Dr Godfrey Chi Fung Chan, Department of Paediatrics and Adolescent Medicine, The University of Hong Kong, Queen Mary Hospital, 102 Pokfulam Road, HKSAR, China. E-mail: firstname.lastname@example.org
Mesenchymal stem cells (MSCs) are an important cellular component of the bone marrow microenvironment for supporting haemopoiesis. However, their response to high-dose chemotherapy remains unknown. We assessed the acute direct effects of individual chemotherapeutic agents on human MSCs (hMSCs). Using an in vitro culture system, the chemosensitivity of hMSCs was determined by XTT (2,3-bis(2-methoxy-4-nitro-5-sulphophenyl)-5-[(phenylamino) carbonyl]-2H-tetrazolium hydroxide) assay in comparison with that of NB-4 cells, a leukaemic cell line, and normal peripheral blood mononuclear cells. The recovery of cell numbers following exposure to chemotherapeutic agents and chemotherapy-induced apoptosis of hMSCs were evaluated. Human MSCs were resistant to chemotherapeutic agents commonly used in bone marrow transplantation (BMT) (i.e. busulphan, cyclophosphamide and methotrexate). However, they were relatively sensitive to a panel of cytotoxic agents, such as paclitaxel, vincristine, etoposide and cytarabine. Furthermore, different recovery patterns were noted. There was sustained suppression in hMSCs following 3 d exposure to paclitaxel, cytarabine and etoposide. In contrast, significant recovery was seen in hMSCs treated with dexamethasone and vincristine respectively. Human MSCs have different patterns of response to a panel of chemotherapeutic agents commonly used in BMT or cancer therapy. Understanding this variation is important in optimizing conditioning regimens for BMT.
Of foremost importance, after conventional HSC transplantation, bone marrow stroma remains recipient in origin and the damage cannot be repaired by donor MSCs (Devine & Hoffman, 2000). Even the infusion of culture-expanded MSCs (up to 10 × 106 cells/kg) still failed to repopulate allogeneic MSCs in recipient bone marrow (Koc et al, 2002; Lee et al, 2002).
In the past, investigations of factors affecting BMT engraftment were focused on the role of HSCs and their progeny. But, despite the clinical manipulation of these factors, engraftment failure continues to be a significant clinical problem (Woolfrey & Anasetti, 1999). Engraftment failure has been considered to be a consequence of inadequate HSCs (Wolff, 2002) and lack of immunological cells in the donor graft (Woolfrey & Anasetti, 1999). The role of MSCs in enhancing the engraftment has been suggested by some recent studies (Devine & Hoffman, 2000; Fibbe & Noort, 2003).
Knowing how to preserve or destroy MSCs is necessary in clinical practice. It is therefore important to understand the MSCs damage pattern caused by chemotherapy. Patients receiving anticancer treatment usually have a combination of chemotherapeutic agents or/and irradiation, so the damage caused by each individual agent cannot be determined individually. One study has suggested that damage caused by chemotherapy occurred during the early post-BMT period (Galotto et al, 1999), implying chemotherapy-induced acute injury to MSCs. In the present study, we assessed the acute direct effects of individual chemotherapeutic agents on human MSCs (hMSCs) using an in vitro culture system. The chemosensitivity of hMSCs was also compared with that of NB-4 cells, a leukaemic cell line and normal peripheral blood mononuclear cells (PBMCs). We also evaluated the recovery of cell numbers following exposure to chemotherapeutic agents and determined whether high-dose chemotherapy induced cell death in hMSCs.
Materials and methods
Isolation and culture of hMSCs
This study was approved by the Ethics Committee (Internal Review Board) of the University of Hong Kong.
Bone marrow samples were collected from three young healthy donors. Heparinized BM was mixed with 2 volumes of phosphate-buffered saline (PBS) and density-separated by Ficoll–Hypaque (Amersham Biosciences, Uppsala, Sweden). Mononuclear cells were collected from the interface and washed twice with PBS. The washed cells were resuspended in hMSC medium, consisting of Dulbecco's modified Eagle's medium (DMEM) low glucose supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, 100 mg/ml streptomycin, and 2 mmol/l l-glutamine. The cells were plated at 5–30 × 106 cells/100 mm dish. Cultures were maintained at 37°C in a humidified atmosphere with 5% CO2. After 24 h, non-adherent cells were removed by changing the medium and the plate was washed twice with PBS. Medium was replaced every 3–4 d thereafter. When the cultures were near confluence, cells were detached by 0·05% trypsin/25 mmol/l EDTA solution and re-plated at a density of 2 × 105 cells/75 cm2 flask.
Culture of NB-4 cells
NB-4 cells (ACC 207; Deutsche Sammlung von Mikroorganismen und Zellkulturen Abt. Menschliche und Tierische Zellkulturen, Braunschweig, Germany) were established from the bone marrow of a 23-year-old woman with acute promyelocytic leukaemia (acute myeloid leukaemia, French–American–British type M3) in second relapse in 1989. These cells carry the t(15;17) PML-RARA fusion gene. Cells were cultured in medium consisting of 90% Roswell Park Memorial Institute (RPMI) 1640 medium, 10% FBS, 100 U/ml penicillin, 100 mg/ml streptomycin and maintained at 37°C in a humidified atmosphere with 5% CO2.
Isolation and culture of PBMCs
The PBMCs were isolated from the buffy coat of a healthy voluntary donor. The buffy coat was mixed with an equal volume of PBS and density-separated by Ficoll–Hypaque. Mononuclear cells were collected from the interface and washed twice with PBS. The washed cells were resuspended in medium consisting of 90% RPMI 1640 medium, 10% FBS, 100 U/ml penicillin, 100 mg/ml streptomycin. Cultures were maintained at 37°C in a humidified atmosphere with 5% CO2.
Immunophenotype of hMSCs
Surface markers were selected according to a previous report (Pittenger et al, 1999). The cells were harvested by treatment with 0·05% trypsin-EDTA at the end of the first passage. For flow cytometry, the detached cells were washed and resuspended in PBS. Aliquots of 2 × 105 cells were labelled with fluorescein isothiocyanate (FITC)- or phycoerythrin (PE)-conjugated monoclonal antibodies including Isotype control, CD14-FITC, CD29-FITC, CD45-PE (Immunotech, Marseille, France), CD34-PE (Becton Dickinson Immunocytometry Systems, San Jose, CA, USA), CD105-PE (Research Diagnostics Inc., Flanders, NJ, USA) and HLA-A,B,C-FITC (PharMingen, San Diego, CA, USA). Flow cytometric analysis was performed with a Coulter Epics Elite Flow Cytometer (Beckman Coulter Corporation, Miami, FL, USA). Ten thousand events per sample were collected into listmode files and analysed by the WinMDI 2·8 analysis software.
The capacity of MSCs to differentiate along adipogenic and osteogenic lineages was assessed as previously reported (Galotto et al, 1999; Pittenger et al, 1999). Briefly, confluent hMSC cultures were induced to adipogenic differentiation by culturing in adipogenic induction medium containing 1 μmol/l dexamethasone and 0·5 mmol/l methyl-isobutylxanthine, 10 μg/ml insulin, 100 μmol/l indomethacin (MDI + I; Sigma, St Louis, MO, USA) and 10% FBS in DMEM for 48–72 h, and then the medium was changed to adipogenic maintenance medium containing insulin (10 μg/ml) and 10% FBS in DMEM for 24 h. The hMSCs were then re-treated with MDI + I for a second or third time. The cells were stained with 1-8-[4-(dimethylphenylazo)dimethylphenylazo]-2-naphthalenol (oil red O). To induce osteogenic differentiation of hMSCs, the cells were treated with medium consisting of DMEM, 10% FBS, 50 μmol/l l-ascorbic acid, 10 mmol/l β-glycerol phosphate, and 100 nmol/l dexamethasone (Sigma). Medium was replaced every 3 d. Calcium deposition was examined by the von Kossa staining.
Cell proliferation was measured using a commercial colorimetric assay (2,3-bis (2-methoxy-4-nitro-5-sulphophenyl)-5-[(phenylamino) carbonyl]-2H-tetrazolium hydroxide; XTT) (Roche Molecular Biochemicals, Mannheim, Germany) according to the manufacturer's instructions. Cells were grown in flat-bottomed 96-well plates in a final volume of 100 μl culture medium per well. An aliquot (50 μl) of the XTT labelling mixture was added to each well, and incubated for 4 h at 37°C in a humidified atmosphere with 5% CO2. The resulting spectrophotometrical absorbance was measured at 490 nm, and results are expressed as a percentage of control cell numbers.
The following chemotherapeutic agents were tested (range of concentration): arsenic trioxide (10−9–10−4 mol/l, Sigma); busulphan (10−8–10−3 mol/l; Busulfex; Orphan Medical, Minnetonka, MN, USA); cytarabine (10−7–10−2 mol/l, Cytosar, Pharmacia and Upjohn, Puurs, Belgium); cyclophosphamide (10−7–10−2 mol/l; Endoxan-Asta, ASTA Medica AG, Frankfurt, Germany); dexamethasone (10−9–10−4 mol/l; Sigma); etoposide (10−8–10−3 mol/l; PCH Pharma Chemie, Harrlem, the Netherlands); methotrexate (10−8–10−3 mol/l; EBEWE Arzneimittel Ges.m.b.H., Unterach, Austria); paclitaxel (10−9–10−4 mol/l, Taxol; Bristol-Myers Squibb Caribean Company, Mayaguez, Puerto Rico); vincristine (10−10–10−5 mol/l, Pharmachemie B.V., Haarlem, NY, USA). These concentration ranges were selected in order to determine the maximal tolerable dose of hMSCs. The dosage range covers the plasma level of patients receiving these chemotherapeutic agents.
Drug sensitivity measurements
Human MSCs were seeded in 96-well plates at a density of 3 × 103 per well in 100 μl of medium. NB-4 cells and PBMC were seeded in 96-well plates at a density of 105 cells per well in 100 μl of medium. After 24 h, the chemotherapeutic agents were added at the final concentrations described above. Cells were incubated with each agent for 3 d. Control wells containing hMSCs with culture medium, but no chemotherapeutic agent, were used to determine the control cell numbers. All tests were assayed in triplicate. Results were expressed as a percentage of control cell numbers. The IC50 value, which is the drug concentration needed to inhibit 50% proliferation of the cells, was used as a measure of sensitivity.
Recovery of hMSCs following treatment of individual chemotherapeutic agents
Human MSCs were seeded in 96-well plates at a density of 3 × 103 per well in 100 μl of medium. After 24 h, individual chemotherapeutic agents were added to the cultures for 72 h and then removed. Human MSCs were cultured in medium for further 9 d and cell numbers were determined at different time points. The concentration used for each agent was determined from available data of plasma concentrations measured in patients receiving intensive chemotherapy (Dorr & VonHoff, 1994; Grochow & Ames, 1998; Toth et al, 1999). The levels used corresponded to the BMT setting for busulphan (16 mg/kg) and cyclophosphamide (180 mg/kg). For cytarabine, methotrexate and dexamethasone, the levels corresponded to high-dose therapy (cytarabine 3 g/m2, methotrexate 2–5 g/m2, dexamethasone 60 mg/m2). The respective levels of the drugs were as follows: arsenic trioxide 2 μmol/l, busulphan 5 μmol/l, cytarabine (high-dose) 100 μmol/l, cyclophosphamide (high-dose) 500 μmol/l, dexamethasone (high-dose) 10 μmol/l, etoposide 10 μmol/l, methotrexate (high-dose) 10 μmol/l, paclitaxel 1 μmol/l, vincristine 0·1 μmol/l.
Evaluation of apoptosis in hMSCs cultured with a chemotherapeutic agent
Annexin V (AV) was used as a marker of apoptosis. Human MSCs were seeded in 60-mm dishes at a density of 2·5 × 105 per dish. After 24 h, chemotherapeutic agents (high-dose cytarabine 100 μmol/l, high-dose dexamethasone 10 μmol/l, etoposide 10 μmol/l, paclitaxel 1 μmol/l, vincristine 0·1 μmol/l) were added for 24–72 h, after which the cells were trypsinized and collected together with the supernatant. Cells were collected at three time points: 24, 48 and 72 h. The cells were labelled with AV-FITC and propidium iodide (PI) (Annexin V-FITC Kit; Immunotech, Marseille, France) and analysed by flow cytometry. Treatment was performed in duplicate and the percentage of labelled cells undergoing apoptosis was determined using the WinMDI software.
Values are given as mean ± standard error of the mean (SEM). Comparisons between mean values were made using unpaired Student's t-test (two-tailed). The difference was statistically significant when P < 0·05.
Growth and immunophenotype of hMSCs
Human bone marrow-derived MSCs were successfully cultured and expanded. Symmetric colonies were observed at approximately 5–7 d after initial plating. A morphologically homogeneous population of fibroblast-like cells with 90% confluence was seen after 10–16 d (Fig 1A). The cultured mesenchymal cells comprised a single phenotypic population with 98% homogeneity after passage 1, by flow cytometric analysis of expressed surface antigens. The hMSCs were uniformly positive for CD29, CD105 and HLA-A, B, C, but negative for CD14, CD34 and CD45 (Fig 1B), which was consistent with the hMSCs phenotype and excluded the contamination of haemopoietic cells.
Differentiation of hMSCs
When subjected to osteogenic medium, hMSCs underwent a morphological change, from spindle-shaped to cuboidal, and showed deposition of calcium, as shown by van Kossa stain (Fig 2A). Adipogenic differentiation was shown by the accumulation of lipid vacuoles stained with oil red O (Fig 2B).
Effect of chemotherapeutic agents on hMSCs numbers
Similar dose-effect–response curves were observed from all three sets of MSCs, and results from one representative experiment are shown in Fig 3. A variety of dose–response curves were produced with the chemotherapeutic agents tested. Increasing concentrations of etoposide and paclitaxel reduced cell numbers in a dose-dependent manner, whereas a plateau occurred with vincristine >10−7 mol/l and cytarabine >10−5 mol/l. As for arsenic trioxide, busulphan and cyclophosphamide, reduction in cell numbers occurred only at extremely high concentrations that well exceeded the highest plasma concentration achieved in patients receiving chemotherapy. Five agents (paclitaxel, vincristine, etoposide, high-dose cytarabine and high-dose dexamethasone) reduced the cell numbers by more than 20% at the clinically relevant concentrations (58·5%, 57·7%, 31·1%, 25·4% and 30·2% respectively).
Chemosensitivity of hMSCs versus NB-4 cells and PBMCs
Human MSCs were more resistant to all chemotherapeutic agents than NB-4 cells. With the exception of paclitaxel and busulphan, the IC50 value of each agent in hMSCs was 10 times higher than that of NB-4 cells. Human MSCs were also more sensitive than PBMCs for the following agents: etoposide (2·8-fold), paclitaxel (101-fold) and vincristine (65·7-fold), but more resistant than PBMCs for cytarabine, busulphan (20·4-fold), arsenic trioxide (7·8-fold), and cyclophosphamide (6·9-fold). The relative sensitivity and IC50 data of the agents from one representative experiment are summarized in Table I.
Table I. Chemosensitivity of hMSCs, NB-4 cells and PBMCs [IC50 (10−7 mol/l)].
The results represent the mean ± SEM of triplicate cultures of one representative experiment.
PBMC, peripheral blood mononuclear cell; hMSC, human mesenchymal stem cell; NB-4 cell, a leukaemic cell line.
R1, the mean IC50 value for hMSC divided by the mean IC50 value for NB-4 cells.
R2, the mean IC50 value for hMSC divided by the mean IC50 value for PBMC.
*Compared with the mean IC50 of NB-4 cells, P < 0·01.
†Compared with the mean IC50 of PBMC, P < 0·01.
‡R1/R2 could not be calculated because the mean IC50 value for hMSCs exceeded the maximum concentration.
Recovery of hMSCs following treatment of individual chemotherapeutic agents
Upon prolonged exposure (72 h) with cytarabine, paclitaxel and etoposide, significant cell loss was noted. When these agents were removed, hMSCs showed sustained suppression of proliferation when compared with the same time-point control. However, in the case of vincristine, hMSCs showed significant recovery after withdrawal of the drug. However, full recovery was not achieved, as the number of hMSCs was still significantly lower than the control (64·3%) after another 9 d of culture. For cells treated with dexamethasone, the number of hMSCs returned to the control level after prolonged culture (9 d), although the cell number was reduced to 69% after the initial 3 d of treatment (Fig 4). The results represent the triplicate culture of one representative experiment from the three sets of MSCs.
Apoptosis of hMSCs induced by chemotherapeutic agents
The induction of apoptosis by the five agents that reduced cell number more than 20% at the clinically relevant concentrations were determined. The other four agents showed no suppressive effect or induced morphological signs of cell death, so they were not included in this assay. All the agents tested increased the percentage of apoptotic cells in human MSCs populations (Fig 5). The percentage of early apoptotic cells (AV+/PI−) and late apoptotic cells (AV+/PI+) are shown in Fig 5A and B respectively. The maximal effect was achieved between 48 and 72 h. Within 24 h, paclitaxel and vincristine increased the numbers of early apoptotic cells (9·25 ± 0·65% and 7·6 ± 0·7% respectively), and apoptotic cell numbers increased when the exposure time was prolonged. Maximal increases of early apoptotic hMSCs were seen with paclitaxel at 48 h (27·6 ± 0·4%; P < 0·01), whereas no significant apoptotic effect was caused by cytarabine, dexamethasone and etoposide after 24-h exposure. High-dose dexamethasone caused only a slight increase of dead cell numbers at 48 h (7·6%). Results represent the mean of duplicate tests of one representative experiment from the three sets of MSCs.
The MSCs are an important cellular component of the bone marrow for supporting haemopoiesis (Majumdar et al, 1998; Devine & Hoffman, 2000; Majumdar et al, 2000). Combination chemotherapy makes it difficult to determine the exact effects of the individual agents. We evaluated the effect of individual chemotherapeutic agents on hMSCs in vitro, and found that five agents (paclitaxel, vincristine, etoposide high-dose cytarabine and high-dose dexamethasone) led to a more than 20% reduction of viable cell numbers at clinically relevant concentrations. Human MSCs were moderately sensitive to high-dose dexamethasone, etoposide and high-dose cytarabine (approximately 30% reduction), and were sensitive to paclitaxel and vincristine (approximately 60% reduction). However, the recovery patterns following withdrawal from these agents did not correlate to their sensitivity to chemotherapy. Full and partial recovery was seen in hMSCs treated with high-dose dexamethasone and vincristine respectively. However, sustained suppression in proliferation was observed among hMSCs treated for 3 d with paclitaxel, etoposide and high-dose cytarabine. To the best of our knowledge, such reduction of hMSCs numbers by paclitaxel, vincristine, cytarabine and etoposide and the different patterns of recovery have not been reported. We also determined whether the reduction in cell numbers by these five agents was the result of apoptosis or suppression of growth. Using flow cytometric analysis, our data suggested that chemotherapeutic agents could induce apoptosis of hMSCs. This may account for the decrease in cell viability and the clinical observation of a reduction in the number of hMSCs (Galotto et al, 1999; Banfi et al, 2001a). Our results showed that high-dose dexamethasone suppressed the growth of hMSCs with minimal evidence of apoptosis. Dexamethasone is a crucial component of the media for adipogenic, chondrogenic and osteogenic differentiation. Inducing differentiation may account for the suppression of proliferation. But at least 7–10 d of treatment is required for differentiation purpose (Cheng et al, 1994; Pittenger et al, 1999). In this short-term exposure setting, transient suppression by dexamethasone with subsequent rapid recovery of MSCs following the withdrawal of the drug is the more logical explanation.
Our results suggested that paclitaxel, etoposide and high-dose cytarabine may irreversibly damage the bone marrow microenvironment and are in accordance with published clinical reports (Carlo-Stella et al, 1997) and animal studies (Ben Ishay & Barak, 2001). As these agents are widely used in clinical practice (e.g. paclitaxel for treating breast, ovarian, lung, oesophageal, bladder and head and neck cancers; and vincristine, cytarabine and etoposide for leukaemia), attention should be paid to the marrow stromal damage caused by these agents.
On the contrary, hMSCs were more resistant to methotrexate, arsenic trioxide and alkylating agents (cyclophosphamide and busulphan) than NB-4 cells and PBMCs. Unlike other agents, cyclophosphamide requires biotransformation to become cytotoxic. Cyclophosphamide is activated by hepatic microsomal enzyme, P450 mixed-function oxidase, to two major intermediates, namely aldophosphamide and 4-hydroxycyclophosphamide. They ultimately form at least two intracellular alkylating metabolites, acrolein and phosphoramide mustard (Colvin et al, 1973). However, most cell cultures contain little, if any, cytochrome P450 mixed function oxidase metabolic capability (Gad, 2000). Although in our assessment system, the cytotoxicity of cyclophosphamide may be underestimated, a study on the effects of 4-hydroperoxycyclophosphamide, a derivative of cyclophosphamide that exhibits properties similar to those of microsomally activated cyclophosphamide, obtained similar results to ours (Siena et al, 1985). In addition, our results are also consistent with clinical evidence that no irreversible injury was caused by high-dose cyclophosphamide alone (Baran et al, 1976; Sensenbrenner et al, 1977; Territo, 1977). However, previous murine studies showed contrasting results to ours. These studies (Fried et al, 1977; Fried & Barone, 1980) suggested that cyclophosphamide and busulphan caused prolonged damage to the haemopoietic stromal function. They used a mouse model with implantation of a femur into an isogeneic host and assessed the haemopoietic stromal function by assaying the number of HSCs that were present in the marrow 6 weeks after implantation. It should be noted that the doses used in these murine studies (cyclophosphamide 500 mg/kg, busulphan 100 mg/kg) were well in excess of those used in humans on a per weight basis. Domenech et al (1994) assessed the damage caused by high-dose therapy on the stromal cell compartments using long-term marrow cultures established from patients after autologous BMT. Using multivariate analysis, they found that patients conditioned with busulphan developed a confluent stromal layer less frequently. However, this might be a result of confounding factors, such as underlying disease, previous therapy and synergistic effect with other agents involved in conditioning regimen.
The effects of these chemotherapeutic agents on the osteogenic and adipogenic capacity were also evaluated. Our experiments showed that hMSCs were able to maintain their osteogenic and adipogenic differentiation potential in vitro after treatment with various chemotherapeutic agents (data not shown).
Although the exact MSC counterpart in the BM remains uncertain, it is known that MSCs are a quiescent cell population in vivo (Gronthos et al, 2003), therefore the response of rapidly dividing MSCs in the cell culture system may not represent their actual behaviour in vivo. Some studies suggested that the stromal response in vivo to acute injury, including chemotherapy, is complex and may involve induction of differentiation, cell migration and proliferation (Islam, 1987; Gimble et al, 1996; Park et al, 1999; Almohamad et al, 2003).
In conclusion, hMSCs were relatively resistant to chemotherapeutic agents commonly used in BMT (i.e. busulphan, cyclophosphamide and methotrexate), however, they were sensitive to a panel of cytotoxic agents that included paclitaxel, vincristine, etoposide and cytarabine. Sustained suppression was observed in hMSCs following 3-d exposure to paclitaxel, etoposide and high-dose cytarabine. In contrast, full and partial recovery was seen in hMSCs treated with high-dose dexamethasone and vincristine respectively. Human MSCs maintained their osteogenic and adipogenic differentiation potential in vitro after a 3-d exposure to all chemotherapeutic agents tested. Our observation implied that the intensive chemotherapy not used in the BMT setting might be toxic to hMSCs. Understanding this variation is important when deciding the appropriate conditioning regimen for BMT. Further definition of the underlying mechanisms responsible for the differences in chemosensitivity of hMSCs to chemotherapeutic agents is important for the design of appropriate conditioning regimens in the BMT setting.
This study is supported by the Research Fund for Paediatric Oncology/Immunology/Transplant (RFPOIT no. 02-03) of the Department of Paediatrics and Adolescent Medicine, the University of Hong Kong and Transplant Training and Research Assistance Scheme (TTRAS nos 02-01-01-13-54 and 03-03-01-13-71), Queen Mary Hospital Charitable Trust. We also thank Mr. Wilfred Wong for statistical advice and Ms Olive Yu for her help in formatting the manuscript.