Human T Lymphocyte Clones: Influence of Culture Conditions and Optimization of Proliferative Assays

Phosphatidylinositol-specific phospholipase C (PI-PLC) treatment of the human natural killer (NK) target cells Molt-4, Jurkat, and U937 reduced their susceptibility to killing by human NK cells in a dose-dependent fashion. This indicates that a cell surface structure, anchored by a glycosyl-phosphatidylinositol (G-PI) moiety, is important in NK cytotoxicity. In contrast, another common NK target cell line, K562, remained susceptible to NK killing after enzyme treatment, suggesting that distinct target structures are expressed by this cell line. PI-PLC treatment of Molt-4 cells also reduced their sensitivity to human lymphokine activated killer (LAK) cells, suggesting that NK and LAK cells share common specificity in the killing of Molt-4. In contrast, PI-PLC had no effect on the killing of the LAK target cell line, Daudi, which is only weakly sensitive to unactivated NK cells. Killing of a variety of murine target cells by murine NK cells was not affected by PI-PLC treatment, but cross-species killing of Molt-4 by murine NK cells was inhibited by PI-PLC, suggesting a common mechanism in the killing of this human target cell line. The PI-PLC treatment of effector cells from either species did not reduce their NK activity. The reduction in sensitivity of the Molt-4, Jurkat, and U937 target cells probably results from a loss of a target specific G-PI linked membrane molecule, but other possible explanations for these results are also discussed.     

Engagement of MHC Class I Proteins on Natural Killer Cells Inhibits their Killing Capacity

We have studied whether engagement of MHC class I (MHC—I) molecules on natural killer (NK) cells can influence the NK killing activity. Human NK effector cells, enriched by nylon wool passage, were incubated with monoclonal antibodies (MoAb) to MHC—I followed by cross-linking with secondary rabbit anti mouse Ig or streptavidin. Cross linking of MHC—I molecules on NK cells resulted in a clear inhibition of the NK activity against the target cells K562, Molt-4 and U937. The inhibitory effect was selective for MHC—I and was not seen with MoAb to MHC—II or CD56 molecules. The inhibition was not mediated via Fc receptors since F(ab)₂ fragments of the MHC—I MoAb W6/32 were as effective as the intact antibody. The best inhibition of NK activity was obtained using biotin-labelled F(ab)₂ fragments of W6/32 and streptavidin as a cross-linker, where up to 70 % reduction in NK cell activity was observed. Antibody dependent cellular cytotoxicity (ADCC) was also inhibited by cross-linking MHC—I molecules on the effector cells.
The results show that antibody mediated cross-linking of MHC—I proteins on NK cells can inhibit their killing capacity. This indicates that MHC—I molecules on NK cells can be involved in the regulation of NK cytotoxicity, perhaps by transmitting inhibitory signals into the NK cell.     

Modulation of natural killing by tumor promoters: the regulatory influence of adherent cells varies with the type of target cell

We have analyzed the regulatory effects of two classes of tumor promoters, phorbol diesters and indole alkaloids, on human natural killer (NK) cell activity in vitro. In accordance with previous reports, we found that 12-O-tetradecanoylphorbol-13-acetate (TPA) inhibited natural killing against K 562 targets by unseparated mononuclear cells. Here, suppression of NK required the presence of adherent cells (macrophages). Contrary to the results obtained with K 562, tumor promoter-induced suppression of NK activity tested against U 937, another cell line of known NK susceptibility, was independent of the presence of adherent cells. Thus, NK cytotoxicity of effector cells rigorously depleted of adherent and Ia-positive cells was still inhibited when assayed against U 937, while it was generally enhanced when tested against K 562. Identical results were obtained with teleocidin and dihydroteleocidin B, two members of the recently discovered indole alkaloid class of tumor promoters. Therefore, we demonstrate that the regulatory effect of tumor promoters on human NK activity (suppression or stimulation) is determined not only by macrophages at the effector cell level but also by the type of target cell under study.     

In vitro binding of natural killer cells to Cryptococcus neoformans targets.

Nylon wool-nonadherent splenic cells from 7- to 8-week-old CBA mice were further fractionated on discontinuous Percoll gradients. Enrichment of natural killer (NK) cells in Percoll fractions 1 and 2 was confirmed by morphological examination, by immunofluorescent staining, and by assessing the cytolytic activity of each Percoll cell fraction against YAC-1 targets in the 4-h51Cr release assay. Cells isolated from each Percoll fraction were tested for growth-inhibitory activity against Cryptococcus neoformans, a pathogenic yeastlike organism, by using an in vitro 18-h growth inhibition assay. The results showed that NK cell enrichment was concomitant with enrichment of anti-Cryptococcus activity in Percoll fractions 1 and 2. Cells from NK cell-rich fractions formed conjugates with the mycotic targets similar to the conjugates reported in NK cell-tumor systems. In addition, the percentage of effector cell-Cryptococcus conjugates was directly proportional to the level of the C. neoformans growth-inhibitory activity of the effector cells used. Scanning electron microscopy of the effector cell-Cryptococcus conjugates showed direct contact between the effector cells and the cryptococcal targets. An immunolabeling method combined with scanning electron microscopy was used to demonstrate that the effector cells attached to C. neoformans were asialo GM1 positive and, therefore, had NK cell characteristics.     

Specificity of γδ Receptor-Bearing Cytotoxic T Lymphocytes Isolated from Human Peripheral Blood

T-cell receptor (TcR)-gamma delta-bearing lymphocytes were isolated from the peripheral blood of two healthy donors by immunomagnetic separation and subsequently cultured. The cell lines generated showed two distinct patterns of cytotoxicity. One TcR-gamma delta + cell line (HG.D) lysed K562 and U937 target cells, three TcR-gamma delta + cell lines lysed Daudi cells, and one TcR-gamma delta + cell line showed a shift from the former to the latter specificity during culture. Cold target inhibition experiments showed that the HG.D effector cells which were cytotoxic against U937 cells also lysed K562 cells. The cytotoxicity against Daudi cells was strongly inhibited by monoclonal antibodies (MoAb) against the CD3 complex, whereas the cytotoxicity of the HG.D cell line against K562 and U937 was unaffected by such antibodies. The cytotoxicity against Daudi cells was also strongly inhibited in the presence of anti-TcR-gamma delta MoAb. However, in two of the Daudi-specific cell lines, strong cytotoxicity against K562 cells was induced by anti-TcR-gamma delta MoAb. Anti-LFA-1 MoAb caused only a partial inhibition of cytotoxicity, while anti-CD2 and anti-TcR-alpha beta MoAb were found to have no effect. The results indicate that human gamma delta receptor-bearing T cells demonstrate a certain degree of target cell specificity, and that recognition of some target cells may be mediated through the TcR-gamma delta.     

Target-Effector-Cell Interactions in the Human Natural-Killer(NK)-System: Isolation of Target Structures

The existence of structures on NK-sensitive target cells selectively recognized by the effector cells have been postulated. To test this hypothesis, four selected human cell lines were investigated for target-cell proteins which could serve as specific ligands for the putative NK-cell receptor(s). NP-40 extracts from two highly NK-sensitive (K 562 and Molt-4) and two rather insensitive cell lines (HL-60 and Reh-6) were fractionated on SDS-polyacrylamide gels and tested for their ability to inhibit binding of effector to target cells as well as NK cytotoxicity. Three fractions with molecular weights (MW) of 200, 120 and 80 ± 10 KD isolated from K 562 cells were able to inhibit binding of large granular lymphocytes (LGL) to K 562. Of the other cell lines, Molt-4 and HL-60, both which were able to inhibit lysis of K 562 in a cold target inhibition assay, showed also two inhibitory fractions with MW 120 and 80 KD, whereas Reh-6, which is not able to compete with ⁵¹Cr-labelled K 562 in a cytotoxicity assay, lacked these structures. The 200, 120 and 80 KD fractions isolated from K 562 and the 120 and 80 KID fractions from Molt-4 and HL-60 were able to inhibit lysis of K 562 cells when added to the cytotoxicity assay. By adsorption/elution of radiolabelled K 562 extracts to/from LGL it was possible to detect an 80 KID target-cell surface protein which became preferentially bound by LGL-enriched but not by LGL-depleted lymphocyte preparations. Our results indicate the existence of target-cell proteins in NK-sensitive cell lines which serve as specific ligands for binding of NK cells. These target-cell structures of human cell lines differed from NK target structures described for mouse-NK-sensitive cell lines.     

An improved flow cytometry-based natural killer cytotoxicity assay involving calcein AM staining of effector cells

Several flow cytometric methods for measuring natural killer cell activity have been developed. Commonly used protocols involve the staining of target cells with various fluorescent dyes. However, these protocols are not applicable to certain experimental settings. Therefore, we used Calcein AM (CAM), which has been reported to be the most suitable dye for use in target cell staining protocols, as a means of developing an improved flow cytometry-based NK cytotoxicity assay involving effector cell staining. Peripheral blood mononuclear cells (PBMCs) isolated by gradient density centrifugation and expanded NK cells were used as effector cells. Cytotoxicity against K562 cells and several hematologic cancer cell lines was measured by a flow cytometry-based method using CAM and propidium iodide. The new assay was compared with a standard (51)Cr release assay (CRA) in terms of its ability to measure the cytotoxicity of NK cells in PBMCs and expanded NK cells against K562 cells. The optimal concentration of CAM for staining effector cells was 0.05 μM, and CAM fluorescence intensity in effector cells was maintained for 4 hours. CAM staining had no significant effect on NK cell activity in human PBMCs or expanded NK cells. Comparison of the CRA and this new assay using K562 cells revealed a good correlation (PBMCs, r = 0.894; expanded NK cells, r = 0.887). Distinct separation between target tumor cells (Daudi, Raji, RPMI8226, U266, U937, and K562 cells) and CAM-stained PBMCs (E:T ratio, 12.5:1 to 50:1) or expanded NK cells (E:T ratio, 0.5 to 4:1) was observed after incubation for 1 or 4 hours. In summary, we successfully developed an effective flow cytometry-based assay for assessing the activity of NK cells in PBMCs and expanded NK cells against K562 cells and various types of hematologic cancer cells.     

Different functional domains on the transferrin receptor molecule defined by monoclonal antibodies.

Three monoclonal antibodies (mAb) FG 1/5, FG 1/6 and FG 2/12, specific for the human transferrin receptor molecule (TR), have been used to define epitopes on the TR molecule and to block natural killer lysis. FG 2/12 mAb but not FG 1/5 or FG 1/6 blocked [125I-] transferrin binding to the cellular receptor. Furthermore, FG 1/5 and FG 1/6 mAbs competed out the binding of each other to the cells but not significantly that of FG 2/12. As expected, the binding of F2/12 but not of FG 1/5 or FG 1/6 was inhibited by transferrin. In addition, FG 2/12 inhibited in a dose-dependent manner the NK activity of purified T3- large granular lymphocyte effector cells against HeLa or Molt-4 cells but not against K-562 or U937 cells. FG 1/5 preferentially inhibited NK activity against HeLa cells and FG 1/6 mAb was completely uneffective. These inhibitions were stronger at low effector to target cell (E:T) ratios than at high E:T ratios, suggesting that NK cells and anti-TR mAbs compete for the same site in the target cell. It was shown that FG 1/5 and FG 2/12 mAbs blocked cells' conjugate formation by acting at the target cell level. Our results confirm the role of TR as a one of the target structures in NK lysis and suggest that the epitope recognized by NK cells is close to but different from the transferrin binding site.     

Effects of Chloroquine, Mefloquine and Quinine on Natural Killer Cell Activity in vitro

Natural killer (NK) cell activity against K 562 target cells was inhibited by pharmacological concentrations of chloroquine, mefloquine and quinine. The most potent were mefloquine and quinine. The drug-induced inhibition of the NK cell activity was abolished by addition of alpha-interferon (IF) or interleukin 2 (Il-2); preincubation of mononuclear cells with IF or Il-2 followed by addition of anti-malarial drugs decreased the inhibitory effects of the drugs. The drug-induced inhibition of the NK cell activity was not dependent on the presence of monocytes. Using monocyte depleted Percoll fractionated NK cell enriched populations in a single cell agarose assay, it was shown that the inhibitory effects of mefloquine, but not of chloroquine and quinine were due to an inhibition of the formation of effector/target cell conjugates.     

The effect of differentiation inducers on the sensitivity of two myeloid cell lines to natural killer (NK) cell-mediated lysis

The effects of differentiation inducers on the sensitivity of two human myeloid cell lines, K562 and HL-60, to natural killer (NK) cell lysis were examined. K562 cells treated with 12-O-tetradecanoyl phorbol-13-acetate (TPA) and HL-60 cells treated with either TPA or dimethylsulfoxide (DMSO) not only became less sensitive NK targets, but also became less competitive in cold target inhibition experiments. Target binding cell assays revealed that TPA-treated target cells bound fewer NK effector cells than did untreated targets. TPA treatment renders K562 and HL-60 target cells more susceptible to hypotonic lysis, indicating that the decreased NK sensitivity is not due to altered osmotic fragility. The observed reduction in sensitivity to NK lysis was not apparently mediated by interferon (IFN) released by the target cells, as neither alpha nor gamma IFN were detected in culture supernatants. Furthermore, the effects of TPA were additive to the known protective effect of IFN on NK target cells. In contrast to the parent line, a subclone of HL-60 resistant to chemically induced differentiation did not become a less sensitive target after exposure to TPA. These observations with myeloid and erythroid target cells imply that the reduced killing of the TPA-treated cells was secondary to impaired recognition of the target by the effector, and that the state of cellular differentiation is a major determinant of susceptibility to NK cell-mediated lysis.     

Interferon-Induced NK Augmentation in Humans

By combining a single-cell cytotoxicity assay in agarose with estimations of the maximal natural killer (NK) cell potential (Vmax) by 51Cr release, the mechanism behind interferon augmentation of human NK cells were analysed. The number of target-binding cells (TBCs) the fraction of active TBCs and NK cell recycling were studied after short-term interferon treatment. The results demonstrate a dual effect of interferon on human NK cells: effector cell recruitment and increased effector cell recycling. Both of these variables were increased when NK cells were tested against the standard target K-562 and against Daudi and BJAB cells, derived from B-type lymphomas. However, when T cell lines derived from acute lymphocytic leukaemia (Molt-4 and 1310) were used as targets, a larger fraction of active NK cells were found among untreated TBCs, whereas interferon treatment only resulted in increased effector cell recycling and not in effector cell recruitment. No increase in TBCs after interferon treatment could be detected with any cell line tested. The difference seen between T and non-T cell lines with regard to interferon-induced effector cell recruitment is discussed in relation to known characteristics of the human NK system.     

Characterization of the in Vitro Effects of Glucocorticosteroids on NK Cell Activity

The NK cell activity of mononuclear cells as well as monocyte-depleted, Percoll-fractionated, NK cell-enriched effector cells against K 562 target cells was inhibited by methylprednisolone (MP) and hydrocortisone (HC) in a dose-dependent manner. The effector/target cell conjugate formation was studied in a single cell agarose assay, and it was shown that MP and HC partly inhibited the NK cell activity by inhibition of the adhesion of effector cells to target.     

Effects of Heat Shock on Cytolysis Mediated by NK Cells, LAK Cells, Activated Monocytes and TNFs α and β

We have recently shown that a heat treatment of a murine target cell line, WEHI 164, induces resistance to lysis mediated by tumour necrosis factors alpha (TNF-α) and beta (TNF-β). In the present study the effect of the heat shock of target cells on cytotoxicity mediated by natural killer cells (NK cells), lymphocyte-activated killer cells (LAK cells), activated monocytes, TNF-α, and TNF-β was investigated, First, WEHI 164 cell line and six human cell lines (ME 180, K 562, U 937, HeLa, MCF 7, and SK-OV 3) were screened for their sensitivity to different forms of lysis, and then sensitive cell lines were heat-treated. Pretreatment of target cells at 42° C for 45-60 min also rendered human target cell lines more resistant to lysis by rTNFs, and the acquired resistance was accompanied by an increased resistance to activated monocytes, but not to NK cells or LAK cells. Thus, the heat-induced resistance mechanisms capable of protecting target cells from lysis by rTNFs and by activated monocytes do not elicit resistance to lysis by NK cells and LAK cells, supporting the hypothesis that mediators other than TNFs are involved in NK cell-and LAK cellmediated killing.     

Cytotoxic Factors Released by Dengue Virus-Infected Human Blood Monocytes

Human monocyte-derived cytotoxic factors (CF) induced by dengue virus were studied. Using several human leukemia cell lines as precursors, the biological activities of CF in conditioned medium from dengue virus-infected monocytes were demonstrated through the measurement of tumor cell growth inhibition. The conditioned medium from dengue virus infected monocytes suppressed significantly growth of CEM, HL60, K562, and U937 cells. In the presence of 10% conditioned medium (v/v) from dengue virus infected monocytes, DNA synthesis of U937 cells, as measured by [³H]thymidine incorporation, decreased by 99% in contrast to their synthesis in conditioned medium from noninfected control monocytes, which did not have any suppressive effect. Partial characterization of CF showed that it is a proteinase-K-sensitive and heat-labile protein with a molecular mass over 100 kDa. Employing a flow cytometric analysis of the cell cycle, it was found that U937 cells, treated either with conditioned medium from dengue virus infected monocytes or with CF, but not treated with conditioned medium from noninfected monocytes, showed cell-cycle arrest in G1 phase by 48 hr. This suppressive effect of CF on U937 growth was dose- and time-dependent. These results suggest that dengue virus-infected monocytes may produce CF to target myeloid cells, resulting in the hematological changes observed in patients with dengue fever. © 1995 Wiley-Liss, Inc.     

The effect of the platelet-activating factor antagonist, BN 52021, on human natural killer cell-mediated cytotoxicity.

The influence of the platelet-activating factor (PAF) antagonist, BN 52021, on human natural killer (NK) cell cytotoxicity against K 562 target cells was determined. Cytotoxicity was measured by a short-term (4 hr) 51Cr-release assay. The cytotoxicity was significantly reduced in the presence of PAF antagonist at concentrations from 30 to 120 microM. This reduction of killing was not due to the impairment of binding of effector cells to target cells. Pretreatment of K 562 target cells with the PAF antagonist led to a greater inhibition of NK cell cytotoxicity compared with that observed when the effector cells were preincubated with BN 52021. Thus, the inhibition of cytotoxicity appears to be due to an effect of BN 52021 on target cells rather than on lymphocytes. Furthermore, the increase in NK activity induced by interferon was less pronounced when BN 52021 was added in the incubation medium. The natural cytotoxicity of platelet-depleted or large granular lymphocyte-enriched effector cell populations was inhibited by the PAF antagonist in a similar manner. The effect of BN 52021 appears to be related to its specific PAF antagonistic activity since a similar action on NK cells was noted with two other structurally unrelated PAF antagonists, BN 52111 and WEB 2086. In contrast, Ginkgolide J (BN 52024), which is structurally related to BN 52021 but lacks PAF antagonistic activity, was ineffective in inhibiting NK cell cytotoxicity. Finally, synthetic PAF induces a dose-dependent cytotoxic action on K 562 cells and this effect of the autacoid is inhibited by BN 52021. These observations provide indirect evidence that PAF could play a role in the mechanism(s) of NK cytotoxity.     

Enhancement of Human Spontaneous Cell-Mediated Cytotoxicity by a Monoclonal Antibody against the Large Sialoglycoprotein (CD 43) on Peripheral Blood Lymphocytes

A murine monoclonal antibody, MoAb B1B6 (IgG1 chi), which recognizes the large sialoglycoprotein (LSGP) on human peripheral blood lymphocytes (PBL) effectively enhanced the spontaneous cytotoxicity of these cells against the natural killer (NK)-sensitive target cells K562 and Molt-4. Whereas preincubation of the lymphocytes with MoAb B1B6 resulted in increased cytotoxicity, preincubation of the target cells had no effect, indicating that the MoAb amplified cytotoxicity at the effector cell level. Kinetic analysis of the data revealed no differences between the control and the MoAb-treated lymphocytes with regard to Vmax, usually considered to reflect the overall lytic potential of the cells. The slopes of the saturation curves, however, differed significantly for the two cell populations, indicating a substantial increment in the activity of the MoAb-treated cells. When studied at the single cell level and with K562 as targets, treatment of PBL with the MoAb resulted in the recruitment of new effector lymphocytes from the pool of non-binding cells. In contrast, when Molt-4 cells were employed as targets, no additional effector cells were recruited. These results indicate that the enhanced cytotoxicity induced by MoAb B1B6 is the result of either recruitment of new effector lymphocytes or of an increased recycling capacity of preexisting effector cells. Together with previous observations, these findings support the conclusion that LSGP belongs to the set of surface molecules which regulate human lymphocyte activation.     

Inhibition of SK cell activity in frogs by certain drugs and sugars

We present similarities between mammalian natural killer (NK) cells and (anuran amphibian) frog spontaneous killer (SK) cells. A cytotoxic assay utilizing allogeneic erythrocytes as target cells was used and lysis assessed by measuring release of hemoglobin. SK effector cells, just as mammalian NK cells, are not sensitive to cycloheximide nor most simple sugars (50 mM glucose, glucose-6-phosphate, galactose, fucose, mannose). However, SK activity is inhibited by chloroquine, colchicine and mannose-6-phosphate. When SK cells were co-incubated with mammalian tumor cells, they were able to lyse only the NK-sensitive target YAC-1, but not other mammalian tumor cell targets including K562, Molt-4, Raji, P815 and EL4. Lysis of YAC-1 cells was also inhibited by colchicine and chloroquine. These results allow speculation on the evolution of cell mediated cytotoxicity since natural cytotoxic cells are present in ectothermic vertebrates.     

Lack of effect of LTB4 on human natural killer cell activity and T lymphocyte transformation

Human natural killer (NK) cell activity was not significantly affected by leukotriene (LT) B4 over a wide concentration range (10(-6)-10(-14) M), whether added directly to the NK cell assay or after preincubation of effector cells for 2 h with the drug before addition of Cr labelled K562 target cells. In addition, LTB4 did not affect the kinetics of NK cell mediated cytotoxicity. Addition of LTB4 (10(-6)-10(-10) M) to concanavalin A (Con A) or phytohaemagglutinin (PHA)-stimulated human peripheral blood mononuclear cells (PBMC) had no significant effect on proliferation measured by tritiated thymidine incorporation, using both optimal and sub-optimal mitogen concentrations. Whilst it is clear that LTB4 is an important mediator of inflammation involving polymorphonuclear (PMN) leukocytes in vivo, it does not affect NK cell or T cell function in vitro.     

Direct evidence for release of pore-forming protein during NK cellular lysis.

In the present study, we have examined the exocytosis model for cellular lysis. Using monoclonal antibodies reactive with human pore-forming protein (PFP), we examined the localization of PFP at the interaction site of natural killer (NK) cells and the NK tumor targets K562 and Molt-4 as well as during antibody-dependent cellular cytotoxicity. Following the interaction of effector-target cell contact, an increased frequency of PFP was detected on the effector surface, in the micro-environment, and on the target surface of the interaction site. This temporal deposition of PFP was paralleled by loss of target cell integrity and release of chromium. In addition, selective deposition of PFP was seen at the interaction site of the target cell compared to other target cell regions. Collectively, these results are consistent with the exocytosis model and further support the hypothesis that PFP is one of the secreted moieties involved in NK cellular cytotoxicity.     

索拉非尼联合柔红霉素对白血病K562细胞的抑制作用

目的:探讨小分子Raf激酶抑制剂索拉非尼(sorafenib)联合柔红霉素(DNR)对白血病细胞K562及U937的抑制作用及可能的分子机制。方法:MTT法测定索拉非尼和柔红霉素单独作用于K562和U937细胞的抑制率及DNR IC10联合不同浓度索拉非尼作用于K562及U937的联合抑制率;流式细胞AnnexinⅤ/PI法测定单药及联合用药后K562细胞的凋亡率以及Hoechst33258染色法观察单药及联合作用后细胞凋亡形态的改变;West-ern blotting法测定索拉非尼、DNR及U0126对K562及U937p-ERK1/2的影响;根据金氏方程证明2种药物联合抑制率及凋亡是否有协同作用。结果:MTT法测定索拉非尼联合DNR对K562及U937均有协同抑制作用(q1.15,P0.01);流式细胞AnnexinⅤ/PI和Hoechst33258染色法,均证明索拉非尼联合柔红霉素能联合诱导K562细胞凋亡(q1.15,P0.05),两者有明显的一致性;K562细胞的基础pERK1/2蛋白水平明显高于U937细胞(P0.01),索拉非尼和U0126都能够显著抑制K562细胞p-ERK1/2水平;U0126联合DNR存在协同抑制K562细胞的作用。结论:索拉非尼联合DNR作用于白血病细胞K562、U937存在协同抑制及凋亡诱导作用;DNR对U937的抑制作用明显高于K562;索拉非尼对K562的敏感性高于U937细胞;索拉非尼可能通过下调p-ERK1/2水平增加柔红霉素抗白血病细胞的效应。