1. Ingeniería

From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
Myeloid and Erythroid Progenitor Cells From Normal Bone Marrow Adhere To
Collagen Type I
By Michael Koenigsmann, James D. Griffin, Jennifer DiCarlo, and Stephen A. Cannistra
One of the mechanisms by which normal hematopoietic
progenitor cells remain localized within the bone marrow
microenvironment is likely t o involve adhesion of these cells
t o extracellular matrix (ECM) proteins. For example, there is
evidence that uncommitted, HLA-DR-negative progenitor
cells and committed erythroid precursors (BFU-E) bind t o
fibronectin. However, fibronectin is not known t o mediate
binding of committed myeloid (granulocyte-macrophage)
progenitors, raising the possibility that other ECM proteins
may be involved in this process. We investigated the binding
of the M 0 7 myeloid cell line t o a variety of ECM proteins and
observed significant specific binding t o collagen type I
(56% 2 5%). minimal binding t o fibronectin (18% k 4%) or
t o laminin (19% 2 5%). and no binding to collagen type 111, IV,
or V. Similarly, normal bone marrow myeloid progenitor cells
(CFU-GM) demonstrated significant specific binding t o collagen type I(46% f 8% and 47% f 12% for day 7 CFU-GM and
day 14 CFU-GM, respectively). The ability of collagen t o
mediate bidding of progenitor cells was not restricted t o the
myeloid lineage, as BFU-E also showed significant binding t o
this ECM protein (40% f 10%). The binding of M 0 7 cells and
CFU-GM was collagen-mediated, as demonstrated by complete inhibition of adherence after treatment with collagenase type VII, which was shown t o specifically degrade
collagen. Binding was not affected by anti-CD29 neutralizing
antibody (anti-p-I integrin), the RGD-containing peptide
sequence GRGDTP, or divalent cation chelation, suggesting
that collagen binding is not mediated by the p-1integrin class
of adhesion proteins. Finally, mature peripheral blood neutrophils and monocytes were also found t o bind t o collagen type
I (25% f 8% and 29% 2 6%, respectively). These data suggest that collagen type Imay play a role in the localization of
committed myeloid and erythroid progenitors within the
bone marrow microenvironment.
o 1992by The American Society of Hematology.
H
providing a potential mechanism for the release of mature
red blood cells (RBCs) from the bone marrow into the
peripheral
Thus, both the pattern of progenitor
cell binding to fibronectin as well as the recognition sites for
binding to this molecule change as the progenitor cell
acquires a more committed phenotype.
Although fibronectin may partly mediate the binding of
uncommitted, HLA-DR-negative progenitors and committed erythroid precursors, it is likely that this protein does
not play a major role in the localization of committed
myeloid progenitors such as CFU-GM within the bone
marrow. However, other ECM proteins in the bone marrow
that could be involved in the binding of committed myeloid
progenitors include hemonectin, collagen types I, 111, and
IV, and laminin. Hemonectin is a 60-Kd protein purified
from rabbit ECM that has been observed to support the
binding of murine CFU-GM and, to a lesser extent,
erythroid burst forming units (BFU-E) in vitro.8 Myeloid
binding to hemonectin appears to be differentiationassociated, with more mature cells such as bands and
polymorphonuclear leukocytes exhibiting decreased adhe-
EMATOPOIESIS OCCURS through a series of orderly steps involving proliferation, commitment, and
differentiation of early bone marrow progenitor cells, ultimately leading to the release of mature cells into the
peripheral blood. To ensure that hematopoietic progenitors
are provided with appropriate conditions for development,
including the availability of humoral growth factors, it is
likely that specific mechanisms exist for the adhesion of
these cells to bone marrow stromal cells or to extracellular
matrix (ECM) proteins during maturation. For example,
early murine progenitors with high proliferative capacity
are preferentially located in the adherent cell layer of
long-term marrow cultures.’ In human studies, uncommitted, HLA-DR-negative progenitors capable of repopulating long-term marrow cultures have been shown to avidly
bind to the stromal cell layer, whereas more committed
progenitors exhibit significantly reduced binding.’ Although
the mechanism for the adhesion of early progenitors to the
stromal cell layer is not fully understood, there is increasing
evidence that the ECM protein fibronectin plays an important role in this process. For example, uncommitted, HLADR-negative progenitors selectively bind in vitro to the
heparin binding domain I1 of fibronectin, an epitope that is
recognized by the VLA-4 molecule of the p-1 integrin
family.”4 Interestingly, as early progenitors become more
committed, as indicated by expression of HLA-DR, their
binding to fibronectin is altered in two major ways. First,
binding of committed progenitors appears to be mediated
through an additional region of the fibronectin molecule
termed the cell binding domain, an area recognized by the
VLA-5 integrin. Second, only the erythroid subset of
committed progenitors appears to maintain significant binding avidity for fibronectin, with other committed progenitors such as granulocyte-macrophage colony-forming units
(CFU-GM) demonstrating significantly reduced binding to
this ECM p r ~ t e i nThe
. ~ binding of erythroid progenitors to
fibronectin is differentiation-dependent, with progressive
loss of binding as the cells mature into erythrocytes, thereby
Blood, Vol79, No 3 (February I),
1992: pp 657-665
From the Divisions of Tumor Immunology and Medical Oncology,
Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA.
Submitted June 13, 1991; accepted September 30, 1991.
Supported in part by Public Health Service Grant Nos. CA 36167,
CA 42802, and CA 34183, Biomedical Research Support Grant No.
SO7RRO5526-26, and by the US Cancer Research Council. M.K. is a
grant recipient of the Deutsche Forschungsgemeinschap. J.D.G. is a
scholar of the Leukemia Society ofAmerica.
Address reprints to Stephen A . Cannistra, MD, Division of Tumor
Immunology, Dana-Farber Cancer Institute, 44 Binney St, Boston,
MA 02115.
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.
0 1992 by The American Society of Hematology.
0006-4971I9217903-0005$3.00/0
657
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
KOENIGSMANN ET AL
658
sion to this molecule when compared with undifferentiated
blasts. Although the molecular structure of hemonectin and
its relationship to other adhesion proteins are unknown,
the ability of this protein to support the binding of CFU-GM
in a differentiation-associated fashion deserves further
investigation.
The role of collagen in the localization and development
of progenitor cells within the bone marrow is of interest for
several reasons. Collagen types I, 111, and IV are major
constituents of the bone marrow ECM,'.'' and hematopoietic precursor cells express molecules such as CD44I3that
are known to recognize ~ o l l a g e n . Furthermore,
~~~'~
inhibition of collagen synthesis by cis-4-hydroxyprolineinterferes
with the generation of a functional stromal microenvironment in vitro.16 In view of these considerations, the goal of
the present study was to assess the binding of hematopoietic cells to a variety of ECM proteins to determine possible
mechanisms by which committed myeloid progenitor cells
remain localized within the bone marrow during development. We report that collagen type I supports significant
binding of normal myeloid and erythroid progenitor cells,
and that this binding is not mediated by the p-1 integrin
family of adhesion proteins; furthermore, mature myeloid
cells also bind to collagen type I. The implications of these
results for the mechanism of progenitor cell localization to
the bone marrow are discussed.
MATERIALS AND METHODS
Reagents. Purified ECM proteins were obtained as follows:
collagen type I (bovine), collagen type IV (murine), and laminin
(murine) from Collaborative Research (Bedford, MA); collagens
type I11 (human) and V (human) from Sigma (St Louis, MO)
(Sigma designations X and IX); fibronectin (human) from Telios
(San Diego, CA). Recombinant human granulocyte-macrophage
colony-stimulating factor (GM-CSF), interleukin-3 (IL-3), and
erythropoietin were kindly provided by Dr Steven Clark (Genetics
Institute, Cambridge, MA). The following murine monoclonal
antibodies that recognize adhesion proteins were used as part of
this study: 4B4 (anti-CD29, pl integrin), 8F2 (anti-VLA-4), 2H6
(anti-VLA-5), and IF7 (anti-CD26) were gifts of Dr Chikao
Morimoto (Dana Farber Cancer Institute [DFCI], Boston, MA);
TS2/7 (anti-VLA-1) was a gift of Dr Martin Hemler (DFCI); 12F1
(anti-VLA-2) was a gift of Dr Virgil Woods (University of San
Diego, San Diego, CA); 5143 (anti-VLA-3) was a gift of Dr
Anthony Albino (Memorial Sloan Kettering Institute, New York,
NY); GoH3 (anti-VLA-6) was a gift of Dr Arnoud Sonnenberg
(University of Amsterdam, Amsterdam, The Netherlands); and
515 (anti-CD44) was a gift of Dr Goeffrey Kansas (DFCI). Fetal
calf serum (FCS) for colony cultures was purchased from HyClone
(Logan, UT). Materials used for serum-free cell culture were
obtained as follows: crystalline bovine serum albumin (BSA)
(globulin-free), bovine insulin, and cholesterol from Sigma; human
iron-saturated transferrin from Boehringer-Mannheim (Indianapolis, IN); Dulbecco's minimal essential medium (DMEM) from
Whittacker (Walkersville, MD). Serum-free medium (SFM) consisted of DMEM containing BSA 15 mg/mL, insulin 1 p,g/mL,
transferrin 7.7 x
mol/L, cholesterol 7.8 & n L . Highly
purified collagenase type VI1 from Clostridium histolyticum was
obtained from Sigma. The synthetic peptides GRGESP and
GRGDTP were from Telios. "Chromium (1 mCi/mL, 200 Ci/g)
and tritiated thymidine (1 mCi/mL, 2 Ci/mmol) were purchased
from New England Nuclear (Boston, MA).
Source of cells. The GM-CSF- and IL-34ependent cell line,
M07, was obtained from Dr Steven Clark, and was originally
derived by Avanzi et all7 from the peripheral blood of an infant
with acute megakaryocytic leukemia. This line was maintained in
DMEM supplemented with 20% FCS (20% DMEM) and 10
ng/mL of both GM-CSF and IL-3. Bone marrow progenitor cells
were obtained from healthy donors after informed consent by
posterior iliac crest aspiration into heparinized syringes. Mononuclear cells were recovered by Ficoll density centrifugation, and
macrophages were removed by plastic adherence for 1 hour at 37°C
as previously described." Neutrophils and monocytes were simultaneously prepared from heparinized peripheral blood donated by
healthy volunteers. After Ficoll density centrifugation, neutrophils
were isolated from the RBC pellets by dextran sedimentation.
Monocytes were obtained from the mononuclear cell fraction by
plastic adherence for 1 hour at 37T, as previously described.Ix
Both neutrophil and monocyte preparations were greater than
90% pure as assessed by Wright-Giemsa staining.
Immunophenotyping. M 0 7 cells were screened for the expression of known collagen adhesion receptors and for the myeloid
marker CD33 by indirect immunofluorescent staining. Briefly, 1 x
lo6cells were stained for 30 minutes, 4°C with 100 FL of either an
irrelevant control antibody (3CllC8, anti-y-interferon antibody),
or with the previously described antibodies that recognize CD29,
VLA-1, VLA-2, VLA-3, VLA-4, VLA-5, VLA-6, CD44, or 1F7.
After washing two times in phosphate-buffered saline (PBS), the
cells were labeled with fluorescein-conjugated goat antimouse Ig
(FITC) (Tago, Burlingame, CA) for 30 minutes, 4°C. After two
additional washes, the cells were analyzed on a Coulter Epics C
flow cytometer (Coulter, Hialeah, FL).
SIChromiumassay for cell adhesion. The binding of M 0 7 cells,
neutrophils, and monocytes to ECM proteins was quantitated by
assessing "chromium release of adherent cells. Cells, lo6, were
labeled with 100 KCi of "chromium (200 Ci/g) for 1 hour at 37T,
and then washed two times in PBS and resuspended at 0.5 to 1 x
106/mL of SFM. Flat-bottom 96-well microtiter wells (Falcon,
Oxnard, CA) were coated with 10 to 20 p,L of individual ECM
proteins (20 to 70 p,g/well, 0.28 cm2/well)and allowed to air dry for
1 to 3 hours at room temperature. Control coating was performed
with either the respective solvents (0.1N acetic acid for collagen
preparations, Tris/NaCI for laminin, H,O for fibronectin) or with
1.5% BSA. BSA was used as a control to ensure that cells were not
adhering to protein in a nonspecific fashion in these experiments.
None of the cell types used in this study was observed to bind to
BSA. "Chromium-labeled cells (50 to 100 x lo3) were added to
each well in a total of 100 p,L of SFM, and binding was allowed to
occur for 2 to 4 hours at 37°C. The nonadherent cells were then
removed by three washes with PBS, followed by lysis of bound cells
with 0.1% NP40. The radioactivity of each lysate was measured in a
gamma counter. The percentage of cells adhering to an individual
ECM protein (percent specific binding) was calculated as follows:
percent specific binding = 100 x (mean cpm [ECM-coated
surface] - mean cpm [control surface])/CPM (total).
For some experiments, ''chromium-labeled cells were pretreated
with anti-CD29 antibody (1:200 dilution) or control antibody
(anti-CD33, which reacts with M 0 7 cells) for 30 minutes at 4°C
before performing the bind, with continued presence of antibody
during the duration of the bind. Likewise, the effects of RGDcontaining peptides was determined by preincubation of cells in
200 p,g/mL of either GRGESP or GRGDTP for 30 minutes at
room temperature, followed by binding in the continued presence
of peptide. The effects of divalent cation chelation were assessed
by performing the bind in EDTA (10 mmol/L final concentration
per well) in some studies. Finally, collagen-coated wells were
pretreated with collagenase in some experiments to determine the
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
PROGENl'iOR CELL BINDING TO COLLAGEN TYPE I
specificityof binding to collagen. For these studies, collagen-coated
or control wells were pretreated with 20 pL of either collagenase
buffer (50 nlthol/L Tris, 2.5 mmol/L CaCI,, pH 7.5) or collagenase
(0.5 U/mL) at room temperature and allowed to air dry before
adding 5'chfomium-labeled cells.
Assay for pkogenitor cell adhesion. Nonadherent bone marrow
mononuclear cells from healthy donors were added to 96-well
plates coated with either ECM protein, control solvent, or with
1.5% BSA as described. To ensure that a reasonable number of
colonies Would be obtained per well after binding, cell numbers
were added in a range of 10 to 100 x 10' per well. Progenitor cells
were allolell to bind for 4 hours at 3 7 T , followed by removal of
nonadherdnt cells by washing three times in PBS. For determination of CPb-GM binding, each well was overlayed with 0.3% agar
in Iscove'S modified Dulbecco's MEM (IMDMEM) containing
10% (vol/vol) 5637 conditioned medium as a source of CSFs.
Myeloid progenitor cells were scored using an inverted microscope.
Clusters (8 to 40 cells) were counted on day 7 and colonies ( > 4 0
cells) were counted on day 14. For determination of CFU-mix and
BFU-E binding, each well was overlayed with 0.9% methylcellulose
in IMDMEM containing 4 U/mL erythropoietin, and 5 ng/mL of
both GM-CSF and IL-3. CFU-mix and BFU-E were enumerated
on day 14. The total number of CFUs added per well was
determined under identical growth conditions in collagen- or
control-coated plastic wells. It should be noted that there was no
difference in the growth of CFU-GM in the presence or absence of
collagen in any of the experiments performed in this study. The
percentage of CFUs adherent to ECM (percent specific binding)
was calculated as follows: percent specific binding = 100 x (mean
CFU [ECM-coated surface] - mean CFU [control surface])/total
CFU. To establish consistency between experiments, all data were
corrected to reflect cluster or colony number per 10s cells added.
For some experiments, the effects of control antibody (anti-CD33),
anti-CD29, EDTA, and collagenase on progenitor cell binding to
collagen type I were assessed under the same incubation conditions
as described for 5'chromium-labeled cells. Specifically, bone marrow mononuclear cells were first pretreated with either control or
anti-CD29 antibody, with continuous exposure to antibody during
the binding period.
Cell proliferation assay. The effect of ECM proteins on the
proliferation of M 0 7 cells was examined in 96-well plates coated
with either ECM protein or control solvent. M 0 7 cells, 20 x 10'
per well, were incubated for 60 hours at 37°C in the presence or
absence of 5 to 25 ng/mL of GM-CSF in either 20% FCS or in
SFM. Tritiated thymidine (1.6 $3, 2 Ci/mmol) was added to each
well for the last 12 hours of incubation, followed by cell harvesting
and determination of radioactivity with a beta counter (Packard
Liquid Scintillation Analyzer 2000 CA).
Assessment of collagenase activity. To determine the specificity
of collagenase VI1 for collagen type I, 7 pL of either laminin,
fibronectin, or collagen (approximately 20 pg of each protein) were
treated with 43 pL of either control buffer (50 mmol/L Tris, 2.5
mmol/L CaCI,, pH 7.5) or with collagenase type VI1 (approximately 430 m u ) in a total volume of 50 pL at room temperature for
1 hour. In some experiments, 7 pL of collagen was treated with 43
pL of neutrophil supernatant (86% vol/vol) under the same
conditions to test for the possibility of endogenous collagenase
production by neutrophils. After addition of 50 pL of 2X sample
buffer with mercaptoethanol, the samples were boiled at 100°C for
5 minutes and loaded onto a 7.5% sodium dodecyl sulfate (SDS)
polyacrylamide gel. After electrophoresis, the protein bands were
visualized by Coomassie blue staining.
Statistical analysis. Significance levels for comparison between
treatment groups were determined using the two-sided Student's ttest for paired samples.
659
RESULTS
Adhesion receptor expression andpattem of ECM binding of
M 0 7 cells. Initial binding experiments were performed
using the factor-dependent myeloid leukemic cell line,
M07. Immunophenotyping analysis was initially performed
to determine whether these cells expressed receptors for
collagen (VLA-1, -2, and -3, CD44, and CD26), fibronectin
(VLA -3, and -4), and laminin (VLA-1, -2, and -6).19 The
percentage of M 0 7 cells specifically positive for each
adhesion protein was: VLA-1 = 0%, VLA-2 = 41%,
VLA-3 = 0%, VLA-4 = 48%, VLA-5 = 94%, VLAd =
64%, CD29 (VLA-p1-chain) = 95%, CD44 = 89%, and
CD26 = 0%. The cells also expressed the CD33 molecule
(90%). These data suggest that M 0 7 cells express adhesion
proteins for collagen, fibronectin, and laminin. Therefore,
5'chromium-labeled M 0 7 myeloid cells were allowed to
bind to microtiter wells previously coated with 20 to 70 kg
of either collagen types I, 111, IV, and V, fibronectin, or
laminin for 4 hours at 37"C, with subsequent washing and
measurement of specifically bound cells as previously described. As shown in Table l, M 0 7 cells demonstrated
significant binding to bovine collagen type I (56% +- 5%),
with a significantly lesser amount of binding to human
fibronectin (18% c 4 % ) and to murine laminin
(19% 2 5%). There was no appreciable binding of M 0 7
cells to collagen type 111, IV, or V. Increasing the absolute
amounts of any coated ECM protein did not increase
binding further. The kinetics of binding of M 0 7 cells to
collagen type I showed peak binding at 2 hours of incubation.
Adhesion of normal bone marrow progenitor cells to collagen type I. In view of the significant degree of M 0 7
binding to collagen type I, the remainder of our studies
were designed to assess the relevance of this observation for
normal bone marrow progenitor cells. Adherence-depleted
bone marrow mononuclear cells were allowed to bind to
collagen type I-coated microtiter wells for 4 hours at 37"C,
followed by washing and overlaying the bound cells with
either 0.3% agar or 0.9% methycellulose under conditions
Table 1. Adhesion of the Myeloid Leukemic Cell Line M 0 7 to ECM
Proteins
Range
Specific Binding
(%)t
ECM Protein*
Collagen type I
Collagen type 111
Collagen type IV
Collagen type V
Laminin
Fibronectin
~
~
(%)
56 f 5
48-63
2 2 2
0
0
19 f 5
0-2
~
5
3
3
3
5
5
-
14-28
14-21
18 2 4
~
No. of
Experiments*
~
~~
~
~
*Microtiter wells were coated with either control solvent or extracellular matrix proteins as described in text before performing bind with
50 x lo35'chromium-labeled M07 cells for 4 hours at 37°C.
tPercent specific binding of cells adhering to each ECM protein was
calculated for each experiment using mean values from quadruplicate
wells as described in text. Background binding was 3% ? 2% for all
experiments. The value shown is mean IT SD percent specific binding
from the indicated number of experiments.
*Indicates number of separate experiments performed for each ECM
protein.
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
660
KOENIGSMANN ET AL
that would support maximum colony growth of CFU-GM,
CFU-GEMM, and BFU-E as described. Bone marrow
progenitors were depleted of accessory cells to reduce the
potential for simultaneous accessory cell binding that might
alter the pattern of colony formation due to the secretion of
endogenous growth factors. As shown in Table 2, significant
binding of day 7 CFU-GM (46% f 8%, n = 6), day 14
CFU-GM (47% f 12%, n = 4), and BFU-E (40% f lo%,
n = 5) to collagen type I was observed (P < .001). Interestingly, there were no quantitative differences between the
binding of early (ie, day 14 CFU-GM or CFU-GEMM)
versus later (day 7 CFU-GM) progenitor cells to collagen
type 1.Although CFU-mix tended to bind to collagen type I,
this effect was not statistically significant, possibly due to
the small numbers of colonies observed.
SpeciJcity of myeloid cell binding to collagen I. To confirm that the observed binding of M 0 7 cells or normal bone
marrow progenitors was mediated by collagen type I, as
opposed to a contaminating protein present in the collagen
preparation, we studied the effects of pretreatment of
collagen-coated microtiter wells with highly purified collagenase type VII. As shown in Fig 1, collagenase pretreatment significantly reduced the binding of both M 0 7 cells
and CFU-GM to collagen type I, whereas similar pretreatment with collagenase buffer had no effect. To determine
whether the collagenase effect was specific for collagen, we
treated fibronectin, laminin, and collagen type I with
collagenase for 1 hour at room temperature, with subsequent visualization of the protein products by polyacrylamide gel electrophoresis (PAGE) followed by Coomassie
blue staining. As show in Fig 2, collagenase had no effect on
the integrity of laminin or fibronectin, whereas it completely degraded the three major bands associated with
collagen type I(115,125, and 212 Kd).
Effects of anti-CD29 antibody, RGD-containing peptide, or
divalent cation chelation on the binding of myeloid cells to
collagen type I. The 6-1 integrin family of adhesion proteins is comprised of heterodimers containing a common
Table 2. Adhesion of Normal Myeloid Progenitor Cells to Collagen Type I
Donor*
Treatmentt
CFU-GM D7
1
Control
Collagen
Total CFU added
% Specific binding
Control
Collagen
Total CFU added
% Specific binding
Cont roI
Collagen
Total CFU added
% Specific binding
Control
Co IIage n
Total CFU added
YOSpecific binding
Cont ro I
Collagen
Total CFU added
% Specific binding
Control
Collagen
Total CFU added
O h Specific binding
Control
Collagen
Total CFU added
O h Specific binding
7*
7I
57 6
158 + 32
32
4 2 1
22 -c 6
31
58
1 2 1
8+2
12
67
020
020
4+2
9 2 4
44
1 2 0
2+1
50
020
020
OtO
o+o
49 + 10
109
45
422
25 + 4
44 + 6
48
9+3
22 2 1
41
1 2 0
7+2
1426
43
7+3
25 4
28
+
2 f 1
20
10
o+o
o+o
9 r 4
22 + a
41
221
422
50
2
3
7
Mean % specific binding11
+
CFU-GM D14
ND§
BFU-E
Or0
5+2
16 + 4
31
CFU-mix
ND
020
22 f 5
44
50
ND
ND
ND
222
74 + 6
167 + 19
43
46 + 8
17 + 4
46 2 a
37
47 + 12
ND
ND
020
o+o
4+2
7+4
57
1+0
3+ 1
33
ND
ND
o+o
40
2
10
3 4 + 17
"Bone marrow progenitor cells from seven different normal donors were depleted of monocytes by plastic adherence before performing binding
assay.
tCells (10-100 x 10') were added t o flat bottom 96-well plates precoated with either 0.1% acetic acid (control) or collagen type I (60 pg). After 4
hours at 37"C, the wells were washed to remove nonadherent cells, followed by addition of either agar or methylcellulose as described in text.
Appropriate controls were performed in order to assess the total number of CFU-GM and BFU-E added to each well.
*Data expressed as mean + SD of colony number from quadruplicate wells. To establish consistency between experiments, all data are expressed
as cluster (day 7) or colony (day 14) number per 1O'totaI cells added.
OND, not determined.
IlCumulative mean + SD of percent specific binding from individual experiments. For each experiment shown, highly significant binding (P < .001,
Student's t-test), of CFU-GM and BFU-E to collagen type Iwas observed. Although CFU-mix tended t o bind to collagen-coated wells, this binding was
not statistically significant, possibly due t o the small number of colonies observed.
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
PROGENITOR CELL BINDING TO COLLAGEN TYPE I
70
66 1
BUFFER
COLLAGENASE
z
E
50
0
u.
w
L
0
+
0
U
w
n
40
30
20
10
0
M07
CFU-GM
Fig 1. Effects of collagenase on binding of M 0 7 cells or normal
bone marrow CFU-GM t o collagen type 1. Control or collagen-coated
wells were incubated for 1 hour at room temperature with 20 pL of
either buffer (50 mmol/L Tris, 2.5 mmol/L CaCI,, pH 7.5) or with
collagenase type VI1 (0.5 U/mL) before adding either 6’chromiumlabeled M 0 7 cells or adherence-depletednormal bone marrow progenitor cells for determination of specific binding as described in text.
Data are expressed as mean 2 SD of percent specific binding from
two separate experiments for M07 cells and for day 7 CFU-GM.
B-1 chain (CD29) and at least six different a subunits.
These six distinct heterodimers are referred to as VLA-1 to
VLA-6, and they are known to bind to a variety of ECM
proteins.IYVLA proteins -1, -2, and -3 have been determined to bind to collagen in a calcium-dependent fashion
that may be partially blocked by peptides containing the
RGD
Because immunophenotypicanalysis of
the M07 cell line showed expression of VU-2, we were
interested in determining whether collagen binding of this
cell line was mediated through this adhesion protein.
Therefore, we studied the effects of a neutralizing antiCD29 antibody (4B4), RGD-containing peptide, or divalent cation chelation with EDTA, on the ability of M07
cells or normal bone marrow CFU-GM to bind to collagen
Fig 2. Specificity of collagenase for collagen type
1. Twenty micrograms of either laminin (lanes 1 and
4). fibronectin (lanes 2 and 5). or collagen type I(lanes
3 and 6) was incubated with either buffer (lanes 1
through 3) or 20 mU of collagenase type VI1 (lanes 4
through 6) for 1 hour at room temperature. After
boiling in sample buffer under reducing conditions,
the proteins were resolved by 7.5% SDS-PAGE, followed by visualization by Coomassie blue staining.
Three bands representing collagenase species are
indicated in lanes 4 through 6. Note that the 115-Kd,
125-Kd, and 212-Kd bands of collagen type I (lane 3)
are selectively degraded by collagenase type I,
whereas this enzyme has not degraded either laminin
or fibronectin.
type I. The 4B4 antibody has been previously shown to
block VLA-4lVCAM-1 interactions,=to neutralize collageninduced proliferation of T cells? and to neutralize binding
of the ovarian cancer cell line OVCAR-3 to both laminin
and fibronectin (S.A.C., unpublished observations, June
1991). As shown in Table 3, neither antibody 4B4 nor
RGD-containing peptide altered adhesion to collagen type
I. Also, divalent cation chelation caused no significant
change in collagen binding of M07 cells or of CFU-GM.
These results suggest that binding of M07 cells or of
CFU-GM to collagen type I occurs through a VLAindependent mechanism.
Because CD44 is also expressed by M07 cells and is
known to mediate collagen binding,I4.”we next tested the
effects of antLCD44 antibody (515) on the adherence of
M07 cells to collagen type I. However, no inhibitory effect
of anti-CD44 on M07 binding was observed (data not
shown). Because the 515 antibody used in these studies is
not known to be neutralizing, M07 cells were continuously
exposed to 515 antibody for 72 hours at 37°C in an attempt
to study the effects of CD44 downregulation on subsequent
binding. However, CD44 expression was not downregulated
despite a 72-hour exposure to the 515 antibody (data not
shown).
Comparative binding of normal neutrophils, monocytes,
and M 0 7 cells to collagen ype I. These data suggest the
possibility that binding to collagen I may represent one of
the mechanisms by which committed myeloid cells remain
within the bone marrow microenvironment during development. In an attempt to determine whether mature myeloid
cells also bind to collagen, we investigated the ability of
S’chromium-labeledneutrophils, monocytes, and M07 cells
to bind to collagen type I, with cumulative results shown in
Fig 3. These experiments were performed over a 2-hour
incubation (as opposed to 4 hours) at 37°C to maintain
neutrophil viability at greater than 90%. In the absence of
116-
COLLAGENASE
,$
77-
1
2
3
4
5
6
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
KOENIGSMANN ET AL
662
Table 3. Effects of Anti-CD29 Antibody, RGD Peptide, or EDTA on
Binding of M 0 7 Cells of CFU-GM to Collagen Type I
Cell Type
CFU-GM
Treatment
M07
(day 7)
Antibody or RGD peptide:*
Controlt
Anti-CD29
loo%§ (n = 3)
118% (n = 3)
100% (n = 1)
114% (n = 3)
ND
100% (n = 5)
100% (n = 3)
97% (n = 6)
RGD
Divalent cation chelation:*
Control
EDTA
79%11 (n = 9)
96% (n = 1)
Abbreviation: ND, not done.
*Binding was performed in quadruplicate wells in the continuous
presence of either anti-CD29, which is a neutralizing antibody recognizing the p-I integrin molecule, or RGD peptide (GRGDTP), with details as
described in Materials and Methods. Cells were pretreated with either
antibody or peptide prior to binding as previously described.
tFor experiments using the anti-CD29 antibody, the control antibody
was anti-CD33, which recognizesa surfaceprotein
with no known
adhesion properties. For experiments using RGD peptide, the control
peptide was GRGESP. Neither anti-CD33 nor GRGESP peptide altered
binding when compared to treatment with media alone.
Sln a separate set of experiments, binding was performed in quadruplicate wells under serum-free conditions with either 20% vol/vol PBS
(control) or 20% vol/vol EDTA in PBS (for a final concentration of 10
mmol/L EDTA in SFM).
§Data for each treatment group are expressed as percent of control
specific binding (SB)as determined from the indicated number of
experimentsasfollows: % of control SB = 100 x (mean SB [treatment])/
(mean SB [control]).
l/Not significantly different when compared with control (P = .06).
EDTA, background binding of monocytes to control wells
was significant (20% to 50% adhesion), precluding an
accurate assessment of specific binding to collagen type I
under these conditions. Because pilot experiments showed
that the binding of monocytes to plastic was significantly
blocked by 10 mmol/L EDTAwithout affecting corresponding binding to collagen I, we performcd binding experiments using neutrophils, monocytes, and M 0 7 cells in the
presence of 10 mmol/L EDTA to achieve acceptable levels
of background binding (5% to 10%). As shown in Fig 3, the
binding of both neutrophils and monocytes to collagen type
I was significantly less than that of M 0 7 cells (25% ? 8%
and 29% 2 6% v 40% 7%, P = .001). In addition, comparison of percent specific binding to collagen type I
obtained with neutrophils, monocytes, and CFU-GM (Table 2) showed significantly lowcr levels for either neutrophils or monocytes when compared with either day 7
(46% ? 8%) or day 14 CFU-GM (47% 12%) ( P < ,001).
Although CFU-GM binding was typically performed Over a
4-hour period, as opposed to 2 hours for neutrophils and
monocytes, we have performed additional experiments
which show that peak CFU-GM binding is achieved after a
minimum Of 1 hour Of inCubation at 37"c (data not shown).
Thus, it is unlikely that the lower binding of neutrophils or
monocytes compared with that of CFU-GM is due to
differences in binding duration.
During terminal differentiation, myeloid cells acquire the
ability to produce co~~agenase.24.2S
Therefore, we considered
the possibility that the obse,.,ed
decrease in neutrophil and
monocyte adhesion to collagcn might result from endogenous collagenase secretcd by these cells during the incubation period. To evaluate this possibility, media were conditioned by either neutrophils, monocytes, or M 0 7 cells for 2
hours at 37"C, (5 x 10' cells/mL SFM). Ccll-free supernatant was collected and used to treat 20 pg of collagen type I
in a total volume of 50 pL (86% vol/vol supernatant) for 1
hour at room temperature, followed by analysis by SDSPAGE as described. NO collagen degradation was observed
in any treatment group (data not shown).
Effects Of collagen lype I on Progenitor cell Proliferation.
To examine whethcr progenitor cells derive a proliferative
advantage during binding to collagen typc I, we evaluated
thymidine incorporation of M 0 7 cells (n = lo), as well as
colony growth of bone marrow progenitors (n = 9), in thc
presence or absence of collagen. Experiments were conducted in the presence or absence of GM-CSF or IL-3 over
a dose range of 5 to 25 ng/mL. No collagen-induced
enhancement of proliferation was observed (data not
shown).
*
*
DISCUSSION
Fig 3. Binding of M 0 7 cells, neutrophils, and mohocytes to collagen type I. 5'Chromium-labeled cells were bound to collagen type
I-coated plates for 2 hours at 37°C and processed as described in text
for assessment of percent specific binding, Binding was performed in
the presence of IO mmol/L EDTAfor all cell types as described in text
to reduce nonspecific binding of neutrophils and monocytes to
plastic. Data are expressed as mean 2 SD of specific binding of each
cell type from the indicated number of experiments. A statistically
significant decrease in the binding of both neutrophils and monocytes
to collagen type I when compared with that of M 0 7 cellS was
observed (P < ,001. Student's t-test).
This study was initially undertaken in an attempt to
investigate the ability of several ECM proteins to support
the adherence of the M 0 7 myeloid leukemia cell line. This
line was chosen for these experiments in view of its
hematopoietic origin and its expression of several potcntial
adhesion proteins, inc1uding vLA-2, vLA-4, VLA-6, CD29,
and CD44. We observed that collagen type I mediated
significant binding of M 0 7 cells, whereas these cells bound
to a considerably lesser degree to fibronectin and laminin,
In view of the high level of M 0 7 binding to collagen type I,
we further investigated thc role that this ECM protcin
might play in thc binding of normal bonc marrow cells.
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
PROGENITOR CELL BINDING TO COLLAGEN TYPE I
Both normal CFU-GM and BFU-E were found to significantly bind to collagen type I, and this interaction was
completely abolished by pretreatment of collagen-coated
plates with collagenase. In contrast to the results of other
investigators, who report proliferative effects of collagen
type I on T lymphocytes? we found no evidence for a
mitogenic effect of collagen type I on myeloid cells in the
presence or absence of exogenous GM-CSF or IL-3. This
observation makes it unlikely that our results are due to a
selective growth advantage of cells exposed to collagencoated wells.
We also observed that mature neutrophils and monocytes bind to collagen type I (Fig 3). Interestingly, this
binding was significantly less when compared with that of
M 0 7 cells (Fig 3), suggesting the possibility that adhesion
to collagen may decrease with myeloid maturation. It is
important to note that although M 0 7 cells have a blast
morphology and represent a less differentiated phenotype
when compared with neutrophils or monocytes, the M 0 7
cell line is leukemic, as well as megakaryocytic,in origin and
may not accurately reflect the binding of normal bone
marrow progenitors. However, the binding of neutrophils
and monocytes to collagen type I was also observed to be
significantly lower when compared with that of normal
CFU-GM (Table 2). It must be pointed out that these data
are not strictly comparable due to technical differences
between the measurement of mature cell binding (”chromium-labeling) and the measurement of CFU-GM binding
(colony assay). Unfortunately, in view of the low number of
CFU-GM present in normal bone marrow, it is difficult to
obtain highly purified progenitor cell populations for use in
51chromium-labelingstudies.
Collagen binds to a heterogenous group of cell surface
receptors. These include the p-1 integrin family of heterodimers (VLA-1, -2, and -3) (24), the CD44 molecule,
also known as the class I11 collagen r e ~ e p t o r , ~and
~ . ’ ~a
110-Kd protein known as dipeptidyl peptidase IV (CD26).23
Although VLA-1, VLA-3, and CD26 were not expressed by
the M 0 7 cell line, both VLA-2 and CD44 were strongly
expressed. The interaction of collagen with p-1 integrins is
known to be inhibited by RGD-containing
by
divalent cation chelation,27and by neutralizing antibody
against the common p-1 chain.” However, as shown in
Table 3, these interventions did not alter the binding of
M 0 7 cells or of CFU-GM to collagen type I. The concentration of RGD peptide used in this study was based on
previous work which demonstrated that this dose is capable
of blocking collagen-induced T-cell proliferation as well as
HeLa cell attachment to a collagen-based gelatin matrix.23328
However, higher concentrations of RGD peptide have been
required for maximal inhibition of binding in other cell
systems.” Nevertheless, our data suggest that the binding of
myeloid cells to collagen type I is not mediated by the p-1
integrins, despite the expression of VLA-2 by M 0 7 cells.
Although VLA expression is sometimes necessary for the
binding of cells to ECM proteins such as collagen, it is clear
that such expression is not sufficient for binding to occur.
For instance, it has been recently shown that activated B
663
cells bind to germinal center follicles through an interaction
between B-cell-expressed VLA-4 and the VCAM-1 molecule expressed within the germinal center.22 However,
B-cell lines which express VLA-4 do not necessarily bind to
germinal centers, suggesting that VLA-4 must be activated
(through a conformational change or through association
with another surface protein) to recognize the VCAM-1
molecule. A similar requirement for activation has been
recently shown for the binding of LFA-1 to ICAM-1F and
for the binding of LAM-1 to mannose residues3’ Therefore,
VLA expression may be necessary, but not sufficient, for
binding in some situations. The fact that the M 0 7 cell line
expresses VLA-2 but does not bind to collagen through this
molecule is consistent with this hypothesis.
Our studies do not exclude the possibility that CD44 is
responsible for mediating the binding of M 0 7 cells or
committed progenitor cells to collagen type I. CD44 expression was noted on greater than 90% of M 0 7 cells in this
report, and it has also been observed in highly purified
populations of normal human bone marrow progenitor
cells.1331 In addition, CD44 expression has been shown to
diminish as myeloid cells differentiate toward mature neutrophil~.’~
Because it is not known whether the anti-CD44
antibody used in this study is able to neutralize collagen
binding, and because the CD44 molecule could not be
downregulated from the surface of M 0 7 cells, the role of
CD44 in mediating the binding of progenitor cells to
collagen type I remains unclear.
It is likely that the binding of hematopoietic progenitor
cells to ECM proteins involves the coordinate interaction of
several types of adhesion proteins. Thus, uncommitted,
HLA-DR-negative progenitors bind to f i b r ~ n e c t i n , ~ . ~
whereas committed erythroid progenitors bind to both
fibronectinS7and collagen type I, and committed myeloid
progenitors bind to both hemonectin’ and collagen type I.
The presence of alternative mechanisms of erythroid progenitor cell binding that do not involve fibronectin was also
suggested by Coulombel et a17 who demonstrated that
BFU-E adherence to stromal cell-derived ECM could only
be partially blocked by antifibronectin antibody. Based on
our studies, it is possible that collagen type I may have
partly contributed to the fibronectin-independent binding
of BFU-E observed by these investigators. Although not
specifically addressed in this study, the role of other
collagens such as the type 111 and type IV species in the
binding of normal progenitor cells deserves further investigation.
Our data suggest the possibility that collagen type I may
partly mediate the adhesion of committed myeloid and
erythroid progenitor cells within the bone marrow microenvironment. In addition, mature myeloid cells such as
neutrophils and monocytes retain the capacity to bind to
collagen type I, with a suggestion that this binding may be
decreased compared with that of CFU-GM. Under certain
conditions, higher levels of bone marrow progenitor cells
circulate in the peripheral blood, such as during rebound
myelopoiesis after ~hemotherapy,~’
during hematopoietic
growth factor administration? or during pathologic condi-
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
KOENIGSMANN ET AL
664
tions such as the stem cell expansion characteristic of
chronic myeloid leukemia?' The ability of hematopoietic
growth factors such as GM-CSF or G-CSF to mobilize bone
marrow progenitors into the circulating white blood cell
pool is particularly noteworthy, because these agents are
also known to alter the expression of several adhesion
proteins on the surface of myeloid cells?6 A comparison of
the binding of circulating versus bone marrow progenitor
cells to collagen type I may permit a better understanding
of the physiologic role of this ECM protein in progenitor
cell adhesion.
ACKNOWLEDGMENT
We thank Dr Martin Hemler for his helpful advice during the
study.
--
REFERENCES
1. Coulombel L, Eaves AC, Eaves CJ: Enzymatic treatment of
long-term human marrow cultures reveals the preferential location
of primitive progenitors in the adherent layer. Blood 62:291,1983
2. Verfaille C, Blakolmer K, McGlave P: Purified primitive
human hematopoietic progenitor cells with long-term in vitro
repopulating capacity adhere selectively to irradiated bone marrow
stroma. J Exp Med 172:509,1990
3. Verfaille C, McCarthy J, McGlave P Differentiation of
primitive human Lin-34+DR- hematopoietic progenitors is accompanied by transition of adhesion from the heparin-binding domain
to the cell-binding domain of fibronectin. Blood 76:125a, 1990
(abstr)
4. Williams DA, Stephen C, Rios M, Patel V: Interactions of
primitive hematopoietic stem cells with the hematopoietic microenvironment: role of fibronectin and V U - 4 integrin. Blood 76:126a,
1990 (abstr)
5. 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
6. Patel VP, Ciechanover A, Platt 0, Lodish HF: Mammalian
reticulocytes lose adhesion to fibronectin during maturation to
erythrocytes. Proc Natl Acad Sci USA 82440,1985
7. Vuillet-Gaugler MH, Breton-Gorius J, Vainchenker W, Guichard J, Leroy C, Tchernia G, Coulombel L Loss of attachment to
fibronectin with terminal human erythroid differentiation. Blood
76:865,1990
8. Campbell AD, Long MW, Wicha MS: Developmental regulation of granulocytic cell binding to hemonectin. Blood 76: 1758,
1990
9. Zuckerman KS, Wicha MS: Extracellular matrix production
by the adherent cells of long-term murin bone marrow cultures.
Blood 61540, 1983
10. Waterhouse EJ, Quesenberry PJ, Balian G: Collagen synthesis by murine bone marrow cell culture. J Cell Physiol127:397,1986
11. Liesveld JL, Abboud CN, Duerst RE, Ryan DH, Brennan
JK, Lichtman MA: Characterization of human marrow stromal
cells: Role in progenitor cell binding and granulopoiesis. Blood
73:1794,1989
12. Paul SR, Yang RE, Donahue RE, Goldring S, Williams D A
Stromal cell-associated hematopoiesis: Immortalization and characterization of a primate bone marrow-derived stromal cell line.
Blood 77:1723,1991
13. Kansas GS, Muihead MJ, Dailey MO: Expression of the
CDllKD18, leukocyte adhesion molecule 1, and CD44 adhesion
molecules during normal myeloid and erythroid differentiation in
humans. Blood 76:2483,1990
14. Carter WG, Wayner EA: Characterization of the class 111
collagen receptor, a phosphorylated, transmembrane glycoprotein
expressed in nucleated human cells. J Biol Chem 263:4193,1988
15. Brown TA, Bouchard T, St-John T, Wayner E, Carter WG:
Human keratinocytes express a new CD44 core protein (CD44E)
as a heparan-sulfate intrinsic membrane proteoglycan with additional exons. J Cell Biol113:207, 1991
16. Zuckerman KS, Rhodes RK, Goodrum DD, Patel VR,
Sparks B, Wells J, Wicha MS, Mayo LA. Inhibition of collagen
deposition in the extracellular matrix prevents the establishment of
a stroma supportive of hematopoiesis in long-term murine bone
marrow cultures. J Clin Invest 75:970,1985
17. Avanzi GC, Lista P, Giovinazzo B, Miniero R, Saglio G,
Benetton G, Coda R, Catorretti G, Pegoraro L Selective growth
response to IL-3 of a human leukemic cell line with megakaryoblastic features. Br J Haematol69:359,1988
18. Cannistra SA, Groshek P, Griffin JD: Monocytes enhance
gamma-interferon-induced inhibition of myeloid progenitor cell
growth through secretion of tumor necrosis factor. Exp Hematol
16:865,1988
19. Hemler ME: VLA proteins in the integrin family: Structures, functions, and their role on leukocytes. Annu Rev Immunol
8:365,1990
20. Ruoslahti E, Pierschbacher MD: New perspectives in cell
adhesion: RDG and integrins. Science 238:491, 1987
21. Dedhar S, Ruoslahti E, Pierschbacher MD: A cell surface
receptor complex for collagen type I recognizes the Arg-Gly-Asp
sequence. J Cell Biol 104:585,1987
22. Freedman AS, Munro JM, Rice GE, Bevilacqua MP, Morimoto C, McIntyre BW, Rhynhart K, Pober JS, Nadler LM: Adhesion
of human B cells to germinal centers in vitro involves VLA-4 and
INCAM-1 10. Science 349:1030,1990
23. Dang NH, Torimoto Y, Schlossman SF, Morimoto C
Human CD4 helper T cell activation: Functional involvement of
two distinct collagen receptors, 1F7 and VLA integrin family. J Exp
Med 172:649,1990
24. Hasty KA, Hibbs MS, Kang AH, Mainardi C L Secreted
forms of human neutrophil collagenase. J Biol Chem 261:5645,
1986
25. Campbell EJ, Cury JD, Lazarus CJ, Welgus HG: Monocyte
procollagenase and tissue inhibitor of metalloproteinases. Identification, characterization, and regulation of secretion. J Biol Chem
262:15862,1987
26. Gehlsen KR, Argraves WS, Pierschbacher MD, Ruoslahti E:
Inhibition of in vitro tumor cell invasion by Arg-Gly-Aspcontaining synthetic peptides. J Cell Biol106:925,1988
27. Springer T A Adhesion receptors of the immune system.
Nature 346:425,1990
28. Beacham DA, Jacobson BS: Mg2' mediates the cellsubstratum interaction of Arg-Gly-Asp-dependent HeLa cell collagen receptors. Exp Cell Res 189:69,1990
29. Hibbs ML, Xu H, Stacker SA, Springer T A Regulation of
adhesion to ICAM-1 by the cytoplasmic domain of LFA-1 integrin
subunit. Science 251:1611, 1991
30. Spertini 0, Kansas GS, Munro JM, Griffin JD, Tedder T F
Regulation of leukocyte migration by activation of the leukocyte
adhesion molecule-1 (LAM-1) selectin. Nature 349:691,1991
31. Lewinsohn DM, Nagler A, Ginzton N, Greenberg P, Butcher
EC: Hematopoietic progenitor cell expression of the H-CAM
(CD44) homing-associated adhesion molecule. Blood 75589, 1990
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
PROGENITOR CELL BINDING TO COLLAGEN TYPE I
32. Coulombel L, Vuillet MH, Leroy C, Tchernia G: Lineageand stage-specific adhesion of human hematopoietic progenitor
cells to extracellular matrices from marrow fibroblasts. Blood
71:329, 1988
33. Richman CM, Weiner RS, Yankee RA: Increase in circulating stem cells following chemotherapy in man. Blood 47:1031, 1976
34. Socinski MA, Cannistra SA, Elias A, Antman KH, Schnipper L, Griffin JD: Granulocyte-macrophage colony stimulating
665
factor expands the circulating haemopoietic progenitor cell compartment in man. Lancet 1:1194,1988
35. Cannistra S A Chronic myelogenous leukemia as a model for
the genetic basis of cancer. Hematol Oncol Clin North Am 4:337,
1990
36. Demetri GD, Spertini 0,Pratt ES, Kim E, Elias A, Antman
K, Tedder TF, Griffin JD: GM-CSF and G-CSF have different
effects on expression of the neutrophil adhesion receptors LAM-1
and CDllb. Blood 76:178a, 1990 (abstr)
From www.bloodjournal.org by guest on February 6, 2015. For personal use only.
1992 79: 657-665
Myeloid and erythroid progenitor cells from normal bone marrow
adhere to collagen type I
M Koenigsmann, JD Griffin, J DiCarlo and SA Cannistra
Updated information and services can be found at:
http://www.bloodjournal.org/content/79/3/657.full.html
Articles on similar topics can be found in the following Blood collections
Information about reproducing this article in parts or in its entirety may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests
Information about ordering reprints may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#reprints
Information about subscriptions and ASH membership may be found online at:
http://www.bloodjournal.org/site/subscriptions/index.xhtml
Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American
Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036.
Copyright 2011 by The American Society of Hematology; all rights reserved.