Plasma Levels of SDF-1 and Expression of SDF-1 Receptor on CD34+ Cells in Mobilized Peripheral Blood of Non-Hodgkin's Lymphoma Patients

Authors


Abstract

CXCR4 is the receptor for the chemokine stromal derived factor-1 (SDF-1), is expressed on CD34+ cells, and has been implicated in the process of CD34+ cell migration and homing. We studied the mobilization of CD34/CXCR4 cells and the plasma levels of SDF-1 and flt3-ligand (flt3-L) in 36 non-Hodgkin's lymphoma patients receiving cyclophosphamide (Cy) plus G-CSF (arm A), Cy plus GM-CSF (arm B), or Cy plus GM-CSF followed by G-CSF (arm C) for peripheral blood stem cell (PBSC) mobilization and autotransplantation.

We observed lower plasma levels of SDF-1 in PBSCs compared to premobilization bone marrow samples. The mean plasma SDF-1 levels were similar in PBSC collections in the three arms of the study. In contrast, SDF-1 levels in the apheresis collections of the “good mobilizers” (patients who collected a minimum of 2 × 106 CD34+ cells/kg in one to four PBSC collections) were significantly lower than the apheresis collections of the “poor mobilizers” (≥0.4 × 106 CD34+ cells/kg in the first two PBSC collections; 288 ± 82 pg/ml versus 583 ± 217 pg/ml; p = 0.0009). The mean percentage of CD34+ cells expressing CXCR4 in the apheresis collections was decreased in the PBSC collections compared with premobilization values from 28% to 19.4%. Furthermore, the percentage of CD34+ cells expressing CXCR4 in the good mobilizers was significantly lower compared with the poor mobilizers (14.7 ± 2.1% versus 33.6 ± 2.1%; p = 0.002). The good mobilizers had also significantly lower levels of flt3-L compared with the poor mobilizers (34 ± 4 pg/ml versus 106 ± 11 pg/ml; p = 0.006), Finally, the levels of flt3-L strongly correlated with SDF-1 levels (r = 0.8; p < 0.0001). We conclude: A) low plasma levels of SDF-1 and low expression of CXCR4 characterize patients with good mobilization outcome, and B) the levels of SDF-1 correlate with flt3-L, suggesting an association of these cytokines in mobilization of CD34+ cells.

Introduction

Autologous peripheral blood stem cells (PBSC) provide a rapid and sustained hematopoietic recovery after the administration of high-dose chemotherapy or chemoradiotherapy in patients with hematological malignancies and solid tumors. PBSC transplantation has become the preferred source of stem cells for autologous transplantation because of the shorter time to engraftment and the lack of a need for surgical procedure necessary for bone marrow (BM) harvesting [1-5]. The mechanism underlying the release of stem cells from the marrow to the peripheral blood is not yet understood. Similarly, it is not yet clear why some patients mobilize CD34+ cells better than others do, although prior therapy has been shown to be a contributing factor [6-13]. G-CSF and GM-CSF are the most commonly used hematopoietic growth factors for PBSC mobilization and although they slightly differ in the profile of cells mobilized to the peripheral blood [14], they are equipotent in mobilizing CD34+ cells [3, 15-23]. Therefore, one must assume that other factors are involved in the regulation of stem cell release from the BM.

flt3 ligand (flt3-L) is an effective mobilizer of hematopoietic stem cells and is capable of inducing multilineage hematopoietic cell differentiation, both in vivo [24] and in vitro [25, 26]. Interestingly, it has also been reported that flt3-L levels are markedly elevated in patients with Fanconi's anemia and aplastic anemia [27, 28], and in patients with leukemia following chemotherapy-induced BM aplasia [29]. Based on these findings, it has been postulated that the increase in flt3-L in those diseases reflects a compensatory mechanism to stimulate hematopoiesis [27-29]. In agreement with these findings, we reported recently that non-Hodgkin's lymphoma (NHL) patients who failed to mobilize CD34+ cells had relatively high plasma levels of flt3-L, both before mobilization and in the apheresis collections, regardless of the growth factor used for mobilization [30].

The CXCR4 molecule constitutes the receptor for the chemokine stromal derived factor-1 (SDF-l) and is expressed on CD34+ cells [31, 32]. Both SDF-1 and its receptor have been implicated in migration of CD34+ cells in vitro [30-35]. However, to date, no studies have been reported comparing the plasma levels of SDF-1 and the efficiency of CD34+CXCR4+ cell mobilization in a controlled prospective study of PBSC mobilization by different mobilization regimens. Herein, we report the results obtained from a randomized study of NHL patients comparing the extent of CD34+CXCR4+ cell mobilization in patients mobilized with cyclophosphamide (Cy) plus G-CSF, Cy plus GM-CSF, or Cy plus GM-CSF followed by G-CSF. We also measured the levels of SDF-1 and flt3-L in the plasma of these patients before mobilization and in the apheresis collections. Our results indicate significant involvement of the CXCR4 receptor and SDF-1 chemokine in the release of CD34+ cells to the peripheral blood.

Materials and Methods

Study Design

We designed a randomized study of stem cell mobilization to compare the efficacy of PBSC mobilization with Cy (3 g/m2) plus G-CSF (10 μg/kg) (arm A), Cy plus GM-CSF (250 μg/m2) (arm B), or Cy plus GM-CSF (250 μg/m2) for seven days, followed by five to eight days of G-CSF (10 μg/kg) (arm C), in NHL patients undergoing high-dose chemotherapy and PBSC rescue. Patients were excluded from the study if they were over 70 years of age, had therapy for over two years, or received therapy within the last four weeks. Patients with abnormal heart, kidney, or liver function and pregnant women were also excluded. All patients were pheresed for four hours a day, for four consecutive days, however, patients who collected the required amount of CD34+CD45dim cells (≥2 × 106 CD34+CD45dim cells/kg) before day 4 were subsequently pheresed for two hours for the remaining days of collection. Most patients in this group were pheresed at least two days for the full four hours with 10 l of blood exchanged. Patients who collected ≥2 × 106 CD34+CD45dim cells/kg in one to four days were defined as “good mobilizers” [9, 15]. These patients collected ≥1 × 106 CD34+CD45dim cells/kg in two days of apheresis. In contrast, patients who collected ≤0.4 × 106 CD34+CD45dim cells/kg, in the first two days of apheresis were considered “poor mobilizers” and subsequently were dropped off protocol after two collections and were remobilized later by other mobilization protocols at the discretion of the referring physician [9, 15]. This definition of good mobilizers and poor mobilizers was adopted following our previous observation that patients who collected ≤0.4 × 106 CD34+CD45dim cells/kg in the first two days of apheresis had >95% probability to fail to collect the desired amount of 2 × 106 CD34+CD45dim cells/kg in ≥4 days of apheresis [9, 15].

Patient Population

Thirty-seven patients enrolled in the study at the Audie L. Murphy Memorial VA Hospital, University Hospital and Wilford Hall Medical Center at San Antonio. Thirteen patients were enrolled in arm A, 10 in arm B, and 12 in arm C. Median age in arm A was 46 years and in arms B and C, 52 and 55 years, respectively. More than 75% of patients in each arm were males (mostly VA patients). Fifty-four percent, 80%, and 83% of patients in arms A, B, and C, respectively, were in relapse at the time of transplant and 42%, 40%, and 50% of patients in arms A, B, and C, respectively, had three or more regimens of therapy prior to mobilization. The distribution of patients with follicular lymphoma was similar in the three arms of the study (41%, 30%, and 40%, in arms A, B, and C, respectively). All patients signed a consent form as required by our institutional review board.

Stem Cell Apheresis Collection, Cryopreservation, and Infusion

Stem cell collection, cryopreservation, and infusion and criteria for engraftment were described previously [9, 15]. All patients started apheresis collections at WBC >500 in two consecutive days (usually on day 12). Patients were collected for four days using the Cobe Spectra apheresis machine and 10 l of blood were exchanged in each procedure. Patients were transplanted within one month with ≥2 × 106 CD34+ cells/kg, following conditioning with CBV (CTX 1.8 g/m2 at d -6 to d -3; etoposide 0.4 g/m2 at d -6 to d -4; BCNU 0.45 g/m2 at d -3). Following transplantation, patients receiving ≥2 × 106 CD34+ cells/kg engrafted neutrophiles (>500, for two consecutive days) and platelets (>20,000 for two consecutive days and with no platelet transfusion) with a median of 11 and 14 days, respectively. Plasma samples for SDF-1 and flt3-L and mononuclear cells for CD34+ cells and CD34+CXCR4+ cells were collected from the peripheral blood prior to mobilization, during four apheresis collections (at mid-point of collection) and six months after transplantation and were stored at -80°C.

Determination of flt3-L by ELISA

Frozen plasma samples were thawed in a 37°C water bath and centrifuged for 10 min at 3,000 g and the plasma supernatants were transferred to new tubes for testing. The plasma level of flt3-L was determined by a sandwich enzyme-linked immunosorbent assay (ELISA), essentially as described before [30]. Briefly, 100 μl/well of rat α-hu-flt3-L (4 μg/ml; M5 monoclonal antibody, Immunex Corporation; Seattle, WA; http://www.immunex.com) were used to coat 96-well plates (MaxiSorp Immunoplate; NUNC; Naperville, IL; htttp://www.nalgenunc.com) and incubated overnight at 4°C. The plates were then thoroughly washed with washing buffer (phosphate-buffered saline [PBS]) containing 0.05% Tween 20. Plasma samples were diluted 1:5 with PBS and dispensed into the wells. To generate a standard curve, soluble Chinese hamster ovary (CHO)-cell-derived, hu-r-flt3-L (Immunex Corporation) was used at a concentration range of 12.5 pg/ml to 2.5 ng/ml, following serial dilutions in normal rat plasma. Standard curves were included for each plate. Rabbit polyclonal α-hu-flt3-L (P9, Immunex) was used at 1:4,000 dilution as second antibody. For detection, peroxidase-conjugated goat α-rabbit IgG (Jackson Immunoresearch Laboratory; West Grove, PA) was used at 1:6,000 dilution. For color development, 100 μl of TMB-peroxidase substrate/chromogen solution (Kirkegaard & Perry; Gaithersburg, MD; http://www.kpl.com) were added to each well and incubated at room temperature for 10 min. The reaction was stopped with 100 μl of 1 M H3PO4. Absorbance optical density (OD) at 450 nm was determined by an automated ELISA reader (THERMOmax, Molecular Devices Corporation; Menlo Park, CA; http://www.moleculardevices.com). Regression curve was used to convert OD units to picograms per milliliter of flt3-L. Samples were run in triplicates. Assay background was determined by inclusion in each plate of triplicate wells containing all reagents, but with normal rat plasma, instead of patient's sample. Background (<0.15 OD) was automatically subtracted from each well.

Determination of SDF-1 by ELISA

The plasma levels of SDF-1 were determined by a sandwich ELISA as follows: aliquots of 100 μl/well of anti-SDF-1 (5-10 μg/ml; monoclonal antibody, R&D Systems; Minneapolis, MN; http://www.rndsystems.com/asp.g_home.asp) were used to coat 96-well plates (MaxiSorp Immunoplate; NUNC) and incubated overnight at 4°C. Plates were blocked with PBS containing 1% bovine serum albumin (BSA) for one hour at room temperature, followed by washing with washing buffer (PBS) containing 0.1% BSA plus 0.05% Tween 20. Plasma samples were diluted 1:4 or 1:2 with PBS and were dispensed into the wells. To generate a standard curve, hu-r-SDF-1 (SDF-1α, R&D Systems) was used at a concentration range of 25 ng/ml diluted to 100 pg/ml in 10 serial dilutions in PBS plus 1% BSA. After a two-hour incubation at room temperature, plates were washed and 100 μl of a 0.5 μg/ml of biotin anti-SDF-1α antibody (clone 79018-111, R&D Systems) were added to each well. After a two-hour incubation at room temperature, plates were thoroughly washed and 100 μl of a 1:10,000 dilution of peroxidase-streptavidin conjugate (Jackson Immunoresearch Laboratory) were added into each well and plates were incubated at room temperature for one hour. After washing of unbound antibody, 100 μl of TMB-peroxidase substrate/chromogen solution (Kirkegaard & Perry) were added to each well and incubated at room temperature for 10-20 min. The reaction was stopped with 100 μl of 1 M H3PO4. Absorbance at 450 nm was determined by an automated ELISA reader (THERMOmax). Patients' samples were run in triplicates. Assay background was determined as was indicated for flt3-L. An example for a typical standard curve is shown in Figure 1.

Figure Figure 1..

Plasma levels of SDF-1 in mobilized peripheral blood.Plasma samples from PBSC of four consecutive apheresis collections were used. For experimental details regarding plasma separation, SDF-1 ELISA and for the definition of the three arms of the study, seeMaterials and Methods.Bars represent mean ± SD. Standard curves for each ELISA plate (inset) were obtained as described in theMaterials and Methodssection. A reproducible r-value between the range of 0.90-0.95 was obtained. The background was <0.16 OD. Results were derived from 10 patients in arm A, eight patients in arm B, and 10 patients in arm C.

Dual Color Staining for CD34+CD45dim and CD34+CXCR4+ Cells

Staining for CD34+CD45dim and the pattern of stem cell mobilization for each patient in this study were published recently in detail [9, 15]. Staining for CD34+CXCR4+ cells was done as described for CD34+CD45dim cells except that anti-CD34 antibody (HPCA-2-PE, Becton Dickinson; Mountain View, CA; http://www.bd.com) was matched with anti-CXCR4-FITC (R&D Systems). One hundred thousand cells were collected, to ensure >200 cells in the double positive quadrant.

Statistical Analyses

Nonparametric Student's t-test was performed for comparing two populations (Mann-Whitney test). Analysis of variance (ANOVA) between multiple groups was performed by the Tukey multiple comparison test, or by the Newman-Keuls multiple comparison test. The statistical package of the GraphPad Prism graphics software (San Diego, CA) was used throughout this study.

Results

Expression of CXCR4 Receptors on CD34+ Cells in Steady-State Bone Marrow and PB and in Mobilized PBSC

We determined the expression of CXCR4 on CD34+ cells in apheresis collections of NHL patients mobilized with three different regimens: Cy plus G-CSF (arm A); Cy plus GM-CSF (arm B); Cy plus GM-CSF followed by G-CSF (arm C) for stem cell collection. The co-expression of CXCR4 receptor on CD34+ cells was determined by dual color staining for CD34+ cells with anti-CD34-PE (red fluorescence) and with anti-CXCR4-FITC (green fluorescence). An example is shown in Figure 2 for a good mobilizer with 1.27% CD34+CXCR4 cells (upper left quadrant) and 0.34% CD34+CXCR4+ cells (upper right quadrant) and 36.88% CXCR4+CD34 cells (lower right quadrant). Using this technique we analyzed premobilization peripheral blood (PB) and BM mononuclear cells as well as PBSC in four consecutive apheresis collections for the number of CD34+ cells co-expressing CXCR4. The results are depicted in Figure 3. The mean percentage of CD34+ cells expressing CXCR4 in the BM was 29 ± 7.2%, 26.5 ± 8.6%, and 26.4 ± 7.1%, in patients in arms A, B, and C, respectively. The mean percentage of CD34+ cells expressing CXCR4 in the PB was 21.7 ± 4.6%, 18.6 ± 8.3%, and 22.6 ± 3.8%, in patients in arms A, B, and C, respectively. The differences observed in CXCR4 expression on CD34+ cells between the combined BM and peripheral blood samples were marginally significant (p = 0.05). The mean percentage of CD34+ cells expressing CXCR4 was significantly lower in the combined apheresis collections compared to the combined BM CD34+ cells (15.3 ± 2.8%, 16.4 ± 3.1%, and 17 ± 2.8%), in patients in arms A, B, and C, respectively. Significant differences were observed between BM and PBSC collections from all patients combined (BM versus PBSC; p = 0.03) and between PB and PBSC collections from all patients combined (PB versus PBSC; p = 0.04). Although a general trend for a decrease in CXCR4 expression was observed between day 1 to day 4 of collection, within each study group, no major differences were observed between the three arms of the study (Fig. 3).

Figure Figure 2..

Dual color staining for CD34+CXCR4+cells: an example for dual staining of an apheresis collection from a “good mobilizer” patient.Left upper panel is the light scatter gating; right upper panel is the background fluorescence of isotype controls. Lower panel is staining for CD34/CXCR4 cells. The upper left quadrant represents CD34+cells only (1.27% of total); the upper right quadrant represents CD34/CXCR4+cells (0.33% of total). Lower right quadrant represents CXCR4+cells (36.88% of total). One hundred thousand cells were analyzed. For further details, seeMaterials and Methods.

Figure Figure 3..

Expression of CXCR4 receptor on mobilized CD34+cells: comparison between the three arms.Staining for CD34+CXCR4+cells was performed as described in theMaterials and Methodssection. d1 to d4 are apheresis collections in four consecutive days. Bars represent mean ± SD.

In contrast, when the patients in arms A, B, and C were regrouped as good mobilizers and poor mobilizers, a statistically significant difference in the percentage of CD34+ cells expressing CXCR4 was observed between the PBSC collections (day 1 + day 2) in the two groups of patients (Fig. 4). Whereas the mean percentage of CD34+ cells expressing CXCR4 in the steady-state BM and PB was similar in the good mobilizers and the poor mobilizers (34 ± 3.1% versus 33.6 ± 2.1% and 31.8 ± 3.3% versus 25.3 ± 2.7% for BM and PB, respectively), the mean percentage of CD34+ cells expressing CXCR4 in day 1 plus day 2 of the apheresis collections of the good mobilizers was 14.7 ± 2.1%, compared with 33.6 ± 5.2% in the combined collections of the poor mobilizers group of patients (p = 0.002) (Fig. 4). Interestingly, we observed low expression of CXCR4 on CD34+ cells in poor mobilizers, upon successful remobilization with high-dose G-CSF or with paclitaxel plus G-CSF (results not shown). We also tested the expression of CXCR4 on normal donors mobilized with G-CSF for allogeneic transplantation. The mean percentage of CD34+ cells expressing CXCR4 in the PBSC collections obtained from four normal donors was much lower compared with the mean obtained for unprimed BM from three normal donors (7.5 + 4.4% versus 24.5 ± 5%; BM versus PBSC p = 0.001) (horizontal lines in Fig. 4).

Figure Figure 4..

Expression of CXCR4 receptor on mobilized CD34+cells: “good mobilizers” versus “poor mobilizers.”The results presented are from the pool of patients in arms A, B, and C as described in Figure 3Figure 3., but grouped as “good mobilizers” in arms A, B, and C (≥2 × 106 CD34+CD45dim cells/kg, in four consecutive collections), or as “ poor mobilizers” (0-0.4 × 106 CD34+CD45dim cells/kg, in two consecutive collections). Bars represent mean ± SD. Other details are as in Figure 3Figure 3.. The two horizontal lines depict the mean percentage of CD34+cells expressing CXCR4 in the PBSC collections obtained from four normal donors (N PB) and the mean obtained for unprimed BM from three normal donors (N BM).

SDF-1 Levels in Apheresis Collections

We determined the plasma levels of SDF-1 in NHL patients in the three arms of the study in four consecutive apheresis collections. No statistically significant differences were observed in the plasma levels of SDF-1 between the three arms of the study. A mean level of SDF-1 of 522 ± 232 pg/ml, 503 ± 165 pg/ml, and 491 ± 188 pg/ml was observed for patients in arms A, B, and C, respectively (Fig. 1). A trend for a decrease in the levels of SDF-1 was observed in the apheresis collections between day 1 and day 4 of collection. In contrast, when the patients were regrouped as good mobilizers and poor mobilizers, a statistically significant difference in the plasma levels of SDF-1 was observed in the PBSC collections between the two groups of patients. The results are depicted in Figure 5. In the group of the good mobilizers, SDF-1 levels before mobilization were relatively high in the BM and PB (426 ± 99 pg/ml versus 358 ± 65 pg/ml) compared with the levels observed in the four apheresis collections (239 ± 81 pg/ml, 220 ± 71 pg/ml, 211 ± 79 pg/ml, and 205 ± 65 pg/ml, for days 1, 2, 3, and 4 of apheresis, respectively). The decrease in plasma levels of SDF-1 was statistically significant with p values for BM versus day 1 of 0.0013, BM versus day 2 of 0.0016, BM versus day 3 of 0.005, and BM versus day 4 of 0.008 (Fig. 5). Significant differences were also observed between PB and PBSC collections with p values of 0.007, 0.003, 0.003, and 0.003 for days 1, 2, 3, and 4, respectively (Fig. 5). The poor mobilizers had even higher levels of SDF-1, before mobilization (759 ± 250 pg/ml, 714 ± 211 pg/ml) and after mobilization, in two consecutive apheresis collections (666 ± 264 pg/ml and 637 ± 175 pg/ml, for days 1 and 2 of collection, respectively). Therefore, poor mobilizers had statistically significant higher levels of SDF-1 both at steady state, before mobilization, and in the PBSC collections (p = 0.0004 and 0.005 for BM versus PBSC and PB versus PBSC, respectively). Furthermore, the levels of SDF-1 observed in the BM of eight normal donors (544 ± 209 pg/ml) were close to the levels observed in the BM, PB, and PBSC of the poor mobilizers group of patients (Fig. 5).

Figure Figure 5..

Plasma levels of SDF-1 pre- and postmobilization.Peripheral blood (PB) and bone marrow (BM) samples were collected before mobilization and during the first four days of collection (apheresis product). For experimental details regarding plasma separation, SDF-1 ELISA and for the definition of “good mobilizers” and “poor mobilizers,” seeMaterials and Methods. Bars represent mean ± SD. Results were derived from 18 good mobilizers (four apheresis collections each) and eight poor mobilizers (two apheresis collections each) and eight normal (N) controls.

To further investigate the differences between the good mobilizers and the poor mobilizers, we compared SDF-1 levels in the pooled apheresis collections of the good mobilizers, the poor mobilizers, and PB from six normal donors. The results are depicted in Figure 6. Thus, whereas the mean level of SDF-1 in the four apheresis collections of the good mobilizers was 248 ± 82 pg/ml, a mean of 583 ± 217 pg/ml SDF-1 was observed for the two apheresis collections of the poor mobilizers (p = 0.0003). The mean SDF-1 levels in PB from six normal donors were close to the levels in the poor mobilizers (610 ± 243) (Fig. 6). Interestingly, we observed decreased plasma levels of SDF-1 in the apheresis collections derived from poor mobilizers, upon successful remobilization with high dose G-CSF or with paclitaxel plus G-CSF (results not shown).

Figure Figure 6..

Plasma levels of SDF-1 in mobilized peripheral blood.Blood samples were collected before mobilization and during the first four days of collection (apheresis product). For experimental details regarding SDF-1 ELISA and for the definition of “good mobilizer” and “poor mobilizer” patients, seeMaterials and Methods. Bars represent mean ± SD.

In a previous publication we reported that poor mobilization was associated with higher levels of flt3-L [30]. Therefore, it was of interest to us to compare the levels of flt3-L and SDF-1 in the same patient in order to determine the correlation between the two cytokines. Indeed, we found a strong correlation (r = 0.8; p < 0.0001) between SDF-1 and flt3-L, indicating a possible linkage between the biological actions of these cytokines (Fig. 7).

Figure Figure 7..

Correlation between SDF-1 and flt3-L levels in mobilized peripheral blood.ELISA for SDF-1 and flt3-L was performed as described inMaterials and Methods. Correlation coefficient between SDF-1 and flt3-L was 0.8 (p < 0.0001).

Discussion

It is not yet understood why some patients do mobilize adequate numbers of CD34+ cells and some do not. Recently, however, several factors were identified as high risk factors for poor mobilization of CD34+ cells. The extent of prior therapy and disease status was the major factor associated with poor mobilization [12-18]. More recently, we demonstrated that the levels of flt3-L could be used to predict the success of mobilization [30]. These markers, while practically useful, do not explain the differences in the efficacy of CD34+ cell mobilization between different patients. Most intriguing in this respect are recent findings associating the CXCR4 receptor on CD34+ cells and its chemokine, SDF-1 in the process of migration of CD34+ cells, in vitro [31-37]. In addition, recent studies implicated adhesion molecules such as VLA-4 and L-Selectin expressed on CD34+ cells, in the mobilization process [38-41]. These recent findings begin to explain the molecular mechanisms of stem cell release from the BM to the peripheral blood.

High Plasma Levels of SDF-1 and High Levels of CXCR4 on CD34+ Cells Characterize Patients with Poor Mobilization of CD34+ Cells

In agreement with previous reports [35], we observed lower percentage of CD34+CXCR4+ cells in premobilization PB compared with BM. We also observed a decrease in the percentage of CD34+ cells expressing CXCR4 in the apheresis collections of all study patients compared with premobilization BM and PB, regardless of the growth factor used for mobilization. Our results agree with previous results reported by Aiuti et al. [33], who determined the expression of CXCR4 on CD34+ cells in BM, PB, and apheresis collections in a small number of patients. They reported 40% CD34+CXCR4+ cells in BM, 32% in PB and 17% in PBSC collections. Mohle et al. [31] reported a variable expression of CXCR4 on CD34+ cells (28.4% to 87.6%) in four collections from patients with different types of cancer [31]. Lee et al. [42] reported 68% CD34+CXCR4+ cells in BM from normal donors. They did not test apheresis collections. Viardot et al. [43] tested one apheresis collection from a total of 21 patients with different types of cancer, the majority of whom had solid tumors; only four had NHL. In contrast to our results, they found a mean proportion of CD34+ cells expressing CXCR4 of 37.5% in PBSC collections and 65.9% in the BM. However, only six patients were matched for the proportions of CD34+CXCR4+ cells pre- and postmobilization. The discrepancy could be due to the patient population size, the number of apheresis collections tested, and the type of cancer. In this respect, only six patients in the Viardot et al. study had BM involvement [43], whereas our results are based on 116 samples from 35 NHL patients, tested before and after mobilization.

Most importantly, however, we demonstrate for the first time a statistically significant decrease in the expression of CXCR4 on CD34+ cells in the PBSC collections of the good mobilizers compared with the poor mobilizers group of patients. These findings suggest an association between SDF-1 and its receptor in the migration and homing of CD34+ cells in the BM, as suggested before [31, 34, 35].

SDF-1 has been shown to modulate the expression of its receptor in vitro and in vivo and was implicated in the process of CD34+ cell migration and mobilization [31, 34-36]. We therefore hypothesized that the plasma levels of SDF-1 in the BM (BM supernatant) and in the apheresis collections will vary in patients before and after mobilization and between the apheresis collections of poor mobilizers versus good mobilizers. Indeed, our results clearly demonstrate a statistically significant decrease in the levels of SDF-1 in the apheresis collections compared with the premobilization BM and PB. Furthermore, we also demonstrated that PBSC collections derived from poor mobilizers had statistically significant higher levels of SDF-1, compared with the good mobilizers group of patients. Our finding that poor mobilizers had lower levels of CD34+CXCR4+ cells and lower plasma SDF-1 upon successful remobilization further supports an association between SDF-1 and CXCR4 and the mobilization process (Gazitt, unpublished observation). Furthermore, as was the case for flt3-L [30], high levels of SDF-1 could be used as a predictor for poor mobilization.

High expression of CXCR4 on CD34+ cells is a prerequisite for migration and proliferation of CD34+ cells in vitro [31, 34-36] and for successful engraftment and BM reconstitution in vivo [34]. However, the expression of CXCR4 on CD34+ cells in apheresis collections of mobilized patients is relatively low. The apparent paradox can be resolved by recent findings indicating that growth factors such as flt3-L, SCF and IL-3, upregulates the expression of CXCR4, in vitro [31, 34-36] and in vivo [34]. Furthermore, a cross-talk between SDF-1 and other growth factors and adhesion molecules (e.g., VLA-4 and VLA-5) has been described recently by Peled et al. [44]. Hence, it is reasonable to assume that endogenous growth factors, or growth factors administered post-transplantation will result in facilitation of the process of engraftment. In this regard, our finding of a strict correlation between plasma levels of flt3-L and SDF-1 suggests perhaps a concerted role for these two cytokines in the regulation of stem cell proliferation, homing and release from the BM.

In summary, our results confirm a general decrease in the expression of CXCR4 and in the plasma levels of SDF-1 in apheresis collections compared to steady-state BM or peripheral blood in NHL patients. The fact that a similar decrease in the expression of CXCR4 was observed also in PBSC collections from normal donors primed for mobilization with G-CSF compared with unprimed normal marrow suggests perhaps a more general association of these molecules in normal growth of hematopoietic stem cells in the bone microenvironment and in the release of hematopoietic stem cells to the circulation. More studies are under way to determine the exact role of CXCR4 and SDF-1 in the process of mobilization.

Acknowledgements

This study was partially funded by a grant from Immunex Corporation. The authors would like to thank Mr. C. Thomas for running the flow cytometer.

Ancillary