View - Blood

~~
Loss of Attachment to Fibronectin With Terminal Human Erythroid Differentiation
By M.H. Vuillet-Gaugler, J. Breton-Gorius, W. Vainchenker, J. Guichard, C. Leroy, G. Tchernia, and L. Coulombel
Human erythroblastic progenitors (colony-forming uniterythroid
[CFU-E] and burst-forming unit-erythroid [BFU-
E]) have been shown to attach to fibronectin (Fn), a
property that might be involved in the local regulation of
erythropoiesis. In this study, we have investigated changes
in cell attachment to Fn upon terminal erythroid differentiation.
We first purified CFU-E from human marrow by
avidin-biotin immune rosetting. This negative selection
procedure yielded a cell population containing -80% blasts
that, after characterization by colony-assays and electron
microscopy, appeared to consist of CFU-E (10% to 15%)
and their immediate progeny (85% to 90%). here defined as
“preproerythroblasts.” In the presence of erythropoietin,
purified cells differentiated into reticulocytes in 7 to 10
days. Cell attachment to Fn was inversely correlated to the
N MANY TISSUES acquisition of differentiated func-
I tions is induced by either cell-cell or cell-matrix
interactions.’.* In the bone marrow also, hematopoietic stem
cells depend for their differentiation upon interactions with a
complex environment composed of nonhematopoietic stromal
cells and their extracellular matrix (ECM).3-6 However,
there is little known about the molecular mechanisms involved
in these interactions. Recently attention has been
focused on the role of growth factors that are synthesized by
stromal cells and act either in a soluble or bound to
the cell-surface9 or to ECM components.” However, most of
these studies have been performed with fibroblasts and
endothelial cells from extramedullary sources that may not
help to elucidate lineage- or differentiation-specific interactions
between hematopoietic and stromal cells in the marrow.
These may be accounted for by multiple specialized marrow
stromal cells. Alternatively, hematopoietic cells themselves
may modulate, as differentiation proceeds, the expression on
their surface of receptors involved in stromal cell or matrix
recognition. Likely candidate molecules for such a mechanism
are integrins.” These glycoproteins exist as heterodimers
of noncovalently associated LY and @ transmembrane
subunits and are widely distributed. Three different @
chains exist, which define three integrin families.” In each
family, @ associates with multiple a chains, resulting in
different ligand specificities. Thus, the al-6 81 family, or
VLA family, includes most of the cell surface receptors for
Fn and ECM proteins.”.’*
There are at least two arguments to implicate integrins
and cell-matrix adhesion in the regulation of hematopoietic
differentiation. First, the ECM produced by marrow stromal
cells has been shown to contribute to the function of the
adherent layer in long-term marrow culture^.'^^'^ Second,
we’’ and ~thersl~.’~ have demonstrated direct adhesion of
progenitor cells to ECM proteins. Thus, colony-forming
units-erythroid (CFU-E) and some burst forming unitserythroid
(BFU-E) show significant attachment to Fn,”.I6
whereas CFU granulocyte-macrophage (GM) interact with
a different ECM component.” Furthermore, in the erythroblastic
lineage, adhesion to Fn is a differentiation-regulated
event, as shown by the higher capacity of CFU-E than
BFU-E to bind Fn,” and by the loss of Fn-adhesion and of
stage of differentiation of the erythroid cell: more than 50%
of the CFU-E population reproducibly adhered to Fn,
whereas at most 30% of the preproerythroblasts had the
same capacity. Adhesion was further lost at late maturation
stages, and a constant finding was the inability of
reticulocytes to adhere to Fn. Finally, CFU-E adhesion to Fn
was blocked by polyclonal IgG raised against the Fn
receptor and by a monoclonal antibody against VLA-I.
These results demonstrate that adhesion to Fn is developmentally
regulated during normal human erythropoiesis.
Restriction of its expression to CFU-E and its first divisions
strikingly correlates with the migratory capacity of these
cells.
0 1990 by The American Society of Hematology.
the Fn receptor in murine erythroleukemic cells when induced
by dimethyl sulfoxide
In the present study, we have investigated how adhesion to
Fn is modulated during terminal erythroid differentiation.
For this purpose, we developed a procedure for obtaining a
highly enriched population of very immature erythroid cells,
ie, CFU-E and their immediate progeny. These purified cells
were then induced to differentiate in vitro, and during this
process, we investigated their changing ability to attach to
Fn. We found that CFU-E attach strongly to Fn, and that
this property is lost at specific stages of terminal erythroid
maturation. Moreover, the binding of CFU-E to Fn is
mediated by a receptor bearing some similarities to the
integrin family.
MATERIALS AND METHODS
Bone marrow samples. Normal marrow samples were obtained
after informed consent from hematologically normal patients undergoing
hip surgery. Cells were flushed from the bone fragments with a
medium containing 100 to 200 pg/mL DNAse (type I, Sigma
Chemical Co, St Louis, MO). Cells were separated on Ficoll-
Hypaque (1.077 g/mL) (Eurobio, Paris, France). Light-density
mononuclear cells were washed twice, resuspended at a concentration
of 3 x lo6 cells/mL in a-medium supplemented with 10% fetal
From the Hematology Laboratory and University Paris XI,
H6pitaI BicBtre, Kremlin-BicZtre: and INSERM U.91, H6pital
Henri-Mondor, CrLteil, France.
Submitted April 24,1989; accepted October 24. 1989.
Supported by Grants No. CRE 862007 from INSERM (L.C.);
Grant No. ARC 6532 from the Association pour la Recherche
contre le Cancer; and a grant from Ligue contre le Cancer (Comitt
Dkpartemental des Hauls-de-Seine). M.H. K-G. is the recipient of a
fellowship from the French Ministere de I’lndustrie et de la
Recherche.
Address reprint requests to Laure Coulombel, MD, PhD, Institut
de Pathologie Cellulaire, H6pital Bici?tre, 94275 Kremlin-Bici?tre,
France.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be hereby marked
“advertisement” in accordance with 18 U.S.C. section 1734 solely to
indicate this fact.
6 1990 by The American Society of Hematology.
0006-4971/90/7504-0028$3.00/0
Blood, Vol 75, No 4 (February 15), 1990: pp 865-873 865

866 VUILLET-GAUGLER ET AL
calf serum (FCS) (Flow Laboratories, Paris, France), and incubated
in 75 cm’ flasks for 1 hour at 37OC in a 5% CO, atmosphere in air to
remove plastic-adherent cells. The nonadherent mononuclear cells
(NA-MNC) were subsequently fractionated by immune rosetting
(see below).
Antibodies. Monoclonal antibodies (MoAbs) against human
HLA-DR (OKIal), human T cells (OKT3), and human CALLA
(CD IO; OKB-CALLA) were purchased from Ortho Diagnostic
Systems (Roissy, France). MoAb recognizing HPCA (CD34) was
from Becton Dickinson (Mountain View, CA), BI (anti-B cells)
from Coulter Immunology (Hialeah, FL). MoAbs CH3 (IgM
recognizing granulocytes and CFU-GM), CS4 (recognizing glycophorin
A) and CK26 (binding to lymphoid and monocytic cells) have
been previously described,20 and were generously provided by Dr
Berthier (INSERM U.217, Grenoble, France). Anti-human HLA-
DR K5 MoAb was a gift from Dr Fermand (INSERM U108,
Hdpital St Louis, France). FA6-152 MoAb,,’ which recognizes an
88 Kd glycoprotein on platelets, monocytes, and erythroid cells and
belongs to the CD36, was a gift from L. Edelman (Institut Pasteur,
Paris, France). Anti-platelet GPIIIa C17 MoAb and anti-glycophorin
A (GPA) plyclonal antibody used in electron microscopy
were kind gifts from Drs F. Tetteroo22 and L.C. Ander~son.~’
Biotin-conjugated affinity-purified goat anti-mouse IgM (p specific)
and IgG (H -t L) F(ab’)2 fragments were obtained from Cappel-
Cooper Biomedical (Malvern, PA). A rabbit antiserum prepared
against the purified human placental fibronectin receptor (HFnR-
Ab) was purchased from Telios Pharmaceuticals (La Jolla, CA).
This antibody recognizes mainly the 115 Kd 81 subunit of the
receptor, and is less reactive with the 160 Kd a5 subunit.24 A rat
MoAb BlE5, which is directed against the a5 subunit of the human
Fn receptor and which selectively inhibits cell attachment to Fn was
kindly provided by C. Damsky (University of California, San
Francisco, CA) and used as hybridoma s~pernatant.~’
Enrichment of erythroblastic cells by immune rosetting. Cell
purification was performed by negative selection using avidin-biotin
immune rosetting according to the original technique described by
Fiirfang and Thierfelder.26 Briefly, 1.2 mg avidin (Vector Laboratories
Inc, Burlingame, CA) was covalently coupled to 2 x lo9
glutaraldehyde-fixed sheep red blood cells (SRBC-A) as described in
detail elsewhere.26 SRBC-A can be kept at 4OC in sterile conditions
for at least 4 to 6 months without loss of reactivity. The purification
was performed in two steps: NA-MNC (usually 0.5 to 3 x IO8 cells)
were first mixed with a 10-fold excess of SRBC-A. The mixture was
centrifuged 6 minutes at 200g, and the pellet was incubated 45
minutes at 4OC. Cells were then gently resuspended at a concentration
of 1 x lo7 cells/mL in a-medium supplemented with 5% FCS
((~5%) and layered over Ficoll-Hypaque. Mature granulocytes that
had phagocytized SRBC-A were eliminated in the pellet (fraction
P-1). Cells harvested at the interface (fraction 1-1) were washed
twice, resuspended at a concentration of 4 x lo7 cells/mL, and
incubated 45 minutes at 4OC with an excess of the following MoAbs:
OKIal, OKT3, OKB-CALLA, B1, anti-HPCA, K5, CH3, CS4,
CK26. At the end of the incubation period, cells were washed twice,
resuspended at a concentration of 4 x lo7 cells/mL in a5% and
incubated with a 1/40 dilution of a mixture of biotin-conjugated
goat anti-mouse IgM and IgG for 45 minutes at 4°C. Excess
biotinylated immunoglobulin was removed by three washes in 015%.
Finally, labeled cells were incubated with SRBC-A in the proportion
of l:lO, as described above, and rosettes were allowed to form. Cells
were then gently resuspended and adjusted to the initial concentration.
Rosettes were removed by centrifugation over Ficoll-Hypaque
(fraction P-2), and unlabeled cells were recovered at the interface
layer (fraction 1-2). Cells recovered in each fraction (1-1, P-I, 1-2,
P-2) were counted and the number of rosettes was determined.
Purified cells (1-2) were characterized morphologically on May-
Griinwald-Giemsa (MGG) stained cytocentrifuged slides, by electron
microscopy (see below), and by their ability to form colonies in
methylcellulose colony-assays.
Electron microscopy. The simultaneous detection of specific
membrane markers (GPIIIa and GPA) by immunogold and intracellular
peroxidatic activities was determined on the purified erythroid
cell population (1-2) as previously described in detaiL2’ Briefly,
immunolabeling was first performed by incubating cells with specific
antibodies recognizing either GPIIIa or GPA, and then with a
second antibody coupled with 15 nm and 5 nm gold particles,
respectively. Cells were then fixed in 1.25% glutaraldehyde in Gey’s
buffer, washed, incubated in diaminobenzidine medium,28 post-fixed
with osmium tetroxide, dehydrated, and embedded in Epn. Sections
were examined with a Philips electron microscope CM 10 (Eindhoven,
The Netherlands) without staining in order to easily identify
ferritin molecules by their own contrast.
Suspension culture of purijed erythroblastic cells. Purified
cells from the 1-2 fraction were cultured at a concentration of 1 x lo6
cells/mL in a-medium containing 20% FCS and 3 U/mL human
urinary erythropoietin (Ep). The batch of FCS used during the study
contained less than 10 pg of Fn per milliliter as measured by an
enzyme-linked immunosorbent assay (ELISA). Cultures were initiated
in 96-flat bottom microtiter plate wells (Falcon, Becton-
Dickinson, France, No. 3072; 200 pL per well) and incubated at
37OC in a water saturated air atmosphere with 5% CO,. Cells were
harvested regularly, and their number, viability (dye exclusion),
morphology, plating efficiency, and ability to adhere to Fn were
determined. Three stages of maturation were defined after MGG
staining: immature (proerythroblasts and early basophilic erythroblasts),
intermediate (late basophilic and polychromatophilic), and
mature (acidophilic and enucleations). Reticulocytes were counted
separately .
Hematopoietic progenitor cell colony-assays. Cells from the
P-1, 1-1, P-2, and 1-2 fractions were plated in a-medium containing
0.8% methylcellulose, 30% FCS, 1% deionized bovine serum albumin
(BSA), mol/L 8-mercaptoethanol, 3 U/mL Ep and 10%
agar leukocyte conditioned medium. Final concentration of cells per
dish was 1.5 x IO’ for NA-MNC and P-1 and 1 to 3 x lo4 for 1-1,
1-2, and P-2. Erythroid colonies were scored as previously described
in detail7
Adhesion assay. The capacity of erythroid cells to adhere onto
Fn was assessed in two different ways: purified 1-2 cells and cells
grown during various periods of time in suspension (see above) were
incubated on Fn-coated 2.5 cm2 tissue culture wells (2 to 4 x 10’
cells per well) in (Y 5% for 2 hours at 37OC in a 5% CO, atmosphere.
Nonadherent cells were then harvested, weakly adherent cells were
removed by two mild washings, and strongly adherent cells were
detached by vigorous pipetting or short trypsinization. Adherent and
nonadherent cells were counted, stained with MGG for morphologic
examination, and plated in methylcellulose colony-assays at 1.5 x
lo4 cells per dish to assess the number of CFU-E in each fraction. In
the second method, cells were grown as described above for suspension
culture, except that they were incubated from the start on
Fn-coated 96-flat bottom microtiter plate wells (1 x lo6 cells/mL,
200 pL per well) and kept on that substrate for the whole culture
period. Every 2 days during the 6 to 10 day-culture period, one well
was sacrificed and nonadherent and adherent cells were harvested
separately, counted, and stained with MGG. Cell viability of both
fractions was checked by dye exclusion (Nigrosin). In both methods,
tissue-culture wells were precoated with 50 to 80 pg/mL of Fn
purified from human plasma by gelatin affinity chromatography, or
human collagens I, 111, and IV in the monomeric form or laminin.
Nonspecific binding sites were blocked by 1-hour incubation in 10
mg/mL BSA.”

I
ERYTHROBLAST ADHESION TO FIBRONECTIN 867
L A
- {
A
Fig 1. Photomicrographs of MGG-stained 1-2 cells before culture (A), and after 7 days (6). and 10 days (C), in suspension culture in the
presence of Ep. There was no difference in the morphology of cells grown in suspension or on an Fn 8ubstrate. (Original
magnification, xs00)
In some experiments, 1-2 cells were first incubated (45 minutes at
4OC) with a 1/10 dilution of the HFnR-Ab or undiluted BlE5
hybridoma supernatant, and then incubated as described above onto
Fn-coated wells for 2 hours in the continuous presence of the
antibody. Negative controls included incubation of cells either with
nonimmune rabbit IgG or with anti-CD36 (FA6-152) or anti-GPA
hybridoma supernatant (CS4).
Indirect immuncfluoresence labeling. Purified erythroblasts were
stained with the HFnR-Ab by indirect immunofluorescence. For
that purpose, 0.5 to 1 x lo6 cells in 100 pL were incubated 30
minutes at 4% with a 1/10 dilution of the HFnR-Ab or control
nonimmune rabbit IgG. Cells were subsequently washed twice and
resuspended in a 1 /50 dilution of fluorescein-labeled goat anti-rabbit
IgG F(ab')2. Incubation was continued for 30 minutes at 4OC. Cells
were washed twice, resuspended at a final concentration of 1 x IO6
cells/mL in phosphate buffered saline (PBS) containing 1% BSA
and examined under a Zeiss microscope equipped for epifluorescence.
RESULTS
Purification to homogeneity of very immature human
marrow erythroblasts. The purification procedure de-
scribed under Materials and Methods yielded in 20 experiments
a population (1-2) containing 79% f 3% of blasts that
were HLA-DR, CD34, and GPA negative by indirect immunofluorescence
(Fig 1). Positive labeling with the FA6-152
MoAb whose expression is restricted to platelets, monocytes,
and erythroid cells suggested the erythroid origin of these
~elIs.2~
The erythroid origin of purified 1-2 cells was clearly
demonstrated by the 1-2 cells' ultrastructural characteristics
and immunophenotype. Eighty percent of 1-2 cells were
classified as erythroid and included only two maturation
stages (Fig 2): the most immature phenotype represented
10% to 15% of this cell fraction. Despite the absence of GPA
and rhopheocytosis, the population was recognized as erythroid
by the presence of Theta granules (8-Gr) and of a few
ferritin molecules located in the 8-Gr and scattered through
the cytoplasm (Fig 2A, B). These cells were presumably
CFU-E, since the 1-2 cell fraction contained the same
proportion of progenitors giving rise to 1-2 hemoglobinized
clusters (see below). Other erythroid cells in the 1-2 fraction
were more mature, as shown by the presence of GPA (Fig
Fig 5. Photomicrographs of
MGG-stained adherent (A) and
nonadherent (6) erythroblasts
observed in a typical experiment
where purified 1-2 cells
were cultured in Fn-coated
wells for 7 days.
t
A

868 VUILLET-GAUGLER ET AL
.
Fig 2. Ultrastructural analysis
of purified 1-2 cells. (A) Pre
sumptive CFU-E. (inset left) At
low magnification (original magnification,
~3,300) a very large
nucleolus is saen in tho nucleus.
Cytoplasmic vacuoles and
granules are clustered in the
golgi zone. At higher magnification
(original magnification,
x 52.600). the structura of 0-Gr
is clearly distinguished. In all of
them, a few ferritin molecules
are identified. There are no gold
particles on the plasma mambrane,
indicating the absence
of GPA. (inset right) High magnification
(original, x 108,300)
of a 0-Gr with its dense rod and
some dense periodic structure.
The typical configuration of ferritin
molecules (f) is recognized
in the cytoplasm and in 0-Gr.
(E) Preproerythroblast. Weak
labaling for GPA is detected at
one place of the cell membrane
(arrow). Two rhopheocytosis invaginations
(r), also called
coated pits, are indicated. In
one of them, ferritin molecules
(f) are presant. Nota also the
presence of intracytoplasmic
free ferritin (f)(original magnification,
~68,800). (C) A more
matura preproarythroblast as
judged by the heavier labeling
(arrow) by GPA-Ab and by the
increased number of free ferritin
molecules (1). and ferritin
molecules associated with theta
granules 0-Gr (original magnification.
~58,800).
2B). However, GPA density was below the threshold of
sensitivity expected for conventional immunofluorescence or
avidin-biotin labeling, which explains why these cells were
not eliminated by immune rosetting. One or 2 sites of
rhopheocytosis were observed (Fig 2B), and the concentration
of ferritin in the 0-Gr and in the cytoplasm was increased
in comparison with the more immature cdls (Fig 2C).
Hemoglobin was absent, as judged by the negative dense
peroxidase reaction. These properties, together with the low
expression of rhopheocytosis and GPA, indicated that this
population was less differentiated than the classical
proerythroblast,*’ and we have therefore designated these
cells “preproerythroblasts” as they appear to represent a
stage intermediate between CFU-E and proerythroblasts.
The 1-2 fraction contained on average 65% preproerythroblasts.
Twenty to twenty-five percent of 1-2 cells were
nonerythroid, as recognized by electron microscopy. Most of
these belong to the megakaryocytic lineage, and a minority
were mast cells, basophil myelocytes, and promyelocytes
(data not shown).
Clonogenic properties of purified 1-2 cells were assessed in
methylcellulose colony-assays (Table 1). If we define a
CFU-E as a progenitor giving rise to a cluster of at least 10 to
15 hemoglobinized cells at day 9 to 10, 11% * 1% of 1-2 cells
were CFU-E (n = 20), whereas preproerythroblasts variably
gave rise to clusters of less than 10 cells that usually had
lysed by day 9. The proportion of CFU-E matched the
proportion of very early GPA negative erythroid cells recognized
by electron microscopy. CFU-E enrichment was 36 2
4-fold relative to NA-MNC and over 50 to 60-fold compared
with native marrow. The enrichment of preproerythroblasts
could not be calculated, as these cells cannot be accurately
quantitated in unpurified marrow cells either by morphologic
or colony-assays criteria. Recovery of CFU-E in the 1-2

ERYTHROBLAST ADHESION TO FIBRONECTIN 869
Table 1. Progenitor Cell Enrichment During the Different Steps of the Purification Procedure
Fraction
NA-MNC
I- 1
P- 1
1-2
P-2
No. Progenitors/l.B x lo5 Cells Plated
CFU-E mBFU-E IBFU-E CFU-GM
490 f 60
1,609 k 314
(3)
62 f 16
17,571 k 1,658
(36)
586 k 123
155 f 24 12 f 2 328 f 54
427 f 57 43 f 7 1,308 f 113
(2.7) (3.5) (4)
27 k 13 924 90 k 26
1,262 f 276 26 f 8 898 f 286
(8) (2) (2.7)
634 f 74 57 f 12 1,316 f 263
Each number represents the mean f SEM of seven different experiments. Numbers in brackets represent the enrichment factor relative to NA-MNC in
the 1-1 and 1-2 fractions.
Abbreviations: mBFU-E, mature BFU-E; iBFU-E, immature BFU-E.
fraction was 41% 3%. In seven experiments where the
different fractions obtained during the purification were
cloned in methylcellulose, 10% CFU-E were found in the P-2
fraction (Table 1). The latter may represent a subset of
HLA-DR+, CD34+ CFU-E. Our results also showed that
contamination by other progenitor cells was negligible since
in 20 experiments, BFU-E and CFU-GM represented 0.7%
and 0.6% of 1-2 cells, respectively (Table 1).
In order to assess the proliferative and differentiative
behavior of the purified 1-2 cells in vitro, we cultured them in
suspension in the presence of 3 U/mL Ep (but no other
hematopoietic growth factors). For the first 2 to 3 days, there
was no net change in the number of cells, whereas during the
next 7 to 9 days, there was a four- to ninefold increase over
the initial number of cells, reflecting the proliferation of the
maturing erythroblasts (Fig 3). Morphologic examination
showed that 1-2 cells differentiated synchronously up to the
enucleation stage, with the production of many reticulocytes
(Fig 1). In the absence of Ep, 1-2 cells did not survive beyond
48 hours. Nonerythroid cells were not maintained in these
cultures.
Adhesion of purified erythroblasts to Fn during their
diferentiation. CFU-E and preproerythroblasts adhesion
to Fn was evaluated in 19 separate experiments using 1-2
cells. The proportion of Fn-adherent CFU-E was precisely
quantitated by colony-assays of adherent and nonadherent
cells and that of Fn-adherent preproerythroblasts was estimated
from the proportion of total Fn-adherent 1-2 cells. The
mean (+SEM) proportion of Fn-adherent CFU-E was
58% f 5%. It was over 65% in 10 of 19 experiments, and
below 30% in only three experiments. In contrast to the high
proportion of adherent CFU-E, only 31% 3% of total
nucleated 1-2 cells were Fn-adherent. As CFU-E represents
10% to 15% of the 1-2 population, and preproerythroblasts
65%, this gives a proportion of Fn-adherent preproerythroblasts
below 25%. This difference in the ability of CFU-E
and preproerythroblasts to attach to Fn was seen in each
experiment, and suggests loss of adhesion to Fn as CFU-E
matures in preproerythroblasts. No significant attachment of
CFU-E and preproerythroblasts occurred onto either human
collagen I, 111, IV or laminin.
The effect of differentiation on the adherence to Fn was
next investigated: seven experiments were performed using
cells removed after varying periods of time from 7 to 10
day-cultures of highly CFU-E-enriched population, and six
experiments were performed by culturing cells 7 to 10 days
on an Fn substrate. Results from these 11 experiments have
been pooled in Fig 4. In each experiment, adhesion to Fn
decreased with maturation and was completely lost at late
maturation stages. Even though the rate of decline varied
between experiments, the pattern observed was similar in
each experiment, ie, in the purified 1-2 population before
culture (day 0), the proportion of adherent CFU-E was
higher (58%) than that of total purified cells (30%) and thus
preproerythroblasts, and slowly declined over the next 8 days
in culture to reach 20% at day 8 to 9 (polychromatophilic
stage). At that stage, the capacity to adhere to Fn appeared
to rapidly decline and was completely lost at the enucleation
- 10-
E
\ 9-
(D
-; 8-
z 7-
6-
LL
s 5-
W
z 4-
0
3-
W 2 2-
1 1 1 1 1 1 1 1 1 1 ,
1 2 3 4 5 6 7 8 9 10 11
DAYS IN CULTURE
2 Imm 81 87 30 3 0
0 0 60 59 22
Mat 0 0 10 38 78
2 1 inter
c
Fig 3. Changes in the number and cytology of cells present
after varying periods of time when purified 1-2 cells were cultured
in suspension in the presence of Ep. Each point represents the
mean f SEM of the cell concentration observed in three to seven
experiments. The SEM is not indicated for the day 10 point since
only two experiments were available. Morphologic examination of
the cells was performed after MGG staining at the indicated day
(see Materials and Methods).
1

870 VUILLET-GAUGLER ET AL
adhere to Fn was paralleled by the disappearance of a
cell-surface receptor (data not shown).
2
W
V
5 40-
W
K W
8 30-
a
I-
Z
W
p 20- Purified 1-2
W 0
10 - cells
DAYS IN CULTURE
Fig 4. Loss of adhesion to Fn with terminal erythroid differentiation.
The percentage of adherent cells and CFU-E was first
determined in the 1-2 population at day 0. Cells were then cultured
either in suspension (n = 7) or on Fn-coated wells (n = 6). and
the proportion of adherent cells was measured sequentially as
described in the text. Each point represents the mean c SEM of
the percent adherent cells. 1-2 cells differentiated synchronously
up to day 7. However, beyond that point, the population was
heterogeneous, and reticulocytes never represented 100% of the
cell population. Therefore, the day 11 point and the dashed line
represent the percent adhesion expected if 100% of day 7 cells
had matured into reticulocytes.
stage. Reticulocytes were always found exclusively in the
nonadherent fraction. This was best appreciated in experiments
in which purified 1-2 cells were cultured on Fn during
several days in the presence of Ep. In a representative
experiment, shown in Fig 5, 20% to 30% of the day 7 cells
were still strongly adherent to Fn. Most of these adherent
cells had a phenotype of polychromatophilic erythroblasts,
and no reticulocytes were seen (Fig 5). However, some
reticulocytes were already present in this culture, but they
had been released from the Fn substrate and represented
25% of the nonadherent population (Fig 5). The same
partitioning was seen for enucleating cells. Failure of enucleating
erythroblasts and reticulocytes produced in culture
(kFn) to attach to Fn was not attributable to any change in
their viability as assessed by dye exclusion. Furthermore,
assessment of enriched reticulocytes, freshly isolated from
the peripheral blood from three patients with hemolytic
anemia showed that these also did not attach to Fn in our
assay (data not shown).
Effect of anti-HFnR antibodies on the adhesion of purified
erythroblasts to Fn. In seven separate experiments,
CFU-E and preproerythroblasts (1-2 cells) were first preincubated
with either a 1/10 dilution of polyclonal IgG against
HFnR or with undiluted B1E5 hybridoma supernatant and
then incubated in Fn-coated wells. As shown in Fig 6, both
antibodies specifically inhibited CFU-E and 1-2 cells (ie,
preproerythroblast) adhesion to Fn, as nonimmune rabbit
IgG, anti-CD 36 (FA6-152), and anti-GPA had no effect.
HFnR-Ab labeled 100% of erythroid 1-2 cells by indirect
immunofluorescence, whereas enucleating cells and reticulocytes
stained negatively, suggesting that the disappearance
of the functional capacity of differentiating erythroblasts to
DISCUSSION
One of the major questions regarding hematopoiesis is the
understanding of the mechanisms whereby progenitor cells
interact with their marrow environment. Our recent
observationls that human marrow CFU-E and mature BFU-
E selectively attach to ECM prepared from marrow fibroblasts
and to purified Fn suggests that progenitor cellsmatrix
interactions might represent one such mechanism. In
the present study, we have demonstrated that attachment to
Fn is a property of most CFU-E and early erythroblasts in
normal human marrow, and that this property is subsequently
lost upon terminal erythroid differentiation.
We have approached this question by designing an in vitro
model that reproduces the successive steps followed by a
CFU-E when it differentiates into reticulocytes. Our strategy
was first to purify human CFU-E directly from fresh marrow
samples. Negative selection by avidin-biotin immune rosetting
reproducibly yielded a homogenous population of very
immature cells identified as erythroid by their positive
labeling with FA6-1 52,29 and by ultrastructural criteria. The
procedure described was nontoxic, and more powerful and
reliable in our hands than panning.30s31 Moreover, large
numbers of marrow cells could be easily processed within a
few hours with minimal cell loss, in contrast to cell sorting.'*
The availability of a highly enriched population of very
immature normal erythroblasts gave us the opportunity to
delineate the ultrastructural changes that are associated with
human erythroid cell differentiation. Previous information
has relied on analyses of cells from patients with erythroblastic
leukemia or of partially purified populations of normal
CFU-E.31 The earliest markers detected in 1-2 cells were
0-Gr and ferritin molecules (Fig 2). Most likely, cells that
exhibited these two features only represent CFU-E. This
4 60
W
I
5 30
g 20
: 10
a.
CFU-E
T
T
PURIFIED CELLS
Fig 6. Inhibition of CFU-E and 1-2 cell attachment to Fn by
anti-HFnR antibodies. 1-2 cells were incubated 2 hours on Fncoated
wells after preincubation with either polyclonal IgG against
the human Fn receptor (a5Dl; ) or B1E5 MoAb against the a5
chain (W). Controls included preincubation of the cells with no
antibody (0). nonimmune rabbit IgG (&I, or with hybridoma
supernatant from an unrelated erythroid-specific antibody (B).
Histograms show the mean * SEM of the percentage of adherent
cells (n = 6).

ERYTHROBLAST ADHESION TO FIBRONECTIN 87 1
conclusion is based on our finding that this morphology
identified the same proportion of cells in the purified population
as were detected by colony-assay as CFU-E. Surprisingly,
these cells did not express platelet-peroxidase-like
activity previously described in presumptive CFU-E.27,31
However, after a few hours in culture, this activity was
recovered (data not shown), suggesting its possible inactivation
by the purification procedure. Cell-surface GPA and
images of rhopheocytosis were seen on the other 1-2 erythroid
cells, but at a much lower density than what has been
described for proerythr~blasts.~~.~~ These cells, or “preproerythroblasts,”
therefore appear to represent an intermediate
stage between CFU-E and proerythroblasts.
Human CFU-E have been purified only indire~tly~’.~~~~~
through the proliferation of partially purified blood or
marrow BFU-E cultured 7 to 9 days either in liquid
su~pension,~~ or in semisolid assay” in the presence of Ep.
This approach also yielded 80% pure populations of very
immature erythroblasts containing up to 50% CFU-E. However,
in contrast to cells directly purified from marrow by
avidin-biotin immune rosetting, which appear homogeneous
with respect to their stage of maturation, BFU-E-derived
cells are more heterogeneous and include many hemoglobin-
ized proerythrobla~ts.~~ CFU-E and early erythroblasts have
also been purified from either thiamphenicol-treated,36 or
Friend virus-infected mice spleen cells,37 but it is difficult to
compare them with human cells.
We have previously shown that the primitive BFU-E,
which give rise to multiclustered colonies, do not attach to
Fn, whereas 30% of CFU-E, which represents the most
mature type of progenitor, express this property.” In this
study, we assessed the attachment to Fn of purified CFU-E,
preproerythroblasts, and their progeny, generated in cultures
that support their terminal erythroid differentiation. Interestingly,
the majority of CFU-E recovered in the purified
population selected for these studies showed a higher proportion
of Fn-adhering members. Although a direct comparison
of Fn-adherence for purified and unpurified CFU-E was not
undertaken, it is possible that cells defined as CFU-E by
colony-assays are heterogeneous with respect to their Fnadherence
properties, and that our purification procedure
isolated a subset of CFU-E with higher Fn-adherence potential.
Alternatively, it is possible that the adhesion assay may
give different absolute results according to the composition of
the population being tested. What the present studies have
revealed is that adherence to Fn also clearly changes during
the later stages of erythroid cell differentiation: (1) CFU-E
attached to Fn in the highest proportion, close to 60% (Fig 4),
(2) the proportion of adherent preproerythroblasts was
consistently half that observed with the copurified CFU-E,
and (3) enucleating erythroblasts and reticulocytes derived
from these immature erythroblasts were unable to attach to
Fn. These data demonstrate that the Fn-adhesion property is
lost in differentiating erythroblasts in two sequential stages:
an initial decline during the transition from CFU-E to
preproerythroblasts, with a subsequent loss of all remaining
adhesive properties at the time of enucleation.
Expression of Fn-adhesion during human erythroid differentiation
appears to be strikingly correlated to the migratory
capacity of these cells. In semisolid culture media, all BFU-E
exhibit an extensive migratory capacity, which is abruptly
lost at the CFU-E level.38 In vivo, marrow cell egress occurs
at the reticulocyte stage, and young reticulocytes are motile,
whereas very few CFU-E circulate. It is thus tempting to
speculate that these changes in erythroid cell motility reflect
changes in cell attachment to Fn and possibly other adhesive
glycoproteins of the marrow ECM. This is strengthened by
the recent observation that glycoprotein IV, an 88 Kd
glycoprotein identified as a thrombospondin re~eptor,’~ is
present on mature BFU-E, CFU-E and erythroblasts, and is
similarly lost during terminal erythroid differentiati~n.’~
Loss of attachment to Fn with terminal differentiation has
been previously observed in DMSO-induced murine erythroleukemic
cells,” and hemin-induced human K 562 cells,”
and ascribed to the loss of the 140 Kd Fn receptor molecule
identified as VLA-5 on uninduced K562.12,40*41 Results obtained
with human reticulocytes isolated from patients with
hemolytic anemia or recently splenectomized have shown
that these cells can attach to Fn?’ which appears to contradict
the results presented in this study. However, reticulocytes
produced in response to hemolysis are not identical to
those produced during steady-state erythropoie~is,”~ and
their adhesion properties might be altered by the underlying
disorder.” Furthermore, the marked susceptibility of cell
adhesion to the conditions in which the functional attachment
assay is performed might explain quantitatively different
results.
Both the anti-a5/31 polyclonal IgG and the anti-d MoAb
inhibited CFU-E and preproerythroblast adhesion to Fn.
This strongly suggests that VLA-5 (a5pl) functions as an Fn
receptor on normal human erythroblastic cells. The Fn
receptor was undetectable from reticulocytes by indirect
immunofluorescence. However, changes in the expression of
the Fn receptor during in vitro erythroblastic differentiation
could not be analyzed in more detail by immunofluorescence:
the polyclonal antibody recognizes not only VLA-5, but all
integrins sharing the /3l subunit, and some of these molecules
may be expressed by erythroblasts, and the B1E5 MoAb is
not suitable for fluorescence labeling (C. Damsky, personal
communication, January 1989).
The above results demonstrate that adhesion to Fn is
precisely regulated during normal human erythroid differentiation
and may influence cell migration in the bone marrow.
A challenging question will be to determine if stage-specific
expression of integrins will influence cell behavior by triggering
early events associated with erythroid cell proliferation
and differentiati~n.~’.~~
ACKNOWLEDGMENT
We gratefully acknowledge the help of the surgeons who provided
marrow samples: Drs Brunet, Missenard, and Lapresle of Clinique
Arago and Drs Bonnet, Judet, and Koechlin of HBpital de la
Croix-Rouge and their nursing staff. We thank Dr A. Eaves (Terry
Fox Laboratory, Vancouver, Canada) for the gift of erythropoietin;
Drs L.C. Anderson, R. Berthier, C. Damsky, L. Edelmann, J.P.
Fermand, and F. Tetteroo for their generous gifts of antibodies; C.
Eaves, D. Baruch, and M. Rosemblatt for helpful discussions and
critical reading of the manuscript, and M. Laroche for excellent
secretarial assistance.

VUILLET-GAUGLER ET AL
1. Edelman GM: Morphoregulatory molecules. Biochem 27:
3533,1988
2. Li ML, Aggeler J, Farson A, Hatier C, Hassel J, Bissell M:
Influence of a reconstituted basement membrane and its components
on casein gene expression and secretion in mouse mammary epithelial
cells. Proc Natl Acad Sci USA 84:136, 1987
3. Dexter TM, Allen TD, Lajtha LG: Conditions controlling the
proliferation of haemopoietic stem cells in vitro. J Cell Physiol
91:335, 1977
4. Lichtman MA: The ultrastructure of the hematopoietic environment
of the marrow. Exp Hematol9:391, 1981
5. Dexter TM, Spncer E, Toksoz D, Lajtha LG: The role of
cells and their products in the regulation of in vitro stem cell
proliferation and granulocyte development. J Supramol Struct
13513, 1980
6. Coulombel L, Eaves C, Eaves A: Enzymatic treatment of long
term human marrow cultures reveals the preferential location of
primitive hemopoietic progenitors in the adherent layer. Blood
62:291, 1983
7. Tsai S, Emerson SG, Sieff C: Isolation of a human stromal cell
strain secreting hemopoietic growth factors. J Cell Physiol 127:137,
1986
8. Zsebo KM, Yuschenkoff VN, Schiffer S, Chang D, McWall E,
Dinarello C, Brolun MA, Altrock B, Bagby GC: Vascular endothelial
cells and granulopoiesis: Interleukin-1 stimulates release of
G-CSF and GM-CSF. Blood 71:99,1988
9. Rettenmeier CN, Roussel MF Differential processing of
colony-stimulating factor 1 precursors encoded by two human
cDNAs. Mol Cell Biol8:5026, 1988
10. Gordon MY, Graham PR, Watt SM, Greaves MF: Compartmentalization
of a haematopoietic growth factor (GM-CSF) by
glycosaminoglycans in the bone marrow microenvironment. Nature
326:403, 1987
11. Ruoslahti E, Pierschbacher M: New perspectives in cell
adhesion: RGD and integrins. Science 238:491, 1987
12. Hemler ME, Huang C, Schwarz L: The VLA protein family.
Characterization of five distinct cell surface heterodimers each with
a common 130,000 molecular weight @ subunit. J Biol Chem
262:3300,1987
13. Campbell A, Wicha MS, Long M: Extracellular matrix
promotes the growth and differentiation of murine hematopoietic
cells in vitro. J Clin Invest 75:2085, 1985
14. Spooncer E, Gallagher JT, Krizsa F, Dexter TM: Regulation
of haemopoiesis in long-term bone marrow cultures. IV: Glycosaminoglycan
synthesis and the stimulation of haematopoiesis by @-
xylosides. J Cell Biol96:510, 1983
15. Coulombel L, Vuillet MH, Leroy C, Tchernia G: Lineageand
stage-specific adhesion of human hematopoietic cells to extracellular
matrices from marrow fibroblasts. Blood 71:329, 1988
16. Tsai S, Patel V, Beaumont E, Lodish HF, Nathan DG, Sieff
CA: Differential binding of erythroid and myeloid progenitors to
fibroblasts and fibronectin. Blood 69:1587, 1987
17. Campbell A, Long MW, Wicha MS: Haemonectin, a bone
marrow adhesion protein specific for cells of the granulocytic
lineage. Nature 329:744, 1987
18. Patel VP, Lodish HF: Loss of adhesion of murine erythroleukemia
cells to fibronectin during erythroid differentiation. Science
224:996, 1984
19. Virtanen I, Ylanne J, Vartio T: Human erythroleukemia cells
adhere to fibronectin: Evidence for a Mr 190000 receptor protein.
Blood 69:578,1987
20. Mouchiroud G, Berthier R, Newton IA, Chapel A, Hollard
D: Monoclonal antibodies against human hematopoietic cells and
REFERENCES
the separation of progenitor cells from bone marrow. Exp Hematol
13:566, 1985
21. Edelman P, Vinci G, Villeval JL, Vainchenker W, Henri A,
Miglierina R, Rouger R, Reviron J, Breton-Gorius J, Sureau C,
Edelman L: A monoclonal antibody against an erythrocyte ontogenic
antigen identifies fetal and adult erythroid progenitors. Blood
67:56, 1986
22. Tetteroo PAT, Lansdorp PM, Leeksma OC, Von Dem Borne
AEG: Monoclonal antibodies against human platelet glycoprotein
IIIa. Br J Haematol55:509, 1983
23. Anderson LC, Gahmberg CG, Teerenhovi L, Vuopio P
Glycophorin A as a cell surface marker of early erythroid differentiation
in acute leukemia. Int J Cancer 23:717, 1979
24. Argraves WS, Suzuki S, Arai H, Thompson K, Pierschbacher
M, Ruoslahti E: Amino acid sequence of the human fibronectin
receptor. J Cell Biol 105:1183, 1987
25. Werb Z, Tremble P, Behrendtsen 0, Crowley E, Damsky C:
Signal transduction through the fibronectin receptor induces collagenase
and stromelysin gene expression. J Cell Biol 109377, 1989
26. Furfang J, Thierfelder S: Effective avidin coating of erythrocytes
by disulfide bound formation. J Immunol Methods 91:123,
1986
27. Breton-Gorius J, Van Haeke D, Pryzwansky KB, Guichard J,
Tabilio A, Vainchenker W, Carmel R: Simultaneous detection of
membrane markers with monoclonal antibodies and peroxidatic
activities in leukaemia: Ultrastructural analysis using a new method
of fixation preserving the platelet peroxidase. Br J Haematol58:447,
1984
28. Graham RC, Karnovsky MJ: The early stages of absorption
of injected horseradish peroxidase in proximal tubules of mouse
kidney. Ultrastructural cytochemistry by a new technique. J Histochem
Cytochem 14:291, 1966
29. Kieffer N, Bettaieb A, Legrand C, Vainchenker W, Breton-
Gorius J, Edelman L: Immunochemical identification of the 88 Kd
platelet glycoprotein IV on human erythroid progenitor cells. Blood
70:156a, 1987 (abstr)
30. Greenberg PL, Baker S, Link M, Minowada J: Immunologic
selection of hematopoietic precursor cells utilizing antibodymediated
plate binding (“panning”). Blood 65:190, 1985
31. Breton-Gorius J, Villeval JL, Mitjavila MT, Vinci G, Guichard
J, Rochant H, Flandrin G, Vainchenker W: Ultrastructural
and cytochemical characterization of blasts from early erythroblastic
leukemias. Leukemia 1:173, 1987
32. Loken MR, Shah VO, Dattilio KL, Civin CI: Flow cytometry
analysis of human bone marrow: I. Normal erythroid development.
Blood 69:255,1987
33. Breton-Gorius J, Reyes F Ultrastructure of human bone
marrow cell maturation. Int Rev Cytol46:251, 1976
34. Fibach E, Manor D, Oppenheim A, Rachilewitch EA: Proliferation
and maturation of human erythroid progenitors in liquid
culture. Blood 73:100, 1989
35. Sawada K, Krantz SB, Kans JS, Dessypris EN, Sawyer S,
Click AD, Civin CI: Purification of human erythroid colony-forming
units and demonstration of specific binding of erythropoietin. J Clin
Invest 80357,1987
36. Nijhof W, Wierenga PK Isolation and characterization of
the erythroid progenitor cells CFU-E. J Cell Biol96:386, 1983
37. Koury MJ, Bondurant MC, Rana SS: Changes in erythroid
membrane proteins during erythropoietin-mediated terminal differentiation.
J Cell Physiol 133:438, 1987
38. Eaves AC, Eaves CJ: Erythropoiesis in culture, in McCulloch
EA (ed): Cell Culture Techniques. Clinics in Haematology, vol 13.
Philadelphia, PA, WB Saunders, 1981, p 371

ERYTHROBLAST ADHESION TO FIBRONECTIN
a73
39. Asch AS, Barnwell J, Silverstein RL, Nachman RL: Isolation
of the thrombospondin membrane receptor. J Clin Invest 79:1054,
1987
40. Takada Y, Huang C, Hemler M: Fibronectin receptor structures
in the VLA family of heterodimers. Nature 326:607, 1987
41. Patel VP, Lodish H F The fibronectin receptor on mammalian
erythroid precursor cells: Characterization and developmental regulation.
J Cell Biol 102:449, 1986
42. Patel VP, Ciechanover A, Platt 0, Lodish HF: Mammalian
reticulocytes lose adhesion to fibronectin during maturation to
erythrocytes. Proc Natl Acad Sci USA 82:440, 1985
43. Coulombel L, Tchernia G, Mohandas N: Human reticulocyte
maturation and its relevance to erythropoietic stress. J Lab Clin Med
94:467, 1979
44. Mohandas N, Evans E: Sickle erythrocyte adherence to
vascular endothelium: Morphologic correlates and the requirement
for divalent cations and collagen binding plasma proteins. J Clin
Invest 76:1605, 1985
45. Dike LE, Farmer SR: Cell adhesion induces expression of
growth-associated genes in suspension-arrested fibroblasts. Proc
Natl Acad Sci USA 85:6792,1988
46. Patel VP, Lodish HF A fibronectin matrix is required for
differentiation of murine erythroleukemia cells into reticulocytes. J
Cell Biol 105:3105, 1987