitors generatesconditions for the accumulation of multiple deleterious mu-tations, while under normal conditions such progenitors areeliminated by terminal differentiation. This increases theprobability to generate abnormal clones prone to accumu-lating more mutations that further favor their proliferationor protect them from apoptosis. SPARC is a de-adhesivemolecule and can induce alteration of integrin expressionwhich promotes malignant cell growth and may also pro-mote survival of cells with aberrant DNA[36–38]. Therole of thrombospondin in modulating the proliferationof diverse cells including hematopoietic progenitors hasbeen reported[39]. Besides the permissive environment forlong-term growth of abnormal clones in the MDS stroma,


adhesive interactions between the hematopoietic cells andthe stroma may favor genomic instability.

The small group of patients studied here is probably notrepresentative of all the MDS. In particular, MDS in adult-hood is frequently associated with chronic environmentalaggression or cancer chemotherapy, in which a primary le-sion of hematopoietic progenitors is most probably the di-rect cause of MDS. In these cases, elimination of malignantclones and allogeneic transplantation can lead to remissionand full re-establishment of normal hematopoiesis, while itdoes not replace the patient’s bone marrow stroma. In child-hood MDS, intrinsic properties of the hematopoietic envi-ronment may favor the appearance of abnormal clones andtheir further evolution leading to the severe outcome of thedisease.

In conclusion, we propose that in patients with the profiledescribed in this study, the early establishment of an abnor-mal stroma consequent to still unknown pathogenic mech-anisms may generate conditions for a progressive drift ofblood cell progenitors to the MDS profile. The transforma-tion to true leukemia will depend upon the accumulation ofdeleterious mutations and the generation of aggressive ab-normal clones.

Acknowledgements

This work was supported by FAPESP (No. 99/12641-1)and CNPq. We thank all the colleagues from BCG-PED-MDSwho directly or indirectly helped us with patient data. Mrs.Maria Helena A. Nicola and Silvia P. Azevedo are acknowl-edged for relevant contributions during the initial stages ofthis study.

Contributions. R. Borojevic contributed to the concept anddesign, interpreted and analyzed the data, provided draft-ing of the article, provided critical revisions and importantintellectual content, gave final approval, supplied statisticalexpertise, collected and assembled the data R.A. Roela in-terpreted and analyzed the data, and provided drafting ofthe article, collected and assembled the data. R.S. Rodarteinterpreted and analyzed the data, collected and assembledthe data and L.S. Thiago and M.I.D. Rossi supplied sta-tistical expertise, and provided administrative support. F.S.Pasini and F.M. Conti interpreted and analyzed the data.L.F.L. Reis interpreted and analyzed the data, supplied sta-tistical expertise, provided administrative support, collectedand assembled the data. L.F. Lopes provided critical re-visions and important intellectual content, provided studymaterials/patients, collected and assembled the data. M.M.Brentani contributed to the concept and design, interpretedand analyzed the data, provided drafting of the article, pro-vided critical revisions and important intellectual content,gave final approval, provided study materials/patients, sup-plied statistical expertise, obtained a funding source, pro-vided administrative support, collected and assembled thedata.

References

[1] Mecucci C. Molecular features of primary MDS with cytogeneticchanges. Leuk Res 1998;22:293–302.

[2] Willman CL. Molecular genetic features of myelodysplasticsyndromes (MDS). Leukemia 1998;12(1):2–6.

[3] Koeffler HP. Myelodysplastic syndrome (preleukemia). SemHaematol 1986;23:284–99.

[4] Borbenyi Z, Cinkotai C, Harrison C, Testa NG. The growth ofmyelodysplastic bone marrow in long-term cultures. Br J Cancer1987;55:291–3.

[5] Coutinho LH, Geary CG, Chang J, Harrison C, Testa NG. Functionalstudies of bone marrow haemopoietic and stromal cells in themyelodysplastic syndrome (MDS). Br J Haematol 1990;75:16–25.

[6] Aizawa S, Nakano M, Iwase O, Yaguchi M, Hiramoto M, Hoshi H, etal. Bone marrow stroma from refractory anemia of myelodysplasticsyndrome is defective in its ability to support normal CD34-positivecell proliferation and differentiation in vitro. Leuk Res 1999;23:239–46.

[7] Alvi S, Shaher A, Shetty V, Henderson B, Dangerfield B, ZoratF, et al. Successful establishment of long-term bone marrowcultures in 103 patients with myelodysplastic Syndromes. Leuk Res2001;25:941–54.

[8] Wetzler M, Kurzrock R, Estrov Z, Estey E, Talpaz M. Cytokineexpression in adherent layers from patients with myelodysplasticsyndrome and acute myelogenous leukemia. Leuk Res 1995;19:23–34.

[9] Kouides PA, Bennett JM. Understanding the myelodysplasticsyndromes. Oncologist 1997;2:389–401.

[10] Flores-Figueroa E, Gutiérrez-Espındola G, Montesinos JJ, Arana-Trejo RM, Mayani H. In vitro characterization of hematopoieticmicroenvironment cells from patients with myelodysplasticsyndrome. Leuk Res 2002;26:677–86.

[11] Lambertenghi Deliliers GH, Annaloro C, Soligo D, Oriani A. Thediagnostic and prognostic value of bone marrow immunostaining inmyelodysplastic syndromes. Leuk Lymphoma 1998;28:231–9.

[12] Duhrsen U, Hossfeld DK. Stromal abnormalities in neoplastic bonemarrow diseases. Ann Hematol 1996;73:53–70.

[13] Rigolin GM, Cuneo A, Roberti MG, Castoldi G. Myelodysplasticsyndromes with monocytic component: hematologic and cytogeneticcharacterization. Haematologica 1997;82:25–30.

[14] Aricó M, Biondi A, Pui CH. Juvenile myelomonocytic leukemia.Blood 1997;90:479–88.

[15] Cambier N, Baruchel A, Schlageter MH, Menot ML, Wattel E,Fenaux P, et al. Chronic myelomonocytic leukemia: from biology totherapy. Hematol Cell Ther 1997;39:41–8.

[16] Sasaki H, Manabe A, Kojima S, Hayashi Y, Ikuta K, Okamura J, etal. Myelodysplastic syndrome on childhood: a retrospective study of189 patients in Japan. Leukemia 2001;15:1713–20.

[17] Lopes LF, Lorand-Metze I, Niero-Melo L, Tone LG, Velloso E,Campanaro CM, et al. The Brazilian Pediatric Myelodysplastic Co-operative Group strategies: are they relevant to improve educationalapproach and correct diagnosis? Leuk Res 2002;26(7):637–42.

[18] Collins LS, Dorshkind K. A stromal cell line from myeloid long-termbone marrow culture can support myelopoiesis and B lymphopoiesis.J Immunol 1987;138:1082–7.

[19] Margis R, Borojevic R. Quantification of attached cells in tissueculture plates and on microcarriers. Anal Biochem 1989;181:209–11.

[20] Liang P, Pardee AB. Differential display of eukaryotic messengerRNA by means of the polymerase chain reaction. Science 1992;257:967–71.

[21] de Souza SJ, Camargo AA, Briones MR, Costa FF, Nagai MA,Verjovski-Almeida S, et al. Identification of human chromosome 22transcribed sequences with ORF expressed sequence tags. Proc NatlAcad Sci USA 2000;97(23):12690–3.

[22] Meireles SI, Carvalho AF, Hirata R, Montagnini AL, Martins WK,Runza FB, et al. Differentially expressed genes in gastric tumorsidentified by cDNA array. Cancer Lett 2003;190(2):199–211.


[23] Kumazaki T, Mitsui Y, Hamada K, Sumida H, Nishiyama M.Detection of alternative splicing of fibronectin mRNA in a singlecell. J Cell Sci 1999;112:1449–53.

[24] Siler U, Seiffert M, Puch S, Richards A, Torok-Storb B,Muller CA, et al. Characterization and functional analysis oflaminin isoforms in human bone marrow. Blood 2000;96:4194–203.

[25] Dennis JF, Charbord P. Differentiation of human and murine stroma.Stem Cells 2002;20:205–14.

[26] Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DAG,Gralnick HR, et al. Proposals for the classification of themyelodysplastic syndromes. Br J Haematol 1982;5:189–99.

[27] Raza A. Hypothesis: myelodysplastic syndromes may have viraletiology. Int J Hematol 1998;68:245–56.

[28] Mundle S, Allampallam K, Rashid KA, Dangerfield B, CartlidgeJ, Zeiter D, et al. Presence of activation-related mRNA for EBVand CMV in the bone marrow of patients with myelodysplasticsyndromes. Cancer Lett 2001;164:197–205.

[29] Raza A, Gregory SA, Preisler HD. The myelodysplastic syndromesin 1996: complex stem cell disorders confounded by dual actions ofcytokines. Leuk Res 1996;20:881–90.

[30] Yoshida Y, Mufti GJ. Apoptosis and its significance in MDS:controversies revisited. Leuk Res 1999;23:777–85.

[31] Winter SS, Hanissian GA, Harville TO, Ware RE. Tumor necrosisfactor-� suppressed hematopoiesis in children with Myelodysplasia.Med Am Ped Oncol 1997;28:69–74.

[32] Dieudonne SC, Kerr JM, Xu T, Sommer B, DeRubeis AR, KuznetsovSA, et al. Differential display of human marrow stromal cells revealsunique mRNA expression patterns in response to dexamethasone. JCell Biochem 1999;76(2):231–43.

[33] Taichman RS, Emerson SG. The role of osteoblasts in thehematopoietic microenvironment. Stem Cells 1998;16(1):7–15.

[34] Visnjic D, Kalajzic T, Gronowicz G, Aguila HL, Clark SH,Lichtler AC, et al. Conditional ablation of the osteoblast lineage inCol2.3deltatk transgenic mice. J Bone Miner Res 2001;16:2222–31.

[35] Benito AI, Bryant E, Loken MR, Sale GE, Nash RA, Gass MJ, etal. NOD/SCID mice transplanted with marrow from patients withmyelodysplastic syndrome (MDS) show long-term propagation ofnormal but not clonal human precursors. Leuk Res 2003;27:425–36.

[36] Yan Q, Sage EH. SPARC, a matricellular glycoprotein with importantbiological functions. J Histochem Cytochem 1999;47(12):1495–506.

[37] Sturm RA, Satyamoorthy K, Meier F, Gardiner BB, Smith DJ, VaidyaB, et al. Osteonectin/SPARC induction by ectopic�3 integrin inhuman radial growth phase primary melanoma cells. Cancer Res2002;62:226–32.

[38] Lewis JM, Truong TN, Schwartz MA. Integrins regulate the apoptoticresponse to DNA damage through the modulation of p53. Proc NatlAcad Sci USA 2002;99:3627–32.

[39] Gupta P, Oegema TR, Brazil JJ, Dudek AZ, Slungaard A, VerfaillieCM. Structurally specific heparan sulfates support primitive humanhematopoiesis by formation of multimolecular stem cell niches.Blood 1998;92(12):4641–51.

Bone marrow stroma in childhood myelodysplastic syndrome: composition, ability to sustain hematopoiesis in vitro, and altered gene expression

Documents