Fludarabine-induced immunosuppression is associated with ... - Nature

© 1999 Nature America Inc. • http://medicine.nature.com

ARTICLES

Fludarabine-induced immunosuppression is associated with inhibition of STAT1 signaling DAVID A. FRANK1,2, SUDIPTA MAHAJAN1 & JEROME RITZ1,2


1

Department of Adult Oncology, Dana-Farber Cancer Institute, 44 Binney St. Boston, Massachusetts 02115, USA 2 Department of Medicine, Harvard Medical School, Boston, Massachusetts 02115, USA Correspondence should be addressed to D.A.F.; email: [email protected]

Fludarabine is a nucleoside analog used in the treatment of hematologic malignancies1 that can induce severe and prolonged immunosuppression2,3. Although it can be incorporated into the DNA of dividing cells, fludarabine is also a potent inhibitor of cells with a low growth fraction4,5, thus it must have other mechanisms of action. STAT1, which is activated in response to many lymphocyte-activating cytokines including the interferons, is essential for cell-mediated immunity, as the absence of this protein is associated with prominent defects in the ability to control viral infections6,7. Here we show that fludarabine, but not the immunosuppressant cyclosporine A, inhibits the cytokine-induced activation of STAT1 and STAT1-dependent gene transcription in normal resting or activated lymphocytes. Fludarabine caused a specific depletion of STAT1 protein (and mRNA) but not of other STATs. This loss of STAT1 was also seen in cells from patients treated with fludarabine in vivo. Brief exposure to fludarabine led to a sustained loss of STAT1, analogous to the prolonged period of immunosuppression induced by exposure to the drug in vivo. Thus, STAT1 may be a useful target in the development of new immunosuppressive and antineoplastic agents. Given the importance of STAT1 for cell-mediated immunity, and the ability of fludarabine to cause severe immunosuppression independent of its incorporation into DNA, we assessed the effect of fludarabine on cytokine-induced STAT activation. We collected peripheral blood mononuclear cells (PBMC) from normal human donors and either left them untreated or exposed them to fludarabine, then treated them with interferon (IFN)-α. We analyzed their nuclear extracts by electrophoretic mobility-shift assay using a double-stranded oligonucleotide known to bind to activated STATs (ref. 8). In cells left untreated, IFN-α induced a specific STAT1-containing protein–DNA complex (Fig. 1a, lanes 1 and 2, complex I), which was abolished by incubation with an antibody against STAT1 (Fig. 1a, lanes 3 and 4) or 100-fold excess unlabeled probe (Fig. 1a, lanes 5 and 6). After exposure to fludarabine for as little as 6 hours, IFN-α-induced STAT activation was substantially reduced, and after 24 hours of exposure to fludarabine, almost no IFN-α-induced DNA binding could be detected (Fig. 1a, lane 8). The inhibition of STAT1 mediated signaling is not common to immunosuppressive agents in general, as cyclosporine A had no effect on STAT1 activation (Fig. 1a, lanes 9 and 10). The loss of DNA binding of STAT1 was associated with an absence of tyrosine-phosphorylated STAT1 from the nuclei or cytoplasm of cells pre-treated with fludarabine (Fig. 1b). To determine whether the loss of STAT1 activation in response to IFN-α represented a defect in IFN-α signaling, or a more gener444

alized loss of STAT1 activation, we incubated untreated and fludarabine-treated PBMC with IL-2 or IL-6 and assessed STAT1 activation. Incubation with fludarabine for 24 hours led to more than 80% inhibition in STAT1 activation in response to either of these cytokines (Fig. 1c). The tyrosine phosphorylation of STAT5 can also be detected by the phospho-STAT1 antibody used in these western blot analyses9,10. In contrast to STAT1, the tyrosine phosphorylation of STAT5 induced by IL-2 was only minimally reduced after fludarabine treatment, which we confirmed by re-probing the membrane with an antibody specific for tyrosine phosphorylated STAT5 (ref. 11) (Fig. 1c, lower panel, lanes 3 and 4). Thus, fludarabine induces a specific defect in STAT1 activation in response to many physiologically important cytokines. Fludarabine had no substantial effect on the tyrosine phosphorylation of either Jak1 or tyk2 in response to IFN-α (data not shown), indicating that the loss of STAT activation is distal to the kinase in the pathway. Therefore, we next determined whether the amount of STAT1 protein was decreased by fludarabine treatment. Resting PBMC treated for 24 hours with fludarabine showed a loss of more than 90% of STAT1 from treated cells (Fig. 2a). Mitogen-activated lymphocytes showed a similar loss of STAT1 (data not shown). This did not represent a nonspecific toxicity of fludarabine, as total cellular protein and levels of STAT2, STAT3, STAT5, STAT6, ERK1 (Fig. 2a), Jak1 and tyk2 (not shown) were unaffected by this treatment. Furthermore, the viability of these cells was more than 95% in both treated and untreated cells, as measured by trypan blue exclusion (data not shown). The depletion of STAT1 after fludarabine treatment was also seen in cell lines derived from human lymphocytes or natural killer cells. NKL or NK3.3 cells treated with fludarabine showed a specific loss of STAT1 protein similar to that seen in primary cells (Fig. 2b). Fludarabine also causes a loss of STAT1 (of between 30 and 95%) in non-lymphoid cell lines, including those derived from human breast carcinoma, colon carcinoma and neuroblastoma, and mouse fibroblasts (data not shown). Thus, loss of STAT1 seems to be a common response to fludarabine treatment. The loss of STAT1 is not caused by all nucleoside analogs, as pentostatin does not induce this effect in PBMC (data not shown). The splice variants STAT1α and STAT1β (ref. 12) are both depleted by fludarabine treatment; therefore, we next determined whether the mechanism of loss of STAT1 was at the level of mRNA. To examine this, we did RT–PCR on RNA from PBMC that were untreated or treated with fludarabine for 24 hours. STAT1 mRNA was reduced to less than 5% of control levels after exposure to fludarabine (Fig. 2c). Thus, fludarabine leads to a loss of STAT1 from lymphocytes through decreased synthesis. NATURE MEDICINE • VOLUME 5 • NUMBER 4 • APRIL 1999


ARTICLES Fig. 1 Fludarabine leads to a loss of anti100X STAT1 activation after cytokine treatSTAT1 Comp. Fludara CsA – Fludara ment of PBMC. a, PBMC incubated IFN-α: IFN-α: with medium alone (lanes 1–6), fludarabine (Flu; lanes 7 and 8) or cyclosporine A (CsA; lanes 9 and 10) STAT 1 α were then left untreated (lanes 1, 3, STAT 1 β 5, 7 and 9) or treated with IFN-α (lanes 2, 4, 6, 8 and 10), and nuclear 1 2 3 4 extracts were assessed by electrophoretic mobility-shift assays Fludara: using radiolabeled STAT binding seIL-6: quence from the IRF-1 promoter (I, right margin). Antibody against IL-2: STAT1 (lanes 3 and 4) or 100-fold exkDa cess of unlabeled probe (lanes 5 and 6) were included as indicated. STAT 5 b, Proteins from nuclear extracts pre90 STAT 1 α pared as in a were assessed by westSTAT 1 β ern blot analysis with an antibody specific for STAT1 phosphorylated on 70 1 2 3 4 5 6 tyr-701 (positions indicated along right margin). Top line: –, no fludarabine; Fludara, with fludarabine. 1 2 3 4 5 6 7 8 9 10 c, PBMC cultured without (lanes 1, 3 and 5) or with (lanes 2, 4 and 6) flu(upper) or tyrosine-phosphorylated STAT5 (lower); the former antibody darabine (Fludara) were then left untreated (lanes 1 and 2) or treated with ILcross-reacts with tyrosine-phosphorylated STAT5 (refs. 9,10). Right margin, 2 (lanes 3 and 4) or IL-6 (lanes 5 and 6). Extracts were assessed by western positions of STAT proteins; left margin, molecular-weight marker positions. blot analysis with an antibody against tyrosine-phosphorylated STAT1

a

b


c

Given that the STAT1 promoter contains STAT binding sites13 and STAT1 protein is increased by IFN-γ (ref. 8), the loss of STAT1 protein may lead to an exponential decrease in STAT1 gene expression. To determine the importance of the loss of STAT1, we assessed STAT1-dependent gene transcription. PBMC untreated or treated with fludarabine for 24 hours were then stimulated with IFN-γ for 30 minutes, and the level of IRF-1 mRNA was determined by semi-quantitative RT–PCR. The induction of IRF-1 in response to IFN-γ depends only on the activation of STAT1 (ref. 14). Whereas IFN-γ led to a prominent induction of transcription of IRF-1 in untreated cells, cells treated with fludarabine showed only minimal induction of IRF-1 after IFN-γ treatment (Fig. 2d). The mRNA levels of GAPDH were unaffected by either fludarabine or IFN-γ treatment. Thus, fludarabine-induced loss of STAT1 leads to a spe-

cific loss of STAT1-mediated gene activation. One prominent characteristic of fludarabine when used in vivo is a prolonged period of immunosuppression that persists long after treatment2,3. Thus, we determined whether the depletion of STAT1 induced by fludarabine might be a long-lived event in vitro. After treating PBMC with fludarabine for 24 hours, we removed the drug from the medium and cultured the cells without fludarabine. For 48 hours after fludarabine treatment, cellular levels of STAT1 remained less than 10% of those of untreated cells (Fig. 3a, upper panel); the loss of STAT1 persisted for at least 96 hours of culture (data not shown). During this time, cells remained viable, and the levels of other STAT proteins and the kinase ERK1 (Fig. 3a, lower panel) were unaffected. Thus, even brief exposure of normal lymphocytes to fludarabine in vitro is associated with a prolonged loss of STAT1.

b

a

NK3.3

Fludara

–

c

NKL

Fludara:

Fludara:

STAT 1

kDa STAT 1 α STAT 1 β

87

1 STAT1

2

3

4

STAT2

5

6

STAT3

7

8

STAT5

9

10

STAT6

11

12

ERK1

1

GAPDH

69 1

2

3

4

d Fig. 2 Fludarabine leads to a selective loss of STAT1 protein and mRNA from lymphoid cells. a, PBMC untreated (–) or treated (+) with fludarabine (Fludara, top row) were assessed by western blot analysis of cell extracts, with antibodies against STAT1 (lanes 1 and 2), STAT2 (lanes 3 and 4), STAT3 (lanes 5 and 6), STAT5 (lanes 7 and 8), STAT6 (lanes 9 and 10) or ERK1 (lanes 11 and 12). b, NK3.3 cells (lanes 1 and 2) and NKL cells (lanes 3 and 4) were cultured without (–) or with (+) fludarabine (Fludara), and whole-cell extracts were assessed by STAT1 western blot analyses. Right margin, positions of STAT proteins c, PBMC untreated (lane 1) or treated with fludarabine (lane 2) were analyzed by RT–PCR to amplify mRNA for STAT1 (upper) or GAPDH (lower). d, PBMC untreated (lanes 1 and 2) or treated with fludarabine (lanes 3 and 4) were then left untreated (lanes 1 and 3) or treated with IFN-γ (lanes 2 and 4) and analyzed by RT–PCR to amplify mRNA for IRF-1 (upper) or GAPDH (lower). NATURE MEDICINE • VOLUME 5 • NUMBER 4 • APRIL 1999

2

Fludara

–

IFN-γ: IRF-1

1

2

3

4 GAPDH

445


a

–

Time (h): 0 kDa

24

48

0

24

48 STAT 1

91 1

2

3

4

5

6 ERK 1

38

c

Fludara: PHA:

d CLL

NKL

Fludara: kDa STAT 1 α STAT 1 β

87 69 1

2

3

Fludara: PHA:

Because STAT1 is phosphorylated during mitogen activation of PBMC (refs. 11,15), we determined whether PBMC in which STAT1 had been depleted by transient treatment with fludarabine could become functionally activated in response to phytohemagglutinin (PHA). We treated PBMC with fludarabine for 24 hours, then washed cells free of drug and activated them by exposure to the mitogen PHA. Before PHA treatment, more than 99% of cells were in G0/G1 (data not shown), and thus fludarabine could not be incorporated into DNA. The ability of the fludarabine-treated PBMC to undergo DNA synthesis (Fig. 3b) and to secrete IFN-γ (Fig. 3c) after mitogen stimulation was severely depressed compared with that of PBMC not treated with fludarabine. Thus, fludarabine-induced STAT1 depletion is associated with a persistent defect in lymphocyte activation. Although fludarabine treatment in vitro causes a loss of STAT1, this may not reflect the action of this drug in vivo. To assess the effect of fludarabine on STAT1 in lymphocytes from patients treated with this agent, we obtained PBMC from a patient with chronic lymphocytic leukemia 24 hours after the patient received an infusion of fludarabine. Compared with cells from a chronic lymphocytic leukemia patient before treatment (Fig. 3d, lane 1) or untreated NKL cells (Fig. 3d, lane 3), PBMC recovered from the fludarabine-treated patient showed a greater than 90% decrease in STAT1 (Fig. 3d, lane 2). Thus, fludarabine induces the specific loss of STAT1 in vivo in humans, as well as in vitro. The normal development but profound immune deficiency of STAT1-deficient mice6,7 indicates that selective inhibition of STAT1 is a useful target for modulating the human immune response. As with any immunosuppressant, there may be increased susceptibility to infections or malignancies16. Although fludarabine can cause STAT1 depletion in a variety of cell types, the absence of non-immune defects in STAT1-deficient mice or in patients treated with fludarabine indicates that, in contrast to current immunosuppressants such as cyclosporine A, alkylating agents and corticosteroids, drugs that target STAT1 may have few non-immune effects. 446

b

Fludara

[3H]-thymidine incorporation (cpm)

Fig. 3 Fludarabine-induced inhibition in vitro and in vivo. a, Brief treatment of PBMC with fludarabine leads to prolonged loss of STAT1. PBMC untreated (–; lanes 1–3) or treated with fludarabine (Fludara; lanes 4–6) for 24 h were then incubated without fludarabine (time ,in hours, above blot). Cell extracts were assessed by western blot analysis for STAT1 (upper) and ERK1 (lower), using the same membrane. b, Transient treatment of PBMC with fludarabine leads to inhibition of mitogen-induced proliferation. PBMC untreated (–) or treated (+) with fludarabine (Fludara) were then cultured with (+) or without (–) PHA, and proliferation was assessed by 3H-thymidine incorporation. Results from a representative experiment done in triplicate are shown. c, Transient treatment of PBMC with fludarabine leads to inhibition of mitogen-induced IFN-γ production. PBMC untreated (–) or treated (+) with fludarabine (Flurdara) were then cultured with (+) or without (–) PHA; the concentration of IFN-γ in the medium was then determined. Results from a representative experiment done in duplicate are shown. d, Fludarabine treatment in vivo leads to a loss of STAT1. PBMC isolated from patients with chronic lymphocytic leukemia either untreated (lane 1) or 24 h after an infusion of fludarabine (lane 2), as well as NKL cells (lane 3), were assessed by western blot analysis for STAT1.

IFN-γ production (pg/ml)


ARTICLES

Methods Cell lines. NKL (ref. 17) and NK3.3 (ref. 18) cells were grown in RPMI 1640 medium supplemented with 15% fetal bovine serum, 1 mM sodium pyruvate, 250 U/ml penicillin and 250 µg/ml streptomycin, at 37 °C in a humidified atmosphere containing 5% CO2. For NKL cells, the medium was supplemented with 100 U/ml IL-2; for NK3.3 cells, with 10% Lymphocult (Biotest Diagnostics, Denville, New Jersey) and 50 U/ml IL-2. Primary cells. PBMC were isolated from anticoagulated blood obtained from normal donors by density gradient centrifugation using Ficoll-Paque (Pharmacia). Approximately 80% of cells recovered were T cells, which were activated by being cultured in RPMI 1640 medium containing 15% fetal calf serum and 2.5 µg/ml PHA (Murex, Dartford, UK) for 48 h. Cell treatments. Both cell lines and primary cells were incubated with fludarabine for 24 h. IFN-α treatment was 15 min; IFN-γ treatment was 60 min. PBMC were isolated from the chronic lymphocytic leukemia patient 24 h after infusion of 25 mg/m2 fludarabine. Cell extracts. After being stimulated, cells were placed on ice, washed once with ice cold phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4, pH 7.4) and extracted in buffer containing 10 mM Tris, pH 8.0, 0.5% NP-40, 250 mM NaCl, 10 mM sodium orthovanadate, 100 µM phenylmethylsulfonyl fluoride (PMSF), 1 µg/ml leupeptin, 1 µg/ml pepstatin and 1 µg/ml aprotinin. Insoluble material was removed by centrifugation at 12,000g for 1 min. Western blot analysis. Protein was separated by SDS–PAGE, transferred to nitrocellulose and blocked with 5% non-fat dry milk (for anti-phosphoSTAT blots) or 5% BSA (for other blots) in TBST (100 mM Tris, pH 8.0, 150 mM NaCl and 0.05% Tween-20). The primary antibody was diluted 1:20,000 in TBST containing 3% BSA and was incubated with the blot for 1 h at room temperature. After being washed extensively, the blot was incubated for 1 h at room temperature with horse radish peroxidaseconjugated secondary antibody (Calbiochem, La Jolla, California) diluted 1:20,000 in TBST containing 1% BSA. After blots were washed extensively, bound antibody was detected by chemiluminescence (Renaissance Kit; NEN). Blots were ‘stripped’ by incubation in 62.5 mM Tris, pH 6.9, containing 2% SDS and 100 mM β-mercaptoethanol for 1 h at 65 °C. Antibodies and reagents. Antibodies against the tyrosine phosphorylated form of STAT1 (ref. 19) and STAT5 (ref. 11) were generated as described. NATURE MEDICINE • VOLUME 5 • NUMBER 4 • APRIL 1999


ARTICLES


Antibodies against STAT1, STAT2, STAT3, STAT5, STAT6, Jak1 and tyk2 were purchased from Santa Cruz Biochemicals (Santa Cruz, California). Antibodies against ERK1 were purchased from New England Biolabs (Beverly, Massachusetts). IFN-α was obtained from Hoffman-LaRoche (Nutley, New Jersey), and used at a concentration of 500 U/ml. IFN-γ was obtained from Genzyme (Cambridge, Massachusetts) and used at a concentration of 500 U/ml. IL-2 was obtained from Amgen (Thousand Oaks, California), and used at a concentration of 100 U/ml. IL-6 was provided by K. Anderson (Dana-Farber Cancer Institute), and used at a concentration of 5 ng/ml. Cyclosporine A was provided by B. Bierer (Dana-Farber Cancer Institute), and used at a concentration of 100 nM. Fludarabine was obtained from Berlex (Richmond, California), and was used at a concentration of 50 µM. Nuclear extracts. Cells were stored on ice, washed once with cold PBS, then resuspended in 5 ml of hypotonic buffer (10 mM Tris, pH 7.4, 10 mM NaCl and 6 mM MgCl2) and incubated on ice for 5 min. Thereafter, the cells were centrifuged and resuspended in 0.8 ml of hypotonic buffer containing 1 mM β-mercaptoethanol, 10 µg/ml phenylmethylsulfonyl fluoride and 1 mM sodium orthovanadate. Cells were disrupted by being sheared in a Dounce homogenizer (type b pestle, 25 strokes). Nuclei were collected by a 10-second centrifugation at 12,000g. The supernatant was removed and cleared by centrifugation at 12,000g for 10 min at 4 °C; this was called the cytoplasmic extract. The nuclear pellet was washed once with hypotonic buffer, then resuspended in three volumes of high-salt buffer (20 mM Hepes, pH 7.9, 420 mM NaCl, 25% glycerol, 1.5 mM MgCl2, 0.2 mM EDTA, 1 mM β-mercaptoethanol, 1 mM sodium orthovanadate, 1 µg/ml leupeptin, 1 µg/ml pepstatin, 1 µg/ml aprotinin and 100 µM phenylmethylsulfonyl fluoride), and rocked at 4 °C for 30 min. Intact nuclei were removed by centrifugation at 12,000g for 3 min at 4 °C, and the supernatant was recovered and called the nuclear extract. DNA binding assay. Nuclear extract (2 µl) was mixed with 1 ng 32Plabeled oligonucleotide (5’–AGCCTGATTTCCCCGAAATGACGGC–3’ and its complement; ref. 8) in 10 µl of binding buffer (25 mM HEPES, pH 7.9, 100 µM EGTA, 200 µM MgCl2, 500 µM dithiothreitol, 1 µg/µl BSA, 0.2 µg/µl poly dI:dC, 1% Ficoll and 0.1 µg/µl herring testis DNA). The incubation was done at room temperature for 15 min. For antibody competition, antiserum (1:200 dilution) was added at the end of the binding reaction and incubated at 4 °C for an additional 15 min. The products of the binding reaction were then separated on a 4% acrylamide gel in 0.2x Trisborate/EDTA. RT–PCR. RNA was collected (RNeasy Kit; Qiagen, Valencia, California) and reverse-transcribed using Superscript IIRT (Life Technologies). PCR to amplify STAT1 used primers for a 306-bp fragment (sense, 5’–AGTGGTACGAACTTCAGCAGC–3’; anti-sense, 5’–TGATCATAGACATCTGGATTGG–3’), PCR to amplify IRF-1 used primers for a 537-bp fragment (sense, 5’–AAGGGAAATTACCTGAGGACATCAT–3’; anti-sense, 5’–ATACACTGGTCTCAGAACCTCATCTT–3’) and PCR to amplify GAPDH used primers for a 226-bp fragment (Perkin Elmer Applied Biosystems, Branchburg, New Jersey). Measurement of IFN-γ production. Untreated or fludarabine-treated PBMC were washed and plated in U-bottom wells at 3 × 104 cells/well (5 × 105 cells/ml) in medium alone or medium supplemented with 2.5 µg/ml PHA. After 48 h, supernatants were collected, and the IFN-γ concentration was measured by ELISA (Endogen, Cambridge, Massachusetts).

NATURE MEDICINE • VOLUME 5 • NUMBER 4 • APRIL 1999

Proliferation assay. Untreated or fludarabine-treated PBMC were washed and plated at 3 × 104 cells/well (5 × 105 cells/ml) in medium alone or medium supplemented with 2.5 µg/ml PHA for 48 h. During the last 6 h of the PHA incubation, 1 µCi of 3H-thymidine was added to the medium. Samples were collected with a 96 Mach II harvester (Tomtec, Orange, California) and 3H-thymidine incorporation was measured with a 1205 Betaplate liquid scintillation counter (Pharmacia, Turku, Finland).

Acknowledgments This work was supported by grants from the National Cancer Institute (CA41619 and CA66996) and a gift from M. and M. Rubin in honor of M. Idelson.

RECEIVED 30 DECEMBER 1998; ACCEPTED 24 FEBRUARY 1999 1. Plunkett, W. et al. Fludarabine: Pharmacokinetics, mechanisms of action, and rationales for combination therapies. Semin. Oncol. 20, 2–12 (1993). 2. Cheson, B.D. New antimetabolites in the treatment of human malignancies. Semin. Oncol. 19, 695–706 (1992). 3. Fenchel, K. et al. Clinical experience with fludarabine and its immunosuppressive effects in pretreated chronic lymphocytic leukemias and low-grade lymphomas. Leuk. Lymphoma 18, 485–492 (1995). 4. Keating, M.J. Fludarabine phosphate in the treatment of chronic lymphocytic leukemia. Semin. Oncol. 17, 49–62 (1990). 5. Hochster, H. & Cassileth, P. Fludarabine phosphate therapy of non-Hodgkin’s lymphoma. Semin. Oncol. 17, 63–65 (1990). 6. Durbin, J.E., Hackenmiller, R., Simon, M.C. & Levy, D.E. Targeted disruption of the mouse Stat1 gene results in compromised innate immunity to viral disease. Cell 84, 443–450 (1996). 7. Meraz, M.A. et al. Targeted disruption of the Stat1 gene in mice reveals unexpected physiologic specificity of the JAK-STAT signaling pathway. Cell 84, 431–442 (1996). 8. Bonni, A., Frank, D.A., Schindler, C. & Greenberg, M.E. Characterization of a pathway for ciliary neurotrophic factor signaling to the nucleus. Science 262, 1575–1579 (1993). 9. Frank, D.A. & Varticovski, L. BCR/abl leads to the constitutive activation of Stat proteins, and shares an epitope with tyrosine phosphorylated Stats. Leukemia 10, 1724–1730 (1996). 10. Sattler, M. et al. The thrombopoietin receptor c-MPL activates JAK2 and TYK2 tyrosine kinases. Exp. Hematol. 23, 1040–1048 (1995). 11. Gollob, J.A. et al. Altered IL-12 responsiveness in Th1 and Th2 cells is associated with the differential activation of STAT5 and STAT1. Blood 91, 1341–1354 (1998). 12. Schindler, C., Fu, X.-Y., Improta, T., Aebersold, R. & Darnell, J.E., Jr. Proteins of transcription factor ISGF-3: one gene encodes the 91 and 84 kDa ISGF-3 proteins that are activated by interferon-α. Proc. Natl. Acad. Sci. USA 89, 7836–7839 (1992). 13. Kolla, V., Lindner, D.J., Weihua, X., Borden, E.C. & Kalvakolanu, D.V. Modulation of interferon (IFN)-inducible gene expression by retinoic acid. Up-regulation of Stat1 protein in IFN-unresponsive cells. J. Biol. Chem. 271, 10508–10519 (1996). 14. Pine, R., Canova, A. & Schindler, C. Tyrosine phosphorylated p91 binds to a single element in the ISGF2/IRF-1 promoter to mediate induction by IFNα and IFNγ, and is likely to autoregulate the p91 gene. EMBO J. 13, 158–167 (1994). 15. Gimeno, R., Codony-Servat, J., Plana, M., Rodriguez-Sanchez, J.L. & Juarez, C. Stat1 implication in the immune response to superantigens in vivo. J. Immunol. 156, 1378–1386 (1996). 16. Kaplan, D.H. et al. Demonstration of an interferon γ-dependent tumor surveillance system in immunocompetent mice. Proc. Natl. Acad. Sci. USA 95, 7556–7561 (1998). 17. Robertson, M.J. et al. Characterization of a cell line, NKL, derived from an aggressive human natural killer cell leukemia. Exp. Hem. 24, 406–415 (1996). 18. Kornbluth, J., Flomenberg, N. & Dupont, B. Cell surface phenotype of a cloned line of human natural killer cells. J. Immunol. 129, 2831–2837 (1982). 19. Frank, D.A., Robertson, M., Bonni, A., Ritz, J. & Greenberg, M.E. IL-2 signaling involves the phosphorylation of novel Stat proteins. Proc. Natl. Acad. Sci. 92, 7779–7783 (1995).

447