You are on page 1of 9

Experimental Hematology 28 (2000) 707–715

Proliferation kinetics and differentiation

potential of ex vivo expanded human bone marrow
stromal cells: Implications for their use in cell therapy
Andrea Banfia,b,c, Anita Muragliaa, Beatrice Dozina,b,
Maddalena Mastrogiacomoa,c, Ranieri Canceddaa,b,c, and Rodolfo Quartoa,b
Centro di Biotecnologie Avanzate, Genova, Italy;
Istituto Nazionale per la Ricerca sul Cancro, Genova, Italy; cDipartimento di Oncologia, Biologiae Genetica, Universita’ di Genova, Italy

(Received 4 January 2000; revised 12 February 2000; accepted 22 February 2000)

Objective. Bone marrow stromal cells (BMSC) are an attractive target for novel strategies in
the gene/cell therapy of hematologic and skeletal pathologies, involving BMSC in vitro expan-
sion/transfection and reinfusion. We investigated the effects of in vitro expansion on BMSC
pluripotentiality, proliferative ability, and bone-forming efficiency in vivo.
Materials and Methods. BMSC from three marrow donors were cultured to determine their
growth kinetics. At each passage, their differentiation potential was verified by culture in in-
ductive media and staining with alizarin red, alcian blue, or Sudan black, and by immu-
nostaining for osteocalcin or collagen II. First passage cells were compared to fresh marrow
for their bone-forming efficiency in vivo. Stromal cell clones were isolated from five donors
and characterized for their multidifferentiation ability. The lifespan and differentiation kinet-
ics of five of these clones were determined.
Results. After the first passage, BMSC had a markedly diminish proliferation rate and gradu-
ally lost their multiple differentiation potential. Their bone-forming efficiency in vivo was re-
duced by about 36 times at first confluence as compared to fresh bone marrow. Experiments
on the clones yielded comparable results.
Conclusions. Culture expansion causes BMSC to gradually lose their early progenitor proper-
ties. Both the duration and the conditions of culture could be crucial to successful clinical use
of these cells and must be considered when designing novel therapeutic strategies involving
stromal mesenchymal progenitor manipulation and reinfusion. © 2000 International Society
for Experimental Hematology. Published by Elsevier Science Inc.
Keywords: Chondrogenesis—Osteogenesis—Adipogenesis—Bone marrow stromal cells—Cell

Introduction also are capable of transferring the hemopoietic microenvi-

The bone marrow stromal system is defined as the connec- ronment [3–5] both in vitro [6] and in vivo [7,8].
tive tissue elements providing structural and functional sup- A mesenchymal stem cell (MSC) is defined as a pluripo-
port for hemopoiesis [1,2] and includes various cell types: tent cell capable of replicating extensively and self-main-
bone marrow fibroblasts–reticular cells, adipocytes, osteo- taining and whose progeny eventually gives rise to skeletal
blasts, macrophages, and endothelial cells. It has long been tissues: cartilage, bone, tendon, ligament, and marrow
recognized that the fibroblastoid component of the stromal stroma [5]. By definition, such cells exist during embryo-
system contains a rich reserve of osteogenic progenitors that genesis, and it is the object of current research to determine
whether these cells persist during adult life.
Bone marrow stromal cells (BMSC) can be cultivated in
vitro and contain progenitors capable of pluridifferentiating
Offprint requests to: Rodolfo Quarto, M.D., Laboratorio di Differenzia-
mento Cellulare, Centro di Biotecnologie Avanzate/Istituto Nazionale per
into the mesenchymal lineages of bone, cartilage, fat, mus-
la Ricerca sul Cancro, L.go R. Benzi n. 10, 16132 Genova, Italy; E-mail: cle, and other connective tissues [3,5,9–11]; therefore, they are an interesting target for use in cell and gene therapy.

0301-472X/00 $–see front matter. Copyright © 2000 International Society for Experimental Hematology. Published by Elsevier Science Inc.
PII S0301-472X(00)0 0 1 6 0 - 0
708 A. Banfi et al./Experimental Hematology 28 (2000) 707–715

BMSC attractiveness is based on the ease with which they tial, and in vivo bone-forming efficiency. The question
can be isolated from the patient’s marrow, expanded many- whether the graduality of this process could make a limited
fold in vitro, manipulated, and reinfused to the same patient. expansion compatible with successful clinical use awaits in
In this setting, BMSC can have many hematologically rele- vivo studies. Nevertheless, the effect of in vitro expansion
vant applications. They are being investigated as vehicles on the stem cell features of BMSC cultures need to be con-
for gene therapy of hemophilia B [12]. Several authors have sidered when designing strategies for use of BMSC in gene
shown that the bone marrow microenvironment is heavily and cell therapy.
and permanently damaged by bone marrow transplantation
(BMT) conditioning regimens [13–21]. A preliminary re-
port has claimed that autologous ex vivo expanded BMSC Materials and methods
accelerated hemopoietic recovery after high-dose chemo-
Cell culture and BMSC growth kinetics determination
therapy with peripheral blood progenitor cells (PBPC) res-
Stromal cells were obtained from iliac crest marrow aspirates from
cue in breast cancer patients [22]. BMSC can be used in
healthy donors for BMT procedures carried out in S. Martino Hos-
BMT as a therapy for skeletal pathologies [23,24] and has pital, Genova, Italy, and G. Gaslini Pediatric Hospital, Genova, It-
been applied to cases of osteogenesis imperfecta [25,26]. aly. Donor ages ranged between 5 months and 30 years; all donors
The use of BMSC in such therapeutic strategies relies on were white Caucasians. Informed consent was obtained from all
the ability of BMSC to proliferate readily and give rise to a donors, and all procedures were approved by the institutional ethi-
differentiated progeny that can substitute for the diseased cal committee. BMSC cultures were performed essentially as pre-
counterpart. It is of interest to determine whether, after the viously described [13], except that they were always carried out
ex vivo expansion necessary before their therapeutic manip- with medium supplemented with 1 ng/ML human recombinant fi-
ulation and reinfusion, these cells possess stem features. In broblast growth factor 2 (FGF-2) (Austral Biologicals, San Ra-
this study, we investigated the stem cell characteristics of ex mon, CA). After 2 weeks, CFU-F number was determined as de-
scribed [13]. Confluent cultures were detached with 0.05%
vivo expanded human BMSC by asking the following ques-
trypsin-0.01% EDTA (Sigma, St. Louis, MO). Cells were replated
tions: (1) Do BMSC maintain a constant doubling time dur-
for the in vitro differentiation experiments and for determination
ing their in vitro culture? (2) Do BMSC maintain their own of BMSC growth kinetics. At the first passage, BMSC were re-
multilineage differentiation potential after a number of mi- plated at a concentration of 5 ⫻ 105 cells in one 10-cm Petri dish.
totic divisions? (3) Do BMSC maintain a constant effi- When subconfluent, this plate was successively split 1:2 unit eight
ciency in their bone-forming ability in vivo? (4) Do clones dishes were obtained. At this point, cells again were detached,
derived from individual colony-forming unit-fibroblasts pooled, counted, and analyzed for differentiation potential. The
(CFU-F) display a multilineage differentiation potential and same procedure was repeated for five passages, and the cumulative
do they maintain it throughout their lifespan? cell doublings of the populations were plotted against time in cul-
To this purpose, we investigated the in vitro growth ki- ture to determine the growth kinetics of BMSC expansion.
netics of BMSC from three different donors in the primary Cell cloning
culture and up to the fourth passage. We analyzed their dif- Marrow samples were used to obtain clones by limiting dilution.
ferentiation potential at each passage to monitor the mainte- Briefly, 2 ⫻ 105 nucleated marrow cells were plated in each of four
nance of pluripotentiality during mitotic expansion. To fur- to five 96-well plates for each marrow sample. Cultures were per-
ther test this point, we compared the in vivo bone-forming formed in standard medium in the presence of FGF-2 and exam-
efficiency of first passage BMSC to that of fresh bone mar- ined daily for the appearance of stromal colonies. Wells containing
row. In both experiments we found that, as a population, ex- more than one colony were not considered. All clones that reached
panded BMSC progressively lost their differentiation poten- confluence after passaging into one well of a 24-well plate were
trypsinized and each of them replated in four wells (replicas). Of
the four replicas of each clone, three were stimulated for the differ-
The nearest approximation to the putative MSC that can
entiation experiments when cells reached confluence and one was
be studied is the CFU-F. We isolated, expanded, and char- frozen to be kept as stock.
acterized 80 nonimmortalized clones originating from sin-
gle CFU-F of human bone marrow. Their osteo-chondro- In vitro differentiation of BMSC cultures and clones
adipogenic potential was assessed by in vitro assays. All BMSC potential to differentiate into the chondrogenic, osteogenic
BMSC clones analyzed were able to differentiate in vitro and adipogenic lineages was verified at the level of primary cul-
into the osteogenic lineage and about one-third displayed tures and at each passage for four passages. Clones were analyzed
at the stage of replicas and at the following passages during their
osteo-chondro-adipogenic potential. The lifespan was deter-
lifespan. In all cases, cells were enzymatically detached from cul-
mined and the differentiation potential was assessed at in- ture dishes on reaching 90% confluence and plated at 5 ⫻ 104
creasing cell doublings in a number of clones. The results cells/cm2 in 24-well culture plates. Some dishes were not replated
paralleled those of BMSC cultures. in the wells, but instead were stimulated under the same conditions
This study provides evidence that, with extended culture and then used for RNA extraction and reverse transcriptase-poly-
in the conditions currently in use, human BMSC display a merase chain reaction (RT ⫽ PCR) analysis. Each well or dish was
tendency to lose their multipotentiality, proliferation poten- kept in complete medium until confluence was reached. Cultures
A. Banfi et al./Experimental Hematology 28 (2000) 707–715 709

were stimulated with the appropriate differentiating medium ac- row nucleated cells, of which the majority are hemopoietic progen-
cording to the following conditions: itors, that produce a bone formation equivalent to the standard
2.5 ⫻ 105 expanded BMSC.
• Chondrogenic differentiation: cultures were stimulated for 1
The fresh marrow content of CFU-F was determined as de-
week in standard medium (F-12 with 10% fetal calf serum
scribed earlier. For BMSC, and additional colony-forming effi-
[FCS] and without FGF-2) supplemented with 50 ␮g/mL
ciency experiment was performed on reaching the first confluence,
ascorbic acid and 1 ng/mL human recombinant transforming
in parallel with the in vivo implants, by plating 103 BMSC per 10-
growth factor ␤1 (TGF-␤1; Austral Biologicals, San Ramon,
cm Petri dish. After 2 weeks of culture, the plates were fixed and
CA). Type II collagen expression was revealed by immu-
stained, and the colonies were counted as for the fresh marrow.
nostaining with the monoclonal antibody CIICI (Develop-
Animals were sacrificed 8 weeks after implantation, and the grafts
mental Studies Hybridoma Bank) and the deposition of pro-
removed and processed for histologic analysis as previously de-
teoglycans was revealed by alcian blue staining.
scribed [27].
• Osteogenic differentiation: cultures were stimulated for 2
weeks in standard medium supplemented with 50 ␮g/mL
Clone lifespan
ascorbic acid, 1.5 mg/mL ␤-glycerophosphate, and 10⫺8M
Five clones were selected after characterization at the stage of rep-
dexamethasone. Osteocalcin expression was revealed by im-
licas for determination of their lifespan and maintenance of differ-
munostaining with an antiserum against bovine osteocalcin
entiation ability: three were tripontential (OCA [osteo-chondro-
kindly provided by Dr. Simon Robbins, and the presence of
adipogenic]) and two were bipotential (OC [osteo-chondrogenic])
calcium deposits was revealed by von Kossa and alizarin S
at this stage. The frozen replica was thawed and plated in a 6-cm
Petri dish. When it reached confluence, cells were trypsinized,
• Adipogenic differentiation: cultures were stimulated for 3
counted for determination of cell doublings, and split into four
weeks in F12 medium supplemented with 1% FCS, 10⫺7M
equal parts: three were plated each in a well of a 24-well plate and
dexamethasone, and 6 ng/mL insulin. Lipid droplets were re-
stimulated for differentiation to the three lineages as described ear-
vealed by staining with Sudan black IV.
lier, and the fourth was replated in a 6-cm Petri dish to continue
As controls, the following culture conditions were used: (a) expansion. For these experiments, osteogenesis and chondrogene-
proliferating unstimulated cells expanded with FGF-2 up to con- sis were assayed as osteocalcin and collagen II expression as deter-
fluence (basal phenotype); and (b) confluent cultures kept in stan- mined by immunohistochemistry. The results could not be quanti-
dard medium for the same period of time as the corresponding fied and were only scored as positive or negative as compared to
differentiation protocols, but without the stimulating factors (sponta- the negative control.
neous over-time phenotype).
The cultures were scored as positive, borderline, or negative. Semiquantitative RT-PCR measurements
For adipogenesis, scores were defined as follows: negative ⫽ no Total RNA was extracted from BMSC and control tissues (human
more than a single adipocyte in from a total of three random mi- articular cartilage, bone, and adipose tissue) with the guanidinium
croscopic fields; borderline ⫽ one or two adipocytes in each of at isothiocyanate procedure by Chomczynski and Sacchi [28]. Semi-
least two of three random microscopic fields; positive ⫽ at least quantitative RT-PCR was performed using the GeneAmp RNA
three adipocytes in each of at least two of three random micro- PCR kit from Perkin Elmer. For each RNA sample, a master RT
scopic fields. For osteogenesis and chondrogenesis, we quantified reaction was performed with 2 ␮g total RNA in a 40 ␮L mixture
alizarin S and alcian blue staining, respectively. Images of the containing 1 ⫻ PCR buffer from the provider, 5 mM MgCl2, 1 mM
stained wells were acquired with a Cohu black-and-white camera each of dCTP, dATP, dGTP, and dTTP, 1 U/␮L of RNAse inhibi-
and analyzed with the NIH-Image 1.60 public domain software. tor, 250 pmol of random examer primers, and 2.5 U/␮L of MuLV
We scored the average of duplicate wells for each condition. Back- RT. Reaction time was 1 hour at 42⬚C. Each master cDNA product
ground was calculated from empty wells and subtracted from all was divided into two equal parts that were used for PCR amplifica-
experimental readings. Differentiation units were calculated as the tion either of the housekeeping gene glyceraldehyde phosphate de-
fold increase in staining induced by the differentiation schedule as hydrogenase (GAPDH) or of one of the following specific genes:
compared with the negative control. Cultures were scored as fol- ␣1(II) collagen, osteocalcin, and lipoprotein lipase (LPL). The
lows: negative ⫽ less than 1.2 differentiation units; borderline ⫽ PCR reactions were performed in a 100-␮L mixture that contained
between 1.2 and 2 differentiation units; positive ⫽ more than 2 dif- 1 ⫻ PCR buffer, 2 mM MgCl2, 1 mM each of dCTP, dATP, dGTP,
ferentiation units. and dTTP, 30 pmol of 5⬘ and 3⬘ specific primers, and 2.5 U of Am-
pliTaq DNA polymerase. Primer sequences for all the aforemen-
In vivo bone-forming efficiency tioned genes were as follows: GAPDH: 5⬘-ACCACAGTCCAT-
In vivo ectopic bone formation was assayed according to a proto- GCCATCAC-3⬘ and 5⬘-TCCACCACCCTGTTGCTGTA-3⬘ [29];
col previously optimized in our laboratory [27]. Briefly, on reach- ␣1(II) collagen: 5⬘-CTGCTCGTCGCCGCTGTCCTT-3⬘ and 5⬘-
ing their first confluence, 2.5 ⫻ 105 BMSC were loaded on a AAGGGTCCCAGGTTCTCCATC-3⬘ [11]; OC: 5⬘-CATGAG-
highly porous hydroxyapatite support and implanted subcutane- AGCCCTCACA-3⬘ and 5⬘-AGAGCGACACCCTAGAC-3⬘ [30];
ously on the back of nude CD-1 nu/nu mice purchased form LPL: 5⬘-GAGATTTCTCTGTATGGCACC-3⬘ and 5⬘-CTGCAA-
Charles River Italia (Calco, Italy). ATGAGACACTTTCTC-3⬘ [30]. The expected product sizes were
When fresh bone marrow was used, nucleated cells were iso- GAPDH 450 bp, ␣1(isoform IIB) collagen 225 bp, OC 310 bp, and
lated by Ficoll gradient centrifugation and counted, and each cube LPL 276 bp. The number of cycles used was in the linear range of
was loaded with the appropriate number of cells. Preliminary ex- amplification for the specific gene product, as preliminary experi-
periments were carried out to determine the number of fresh mar- ments were carried out by the authors originally describing the
710 A. Banfi et al./Experimental Hematology 28 (2000) 707–715

tirety of the products. The intensity of individual bands was quan-

tified by image analysis using the public domain software NIH
1.60. The amount of PCR product for each single gene was nor-
malized according to the corresponding GADPH PCR product.


Growth kinetics and maintenance

of differentiation potential of primary BMSC cultures
The growth kinetics of three different BMSC cultures was
investigated from the primary culture through the fifth pas-
sage, corresponding to about 24 doublings. Expanded
BMSC were tested for their ability to undergo osteogenic,
chondrogenic, and adipogenic differentiation after appropri-
ate induction until the fourth passage.
Primary cultures reached the first confluence in about 3
weeks and 12–15 doubling (Fig. 1). On replating, BMSC
slowed their proliferation rate considerably and by the sec-
Figure 1. BMSC culture growth kinetics and differentiation potential ond passage stabilized their growth rate at about 10–14 days
maintenance. Growth kinetics of three BMSC cultures was determined. per population doubling (Fig. 1 and Table 1). The average
Primary cultures reached the first confluence in about 3 weeks and 12–15 doubling time increased from 1.3 days at the primary cul-
doublings. On replating, BMSC slowed their proliferation rate consider- ture stage to 7.7 at the first passage and 14.7 at the third pas-
ably. The curves represent the cumulative doubling number of the three
sage (Table 1). Concurrent with this slowing of growth rate,
individual cultures versus time in culture.
BMSC showed a change in appearance, from the initial
spindle shape to a more flattened square morphology (data
not shown).
primer sets. The reaction profiles used were as follows, with all Differentiation in the three lineages was induced as de-
preceded by a first denaturation step at 95⬚C for 4 min: GAPDH: scribed in the Materials and methods section. Figure 2
95⬚C for 1 minute, 55⬚C for 1 minute, and 72⬚C for 1 minute, for shows the results of a representative experiment with
24 cycles; OC: 95⬚C for 1 minute, 55⬚C for 1 minute, and 72⬚C for BMSC at their first confluence. As revealed by immu-
1 minute, for 30 cycles; LPL: 95⬚C for 1 minute, 60⬚C for 1 nostaining with specific antibodies, OC expression was evi-
minute, and 72⬚C for 1 minute, for 35 cycles; ␣1(II) collagen:
dent after the second week in osteogenic culture medium
95⬚C for 3 minutes, 58⬚C for 2 minutes, and 72⬚C for 1.5 minutes,
for one cycle and then 95⬚C for 1 minute, 58⬚C for 1 minute, and
(Fig. 2A), type II collagen was observed earlier, after 1
72⬚C for 1.5 minute, for 30 cycles. All PCR reactions ended with a week in chondrogenic medium (Fig. 2C), and adipogenic
7-minute incubation at 72⬚C. RT-PCR products were analyzed by differentiation was always the last to be observed. Although
electrophoresis of 20-␮L aliquots in 1% agarose gels and visual- sporadic adipocytes already were evident after several days
ized by ethidium bromide, except for the ␣1(II) fragment obtained in adipogenic medium, only after 3 weeks of stimulation did
by the BMSC RNAs that were ethanol precipitated to load the en- cultures display several large adipocytic colonies, as re-

Table 1. Growth kinetics and differentiation of one BMSC primary culture


Passage Total doublings Average doubling time (days) Osteogenesis Chondrogenesis Adipogenesis

0 14.6 1.3 ⫹ ⫹ ⫹
1 17.2 7.7 ⫹ ⫹ ⫹
2 18.7 10.0 ⫹ ⫹ ⫾
3 20.2 14.7 ⫹ ⫹ ⫾
4 22.2 9.5 ⫾ ⫾ ⫺
5 24.1 15.8 ND ND ND

Population doublings and differentiation potential of one BMSC culture were determined at the first confluence (passage 0) and at each passage thereafter.
Average doubling time at each passage was determined by dividing the number of doublings undergone by the culture since the previous passage by the time
(in days) taken to reach confluence. Differentiation was assayed by staining with alizarin S (osteogenesis), alcian blue (chondrogenesis), and Sudan black (ad-
ipogenesis). Scoring was as follows: ⫹ ⫽ positive; ⫾ ⫽ borderline; ⫺ ⫽ negative, as defined in the Materials and methods section. ND ⫽ not done.
A. Banfi et al./Experimental Hematology 28 (2000) 707–715 711

Figure 3. RT-PCR analysis for expression of osteoblast, chondrocyte, and

adipocyte-related genes in BMSC, bone, cartilage, and adipose tissue.
RNA was extracted from unstimulated proliferating BMSC ( lanes 2 and 5),
from confluent cells stimulated for the individual lineage (lanes 3 and 6),
and from parallel cultures of confluent unstimulated cells (lanes 4 and 7).
RNA also was extracted from native tissues as control for PCR products
(lanes 8 and 9): bone (A), cartilage (B), and adipose tissue (C). Reverse
transcribed cDNA was amplified by using specific oligonucleotide primers
(described in the Materials and methods section). PCR products were visu-
alized in ethidium bromide stained gels. Lanes 2, 3, 4, and 8 correspond to
GADPH product, lanes 5, 6, 7, and 9 (A) to OC product, lanes 5, 6, 7, and 9
Figure 2. Differentiation of human BMSC. Confluent human BMSC cul- (B) to ␣1(II) product, and lanes 5, 6, 7, and 9 (C) to LPL product. A 123-bp
tures were maintained in differentiation medium (A, C, E) and in standard ladder was used as standard (lane 1: from top to bottom 492, 369, 246,
control medium (B, D, F) for different time intervals. Osteocalcin deposi- 123 bp).
tion in the extracellular matrix, as revealed by immunostaining, was posi-
tive after the second week in osteogenic medium (A). After 1 week in
chondrogenic medium, discrete matrix noduli were positive for type II col-
lagen (C). Several large adipocytic colonies were observed with Sudan
black staining after 3 weeks of stimulation (E). Without the stimulating With respect to the basal level, the OC, ␣1(II), LPL levels
factors, cultures kept in the same conditions for the same time period dis- were increased by a factor of 16, 3.8 and 4.8, respectively
played only basal levels of differentiation (B, D, F). Immunostaining for (Table 2). In long-term unstimulated cultures, the levels of
osteocalcin (A, B), immunostaining for type II collagen (C, D), and Sudan ␣1(II) and LPL were similar to or lower than the basal level,
black staining (E, F). Bar ⫽ 1 mm.
showing absence of spontaneous differentiation (Table 2
and Fig. 3 B and C lane 7). In contrast, the level of OC
expression in long-term unstimulated cultures was about
three times higher than the basal level (Table 2 and Fig. 3A,
vealed by phase contrast microscopy and Sudan black stain-
lane 7).
ing (Fig. 2E). von Kossa, alizarin S, and alcian blue staining
For the three markers considered, the expected sizes of
were positive on similar cultures maintained in osteogenic
the PCR products were consistent with those obtained with
and chodrogenic medium, respectively (data not shown),
Control unstimulated cultures yielded barely detectable lev-
els of expression of the differentiation markers (Fig. 2B, D,
and F).
Table 2. Quantification of PCR products in differentiating
Differentiation in the three lineages induced by culture BMSC cultures
with inductive media was verified by semiquantitative RT-
PCR analysis using primers for type II collagen, OC, and Condition OC ␣1(II) LPL
LPL (Fig. 3 and Table 2). As shown in Figure 3 (lane 5), un- Basal expression 1 1 1
stimulated cultures already show some, although variable, Induced differentiation 16 3.8 4.8
basal expression of the specific makers studied: ␣1(II) Spontaneous differentiation 2.9 0.7 0.6
(panel B) and LPL (panel C) were clearly detectable,
RT-PCR reaction products shown in Figure 2 were quantified by image
whereas the OC level (panel A) was barely above the back- analysis. For each set of reactions, product intensity was corrected accord-
ground. With appropriate stimulation (Fig. 3, lane 6), the ing to the amount of the corresponding GAPDH PCR product and normal-
expression of the three markers was significantly induced. ized to the relative basal level.
712 A. Banfi et al./Experimental Hematology 28 (2000) 707–715

RNA extracted from the corresponding native tissues (Fig. mm3 of bone could be formed for every 161 CFU-F im-
3A–C, lane 9). planted, whereas, with expanded BMSC, 1 mm3 could be
Parallel with the loss of proliferation velocity was a loss produced for every 5,781 CFU-F implanted, on average.
of differentiation potential (Table 1). By the 19th doubling,
BMSC greatly reduced their response to adipogenic stimu- Lifespan and maintenance
lation to a borderline level and completely lost it by the of differentiation potential of BMSC clones
22nd passage. Chondrogenesis, as revealed by alcian blue One hundred twenty BMSC clones were obtained from five
staining, and osteogenesis, as assessed by alizarin S stain- different marrow samples, of which 80 were able to reach
ing, were maintained longer, but dropped to borderline liv- confluence after passing from the cloning 96 well into one
els by the fourth passage, corresponding to about 22 dou- well of a 24-well plate. At this stage, they were divided into
blings. four replicas for induction of differentiation. Only three of
the seven theoretically possible phenotypes (O, C, A, OC,
In vivo bone-forming efficiency OA, CA, OCA) were observed: 34% of the clones were os-
Osteogenic potential of freshly harvested bone marrow teo-chondro-adipogenic (tripontential or OCA clones), 61%
cells, which include both osteogenic and hemopoietic cells, were osteo-chondrogenic (bipotential or OC clones), and
was investigated and compared to that of expanded primary 5% were purely osteogenic (O clones).
BMSC, which reached confluence in about three weeks, un- Lifespan and the ability to maintain differentiation po-
dergoing 17.5 population doublings. Samples of fresh bone tential with time in culture were determined for three OCA
marrow were adsorbed onto HA ceramic cubes and subcuta- and two OC clones (Table 4). OCA clones were able to
neously implanted in nude mice as described in the Materi- reach 22–23 doublings after about 80 days of culture (Fig.
als and methods section. We first determined that when 5 4). All had lost adipogenic potential when tested at passage
million nucleated cells from fresh bone marrow after sepa- 3 (equivalent to 22 doublings), but maintained their osteo-
ration on Ficoll were implanted, the amount of bone tissue chondrogenic potential. The two OC clones had a slightly
formed was roughly comparable to that observed in the ex- shorter lifespan, reaching 19 and 22 doublings, respectively.
periments where the standard 250,000 expanded BMSC At passage 3, one clone lost its chondrogenic poten-
were used. Hemopoietic marrow was detected in both speci- tial, whereas the other clone maintained its bipotentiality
mens (data not shown). (Table 4).
Several independent experiments were performed, and
the data from one representative experiment are given in
Table 3. It should be noted that over the 3-week culture pe- Discussion
riod, only about 14% of the expanded BMSC remained clo- The data presented in this study indicate that, after in vitro
nogenic (data not shown). When the capacity to promote culture, the population of expanded BMSC progressively
bone formation in the nude mice assay was evaluated, we lost their ability to proliferate, differentiate in multiple mes-
observed that the osteogenic potential of these expanded enchymal lineages, and efficiency in generating mature
CFU-F decreased about 36 times with respect to CFU-F in bone tissue. This effect is clearly evident after the first pas-
fresh bone marrow. We found that, with fresh marrow, 1

Table 4. Lifespan and differentiation potential of tripotential and

Table 3. Osteogenic efficiency of fresh bone marrow versus primary bipotential clones
expanded BMSC
Doubling Doubling
Fresh marrow* Expanded BMSC† Clone no. O C A no. O C A

Population doublings 0 17.5 1 19 ⫹ ⫹ ⫹ 22 ⫹ ⫹ ⫺

Total bone formed (mm3) 2 19 ⫹ ⫹ ⫹ 23 ⫹ ⫹ ⫺
by 103 CFU-F 6.19 0.17 3 19 ⫹ ⫹ ⫹ 23 ⫹ ⫹ ⫺
Osteogenic unit‡ 161 5,781 4 18 ⫹ ⫹ ⫺ 19 ⫹ ⫹ ⫺
5 19 ⫹ ⫹ ⫺ 22 ⫹ ⫺ ⫺
In vivo implants were performed with 2.5 ⫻ 105 expanded BMSC or 5 ⫻
106 nucleated cells from fresh marrow and the CFU-F content of the cul- Clone lifespan ranged between 19 and 23 cell doublings. OCA clones
tures was determined as described in the Materials and methods section. reached 22–23 doublings, whereas OC clones had a slightly shorter
From the results of these experiments, the expected yield of 103 fresh or ex- lifespan, reaching 19 and 22 doublings. The three OCA clones lost adipo-
panded CFU-F and the sizes of the respective osteogenic units were calcu- genic potential by 22 doublings, although they maintained osteo-chondro-
lated. genic potential. Of the two OC clones tested, one lost its chondrogenic poten-
* Average of triplicate samples of one representative experiment. tial, whereas the other clone maintained both osteogenic and chondrogenic

Average of duplicate samples of one representative experiment. potential.

Number of clonogenic CFU-F necessary to induce the formation of 1 ⫹ ⫽ positive; ⫺ ⫽ negative.
mm3 of bone when implanted as described in the Materials and methods A ⫽ adipogenic phenotype; C ⫽ chondrogenic phenotype; O ⫽ osteo-
sections. genic phenotype.
A. Banfi et al./Experimental Hematology 28 (2000) 707–715 713

tain healthy osteogenesis for the whole life of the recipient,

possibly a young child, clearly what will determine the last-
ing success of the procedure is the number of stem cells or
very early progenitors present in the reinfused population.
We sought, therefore, to study BMSC stem characteristics
during culture expansion.
Concepts currently accepted as defining a stem cell are
the self-renewing capability, the high capacity for cell divi-
sion (lifespan) maintained throughout the lifetime of the or-
ganism, and the multipotential differentiation capability
[31]. These criteria might not apply to cells of postnatal
mesenchymal tissues, where the apparent cell plasticity
might play the same role without the need for a real stem
cell compartment, given the low tissue turnover in adult-
hood [4,9]. In other words, the existence of the MSC in the
postnatal organism is still debatable.
To test the self-renewing ability of expanding BMSC, we
investigated the growth kinetics of primary cultures and
monitored their ability to differentiate into the three lin-
eages considered at each passage.
Gene expression for lineage-specific markers was ana-
Figure 4. Lifespan and differentiation potential of BMSC tripotential
clones. The lifespan of three OCA clones was determined and the mainte- lyzed by RT-PCR and confirmed the immunochemical data.
nance of the original differentiation potential at increasing cell doublings Interestingly, low but detectable levels of expression for the
was verified. They reached 22–23 doublings in about 80 days of culture. three markers considered were present before stimulation,
By then, they had lost any apparent capability of proliferating. The threee indicating a possible commitment of some the these cells to
clones maintained the original differentiation potential up to 19 doublings,
differentiate into the three lineages tested already at the pri-
but they lost the adipogenic potential by the 22nd doubling still retaining
the OC potential The curve represents the average cumulative doubling mary culture stage. Noteworthy is that, in long-term unstim-
number of the three clones versus time in culture (⫾ SE). The differen- ulated cultures, only the level of osteocalcin expression was
tiation ability is represented by the underlying histograms. Arbitrary units: increased spontaneously with respect to the basal value.
0 ⫽ negative; 1 ⫽ borderline; 2 ⫽ positive, as defined in the Material and This may indicate that the osteogenic pathway is the pre-
methods section.
ferred lineage through which these cells progress, possibly
because of an intrinsic commitment or because the in vitro
culture conditions represent a microenvironment favoring
sage and is possibly already occurring during the primary osteogenesis.
culture, as indicated by the bone-forming efficiency assay. The kinetics of the differentiation to the three lineages
This behavior is consistent with a population undergoing during mitotic expansion show that adipogenesis starts fad-
commitment and differentiation, although this appears to be ing by 18 doublings and is totally lost by 22 passages, when
a gradual process. chondrogenesis and osteogenesis start weakening. In agree-
The presence in the BMSC population of progenitors for ment with Bruder et al. [32], we found a slowing of cell pro-
both the mesenchymal family of tissues and for hemopoi- liferation as a function of increasing passages. Although
esis-regulating stroma, together with the fact that these cells these authors report a faster proliferation rate at every pas-
can be easily isolated from bone marrow aspirates and can sage considered, this may reflect different culture condi-
be conveniently expanded and transfected in vitro, has tions, whereas the trend toward decreasing replication poten-
prompted a wave of interest in their exploitation as vehicles tial with increasing cumulative doublings is in concordance
for cell and gene therapy of both hematologic and musculo- with our data.
skeletal pathologies. This orderly loss of differentiation potential, paralleling
In the various therapeutic settings currently under inves- that of proliferation velocity, strongly points to a progessive
tigation, the need for BMSC to display stem cell features commitment of the population as a whole and to the inabil-
varies considerable. When used for hemophilia B gene ther- ity, in these culture conditions, of the putative stem cells in
apy, BMSC are required to engraft and survive in a site with the expanding population to self-renew and maintain a
access to the general circulation and secrete factor IX with- pluripotent differentiative capability.
out necessarily proliferating and differentiating; therefore, Interestingly, the largest drop in proliferation rate in all
the presence of stem cells is not the main issue. On the other three primary cultures occurs on the first passage, whereas
hand, if the reinfused BMSC are to reconstitute the he- the differentiation potential is maintained for one passage
mopoiesis-supporting stroma or, even more so, are to sus- longer. At the primary culture stage, CFU-F are plated to-
714 A. Banfi et al./Experimental Hematology 28 (2000) 707–715

gether with the whole marrow microenvironment: a part of sponsible for BMSC maintenance in an earlier progenitor
this is made of nonadherent cells (mainly hemopoietic) and state is desirable.
is removed with the medium changes, but a part is surely
adherent to the plastic or the stromal cells themselves (mac-
rophages, endothelial cells, and early hemopoietic progeni- Partially supported by grants from Associazione Italiana Ricerca
tors at least) and can be responsible for a complex network sul Cancro, Italian Space Agency (ASI) and European Space
of soluble and cell-contact signals to the BMSC. It is tempt- Agency (ESA). We wish to thank the Bone Marrow Transplant
ing to speculate that dilution of this nonproliferating com- Centres at both S. Martino Hospital and G. Gaslini Pediatric Hos-
partment with the first passage can be related to BMSC loss pital for their factive collaboration and Prof. G. Corte for helpful
of proliferation and multidifferentiative ability. Further ex- discussion.
periments are needed to test this hypothesis.
It is possible that stem cells are present in the initial cul-
ture and that they undergo unequal division during expan- References
sion, producing a vast progeny of committed and differenti- 1. Bentley SA (1982) Bone marrow connective tissue and the haemopoi-
ating daughter cells and becoming progressively diluted etic microenvironment. Br J Haematol 50:1
during the following passages. The nearest approximation 2. Dexter TM (1982) Stromal cell associated haemopoiesis. J Cell Phys-
iol Suppl 1:87
to the putative MSC that can be studied is the CFU-F.
3. Friedenstein AJ (1990) Osteogenic stem cells in the bone marrow.
Therefore, we compared the osteogenic efficiency of native Bone Miner Res 7:243
CFU-F in fresh marrow to that of expanded CFU-F in first 4. Beresford J (1989) Osteogenic stem cells and the stromal system of
confluence BMSC. The result was that first confluence ex- bone and marrow. Clin Orthop Rel Res 240:270
panded CFU-F were 36 times less efficient than fresh mar- 5. Caplan AI (1991) Mesenchymal stem cell. J Orthop Res 9:641
6. Brockbank KG, de Jong JP, Piersma AH, Voerman JS (1986) He-
row samples at generating bone. The in vivo bone formation
mopoiesis on purified bone-marrow-derived reticular fibroblast in
assay depends on the presence in the sample of a sufficient vitro. Exp Hematol 14:386
number of putative stem cells and/or early progenitors able 7. Friedenstein AJ, Chailakhyan RK, Latsinik NV, Panasyuk AF, Keiliss-
to proliferate and give rise to a consistent differentiated Borok IV (1974) Stromal cells responsible for transferring the mi-
progeny. Therefore, after in vitro expansion of about 17 croenvironment of the hemopoietic tissues. Cloning in vitro and re-
transplantation in vivo. Transplantation 17:331
doublings, which is in the range necessary to obtain a thera-
8. Tavassoli M, Friedenstein A (1983) Hemopoietic stromal microenvi-
peutically useful numbers of cells, the clonogenic popula- ronment. Am J Hematol 15:195
tion in a BMSC primary culture is significantly depleted of 9. Owen M (1988) Marrow stromal stem cells. J Cell Sci 10:63
osteogenic early progenitors/putative stem cells in compari- 10. Beresford JN, Bennett JH, Devlin C, Leboy PS, Owen ME (1992) Evi-
son with a fresh bone marrow sample. dence for an inverse relationship between the differentiation of adipo-
cytic and osteogenic cells in rat marrow stromal cell cultures. J Cell
Results of experiments we performed on a number of
Sci 102:341
clones to determine their actual lifespan are consistent with 11. Johnstone B, Hering TM, Caplan AI, Goldberg VM, Yoo JU (1998) In
the results obtained from primary cultures. Results indicate vitro chondrogenesis of bone marrow-derived mesenchymal progeni-
that these cells are able to undergo a limited number of mi- tor cells. Exp Cell Res 238:265
totic divisions, with characteristic kinetics, progressing 12. Cherington V, Chiang GG, McGrath CA, et al. (1998) Retroviral vec-
tor-modified bone marrow stromal cells secrete biologically active fac-
faster at the primary culture level, gradually slowing down
tor IX in vitro and transiently deliver therapeutic levels of human fac-
during the following passages, and reaching a complete stop tor IX to the plasma of dogs after reinfusion. Hum Gene Ther 9:1397
after 22–23 cell doublings. Futhermore, the three OCA 13. Galotto M, Berisso G, Delfino L, et al. (1999) Stromal damage as con-
clones analyzed displayed a loss of adipogenic potential by sequence of high-dose chemo/radiotherapy in bone marrow transplant
doubling 22, and one of two OC clones lost its chondro- recipients. Exp Hematol 27:1460
14. Barbot C, Rice A, Vanes I, Mahon FX, Jazwiec B, Reiffers J (1994)
genic potential by 22 cell doublings.
Quality and functional capacity of the bone marrow microenvironment
Ex vivo expansion is necessary to obtain a sufficient of autologous blood stem cell transplantation (ABSCT) recipients.
number of cells for in vitro engineering and use in novel Nouv Rev Fr Hematol 36:325
strategies of gene and cell therapy. When the reinfused 15. Carlo-Stella C, Tabilio A, Regazzi E, et al. (1997) Effect of chemo-
BMSC are used to sustain healthy histogenesis for the rest therapy for acute myelogenous leukemia on hematopoietic and fibro-
blast marrow progenitors. Bone Marrow Transplant 20:465
of the life of the patients, possibly young children, as in the
16. del Canizo C, Lopez N, Caballero D, et al. (1999) Haematopoietic
case of osteogenesis imperfecta, the stem cell content of the damage persists 1 year after autologous peripheral blood stem cell
population delivered clearly is of paramount importance transplantaion. Bone Marrow Transplant 23:901
and the effects of the in vitro expansion stage must be con- 17. Domenech J, Gihana E, Dayan A, et al. (1994) Haemopoiesis of trans-
sidered. However, it is not known how many stem cells/ planted patients with autologous marrows assessed by long-term mar-
row culture. Br J Haematol 88:488
early progenitors are needed in the reinfused population to
18. Domenech J, Roingeard F, Herault O, et al. (1998) Changes in the
obtain a therapeutic benefit, and it is possible that a limited functional capacity of marrow stromal cells after autologous bone mar-
expansion would be compatible with a successful proce- row transplantation. Leuk Lymphoma 29:533
dure. A better understanding of the culture conditions re- 19. Radford JA, Testa NG, Crowther D (1990) The long-term effects of
A. Banfi et al./Experimental Hematology 28 (2000) 707–715 715

MVPP chemotherapy for Hodgkin’s disease on bone marrow function. ity and therapeutic effects of bone marrow-derived mesenchymal cells
Br J Cancer 62:127 in children with osteogenesis imperfecta. Nat Med 5:309
20. Rice, A, Reiffers J, Bernard P, et al. (1992) Incomplete stroma forma- 27. Martin I, Muraglia A, Campanile G, Cancedda R, Quarto R (1997) Fi-
tion after allogeneic marrow or autologous blood stem cell transplanta- broblast growth factor-2 supports ex vivo expansion and maintenance
tion. Nouv Rev Fr Hematol 34:167 of osteogenic precursors from human bone marrow. Endocrinology
21. Testa NG, Hendry JH, Molineux G (1985) Long-term bone marrow 138:4456
damage in experimental systems and in patients after radiation or che- 28. Chomczynski P, Sacchi N (1987) Single-step method of RNA isolation
motherapy. Anticancer Res 5:101 by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal
22. Koc ON, Gerson SL, Cooper BW, et al. (1998) Rapid hematopoietic Biochem 162:156
recovery after coinfusion of autologous culture-expanded human mes- 29. Hicok KC, Thomas T, Gori F, Rickard DJ, Spelsberg TC, Riggs BL
enchymal stem cells (hMSCs) and PBPCs in breast cancer patients re- (1998) Development and characterization of conditionally immortal-
ceiving high-dose chemotherapy. Blood 92:274a (abstr) ized osteoblast precursor cell lines from human bone marrow stroma. J
23. Onyia JE, Clapp DW, Long H, Hock JM (1998) Osteoprogenitor cells Bone Miner Res 13:205
as targets for ex vivo gene transfer. J Bone Miner Res 13:20 30. Rickard DJ, Kassem M, Hefferan TE, Sarkar G, Spelsberg TC, Riggs
24. Hou Z, Nguyen Q, Frenkel B, et al. (1999) Osteoblast-specific gene BL (1996) Isolation and characterization of osteoblast precursor cells
expression after transplantation of marrow cells: implications for skel- from human bone marrow. J Bone Miner Res 11:312
etal gene therapy. Proc Natl Acad Sci U S A 96:7294 31. Morrison SJ, Shah NM, Anderson DJ (1997) Regulatory mechanisms
25. Pereira RF, O’Hara MD, Laptev AV, et al. (1998) Marrow stromal in stem cell biology. Cell 88:287
cells as a source of progenitor cells for nonhematopoietic tissues in 32. Bruder SP, Jaiswal N, Haynesworth SE (1997) Growth kinetics, self-
transgenic mice with a phenotype of osteogenesis imperfecta. Proc renewal, and the osteogenic potential of purified human mesenchymal
Natl Acad Sci U S A 95:1142 stem cells during extensive subcultivation and following cryopreserva-
26. Horwitz EM, Prockop DJ, Fitzpatrick LA, et al. (1999) Transplantabil- tion. J Cell Biochem 64:278