The pathophysiology of the myelodysplastic syndromes (MDS) is incompletely understood. Tumor necrosis factor (TNF)α levels are elevated, particularly in early-stage MDS, and apoptosis in marrow cells is upregulated. Observations in other models have shown a role for insulin-like growth factor binding protein 3 (IGFBP-3) in TNFα-mediated apoptosis. We observed increased levels of IGFBP-3 in the marrow plasma of patients with MDS (P=0.005) and hypothesized that altered IGFBP-3 levels contribute to the dysregulation of hemopoiesis in MDS by affecting proliferation and apoptosis. Western analysis of marrow plasma from MDS patients revealed an increase in the ratio of intact vs fragmented IGFBP-3 in early-stage MDS (relative to controls) that decreased with MDS disease progression, suggesting increased proteolysis with more advanced disease. Thus, these results provide evidence for dysregulation of IGFBP-3 in patients with MDS. While the data are complex, they are consistent with a modulatory effect of IGFBP-3 on hemopoiesis in MDS. Conceivably, understanding these mechanisms may allow for the development of novel therapeutic strategies.
Myelodysplastic syndromes (MDS) comprise a spectrum of hemopoietic stem cell disorders that occur predominantly in older individuals (median age at diagnosis 70 years). Patients generally present with cytopenias in peripheral blood, while the marrow is usually normo- or hypercellular. A high myeloblast count in the marrow, multiple cytopenias, and certain clonal chromosome abnormalities (chromosome 7 deletion; complex abnormalities) are associated with a poor prognosis.1, 2
The pathophysiology of MDS is incompletely understood. The discrepancy in cellularity between marrow and peripheral blood is thought to be due to excessive proliferation concurrently with upregulation of programmed cell death (apoptosis). Tumor necrosis factor (TNF)α, Fas-L, and TNF-related apoptosis-inducing ligand (TRAIL) are dysregulated in patients with MDS.3 These cytokines and their receptors influence expression of various components of the insulin-like growth factor (IGF) system, including IGF-binding protein-3 (IGFBP-3), which has been shown to be required for TNFα-induced apoptosis in some models.4 The IGF system is a key modulator in cell proliferation, cell regulation, and cell survival in various tissues.5, 6, 7 Its role in hemopoiesis is less well defined.
IGFBP-3 is the predominant IGFBP present in blood that regulates IGF availability to cells and tissues. However, IGFBP-3 has multiple functions that are tissue/cell specific and are modulated by concurrent exposure to other cytokines. IGFBP-3 enhances proliferation and protects cells from apoptotic stimuli, and also has potent antiproliferative properties, for example, against the potentially transforming effects of growth hormone and IGF-I.8 IGFBP-3 alone has little effect on growth inhibition, but it can dramatically enhance cell death responses to apoptosis-inducing agents.9, 10 Cleavage by serine proteases, including plasmin, may further shift the functional properties of IGFBP-3 from proproliferative to proapoptotic.11, 12, 13, 14 IGFBP-3 may also be able to induce apoptosis via an IGF-independent pathway.4, 15, 16, 17 We observed in ancillary studies that IGFBP-3 levels in the marrow plasma of patients with MDS tended to be higher than in age-matched controls. As TNFα levels are also elevated in MDS, and TNFα has been shown to upregulate IGFBP-3, we hypothesized that altered IGFBP-3 levels contribute to the dysregulation of hemopoiesis in patients with MDS. Furthermore, IGFBP levels have been linked to aging, and MDS is primarily a disease of older patients. Elevated levels of IGFBP-3 have been observed in cultures of fibroblasts with increasing donor age, in vitro senescence, and increasing confluency of cell cultures.18, 19, 20, 21 Several studies show tissue specificity of IGFBP expression and evidence of increased proteolytic degradation in senescent cells.22, 23 Such degradation may decrease the ability of these proteins to present IGFs to cell surface receptors19, 24, 25 and as a result may alter proliferative signals. The function of IGFBP-3 is thought to be modulated by numerous factors, in particular, the milieu in which it is acting.26, 27 These considerations are relevant in the context of MDS, where multiple growth and inhibitory molecules are dysregulated, where clonal and normal hemopoietic precursors coexist for extended periods of time, and where interactions of hemopoietic precursors with cellular and humoral components of the marrow microenvironment are involved in the disease pathophysiology (reviewed in Steensma and Tefferi28 and Vergilio and Bagg29).
The objective of this study was to determine IGFBP-3 levels in patients with MDS, and to define a potential role of this molecule in the pathophysiology of MDS. An understanding of the role of IGFBP-3 in MDS might provide further insights into the disease pathophysiology and may allow for the development of novel therapeutic strategies.
Materials and methods
Rational for using leukemic cell lines to elucidate IGFBP-3 function
We faced several difficulties in designing functional experiments with MDS cells. First, because of the characteristics of MDS, that is, increased apoptosis of normal cells, no stable MDS-derived cell lines were available. Secondly, primary cells from MDS patients are already ‘sensitized’ to apoptosis (a hallmark of the disease), and therefore, high background levels of apoptosis tended to obscure actual induced changes. Therefore, KG1a and ML-1 myeloid leukemic cell lines in addition to primary MDS marrow cells were used.
Primary marrow cells and plasma
Marrow cells and plasma were obtained from normal volunteers and from patients with MDS who had given informed consent according to the requirements of the Institutional Review Board of the Fred Hutchinson Cancer Research Center, as described.30
IGFBP-3 protein levels were determined in 19 normal bone marrows (NBM, age 27–78 years), and marrow from 49 patients with MDS (age 28–80 years). The samples from MDS patients included 23 patients with refractory anemia (RA, six with chromosome 7 deletions (RA-7)), nine with RA with excess blasts (RAEB), six with RAEB in transformation (RAEBT), and 11 with MDS, which had transformed to acute myeloid leukemia (AML). As the IGFBP-3 gene is located on chromosome 7, we included six RA-7 patients in our studies in an attempt to determine if the loss of one of the chromosomes altered plasma levels of IGFBP-3.
ML-1 and KG1a cells, and primary marrow mononuclear cells, separated by Ficoll–Hypaque density centrifugation, were cultured in 1640 RPMI (GIBCO BRL, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS), L-glutamine, and 1 mM sodium pyruvate. For analysis of secreted IGFBP-3, cells were cultured in serum-free medium (SFM) consisting of RPMI supplemented with 1 mM sodium pyruvate, L-glutamine, and 0.05% bovine serum albumin (BSA, tissue culture tested; Sigma, St Louis, MO, USA). Cells were incubated at 37°C in a water vapor-saturated 5% CO2/air atmosphere.
Glycosylated human recombinant (hr) IGFBP-3 (gly-IGFBP3) was purchased from GroPep Limited (Adelaide SA, Australia). Nonglycosylated hrIGFBP-3 (ng-IGFBP3) was purchased from Diagnostic System Laboratories (Webster, TX, USA). After considerable review of published data, we chose a concentration of 400 ng/ml of IGFBP-3 (gly or ng) for these present experiments, as this is the highest physiological level of free, unbound IGFBP-3 observed in vivo (total IGFBP-3 protein levels can vary from 100 to 8000 ng/ml depending on the tissue source). The following apoptosis-inducing ligands were used: hrTNFα (Peprotech, Rocky Hill, NJ, USA) and hrTRAIL/APO-2L, provided by Dr A Ashkenazi, Genentech (South San Francisco, CA, USA). Affinity-purified anti-IGFBP-3 antibody was purchased from Diagnostic Systems Laboratories Inc. (DSL Inc., Webster, TX, USA).
ELISA analysis of IGFBP-3 in marrow plasma
Total IGFBP-3 protein levels were determined with IGFBP-3 ELISA kits (DSL Inc.) according to the manufacturer's instructions. This kit does not detect cleaved IGFBP-3.
In vitro hematopoiesis
CD34+ cells were purified from aspirated bone marrow samples taken from healthy volunteers and patients with MDS using magnetic cell sorting beads conjugated with anti-human CD34+ antibody (Miltenyi Biotech, CA, USA) and passaged through a one-directional magnetic bead sorter (Miltenyi Biotec AutoMACS). CD34+ cells were assayed for in vitro colony formation as described.3 Gly-IGFBP3 or ng-IGFBP3 was added at 400 ng/ml of medium to cells for 72–96 h at 37°C in 5% CO2/air, cells were then washed twice with 1 × PBS, resuspended in Iscove's medium with 2% FBS, and plated in MethoCult GF+ (Stemcell Technologies, Vancouver, BC, Cananda), as described.3 Untreated cells served as a reference for each sample, and results with treated cells were compared to the reference samples whose values were set to 100%. Plates were incubated at 37°C in 5% CO2/air for 18 days. On day 18, burst-forming units-erythroid (BFU-E) and colony-forming units-granulocyte/macrophage (CFU-GM) were counted. Colonies contained a minimum of 50 cells.
Immunoblot analysis of IGFBP-3 in marrow plasma
Marrow plasma was concentrated onto 0.45 μm nitrocellulose membrane (Bio-Rad, Hercules, CA, USA), and eluted into 1 × sample buffer (0.5 M Tris, 10% glycerol, 8 M urea, and 2% SDS) by incubation at 96°C for 6 min. Each sample was separated on a 4–20% gradient Tris-HCl SDS-PAGE precast gel (Bio-Rad) and transferred onto Immuno-Blot PVDF membrane (Bio-Rad). Membranes were incubated in 10% H2O2/0.05% Tween-20/1 × PBS (TPBS) for 10 min, blocked in 5% milk for 1 h at room temperature, and incubated overnight at 4°C with 1:1000 affinity-purified goat anti-human IGFBP-3 antibody (DSL Inc.) in 5% nonfat dry milk (Biorad)/1% BSA (Gemini Bioproducts, CA, USA) in TPBS. Proteins were detected with SuperSignal Chemiluminescent Substrate (Pierce, Rockford, IL, USA) and CL-Xposure film (Pierce).
Determination of secreted IGFBP-3 from cell lines treated with TNFα
Myeloid leukemic cell lines ML-1, KG1a, K562, and U937 were plated at 1 × 106 cells/ml in SFM containing graded doses of TNFα. HS-5 stroma cells,31 grown to 90–100% confluency, washed twice with 1 × PBS, and serum-containing medium was replaced with SFM containing various concentrations of TNFα. After 24 h, conditioned medium from each sample was collected and immediately concentrated onto 0.45 μm nitrocellulose membrane.32 Secreted proteins were separated and analyzed for IGFBP-3 protein levels as described above.
Immunoblot analysis of whole-cell lysates
ML-1 cells were lysed in RIPA buffer containing protease inhibitors (Mini Complete Protease Inhibitor Tablets, Roche, Indianapolis, IN, USA) and PMSF (Sigma). Protein samples were analyzed by immunodetection as described above. The following antibodies and blocking agents were used for immunodetection: phospho-Akt (Cell Signaling Technology (CST), Beverly, MA, USA), 1:1000 in 5% BSA; BAD (BD Transduction Laboratories, Lexington, KY, USA), 1:500 in 2.5% milk; phospho-BAD (Ab-1, Ser 112) and phospho-BAD (Ab-2, Ser 136) (Oncogene Research Products, San Diego, CA, USA), 1:500 in 3% BSA; BclXL (CST), 1:1000 in 5% milk; caspase-8 (C15, Alexis, San Diego, CA, USA), 1:5000 in 5% milk; caspase-9 (CST), 1:2000 in 5% milk; cytochrome c (BD Pharmingen, San Diego, CA, USA), 1:5000 in 5% milk; FLIP (NF6, gift from P Krammer, Deutsches Krebsforschungszentrum, Heidelberg), and 1:500 in TPBS. Immunoblots were blocked with 5% milk (except when using anti-FLIP antibody which was blocked in 2% milk). Horseradish peroxidase-labeled antibodies were purchased from Pierce and used at 1:20 000 dilution in 2 or 5% milk.
Flow cytometric analysis of apoptotic cells
Cells were harvested and prepared for FACS analysis of apoptosis as described previously.33 For determination of apoptosis in CD14+ or CD34+ subpopulations among nonfractionated marrow cells, the cell suspension was incubated with R-phycoerythrin (PE)-labeled anti-CD14 antibody (Caltag) or PE-labeled anti-CD34 antibody (Caltag), respectively, for 20 min at 4°C in the dark and washed in cold Flow Buffer.33 Data were analyzed using Cell Quest software.
For all experimental data, mean and standard errors were calculated. ELISA data were analyzed using the Student's t-test. Apoptosis and proliferation data were analyzed using the Wilcoxon's test.
IGFBP-3 protein levels are increased in marrow plasma in early-stage MDS
Plasma from bone marrow aspirates from 19 normal donors and 49 patients with MDS were analyzed for total IGFBP-3 protein by ELISA. IGFBP-3 levels were highest in patients with RA, that is, the least advanced MDS patients who also showed elevated levels of TNFα and the highest rates of apoptosis.3 IGFBP-3 levels were significantly higher than among age-matched healthy volunteers (Figure 1a, P<0.005).
Increased cleavage of IGFBP-3 in marrow plasma from patients with MDS
Total levels of IGFBP-3 in marrow plasma declined toward control values with progressive stages of MDS (Figure 1b). However, as indicated in Figure 2, cleaved IGFBP-3 levels increased with more advanced MDS. Therefore, the ratio of intact IGFBP-3 to cleaved IGFBP-3 was high with early-stage MDS and declined progressively as MDS progressed.
IGFBP-3 protein secretion is regulated by TNF-α
The higher levels of IGFBP-3 protein detected in marrow plasma from patients with RA suggested a possible correlation with increased TNFα levels. To further test this possibility, ML-1 myeloid leukemic cells were treated with hrTNFα, and secreted IGFBP-3 protein was measured (Figure 3). In a TNFα dose-dependent manner, IGFBP-3 protein was upregulated. In contrast, in human stroma-derived HS-5 cells, which provide effective supportive layers for in vitro hematopoiesis, TNFα induced a downregulation of IGFBP-3.31 Thus, as suggested by observations in other models, TNFα modulated levels of IGFBP-3 in a dose- and tissue-specific pattern.
Effect of IGFBP-3 on in vitro hemopoiesis
In vitro hemopoiesis assays were carried out with purified CD34+ cells from bone marrow aspirates from normal donors and MDS patients. CD34+ cells were cultured in the presence of gly- or ng-hrIGFBP-3. Gly- and ng- IGFBP3 had no consistent effect on CFU-GM or BFU-E in CD34+ from normal samples. However, while gly-IGFBP3 either demonstrated no effect or slightly enhanced BFU-E, ng-IGFBP3-treated cells showed a marked decreased in BFU-E numbers compared to the untreated and gly-IGFBP-3-treated CD34+ cell samples from MDS patients. This is of interest as it is the erythrocyte lineage that is most frequently affected in patients with MDS.
IGFBP-3 interferes with apoptosis induced by TNFα and TRAIL
Two myeloid leukemic cell lines, ML-1 and KG1a, were tested for apoptotic responses to TNFα in the presence and absence of exogenous gly-IGFBP3. KG1a cells proved almost completely resistant to apoptosis, while ML-1 cells were apoptosis sensitive (data not shown). Therefore, subsequent experiments were carried out with ML-1 cells. To determine whether IGFBP-3 was involved in pathways other than TNFα-mediated apoptosis, we tested in parallel the effects of IGFBP-3 on TRAIL-mediated apoptosis.
ML-1 cells treated with TNFα or TRAIL in the presence of gly-IGFBP-3 showed decreases in apoptosis (relative to results without added IGFBP-3) of 20% or greater for both ligands (Figure 4a). Concurrent analysis of the effects of TNFα and TRAIL on selected components of apoptotic or proliferative signaling pathways in ML-1 cells showed striking alterations (Figure 4b). TNFα resulted in decreased levels of Akt and phosphorylated BAD, both of which were partially restored in the presence of IGFBP-3. TRAIL alone resulted in a loss of Akt, phosphorylated BAD and BclXL signals and partial cleavage of FLIPLong. In the presence of IGFBP-3, phosphorylated BAD (S136) and BclXL were partially restored, and FLIPLong cleavage was reduced. These findings are consistent with the observed decrease in apoptosis in the presence of IGFBP-3. However, the pattern of response involving pro- and antiapoptotic factors, as well as signals involved in cell proliferation, suggests a complex mechanism of IGFBP-3 action. As most reports on IGFBP-3-induced apoptosis describe the use of ng-IGFBP-3,34, 35 we compared the effects of gly- and ng-IGFBP-3 on ML-1 cells in the presence of TNFα or TRAIL. As shown in Figure 4c, both isoforms of IGFBP-3 inhibited apoptosis in response to proapoptotic signals. IGFBP-3 alone, regardless of glycosylation status, did not induce apoptosis, nor enhance proliferation. These studies suggest that IGFBP-3 functions as a regulator of cell response to external stimuli.
The objective of this study was to begin to elucidate a potential role of IGFBP-3 in the pathophysiology of MDS. IGFBP-3 protein levels were increased in early-stage MDS patient bone marrow plasma and an increase in cleaved IGFBP-3 was observed with progression of MDS to AML. These results are consistent with previous studies, which showed that myeloid leukemic cells express high levels of proteases capable of cleaving IGFBP-3.36 Thus, as MDS progresses to AML, more leukemic cells would be present to cleave IGFBP-3 and potentially alter the biologic function of IGFBP-3.
TNFα, which is increased in bone marrow plasma from early-stage MDS patients, upregulated secretion of gly-IGFBP-3 in myeloid leukemic cell cultures, further supporting a relationship between the dysregulation of IGFBP-3 and MDS. Interestingly, TNFα stimulated secretion of gly-IGFBP-3 in HS-5 cells, the stromal marrow cells that provide optimal growth conditions for the culturing of primary marrow cells harvested from MDS patients. The upregulation of gly-IGFBP-3 in response to TNFα suggests that HS-5 cells may secrete IGFBP-3 as a protective response to apoptotic effects of TNFα. The subsequent experiments described here show, indeed, that IGFBP-3 protected myeloid cells against apoptosis induced by TNFα, Fas ligand, and TRAIL.
Several reports show ng-IGFBP-3 to have a two- to three-fold higher cell binding affinity than gly-IGFBP-337, 38 and suggest that glycosylation status may play a role in the distribution of IGFBP-3 in the circulation and for its functional activities. Gly-IGFBP-3 may promote cell proliferation by the availability of IGFBP-3 to present IGFs to IGF receptors, while ng-IGFBP-3 may be taken out of circulation through cell surface binding or contribute to the growth inhibition via IGF-independent receptor activation of a growth inhibitory pathway. The IGF system is involved in erythropoiesis.39, 40 However, there is minimal data describing the function of IGFBP-3 in or on myeloid hemopoietic cells.36, 41 Preliminary in vitro hemopoiesis experiments demonstrated that ng-IGFBP-3 inhibited colony formation of BFU-E, while glyc-IGFBP-3 either showed no effect or minimally enhanced BFU-E colony numbers (data not shown). This is of interest as it is the erythrocyte lineage that is most frequently affected in MDS patients with cytopenia. These preliminary data are the first to show a link between glycosylation status and IGFBP-3 activity as no published reports indicate that glycosylation status affects IGFBP-3 function.
To determine the potential mechanism(s) by which the glycosylation status of IGFBP-3 was affecting in vitro hemopoiesis, we treated myeloid leukemic cells (KG1a and ML-1 cell lines) with the proapoptotic ligands TNFα and TRAIL in the absence and presence of IGFBP-3. In contrast to our original prediction that IGFBP-3, described as an inducer of apoptosis in other models,25, 42 would contribute to higher levels of apoptosis in MDS patients, our studies revealed a cytoprotective effect of IGFBP-3 against TNFα- and TRAIL-induced apoptosis. As we have shown that IGFBP-3 is upregulated in early-stage MDS, a preliminary step in the pathophysiology of MDS may be the generation of clonal populations with enhanced responsiveness to the cytoprotective effects of IGFBP-3. Thus, clonal cells could achieve a growth advantage over the surrounding normal cells when environmental concentrations of the apoptosis-inducing ligands TNFα, FasL, and TRAIL are increased, as observed in early-stage MDS marrow plasma.
In summary, we present data on abnormal expression and a possible role of IGFBP-3 in the regulation of apoptosis and proliferation of hemopoietic cells in patients with MDS. As TNFα is known to be upregulated in MDS and the present results show a regulatory effect of TNFα on IGFBP-3 expression, it is conceivable that IGFBP-3 levels change in response to TNFα and, thus, may be involved in the pathophysiology of MDS. If confirmed in additional studies and over a wider range of IGFBP-3 concentrations (and glycosylation patterns), insights from this work may point toward new therapeutic strategies to treat hemopoietic failure and possibly interfere with disease evolution in patients with MDS.
We thank Dr Deborah Banker for providing the ML-1 and KG1a cell lines and Dr Beverly Torok-Storb for providing the HS-5 stroma cell line. We also thank Dr Yansong Gu, Dr Uwe Platzbecker, Dr Jeffrey Schwartz, Dr Martin Benesch, and Cassandra Beckham for helpful discussions. We thank Kathleen Haugk for immunoblot analysis of phosphorylated IGF receptor in ML-1 cells. This work was supported by PHS Grant CA87948 and HL36444 Grant.