Leukemia (2002) 16, 112–119  2002 Nature Publishing Group All rights reserved 0887-6924/02 $25.00 www.nature.com/leu Expression of endothelial cell-associated molecules in AML cells M Watarai1, H Miwa1, M Shikami1, K Sugamura1, M Wakabayashi1, A Satoh1, K Tsuboi1, A Imamura1, H Mihara1, Y Katoh1, K Kita2 and M Nitta1

1Department of Internal Medicine, Division of Hematology, Aichi Medical University School of Medicine, Nagakute, Aichi, Japan; and 2Tokura Hospital, Uji, Kyoto, Japan

Recently, it has been clarified that interaction between hemato- hematopoietic cells is also mediated by adhesion molecules poietic cells and endothelial cells is important in normal hema- such as integrins, as well as soluble factors.10,11 topoiesis and leukemogenesis. In this study, we examined the relationship between AML cells and endothelial cells by analyz- Angiopoietins and VEGF are recognized to act coordinately ing the expression profile of angiogenic factors, angiopoietin- during vascular growth and remodeling. Blood vessels remain 1 (Ang-1), Ang-2, Tie-2 (a receptor for angiopoietins) and vascu- in a stable state when Tie-2 receptor is constitutively engaged lar endothelial growth factor (VEGF). Our results demonstrated with angiopoietin-1 (Ang-1) by stabilizing blood vessels -that CD7(؉)AML expressed Ang-2 mRNA frequently and inte- through interactions with perivascular cells and the extracellu grin-family adhesion molecules (CD11c and CD18) intensively, lar matrix. When Ang-2 expression is up-regulated, the inter- suggesting the close correlation with endothelial cells. On the other hand, in t(8;21) AML cells, expression of Ang-2 was action between Tie-2 and Ang-1 is disrupted, and the vessel infrequent and expression of integrin-family adhesion mol- is destabilized. Endothelial cells, separated from support cells ecules (CD11b, CD11c and CD18) was weak, suggesting the and matrix, become plastic and angiogenesis is promoted in sparse association with endothelial cells. As for CD7(؉)AML the presence of VEGF.12–15 In addition, various types of cancer cells, despite the frequent and intense expression of endo- cells have been shown to produce factors stimulating angio- thelial cell-associated molecules (such as Ang-2, CD11c and genesis in the tumor.16,17 Inhibitors for these angiogenicfac- CD18), intensity of Tie-2 expression was quite low (P < 0.05). ؉ tors are now considered to be candidates for novel therapeutic Ang-2 expressed in CD7( )AML cells is not considered to act in 18,19 an autocrine fashion, but to work on endothelial cells to ‘feed’ tools for cancer. leukemic cells. Although Ang-2 is recognized as a natural In this study, we would like to clarify the relationship antagonist for Tie-2, our data presented here suggested the between AML cells and endothelial cells by analyzing the alternative role of Ang-2 in the relationship between endothelial expression profile of angiogenicfactors,Ang-1, -2, Tie-2 and cells and leukemia cells, at least in a subset of leukemia such VEGF. Phenotype- and karyotype-specific expression pattern as CD7(؉)AML. These results were supported by the study using AML cell lines, KG-1 (CD7 negative) and its subline KG- was observed. In addition, expression of integrin-family 1a (CD7 positive); KG-1 had mRNA expression profile of Ang- adhesion molecules also showed distinct pattern. Co-culture -1؉Ang-2−Tie-2+, while KG-1a showed Ang-1؉Ang-2؉Tie-2؊. experiment of AML cells and human umbilical vein endo These difference in the expression profile of angiogenic factors thelial cells was done to investigate the functional interaction .between CD7(؉)AML and t(8;21)AML may explain the charac- between AML cells and endothelial cells teristic morphological features of these leukemias (CD7(؉)AML as blastic type and t(8;21)AML as differentiative type). Leukemia (2002) 16, 112–119. DOI: 10.1038/sj/leu/2402326 Keywords: AML; angiopoietins; VEGF; endothelial cell Materials and methods

Samples and cell lines Introduction Bone marrow (BM) and peripheral blood (PB) smears were Recently, it has been clarified that hematopoietic stem cells prepared for May–Gru¨nwald Giemsa, peroxidase, naphthol and endothelial cells originated from common precursor cell, AS-D chloroacetate esterase, and ␣-naphthyl butylate esterase hemangioblast.1–3 In the normal developmental process, inter- staining. The morphological diagnosis of AML was made action between hematopoietic stem cells and endothelial cells according to FAB classification. Karyotypic analysis was per- is thought to be important.4–6 Endothelial cells have been formed on aspirated BM cells. All samples contained more shown to produce many cytokines known to play a role in than 80% leukemic cells morphologically. All patients gave the proliferation and differentiation of hematopoieticprogeni- their informed consent for the procedure involved. Leukemic tors.4,5 Conversely, it has been shown that hematopoieticstem cells from 36 AML patients including 12 M1, 12 M2, three cells play important roles for angiogenesis.6 In addition, some M3, five M4 and four M5 cases were studied. In addition, 375 reports suggested that leukemia cells also have intimate AML samples were studied for phenotypical study, especially correlation with endothelium in bone marrow. Increased angi- of integrin-family adhesion molecules. An AML cell line, KG- ogenesis was seen in the bone marrow of acute and chronic 1 and its subline, KG-1a (provided from Cell Resource Center leukemia patients, where some angiogenic factors such as vas- for Biomedical Research, Institute of Development, Aging and cular endothelial growth factor (VEGF) and basic fibroblast Cancer, Tohoku University, Japan) were also studied. growth factor (bFGF) were implicated to be mediators.7–9 The interaction between endothelial cells and normal or malignant Cell separation

Mononuclear cells (MNC) were separated from Correspondence: H Miwa, Department of Internal Medicine, Division of Hematology, Aichi Medical University School of Medicine, heparinized PB and BM by Ficoll–Hypaque centrifugation. Nagakute, Aichi, 480–1195, Japan; Fax: 81 561 63 3401 Stocked samples, which had been isolated and frozen at Received 28 May 2001; accepted 28 August 2001 −196°C in RPMI 1640 medium with 20% heat-inactivated Relationship between AML cells and endothelial cells M Watarai et al 113 fetal calf serum (FCS) and 10% dimethyl sulfoxide, were also VEGF; 1 min at 94°C, 1 min at 60°C, 2 min at 72°C (23 cycles) used as needed. for ␤-actin. In each experiment, these cycles were preceded by 10 min incubation at 95°C to activate the AmpliTaq GOLD (Roche Molecular Systems, Branchburg, NJ, USA), and fol- Immunophenotyping lowed by 7 min of 72°C for final extension. Primers for PCR are listed in Table 1.12,22–25 Products of RT-PCR, electrophor- Before immunostaining, MNC were treated with 1 mg/ml esed in 3.5% agarose, were stained with ethidium bromide human ␥-globulin for blocking of the binding due to receptors and visualized by UV light. The intensity of the band was for immunoglobulin G (IgG) Fcportion (Fc ␥R). CD13 calculated by densitometer. (recognized by My7, Coulter, Hialeah, FL, USA), and CD33 (by My9, Coulter) were tested as myeloid differentiation mark- ers. CD7 (by M-T701, PharMingen, San Diego, CA, USA) and Cell culture CD19 (by HIB19, PharMingen) were tested as lymphoid asso- ciated markers. CD34 (by HPCA-1, Becton Dickinson, Moun- Human umbilical vein endothelial cells (1.5 × 104) tain View, CA, USA) was examined as stem cell marker. (HUVEC)/cm2 culture dish were plated, then after 2 days, leu- CD11b (by Leu15, Becton Dickinson), CD11c (by B-ly6, kemia cells (4.0 × 105 cells/ml) were added on the culture. PharMingen), and CD18 (by 7E4, Immunotech, Marseille, Simple culture of leukemia cells was started simultaneously. France) were tested as integrin-family adhesion molecules. Fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse immunoglobulins was used as the second reagent. Samples Table 1 The primers for PCR were examined by a flowcytometer (FACSCalibur; Nippon Becton Dickinson, Tokyo, Japan). We focused on the leu- Ref. kemic cell fraction of the cytogram to clarify its phenotype, as previously described.20 At least 5000 cells were examined Angiopoietin-1 sense 5Ј-AGTCCAGAAAACAGTGGGAG-3Ј 22 in the surface marker study. The whole mouse Igs, IgG1, and antisense 5Ј-AGCAGCTGTATCTCAAGTCG-3Ј IgG2 used as the controls were from Chemicon International (Temecula, CA, USA). Angiopoietin-2 sense 5Ј-AAGAGCATGGACAGCATAGG-3Ј 12 antisense 5Ј-GAGTGTTCCAAGAGCTGAAG-3Ј Reverse transcription-polymerase chain reaction (RT- Tie-2 PCR) sense 5Ј-CTGCAGTGCAATGAAGCATG-3Ј 23 antisense 5Ј-TGAAGGGCTTTTCCACCATC-3Ј Expression of angiopoietin-1, 2 (Ang-1, 2), Tie-2, and vascular VEGF endothelial growth factor (VEGF) and ␤-actin genes was sense 5Ј-TCGGGCCTCCGAAACCATGA-3Ј 24 detected by the reverse transcription polymerase chain reac- antisense 5Ј-CCTGGTGAGAGATCTGGTTC-3Ј tion (RT-PCR) method as previously described.21 PCR was per- ␤-actin formed by using cDNA reverse-transcribed from 200 ng of sense 5Ј-GTGGGGCGCCCCAGGCACCA-3Ј 25 total RNA in the following conditions: 1 min at 94°C, 2 min antisense 5Ј-GTCCTTAATGTCACGCACGATTTC-3Ј at 60°C, 2 min at 72°C (27 cycles) for Ang-1, 2, Tie-2 and

Figure 1 RT-PCR of angiopoietin-1, -2, Tie-2, and VEGF in various AML cells. FAB subtypes are demonstrated in the uppermost case. Lower cases show the presence or absence of CD7 antigen expression or t(8;21) chromosome abnormality. RT-PCR of ␤-actin showed that the equal amount of RNA was loaded.

Leukemia Relationship between AML cells and endothelial cells M Watarai et al 114

Figure 2 Expression of RNA for angiopoietin-1, -2, Tie-2 and VEGF in association with CD7 or t(8;21). Left panels (a, c, e, g) show the comparisons of RNA expression between CD7(+) AML and CD7(−) AML. Right panels (b, d, f, h) shows the comparison between t(8;21)(+)AML and t(8;21)(−) AML. Each data point depicts the mean of the expression levels by the densitometric data of each molecule, standardized by ␤- actin. Each box shows ±1 s.e., and each bar ±1.96 s.e. P value is shown in each panel. NS, not statistically significant.

After 10 days co-culture, leukemia cells were harvested and Cell cycle analysis of cell line cells after co-culture on the cell number was counted. The medium, EBM-2, used in HUVEC this co-culture study contained hydrocortisone, basic fibro- × 5 blast growth factor, VEGF, insulin-like growth factor-1, ascor- KG-1 or KG-1a cells (1.0 10 cells/ml) were co-cultured for bic acid, fetal bovine serum (5%) and epidermal growth factor 24 h on the HUVEC. In the case of KG-1a, co-culture contain- (BioWhittaker, Walkersville, MD, USA). ing TIE-2/FC (which binds to angiopoietins; Sigma, St Louis, MO, USA) at the concentration of 1 ␮g/ml or 5 ␮g/ml was

Leukemia Relationship between AML cells and endothelial cells M Watarai et al 115

Figure 3 Expression of integrin-family adhesion molecules in association with CD7 or t(8;21). Left panels (a, c, e) show the comparisons of expression of integrins between CD7(+) AML and CD7(−) AML. Right panels (b, d, f) shows the comparison between t(8;21)(+) AML and t(8;21)(−) AML. Each data point depicts the mean of the positive percent of each adhesion molecule. Each box shows ±1 s.e., and each bar ±1.96 s.e. P value is shown in each panel. NS, not statistically significant.

Table 2 Frequency of expression in association with FAB classi- fication

Angiopoietin-1 Angiopoietin-2 Tie-2 VEGF

M1 10/12 5/12 3/12 6/12 ] P Ͻ 0.05 M2 9/12 4/12 8/12 7/10 P Ͻ 0.05 M3 2/3 1/3 0/3] 3/3 M4 3/5 0/5 3/5 2/5 M5 3/4 2/4 1/4 3/4

also done simultaneously. Then, the cell cycle was analyzed by using In Situ Cell Proliferation Kit, FLUOS (Roche Diagnos- tics, Mannheim, Germany). Briefly, 5-bromo-2Ј-deoxyuridine (BrdU) was added to the culture at the concentration of 10 ␮M for 60 min. Then, the cells were harvested gently without tearing HUVEC from the dish. The cells were fixed in 15 mM glycine and 70% ethanol for 30 min at 4°C, denatured in 4 M HCl for 10 min at room temperature, then incubated in PBS containing 0.5% bovine serum albumin (BSA) and 0.1% Figure 4 The effect of HUVEC for the survival of AML cells. The ratio of viable AML cells on the HUVEC/viable AML cells of simple Tween20 for 10 min at room temperature. Anti-BrdU-FLUOS + − antibody was added and incubated for 45 min at 37°Cina culture at day 10 were compared between CD7( ) and CD7( ) AML ␮ cells. Each data point depicts the mean of the ratio. Each box shows humid chamber. After the addition of 1 g/ml propidium iod- ±1 s.e., and each bar ±1.96 s.e. P value is shown in the panel. ide (PI), the cell cycle of the cell line cells was analyzed by flow cytometer. Simple culture of both cell lines was also sub- jected to cell cycle analysis. Cells having green fluorescence (BrdU) intensity beyond the negative control level were con- Statistical analysis sidered to be in S phase. Negative BrdU fluorescence with high PI fluorescence were considered to be in G2/M phase Differences among subtypes of AML were evaluated by t-test. and negative BrdU fluorescence with low PI fluorescence in The statistical analyses were performed with the STATISTICA

G0/G1 phase. software program (StatSoft Inc, Tulsa, OK, USA).

Leukemia Relationship between AML cells and endothelial cells M Watarai et al 116

Figure 5 Phenotypes and RT-PCR study of KG-1 and KG-1a. (a) Phenotypes (CD34, CD13, CD33 and CD7) of KG-1 (upper panel) and KG- 1a (lower panel). (b) RT-PCR study of Ang-1, Ang-2 and Tie-2 in KG-1 and KG-1a. RT-PCR of ␤-actin showed that the equal amount of RNA was loaded.

Results Ang-1, 2, Tie-2, VEGF mRNA expression in association with CD7 or t(8;21) abnormality Expression of mRNA for angiopoietin-1, 2 (Ang-1, 2), Tie-2, vascular endothelial growth factor (VEGF) in Since the difference of frequency was observed in association AML samples with CD7 or t(8;21), intensity of each mRNA expression was compared between CD7(+) and CD7(−) cases, or t(8;21)(+) mRNA expression for ang-1, 2, Tie-2 and VEGF was detected and t(8;21)(−) cases (Figure 2). Tie-2 mRNA expression was in 27, 12, 15 and 21 of 36 AML cells examined, respectively weaker in CD7(+) cases than in CD7(−) cases (P Ͻ 0.05). Ang- (34 AML cells were tested in the case of VEGF) (Figure 1). 1 and Ang-2 expression was stronger in CD7(+) cases than Significant difference of frequency of mRNA expression CD7(−), although the difference was not statistically signifi- among the FAB categories was not seen except for the differ- cant. In addition, t(8;21) AML cases showed weaker ence of Tie-2 between M2 (8/12) and M1 (3/12) or M3 (0/3) expression of Ang-1 and 2, however, without statistical cases (P Ͻ 0.05) as shown in Table 2. Frequency of expression significance. Intensity of VEGF mRNA expression was not in association with phenotypes and karyotypes is shown in influenced by CD7 expression or the presence of t(8;21) Table 3. Ang-2 mRNA expression was more frequently abnormality. observed in CD7(+) AML cells than in CD7(−) AML cells. Other phenotypicmarkers suchas CD34, CD19, CD13 and CD33 did not affect the frequency of mRNA expression for Expression of integrin-family adhesion molecules Ang-1, 2, Tie-2 or VEGF. As for the karyotypes, AML cells carrying t(8;21) chromosome abnormality showed less fre- Expression of integrin-family adhesion molecules (CD11b, quent expression of Ang-2. AML cases with t(15;17) were too CD11cand CD18), known to be important for the interaction few to evaluate statistically. between hematopoieticand endothelial cells,was studied by

Leukemia Relationship between AML cells and endothelial cells M Watarai et al 117

Figure 6 Cell cycle analysis of KG-1 and KG-1a (simple culture vs co-culture on HUVEC) After 24 h co-culture on HUVEC, 10 ␮M BrdU was added to the culture for 60 min, then the cells were fixed and denatured. Anti-BrdU-FLUOS antibody was applied and stained with 1 ␮g/ml propidium iodide, then the cells were analyzed by two-color flow cytometry. In the case of KG-1a, the co-culture containing TIE-2/FC (1 ␮g/ml or 5 ␮g/ml) was also done simultaneously.

Table 3 Frequency of expression in association with phenotypes Study of KG-1 and KG-1a and karyotypes An immature AML cell line, KG-1 and its variant subline KG- Angiopoietin-1 Angiopoietin-2 Tie-2 VEGF 1a were also studied. Phenotypes of these two cell lines were quite similar except for CD7 expression, which was positive CD7 for KG-1a and negative for KG-1 (Figure 5a). Expression of + 12/14 8/14 4/14 7/14 NS P Ͻ 0.05 NS NS − 15/22]] 4/22 11/22 ]] 14/20 mRNA for Ang-1, Ang-2 and Tie-2 of KG-1 and KG-1a was depicted in Figure 5b. Ang-1 mRNA was distinctly detected CD19 in both cell lines, although Ang-2 was expressed only in KG- + 9/11 5/11 4/11 5/11 NS NS NS NS 1a. Tie-2 mRNA expression was weakly detected in KG-1, but − 18/25]] 7/25 11/25 ]] 16/23 was not detected in KG-1a. The cell cycle study using BrdU CD34 and propidium iodide demonstrated that KG-1a cells, but not + 20/25 8/25 11/25 15/24 NS NS NS NS KG-1 cells, increased S/G /M phase cell population by − 7/11]] 4/11 4/11 ]] 6/10 2 co-culture on HUVEC. The increase of S/G2/M phase cells of t(8;21) KG-1a was cancelled by the addition of TIE-2/FC in a dose- + 4/8 0/8 5/8 4/7 NS P Ͻ 0.05 NS NS dependent manner (Figure 6). − 23/28]] 12/28 10/28 ]] 17/27

Discussion

We described here the expression pattern of angiogenic fac- flow cytometry. As shown in Figure 3, expression of CD11c tors (Ang-1, -2 and VEGF), Tie-2 receptor and integrin family and CD18 was more intensely expressed in CD7(+) cases than adhesion molecules in various types of AML cells. Our results CD7(−) cases with statistical significance. On the other hand, demonstrated that CD7(+)AML expressed Ang-2 mRNA fre- t(8;21) AML cells showed less expression of CD11b, CD11c quently and integrin-family adhesion molecules (CD11c and and CD18 than non-t(8;21) AML cells. CD18) intensively, suggesting the close correlation with endo- thelial cells. On the other hand, in t(8;21) AML cells, expression of Ang-2 was infrequent and expression of integrin- family adhesion molecules (CD11b, CD11c and CD18) was Co-culture of AML cells and human umbilical vein weak, suggesting the sparse association with endothelial cells. endothelial cells (HUVEC) It has been demonstrated that both normal hematopoiesis and leukemogenesis are closely correlated with endothelial cells Leukemia cells cultured for 10 days on the monolayer of in bone marrow.4,5,7–9 However, precise mechanisms for the HUVEC were harvested and the number of viable cells were interaction have not been elucidated. CD7(+)AML cells often estimated. The ratio of viable cells on HUVEC/viable cells of show blasticmorphology, and t(8;21)AML is well known to simple culture were compared. As shown in Figure 4, the ratio have a tendency for prominent granulocytic differentiation. It was significantly higher in CD7(+) AML cells (four cases) than is of interest that CD7(+)AML and t(8;21)AML showed good in CD7(−) AML cells (five cases) (P Ͻ 0.05). contrast in expression of molecules associated with interaction

Leukemia Relationship between AML cells and endothelial cells M Watarai et al 118 with endothelial cells. In addition, CD7(+)AML cells were Kogo H, Tsuji K, Nakahata T, Miyajima A. In vitro expansion of revealed to survive on HUVEC longer than CD7(−) AML cells. murine multipotential hematopoieticprogenitors from the embry- This finding confirmed the close association between onicaorta-gonad-mesonephros region. Immunity 1998; 8:105– + 114. CD7( )AML cells and endothelial cells functionally. One 6 Takakura N, Watanabe T, Suenobu S, Yamada Y, Noda T, Ito Y, possible explanation for this contrast is that leukemia cells Satake M, Suda T. A role for hematopoieticstem cellsin promoting having close association with endothelial cells maintain angiogenesis. Cell 2000; 102: 199–209. immaturity and blasticmorphology. 26,27 On the other hand, 7 Hussong JW, Rodgers GM, Shami PJ. Evidence of increased angio- leukemia cells having less association with endothelium tend genesis in patients with acute myeloid leukemia. Blood 2000; 95: to differentiate to mature cells. 309–313. + 8 Padro T, Ruiz S, Bieker R, Burger H, Steins M, Kienast J, Buchner As for CD7( )AML cells, despite the frequent and intense T, Berdel WE, Mesters RM. Increased angiogenesis in the bone expression of endothelial cell-associated molecules (such as marrow of patients with acute myeloid leukemia. Blood 2000; 95: Ang-2, CD11cand CD18), intensity of Tie-2 expression was 2637–2644. quite low (P Ͻ 0.05). (Tie-2 has been demonstrated to be 9 Aguayo A, Kantarjian H, Manshouri T, Gidel C, Estey E, Thomas expressed in normal hematopoieticcellsand some immature D, Koller C, Estrov Z, O’Brien S, Keating M, Freireich E, Albitar M. leukemiccells. 28,29) Here, Ang-2 expressed in CD7(+)AML Angiogenesis in acute and chronic leukemias and myelodysplastic cells is not considered to act in an autocrine fashion, but to syndromes. Blood 2000; 96: 2240–2245. 10 Rood PM, Gerritsen WR, Kramer D, Ranzijn C, von dem Borne work on endothelial cells to ‘feed’ leukemia cells. It has been AE, van der Schoot CE. Adhesion of hematopoietic progenitor cells reported that Ang-2 is recognized as a natural antagonist for to human bone marrow or umbilical vein derived endothelial cell Tie-2.12 However, our data presented here suggested the alter- lines: a comparison. Exp Hematol 1999; 27: 1306–1314. native role of Ang-2 in the relationship between endothelial 11 Peled A, Kollet O, Ponomaryov T, Petit I, Franitza S, Grabovsky cells and leukemia cells, at least in a subset of leukemia such V, Slav MM, Nagler A, Lider O, Alon R, Zipori D, Lapidot T. The as CD7(+)AML. Ang-2 can act as a direct stimulator for Tie-2 chemokine SDF-1 activates the integrins LFA-1, VLA-4, and VLA- 5 on immature human CD34(+) cells: role in transendothelial/ receptor, which augments the leukemia–endothelium inter- stromal migration and engraftment of NOD/SCID mice. Blood action. 2000; 95: 3289–3296. The study using KG-1 (CD7-negative) and its subline KG- 12 Maisonpierre PC, Suri C, Jones PF, Bartunkova S, Wiegand SJ, Rad- 1a (CD7-positive) corresponded well to the results obtained ziejewski C, Compton D, McClain J, Aldrich TH, Papadopoulos from clinical materials. First, Ang-2 expression was detected N, Daly TJ, Davis S, Sato TN, Yancopoulos GD. Angiopoietin-2, only in KG-1a, and Tie-2 expression was detected in KG-1 a natural antagonist for Tie2 that disrupts in vivo angiogenesis. and was not detected in KG-1a. Second, by the contact with Science 1997; 277: 55–60. 13 Asahara T, Chen D, Takahashi T, Fujikawa K, Kearney M, Magner HUVEC, KG-1a cells increased S/G2/M phase population, and M, Yancopoulos GD, Isner JM. Tie2 receptor ligands, angiopoie- the effect was cancelled by TIE-2/FC. The result indicated that tin-1 and angiopoietin-2, modulate VEGF-induced postnatal neo- angiopoietins (especially Ang-2) secreted by KG-1a were vascularization. Circ Res 1998; 83: 233–240. bound to TIE-2/FC, resulting in the disconnection between 14 Holash J, Maisonpierre PC, Compton D, Boland P, Alexander CR, KG-1a and HUVEC and loss of ‘feeding’ KG-1a cells by Zagzag D, Yancopoulos GD, Wiegand SJ. Vessel cooption, HUVEC. Further study should be done to clarify the relation- regression, and growth in tumors mediated by angiopoietins and VEGF. Science 1999; 284: 1994–1998. ship between leukemia cells and endothelial cells through 15 Yancopoulos GD, Davis S, Gale NW, Rudge JS, Wiegand SJ, Hol- angiogenicfactors. ash J. Vascular-specific growth factors and blood vessel formation. Nature 2000; 407: 242–248. 16 Folkman J. Angiogenesis in cancer, vascular, rheumatoid and Acknowledgements other disease. Nat Med 1995; 1: 27–31. 17 Nicosia RF. What is the role of vascular endothelial growth factor- related molecules in tumor angiogenesis? Am J Pathol 1998; 153: Thanks are due to Drs I Tanaka (Suzuka Kaisei Hospital, 11–16. Suzuka), T Ueda (Fukui Medical School, Fukui), S Doi (Osaka 18 Boehm T, Folkman J, Browder T, O’Reilly MS. Antiangiogenic Red Cross Hospital, Osaka), T Kyo and H Dohy (Hiroshima therapy of experimental cancer does not induce acquired drug Red Cross Hospital, Hiroshima) for providing important resistance. Nature 1997; 390: 404–407. patient information. We also thank Ms A Usui and E Naitoh 19 Benjamin LE, Golijanin D, Itin A, Pode D, Keshet E. Selective for their technical and secretarial assistance. ablation of immature blood vessels in established human tumors follows vascular endothelial growth factor withdrawal. J Clin Invest 1999; 103: 159–165. 20 Kita K, Nakase K, Miwa H, Masuya M, Nishii K, Morita N, Takak- References ura N, Otsuji A, Shirakawa S, Ueda T, Nasu K, Kyo T, Dohy H, Kamada N. Phenotypical characteristics of acute myelocytic leu- 1 Eichmann A, Corbel C, Nataf V, Vaigot P, Breant C, Le Douarin kemia associated with the t(8;21)(q22;q22) chromosomal abnor- NM. Ligand-dependent development of the endothelial and hemo- mality: frequent expression of immature B-cell antigen CD19 poieticlineages from embryonicmesodermal cellsexpressing vas- together with stem cell antigen CD34. Blood 1992; 80: 470–477. cular endothelial growth factor receptor 2. Proc Natl Acad Sci USA 21 Miwa H, Kita K, Nishii K, Morita N, Takakura N, Ohishi K, Mah- 1997; 94: 5141–5146. mud N, Kageyama S, Fukumoto M, Shirakawa S. Expression of 2 Choi K, Kennedy M, Kazarov A, Papadimitriou JC, Keller G. A MDR1 gene in acute leukemia cells: association with CD7+ acute common precursor for hematopoietic and endothelial cells. Devel- myeloblasticleukemia/acutelymphoblasticleukemia. Blood opment 1998; 125: 725–732. 1993; 82: 3445–3451. 3 Schuh AC, Faloon P, Hu QL, Bhimani M, Choi K. In vitro hemato- 22 Davis S, Aldrich TH, Jones PF, Acheson A, Compton DL, Jain V, poieticand endothelial potential of flk-1(-/-) embryonicstem cells Ryan TE, Bruno J, Radziejewski C, Maisonpierre PC, Yancopoulos and embryos. Proc Natl Acad Sci USA 1999; 96: 2159–2164. GD. Isolation of angiopoietin-1, a ligand for the TIE2 receptor, by 4 Fennie C, Cheng J, Dowbenko D, Young P, Lasky LA. CD34+ secretion-trap expression cloning. Cell 1996; 87: 1161–1169. endothelial cell lines derived from murine yolk sac induce the 23 Ziegler SF, Bird TA, Schneringer JA, Schooley KA, Baum PR. Mol- proliferation and differentiation of yolk sacCD34 + hematopoietic ecular cloning and characterization of a novel receptor protein progenitors. Blood 1995; 86: 4454–4467. tyrosine kinase from human placenta. Oncogene 1993; 8: 663– 5 Mukouyama Y, Hara T, Xu M, Tamura K, Donovan PJ, Kim H, 670.

Leukemia Relationship between AML cells and endothelial cells M Watarai et al 119 24 Fiedler W, Graeven U, Ergun S, Verago S, Kilic N, Stockschlader 27 Rosler E, Brandt J, Chute J, Hoffman R. Cocultivation of umbilical M, Hossfeld DK. Vascular endothelial growth factor, a possible cord blood cells with endothelial cells leads to extensive amplifi- paracrine growth factor in human acute myeloid leukemia. Blood cation of competent CD34+CD38− cells. Exp Hematol 2000; 28: 1997; 89: 1870–1875. 841–852. 25 Nakajima-Iijima S, Hamada H, Reddy P, Kakunaga T. Molecular 28 Kukk E, Wartiovaara U, Gunji Y, Kaukonen J, Buhring HJ, Rappold structure of the human cytoplasmic beta-actin gene: interspecies I, Matikainen MT, Vihko P, Partanen J, Palotie A, Alitalo K, Alitalo homology of sequences in the introns. Proc Natl Acad Sci USA R. Analysis of Tie receptor tyrosine kinase in haemopoietic pro- 1985; 82: 6133–6137. genitor and leukaemia cells. Br J Haematol 1997; 98: 195–203. 26 Ye ZQ, Burkholder JK, Qiu P, Schultz JC, Shahidi NT, Yang NS. 29 Takakura N, Huang XL, Naruse T, Hamaguchi I, Dumont DJ, Yan- Establishment of an adherent cell feeder layer from human umbili- copoulos GD, Suda T. Critical role of the TIE2 endothelial cell cal cord blood for support of long-term hematopoietic progenitor receptor in the development of definitive hematopoiesis. Immun- cell growth. Proc Natl Acad Sci USA 1994; 91: 12140–12144. ity 1998; 9: 677–686.

Leukemia