Introduction

Signal transducer and activator of transcription (STAT) proteins are transcription factors crucial in the transduction of cytokines and growth factors signaling. Common sets of STAT factors can be activated by different stimuli, thus contributing to finely tuning the response of cells to cytokine treatment according to the different steps of differentiation or the physiological or pathological context.1 Cytokine activity is negatively controlled at different levels2 and, in particular, STAT1 and STAT3 can reciprocally regulate each other, thus contributing to maintain the specificity of cytokine signaling. For example, in STAT3−/− mouse embryonic fibroblasts, interleukin (IL)-6 can induce prolonged activation of STAT1, thus mediating interferon (IFN)-γ-like responses.3 On the other hand, IFN-γ treatment of STAT1−/− mouse embryonic fibroblasts leads to prolonged activation of STAT3 instead of STAT1.4 In STAT1−/− T cells, type I IFNs favor cell proliferation and survival rather than exerting its typical anti-proliferative effect.5, 6 The alternative pathways activated in these cells have not been fully characterized yet, to understand whether the unbalanced expression of STAT1 may favor STAT3 activation that in turn positively regulates cell proliferation5 or whether the proliferative effect is mediated by the activation of other pathways.6 Thus, an unbalanced expression of STAT1 or STAT3 may result in altered cell responses to cytokines such as IL-6 or IFNs. This is of particular interest as STAT1 and STAT3 exert opposite effects on the control of cell proliferation, survival and apoptosis, the former being antiproliferative and proapoptotic, whereas the latter favoring cell proliferation and survival.7, 8

Interferon-γ is critical in the control of cell growth and apoptosis9, 10 and the binding to its specific receptor complex IFN-γR leads to the activation of STAT1, driving the expression of many IFN-inducible genes. Unfortunately, tumor cells can become unresponsive to IFN-γ by deregulating the expression of components of its transducing machinery.11, 12 In particular, in human neoplastic T cells, IFN-γ unresponsiveness is often due to constitutive internalization of the transducing IFN-γR2 chain.13 In an attempt to bypass this unresponsiveness, we thus decided to analyze the consequences of abolishing STAT3 expression in human neoplastic T cells in terms of re-activation of the STAT1-mediated apoptotic pathway. We thus stably knocked down STAT3 in two human neoplastic T cell lines, ST4 and Molt4, and investigated the effects of both IFN-γ and IL-6 treatment. We show that, in the absence of STAT3, STAT1 activation by IL-6 is strongly enhanced and cells acquire an IL-6-dependent apoptotic response.

Materials and methods

Media and malignant cells

All cell lines were cultured in RPMI 1640 (BioWhittaker Inc., Walkersville, MD, USA) supplemented with gentamycin (Schering-Plough, Milan, Italy) and 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA) (complete medium). During experiments, cells were cultured in RPMI containing 2% FBS. All in vitro cultures were maintained at 37 °C in a 5% CO2 humidified atmosphere. Human ST4 cells were derived from a childhood convoluted-type T-cell lymphoma,14 whereas Molt4 (ATCC, CRL1582, Manassas, VA, USA) is a human T cell acute lymphoblastic leukemia.15

shRNA sequences, lentiviral preparation and cell infections

See Supplementary Materials and Methods.

Flow cytometry

To evaluate STAT1 activation, T-cell lines were cultured for 24 h in the presence or absence of either IFN-γ (kindly provided by Dr M Brunda, Hoffman-La Roche, Nutley, NJ, USA) or IL-6 and its soluble receptor (sIL-6R) (R&D Systems, Minneapolis, MN, USA). In all the experiments, IFN-γ was used at 1000 U/ml, and IL-6 and sIL-6R at 200 ng/ml. Intracellular staining of phospho-STAT1 (pY701) was carried out as previously described.16 To assess surface expression of major histocompatibility complex class I (MHC I) molecules, cells were cultured in the presence or the absence of cytokines for 24 h, and then stained and analyzed as previously described.17 To evaluate apoptosis, cells were cultured in the presence of cytokines or left untreated for 120 h and then stained, at different times, with phycoerythrin-conjugated Annexin-V kit (BD Biosciences Pharmingen, San Diego, CA, USA). All experiments were carried out with a FACSCalibur flow cytometer (Becton Dickinson, San Diego, CA, USA). Each plot represents the results from 10 000 events, except for the apoptosis experiments where we analyzed 20 000 events.

Western blot analysis

Western blot experiments were carried out as described elsewhere,18 and the following primary antibodies were used for the analysis: rabbit anti-STAT1, rabbit anti-STAT3, rabbit anti-phospho-Tyr(701)-STAT1 and rabbit anti-cleaved caspase-3 (Asp 175) from Cell Signaling (Beverly, MA, USA); goat anti-caspase-3 p20 (N-19) from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA); rabbit anti-actin from Sigma-Aldrich (St Louis, MO, USA). When indicated, 2 ng/ml IFN-α (Peprotech Inc., Rock Hill, NJ, USA) were used as a positive control.

In vivo experiments

Five-week-old female severe combined immunodeficient (SCID) (CB17 scid/scid) mice (Charles River Laboratories, Calco, Italy) were fed and maintained under specific pathogen-free conditions in the animal facility of the Department of Clinical and Biological Sciences, University of Turin, and treated in accordance with the European guidelines. ST4-C and ST4-S1 (10 × 106 per mouse) were washed twice with phosphate-buffered saline (PBS), re-suspended in 0.4 ml of PBS containing 0.01% murine serum albumin (Sigma-Aldrich) and injected subcutaneously in the inguinal region. At 24 h before tumor challenge, and then weekly, all mice received intraperitoneal injections with 0.2 ml of a 1/20 dilution of anti-asialo GM1 rabbit antiserum (Wako Chemicals GmbH, Dusseldorf, Germany) in PBS. At 2 h after the inoculation of cells, mice were injected in the same site with 0.4 ml of PBS/murine serum albumin alone (vehicle), 1000 U of IFN-γ or 200 ng of IL-6/sIL-6R. This treatment was repeated daily for 10 days. In all the experiments tumor incidence and growth were evaluated twice a week by a blinded observer. Neoplastic masses were measured with calipers along the two perpendicular diameters for 41 days. At the end of this period, tumor-free mice were classified as survivors. Latency time was considered as the period (in days) between challenge and the growth of neoplastic masses to a mean diameter of 3 mm. Mice were killed for human reasons when the tumor exceeded 12-mm mean diameter.

Statistical analysis

Pearson's t-test (GraphPad Prism 3, GraphPad Software, Inc., San Diego, CA, USA) was used to analyze the effect of IFN-γ or IL-6/sIL-6R on the in vivo growth of ST4 cells with P<0.05 as the significance cutoff.

Results

STAT1 activation in response to IL-6 or IFN-γ in the absence of STAT3

To investigate the effect of STAT3 knockdown on the cytokine responsiveness of human neoplastic T cells, ST4-WT and Molt4-WT cells were infected with lentiviral vectors delivering short hairpin RNAs (shRNAs) targeting STAT3 to generate the STAT3-silenced cell lines, ST4-S1, Molt4-S1 and Molt4-S2. As a control, the cell lines ST4-C and Molt4-C were generated using a scrambled shRNA (see Supplementary Materials and Methods and Supplementary Figure 1).

ST4 cells were treated with IFN-γ or with IL-6 in combination with sIL-6R, and the kinetics of STAT1 activation was evaluated by western blot analysis. In all experiments, a 15-min treatment with IFN-α was used as a positive control for STAT1 activation. IL-6/sIL-6R treatment in both ST4-WT and ST4-C cells was only able to trigger a weak and very transient STAT1 activation. In contrast, STAT1 phosphorylation by IL-6 was enhanced and prolonged in ST4-S1 cells, peaking at 4 h from stimulation (Figure 1a, right panels). As expected, IFN-γ induced STAT1 activation already at 15 min, which slightly decreased after 4 h in ST4-WT and ST4-C cells. STAT1 activation kinetics was slightly prolonged in ST4-S1 cells (Figure 1a, left panels). STAT1 phosphorylation was then assessed after longer treatments. After 48 h, IFN-α-induced STAT1 phosphorylation was still high in all cell lines, although much weaker on IFN-γ treatment. Remarkably, in ST4-S1 cells, IL-6 was able to elicit STAT1 phosphorylation levels as high as those induced by IFN-α treatment, whereas no activation was detected in ST4-WT or ST4-C cells (Figure 1b). Sustained STAT1 activation in ST4-S1 cells cannot be mediated by increased IFN-γR2 expression, as all cell lines displayed equally low levels of IFN-γR2 surface expression (data not shown). Long-term STAT1 activation in response to cytokine treatment was also assessed in the Molt4 cell lines. Similar to what observed in ST4-S1 cells, IL-6 acquired the ability to trigger STAT1 activation specifically in the STAT3-silenced Molt4-S1 and Molt4-S2 cells, whereas IFN-γ-dependent STAT1 phosphorylation was similar in all Molt4 cell lines (Figure 1c). In these experiments, the more sensitive fluorescence-activated cell sorting-based analysis of STAT1 phosphorylation was used because of the low STAT1 levels in Molt4 cells.19

Figure 1
figure 1

Interferon (IFN)-γ or interleukin (IL)-6/sIL-6R induce prolonged STAT1 activation in cells with reduced STAT3 expression. (a) After 1 h-serum starvation, ST4-WT, ST4-C and ST4-S1 cells were stimulated with IFN-γ or IL-6/sIL-6R for 4 h. At different times, cells were recovered, total proteins were extracted and STAT1 activation was evaluated by western blot. Membranes were subsequently probed with an anti-STAT1 antibody to confirm equal protein loading in each lane of the gel. As a positive control, cells were stimulated for 15 min with IFN-α. (b) STAT1 phosphorylation was also evaluated in ST4-WT, ST4-C and ST4-S1 cells after 48 h from cytokine stimulation. At this time, STAT1 activation caused an increase in total STAT1 expression, so the membranes were also probed with anti-actin antibody as loading control. (c) Molt4-WT, Molt4-S1 and Molt4-S2 cells were starved for 1 h and then treated, or not, for 24 h with IFN-γ or IL-6/sIL-6R. The intracellular staining of STAT1 phosphorylation was detected by fluorescence-activated cell sorting analysis. Gray histograms represent non-specific fluorescence detected with isotype-matched control Ig. The mean fluorescence intensity (MFI) values indicate the mean fluorescence intensity of positive cells. All experiments were carried out independently at least three times.

Full size image

IL-6 induces MHC I expression in cells lacking STAT3

To assess whether the active STAT1 detected in response to IL-6 in STAT3-silenced cells is functional, we compared the effects of IL-6 and IFN-γ on the expression of MHC class I antigens, a downstream target of STAT1 activation in response to IFNs.20 All ST4 and Molt4 cell lines were cultured with or without recombinant cytokines for 24 h and then stained for MHC I expression and analyzed by fluorescence-activated cell sorting. Compared with the constitutive levels of untreated cells (Figure 2, gray histograms), IFN-γ induced equivalent MHC I upregulation in all ST4 and Molt4 cell lines as already reported21 (Figure 2, thin lines). As expected, IL-6/sIL-6R treatment could not induce MHC I upregulation in STAT3 expressing cells (Figure 2, bold line), but it acquired the ability to trigger MHC I induction equivalent to that triggered by IFN-γ in all STAT3-silenced cell lines.

Figure 2
figure 2

Interleukin (IL)-6 induces major histocompatibility complex class I (MHC I) expression in ST4-S and Molt4-S. All cell lines were treated with interferon-γ (thin line) or IL-6/sIL-6R (bold line) for 24 h. The expression of MHC I molecules was evaluated by FACS analysis. Gray histograms represent the basal surface expression of MHC I in each cell line. The results are from one representative experiment out of the three experiments carried out independently.

Full size image

IL-6 induces apoptosis in STAT3-depleted cells

To assess whether the enhanced STAT1 activation observed in ST4-S1 cells (Figures 1a and b) had an effect also on cytokine-driven apoptosis, ST4-WT, ST4-C and ST4-S1 cells were grown for 120 h in the presence or absence of IFN-γ or IL-6. Cells were collected at different time points and stained with Annexin-V-phycoerythrin to evaluate apoptotic cell death by fluorescence-activated cell sorting analysis. All cell lines showed a spontaneous rate of apoptosis, increasing at 120 h, particularly in the STAT3-silenced ST4-S1 cells. As expected, apoptosis rates were not increased by either cytokine treatment in ST4-WT and ST4-C cells (Figure 3a). In contrast, apoptosis was progressively enhanced, particularly after 120 h of treatment, by IFN-γ and, to a greater extent, by IL-6 (64 and 83% of Annexin-V positivity, respectively) (Figure 3a). These findings correlated well with the levels of cleaved caspase 3 (representing its activated form), a typical feature of apoptotic cell death (Figure 3b). Indeed, only low levels of cleaved caspase 3 were detected in either cytokine-treated or -untreated ST4-WT or ST4-C cells. In contrast, ST4-S1 cells already displayed increased basal levels of cleaved caspase 3, further enhanced by IL-6 and, to a lesser extent, by IFN-γ treatment, and correlating with a decrease in the amount of uncleaved protein.

Figure 3
figure 3

Cytokine-mediated apoptosis in ST4-S1 cells. (a) ST4-WT, ST4-C and ST4-S1 cells were treated with interferon (IFN)-γ (•), interleukin (IL)-6/sIL-6R (▾) or left untreated (▪) until 120 h. At 24, 48, 72 and 120 h, cell aliquots were recovered and stained with Annexin-V-phycoerythrin to detect apoptotic cells by FACS analysis. The percentages of Annexin-V-positive cells are shown as means ± s.e.m. from two independent experiments at each time point. (b) After 48 h of cytokine treatment, total proteins were extracted from cells and the presence of cleaved and uncleaved forms of caspase 3 was detected by western blot. Actin was used as a control for equal protein loading.

Full size image

IL-6, but not IFN-γ, can inhibit in vivo growth of human malignant T cells devoid of STAT3

Next, we decided to test the abilities of IL-6 and IFN-γ to block ST4-S1 cells growth in vivo. SCID mice were subcutaneously inoculated with 10 × 106 ST4-C or ST4-S1 cells, treated for 10 days with PBS/murine serum albumin (control group, n=4 for ST4-C and n=5 for ST4-S1), 1000 U IFN-γ (n=4 for ST4-C and n=6 for ST4-S1) or 200 ng IL-6/sIL-6R (n=4). Tumor growth (evaluated as mean tumor diameter) was monitored. ST4-C cells grew in all groups with the same kinetics, with palpable tumors becoming detectable 35 days after the inoculum and rapidly growing thereafter. In all, 100% of IFN-γ-treated or -untreated mice inoculated with ST4-C cells developed tumors of 10-mm diameter within 50 days of the inoculum, whereas one out of four mice treated with IL-6/sIL-6R was still tumor free after 42 days (Figure 4a). ST4-S1 cells grew more rapidly than ST4-C, giving rise to palpable tumors after 20 days in 100% of both control and IFN-γ-treated groups of mice. In contrast, IL-6 treatment could significantly delay and even inhibit tumor growth. Indeed, three out of four mice inoculated with ST4-S1 cells and treated with IL-6 were still tumor-free after 41 days, and two out of four never developed tumors up to 70 days after the inoculum, when they were killed. In addition, tumor growth could be detected only at day 28 or 52 after the inoculum in the two out of four mice that did develop tumors (Figure 4b).

Figure 4
figure 4

Interleukin (IL)-6 can inhibit in vivo growth of ST4-S1 cells. Severe combined immunodeficient mice were inoculated subcutaneously with 10 × 106 ST4-C (a) or ST4-S1 cells (b). After 2 h, mice were injected in the same site with phosphate-buffered saline/murine serum albumin alone (•), interferon-γ (○) or IL-6/sIL-6R (▪), and the treatment was repeated daily for 10 days. Results are shown as diameter of the single tumors (left panels) and as percentage of tumor-free mice (right panels).

Full size image

Discussion

It is well known that STAT1 activation is fundamental for IFN-γ-mediated anti-proliferative and/or pro-apoptotic responses in different cell types. 22 In human neoplastic T cells, this pathway is impaired because of the downregulated expression of the IFN-γR2 signaling chain.13 Starting from the observation that in STAT3−/− mouse embryonic fibroblasts IL-6 acquires the completely new ability to mediate an IFN-γ-like response due to prolonged STAT1 activation,3 we decided to assess whether a similar phenomenon may also occur in human neoplastic T lymphocytes, leading to bypass their block in IFN-γ signaling. Our data show that, indeed, the responses to IL-6 can be driven towards STAT1 activation and STAT1-mediated functions by interfering with STAT3 in human neoplastic T lymphocyte cell lines. The sustained STAT1 activation triggered by IL-6 in STAT3-depleted cell lines is functional because it could induce the expression of MHC I molecules. This was true in both cell lines tested and with two different shRNA sequences, supporting the idea that the phenomenon could be of general validity for human neoplastic T lymphocytes and not due to off-target effects of shRNA sequences. To further confirm that the phenotype observed was only due to loss of STAT3, many attempts were made to re-instate the expression of a shRNA-resistant form of STAT3 into the silenced cells, but with no success, probably because of the extremely inefficient infection rates achieved in these cells (data not shown). It is not surprising that the induction of MHC I by IL-6 in Molt4-S1 and Molt4-S2 was weaker than that elicited by IFN-γ, as the degree of STAT3 silencing obtained in these cells was lower than in the ST4-S1 cells. Likely, in T lymphocytes even low levels of STAT3 activity are still sufficient to restrain the onset of a STAT1-mediated IFN-γ-like response to IL-6. Despite the prolonged STAT1 activation elicited by IFN-γ in ST4-S1 cells, MHC I molecules induction was unchanged, suggesting that low levels of active STAT1 are sufficient to achieve maximal induction. This observation fits with the data showing that IFN-γ optimally induces the upregulation of MHC I in human malignant T cells cultured in the presence of serum,23 an experimental condition that promotes IFN-γR2 internalization and weak and transient STAT1 activation by IFN-γ.18, 21, 24

Importantly, IL-6 was able to induce apoptosis in ST4-S1 cells more efficiently than IFN-γ, correlating with the growth inhibition activity observed in vivo. Indeed, although IFN-γ did not affect tumor growth, IL-6 treatment could delay the growth of ST4-S1 cells in SCID mice and in some cases even completely prevent it. This may reflect the intensity of STAT1 activation observed in vitro after 48 h, much stronger in response to IL-6 than in response to IFN-γ and reaching levels similar to those triggered by the positive-control IFN-α. Accordingly, in vitro IFN-α-induced apoptotic response was similar to that triggered by IL-6 (data not shown). IFN-γ-dependent apoptosis is known to be strictly linked to the levels of IFN-γR2 expression.15, 18, 24 In ST4-S1 cells, IFN-γR2 expression levels were as low as those of the ST4-WT and ST4-C control cells (data not shown), correlating with the limited apoptotic response induced by IFN-γ. This limitation is likely bypassed by IL-6 treatment, which triggers STAT1 activation through the IL-6R and not the defective IFN-γR system. In addition, other signals besides STAT1 activation might be required to elicit efficient apoptosis, particularly in vivo. These may be delivered by the IL-6R system but not by the impaired IFN-γR system.

One could argue that the ex novo IL-6-dependent STAT1 activation in the absence of STAT3 might be due to altered activation of some of the known negative regulators of the Jak/STAT pathway. Among these, we analyzed SHP1, SOCS1 and SOC3 levels, and did not observe any significant change in their expression in STAT3-depleted cells, suggesting that they are not involved in this phenomenon (data not shown). In conclusion, the advantage afforded by IL-6 treatment is the ability to reinstate a full IFN-γ pro-apoptotic response through STAT1 activation, bypassing the defective IFN-γR system.

The rate of in vivo growth of ST4-S1 was greater than that of ST4-C cells, correlating with the enhanced in vitro proliferation (not shown). ST4 cells do not show a constitutive activity of STAT3, and these data show that the absence of STAT3 does not interfere with their normal growth, differently from other tumor cell lines where STAT3 is constitutively activated and required for survival and proliferation.25, 26 In addition, the absence of STAT3 may unmask the activity of other signaling pathway involved in controlling proliferation. The greater rate of basal apoptosis observed in ST4-S1 cells compared with control cells may also be a consequence of this enhanced proliferation. In contrast to IL-6, IFN-γ responses were mostly unaffected on STAT3 silencing, confirming that STAT3 has a limited function in the IFN-γ response.

STAT3 is considered an oncogene and many laboratories are actively seeking suitable means to interfere, either genetically or pharmacologically, with its activity as an anti-cancer therapy. Our data show that the abolishment of STAT3 expression could be a suitable way to block neoplastic cell growth also in those tumors whose survival and proliferation are not linked to STAT3 activation, thus extending the spectrum of tumors that could be targeted by these strategies to T-cell tumors. Further experiments will be required to rule out potentially negative consequences of an unbalanced STAT1/STAT3 expression in patients affected by neoplastic cell diseases.

Conflict of interest

The authors declare no conflict of interest.