Thrombopoietin Supports the Continuous Growth of Cytokine-Dependent Human Leukemia Cell Lines HG Drexler, M Zaborski and H Quentmeier

Home , Thrombopoietin

Thrombopoietin supports the continuous growth of cytokine-dependent human leukemia cell lines HG Drexler, M Zaborski and H Quentmeier

DSMZ-German Collection of Microorganisms and Cell Cultures, Department of Human and Animal Cell Cultures, Mascheroder Weg 1 B, D-38124 Braunschweig, Germany

Hematopoiesis is a complex process of regulated cellular pro- nate membrane receptor. This binding triggers a series of intra- liferation and differentiation from the primitive stem cells to the cellular mediators involved in the growth factor’s signaling final fully differentiated cell. The long and extensive search for a factor specifically regulating megakaryocytopoiesis led to the pathways. Recently, a novel hematopoietic growth factor, cloning of a hormone, here called thrombopoietin (TPO), that termed thrombopoietin (TPO), was cloned and shown to be a specifically promotes proliferation and differentiation of the megakaryocytic lineage-associated growth and differentiation megakaryocytic lineage. The availability of recombinant TPO factor. Binding of TPO to its receptor, c-MPL, mediates plei- and its imminent clinical use has made a more detailed under- otropic effects on megakaryocyte development in vitro and in standing of its effects on hematopoietic cells more urgent. Nor- vivo. TPO is clearly the primary regulator of this cell lineage mal megakaryocyto- and thrombopoiesis occurs predomi- nantly in the bone marrow, a difficult organ to study in situ, acting at all levels of megakaryocytopoiesis and thrombopo- particularly in humans, due to the low numbers of megakary- iesis (reviewed in Ref. 1). ocytic progenitors and the consequent difficult isolation as The availability of TPO will be of considerable clinical pure populations. Thus, we developed an in vitro system which importance for the treatment of thrombocytopenias following may allow us to address questions regarding the biology of high-dose chemotherapy (of solid tumors and hematological TPO. The acute myeloid leukemia (AML)-derived cell lines HU- malignancies), irradiation (for instance, in the context of bone 3, M-07e, M-MOK and TF-1 have absolute dependence on gra- marrow transplantation) and other bone marrow failures (eg nulocyte–macrophage colony-stimulating factor (GM-CSF). We 2 cultured these cells long term (Ͼ6 months) in the continuous due to myelodysplastic syndromes, aplastic anemias). Still, presence of TPO (omitting GM-CSF). TPO alone supported the while TPO has moved tremendously quickly from theory to maintenance and expansion of these sister cell lines, HU- clinical trials in just about 2 years, more insight into the physi- 3/TPO, M-07e/TPO, M-MOK/TPO and TF-1/TPO, that displayed ology and pathophysiology of TPO, its effects and its regu- somewhat longer doubling times, a larger cell size, and a higher lation, is clearly required. To that end, it is absolutely essential percentage of polynucleated giant cells and slightly adherent cells than the corresponding countercultures grown with GM- that both in vivo and in vitro TPO research will be performed. CSF. In the absence of TPO the cells died quickly, within a few Studies on megakaryocytopoiesis using bone marrow speci- days; thus, the TPO-grown cell lines have an absolute depen- mens are hampered by the paucity of megakaryocytic pro- dence on this factor, but could all be switched back to growth genitors and more committed cells within the total population with GM-CSF. In comparison with the GM-CSF-treated cells, the (requiring sophisticated purification steps) and the striking het- receptors for GM-CSF and interleukin-3 (IL-3) were down-regu- erogeneity of the bystander cells. Continuous cell lines, usu- lated and the receptors for stem cell factor (SCF) and TPO were up-regulated in the TPO-exposed cells. A short-term prolifer- ally derived from leukemias involving this cell lineage, rep- ation assay showed a stronger response of the TPO-cell lines resent excellent alternative in vitro models due to their to erythropoietin, GM-CSF, IL-3, PIXY-321, SCF and TPO than monoclonal homogeneity and unlimited availability. How- the GM-CSF-cell lines. Flow cytometric analysis of the GM-CSF- ever, only a few cell lines appear to respond to TPO, at least and TPO-cultured lines displayed an up-regulation of the in terms of enhanced proliferation or induced differen- megakaryocytic surface markers CD41, CD42 and CD61, and a tiation.1,3 In order to facilitate future investigations in TPO, down-regulation of the erythroid marker glycophorin A in the latter cell lines, suggesting some differentiation along the we attempted to establish and characterize cell lines that are megakaryocytic lineage. Thus, in long-term exposure, TPO absolutely dependent on TPO for growth and survival. These appears to have both a proliferative and a differentiative effect TPO-dependent cell lines provide an experimental setting in on responsive cells. Under serum-deprived culture conditions, which the study of TPO regulation and its effects can be TPO acted as a survival factor on the TPO-cell lines. Taken adequately performed. Furthermore, AML M7-derived cell together, these findings indicate that the TPO-dependent cell lines as in vitro systems of human megakaryocytes will con- lines represent important biological reagents for further characterization of the biology of TPO and should also provide tinue to be useful for the evaluation of any aspect of megakar- a great aid for future in vitro experiments aimed at elucidating yocyto- and thrombopoiesis. megakaryocyto- and thrombopoiesis. Keywords: leukemia; cell lines; TPO; cytokines Materials and methods

Introduction Leukemia cell lines

Regulation of blood cell production, differentiation, and cer- The continuous cell lines were taken from the stock of the cell tain functional responses is mediated in part by speciﬁc hema- bank of the DSMZ (German Collection of Microorganisms and topoietic growth factors (or cytokines). The initial event in Cell Cultures)4,5 or were generously provided by the investi- growth factor action consists of the factor’s binding to its cog- gators who established the cell lines for research purposes (Table 1): HU-3 (kindly provided by Dr D Morgan, Philadel- phia, PA, USA);6,14 M-07e (DSM ACC 104);7 MB-02 (Dr D Correspondence: HG Drexler Morgan);8 M-MOK (Dr T Kudo, Sendai, Japan);9 MUTZ-2 Received 23 October 1996; accepted 17 December 1996 (DSM ACC 271);10 MUTZ-3 (DSM ACC 295);10 OCI-AML-1 TPO-dependent leukemia cell lines HG Drexler et al 542 Table 1 Leukemia cell lines used

Cell line Ref. Phenotype of cell linea Year of Original Tissue of Age/Sex 48 h Incubation with establishment diseaseb originb (patient) proliferative reagents during response to cytokine switchingd TPOc

HU-3 6 megakaryocytic/erythroid 1991 AML M7 BM 69 F 11.4 GM-CSF vs TPO M-07e 7 megakaryocytic 1987 AML M7 BM 0.5 F 8.4 5637 CM vs TPO MB-02 8 megakaryocytic/erythroid 1986 AML M7 PB 70 M 1.1 GM-CSF vs TPO M-MOK 9 megakaryocytic 1989 AML M7 BM 1 F 11.0 GM-CSF vs TPO MUTZ-2 10 myeloid 1993 AML M2 PB 62 M 1.5 SCF vs TPO MUTZ-3 10 monocytic 1993 AML M4 PB 29 M 2.2 5637 CM vs TPO OCI-AML-1 11 myeloid 1987 AML M4 PB 73 F 3.1 5637 CM vs TPO OCI-AML-5 11 myeloid 1990 AML M4 PB 77 M 1.5 5637 CM vs TPO TF-1 12 erythroid 1987 AML M6 BM 37 M 14.5 5637 CM vs TPO UT-7 13 megakaryocytic/pluripotent 1988 AML M7 BM 64 M 1.4 GM-CSF vs TPO

aPhenotype of the cell lines based on the expression of immunological surface markers, capability to differentiate upon stimulation, functional features, etc. bAs described in the original publication on the respective cell line: AML: acute myeloid leukemia (with the morphological subtypes of M2 for myeloblastic, M4 for myelomonocytic, M6 for erythroid and M7 for megakaryocytic); BM: bone marrow; PB: peripheral blood. cCells were extensively washed and then incubated in the absence or presence of 100 U/ml TPO for 48 h; 3H-thymidine was added for the last 4 h of incubation; 3H-thymidine incorporation was determined by standard ␤ scintillation counting; stimulation indices (SI) refer to the untreated control culture; an SI Ͼ2 is considered to be signiﬁcant. dThe cytokines were used at the following concentrations: GM-CSF (5 ng/ml); SCF (50 ng/ml); TPO (20 ng/ml); 5637 CM (10% vol).

(Dr EA McCulloch, Toronto, Canada);11 OCI-AML-5 (DSM 10 cell lines were washed, divided into two aliquots (starting ACC 247);11 TF-1 (DSM ACC 334);12 UT-7 (Dr M Souyri, Paris, concentration of 0.5 × 106 cells/ml in 1 ml) and incubated in France).13 They were grown at 37°C in a humidiﬁed atmos- medium containing one of the following cytokines or CM:

phere of air containing 5% CO2. The basal growth media 5637 CM (at 10% vol); GM-CSF (at 5 ng/ml; kindly provided (RPMI 1640 or alpha-MEM; Gibco BRL, Eggenstein, Germany) by Dr A Mire-Sluis, London, UK); SCF (at 50 ng/ml; kindly were supplemented with 5–20% heat-inactivated (at 56°C for provided by Dr A Mire-Sluis); or TPO (at 20 ng/ml; kindly pro- 45 min) fetal bovine serum (FBS) (Sigma, Deisenhofen, vided by Amgen, Thousand Oaks, CA, USA). The optimal con- Germany). For selective experiments, cells were incubated in centration of each cytokine was determined empirically by the serum-free medium (Macrophage-SFM Medium; Gibco BRL). 3H-thymidine uptake as outlined below. During the first week, Freedom of mycoplasma contamination was determined after the cells were weaned off the initially added cytokine by pro- thawing the cell lines by cultivation on agar and by frequent gressively decreasing the concentrations of this reagent while DAPI staining.15 Cultures were passaged according to stan- simultaneously adding and increasing daily the new cytokine. dard procedures: spent culture medium of suspension cell lines was exchanged at regular intervals (usually after 24– 48 h); confluent adherent monolayers of cell line 5637 (DSM Immunofluorescence analysis of antigen cell surface ACC 35) were detached and split using trypsin-EDTA (Gibco expression BRL).5 The growth factor-dependent cell lines were cultured routinely with recombinant cytokines or with supernatant (10– Surface protein expression of the following four cytokine 20% v/v) from the bladder carcinoma cell line 5637 produc- receptors (R) was analyzed by flow cytometry applying ing several cytokines including granulocyte colony-stimulat- specific monoclonal antibodies (McAbs): anti-GM-CSFR␣ ing factor CSF (G-CSF), granulocyte–macrophage CSF (GM- (CDw116; a mouse IgG1McAb; from Immunotech, Hamburg, CSF), macrophage-CSF (M-CSF) and stem cell factor (SCF), but Germany); anti-Kit (CD117; a mouse IgG1 McAb; from not interleukin-3 (Il-3).16 All cell lines were examined daily in Immunotech); anti-IL-3R␣ (CD123; a mouse IgG1 McAb; from their culture vessels under an inverted microscope. Cells were R&D Systems, Wiesbaden, Germany); and anti-MPL (a mouse harvested in their logarithmic growth phase with viabilities IgG1 McAb; from Genzyme, Ru¨sselsheim, Germany).17 Sev- exceeding 90% as determined by trypan blue dye exclusion; eral cell lineage-associated differentiation surface markers the cells were processed immediately or stored as pellets at were examined with the following McAbs: anti-gpIIb (CD41; −20°Cor−80°C. The morphology of the cells was evaluated a mouse IgG1 McAb; from Immunotech); anti-gpIb (CD42b; in May–Gru¨ nwald–Giemsa-stained cytospins prepared in a a mouse IgG1 McAb; from Immunotech); anti-gpIIIa (CD61; Shandon Cytospin 2 (Shandon, Frankfurt, Germany). The cell a mouse IgG1 McAb; from Dako, Hamburg, Germany); anti- counts and viabilities were examined in standard hematocyto- glycophorin A (VIE-G4; a mouse IgM McAb; kindly provided meters. by Prof W Knapp, Vienna, Austria); anti-HLA-DR (RFDR-2; a mouse IgG1 McAb; kindly provided by Prof G Janossy, Lon- don, UK). Non-reactive isotype-specific reagents were applied Establishment of TPO dependent cell lines as controls. Binding of McAbs was assessed by the indirect immunofluorescence technique using fluorescein isothiocyan- The original cultures of these 10 cell lines (Table 1) had all ate-conjugated goat anti-mouse immunoglobulin heavy been maintained in 10–20% vol 5637 conditioned medium chain-specific antisera (Dunn, Asbach, Germany). Distri- (CM), the main cytokine ingredients of which are G-CSF bution of antigens was analyzed by flow cytometry (FACScan; (40 ng/ml) and GM-CSF (2 ng/ml).16 The original cells of the Becton Dickinson, Heidelberg, Germany). Addition of propid- TPO-dependent leukemia cell lines HG Drexler et al 543

Figure 1 Long-term culture of growth factor-dependent cell lines with GM-CSF, SCF, 5637 CM or TPO. Aliquots of the megakaryocytic– myeloid–monocytic leukemia cell lines MB-02, UT-7, MUTZ-2, MUTZ-3, OCI-AML-1 and OCI-AML-5 were cultured under identical conditions in the presence of either 10% (v/v) 5637 CM (containing mainly GM-CSF, G-CSF and SCF), 5 ng/ml GM-CSF, 50 ng/ml SCF or in the presence of 20 ng/ml TPO (starting concentration of 0.5 × 106 cells/ml in 1 ml). Prior to the start of this culture, cells now grown with TPO had been slowly weaned of the original cytokine (GM-CSF, SCF or 5637 CM) during the simultaneous, slowly increasing addition of TPO over 1–2 weeks. Viable cell numbers were determined at the indicated time-points in a hematocytometer. At the days indicated, the cultures were split 1:4, spent culture supernatants were exchanged at regular intervals without loss of any cells in order to expand maximally the cell populations. Note the lack of cell growth for the TPO-exposed cultures except for OCI-AML-1 where there was a steady increase in cell numbers in the TPO-culture starting at about day 21. TPO-dependent leukemia cell lines HG Drexler et al 544

Figure 2 Establishment of TPO-dependent cell lines. Cultures were set up as described in Figure 1. While growth of the GM-CSF- and TPO- treated cultures of HU-3, M-07e, and M-MOK was similar or identical, proliferation of the TF-1/TPO cell line started only very slowly at about day 15–17.

Table 2 Surface expression of cytokine receptors and differentiation markers on the GM-CSF- and TPO-dependent cell lines

Cell line CDw116 CD123 CDw117 MPL CD41 CD42 CD61 Glycophorin HLA-DR FSCa Doubling time (GM-CSFR␣) (IL-3R␣) (c-kit) (TPO-R) (gpIIb/IIIa) (gpIb) (gpIIIa) A

HU-3/GM 97b 95 85 14 52 16 99 10 70 160c 60 ± 11d HU-3/TPO 92 89 99 31 72 21 99 8 47 146 65 ± 15 M-07e/GM 98 89 100 22 50 89 98 45 0 118 46 ± 12 M-07e/TPO 84 64 100 30 56 96 97 43 0 119 59 ± 10 M-MOK/GM 95 75 99 37 60 77 98 18 0 104 45 ± 6 M-MOK/TPO 76 73 99 60 91 99 100 4 0 115 54 ± 11 TF-1/GM 92 63 56 18 97 39 100 18 99 140 39 ± 6 TF-1/TPO 86 36 69 26 98 30 100 6 99 155 74 ± 5

aForward scatter indicating the size of the cells. bPercentages of positive cells as detected by ﬂow cytometry. Cells were tested at days 87–108. Similar results were obtained at earlier passages (days 68–83). Differences of 5% or more between the GM-CSF- and TPO-treated cultures were considered as signiﬁcant. See also Figure 4. cMean channel on an arbitrary scale. dDoubling times are given in hours; cells were incubated with 10 ng/ml GM-CSF or 10 ng/ml TPO.

ium iodide (10 ␮g/ml; Sigma) allowed gating out of dead cells. and ␤-scintillation counting (LKB 1209 Rackbeta; Pharmacia, Sizes of the cells were determined by their forward light scat- Freiburg, Germany). The cells were seeded in triplicate in tering properties. 100 ␮l medium in ﬂat-bottomed 96-well plates and incubated in the absence or presence of cytokines; for the last 4 h of the 48-h incubation period, 1 ␮Ci [methyl-3H]thymidine Proliferation assay (Amersham-Buchler, Braunschweig, Germany) was added to each well. Cells were seeded at 2.5 or 5 × 105 cells/ml. All The proliferative response of the cell lines to the various effec- 3H-thymidine incorporation assays were performed at least tors was examined by standard 3H-thymidine incorporation twice on different occasions. The following commercially sup- TPO-dependent leukemia cell lines HG Drexler et al 545

Figure 3 Dependency of cell lines on GM-CSF or TPO for survival and proliferation. At days 97–118, aliquots of the four GM-CSF- or TPO- dependent cell lines were incubated with either 5 ng/ml GM-CSF, 20 ng/ml TPO or medium only according to their previous treatment (GM- CSF-variants received GM-CSF and TPO-variants received TPO) or switched after thorough washings (but without any progressive adaptation period) to TPO or GM-CSF, respectively. Both GM-CSF- and TPO-variants died quickly in the absence of any growth factor. TPO-variants switched to GM-CSF showed vigorous and immediate growth. GM-CSF-variants switched to TPO grew poorly during this ﬁrst week of exposure to TPO. Similar results were obtained at earlier passages of these cell lines (at days 36–46).

Figure 4 Flow cytometric analysis of surface proteins. Cultures of M-MOK cells were grown continuously with GM-CSF or TPO and were analyzed for MPL, CD41, CD42 and CD61 expression on day 87. Open curves, GM-CSF-treated cells; solid curves, TPO-treated cells. Isotype- speciﬁc irrelevant antibodies were used as negative controls (not shown). Note the up-regulation of the expression of all four proteins on the TPO-exposed cells in comparison with the GM-CSF-treated cells. TPO-dependent leukemia cell lines HG Drexler et al 546

Figure 5 Morphology of TPO-dependent cell lines. Cytospin slide preparations of HU-3/TPO (a), M-07e/TPO (b), M-MOK/TPO (c) and TF- 1/TPO (d) cells were stained with May–Gru¨nwald–Giemsa stain (micro-photographs were taken with a Nikon Labophot photomicroscope using oil immersion and a ×63 objective). Note the prominent membrane blebbing of the cells.

plied recombinant cytokines were used (in parentheses are the electrophoresis using the Protean II chamber (BioRad, concentrations at which they were applied and the speciﬁc Mu¨nchen, Germany) and were blotted onto nitrocellulose activities of the reagents): erythropoietin (EPO; 5 U/ml; membranes (Trans-Blot Transfer Medium; BioRad) by electro- Ͼ70 000 U/mg); G-CSF (100 U/ml = 10 ng/ml; Ͼ1.0 × 107 blotting (Sartoblot II-S; Sartorius, Go¨ttingen, Germany). For U/mg); GM-CSF (100 U/ml = 10 ng/ml; Ͼ1.0 × 107 U/mg); IL- STAT 3 and STAT 5 detection, the membranes were labelled 3 (100 U/ml = 10 ng/ml; Ͼ1.0 × 107 U/mg); SCF (50 ng/ml; with the respective murine McAbs (Dianova, Hamburg, 0.5–1.0 × 106 U/mg) (Boehringer Mannheim, Germany; R&D Germany) and speciﬁc bands were visualized applying a Systems). Recombinant human TPO (used at 100 U/ml) was biotin/horseradish peroxidase system (Amersham) in combi- kindly provided by Zymogenetics (Dr DC Foster, Seattle, WA, nation with the Renaissance Western Blot Chemoluminesc- USA).18 Recombinant PIXY-321 (a GM-CSF/IL-3 fusion pro- ence Reagent protocol (Du Pont NEN, Bad Homburg, tein; used at 10 ng/ml) was kindly provided by Immunex Germany). (Seattle, WA, USA). As the cell lines were constitutively dependent on externally added growth factors and normally grown in medium containing such factors, the cells were Results washed extensively prior to the experiments. Establishment of TPO-dependent cell lines

Western blot analysis In order to establish TPO-dependent sublines, we cultured cells from 10 constitutively growth factor-dependent leukemia For Western blot analysis, 2 × 106 cells were pelleted, resus- cell lines (Table 1) in the presence of 20 ng/ml TPO and the pended in 25 ␮l protease inhibitor buffer containing 0.1 M respective control cultures in GM-CSF, SCF or 5637 CM for NaCl, 10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 1 ␮g/ml aproti- more than 5 weeks (Figures 1 and 2). In previous studies,1,3 nin and 100 ␮g/ml phenylmethylsulfonyl ﬂuoride, and lysed we found that the cell lines HU-3, M-07e, M-MOK, MUTZ- by boiling in 25 ␮l SDS-PAGE (polyacrylamide gel) sample 3, OCI-AML-1 and TF-1 responded proliferatively in a short- buffer containing 15% glycerol, 125 mM Tris-HCl (pH 6.8), term (48 h) thymidine uptake assay to incubation with TPO (a 5mM EDTA, 2% SDS, 0.1% bromphenol blue and 1% ␤-mer- response with a stimulation index Ͼ2 was considered captoethanol. Proteins were separated in a 9% SDS-PAGE by signiﬁcant) (Table 1). Of the remaining four cell lines, two cul- TPO-dependent leukemia cell lines HG Drexler et al 547

Figure 6 Western blot analysis of STAT 3 and STAT 5 expression by the GM-CSF- and TPO-treated cell lines. Proteins from cell lysates of M-MOK (lanes 1 and 2), TF-1 (lanes 3 and 4), HU-3 (lanes 5 and 6) and M-07e (lanes 7 and 8) were separated by PAGE and blotted onto nitrocellulose membranes. Even numbered lanes show STAT expression of TPO-treated cells, odd numbered lanes show expression of cells grown with GM-CSF. Positive controls for STAT 3 and STAT 5 proteins were included on the same blot. Molecular weight estimates were based on MW markers (left).

Table 3 Proliferative response of the GM-CSF- and TPO-dependent cell lines to various cytokines

Cell line 5637 CM EPO G-CSF GM-CSF IL-3 PIXY-321 SCF TPO

HU-3/GM 972a 639 156 2782 1971 2915 1357 711 HU-3/TPO 1113 954 107 4014 3442 4552 2942 1430 M-07e/GM 554 108 91 1012 881 990 805 411 M-07e/TPO 977 112 102 1916 1507 2289 1371 1404 M-MOK/GM 1272 117 140 2475 2186 3142 5026 1284 M-MOK/TPO 1015 91 99 1940 2026 2604 2465 1624 TF-1/GM 948 450 98 960 908 948 426 312 TF-1/TPO 1094 650 95 1554 1264 1566 598 664 aResults are expressed as percentages of the respective untreated control cells. Cells were tested at days 81–110. Similar results were obtained at earlier passages (day 68). tures, MB-02 and UT-7, were chosen for this study because lines were kept in continuous culture for more than 6 months they are of megakaryocytic origin and display markers of this without loss of their dependency on TPO. We confirmed the cell lineage; the myeloid cell lines MUTZ-2 and OCI-AML-5 dependency of these newly established cell lines on TPO for were used as controls. MB-02, MUTZ-2, MUTZ-3, OCI-AML- cell survival and growth and thus excluded the outgrowth of 5 and UT-7 clearly did not grow in the continuous presence a cytokine-independent subclone: in the presence of the of TPO while vigorously growing in the presence of GM-CSF, respective cytokine (GM-CSF or TPO), the HU-3, M-07e, M- SCF or 5637 CM (Figure 1). The three cell lines HU-3, M-07e MOK and TF-1 cells proliferated normally, but stopped pro- and M-MOK grew with TPO as well as they did with GM-CSF liferation and eventually died when the cytokines were omit- or 5637 CM whereas OCI-AML-1 and TF-1 showed a some- ted from the culture medium (Figure 3). When the cells of the what delayed, but still significant proliferation upon exposure four TPO-dependent variants were switched back to GM-CSF to TPO (Figures 1 and 2). These long-term cultures were exposure (without an adaption period), the cells grew immedi- named in accordance with the respective cytokine, for ately well or even better than with TPO in all four instances; instance, HU-3/GM vs HU-3/TPO (Table 2). In the case of in contrast, cultures continuously incubated with GM-CSF and OCI-AML-1, it appeared that a cytokine-independent sub- switched to TPO (in the absence of GM-CSF), did not show clone had grown out as cytokine-deprived OCI-AML-1/TPO immediately vigorous growth (Figure 3). We determined the cells grew as well as their sister cells incubated with either doubling times of the cell lines by incubating them with GM-CSF or TPO. Thus, all further experiments were perfor- 10 ng/ml GM-CSF or 10 ng/ml TPO: the TPO-variants showed med with the GM-CSF- and TPO-dependent variants of HU- somewhat longer doubling times (Table 2). 3, M-07e, M-MOK and TF-1. These four TPO-dependent cell TPO-dependent leukemia cell lines HG Drexler et al 548 while expression of the erythroid marker glycophorin A was decreased in 2/4 cell lines. M-MOK/TPO and TF-1/TPO cells were significantly larger than their GM-CSF-counterparts in the forward scatter analysis; while the GM-CSF- and TPO-variants of M-07e were of similar size, only the HU-3/GM cells were larger than their TPO-treated sister cells.

Morphological analysis

Both the GM-CSF- and the TPO-exposed variants of the cell lines HU-3 and M-MOK grew as single cells in suspension with only few cells forming small aggregates. Compared with their GM-CSF-treated sister cultures, significantly higher percentages of the M-07e/TPO and TF-1/TPO cells adhered to the plastic bottom of the plates with long cellular extensions. Of note also are the giant, polynucleated cells that were seen in the TF-1/TPO cultures accounting for about 1–3% of the population; none of the other cultures, neither the GM-CSF- nor the TPO-treated lines, displayed significant numbers of these giant cells. With regard to the morphological appear- ance of the May–Gru¨ nwald–Giemsa-stained mononucleated cells, there were no significant differences between the respective pairs of GM-CSF- and TPO-incubated cells. The GM-CSF- and the TPO-treated cells displayed very similar or identical morphological features: the surface of the cells was highly deformed with cytoplasmic extrusions, reminiscent of the physiological release of platelets from the periphery of the cell (‘membrane blebbing’) (Figure 5).

Expression of STAT 3 and STAT 5 molecules

The long-term cultivation of the four cell lines with either GM- CSF or TPO did not alter the pattern of STAT 3 and STAT 5 expression (Figure 6). All variants showed constitutively strong expression of both proteins.

TPO as survival factor Figure 7 TPO as survival factor for TPO-dependent cell lines cultured in serum-free medium. HU-3/TPO, M-MOK/TPO and TF-1/TPO We next evaluated the ability of TPO to protect the four TPO- cells were incubated in serum-free medium in the presence of dependent cell lines under serum-free conditions (Figure 7). 50 ng/ml TPO (TPO50), 20 ng/ml (TPO20), 5 ng/ml GM-CSF or in the absence of cytokines. Viable cell numbers were determined at the TPO could clearly extend the survival of HU-3/TPO, M- indicated time-points. Note the quick deterioration of the control cul- 07e/TPO and M-MOK/TPO. TF-1/TPO cells even showed tures and the survival or even growth of the TPO-exposed cultures. some degree of proliferation in the serum-free medium. Similar results were obtained for M-07e/TPO cells (not shown).

Surface expression of cytokine receptors and Mitogenic effects of several cytokines differentiation markers As shown in Table 3, the TPO-treated variants responded pro- Using flow cytometry, we examined the GM-CSF- and TPO- liferatively to the same cytokines as their parental GM-CSF- treated cell lines for differences in the expression of several treated cells. Apart from TPO, 5637 CM, EPO (only on HU- cytokine receptors and surface proteins associated with 3 and TF-1), GM-CSF, IL-3, PIXY-321 and SCF had mitogenic erythroid and megakaryocytic differentiation (Table 2, effects on these cell lines; G-CSF did not stimulate growth in Figure 4). We considered a (reproducible) 5% difference in any of these cultures. The TPO-exposed variants of HU-3, M- antigen expression to be significant. In comparison with the 07e and TF-1 showed higher 3H-thymidine incorporations GM-CSF-dependent variants, the GM-CSFR and IL-3R were relative to the respective untreated control cells than the sister down-regulated in 4/4 and in 3/4 TPO-dependent cell lines, cultures grown in GM-CSF. The reverse, a stronger proliferat- respectively; c-kit and MPL were up-regulated in 2/4 and 4/4 ive response of the GM-CSF-variant to incubation with 5637 cell lines, respectively; levels of the megakaryocytic markers CM, GM-CSF, IL-3, PIXY-321 and SCF was seen for M-MOK. CD41/CD42 and CD61 were increased in 3/4 (percentages of Dose–response curves demonstrated a significantly stronger positive cells) and 2/4 cell lines (intensity of expression proliferative stimulus of TPO on the TPO-cell lines than on according to the mean fluorescence channel), respectively, the GM-CSF-cell lines (Figure 8). TPO-dependent leukemia cell lines HG Drexler et al 549

Figure 8 Dose–response curves to TPO of the GM-CSF- and TPO-dependent variants. Cells were incubated in 96-well plates with decreasing concentrations of TPO for 48 h. Proliferation was measured by the 3H-thymidine assay. Results are expressed as percentages of thymidine incorporation in comparison with the respective untreated control cells. Note the signiﬁcantly stronger cell proliferation induced by TPO in the TPO-variants than in the GM-CSF-variants.

Discussion models for gaining insights into the molecular genetic regulation of differentiation and proliferation events has become Studies on hematopoietic stem cells, precursor cells and their evident in the plethora of studies employing leukemia cell immature progeny are usually hampered by the paucity of lines.4,19 these cell elements in the normal hematopoietic tissues and Thus, in this study TPO was used to establish TPO-dependent the necessity for extensive purification procedures. Therefore, continuous human leukemia-derived cell lines. The four lines such studies often take recourse to continuous hematopoietic (HU-3/TPO, M-07e/TPO, M-MOK/TPO and TF-1/TPO) derived cell lines established from patients with leukemia which from constitutively growth factor-dependent cell lines were con- recapitulate many of the developmental steps of normal cells. tinuously grown for more than 6 months in TPO-supported While there are certainly disadvantages of the work with cell liquid culture. The cells are TPO-dependent rather than merely lines and not every result should be uncritically extrapolated TPO-responsive as omission of TPO in the cell culture com- to the normal situation, leukemia-derived cell lines also pos- pletely abrogated cell proliferation. While the swift change from sess on the other hand some indisputable advantages, not the incubation with GM-CSF to TPO was not successful, rather least being their unlimited availability, monoclonality, and requiring a certain adaptation period, all TPO-treated cell lines ready accessability. Thus, these cultures are amenable to vari- couldbeeasilyswitchedbacktoGM-CSF. ous in vitro manipulations. Again, without ignoring the limi- Under serum-free conditions TPO promoted cell viability by tations of these in vitro models (it might be argued that any repressing apoptosis in the TPO-dependent cultures. Thus, in vitro scenario is artificial), cell lines have certainly provided depending on the experimental setting (without or with FBS- unique and important models to perform any kind of studies containing culture medium – there are probably co-stimulatory on hematopoietic cells.19 factors in the FBS), TPO can act as a survival factor or as a TPO has a tremendous clinical potential and there is an proliferation-inducing agent.20 The survival signal provided by explosion of research activities in this field. However, it is TPO is, however, not specific as GM-CSF also rescued TPO- often difficult to evaluate the significance of these obser- treated cells from apoptosis upon withdrawal of TPO. vations in vivo because of the scarcity of the cells of interest. While GM-CSF and IL-3 receptors were down-regulated in Therefore, we felt it was important to generate a stable, repro- the TPO-treated cells, c-kit (SCF receptor) and MPL (TPO ducible, accessible and significant in vitro system that is suf- receptor) were up-regulated. The cytokines EPO, GM-CSF, IL- ficiently representative to perform the extensive examinations 3, PIXY-321, SCF and TPO had stronger growth-stimulating of the biological functions of TPO. The utility of such cell line effects on three or all four of the TPO-dependent cell lines TPO-dependent leukemia cell lines HG Drexler et al 550 than on their GM-CSF-exposed counterparts. The immuno- In: Hay RJ, Park JG, Gazdar A (eds). Atlas of Human Tumor Cell phenotypic shift of the TPO-cell lines to a stronger expression Lines. Academic Press: Orlando, 1994, pp 213–250. of the megakaryocytic surface markers CD41, CD42 and 5 Drexler HG, Dirks W, MacLeod RAF, Quentmeier H, Steube K (eds). DSM Catalogue of Human and Animal Cell Lines, 5th edn. CD61, accompanied by the down-regulation of the erythroid Braunschweig, Germany, 1995. lineage-associated surface protein glycophorin A, suggests dif- 6 Morgan D. Differential cytokine response of human lineage spe- ferentiation along the megakaryocytic cell axis induced by the cific progenitor cell lines. Blood 1993; 82: 373a. long-term exposure to TPO. Other mature megakaryocytic 7 Avanzi GC, Lista P, Giovinazzo B, Miniero R, Saglio G, Benetton properties such as platelet factor-4 and ␤-thromboglobulin G, Coda R, Cattoretti G, Pegoraro L. Selective growth response to have not been tested. IL-3 of a human leukaemic cell line with megakaryoblastic features. Br J Haematol 1988; 69: 359–366. The STAT (signal transducers and activators of transcription) 8 Morgan DA, Gumucio DL, Brodsky I. Granulocyte–macrophage proteins are a new family of transcription factors involved in colony-stimulating factor-dependent growth and erythropoietin- growth factor and hormone signal transduction.21 STAT 3 and induced differentiation of a human cell line MB-02. Blood 1991; STAT 5 have been described as being involved in the signal 78: 2860–2871. transduction pathway activated by TPO.1 Here, we found that 9 Itano M, Tsuchiya S, Minegishi N, Fujie H, Minegishi M, Morita both STAT 3 and STAT 5 proteins are constitutively expressed S, Yambe T, Ohashi Y, Masuda T, Koike T, Konno T. Establishment in all cell lines and at similar intensities in the GM-CSF- and and characterization of a novel human immature megakaryoblastic leukemia cell line, M-MOK, dependent on fibroblasts for its TPO-treated cells. Clearly, these cells provide attractive mod- viability. Exp Hematol 1995; 23: 1301–1309. els for studying the role of these and other pathways in cyto- 10 Hu ZB, Ma W, Zaborski M, MacLeod RAF, Quentmeier H, Drexler kine signaling. HG. Establishment and characterization of two novel cytokine- UT-7 is an AML M7-derived cell line with absolute depen- responsive acute myeloid and monocytic leukemia cell lines, dence on IL-3, GM-CSF or EPO.13 We were not able to adapt MUTZ-2 and MUTZ-3. Leukemia 1996; 10: 1025–1040. the UT-7 cells grown under our culture conditions to prolifer- 11 Wang C, Koistinen P, Yang GS, Williams DE, Lyman SD, Minden MD, McCulloch EA. Mast cell growth factor, a ligand for the ate during supplementation with exogenous TPO only (in the receptor encoded by c-kit, affects the growth in culture of the blast absence of GM-CSF); normally, UT-7 cells also do not cells of acute myeloblastic leukemia. Leukemia 1991; 5: 493–499. respond proliferatively to short-term incubation with TPO in 12 Kitamura T, Tojo A, Kuwaki T, Chiba S, Miyazono K, Urabe A, the 3H-thymidine 48 h assay.22 Komatsu et al23 succeeded in Takaku F. Identification and analysis of human erythropoietin isolating a subline, designated UT-7/TPO, that was absolutely receptors on a factor-dependent cell line, TF-1. Blood 1989; 73: dependent on TPO for growth and survival, showing several 375–380. mature megakaryocytic properties. Other studies showed that 13 Komatsu N, Nakauchi H, Miwa A, Ishihara T, Eguchi M, Moroi M, Okada M, Sato Y, Wada H, Yawata Y, Suda T, Miura Y. Estab- it is possible to transfer or extend the cytokine dependency lishment and characterization of a human leukemic cell line with of cell lines from one to another or several factors: an EPO- megakaryocytic features: dependency on granulocyte–macro- dependent subline, UT-7/EPO, had been isolated from the phage colony-stimulating factor, interleukin 3, or erythropoietin parental UT-7/GM cells (grown in GM-CSF); the cell line M- for growth and survival. Cancer Res 1991; 51: 341–348. TAT (cultured with GM-CSF as M-TAT/GM-CSF) proliferated 14 Morgan D, Class R, Soslau G, Brodsky I. Cytokine-mediated + similarly for more than 1 year in EPO (M-TAT/EPO) or SCF erythroid maturation in CD4 megakaryoblastic human cell line 24,25 HU-3. Exp Hematol 1996 (submitted). (M-TAT/SCF). 15 Uphoff CC, Brauer S, Grunicke D, Gignac SM, MacLeod RAF, However, this switch to incubation of a growth factor- Quentmeier H, Steube K, Tu¨ mmler M, Voges M, Wagner B, dependent cell line with a different cytokine is not always Drexler HG. Sensitivity and specificity of five different myco- possible; the fact that a cell line responds proliferatively for a plasma detection assays. Leukemia 1992; 6: 335–341. short period of time (24–72 h) to a given cytokine does not 16 Quentmeier H, Zaborski M, Drexler HG. The human bladder car- automatically imply that these cells can be grown over longer cinoma cell line 5637 constitutively secretes functional cytokines. periods of time with this particular factor (in the absence of Leukemia Res 1996 (in press). 17 Debili N, Wendling F, Cosman D, Titeux M, Florindo C, Dusanter- the ‘usual diet’). In addition, there is always the possibility Fourt I, Schooley K, Methia N, Charon M, Nador R, Bettaieb A, that the cells become quickly cytokine-independent, showing Vainchenker W. The Mpl receptor is expressed in the megakary- vigorous growth, but no longer requiring supplementation of ocytic lineage from late progenitors to platelets. Blood 1995; 85: any external growth factor. 391–401. In summary, we report here the establishment and charac- 18 Lok S, Kaushansky K, Holly RD, Kuijper JL, Lofton-Day CE, Oort terization of four TPO-dependent leukemia cell lines isolated PJ, Grant FJ, Heipel MD, Burkhead SK, Kramer JM, Bell LA, Sprecher CA, Blumberg H, Johnson R, Prunkard D, Ching AFT, from constitutively growth factor-dependent cell lines. The Mathewes SL, Bailey MC, Forstrom JW, Buddle MM, Osborn SG, present data show that the culture systems described here may Evans SJ, Sheppard PO, Presnell SR, O’Hara PJ, Hagen FS, Roth be useful to dissect basic mechanisms regulating megakaryo- GJ, Foster DC. Cloning and expression of murine thrombopoietin cyto- and thrombopoiesis in a simplified and isolated setting. cDNA and stimulation of platelet production in vivo. Nature 1994; 369: 565–568. 19 Drexler HG, MacLeod RAF, Borkhardt A, Janssen JWG. Recurrent chromosomal translocations and fusion genes in leukemia–lym- References phoma cell lines. Leukemia 1995; 9: 480–500. 20 Ritchie A, Vadhan-Raj S, Broxmeyer HE. Thrombopoietin sup- 1 Drexler HG, Quentmeier H. Thrombopoietin: expression of its presses apoptosis and behaves as a survival factor for the human receptor MPL and proliferative effects on leukemic cells. Leukemia growth factor-dependent cell line, M07e. Stem Cells 1996; 14: 1996; 10: 1405–1421. 330–336. 2 Kaushansky K. Thrombopoietin: biological and preclinical proper- 21 Ihle JN. Cytokine receptor signalling. Nature 1995; 377: 591–594. ties. Leukemia 1996; 10 (Suppl. 1): S46–S48. 22 Porteu F, Rouyez MC, Cocault L, Benit L, Charon M, Picard F, 3 Quentmeier H, Zaborski M, Graf G, Ludwig WD, Drexler HG. Gisselbrecht S, Souyri M, Dusanter-Fourt I. Functional regions of Expression of the receptor MPL and proliferative effects of its the mouse thrombopoietin receptor cytoplasmic domain: evi- ligand thrombopoietin on human leukemia cell lines. Leukemia dence for a critical region which is involved in differentiation and 1996; 10: 297–310. can be complemented by erythropoietin. Mol Cell Biol 1996; 16: 4 Drexler HG, Gignac SM, Minowada J. Hematopoietic cell lines. 2473–2482. TPO-dependent leukemia cell lines HG Drexler et al 551 23 Komatsu N, Kunitama M, Yamada M, Hagiwara T, Kato T, Miya- 7/Epo, derived from human leukemia cell line, UT-7. Blood 1993; zaki H, Eguchi M, Yamamoto M, Miura Y. Establishment and 82: 456–464. characterization of the thrombopoietin-dependent megakary- 25 Minegishi N, Minegishi M, Tsuchiya S, Fujie H, Nagai T, Hayashi ocytic cell line, UT-7/TPO. Blood 1996; 87: 4552–4560. N, Yamamoto M, Konnop T. Erythropoietin-dependent induction 24 Komatsu N, Yamamoto M, Fujita H, Miwa A, Hatake K, Endo T, of hemoglobin synthesis in a cytokine-dependent cell line M-TAT. Okano H, Katsube T, Fukumaki Y, Sassa S, Miuta Y. Establishment J Biol Chem 1994; 269: 27700–27704. and characterization of an erythropoietin-dependent subline, UT-