Reconstitution of the response to leukemia inhibitory factor, oncostatin M, and ciliary neurotrophic factor in hepatoma cells.

Ciliary neurotrophic factor (CNTF) has been described as a neuro-active cytokine that shares functional similarities with the leukemia inhibitory factor (LIF). We demonstrate here that, like LIF, CNTF stimulates expression of acute phase plasma proteins in rat H-35 hepatoma cells. Transfection of the LIF receptor into Hep3B hepatoma cells reconstituted LIF and oncostatin M regulation of acute phase plasma protein genes. Co-expression of the LIF receptor and the CNTF receptor, but not expression of either subunit alone, generated CNTF responsiveness in Hep3B cells, suggesting cooperativity of these receptor subunits. Evidence is presented for direct interaction of the LIF receptor with the intracellular signal transduction machinery.

The coordinated synthesis of a set of plasma proteins in the liver is one of the manifestations of the systemic response to inflammation (1). Several distinct cytokines, including interleukin-6 (IL-6),' LIF, IL-11, and OSM are able to elicit a qualitatively similar response in liver cells (2)(3)(4)(5)(6)(7). The usage of common receptors (LIF and OSM) (8) or a common signal transducing subunit (gp130) (9)(10)(11) has been proposed to be the cause for the redundancy of action of these cytokines. Characterization of different hepatic cell lines indicated the existence of different spectra of cytokine responses. H-35 and Fao rat hepatoma cells respond to IL-6, LIF, OSM, and IL-11; HepGZ human hepatoma cells to IL-6, LIF, and OSM; and Hep3B cells to IL-6 alone. A common affinity convertor, gp130, has recently been described for the IL-6 and LIF/OSM receptors (9), but, conversely, differences do exist in the * This work was supported in part by Grant CA26122 (to H. B.) from the National Cancer Institute, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ''aduertkement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Subunits of the receptors for IL-6 (10, 13, 14), LIF (15), and OSM (9,16) are related in structure to a large family that includes subunits of the GCSFR (17) and CNTFR (18). The intracellular domains of the LIFR, gp130, and GCSFR display extensive homology (15). GCSFR is thought to act as a highaffinity homodimer (19), whereas the LIFR and gp130 combine to form a heterodimeric high affinity LIF/OSM receptor (9). A three-subunit model for the CNTF receptor complex has been suggested involving the low affinity CNTF receptor, gp130, and the LIF receptor (11). In order to define the nature of the functional LIF/OSM receptor, we attempted to reconstitute the cloned LIF receptor into non-LIF/OSM-responsive Hep3B cells. Similarly, transfection of the LIF and CNTF receptor cDNAs into Hep3B and HepG2 cells was used to reconstitute functional CNTF receptors in hepatoma cells.

EXPERIMENTAL PROCEDURES
Expression Constructs-For all expression constructs, the cytomegalovirus promoter-based plasmid vector pDC302 was used (15). The full-length human LIFR construct (pHLIFR-FL) was made by ligation of a genomic clone encoding the cytoplasmic domain to the truncated cytoplasmic domain of pHLIFR-65 via a common CellII restriction site (15). A 2.4-kilobase pair Asp718-Xmml fragment of pHLIFR-65 was ligated to Asp718-SmaI-digested pDC302 and resulted in a 746-amino acid soluble form of the hLIFR, terminating immediately in vector sequences (LIFR-soluble). A DNA fragment encoding amino acid residues 1-862 of the hLIFR was produced using the polymerase chain reaction and resulted in a clone encoding the entire signal, extracellular and transmembrane domains and 3 amino acids of the cytoplasmic domain as a membrane anchor (LIFRAcyto). An expression vector for the soluble form of the mouse LIFR has been described previously (15). A GCSFR-LIFR chimera encoding the extracellular domain of the GCSFR (residues 1-601) (17) and the transmembrane and complete cytoplasmic domain of the hLIFR (residues 823-1097) (15) was constructed by ligation of two polymerase chain reaction fragments. The clone encoding the human CNTFR was isolated from a human brain cDNA library using oligonucleotides based on the published sequence (18). The inserts of all constructions were sequenced. Binding affinities for LIF, CNTF, or GCSF (where appropriate) corresponded to those previously described (15,17,18) when the cDNAs were expressed in COS-7 cells.
Transfections and Plasma Protein Gene Expression Assays-DNA was transfected into HepG2 and Hep3B cells as a calcium phosphate precipitate (26) and into H-35 cells as a DEAE-Dextran complex (27). The following reporter gene constructs were used pBFB(2xIL-6RE)CT, containing two tandem copies of the 34-base pair IL-6RE of the rat 6-fibrinogen gene in pCT (23), and pHP(19O)-OCT, containing the rat haptoglobin gene promoter from -165 to +25 in pOCT (24). PIE-MUP served as an internal marker for transfection efficiency in all experiments (25). Following overnight recovery, the cell cultures were subdivided, and 24 h later, the subcultures were treated with cytokines and/or dexamethasone. After another 24 h, the me-dium was collected, and the amount of secreted major urinary protein (MUP) and plasma proteins was determined by immunoelectrophoresis. The chloramphenicol acetyltransferase (CAT) activity in the cell extract was normalized to the amount of MUP expressed by the same cells (% conversion/h X pg MUP). T o compare between experiments, the values in each experimental series were normalized to the untreated control (defined as 1.0).
To obtain stable LIFR transfectants of Hep3B cells, pHLIFR-FL, together with pSV2ne0, was transfected and cells selected using 1 mg/ml G418. Subclones were chosen on the basis of LIFR mRNA expression, LIF and OSM surface binding, and LIF-induced stimulation of haptoglobin and fibrinogen synthesis. Polyadenylated RNA was subjected to Northern blot analysis according to standard procedures.

RESULTS AND DISCUSSION
Transient transfection of a LIFR expression construct with the cytokine-responsive CAT gene reporter constructs pHP(19O)OCT or pPFG(2xIL-GRE)CT into Hep3B cells, resulted in two effects. 1) The basal activity of the CAT gene construct was elevated %fold (Fig. l . 4 ) ; 2) LIF and OSM stimulated CAT expression above this basal level and were equally effective (Table I). The IL-6 response was unaffected.
The elevated basal stimulation was caused, in part, by endogenous LIF production. Northern blot analysis revealed the presence of LIF mRNA, and incubation of Hep3B cells with neutralizing LIF antibodies decreased basal activity of the reporter gene by as much as 60% (Fig. IC). Antibodies against OSM were not effective (data not shown). LIFR appeared to function as well as the endogenous IL-6 receptor, since stimulation by OSM (in the presence of anti-LIF) was comparable to IL-6; no additive action was observed with any combination of LIF, OSM, and IL-6. Endogenous acute phase protein genes were similarly affected by transfection of LIFR. A stable cell line, Hep3B-LIFR, was derived which expressed LIFR mRNA and displayed small numbers of high affinity cell surface LIF and OSM binding sites (LIF site number per cell ( R ) = 760, equilibrium dissociation constant (KD) = 1.1 X 10"O; OSM: R = 129, KD = 1 X lo-"), and responded to LIF treatment by an increased synthesis of fibrinogen (Fig.  lB), and haptoglobin (data not shown). Hep3B-LIFR cells showed a 2-3-fold increase in basal fibrinogen and haptoglobin expression that was partially reduced with anti-LIF, suggesting that these cells also experience autocrine stimulation.
Expression of truncated LIFR forms revealed two interesting features (Table I) Transfected cells were subdivided and treated as indicated. CAT activity was normalized to MUP production and expressed relative to untreated control cells (numbers aboue autoradiogram; in this experiment, the value 1 represents a specific CAT activity of 6% conversion/h X pg MUP).     (Table I), which is capable of binding human LIF ( E ) , or addition of purified soluble mouse LIFR and LIF (data not shown) was unable to elicit a basal level stimulation of the reporter gene construct, suggesting that interaction of the extracellular domain of LIFR with any signal transducing receptor subunit is dependent upon species-specific structural features.
Since the extracellular domain of LIFR exerted such a prominent effect on the signaling event, the precise contribution of the cytoplasmic domain, if any, was not clearly apparent. gp130 has been described as a common signal transducing subunit for IL-6-type cytokine receptors (11,18,28). LIFR has a 238-amino acid cytoplasmic domain, which is highly conserved between man and mouse (87% identity)2 and which is homologous to gp130 (9,15). This extensive homology between LIFR and gp130 suggests that the cytoplasmic domain of LIFR may act as a signal transducer and may contribute to the LIF/OSM response. Since there are no hepatic cell lines that lack gp130, we decided to assess the function of the cytoplasmic domain of LIFR as part of a GCSFR extracellular domain-LIFR transmembrane and cytoplasmic domain (GCSFR-LIFR) chimera (Table 11). The functional form of the GCSFR is a homodimer that does not require gp130 for function (19,29). We have previously observed that transfected GCSFR is functional in hepatic cells and that substitution of the cytoplasmic domain of gp130 onto the GCSFR extracellular domain confers GCSF responsiveness to Hep3B cells (Table I1 and Ref. 30). A GCSFR-LIFR chimera expression construct was able to confer GCSF responsiveness on Hep3B cells (Table 11), HepG2 cells, and H-35 cells (data not shown). The chimeric receptor did not yield a basal level activation of the reporter gene construct as seen for LIFR but was consistently less effective than the GCSFR-D. P. Gearing gp130 construct. The inverse receptor combination, the extracellular domain of LIFR with the trans-membrane and the cytoplasmic domain of GCSFR, was unable to generate a ligand-dependent stimulation (Table 11). Taken together, these data suggest that the LIFR cytoplasmic domain can lead to productive intracellular signal transduction and that a signaling event appears possible in the absence of gp130. However, definitive proof for a signaling mechanism independent of gp130 demands further structural and functional analyses of the LIFR cytoplasmic domain.
To date, CNTF has only been reported to function on neural cells, but it shares many of these functions with LIF and, like LIF, can be toxic when given systemically (31, 32). A recent study concluded that the functional CNTFR complex involves gp130 and perhaps LIFR (11). In order to assess whether CNTF affects other systems and requires LIFR for function, we tested it on hepatic cells. CNTF, like LIF, stimulated, in a dose-dependent fashion, acute phase plasma protein genes in H-35 cells (Fig. 2), although the specific activity of CNTF was 10-fold lower. Similar CNTF and LIF action was demonstrated by the comparable inhibitory response of dexamethasone on the expression of the thiostatin gene. By contrast, dexamethasone enhanced IL-6-induced expression of the thiostatin gene (Fig. 2). CNTF also stimulated a LIF-like acute phase protein response in HepG2 cells. As with LIF and OSM, CNTF was also inactive on Hep3B cells.
Since Hep3B cells expressed a 1 3 0 and failed to respond to LIF and CNTF, we tested whether LIFR expression was required for CNTF function. Hep3B cells were transiently transfected with CNTFR, LIFR, or the combination, together with pHp(l9O)OCT (Table 111)  of endogenous LIF, the cytokine treatments were carried out in the presence of anti-LIF antibodies. While LIFR-transfected cells responded to OSM, CNTF was only minimally active. An equally low CNTF response was recorded in Hep3B cells that had been transfected with CNTFR. Only co-expression of both receptors yielded a 3-fold enhanced CNTF response, although this was only half of the OSM response. Using LIFR with a deleted cytoplasmic domain, no CNTF response was achieved. These results support the proposal (11) that CNTFR cooperates with gp130 and LIFR in order to generate CNTF signal transduction. However, the data also indicate that the subunit combination in Hep3B cells was less effective in CNTF signal transduction than it was for OSM signal transduction. To achieve both CNTF and LIF/OSM response, intact LIFR was required.
HepG2 cells responded to CNTF in a fashion similar to LIF. CNTF and LIF were not additive in action, and OSM caused a much greater stimulation (Table IV). HepG2 cells express both the LIF/OSM receptor and the OSM-specific re~eptor.~ CNTFR-transfected HepG2 cells displayed a higher response to CNTF than control transfectants but no elevated LIF or OSM response (Table IV). These data suggest that either CNTFR does not enhance LIF or OSM signaling by the LIFR.gp130 complex, or that CNTFR can interact with other receptors on the surface of HepG2 cells. One candidate on HepG2 cells is a second receptor that binds OSM but not LIF (8).
These data, together with the previous findings (9, 11), suggest an expansion of the model for the function of the IL-6 receptor-related subgroup of the hematopoietin receptor family. In order to transduce a signal, dimers of intracellular domains may be required. Such active dimers have been documented for LIFR (Table 11), gp130 (30), and LIFR. g~1 3 0 .~ A variety of subunit combinations between ligandbinding subunits (IL-6R, CNTFR and probably IL-11R) and signaling subunit dimers is likely. Involvement of heterotrimeric combinations cannot yet be ruled out, as previously suggested for the high affinity receptor for LIF, OSM, and IL-6 (9). Heterotrimeric complexes seem to be necessary for CNTF responsiveness and would include CNTFR . gp130. LIFR (Table 111)  IV). We have been unable to detect CNTFR mRNA by Northern blot analysis in HepG2 or H-35 cells despite their ability to respond to CNTF. We concluded that either CNTF might be able to interact weakly with the LIFR. gp130 dimer and that expression of CNTFR enhances this interaction (Tables I11 and IV), or CNTF is acting in these cells via a yet to be defined second form of CNTFR. Additional work will be required to understand which combinations of subunits direct cellular responses and to determine how these subunits are linked to the intracellular signal transduction apparatus.