STAT3 and STAT5B are targets of two different signal pathways activated by hematopoietin receptors and control transcription via separate cytokine response elements.

Transient transfection of expression vectors for various members of the hematopoietin receptor family and STAT proteins into COS-1 cells indicated that each receptor was capable of stimulating the DNA binding activity of STAT1, STAT3, and STAT5B. However, gp130 preferentially activated STAT1 and STAT3. Activation of STAT5B differed from that of the other two in that the box 3 sequence motif in the cytoplasmic domain of gp130 was not required. Moreover, STAT5B and STAT3 enhanced gene transcription via separate regulatory elements. This study has identified two potential signal transduction pathways by which hematopoietin receptors, including the interleukin-6 receptor, control transcription of acute phase plasma protein genes in hepatic cells.

The transcriptional regulation of acute phase plasma protein genes in hepatic cells by IL-6 1 has been correlated with the activation of DNA binding properties of STAT3 by the signaling activity of gp130 (1)(2)(3)(4)(5). The suggested function of STAT3 as a transcription factor was supported by the finding that overexpression of STAT3 and its activation by cytokine receptor and Janus kinases resulted in enhanced transcription via IL-6responsive gene elements (6). However, the model proposing a principal role for STAT3 as a mediator of acute phase response (3,7) needed to be refined because: 1) gp130-dependent transcription via certain elements, such as the HRRE, was found to be independent of STAT3 (8); and 2) the acute phase response of the liver included activation of not only STAT3 but also of other members of the STAT protein family, such as STAT5B (9). 2 The study of the transcription control mechanisms by specific STAT isoforms has been difficult primarily due to the lack of adequate experimental assay systems. Recently, we have developed techniques to reconstitute the function of hepatic and non-hepatic hematopoietin receptors in transiently transfected hepatoma cells (10). We could define the cytoplasmic domains of the signal transducing receptor subunits required for the induction of transcription through specific regulatory elements (8,11). This cell assay system was used to characterize the specificity of STAT protein activation by hematopoietin receptors. Two distinct signaling pathways were identified: one specified by the box 3-dependent activation of STAT3 and transcriptional stimulation via an IL-6RE, and the other specified by the box 3-independent activation of STAT5B and transcriptional stimulation via HRRE.
Cell Transfection and Analysis-COS-1 cells were transfected with plasmid DNA by the DEAE-dextran method (21) and HepG2 cells by the calcium phosphate method (22). The cell cultures were subdivided. Subcultures of COS cells were maintained for 16 h in serum-free medium prior to the activation of STAT proteins by treatment with cytokines for 15 min. DNA binding activities of STAT proteins in whole cell extracts were determined by EMSA (23). Double standard oligonucleotides for the high affinity SIEm67 (23), the mammary gland factor binding site of the rat ␤-casein gene (PRL response element, or PRE; Ref. 24), and TB-2, a duplicated IL-6RE sequence (TB-1) of the rat ␣ 2 -macroglobulin gene (2), served as EMSA substrates. The STAT1, STAT3 and STAT5B-containing complexes with PRE and TB-2 were identified by supershift assay using 2 g of monoclonal anti-STAT1 (Transduction Laboratories), 0.5 g of rabbit anti-STAT3 (Santa Cruz Biotechnology), or 0.5 l of antiserum against STAT5 (generous gift of Dr. H. Wakao, DNAX). Aliquots (5 g) of whole cell extracts were electrophoresed on a 6% SDS-polyacrylamide gel, and the proteins were transferred to Immobilon membrane (Millipore). The membranes were reacted with either rabbit anti-STAT3 or anti-STAT5 (Transduction Laboratory) and then processed for chemiluminescent reactions (Amer-* This study was supported by Grants CA26122 from the National Institutes of Health (to H. B.) and DFGHi291/5-4 from the Deutsche Forschungsgemeinschaft (to G. H. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
sham). Transfected HepG2 cell cultures were treated for 24 h with cytokines. CAT activities were quantitated by testing serial dilutions of cell extracts, normalized to the expression of the cotransfected marker plasmid pIE-MUP (11), and expressed relative to the value of the untreated control cultures in each experimental series (defined as ϭ 1.0).

RESULTS AND DISCUSSION
Activation of Multiple STAT Proteins by gp130 -To identify the STAT proteins activated by gp130, we employed three oligonucleotides as diagnostic substrates: the high affinity SIE (23), the PRE of the ␤-casein gene (24), and TB-2 (2). Each probe yielded a specific EMSA pattern of IL-6-regulated DNA binding proteins in H-35 cells (Fig. 1). Both STAT1 and STAT3 bound to the SIE, giving rise to the complexes SIF-A, -B, and -C (23). The ␤-casein element was recognized primarily by the STAT1 homodimers (co-migrating with SIF-C). The STAT proteins contributing to the slow mobility TB-2 complex could not be deduced from the relative position with co-migrating SIF complexes.
Differences in binding specificity for the probes emerged when tested with STAT5B ( Fig. 1). STAT5B did not detectably bind to the SIE. The ␤-casein element yielded two complexes: one co-migrating with SIF-B and likely containing the predicted STAT5 dimer, and the other showing a much slower mobility and co-migrating with the single complex observed with the dimeric TB-2. Since both complexes reacted with anti-STAT5 detectable by supershift EMSA (data not shown), the slow mobility complexes were interpreted to be consistent with a STAT5B tetramer. The comparison of the patterns in Fig. 1 also indicated that in extracts of IL-6-treated H-35 cells, STAT1 and STAT3 predominate, obscuring detection of STAT5-containing complexes. We could, however, detect STAT5 in nuclear extract of IL-6-treated cells by Western blot analysis (data not shown).
To define the activation of specific STAT proteins by gp130, we transiently over expressed rat STAT1, STAT3, and STAT5B together with the chimeric receptor G-CSFR-gp130 in COS-1 cells. The chimeric construct permitted analysis of its signaling independent of the endogenous gp130 (11). We verified that the STAT expression vectors yielded abundant and approximately equal amounts of STAT proteins ( Fig. 2A; Ref. 6). To grade the preference of gp130 for specific STAT proteins, we compared its action with that of co-introduced IL-3R (Fig. 2B). IL-3R was chosen because of its strong stimulation of transcription via HRRE in hepatoma cells (12) and its difference from gp130 in lacking a box 3 motif.
Both receptor types were capable of transducing the signal to each STAT isoform. The comparison also indicated that IL-3R preferred STAT5B, whereas gp130 preferred STAT1 and STAT3, consistent with the action of native IL-6R in H-35 cells (Fig. 1). The EMSA patterns as well as antibody supershift assays (not shown) did not provide any evidence for appreciable formation of heterodimers between STAT1 and STAT5B or between STAT3 and STAT5B. TB-2 contains two IL-6REs in a 20-base pair span and was thus predicted to bind two STAT dimers. Surprisingly, this probe did not produce detectable amounts of mixed tetrameric complexes containing one homodimer each of STAT1 and STAT5B, as judged from the absence of complexes with intermediate mobilities on EMSA (Fig. 2B) or reacting to both anti-STAT1 and anti-STAT5 (not shown). We hypothesized that either two different STAT dimers were incompatible in simultaneous binding to TB-2, or that STAT5B binds as a preformed tetramer. The results obtained with pairwise combinations of STATs indicated a certain degree of competition among STATs for activation by the receptors that was correlated with the preferred usage of the particular STAT by the receptor.
Activation of STAT5B by gp130 Is Independent of the Box 3 Motif-Since gp130 activated STAT5B (Fig. 2B), we asked whether this process, like the activation of STAT1 and STAT3 (8,11), was dependent on a functional box 3. We transfected COS-1 cells with G-CSFR-gp130 in which the cytoplasmic domain had been truncated to 133 residues. This minimal size gp130 with three of the four box 3 motifs removed still activated STAT3 and transcription via IL-6RE (8). As expected, this construct also mediated STAT5B activation that was comparable to the full-length gp130 (Fig. 3). The mutation Y125A in box 3 (M3; Ref. 8) abolished activation of STAT3 (no detectable complex with TB-2 ( Fig. 3) or SIE (data not shown; Ref. 8)), but did not affect activation of STAT5B (Fig. 3). Based on data not shown, a similar level of STAT5B activation was achieved with gp130 truncated to 65 residues. With the deletion of box 2 (40 residues left), STAT5B regulation was substantially reduced. Hence, the cytoplasmic domain of gp130 required for STAT5B regulation was the same as needed for stimulating transcription via HRRE (8). When an expression vector for JAK2 was cotransfected, a ligand-and box 3-independent activation of STAT5B was gained (Fig. 3) that was equivalent to that observed previously for STAT1 and STAT3 (6,8).
From these results, we concluded that gp130 utilizes separate mechanisms to activate STAT3 and STAT5B. The two pathways have distinct requirements for cytoplasmic receptor domains, but both are dependent on the action of JAKs. One consequence of JAK action is the tyrosine phosphorylation of STATs, thereby inducing dimerization and manifestation of DNA binding activity (24 -26). Since direct binding of STATs via their SH2 domain to a phosphotyrosine residue in the receptor molecule is not mandatory for their activation (27), an alternative mechanism for bringing STATs in contact with receptor-bound kinases must exist.
One possibility yet to be tested is that STATs directly interact with the kinase or connector molecules.
One prediction for the proposed mechanism of interaction between receptors and STATs was that other hematopoietin receptors lacking a box 3 sequence but utilizing JAK2 should also activate STAT5B with comparable efficiency to IL-3R in our assay system. As anticipated, PRLR and GHR, which have been associated with the STAT5 pathway (24,26,28), promoted activation of STAT5B (Fig. 4). Nevertheless, these receptors also showed activation of STAT3, albeit at variable and much lower levels and only when STAT3 was provided in sufficiently high concentrations (Fig. 4; Ref. 6).
STAT5B and STAT3 Do Not Stimulate Transcription through the Same Regulatory Elements-The data obtained with COS-1 cells illustrated the preferences of hematopoietin receptors for STATs. The question arose whether the differences in STAT patterns, in particular the use of STAT5B, could be detected at the level of transcription. We selected HepG2 cells as an assay system because transcriptional regulation by various hematopoietin receptors could be reconstituted in these cells (10,12,16), and we expected to find an enhanced responsiveness by overexpressed and receptor-activated STAT proteins.
Functional screening of various regulatory elements indicated that all elements containing a core sequence related to the ␥-activating site (GAS) including the IL-4RE/GAS of the Fc␥R1 gene, ␤-casein PRE, the IL-6RE of the ␣ 2 -macroglobulin gene, and, foremost, HRRE, were responsive to STAT5B. Transfection of GHR and PRLR together with the HRRE-CAT construct and increasing amounts of STAT5B expression vector established a dose-dependent increase of CAT gene transcription (Fig. 5). The enhancing effect of STAT5B was noted on both the basal and receptor-mediated transcription. The change in transcriptional regulation via HRRE was STAT5B-specific because STAT3 (Fig. 5), STAT1, or STAT6 (data not shown; Ref. 6) proved to be ineffective.
In HepG2 cells, both GHR and PRLR were able to control transcription, in part via STAT5B. This finding differs from similar experiments with STAT5A in COS cells reported by Gouilleux et al. (29), who observed that PRLR but not GHR or erythropoietin receptor mediated enhanced gene transcription via the ␤-casein element. Both studies concur in that the magnitude of STAT5 activation by the receptor did not correlate well with the magnitude of transcriptional induction. For instance, in Fig. 5, neither GHR nor PRLR reconstituted an increase in transcription of comparable magnitude with that by the endogenous IL-6R.
To confirm the role of STAT5B in hematopoietin receptor signaling leading to gene transcription, we extended the analyses to other receptor combinations. We focused on those hematopoietin receptors that had shown low action in HepG2 cells and for which no autocrine pathway or response to serum factors was detectable. Moreover, for some of these receptors, the activation of STAT5 had been already reported (30 -32). We achieved transcription-enhancing effects via HRRE and related GAS sequences with c-Mpl, and box 3-deleted constructs of G-CSFR-LIFR(140), G-CSFR-gp130(40), and G-CSFR (27). The most prominent effect of STAT5B on transcription was, however, obtained with IL-2R and IL-4R (Fig. 6). STAT5B enhanced the signal of the transfected receptor, resulting in a transcription of the reporter gene that equaled or even exceeded the effect of the endogenous IL-6R. Moreover, when this optimal experimental receptor system was used in combination with the IL-6RE-CAT construct, the results illustrated the clear difference in the regulatory actions of STAT5B and STAT3. STAT5B enhanced transcription via HRRE but not via IL-6RE, and STAT3 was ineffective via HRRE (Fig. 5) but enhanced transcription via IL-6RE (Fig. 6). What still needs to be determined is the extent to which STAT3 and STAT5, or functionally redundant factors, contribute to the regulation of FIG. 3. Influence of box 3 motif on STAT activation. COS-1 cells were transfected with expression vector for G-CSFR-gp130(133) wild type (wt) or M3 (1 g/ml each) and STAT5B or STAT3 (2.5 g/ml) and JAK2 (0.1 g/ml). Whole cell extracts were used for EMSA with TB-2 as probe. One reaction with extract containing active STAT3 was incubated with anti-STAT3 to supershift the STAT3-containing complex (wt ϩ Ab).
FIG . 4. Action of GHR and PRLR. COS-1 cells were transfected with expression vectors for GHR and PRLR (1 g/ml each) and for STAT3 or STAT5B (2.5 g/ml) as indicated. STAT activities were determined as in Fig. 2. genes by IL-6 in hepatoma cells.
This study demonstrates that hematopoietin receptors use two distinguishable pathways to regulate gene transcription. The following model is proposed. The basic signaling pathway probably exerted by most if not all hematopoietin receptors requires minimally the box 1 motif. Box 1 engages one of the JAK family members (33), which recruits primarily STAT5 and much less the other STATs. Upon phosphorylation, the dimeric or oligomeric STAT5 interacts with GAS/PRE/TB-2-related sequences. If the receptor bears a box 3 or related sequence, STAT 1, STAT3 (8), or STAT6 (34) are also recruited to receptors by using box 3 as docking site. The subsequent activation process by the receptor-bound JAK may be similar to that affecting STAT5. The genetic targets of STAT3 differ from that of STAT5; thus, separate sets of genes may be controlled by specific members of the STAT family. Although the precise action of STAT proteins as transcription factors has yet to be defined, it is commonly assumed that the binding of STATs to the DNA element enhances transcription of the genes in cis. Although STAT1 and STAT6 are also regulated by some of the hematopoietin receptors used in our study, no transcriptioncontrolling action of these STATs could be detected by the approach of overexpression (6). It is conceivable that under physiological conditions STAT1 or STAT6 can interfere with or modulate the action of the other STAT proteins by competition for binding to the same DNA sequences.
Unquestionably, the transfection experiments are crude reconstitutions of the normal regulatory systems operative in cells and may yield artificially exaggerated responses. How-ever, these experiments have emphasized the importance of (a) the relative concentrations of receptors, kinases, and STATs and (b) the sequence specificity of the binding of STATs to DNA response elements. A most notable effect also is that excess amounts of JAKs lead to ligand-independent activation of STATs and transcription, suggesting that under such conditions alternative pathways that are generally unnoticed have become predominant. Moreover, every transcription assay is performed in cells that contribute a complex combination of endogenous factors, some of which may significantly influence the observed regulatory events. FIG. 5. Stimulation of transcription by STAT5B. HepG2 cells were transfected with pHRRE-CAT (10 g/ml) and expression vectors for GHR and PRLR (1 g/ml each) and increasing amounts of STAT5B or STAT3 as noted. Subcultures were treated without cytokines (control) or with PRL, GH, or IL-6. The change in CAT activity was calculated relative to control cells without STAT supplementation. Values represent means of two separate experimental series.

FIG. 6. Transcription control by IL-2R and IL-4R via STAT5B.
HepG2 cells were transfected with either pHRRE CAT or pIL-6RE-CAT (10 g/ml), expression vectors for IL-2R␤, IL-2R␥ and IL-4R (1 g/ml each) and either increasing amounts of STAT5B (noted in left panel) or STAT3 or STAT5B (5 g/ml) (right panel). Subcultures were treated with IL-2, IL-4, or IL-6, and the change in CAT activity was calculated relative to control culture. Similar regulatory effects were measured in three additional experiments.