Regulation of IL-6 Signaling by p53: STAT3- and STAT5-Masking in p53-Val¹³⁵-Containing Human Hepatoma Hep3B Cell Lines¹

Cytokines such as the IFNs, various ILs, and hemopoietic growth factors are currently in widespread therapeutic use against various forms of human cancer (1, 2). Many cytokines such as IL-6 are almost invariably present at the host-tumor interface (1). A common feature of all these cytokines and growth factors is that they engage cell surface receptors that signal to the cell nucleus via the JAK³-STAT signaling pathway (3, 4, 5, 6, 7, 8). Thus, alterations in the regulation of JAK-STAT signaling in cancer cells could determine the responsiveness of such cells to therapeutic cytokines. In this article we report studies that explore a novel indirect cellular process, termed STAT-masking, by which the p53 status of hepatoma cells modulates the response of these cells to cytokines such as IL-6.

Despite clear evidence of the influence of cytokines upon various p53-induced cellular processes, for example the rescue of p53-induced apoptosis in myeloid M1 cells by IL-6 (9, 10, 11), there is little or no information concerning the influence, direct or indirect, of the transcription factor p53 upon cytokine-elicited cellular signaling through the JAK-STAT pathway. While mutations in p53 are among the commonest alterations observed in human cancer (12, 13, 14, 15, 16), many human cancers are characterized by no or only rare mutations in p53 (17, 18, 19, 20, 21, 22, 23). For example, B cell neoplasia such as myelomas rarely display mutations in p53, with the frequency of mutations in p53 rising only in advanced cases no longer responsive to therapy (17, 18, 19, 20, 21, 22). Furthermore, more than half of all human cancers are characterized by an increase in the level of p53 expression, and in many instances this increased expression is that of the normal wild-type (wt) p53 allele (12, 24, 25, 26, 27). Indeed, increased levels of wt p53 have now been found in subsets of human cancers, including neuroblastomas, mesotheliomas, and breast, colon, and pancreatic cancer (24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39). Today, p53 is considered to be a promising target tumor Ag against which to mount an active specific anti-cancer immune response as a therapeutic modality (12, 28, 29, 30, 31, 32, 33). However, we have observed recently that the use of different active immunization regimens against human cancers resulted in the development of long-lived high levels of circulating IL-6 in the cancer patient, an outcome that may well prove to be a general feature of successful active specific anti-cancer immunization regimens (40). Thus, we asked whether the p53 status of tumor cells could affect signal transduction pathways triggered by IL-6.

To investigate the effects of p53 on the signal transduction pathways engaged by IL-6, as distinct from the previously well-studied modulation by p53 of cell proliferation, apoptosis, or rescue from DNA damage, we used the p53-free human hepatoma Hep3B cell line to derive 11 cell lines that constitutively express the temperature-sensitive (ts) p53-Val¹³⁵ mutant of p53 (41). These p53-Val¹³⁵-expressing hepatoma Hep3B cell lines provided a unique opportunity to investigate the ability of p53 to modulate the rapid response of liver cells to cytokines; namely, the induction of acute phase plasma protein secretion by IL-6 as a function of incubation temperature (p53-Val¹³⁵ has a mutant conformation at 37°C and a wt-like conformation at 32.5°C (see ref. 42 for a discussion of the gain-in-function phenotype of p53-Val¹³⁵ at 32.5°C as contrasted with the wt phenotype per se). We observed a reduced ability of these p53-Val¹³⁵-containing Hep3B cell lines to secrete fibrinogen and α₁-antichymotrypsin in response to IL-6 at the wt p53 temperature but not at the mutant p53 temperature (41). In investigating the possible biochemical basis for this reduced responsiveness and in a departure from previous studies that had reported the cytoplasm to nuclear translocation of STAT3 and STAT5 immunofluorescence upon treatment of human hepatoma cells with cytokines (43, 44), we discovered that the exposure to IL-6 of p53-Val¹³⁵-containing hepatoma cells previously incubated at the wt p53 temperature led to a marked loss of STAT3 and STAT5 cytoplasmic and nuclear immunofluorescence without a commensurate degradation of the respective STAT proteins (STAT-masking) (45).

In this article we report a characterization of IL-6-induced STAT-masking in p53-Val¹³⁵-containing Hep3B cells at the cellular level, define the biochemical basis of its regulation, and provide evidence of its functional consequences. The data are consistent with the hypothesis that a p53-Val¹³⁵-dependent gene product(s) that accumulates in Hep3B cells at 32.5°C modulates the immunologic accessibility to and function of STAT3 and STAT5 proteins. Thus, STAT-masking appears to be a new cellular process by which p53 can indirectly regulate the response of cells to cytokines.

A series of 11 stably transfected cell lines constitutively expressing p53-Val¹³⁵ and seven control cell lines expressing pSVneo alone were derived from the p53-free human hepatoma Hep3B line as has been described previously (41). Using immunofluorescence assays, all 11 p53-Val¹³⁵-containing Hep3B cell lines were confirmed to display a ts p53 conformation (mutant at 37°C and wt-like at 32.5°C) and to display IL-6-induced STAT-masking, but none of the seven pSVneo-only lines or the parental Hep3B cells displayed the STAT-masking phenotype (45). Cell lines 1 and 5, both of which express p53-Val¹³⁵, and the parental Hep3B cell line were used in the present experiments. The level of cellular p53 expression in lines 1 and 5 cell lines was comparable to that observed in T47D human breast carcinoma cells as assayed using a pan-specific anti-p53 mAb in immunofluorescence studies.

These were conducted in eight-well chamber slides (Nunc, Naperville, IL; 2 × 10⁴ cells/well) as described previously (45). Immunostaining using various monoclonal and polyclonal Abs was conducted according to the instructions provided by the respective Ab suppliers using goat serum as the blocking agent and fluorescein-tagged goat anti-murine IgG or rhodamine-tagged goat anti-rabbit IgG (Cappel Organon Teknika, West Chester, PA) as the second Ab. Cellular immunofluorescence was evaluated using a Bio-Rad MRC 1000 dual laser confocal microscopy system (Bio-Rad, Richmond, CA). All data within one experiment (controls and all experimental groups) were collected at the same laser intensity, black level, and gain settings. In control experiments, the secondary Abs by themselves did not reveal any immunofluorescence (data not shown). Murine mAbs to STAT1, STAT3, and STAT5a (marketed as anti-STAT5) were purchased from Transduction Laboratories (Lexington, KY). Murine mAb to STAT5b and rabbit polyclonal Ab (pAb) to STAT3 (C-20) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The rabbit polyclonal anti-PY-STAT3 was purchased from New England Biolabs (Beverley, MA). The anti-STAT3 mAb from Transduction Laboratories was derived using the amino-terminal residues 1 to 175 in STAT3 as immunogen, while the anti-STAT3 pAb (C20) from Santa Cruz Biotechnology was raised to a peptide corresponding to the carboxyl-terminal residues 750 to 769 in the STAT3 protein. The anti-PY-STAT3 pAb was raised to a synthetic phospho-Tyr⁷⁰⁵ peptide corresponding to residues 701 to 709 of mouse STAT3.

Hepatoma cells cultured in 100-mm plastic petri dishes were harvested and fractionated into cytoplasmic and nuclear fractions using hypotonic swelling (1 ml/cell pellet derived from one 100-mm culture) and Dounce homogenization essentially as described previously (46, 47, 48). The cytoplasmic fraction was clarified by centrifugation at 15,000 × g for 15 min. The nuclear pellet was washed with hypotonic buffer containing 0.2% Nonidet P-40 to remove cytoplasmic contamination. In some experiments a whole cell lysis buffer containing 1% Triton X-100 and 0.5% Nonidet P-40 (according to the protocol from Transduction Laboratories) was used. All buffers contained protease inhibitors and orthovanadate (100 μM) as described previously (45, 48).

Immunoprecipitation of STAT3 proteins was conducted using rabbit anti-STAT3 pAb C20 or rabbit anti-PY-STAT3 pAb in immunoprecipitation buffer that included 0.5% Triton X-100 and 0.1% SDS essentially as previously described (49).

Western blot procedures used were according to the protocol provided by Transduction Laboratories and the ECL detection kit (Amersham International, Aylesbury, U.K.). When comparing whole cell extracts, aliquots containing equal total protein amounts in the range 30 to 40 μg (Bio-Rad Micro Assay) were loaded in each lane of particular Western blots. In analyzing immunoprecipitates, the entire sample was Western blotted.

The chloramphenicol acetyltransferase (CAT) reporter construct designated pβFibCAT (41), which contains two copies of the 36-bp IL-6 response element from the rat β-fibrinogen promoter (from −168 to −134), was a gift from Dr. Heinz Baumann (50). Transient transfections of pβFibCAT (10 μg/culture) using the calcium phosphate method into 100-mm cultures of Hep3B and Hep3B-derived cell lines together with the plasmid pRSVβgal (5 μg/culture) as a marker for transfection efficiency and assays for IL-6 responsiveness of the CAT reporter construct were conducted, each in duplicate, as described previously (41, 51). β-Galactosidase activity in the cell extracts was used as the basis for normalization of CAT assay data.

Nuclear extracts for assays of DNA binding activity were prepared essentially as described previously (45, 48), and STAT-specific DNA binding activity was assayed using a IFN-γ-activated site, double-stranded DNA element from the IFN response factor-1 promoter (top strand, 5′-gatcGATTTCCCCGAAATcgagatc-3′; lowercase letters are linkers) that yields the typical pattern of A, B, and C complexes in gel-shift assays corresponding to STAT3 homodimer, STAT1/3 heterodimer, and STAT1 homodimer, respectively (5, 45, 48).

Human IL-6 (Escherichia coli derived) was a gift from Sandoz (East Hanover, NJ). The inhibitor PD98059 was purchased from BioMol Research Laboratories (Plymouth Meeting, PA); U-73122 and U-73342 were obtained from Calbiochem-Novabiochem (La Jolla, CA), and D609, okadaic acid, and calyculin A were purchased from Alexis (San Diego, CA). Rabbit anti-peptide Abs against phospholipase C (PLC) types β₁, β₂, β₃, β₄, γ₁, γ₂, and δ₂ for Western blotting were purchased from Santa Cruz Biotechnology.

Student’s t test (two-tailed) was used for statistical evaluation.

We have previously observed that compared with the parental p53-free Hep3B cells, which responded equally well to IL-6 at both 37 and 32.5°C in terms of the cytokine-elicited increase in the synthesis and secretion of Bβ-fibrinogen and α₁-antichymotrypsin, the p53-Val¹³⁵-containing cell lines had a reduced response to IL-6 at 32.5°C (41). Table I extends these previous observations to the level of the IL-6 response of the β-fibrinogen reporter construct (pβFibCAT) in transient transfection assays. The IL-6 inducibility of a reporter construct containing two copies of the 36-bp IL-6-responsive element from the β-fibrinogen promoter is significantly reduced in cells first transfected at 37°C and then shifted to 32.5°C for 1 day provided that the cells expressed p53-Val¹³⁵ (Line 1 in Table I is an example). Consistent with our previous data (41), the parental Hep3B cells did not display a significant ts phenotype with respect to the inducibility of the β-fibrinogen promoter construct. Because the 36-bp enhancer DNA element in pβFibCAT contains the IL-6-responsive STAT binding motifs, the data in Table I suggested the possibility that the reduced responsiveness of p53-Val¹³⁵-containing cells at the wt p53 temperature may be the result of a modulation of IL-6-induced JAK-STAT cell signaling. In investigating possible alterations in cytoplasm to nuclear translocation of IL-6-activated transcription factors in these cells at the two temperatures, we observed that IL-6-induced a marked loss of both cytoplasmic and nuclear immunofluorescence of STAT3 and STAT5 transcription factors, but without a commensurate degradation of these proteins, a cellular process we termed STAT-masking (45).

All 11 p53-Val¹³⁵-containing Hep3B cell lines, but none of the seven pSVneo-containing control cell lines, incubated at 32.5°C for 1 day and then exposed to IL-6 for 30 min displayed a marked loss of STAT3 and STAT5 immunostaining. Figure 1,A illustrates the dependence of the loss of STAT3 immunostaining on IL-6 concentration in one such p53-Val¹³⁵-containing cell line (Line 1); a detectable loss occurred at 1 ng/ml, and the loss was near maximal at 30 ng/ml. This IL-6 concentration dependence parallels the concentration dependence of the ability of IL-6 to induce acute phase plasma protein synthesis in hepatoma cells (5, 41). Figure 1,B illustrates the time course of the loss of STAT3 and STAT5 immunostaining in Line 1 cells treated with IL-6 at a concentration of 30 ng/ml. The loss of STAT3 and STAT5 immunostaining was near maximal by 20 to 30 min, but was transient, in that STAT3 and STAT5 immunostaining was restored by 120 to 240 min. This time course parallels the activation of STAT3 and STAT5 in hepatoma cells to generate DNA-binding competent transcription factors and the subsequent decline by 2 to 4 h of activated STAT3 and STAT5 (3, 4, 5, 6, 7). The loss of STAT3 cytoplasmic and nuclear immunostaining upon IL-6 treatment was observed using 1) an anti-peptide mAb to the amino-terminal 1 to 175 residues of STAT3, 2) a polyclonal anti-peptide STAT3 Ab to the carboxyl-terminal 750 to 769 residues in the STAT3 molecule, and 3) a polyclonal anti-PY-STAT3 Ab to the carboxyl-terminal phosphopeptide corresponding to residues 701 to 709 (Fig. 1,C). It is noteworthy that 1) the same secondary Ab was used in assays in Figure 1,C using the anti-STAT3 pAb and the anti-PY-STAT3 pAb; and 2) in additional control experiments, immunofluorescence due to anti-STAT3 pAb was inhibited by the cognate synthetic peptide used to produce this pAb but not by an irrelevant peptide (data not shown), thus verifying the specificity of the immunofluorescence data illustrated. Although cytoplasm to nuclear translocation of Tyr-P-containing STAT3 was clearly verified in Line 1 cells at 37°C using the anti-PY-STAT3 Ab, a considerable pool of Tyr-P-containing STAT3 remained cytoplasmic (Fig. 1,C, compare with Fig. 6). The monoclonal and polyclonal anti-STAT3 Abs both also showed evidence of some nuclear translocation of STAT3 in IL-6-treated Line 1 cells at 37°C (Fig. 1,C, left side), although the major immunofluorescence remained cytoplasmic in these Hep3B cells at all times (compare with Fig. 6 and see comment in Discussion).

In as much as IL-6-induced STAT-masking in cultures incubated at 32.5°C was observed using three different anti-STAT3 Abs raised against different epitopes in STAT3, the data in Figure 1,C (right side) suggest a dramatic reduction of immunologic accessibility to the entire STAT3 molecule during STAT-masking. Similarly, both the anti-STAT5a and the anti-STAT5b mAbs used revealed a loss of immunologic accessibility to STAT5 (Figs. 1,B and 2). However, the IL-6-induced loss of STAT3 and STAT5 immunofluorescence was selective in that it was not observed for STAT1; STAT4; STAT6; C/EBP-α, -β, or -δ; or Sp1 transcription factors using the same secondary Abs as those used in Figure 1 (45) (data not shown). Additionally, STAT-masking was not observed in human hepatoma HepG2 cells or in human diploid fibroblasts (data not shown).

IL-6-induced STAT-masking was not elicited in Line 1 cells incubated throughout at 37°C (Fig. 2,A). STAT-masking was elicited in only those Line 1 cells that had been incubated at 32.5°C for 20 h and not in those incubated at the wt-like p53 temperature (32.5°C) for either 5 or 10 h (Fig. 2,A). As the simplest possibility, the data suggest a need to accumulate a wt p53-induced gene product(s) during the 20-h incubation at 32.5°C before the STAT-masking phenotype can be elicited by IL-6. As an alternative possibility, the p53-Val¹³⁵ mutant may exhibit a gain-in-function phenotype at 32.5°C not seen with wt p53 (42). Once this hypothetical p53-Val¹³⁵-dependent gene product has accumulated in the cells at 32.5°C for 18 to 20 h, the cells could be shifted to 37°C for 30 min, and the STAT-masking phenotype could be fully elicited at 37°C upon IL-6 addition (Fig. 2 B).

Figure 3,A shows that neither IL-6, IFN-γ, nor epidermal growth factor elicited loss of STAT1 immunostaining. STAT3-masking was observed with IL-6 and to a lesser extent with IFN-γ, but not with epidermal growth factor; STAT5-masking was observed with only IL-6 (Fig. 3 A).

To identify the biochemical events underlying IL-6-induced STAT-masking, we examined the effects of a panel of inhibitors that block cell signaling pathways activated by IL-6. In these experiments, the inhibitors were added either 30 min before IL-6 to examine the effect of the inhibitor on the onset of STAT-masking or 30 min after the addition of IL-6 to examine the effect of the inhibitor on the continued maintenance of STAT-masking.

We have previously shown that addition of the tyrosine (Tyr) kinase inhibitors genistein and staurosporine, but not the serine (Ser) kinase inhibitor H7, to Line 1 cultures 30 min before IL-6 blocked STAT-masking, indicating a requirement for Tyr kinase activity in the onset of STAT-masking (45). Genistein and staurosporine added 30 min after the onset of STAT-masking did not affect the maintenance or the slow reversal of STAT-masking (data not shown), suggesting that continued Tyr kinase activity was not required for maintaining STAT-masking. Additionally, the proteasomal inhibitors MG132 and lactacystin blocked the onset, but not the maintenance, of STAT-masking (45) (data not shown).

Both orthovanadate and pervanadate, which inhibit protein Tyr phosphatases (3, 4, 5, 6, 7, 52, 53, 54, 55, 56, 57), blocked the onset of IL-6-induced STAT3-masking (Fig. 3,B) as well as that of STAT5 (data not shown). However, neither orthovanadate nor pervanadate added 30 min after IL-6 (i.e., after STAT-masking was established) affected the kinetics of reversal of STAT-masking that occur by 2 to 4 h (as in Fig. 1 B). In comparison, okadaic acid and calyculin A, which inhibit Ser-threonine phosphatases, each used at 100 nM, did not affect the onset or the maintenance of IL-6-induced STAT-masking (data not shown). These phosphatase inhibitor data are consistent with the possibility that recruitment of protein Tyr phosphatase(s) to Tyr-P-containing STAT3 and STAT5 may be involved in the production of a STAT-masking complex in p53-Val¹³⁵-containing cells.

The involvement of a Ser phosphorylation (Ser-P) reaction in mediating IL-6-induced STAT-masking is suggested by the ability of PD98059, a selective inhibitor of mitogen-activated kinase kinase 1 (MAPKK1/MEK) (58), to block the onset of STAT-masking (Fig. 3 B). The inability of the Ser kinase inhibitor H7 to block IL-6-induced STAT-masking (45) despite the ability of PD98059 to block STAT-masking suggests that IL-6-induced Ser-P of STAT proteins may itself occur through multiple pathways, not all of which are blocked by H7.

The data in Figure 2 indicated that IL-6-induced STAT-masking required accumulation of an induced gene product(s) that accumulated in p53-Val¹³⁵-containing cells over a period of 18 to 20 h. A recent report has described the increased accumulation of PLC-β4 in p53-Val¹³⁵-containing myeloid M1 cells incubated at the wt p53 temperature (32.5°C) (59). In contrast to observations in p53-Val¹³⁵-containing M1 cells, no increases in the cellular content of PLC-β₁, -β₃, -β₄, -γ₁, or -γ₂ were detected in Western blot analyses of total cellular extracts of p53-Val¹³⁵-containing Hep3B cell lines incubated at 32.5°C for 18 to 20 h compared with those incubated continuously at 37°C (data not shown).

Because IL-6-type cytokines are known to cause Tyr-P and activation of PLC-γ including the physical association between gp130 and PLC-γ (60), we investigated the role of IL-6-triggered PLC-mediated signaling in STAT-masking. We examined the effect on IL-6-induced STAT-masking of U-73122, an inhibitor of the agonist-coupled PI-PLC pathway (61, 62), and compared it to that of U-73342, an inactive congener of U-73122 (61, 62), and to that of D609, an inhibitor of phosphatidylcholine- and phosphatidylethanolamine-specific PLC and of phospholipase D activities (63). Figure 4,A shows that IL-6-induced STAT3- and STAT5b-maskings were blocked in Line 1 cells treated with U-73122 30 min before IL-6. Remarkably, U-73122 added 30 min after elicitation of STAT-masking, was able to rapidly unmask the masked state (Fig. 4,A), suggesting a requirement for continued PI-PLC signaling to maintain STAT-masking. In comparison, neither the inactive congener U-73342 nor D609 affected the onset of IL-6-induced STAT-masking nor its maintenance (Fig. 4,B). By Western blotting, the cellular levels of STAT3 and STAT5b proteins remained unchanged throughout the masking elicited by IL-6 and its reversal by U-73122 as in Figure 4 (data not shown).

The reversal of already established STAT-masking by the PI-PLC inhibitor U-73122 was rapid, in that the return of cytoplasmic STAT3 immunofluorescence was detectable within 3 min, and the reversal of both cytoplasmic and nuclear STAT3 immunostaining was complete within 15 to 20 min (Fig. 4,C). Taken together, the data in Figure 4 are consistent with a major role for a PI-PLC-mediated cell signaling pathway in the regulation of IL-6-induced STAT-masking.

The reduced responsiveness to IL-6 of Line 1 cells cultured at the wt p53 temperature (45) (Table I) suggested that there might be a reduction in the level of STAT3 DNA binding activity in nuclear extracts of cells during STAT-masking. This possibility was investigated by preparing cytoplasmic and nuclear extracts from Hep3B cells and from Line 1 cells induced with IL-6 at the two temperatures and assaying for DNA-binding-competent STAT3 in such extracts.

Figure 5,A shows that in replicate experiments there was a clear reduction in the level of STAT3 DNA binding activity in nuclear extracts prepared from Line 1 cells induced with IL-6 at 32.5°C compared with those derived from cells incubated at 37°C when assayed 30 min after cytokine addition. In comparison, the activation of STAT1 DNA binding activity by IFN-γ was unaffected at the two temperatures in such cells, and a ts phenotype was not observed when the parental Hep3B cells were induced with IL-6 at the two temperatures (data not shown). Furthermore, the data in Figure 5,B and the quantitative evaluation of replicate experiments included in Figure 5 B, suggest that the rapid reversal of STAT3 masking upon addition of the PI-PLC inhibitor U-73122, but not its inactive congener U-73422, is also accompanied by an increase in STAT3 DNA binding activity in the nuclear compartment when assayed 60 min after IL-6 addition.

In contrast to the data in Figure 5, no STAT3 DNA binding activity was detected in the cytoplasmic compartment in IL-6-induced Line 1 cells under nonmasked or masked conditions (data not shown) despite the presence of a considerable pool of Tyr-P-containing STAT3 in the cytoplasmic compartment (Figs. 1 C and 6).

The reduction in levels of STAT3 DNA binding activity assayed at 30 min during STAT-masking (Fig. 5,A) are in dramatic contrast to the marked increases in Tyr-P-containing STAT3 in both the cytoplasmic and nuclear compartments during STAT-masking (Fig. 6). To evaluate the state of Tyr-P of STAT3 during STAT-masking, Line 1 cells cultured at either 37 or 32.5°C were treated with IL-6 for different lengths of time, and cytoplasmic and nuclear fractions were prepared. STAT3 proteins were immunoprecipitated with anti-STAT3 pAb or anti-PY-STAT3 pAb using a buffer containing 0.5% Triton X-100 and 0.1% SDS (to overcome immunologic masking), followed by SDS-PAGE and Western blotting using an anti-STAT3 mAb (Fig. 6). The top panel in Figure 6,A shows undiminished cytoplasmic levels of STAT3 in all lanes at all times, indicating that the fraction of STAT3 that translocates from the cytoplasm to the nucleus in these cells is small. It is clear from the data in Figure 6, A and B, that not only did Tyr-phosphorylated STAT3 proteins (91-kDa STAT3 and 84-kDa STAT3β) accumulate in the cytoplasmic and nuclear compartments of Line 1 cells during STAT3-masking, but there was also an increase and a prolongation in the accumulation of Tyr-P-containing STAT3 proteins during STAT3-masking. While at 37°C, peak levels of Tyr-phosphorylated STAT3 were observed in both the cytoplasmic and nuclear compartments at 30 min, with a rapid decline by 60 min, considerable levels of Tyr-phosphorylated STAT3 were observed in the cytoplasmic and nuclear compartments under masking conditions (i.e., at 32.5°C) at both 30 and 60 min after cytokine addition. IL-6-induced Tyr-P of STAT3β at 60 and 120 min is also evident in the cytoplasmic compartment to equivalent levels at both temperatures, as is the IL-6-induced Tyr and Ser-P of STAT3β (the appearance of an IL-6-induced doublet of STAT3β in Fig. 6 B, lower panel).

Overall, the most dramatic aspect of the data in Figure 6 is the paradoxical result that even though there is no STAT3 DNA binding activity in the cytoplasmic compartment and there is a reduction in peak STAT3 DNA binding activity in the nuclear compartment (Fig. 5 A), there is a clear increase in the accumulation of Tyr-P-containing STAT3 in these compartments during STAT-masking. Thus, STAT-masking is indicative of a novel p53-dependent cellular process that regulates the DNA binding capacity of Tyr-phosphorylated STAT3 proteins.

The dramatic IL-6-induced STAT-masking observed in p53-Val¹³⁵-containing Hep3B cells represents the selective and regulated loss of STAT3 and STAT5 immunologic accessibility and appears to be a reflection of an inhibitory mechanism regulating JAK-STAT signaling (Fig. 7). The discovery of STAT-masking represents a departure from previous observations of a cytoplasm to nuclear translocation of STAT3 and STAT5 immunofluorescence upon treatment of hepatoma cells with cytokines (43, 44). It is clear from a comparison of the data in Figure 1,C with those in Figure 6 that while there occurred nuclear translocation of STAT3 during IL-6-induced STAT-masking, this nuclear STAT3, including Tyr-phosphorylated STAT3 (Fig. 6), was not visible by immunomicroscopy (Fig. 1 C). Furthermore, despite the increased and prolonged accumulation of Tyr-phosphorylated STAT3 in the nuclear compartment during STAT-masking, there was a reduction in peak levels of DNA-binding-competent STAT3 proteins and of functional signaling. As the simplest possibility, this inhibitory mechanism appeared to be dependent upon the accumulation of a wt p53-inducible gene product in the p53-Val¹³⁵-containing Hep3B cells. As an alternative possibility a gain-in-function phenotype of p53-Val¹³⁵ at 32.5°C may account for STAT-masking.

The onset of IL-6-induced STAT-masking required protein Tyr kinase, protein-Tyr phosphatase, proteasomal, and PLC activities. Unlike the inhibitors of Tyr kinase, protein-Tyr phosphatase, and proteasomal activity, all of which blocked the onset of IL-6-induced STAT-masking but not its maintenance, U-73122, an inhibitor of receptor-coupled PI-PLC activity, not only blocked the onset of masking but also rapidly reversed STAT-masking when added 30 min after IL-6, indicating a role for IL-6R-coupled PLC signaling in the negative regulation of JAK-STAT signaling and the termination of STAT-masking. Although, the cellular protein(s) that associates with STAT3 and STAT5 during IL-6-induced masking in p53-Val¹³⁵-containing Hep3B cells remains to be elucidated, the p53-Val¹³⁵-induced gene product(s) regulating STAT-masking is not in itself ts, in that once accumulated it functions at both 32.5 and 37°C (Fig. 2 B). The observation that human HepG2 cells and diploid fibroblasts that contain low levels of wt p53 do not exhibit STAT-masking suggests either 1) that there is a dosage effect with respect to the ability of wt p53 to induced this cellular phenotype; or 2) that p53-Val¹³⁵ exhibits a novel gain-in-function phenotype at 32.5°C that reveals itself as STAT-masking.

In Figure 1,C the same goat anti-rabbit IgG was used as the secondary Ab together with the anti-STAT3 pAb or the anti-PY-STAT3 pAb. The specificity of the immunofluorescence assay shown in Figure 1,C is verified by the strong cytoplasmic staining observed using anti-STAT3 pAb (which could be inhibited by the cognate but not by irrelevant synthetic peptide) but only weak cytoplasmic staining using anti-PY-STAT3 in control untreated cells. Upon addition of IL-6 at 37°C, the anti-PY-STAT3 pAb showed a marked increase in both cytoplasmic and nuclear immunostaining (Fig. 1,C). In contrast, the anti-STAT3 pAb and the anti-STAT3 mAb showed only a minor, but detectable, nuclear translocation in IL-6-treated cells at 37°C (Fig. 1,C, left side). It is important to emphasize that in contrast to the immunofluorescence observations for STAT1 in IFN-γ-treated human fibroblasts (3), translocation of STAT proteins from the cytoplasm to the nucleus usually involves only a minor fraction of the pool of cytoplasmic STAT proteins (52) (also see Fig. 6). Even in IFN-γ-induced fibroblasts, when metabolically labeled STAT1 was followed, at most 10 to 20% of cytoplasmic STAT1 was translocated to the nucleus (52).

The closest analogy to IL-6-induced STAT-masking, although in reverse, is the unmasking of p65 (Rel-A) immunofluorescence upon kainate-induced activation of NF-κB through the glutamate receptor in cultures of rat cerebellar granule cells provided that the immunostaining was assayed using an anti-peptide mAb to the nuclear localization sequence in p65 (64). The nuclear localization signal in p65 is complexed with I-κBα and is therefore immunologically inaccessible in the inactive state (reviewed in Refs. 64–67).

A functional consequence of STAT-masking is impaired responsiveness of the wt p53-containing hepatoma Hep3B cells to IL-6. This impairment has now been demonstrated at the level of transient transfection assays using the β-fibrinogen reporter construct (pβFibCAT), at the level of a reduction in the pool of DNA-binding-competent STAT3 in the nuclear compartment, and at the level of a reduced synthesis and secretion of IL-6-induced β-fibrinogen and of α₁-antichymotrypsin.

Despite reduced peak levels of DNA-binding-competent STAT3 in the nuclear compartment during IL-6-induced STAT-masking, there was increased and prolonged accumulation of Tyr-P (and Ser-P)-containing STAT3-proteins in the nuclear compartment, suggesting an inhibition of STAT3-PY-specific Tyr phosphatase activity during STAT-masking. The hypothesis that an inhibitor of STAT-PY-specific Tyr phosphatase may regulate IL-2-induced JAK1/3-STAT5 signaling has been suggested (68) based upon the observation that the proteasomal inhibitor MG132 prolonged Tyr-P-STAT5 accumulation in the nuclear compartment. We have previously reported that MG132 could partially block the reduction in STAT3 DNA binding activity during IL-6-induced STAT masking (45).

The ability of orthovanadate and pervanadate to block IL-6-induced STAT-masking suggests the recruitment of SH2 domain-containing protein Tyr phosphatases in the onset, but not in the maintenance, of STAT-masking. Several SH2-domain-containing proteins that regulate cytokine signaling have been recently described. As one example, a family of at least 15 SH2-domain-containing proteins, called signal regulatory proteins, that associate with Tyr-phosphorylated (i.e., activated) protein-Tyr phosphatases, such as SHP-1 and -2, and with adapter proteins such as Grb2 to inhibit growth-factor dependent Tyr kinase-dependent signaling has been characterized recently (69). A family of 20- to 25-kDa SH2 domain-containing, IL-6-inducible STAT3 inhibitory proteins has been recently characterized (70, 71, 72). Suppressor of cytokine signaling-1 inhibits both IL-6-induced receptor and STAT3 Tyr-P (70, 71, 72). In as much as Tyr-P of STAT3 is undiminished and prolonged during IL-6 STAT-masking in wt p53-containing Hep3B cells, the data suggest that suppressor of cytokine signaling-1 or its family members are not involved in STAT-masking. Additionally, Fiscella and colleagues (73) have characterized a protein phosphatase, designated Wip1, that is wt p53 dependent in its expression in irradiated cells. The possibility that a wt p53-inducible protein Tyr phosphatase may participate in IL-6-induced STAT-masking is intriguing. Very recently, Chung and colleagues (74) have described a protein designated PIAS3 that associates with Tyr-P-containing STAT3 and inhibits its DNA binding competence without affecting its Tyr-P status. The involvement of PIAS3 in STAT-masking remains to be investigated.

Both the onset and the maintenance of IL-6-induced STAT-masking are dependent upon receptor-coupled PLC signaling (Fig. 4). A requirement for PI-PLC signaling in the onset of STAT-masking may reflect novel IL-6-induced gp130-mediated PLC activation and the intracellular signaling pathway (75). Indeed, the rapid reversal within 15 to 20 min of already established STAT-masking by U-73122, an inhibitor of agonist receptor-coupled PI-PLC activity, but by no other inhibitor tested, provides additional evidence for the involvement of a novel IL-6R-coupled PI-PLC pathway in the regulation of JAK-STAT signaling and STAT-masking.

The identification of the p53-Val¹³⁵-dependent proteins involved in STAT-masking and the elucidation of their physiologic function represent an exciting challenge (Fig. 7). In preliminary experiments we have detected high molecular mass sedimentable complexes of STAT3 in the postmitochondrial cytoplasmic extracts of Line 1 cells during STAT-masking using the procedure of zonal centrifugation through sucrose density gradients. The characterization of proteins associated with STAT3 in these sedimentable complexes produced during STAT-masking is likely to provide insights into this new p53-dependent cellular process.

We thank Manohar V. N. Shirodkar, Pramila Warke, Elyse S. Goldweber, Josephine Lauriello, Suzanne Andrews, Kimberly A. Sorrentino, and Benjamin Z. Holczer for their encouragement and support, and Drs. Victor Fried, Sansar Sharma, and Richard Pine for helpful discussions and numerous suggestions.