[1] 50w at a rate approximately proportional to the density of the material traversed. Beam penetration is shallower in bone than in muscle; in air and air-filled tissue penetration is increased by an amount about equal to the length of the air path traversed. 3. Attnibuted to C. A. Tobias in (6).

[1] 43w The cultured cells morphologically re- semble acute myelogenous leukemia cells, with 98 percent myeloblasts and promyelocytes, 2 percent intermediate and late stage granulocytes, and rare macrophages (Fig. 1). More than 85 percent of the cells stain positively for naphthol AS-D chloroacetate esterase (Sigma);

[2] 65w SCIENCE, VOL. 200, 9 JUNE 1978 5. G. W. Bennett and J. 0. Archambeau, Nucl. In- strum. Methods 125, 333 (1975). 6. H. D. Maccabee, U. Madhvarath, M. R. Raju, Phys. Med. Biol. 14, 212 (1969). 7. Supported by grant CA-188347 from The Na- tional Institutes of Health and by contract EY- 76-C-02-0016 with the Department of Energy. 5 December 1977; revised 28 February 1978

[3] 183w 98 percent stain positively for acid phosphatase; 52 percent give a positive reaction with periodic acid-Schiff reagent; 10 percent are peroxidase positive; and a rare macrophage-like cell has a positive reaction with a-naphthyl butyrase (Fig. 1). These cytochemical characteristics are typical of moderately well-differentiated acute myelogenous leukemia. Medium from 5-day cultures that had been initiated with 3 x 106 cells contained lysoenzymes at 5.5 + 1.2 ,ug/ml. Chromosomal abnormalities were consistently observed upon cytogenetic ex- amination with standard methodology and banding technique (2). Metaphase studies of 19 consecutive cells showed 45 chromosomes, and banding of five cells showed 45,XY,(5p+9q-),-12,-17, +MAR,+fragment. Epstein-Barr virus (EBV) nuclear antigen and the EBV capsid antigen were not present. The leukemic cells had HLA-A30 and HLA-B35 histocompatibility antigens, assayed by microcytotoxicity (3). A human granulocyte-specific antigen was detected by microcytotoxicity on KG-1 cells by 8 of 40 human granulocyte-reactive anti- serums (4). A B lymphocyte-associated or Ia-like antigen was detected on 20 per- cent of the cultured cells ( 5). This anti- gen was previously found in leukemic cells from 70 percent of patients with acute myelogenous leukemia (5).

[4] 192w Surface membrane immunoglobulin was not demonstrable by immunofluorescence. No rosette formation oc- curred with sheep erythrocytes, but 7 ± 3 percent of the cells formed rosettes with antibody-coated sheep red blood cells. The cells did not respond to phytohemagglutinin (PHA) by an in- crease in [3H]thymidine incorporation. The cultured cells used had a minimal (3 percent) ability to lyse chromium-la- beled murine lymphoblasts (EL-4) by antibody-dependent cell-mediated cytotoxicity (6). The cultured leukemic cells did not phagocytose opsonized Candida albicans ( 7), but 29 + 5 percent ingested one or more l-,um latex particles. The 0036-8075/78/0609-1153$00.50/0 Copyright © 1978 AAAS Acute Myelogenous Leukemia: A Human Cell Line Responsive to Colony-Stimulating Activity Abstract. A permanent human cell line that maintains the granulocytic characteristics ofacute myelogenous leukemia cells has been established. The cells of this line form myeloid colonies in soft gel culture in the presence ofhuman colony-stimulating activity. The cell line may be usefulfor studying human acute myelogenous leukemia and the mechanism of response to colony-stimulating activity. leukemic cells did not exhibit chemotax- is in a Boyden chamber when either zymosan-activated or normal human type AB serum was used as the chemoat- tractant (7).

[5] 141w Cultured KG-1 cells are dependent on CSA for colony formation in vitro. Myeloid colonies develop from cells that are plated in 0.3 percent agar overlying I x 106 normal human peripheral white blood cells in 0.5 percent agar as the source of CSA (8). Colonies containing more than 50 cells were enumerated with an inverted microscope on day 14 of cul- ture. A linear relation exists between the number of myeloid colonies formed and the number of cells plated (Fig. 2). The cloning efficiency was 2.5 to 3.0 percent, and there was no colony formation in the absence of CSA. When random colonies were aspirated from the cultures, smeared on slides, and stained with Wright-Giemsa or chloroacetate ester- ase, they revealed predominantly myeloblasts and promyelocytes. However, 30 percent of the colonies contained a small number of intermediate and mature gran- ulocytes.

[6] 168w Several sources of CSA effective in stimulating colony formation of normal myeloid progenitors were tested for their ability to induce colony formation in KG-1 cells (Fig. 3). The supernatants from PHA-stimulated human peripheral blood lymphocytes, or conditioned me- dium from either human peripheral blood leukocytes (9) or continuous culture of T lymphocytes derived from the spleen of a patient with hairy-cell leukemia (10), stimulated colony formation when added directly to the agar cultures. A sigmoid dose-response curve was seen for each CSA tested; the ability of CSA to stimu- late colonies reached a plateau at about 50 Al/ml. A similarly shaped sigmoid CSA dose-response curve has been noted when monkey lung-conditioned medium was the source of CSA and nor- mal human bone marrow was the target cell population (11). When KG-I cells were exposed to CSA in liquid culture, washed, and plated in agar without CSA, no colony formation occurred (12). Nor- mal human myeloid precursors also re- quire the continuous presence of CSA for colony formation (13).

[7] 68w Murine CSA derived from pregnant mouse uterus (14), when tested over a wide range of concentrations, did not in- duce colony formation by the leukemia cells. Murine CSA is also a poor inducer of normal human myeloid colony forma- tion. Conditioned medium (25 to 100 lI) from liquid cultures of KG-I growing at a stationary density of 3.0 x 106 cells per milliliter was tested for possible effects

[8] 16w 1154 700 .u 500/ 0 0 0 300 X E z 100 0.5 1.0 1.5 2.0

[9] 35w Cells plated x 104 Fig. 2. Myeloid colony formation in agar is linearly related to the number of KG-1 cells plated. Each point represents the mean the standard error of four experiments done in triplicate.

[10] 78w on normal human myeloid colony forma- tion. When the conditioned medium was added to duplicate cultures of I x 105 nucleated bone marrow cells from three normal volunteers, colony formation did not differ from the control. When condi- tioned medium was added to agar cul- tures containing 1 x 105 normal bone marrow cells in the presence of a sub- optimal concentration of CSA (25 ,l of leukocyte-conditioned medium), colony formation did not differ significantly from the control.

[11] 60w Acute myelogenous leukemia cells from humans are difficult to grow in longterm culture, although some cell lines have been established (1). These lines are generally composed of poorly dif- ferentiated lymphoblastoid-like cells, containing EBV-associated antigens and lymphocyte cell markers. Two well- characterized myeloid cell lines are the murine M-1 cell line (15) and a human chronic myelogenous leukemia cell line

[12] 16w 700 500 300 .c 0 0 E z 100 001 05 5 10 25 50 100

[13] 65w CSA (AM) Fig. 3. Colony-stimulating activity dose-response curves. Colony formation by KG-1 cells is shown as a function of CSA concentra- tion for three different CSA sources; X, hu- man peripheral blood leukocyte-conlitioned medium; 0, PHA-stimulated human peripheral blood lymphocyte supernatant; 0, conditioned medium from continuous line of leukemic T lymphoblasts. Each point represents the mean the standard error of three experiments done in triplicate.

[1] 47w Research on acute myelogenous leu- kemia in humans has been impeded by the lack of an adequate model system. Leukemia in rodents does not parallel the human disease, and continuous cell lines from patients with acute myeloge- nous leukemia usually lose their myeloid characteristics in vitro (1).

[2] 196w We describe here a cell line (KG-1) which was derived from a 59-year-old man with erythroleukemia that devel- oped into acute myelogenous leukemia. This cell line retains its granulocytic na- ture in vitro and forms myeloid colonies in soft gel culture in response to colonystimulating activity (CSA). The patient's bone marrow contained 30 percent myeloblasts and 10 percent erythroblasts with marked megaloblastic and dysplastic changes. The patient died of sepsis 2 months after study. Cells obtained by bone marrow aspiration were placed in T flasks with alpha medium (Flow) con- taining 20 percent fetal calf serum, pen- icillin, streptomycin, and 10-4M a-thio- glycerol; the cultures were incubated at 37°C in an atmosphere of 5 percent CO2 in air. After 24 days in culture, the cells were actively proliferating. They have since been passaged 52 times in an 8- month period. The cells grow in suspen- sion and do not form clumps. Whenever the cells were transferred in the liquid culture, there was an initial 1-day lag phase followed by growth with a mean doubling time of 70 to 80 hours, until cell densities of 2.5 x 101 to 3.5 x 106 cells per milliliter were reached.

[3] 59w that contains a Philadelphia chromo- some, but shows little evidence of differentiation and no response to CSA (16). Gallagher and co-workers reported continuous growth and some dif- ferentiation of acute human myeloge- nous leukemia cells with conditioned medium from human embryonic cells (17), and Collins et al. have derived a line similar to the one that we describe (18).

[4] 50w We have established an acute myeloidleukemia cell line from humans that re- tains its myeloid characteristics and forms granulocytic colonies in response to human CSA. This cell line should pro- vide a valuable tool for studying the pathophysiology of acute myelogenous leukemia and the interaction of CSA with myeloid cells.