A Preferential Role for STAT5, not Constitutively Active STAT3, in Promoting Survival of a Human Lymphoid Tumor¹

Signal transducers and activators of transcription are an evolutionarily conserved family of transcription factors that play diverse roles in embryonic cell development, differentiation, proliferation, migration, survival, and apoptosis. Their activity can be regulated by numerous cytokines, hormones, and growth factors. In lymphoid cells, STAT1, STAT3, STAT5a, and STAT5b are thought to be responsible for cell survival and growth, while STAT2, 4, and 6 are preferentially involved in differentiation (1).

Presently, STAT3, STAT5a, and STAT5b are found hyperactivated in a number of malignancies, including human T cell leukemia virus-1 (HTLV-1),³ Src, v-Abl, EBV-transformed cell lines, and some patient lymphocytes (2, 3, 4, 5, 6, 7, 8, 9), as reviewed previously (1). These findings suggest that constitutively tyrosine-phosphorylated STATs play key roles in lymphoid tumorigenesis.

STAT3, first cloned as acute-phase response factor (10), is ubiquitously expressed and activated by many ligands, including IL-6, epidermal growth factor, platelet-derived growth factor, oncostatin M, and leukemia-inhibitory factor. In some instances, Jaks, growth factor receptor tyrosine kinases, and nonreceptor tyrosine kinases such as c-Src may serve as activators. Although STAT3 is clearly essential for embryogenesis, it is not critical for T cell development, which may only alter a loss of IL-6 protection against apoptosis (11, 12). STAT3-deficient T cells show a partial defect in IL-2-induced cell proliferation and impaired IL-2Rα chain expression (13). Many groups have shown that STAT3 is constitutively activated in numerous malignancies, including leukemias, lymphomas, breast carcinoma, multiple myeloma, and head and neck, brain, lung, and prostate cancers (14, 15, 16, 17, 18, 19, 20), as reviewed previously (21), supporting the notion of STAT3 as a viable therapeutic target for these cancers (22, 23, 24, 25, 26).

Like STAT3, STAT5a (originally identified as a mammary gland factor) and STAT5b are activated by multiple cytokines in addition to IL-2, including IL-3, IL-5, IL-7, IL-9, IL-15, various growth factors, as well as prolactin, growth hormone, and erythropoietin (1). Mice deficient in both STAT5 genes exhibit impaired immune function, incapable of normal proliferation in response to IL-2 (27), suggesting that STAT5a and STAT5b exert compensatory functions within lymphocytes. Thus, STAT3 and STAT5 most likely contribute to lymphoid cell survival (21, 28, 29). Earlier work from our group found that depletion of STAT5a and STAT5b in mature human peripheral T cells promotes apoptosis (30), supporting this model.

The present work sought to determine whether constitutively active STATs are present and contribute to the survival of a human lymphoma-derived tumor model, YT. STAT3 was the only STAT found to be constitutively tyrosine phosphorylated. Depletion of STAT3 by antisense oligonucleotides caused cell death with protracted kinetics (72–96 h). Interestingly, depletion of nonhyperactivated STAT5 resulted in more rapid and significant death following their depletion (30). To assess differences in survival genes regulated by STAT3 vs STAT5, microarray analysis was performed. The results indicated that regulation of the NF-κB-dependent pathway via its activators such as TNFR-associated factor (Traf)2, Traf5, and B cell leukemia (Bcl)/lymphoma 10 (Bcl10) may contribute to the antiapoptotic pathways critical for survival of this tumor and possible differences between STAT3 and STAT5.

The human YT cell line was maintained in RPMI 1640 medium with 10% FCS, and in some instances stimulated with 100 nM human rIL-2 (Hoffmann-LaRoche) at 37°C for 10 min, as previously reported (31). Helenalin (Axxora Platform) treatment conditions are noted within the figure legend.

Cells lysis, immunoprecipitation, and Western blots with Abs to STAT1, STAT3, STAT5a, or STAT5b were performed, as previously described (31). Anti-phosphotyrosine mAb was obtained from Upstate Biotechnology, monoclonal anti-STAT5 Ab from BD Pharmingen, monoclonal Bcl2 from Sigma-Aldrich, and Bcl10 mAb from Imgenex, and all Abs were used at a dilution of 1/1000. For all samples, total protein was determined by the bicinchoninic acid method (Pierce).

YT cells (20 × 10⁶/ml) were metabolically labeled with [³⁵S]methionine (0.5 mCi/ml; NEN) for 2 h at 37°C and stimulated with 100 nM IL-2 for 10 min. Cell lysis and immunoprecipitation were performed, as described above. IL-2Rβ Ab was obtained from Santa Cruz Biotechnology. Immunoprecipitated proteins were subjected to NEPHGE in the first dimension at 300 V for 8 h using 2% ampholytes (ranging from pH 3.5 to 10; Pharmacia). After equilibration of the tube gels in SDS-sample buffer at room temperature for 1 h, second dimension SDS-PAGE was performed (7.5%) and proteins were visualized by autoradiography.

YT cells were treated with antisense ODN to STAT3 and STAT5 by electroporation, as described earlier (32). The ODN were synthesized by ISIS Pharmaceuticals using a phosphorothioate backbone with 2′-O-methoxyethyl modification of the five terminal nucleotides (underlined) to increase their stability (30, 33). The oligonucleotide sequences were as follows: STAT5a/b 5′-GGG CCT GGT CCA TGT ACG TG-3′. Control ODN was synthesized as a mixture of all four bases, resulting in a preparation that contains an equimolar mixture of all possible ODNs. STAT3 ODN: 5′-ACT CAA ACT GCC CTC CTG CT-3′ or mismatched control oligonucleotide: 5′-TCT GGC AAA GTG TCA GTA TG-3′.

Nuclear extracts were prepared, and the assays were conducted, as published earlier (31). Oligonucleotide sequences corresponding to the β-casein gene promoter for STAT5 5′-AGATTTCTAGGAATTCAATCC-3′, the c-fos promoter sis-inducible element (SIE) for STAT3 5′-TGTCGACATTTCCCGTAAATC-3′, and NF-κB/Rel consensus site 5′-AGT TGA GGG GAC TTT CCC AGG C-3′ were obtained from Santa Cruz Biotechnology.

Cell viability was assessed with 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) reagent (Promega) in triplicate, according to the manufacturer’s instructions. Three independent experiments were performed. The error bars represent SD.

Seventy-two-hour postelectroporation cells were deposited on slides by standard cytospin procedure, fixed in 4% paraformaldehyde, and incubated with TdT reaction mixture containing fluorescein-12-dUTP as a substrate, according to the manufacturer’s instructions (Roche Applied Science). Slides were mounted in Vectashield (Vector Laboratories) with propidium iodide to counterstain DNA for quantification. The slides were visualized by fluorescence microscopy. The values plotted on the y-axis represent the average of the percentage of apoptotic cells per field. Two independent experiments were performed.

Total RNA was isolated from ∼4–5 × 10⁶ cells using the RNeasy kit (Qiagen). The GEArray gene expression array (Superarray) was used to examine 96 apoptosis-related genes. Data were normalized to the expression level of housekeeping genes and analyzed with GEArrayAnalyzer software. To avoid representing genes below the detection level, those genes with a difference between the signal intensities (background subtracted/normalized values, Exp1 (control) − Exp2 (antisense)) of the corresponding spots lower than 1% of the intensity of GAPDH were ignored. Gene expression changes were plotted as log₂ ratio (antisense/control) to visualize the direction of changes, where log₂ ratio = ±1 equals 2-fold down/up-regulation.

Total RNA was isolated, as described above; DNase treated; and reverse transcribed (100 ng, 50°C for 30 min; Superscript II; Invitrogen Life Technologies) with specific reverse primers. Quantification based on real-time monitoring of amplification was determined using an ABI 7700 with SYBR Green II for Traf2 and Traf5 and with an amplicon-specific TaqMan probe for Bcl10. Absolute numbers of mRNA molecules were normalized to 18S rRNA to correct for RNA concentration differences. Samples were run in triplicates with one control reaction containing no reverse-transcriptase enzyme to test for potential DNA contamination. Values of transcripts in unknown samples were obtained by interpolating threshold cycle (PCR cycle number at threshold) values on a standard curve. Standard curves were prepared from known amounts of purified, PCR-amplified amplicon (for Traf2 and Traf5), and a sequence-specific synthetic oligonucleotide for Bcl10.

Primer sets for human Traf2 and Traf5 were obtained from PrimerBank at 〈http://pga.mgh.harvard.edu/primerbank〉. The identification numbers were 22027614a1 and 11321603a1, respectively. Primers and TaqMan probe sequences for human Bcl10 were as follows: forward, CAGGCTGGTCTTGAACT; reverse, GCTCGTGTCTGTAATCC; probe, CCTGGACTCAAGCAACCTACCCA.

YT is a human acute lymphoblastic lymphoma-derived NK-like tumor cell line in type II latency program of EBV infection (34) expressing latent membrane protein (LMP)-1, the transforming cell surface molecule capable of activating survival-promoting signaling pathways, including NF-κB and Jak/STAT molecules (35). Because STATs can be found constitutively activated (or tyrosine phosphorylated) in many lymphoid-derived tumors, we first investigated STAT activation status in YT cells.

All STAT proteins expressed in YT cells were immunoprecipitated and assessed for constitutive tyrosine phosphorylation. In contrast to the other STATs (data not shown), only STAT3 was determined to be constitutively active. STAT3 was immunoprecipitated from unstimulated (−) or IL-2-stimulated (+) YT cells, and then Western blotted with anti-phosphotyrosine and anti-phosphoserine STAT3 (S725) Abs (Fig. 1,A). To verify equal protein loading, the blot was reprobed with anti-STAT3 Ab (lower panel). The basal phosphorylation level of both the conserved tyrosine and serine residues (Fig. 1,A, lane a) indicates that STAT3 is persistently activated in YT cells, but IL-2 was able to cause additional phosphorylation of this population (Fig. 1 A, lane b).

Jak tyrosine kinases are believed to be the primary upstream regulators of STATs. It has been recognized previously that disruption of constitutively active Jak activity by selective inhibitors (such as AG490) can block STAT activation (36). To investigate whether any aberrant Jak activity might be present in YT cells, the cells were left untreated (−) or stimulated with IL-2 (+), and immune complexes were captured with Abs to Jak1 (Fig. 1,B, lanes a and b), Jak2 (c and d), Jak3 (e and f), and Tyk2 (g and h), followed by Western blot performed with anti-phosphotyrosine Abs. As shown in Fig. 1 B, none of the Jak kinases was active in the nonstimulated cells, but Jak3 was readily inducible upon IL-2 stimulation. Equal protein loading was verified by reprobing all blots with corresponding Abs (lower panel). These data suggest that Jak tyrosine kinases do not serve as the primary agent driving YT cell survival and a constitutively active STAT3.

Binding of cytokines induces receptor oligomerization and activation of downstream signaling molecules, including STATs (37, 38). Preformed receptor complexes for cytokines such as prolactin or erythropoietin in the absence of ligand can drive aberrant cell proliferation, survival, and leukemia formation (39, 40, 41). YT cells express ∼12,000 IL-2 binding sites (42, 43, 44), suggesting the high levels of IL-2R β and γ might exist as a preformed complex to drive STAT3 activity in the absence of ligand. To test this hypothesis, two-dimensional gel electrophoresis of YT cells metabolically labeled with [³⁵S]methionine stimulated with medium (−) or IL-2 (+) (Fig. 1 C) was performed. Immunocaptured IL-2Rβ chain showed IL-2Rγ is in the complex only in the presence of IL-2 (right panel). These data indicate that hyperactive STAT3 is not attributable to an activated pre-existing intermediate affinity IL-2R complex.

Depending on the cytokine stimulus and cell type, activated STATs are able to form homo- and/or heterodimers that can contribute to their signal specificity (45, 46, 47). This model suggests that constitutively active STAT3 may dimerize with other STATs in the absence of cytokine. To determine whether the malignant state of YT cells is due to STAT3 forming heterodimers with STATs generally involved in proliferative signals (e.g., STATs 1, 3, 5a, 5b), cells were stimulated with IL-2, lysed, and immunoprecipitated with Abs to STAT1 (Fig. 1 D, lanes a and b), STAT3 (lanes c and d), STAT5a (lanes e and f), and STAT5b (lanes g and h). Western blot with anti-phosphotyrosine Abs revealed that only STAT3 is hyperphosphorylated in unstimulated quiescent cells, whereas STAT5a and STAT5b retain latency, but become activated upon IL-2 stimulation. Reprobing of this blot with Abs to STAT1, STAT3, STAT5a, and STAT5b failed to detect STAT heterodimers, regardless of cytokine stimulus. These findings suggest that STAT3 does not cooperate with other STATs to regulate YT gene transcription.

Due to poor transient transfection efficiency of lymphoid cells to several methods (e.g., lipids, adenoviruses, etc.), we had developed and optimized means of successfully delivering antisense ODN into lymphoid cells with minimal effect on cell viability (32). Thus, to assess the functional role of hyperactive STAT3 in YT cell survival, it was depleted by specific phosphorothioate/methoxy-ethyl-modified STAT3 antisense ODN delivered via this methodology. EMSAs were performed to confirm STAT3 antisense efficiency and uncoupling of its DNA-binding properties. As shown in Fig. 2,A (upper panel), YT cells were left untreated (lane a); electroporated without ODN (lane b), 5 or 10 μM antisense STAT3 ODN (lanes c and d), or 10 μM mismatched control (lanes e–h); and cultured for 72 h. Cells were then stimulated with medium (−) or IL-2 (+), and nuclear extracts (5 μg) were incubated with a ³²P-labeled DNA probe corresponding to the c-fos promoter SIE m67, to confirm the presence of STAT3 in the complex via supershift analysis with either anti-STAT3 (lane g) or normal rabbit serum (nrs; lane h). STAT3 DNA binding was reduced by 15 and 75% by 5 μM (lane c) or 10 μM (lane d) antisense STAT3 ODN treatments, respectively, as compared with mismatched control (lane e). Specificity of the treatment was confirmed within identical samples incubated either with nrs (lane h) or anti-STAT5 (lane g) in combination with a ³²P-labeled STAT5 oligonucleotide probe (Fig. 2 A, lower panel), which were not affected.

Because STAT3 signaling controls transcription of genes that regulate cell growth and survival (48), we sought to test whether abrogated STAT3 activity interferes with these cellular events. As shown in Fig. 2 B, YT cells were treated with 10 μM STAT3 antisense ODN (upper right panel), mismatched control ODN (lower left panel), or no ODN (upper left panel), and at 72 h posttransfection the number of apoptotic cells was determined by TUNEL assay and quantified (lower right panel), as described in Materials and Methods. Depletion of STAT3 maximally induced apoptosis in 35% of the cells only after 72 h, with minimal effects on the electroporated or mismatched control ODN-treated samples (from two separate experiments). Analysis of STAT3-depleted YT cells at earlier time points (24, 48 h) failed to identify significant cell death (data not shown).

IL-2 activates several signaling cascades responsible for cell proliferation and survival (reviewed in Ref. 49). To test whether exogenous IL-2 can induce signaling pathways that can compensate for the loss of STAT3, YT cells were electroporated with STAT3 antisense ODN or mismatched control ODN (described above), and at 24 h transferred into medium containing 2% FCS in the absence (−) or presence (+) of 1 nM IL-2. Viability was measured by MTS assay at 48, 72, and 96 h postelectroporation. Representative data from three independent experiments are shown in Fig. 3 A. As a result of STAT3 depletion, viability was decreased by 35 and 70% at 72 and 96 h posttransfection, respectively, but the viability of the cells in the presence of physiological amounts of IL-2 was comparable to the control (left panel). Previously, we reported that STAT5 depletion of lymphoid cells can cause >40% apoptosis at 24 h. To determine whether IL-2 can compensate for the loss of STAT5, YT cells were electroporated with 7.5 μM antisense STAT5 or control ODN and immediately transferred into medium containing 2% FSC in the absence (−) or presence (+) of 1 nM IL-2. Viability was measured by MTS assay at 24 and 48 h postelectroporation. In contrast, cell viability was diminished by 40 and 50% by STAT5 depletion at 24 and 48 h, respectively. However, STAT5-depleted cells were not affected by the presence of IL-2 in the culture medium (right panel). These results suggest that an IL-2-induced pathway can rescue the cells from death triggered by loss of constitutively active STAT3, but not that of nonconstitutive STAT5.

Because IL-2-activated STAT5 can drive transcription of genes involved in the regulation of cell growth and survival (27, 50, 51), we next tested whether any correlation exists between the IL-2-mediated survival of STAT3-depleted cells and activation of STAT5. As shown in Fig. 1,C, only STAT3, STAT5a, and STAT5b become tyrosine phosphorylated following stimulation of YT cells with IL-2. Cells obtained from the experiment described previously (Fig. 3,A) (72 h) were lysed and immunoprecipitated for STAT3 and STAT5b (indicated to the right), resolved by SDS-PAGE, and Western blotted with anti-phosphotyrosine, anti-STAT3, anti-phosphotyrosine STAT5, and anti-STAT5 Abs (Fig. 3,B). Antisense to STAT3 significantly reduced tyrosine-phosphorylated and total STAT3 protein levels, as compared with the mismatched control ODN-treated or electroporated samples. In the presence of IL-2 (lanes e–h), STAT5 became tyrosine phosphorylated. In the antisense-treated cells (lanes f and g), pronounced hyperphosphorylation of STAT5b was observed compared with the electroporated (lane f) or mismatched control ODN-treated samples (lane h) without altering basal levels of total STAT5b (lower panel). Similar results were observed with STAT5a (data not shown). These data suggest that addition of IL-2 is able to rescue cells from cellular death most likely via activating STAT5 that can compensate for the diminished survival signals provided by hyperactive STAT3. The level of the antiapoptotic effector protein Bcl2, a known STAT5 target gene, was also assessed (Fig. 3,C). Equivalent amounts of total cell lysate from the above treatment were separated on a 12% SDS-PAGE and blotted with anti-Bcl2 Ab. The membrane was reprobed with anti-GAPDH Ab to confirm equal loading. Representative data (n = 3) showed an up-regulation of Bcl2 protein in STAT3-depleted IL-2-cultured cells (Fig. 3 C, upper panel, lanes f and g), suggesting one possible mechanism by which activation of an IL-2/STAT5 survival pathway compensates for depletion of hyperactive STAT3.

To understand in greater detail the differences in the ability of STAT3 and STAT5 to sustain aberrant cell survival, total RNA from STAT3- or STAT5-depleted cells was isolated and subjected to microarray analysis. Based on the earlier data of cell survival, RNA samples obtained from STAT3 antisense ODN- vs control ODN-treated cells (24 h) as well as STAT5 antisense ODN- vs control ODN-treated cells (12 h) were subjected to reverse transcription using [α-³²P]dATP and hybridized to nylon membrane arrays designed to profile the expression of 96 key genes involved in apoptosis. Genes with changes greater than 2-fold were plotted in Fig. 4 A, as described in Materials and Methods. Overlapping genes included antiapoptotic Bcl-x (down-regulated by both STAT3 and STAT5 depletion 2.6- and 2.48-fold, respectively) and TNF-α (opposing effects: reduced by STAT3 depletion 4.9-fold and increased by STAT5 depletion 2.045-fold). STAT5 depletion decreased mRNA levels of NF-κB signaling pathway intermediates Traf2 (2.38-fold), Traf5 (2.27-fold), and Bcl10 (2-fold), genes that were unaffected by blocking STAT3 protein expression.

To determine the functional differences between STAT3 and STAT5, NF-κB regulatory molecules obtained from this screen were validated. Changes in the mRNA levels of Traf2, Traf5, and Bcl10 upon STAT3 vs STAT5 depletion by quantitative real-time RT-PCR (Q-RT-PCR) were confirmed. RNA samples obtained from STAT3 or STAT5 antisense ODN- vs control ODN-treated cells (24 h) were reverse transcribed, followed by Q-RT-PCR, and then normalized (to 18S rRNA, expressed as percentage values), and the results were plotted (Fig. 4,B). The levels of Traf2 (a), Bcl10 (b), and Traf5 (c) mRNAs were reduced by 50, 50, and 61%, following STAT5 depletion, respectively. To detect changes in Bcl10 protein levels, Western blot was performed on total cell lysates from antisense STAT3-, STAT5-, and control ODN-treated cells (Fig. 4 Bd) at 12 h posttreatment on a 10% SDS-PAGE and blotted with anti-Bcl10 Ab, followed by reprobing the membrane with anti-β-actin to correct for possible loading differences. As shown, Bcl10 protein levels decreased 12% upon STAT3 depletion in contrast to STAT5 depletion, which reduced Bcl10 protein level by 48%.

To assess the effect of STAT5 depletion on NF-κB activity, EMSAs were performed. YT cells were electroporated with STAT5 antisense ODN (lane a–d) or control ODN (lanes e–h), and cultured for 24 h in the absence or presence of 1 nM IL-2. Cells were then stimulated with medium (−) or TNF-α (+) for 30 min at 37°C, and nuclear extracts were incubated with a ³²P-labeled DNA probe corresponding to the consensus binding site of NF-κB/Rel heterodimer (Fig. 5,A, upper panel). To examine possible changes in STAT5 DNA binding, identical samples were incubated with a ³²P-labeled STAT5 oligonucleotide probe (Fig. 5,A, lower panel). Normalized densitometric analysis revealed that STAT5 DNA binding was reduced by 40% following antisense STAT5 ODN treatments (Fig. 5,B), as compared with control ODN-treated samples. Upon STAT5 depletion, both NF-κB constitutive and TNF-α-inducible NF-κB DNA-binding activity were reduced by 40 and 50% (Fig. 5 B), respectively, suggesting that NF-κB activity is regulated by STAT5 expression.

Helenalin, an inhibitor of NF-κB DNA-binding activity by selectively alkylating the p65 subunit of NF-κB, was used to assess the direct role of NF-κB on YT cell viability. In triplicate, YT cells (5 × 10³) were treated with increasing concentrations of Helenalin (0.025–10 μM) or 0.2% DMSO (vehicle control) and cultured for 16 h. Cell viability assessed by MTS assay (Fig. 5 C) showed that Helenalin reduced YT cell viability in a dose-dependent manner, suggesting a correlation exists between YT cell survival and NF-κB activity, which is dependent upon STAT5 expression.

In the present work, we provide evidence that uncoupling hyperactive STAT3 in a lymphoid tumor cell line (YT) induced apoptosis (demonstrated by TUNEL assay) with slower kinetics (72–96 h) compared with STAT5 depletion (24 h). The presence of hyperactive STAT3 was not due to aberrant Jak tyrosine kinase activation or a preassembled intermediate affinity IL-2R complex. Moreover, IL-2 was able to rescue STAT3-depleted cells, which paralleled hyperphosphorylation of STAT5 and elevated Bcl2 protein expression. In contrast, STAT5-depleted cells could not be rescued by the addition of IL-2 (data not shown). We concluded that although both STAT5 and STAT3 can promote cell survival in YT cells, STAT5 remains the dominant factor. To identify the mechanism that accounts for this difference in survival, selective blockade of STAT3 or STAT5 protein expression with antisense ODN was assayed via apoptosis-specific microarrays. Compelling evidence from these studies suggest that STAT5 is important for the expression of Traf2, Traf5, and Bcl10 genes involved in the regulation of NF-κB pathway, which were not identified in STAT3-depleted cells. Indeed, further disruption of NF-κB by Helenalin promoted YT cell death.

STATs have traditionally been categorized into two groups. The first group is comprised of STAT2, STAT4, and STAT6 used by ligands that promote specialized T and B cell functions. The second group containing STAT1, STAT3, STAT5a, and STAT5b displays more diverse functions protecting lymphocytes against apoptosis and driving cell-cycle progression (1, 52) and also detected in several malignancies and virus-transformed cells of the immune system (14, 19, 53, 54), as reviewed in Ref. 21 . Also, enforced expression of a constitutively active disulfide-linked STAT3 homodimer is able to drive tumor formation in nude mice (48). In many instances, STAT5 is also found hyperactivated in a number of malignancies, including HTLV-1 and v-Abl-transformed cell lines, and some patient lymphocytes (2, 8, 9).

STAT3 is constitutively phosphorylated on the highly conserved Y701 and S727 residues in YT cells. IL-2 was able to induce further phosphorylation of this STAT3 population (Fig. 1,A). The fundamental importance of STAT signaling in YT cell survival was underlined by our earlier published results when blockade of several IL-2-inducible pathways in YT cells, including MAPK, PI3K, mammalian target of rapamycin, and A-, B-, and C-Raf failed to affect YT cell growth and viability (32). YT is an acute lymphoblastic lymphoma-derived NK-like tumor cell line in type II latency program of EBV infection (34). At this stage, the cells express LMP-1, the transforming cell surface molecule capable of activating survival-promoting signaling pathways, such as NF-κB and Jak/STAT signaling (35). However, blocking the expression of LMP-1 with antisense ODN did not induce apoptosis of YT cells (55), suggesting that cell proliferation, inhibition of apoptosis, and Bcl2 expression may not be directly regulated by LMP-1 as in EBV-transformed B cells. Aggregated receptor complexes have been shown to drive aberrant cell proliferation and survival (39, 40) that correlate with leukemia formation (41). As shown in Fig. 1 C, however, IL-2R β and γ only form a complex upon IL-2 stimulation, excluding the possibility that aberrant receptor coupling is responsible for STAT3 hyperactivation.

Within this study, none of the Jak family members exhibited constitutive autophosphorylated or activated states (Fig. 1 B); therefore, it seems likely that a Jak-independent pathway is responsible for the hyperactivity of STAT3. The identity of this putative kinase is unknown. In addition to the Jak family members, neither were src family tyrosine kinases hyperactivated within YT cells, including pp60src, Lck, Zap70, Hck, Yes, or Fyn. Indeed, kinase inhibitors, such as PP2 (Fyn), failed to block STAT3 tyrosine phosphorylation (data not shown). Also, tyrosine phosphatases such as SHPTP1, SHPTP2, and PTP1B were tested and not found hyperactivated (data not shown). Whether hyperphosphorylated STAT3 is due to other defective downstream phosphatases or negative regulators of Jak/STAT signaling, such as suppressor of cytokine signaling, is not readily known.

Heterodimer formation upon cytokine stimuli between different STAT family members, such as STAT1 and STAT3, STAT1 and STAT2, and STAT5a and STAT5b, has been shown earlier (45, 46, 47). In YT cells, we failed to identify any other STAT competent to dimerize with STAT3 via coimmunoprecipitation reactions, suggesting that STAT3 only forms homodimers within these cells (Fig. 1 D). Confirmatory evidence by EMSA using the STAT3 SIE probe and supershift Abs to all STAT molecules also revealed only STAT3 within the complex (data not shown).

STAT3 can protect cells against apoptosis. Specific inhibition of STAT3 using dominant-negative constructs or antisense molecules can induce apoptosis in different types of cancer cells (14, 19, 53, 56, 57). The mechanism underlying cell survival supported by STAT3 has been linked to the transcriptional control of apoptotic regulatory genes, such as bcl-x_L, bax (48), and mcl-1 (17), as well as cell cycle regulator p21^waf1 (58). However, in YT cells, successful disruption of STAT3 activity (Fig. 2) resulted in weaker and delayed cellular death (<40% at 72 h) as compared with STAT5 depletion (>40% at 24 h) (30) (measured by TUNEL assays), and STAT3-depleted cells were readily rescued by physiological amounts (1 nM) of IL-2, while STAT5-depleted cells were unresponsive (Fig. 3,A). Intriguingly, blocking STAT3 expression had no effect on YT cell cycle progression (data not shown) or viability of the human IL-2-dependent T cell leukemia line Kit225 that do not exhibit overactive STAT3 (data not shown). The reduction of protein levels and DNA-binding activity of either STAT3 or STAT5 was comparable following antisense ODN treatments (∼40% at 24 h; data not shown and Fig. 5 A, respectively).

To understand the apparent differences in survival mechanisms exhibited by STAT3 and STAT5, RNA samples isolated from STAT3- and STAT5-depleted cells, before cell death, were assayed by microarrays designed to profile 96 apoptosis-related genes. Based on the normalized results (for details, see Materials and Methods), Bcl-x, which is a known target of both STAT3 and STAT5 (28, 29, 48), was diminished by depleting both proteins (2.6- and 2.48-fold, respectively), while Bcl2 expression presumably controlled by STAT5 (9) was down-regulated 4-fold. To determine the mechanism of greater sensitivity to STAT5 depletion, genes differentially expressed were studied. Indeed, upstream regulators of NF-κB, including Traf2, Traf5, and Bcl10, were reduced only in STAT5-, but not STAT3-depleted cells. Interestingly, NF-κB DNA binding is constitutively active in YT cells, which can be further activated following TNF-α stimulation (Fig. 5 A) and may cooperate to protect against apoptosis (59). Bcl10 (reduced 2-fold on microarray) is an adapter molecule implicated in Ag receptor-mediated NF-κB signaling by linking to the IκB kinase complex. Mice deficient in Bcl10 expression exhibit impaired B/TCR signaling and nonfunctional B/T cells, as a consequence of impaired NF-κB signaling (60, 61). Traf are adapter molecules mediating TNFR signaling. Effective Traf signaling occurs via activation of NF-κB. The highly related Traf2 and Traf5 molecules were down-regulated 2.8- and 2.2-fold, respectively, following STAT5, but not STAT3 depletion. It is known that Traf2 deficiency causes premature death of otherwise normally developed newborn mice, with thymus and spleen atrophy and hemopoietic progenitors being highly sensitive to TNF-α-induced cell death (62). Furthermore, Traf2-depleted Jurkat cells were unable to produce IL-2 upon CD3 and CD28 ligation, due to the lack of NF-κB activation (63). Similarly, costimulatory signaling, such as response to CD40 and CD27 ligation in T and B cells, respectively, derived from Traf5-deficient animals, was shown to be impaired (64). Traf2 and Traf5 double-knockout mice also exhibit severe reduction in TNF-α-mediated NF-κB activation (65). Finally, another group has recently showed that blocking Traf2 and Traf5 simultaneously leads to the inactivation of NF-κB (66). Q-RT-PCR confirmed the microarray results for Bcl10, Traf2, and Traf5, showing reduction in the transcript levels of ∼50–60% upon STAT5 depletion, whereas STAT3 depletion failed to exhibit significant changes. Similar findings were observed for Bcl10 protein expression following disruption of STAT5, but not STAT3. From these studies, we concluded that these molecules provide key regulatory signals mediated via NF-κB that is important for lymphoid cell survival.

To better understand how STAT5 depletion might affect NF-κB activity at the level of DNA binding, EMSA analysis was performed in YT cells. Indeed, STAT5-depleted cells showed reduced constitutive (40% loss) and TNF-α-inducible (50%) NF-κB DNA-binding activity, similar to reduced STAT5 DNA-binding levels (40%) (Fig. 5, A and B). Concomitantly, STAT3 antisense treatment did not alter NF-κB DNA-binding activity (data not shown). These results suggest that NF-κB DNA binding is dependent on STAT5 expression. Because Bcl10, Traf2, and Traf5 act as upstream activators of NF-κB and regulate the activity of the IκB kinase complex (61, 65), one possible model to explain these findings is that STAT5 depletion reduces the expression of these effectors and consequent NF-κB activation that is important for cell survival. Indeed, treatment of YT cells with Helenalin that selectively alkylates the p65 subunit of NF-κB and interferes with its DNA-binding activity, but not STAT5 activity (67), greatly reduced YT cell viability in a dose-dependent fashion. These data suggest that STAT5 and NF-κB may act in concert to maintain YT cell survival.

In conclusion, these results suggest that within this lymphoma, and perhaps other tumors, STAT5 can be a preferential regulator of cell survival as compared with other STATs. Our data suggest that STAT5 exerts its critical survival function via dictating the expression of upstream activators of NF-κB, which include Bcl10, Traf2, and Traf5 that are weakly regulated by STAT3. These data also put forth the notion that nonconstitutively active STATs may serve as therapeutic targets for controlling certain types of cancers.

The Kit225 cells were a gift from Dr. James Johnston (Queens University, Belfast, U.K.). We thank Dr. Gregory Shipley and the Quantitative Genomic Core Laboratory at University of Texas Health Science Center for generating the Q-RT-PCR results.

The authors have no financial conflict of interest.