Felipe Prosper, Department of Haematology and Cell Therapy, Clinica Universitaria de Navarra. Pamplona 31080, Spain. E-mail: firstname.lastname@example.org
We have investigated the influence of ex vivo expansion of human CD34+ cord blood cells on the expression and function of adhesion molecules involved in the homing and engraftment of haematopoietic progenitors. Ex vivo expansion of umbilical cord blood CD34+ cells for 6 d in the presence of interleukin 3 (IL-3), IL-6 and stem cell factor (SCF) or IL-11, SCF and Flt-3L resulted in increased expression of α4, α5, β1, αΜM and β2 integrins. However, a significant decrease in the adhesion of progenitor cells to fibronectin was observed after the ex vivo culture (adhesion of granulocyte-macrophage colony-forming units (CFU-GM) was 22 ± 4% in fresh cells versus 5 ± 2% and 2 ± 2% in each combination of cytokines). Incubation with the β1 integrin-activating antibody TS2/16 restored adhesion to fibronectin. Transplantation of ex vivo expanded umbilical cord blood CD34+ cells was associated with an early delayed engraftment in non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice. Incubation of cells with the monoclonal antibody TS2/16 before transplantation almost completely abrogated NOD/SCID repopulating ability of both fresh and expanded CD34+ cells. The seeding efficiency of fresh and expanded CD34+ cells was similar, but markedly reduced after incubation with the TS2/16 monoclonal antibody. Our results show that functional activation of β1 integrins could overcome the decreased very late antigen (VLA)-4- and VLA-5-mediated adhesion observed after ex vivo expansion of haematopoietic progenitors. However, in vivo, these effects induced an almost complete abrogation of the homing and repopulating ability of CD34+ UCB cells.
Ex vivo expansion of haematopoietic progenitors constitutes an attractive approach for decreasing the morbidity and mortality associated with haematopoietic stem cell transplantation. Although a number of studies have shown the efficacy of this approach in both experimental models and humans (Brugger et al, 1995; Conneally et al, 1997; Reiffers et al, 1999; Ueda et al, 2000), transplantation of ex vivo expanded progenitors has also been associated with delayed haematopoietic engraftment (Guenechea et al, 1999). These observations, together with the results of recent studies (Guenechea et al, 1999; Zanjani et al, 1999), indicate that the engrafting ability of haematopoietic cells is not only dependent on its haematopoietic progenitor cell content but suggest that at least certain ex vivo expansion procedures can impair the homing and/or engraftment capacity of haematopoietic cells.
Successful engraftment after transplantation requires stem cells to lodge in the bone marrow cavity and to proliferate and differentiate into mature blood cells while preserving the stem cell compartment. The mechanisms underlying homing of haematopoietic cells to the bone marrow and engraftment are poorly understood (Hardy, 1995; Prosper et al, 1998; Quesenberry & Becker, 1998). Interactions between cell adhesion molecules (selectins and integrins) and extracellular matrix components have been implicated in the homing process of haematopoietic stem cells (HSC) in animal models (Williams et al, 1991; Papayannopoulou et al, 1995, 1998; Pruijt et al, 1998; Quesenberry & Becker, 1998; van der Loo et al, 1998; Vermeulen et al, 1998). Experiments using α4-blocking antibodies have demonstrated the participation of the very late antigen (VLA)-4 integrin in haematopoietic cell homing to the bone marrow (Williams et al, 1991). A number of studies have also indicated a role for the VLA-4 integrin in mobilization of HSC in mouse and non-human primate models (Papayannopoulou et al, 1995, 1998; Vermeulen et al, 1998). Treatment of primates with blocking antibodies against the VLA-4 results in mobilization of progenitor cells into the peripheral blood, whereas incubation of bone marrow cells with an anti-VLA-4 significantly decreases homing and engraftment to the bone marrow. Interactions between chemokine receptors and their ligands, in particular CXCR4 and SDF-1, have been implicated recently in the engraftment of HSC in non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice (Peled et al, 1999). In the present study, we have investigated the effect of ex vivo expansion on integrin expression and function, and whether changes in the expression and function of these molecules could account for the short-term repopulating defects observed previously in human haematopoietic samples subjected to two different conditions of ex vivo expansion.
Materials and methods
Samples Cells were obtained from umbilical cord blood (UCB) after normal full term delivery and the informed consent of the mother. All samples were obtained from the Obstetric Department at the Hospital Clínico Universitario (Valencia, Spain). UCB was collected in placental collection bags using acid-citrate dextrose (ACD) as anticoagulant (Baxter, Deerfield, IL, USA). UCB mononuclear cells were separated using Ficoll–Hypaque centrifugation (specific gravity, 1077) (Sigma, St Louis, MO, USA). CD34+ cells were selected by performing two passages over the MACS CD34 Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) as previously described (Prosper et al, 1998). After positive selection the CD34+ populations were > 95% pure using flow cytometry.
Ex vivo expansion of CD34+ cells Purified CD34+ cells were incubated in Iscove's modified Dulbecco's medium (IMDM, GIBCO-BRL, Gaithersburg, MD, USA) supplemented with 20% fetal calf serum (FCS), 100 U/ml penicillin and streptomycin (GIBCO-BRL) and 100 ng/ml of recombinant human interleukin 3 (IL-3), interleukin 6 (IL-6) and stem cell factor (SCF) (R & D, UK) (EE1) or SCF, IL-11 (a kind gift from Genetics Institute, Cambridge, MA, USA) and Flt-3L (kindly provided by Immunex, Seattle, WA, USA) (EE2). Cytokines were added only at the beginning of the culture. Cell concentration was 50 000 CD34+ cells per ml. Cultures were maintained for 6 d at 37°C and 5% CO2. At the indicated time points, cells were harvested and analysed using flow cytometry, assayed in ex vivo cultures and adhesion experiments or transplanted into NOD/SCID mice.
Flow cytometry analysis The following antibodies coupled to fluorescein isothiocyanate (FITC) or phycoerythrin (PE) were used: mouse antibodies directed to CD29 (β1 integrin), CD34, CD38, CD49d (α4 integrin), CD49e (α5 integrin), CD11a (LFA-1), CD11b (Mac-1), CD18 (β2 integrins), CD117 (c-kit), CD62L (L selectin) and CD45. Antibodies purchased from Becton Dickinson (San Jose, CA, USA) were CD38, CD45, CD62L, CD11a, CD117 and CD34; CD29, CD49d, CD11b, CD18 and CD49e were from Immunotech (Marseille, France). Fresh and cultured cells or cells obtained from NOD/SCID mice were resuspended in 100 µl of phosphate-buffered saline (PBS) + 0·3% BSA labelled with 10 µl of antibody and incubated for 10 min at 4°C in the dark, washed and resuspended in 0·5 ml of PBS. For analysis, 40 000 cells were acquired in list-mode using a flow cytometer (FACSCalibur, Becton Dickinson) and software (cellquest, Beckton Dickinson). FITC and PE conjugated isotype-matched immunoglobulins were used as controls.
Short-term methylcellulose progenitor cell cultures The appropriate number of fresh or cultured cells were seeded in methylcellulose based short-term cultures for 14 d at 37°C in a humidified atmosphere at 5% of CO2 in air as previously described (Prosper et al, 1998).
Adhesion experiments Bovine serum albumin (BSA), fibronectin (FN) (Sigma), FN80k and FN H89 (de la Fuente et al, 1999) (fibronectin-derived peptides, a generous gift from Dr Angeles García-Pardo) were diluted in PBS to the appropriate concentrations and adsorbed to 48-well plates as described (Verfaillie et al, 1991; Prosper et al, 1998). Fresh or cultured cells were resuspended in serum-free IMDM and plated in contact with fibronectin, FN H89 or FN 80 k for 2 h in a humidified atmosphere at 37°C. Non-adherent cells were removed by four standardized washes using warm IMDM and adherent cells were removed by trypsinization (Verfaillie et al, 1991). To determine the percentage of adherent CFC, adherent and non-adherent fractions were replated in methylcellulose assays. Per cent adhesion was calculated as number of CFC in the adherent fraction/(number of CFC in adherent + non-adherent fractions). In some experiments, cells were incubated with the activating antibody TS2/16 (1:10 dilution of hybridoma culture supernatant) (generously provided by Dr Sanchez Madrid), with the integrin blocking monoclonal antibodies P4C2 (1:400 dilution), P4C10 (1:400 dilution) or P1D6 (1:400) or with control mouse IgG or media alone for 30 min before adhesion assays (Bhatia et al, 1994). Monoclonal antibodies P4C2 (murine anti–human α4), P4C10 (murine anti–human β1), and P1D6 (murine anti–human α5) were purchased from GIBCO-BRL and used as dilutions of mouse ascites (Lundell et al, 1996). Control mouse IgG (Sigma) was diluted in PBS and used at a concentration of 20 µg/ml.
Transplantation into NOD/SCID mice and analysis of homing and engraftment NOD/LtSz-scid/scid (NOD/SCID) mice (deficient in Fc receptors, complement function, natural killer, B-, and T-cell function) were used as recipients of human haematopoietic cells. Mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). All animals were handled under sterile conditions and maintained under microisolators. Before transplantation, 6- to 8-week-old mice were total body irradiated with 2·5–3·0 Gy of X-rays (300 kV, 10 mA; Philips MG-324, Hamburg, Germany). CD34+ purified cells (100 000 CD34+ cells per mouse) or equivalent numbers of ex vivo expanded cells were transplanted into irradiated NOD/SCID mice (cells obtained from 100 000 CD34+ cells plated on d 0 and expanded for 6 d). In some experiments, cells were pre-incubated with the activating antibody TS2/16 (1:10 dilution) against the β1 integrin for 30 min, centrifuged and resuspended in PBS before infusing into the mice.
At periodic intervals after transplantation (20–90 d), bone marrow samples were aspirated from one femur by puncture through the knee joint, according to a previously described procedure (Guenechea et al, 1999). At the end of the experiments, mice were killed and bone marrow, peripheral blood and spleen analysed using flow cytometry for the presence of human cells as described (Guenechea et al, 1999). The bone marrow cells were harvested by crunching the bones as described (Guenechea et al, 1999). The spleen was carefully cut in parts and sieved through a 100-µm-mesh filter. The cells were then washed and resuspended in IMDM for immunophenotyping. Aliquots of 1–5 × 105 cells/tube were stained for 25 min at 4°C with anti-human –CD45-FITC (clone HI30; Pharmingen, San Diego, CA, USA) or PECy5 (Clone J33, Immunotech) in combination with anti-human–CD34-PE (Anti-HPCA-2; Becton Dickinson Immunocytometry).
For homing experiments the number of transplanted cells per mouse was 500 000 CD34+ cells (fresh) or the equivalent after ex vivo expansion. At 24 h after transplantation mice were killed and femora and spleen were isolated. To determine the seeding efficiency to the bone marrow, we assumed that one femur contains approximately 6% of the total bone marrow cellularity (van Hennik et al, 1999). The bone marrow and spleen seeding efficiencies were calculated on the basis of the number of infused and retrieved CD34+ cells. After infusion of CD34+ UCB cells the seeding efficiency was determined by dividing the number of retrieved/infused CD34+ UCB corrected for the purity of the population. The number of CD34+ cells infused after ex vivo expansion was calculated as the total number of cells by the percentage of CD34+ cells. Seeding efficiency is expressed as the percentage of CD34+ cells recovered.
Statistics Data are presented as the median (range) or mean (SEM). The significance of differences between groups was determined by using the two-tailed Student's t-test. The processing and statistical analysis of the data was performed by using the software spss 9·0 (SPSS, Inc, Chicago, IL, USA).
Expression of integrins in normal and ex vivo expanded CD34+ CB cells
CD34+ CB cells were expanded in vitro for 6 d in the presence of different cytokine combinations. After culture with IL-3, IL-6 and SCF (EE1) there was a 10·8-fold (4–17·2) expansion (n = 7) in the number of cells and a 10·7-fold (2·1–41·8) expansion for the combination of IL-11, SCF and Flt-3L (EE2) (n = 6) (median and range). Immature progenitors ,defined as CD34+ CD38 cells, were expanded 1·3- and 2·1-fold respectively (data not shown). After ex vivo expansion, there was a statistically significant increase in the expression of α4, α5, β1, LFA-1 (CD11a), Mac-1 (CD11b) and β2 integrin by the CD34+ cell population both in the percentage of positive cells as well as in the intensity of expression (Mean Fluorescence Intensity, MFI) (Fig 1) (P < 0·01). No changes were observed in the percentage of cells expressing l-selectin or c-kit, and only the MFI of the l-selectin was modified by the culture.
Ability of normal and ex vivo expanded cells to adhere to fibronectin
The capacity of progenitor cells to adhere to the bone marrow extracellular matrix has been implicated in homing and engraftment (Prosper et al, 1998; Quesenberry & Becker, 1998; Zanjani et al, 1999). Because adhesion through integrins depends not only on the expression but also on their functional status (Hynes, 1992; Stuiver & O'Toole, 1995; Bhatia et al, 1996) we performed progenitor cell adhesion experiments after ex vivo expansion to determine whether increased expression of integrins was associated with increased adhesion capacity. Unexpectedly, UCB CD34+ cells expanded with either combination of cytokines showed a decreased adhesion to FN (Fig 2A) in comparison to adhesion of fresh UCB CFU-GM (5 ± 2% and 2 ± 2% versus 22 ± 4%, P < 0·001). Adhesion of fresh UCB progenitors, but not of ex vivo expanded progenitors, was inhibited by incubation with blocking antibodies against α4, α5 or β1. However, both expanded and fresh UCB CFU-GM adhesion was upregulated to the same degree by activating antibodies against the β1 integrin (Fig 2). To further assess whether decreased adhesion was dependent on α4 and α5 integrins, the recombinant fragments containing the FN region responsible for binding through α4 (FN H89) and through α5 (FN 80 k) integrins were used for adhesion assays (Moyano et al, 1997,1999; Mould et al, 1997, 1998). As shown in Fig 2B, adhesion to both FN H89 and FN 80 k were significantly lower after ex vivo expansion but could be upregulated by activating the β1 integrin. These results suggest that ex vivo expansion of UCB progenitors is associated with a decreased α4β1- and α5β1-mediated adhesion to FN.
Transplantation of ex vivo expanded UCB cells into NOD/SCID mouse
Previously, we have demonstrated that transplantation of ex vivo expanded UCB progenitors into NOD/SCID is associated with a decrease in short-term repopulation with human cells 20 d after transplantation, whereas long-term repopulation (120 d after transplant) is similar after infusion of fresh or ex vivo expanded cells (Guenechea et al, 1999). Based on our in vitro results, we proposed a potential implication of β1 integrin in this effect, and hypothesized that increasing the functional status of the β1 integrin could restore the engraftment ability of the expanded samples. Therefore, we transplanted fresh and expanded CD34+ cells incubated with or without TS2/16 before transplantation. Consistent with our previous observations (Guenechea et al, 1999), the proportion of human cells found in the femoral bone marrow of the NOD/SCID mice short-term after transplantation was significantly lower when expanded rather than fresh samples were used as donor grafts. The percentage of engrafting human cells was 13·5% ± 7·3% and 3·86% ± 2·3% (mean ± SE) for fresh and ex vivo expanded cells respectively. However, activation of β1 integrins with TS2/16 almost completely abrogated the repopulating ability of both fresh and expanded samples (Fig 3). Engraftment was 0·5% ± 0·45% and 0·03% ± 0·02% for fresh and ex vivo expanded cells respectively. Differences in engraftment between cells incubated or not with TS2/16 were statistically significant (P < 0·05). The use of either combination of cytokines did not affect engraftment of expanded UCB cells (data not shown).
Ex vivo expansion of UCB is not associated with abnormal homing of progenitors
Decreased engraftment may be caused by an impaired homing ability of the transplanted progenitors (abnormal lodgement) or to alteration in proliferation of transplanted cells. To determine whether abnormal lodgement of UCB progenitors could be responsible for decreased engraftment, the numbers of human CD34+ cells found in the bone marrow and spleen of NOD/SCID recipients were determined 24 h after transplantation. When fresh CD34+ cells were transplanted, CD34+ seeding efficiency ranged between 0·1% and 6% (see filled points in Fig 4), in good agreement with previous observations (van Hennik et al, 1999). No significant differences were observed when fresh or expanded cells were transplanted, suggesting that reduction in the short term engraftment is not related to a decreased homing ability of the cells. In contrast, a marked reduction in the seeding efficiency of human CD34+ cells was observed in fresh and expanded samples treated with TS2/16. Identical conclusions were derived from the analysis of NOD/SCID spleens, indicating that the TS2/16-mediated activation of β1 integrins is markedly detrimental for homing and engraftment of UCB progenitors into NOD/SCID haematopoietic organs (Table I).
Table I. Homing to the spleen and bone marrow of fresh and ex vivo expanded cells transplanted in NOD/SCID mice.
Bone marrow (%)
3·52 ± 0·87
0·68 ± 0·17
5·34 ± 2·12
1·26 ± 0·65
0·5 ± 0·1
0·07 ± 0·02
1·43 ± 0·87
0·4 ± 0·18
Most studies of ex vivo expansion of haematopoietic progenitors have focused on the potential of ex vivo culture to expand progenitors at different stages of maturation, including putative stem cells (Brugger et al, 1993; Conneally et al, 1997; Piacibello et al, 1997, 1999; Albella et al, 1999; Shih et al, 1999; Ueda et al, 2000). However, whether in vitro culture alters the homing or engraftment process has not been fully addressed. Previous studies have indicated that expression of adhesion molecules is altered during in vitro culture (Reems et al, 1997; Chute et al, 1999). Both upregulation and downregulation of β1 and β2 integrin expression has been reported depending on the culture conditions, cytokines and cells used for expansion (Becker et al, 1999). Our results provide evidence suggesting that ex vivo expansion of UCB progenitors upregulates expression of cell adhesion molecules but significantly reduces the adhesive potential of progenitor cells. Moreover, in accordance with others (Becker et al, 1999), we observed that ex vivo expansion of UCB cells results in a reduced α4β1- and α5β1-dependent adhesive capacity of expanded progenitors.
Previously, we have demonstrated that ex vivo expanded CD34+ UCB cells show a delayed engraftment when transplanted into NOD/SCID mice (Guenechea et al, 1999). However, although activation of β1 integrins increases the adhesive capacity to FN, it clearly impairs the engraftment capacity of expanded as well as fresh cells.
Engraftment requires UCB progenitors and stem cells to initially home to the bone marrow. There is evidence suggesting that haematopoietic cells do not home specifically to the bone marrow (Vos et al, 1972; Samlowski & Daynes, 1985; Papayannopoulou et al, 1995). The initial steps of homing consist of the recognition of endothelial cells by cell adhesion molecules expressed by the progenitor cells that include selectins (Frenette et al, 1998). After lodgement into the bone marrow, progenitor cells establish interactions between integrins of the β1 family (VLA-4 and VLA-5) and extracellular matrix components [FN and vascular cell adhesion molecule (VCAM)-1] and are selectively retained within the bone marrow (Aizawa & Tavassoli, 1988; Williams et al, 1991; Prosper et al, 1998). After intravenous infusion, haematopoietic progenitor cells can be detected temporarily in other organs suggesting that transplanted haematopoietic cells do not selectively home to the bone marrow but are selectively retained once they lodge there (Papayannopoulou et al, 1995). According to our results, ex vivo expansion did not result in an abnormal seeding capacity of UCB CD34+ cells as demonstrated by the similar seeding efficiency between ex vivo expanded and fresh UCB CD34+ cells, suggesting that the decrease in early engraftment is not caused by an abnormal homing capacity. A recent report (Szilvassy et al, 1999) has suggested that ex vivo expansion of haematopoietic cells results in an impaired homing capacity. Differences in methods may explain these discrepancies. Our results were obtained using human cord blood cells expanded ex vivo and transplanted into mice compared with using a mouse into mouse model (Szilvassy et al, 1999). There were also differences regarding the conditions of ex vivo expansion (serum-free media, cytokines) that could be responsible for the differences observed in the two studies.
Recently, Zanjani et al (1999), using a different animal model, has demonstrated a role for VLA-4 in the lodgement of haematopoietic cells into the bone marrow. In their study, using a human into sheep xenogeneic model, blocking of the α4 integrin before infusion of HSC resulted in a profound reduction of human cell lodgement to the bone marrow with a corresponding increase of human cells in the circulation. Further, activation of the β1 integrin also induced an abnormal homing of donor cells into the fetal liver instead of the bone marrow. We observed that β1 integrin activation altered the homing to the bone marrow and spleen both in fresh and ex vivo expanded UCB progenitor cells. We believe that enhanced adhesion to endothelial cells of vessels within other organs may explain the abnormal homing to the bone marrow and spleen as well as the very significantly reduced engraftment. Taken together with our own results, we suggest that changes in the functional status of β1 integrins can impair the engrafting ability of transplanted samples, either by reducing their adhesion to the haematopoietic stroma (down-modulated adhesion function), or by promoting their adhesion to non-hematopoietic tissues (up-modulated adhesion function). These observations support the relevance of modulating ex vivo the activity of β1 integrins on the homing and engrafting ability of human haematopoietic samples in an in vivo animal model.
An explanation for the decreased engraftment of ex vivo expanded cells may be related to changes in cell cycle status of progenitor cells. Recent studies have suggested that cell cycle progression upregulates expression of VLA-4 (Yamaguchi et al, 1998) and that progenitor cells moving into the G1 phase of the cell cycle have a reduced capacity for repopulating NOD/SCID mice (Gothot et al, 1998). Unfortunately, we did not address the cell cycle status of ex vivo expanded cells so we cannot rule out the possibility that cell cycle changes are responsible for the reduced adhesion of UCB progenitors. Taking into account potential differences in the homing capacity of CD34+ subpopulations involved in the short-term and long-term repopulation we cannot definitely exclude that ex vivo expansion alters the homing capacity.
Early acting cytokines such as Flt-3L ligand, SCF or thrombopoietin play a key role in preservation of immature progenitors (Bhatia et al, 1997; Conneally et al, 1997; Piacibello et al, 1998; Yagi et al, 1999). Addition of other cytokines such as IL-3 may result in differentiation of immature progenitors and loss of self-renewal capacity (Brugger et al, 1993; Piacibello et al, 1998). We did not find any differences in the adhesive behaviour or engraftment capacity after ex vivo expansion between the two different combinations of cytokines used suggesting that the reduced engraftment of ex vivo expanded cells is not likely to be related to differentiation of immature progenitors and lost of engrafting cells. In fact, as we have previously demonstrated, long-term engraftment was not altered by ex vivo expansion (Guenechea et al, 1999). Although different authors have used other combinations of cytokines and longer duration of ex vivo expansion, most studies suggest that ex vivo expansion beyond 7–10 d reduces the number of SCID-repopulating cells (SRC). We elected to use the current conditions both to maintain consistency to our previous study and to assure preservation of SRC (Guenechea et al, 1999).
In summary, although the decreased early engraftment observed after transplantation of ex vivo expanded progenitors is associated with a reduced adhesion capacity of UCB progenitors mediated through α4β1 and α5β1 integrins, the homing capacity of CD34+ cells to the bone marrow and spleen is not altered by ex vivo culture. The low functional state of β1 integrins may be physiological as activation of the β1 integrin results in a very significant reduction in homing and engraftment of haematopoietic cells.
We are grateful to Dr Francisco Sanchez Madrid and Dr Angeles García-Pardo for generously providing the antibody TS2/16 and the FN peptides and Dr Catherine Verfaillie for her excellent critical review. The authors thank the midwifes from the Hospital Clinico Universitario for their tireless efforts to provide umbilical cord blood. We thank Eva Juan for her excellent technical support with the experiments. This work was supported in part by a grant from Fondo de Investigaciones Sanitarias FIS 98/0863 to FP and SAF98-0008-CO4-01 to JAB.