Leptin Receptor Action in Hepatic Cells*

Leptin, an adipocyte-secreted hormone, is one of the central regulators of body weight homeostasis. In humans and rodents, two major forms of leptin receptors (OB-R) are expressed. The short form (OB-RS), considered to lack signaling capability, is detected in many organs. In contrast, OB-R long form (OB-RL) predominates in the hypothalamus, but is also present at low levels in peripheral tissues. Transient transfection experiments have demonstrated that OB-RL transduces an intracellular signaling similar to interleukin (IL)-6 type-cytokine receptors. To define the specificity by which OB-R induces genes and cooperates with signal transduction pathways utilized by other hormones and cytokines, rat and human hepatoma cell lines were generated which stably express human OB-RL. Hepatoma cell lines selected for appreciable levels of OB-RL mRNA display enhanced leptin binding and responded to leptin with an IL-6 receptor-like signaling that includes the activation of STAT proteins, induction of acute-phase plasma proteins, and synergism with IL-1 and tumor necrosis factor-α. A leptin-mediated recruitment of phosphatidylinositol 3-kinase to insulin receptor substrate-2 was also detected. However, no significant tyrosine phosphorylation of insulin receptor substrate-2 and modulation of the immediate cell response to insulin were observed. The data suggest that OB-RL action in hepatic cells is equivalent to that of IL-6 receptor. However, leptin does not play a specific role in muting insulin action on hepatoma cells and therefore may not contribute to the diabetic symptoms associated with obesity.

Leptin, a 16-kDa non-glycosylated secretory protein, is produced by adipose cells, released into the circulation, transported across the blood-brain barrier, and acts in a central manner on the hypothalamus to regulate mammalian energy homeostasis (1)(2)(3)(4)(5). Leptin signaling is mediated by the leptin receptor (OB-R), 1 a member of the hematopoietin receptor family that appears most closely related to the signal-transducing subunits of the IL-6 type cytokine receptors (6 -8). In humans and rodents, two predominant forms of OB-R are detected. Both isoforms have identical extracellular domains and ligand-binding affinity, but differ in the intracellular domains which represent alternative splice products. The major OB-R short form (OB-R S ) has a 34-amino acid cytoplasmic domain and is found in many organs. However, despite normal ligand binding activity, OB-R S has been described as being incapable of signaling (9,10). In contrast, the long form (OB-R L ) containing a 302-amino acid cytoplasmic domain, is primarily expressed in specific nuclei of the hypothalamus (11) and is considered to be the signaling-competent receptor isoform (9,10,12).
Reverse transcriptase-polymerase chain reaction and RNA protection analyses have revealed that various peripheral organs, including liver, have detectable levels of mRNA encoding the OB-R L . The relative amount ranges from 2 to 11% (liver 5%) of the total OB-R mRNA signal (9). Recent reports have suggested that leptin can induce a biological response in cells derived from peripheral tissues. For example, treatment of preadipocytes with leptin reduced the expression of acetyl-CoA carboxylase mRNA and lipid synthesis (13). Moreover, in human hepatoma (HepG2) cells, it has been proposed that leptin can oppose the action of insulin by reducing phosphorylation of IRS-1 and enhancing the stimulation of phosphoenolpyruvate carboxykinase mRNA level (14). However, in these studies, leptin cooperated with insulin to enhance association of phosphatidylinositol 3-kinase (PI3K) with IRS-1. Of note, the response of HepG2 cells in these experiments was attributed to the action of the OB-R S because mRNA corresponding to OB-R L could not be detected.
Our recent studies utilizing receptor-transfected hepatoma cells indicated that OB-R L functions independently of gp130 (10), and probably involves receptor homo-oligomerization (15). In these experiments, signaling by the OB-R L was characterized as comparable to that of the structurally related gp130, and included the activation of STAT1, STAT3, and STAT5 and the induction of gene transcription through regulatory elements derived from acute-phase plasma protein (APP) genes. Since these experiments relied on transient overexpression of OB-R L , the regulatory effect of the OB-R L signal on endogenous genes or on other hormone response pathways, such as for insulin, could not be adequately addressed. Since the nontransfected hepatoma cells did not respond to leptin by activation of STAT proteins or reporter gene regulation, these experiments suggested that these cells are deficient in (or may have lost) the expression of biologically important level of the OB-R L .
To compare the OB-R L -induced cell response to that mediated by the related IL-6-type cytokine receptors, and to assess cross-talk between signaling pathways for leptin and insulin, we established hepatoma cell lines stably expressing OB-R L . Analysis of these cell lines demonstrated that OB-R L functions indistinguishably from the endogenous IL-6R. Moreover, although both rat and human hepatoma contain leptin binding activity, only in the presence of introduced OB-R L could significant activation of STAT proteins and recruitment of PI3K to IRS-2 be achieved. The latter action is similar to that observed for the endogenous receptors for LIF and OSM.

MATERIALS AND METHODS
Tissue Culture Cells and Hormone Treatment-Subclonal line H-35 rat hepatoma cells (T-7-18-1) (16), and HepG2 human hepatoma cells; a clonal line established from the 78th passage of the originally described HepG2 cell culture (17), were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, penicillin, streptomycin, and gentamycin. All hormone treatments were started after maintaining the cells for 16 h in minimal essential medium with 0.5% fetal calf serum. Treatments were carried out in minimal essential medium with 0.5% fetal calf serum alone (ϭ control) or containing 1 M dexamethasone, 0 -100 ng/ml human IL-6 (Genetics Institute), 0 -100 ng/ml mouse leptin (R&D Systems), 100 ng/ml oncostatin M (OSM), 100 ng/ml leukemia inhibitory factor (LIF) (Immunex Corp.), 50 ng of human TNF␣, 10 ng/ml IL-1␤, 100 ng/ml human growth hormone (Genentech), or 500 ng/ml porcine insulin (Sigma). Treatment length ranged from 5 (insulin-induced phosphorylation of IRS proteins) to 15 min (activation of STAT proteins) and 24 h (stimulation of APP production). For combination treatments, cells were first incubated for 10 min with leptin and then, for an additional 5 min with leptin plus insulin.
Generation of Cell Lines Stably Expressing the Long Form of Human OB-R-A retroviral expression vector for human OB-R (MINV-hOB-R) was constructed by replacing an internal phosphoglycerate kinase gene promoter present in a derivative of the MSCV retroviral vector, MINV (18,19) with a 3.1-kb ClaI-NotI fragment of the human OB-R cDNA (6). In MINV-hOB-R, the hOB-R cDNA is expressed from the MSCV long terminal repeat on a bicistronic transcript which also contains a downstream neomycin phosphotransferase (neo) gene linked via an internal ribosome entry site from the encephalomyocarditis virus (20). Recombinant MINV-hOB-R retroviruses were produced as replication-defective pseudotyped particles carrying either the amphotropic or the Gibbon ape leukemia virus envelope proteins from stably transduced PA317 (Ref. 21 (24). Producer lines were maintained in Dulbecco's modified Eagle's medium with 4.5 g/liter glucose and 10% fetal calf serum. Virus-containing supernatant was collected from confluent cultures 24 h after medium change and incubation at 32°C. Supernatant from PA317/MINV-hOB-R cells were used to transduce H-35 cells and that from PG13/MINV-hOB-R cells was used to transduce HepG2 cells, essentially as described (25). Viral transductions were carried out in the presence of 8 g/ml Polybrene (Sigma) for 24 h at 37°C with two replacements of the virus-containing supernatant during the 24-h period. The cells were then subjected to selection by culturing in medium containing 2 mg/ml G418 (Life Technologies, Inc.). Proliferating cells recovered after 2 weeks of G418 selection, termed H-35(OB-R) or HepG2(OB-R), were cloned by limiting dilution. For each cell line, 36 individual clones were analyzed for expression of functional OB-R by determining human OB-R mRNA level, the magnitude of leptin-mediated STAT3 activation and the increase in APP secretion. Long-term stability of OB-R expression was determined in subclones that were obtained after 8 weeks of continuous propagation of the initially selected clonal cell lines.
Leptin Binding-Aliquots of 2.5 g of mouse leptin were 125 I labeled with chloramine T on ice for 30 s (26) yielding a specific activity of 1.5-4 ϫ 10 18 dpm/mol. The labeled protein was purified by chromatography on a column of Sephadex G-25 superfine in 50 mM sodium phosphate, pH 7.5, containing 0.25% gelatin, followed by chromatography on a column of Sephacryl S-200 superfine in binding medium (50 mM Hepes, pH 7.5, in minimal essential medium containing 1% bovine serum albumin). The final preparations of 125 I-leptin yielded a single radioactive band with M r ϭ 16,000 on a 15% polyacrylamide gel and had full biological activity based on the stimulation of APP production in H-35(OB-R 13) cells (see Fig. 5 for assay). For saturation binding assay, cell monolayers in 6-well culture plates were incubated for 2 h at 4°C in 1 ml of binding medium containing increasing concentrations of 125 I-leptin. After washing the cultures, cell-associated radioactivity was determined by scintillation counting and values were analyzed according to Scatchard (27).
EMSA and Northern Blot-Whole cell extracts were prepared and analyzed by EMSA as described (28). The 32 P-end-labeled 23-base pair SIEm67 (28) served as high affinity substrate for STAT1 and STAT3 and the 40-base pair TB-2 (29) as high affinity substrate for STAT5 but also as minor binding substrate for STAT1/STAT3 heterodimer (30). Anti-STAT3 (C-20) and anti-STAT5 (C-17) (Santa Cruz, Biotechnology) were used for supershift assay. Total cell RNA was isolated (31) and polyadenylated fraction prepared by chromatography on minispin columns of oligodeoxythymidine cellulose. RNA were electrophoresed in formaldehyde-containing agarose gel, transferred to Nytran Plus membrane (Schleicher and Schuell), and analyzed by hybridization with 32 P-labeled random primed cDNA to human OB-R L , mouse OB-R L (6), rat hemopexin, or rat thiostatin (32). Ethidium bromide staining patterns of the separated RNAs served as indicators of RNA loading.
Proteins were separated on mini-SDS gels (Bio-Rad), electrotransfered to nitrocellulose membrane (Schleicher and Schnell), and reacted sequentially with antibodies as indicated in the figures. For the analysis of F13K with coimmunoprecipitated IRS-1 or IRS-2, the membrane containing transferred proteins was divided at the 120-kDa size position. The lower section was reacted with anti-PI3K 85-kDa regulatory subunit (Upstate Biotechnology, Inc.) and the upper section with anti-IRS-1 or -IRS-2. The immunocomplexes were visualized by enhanced chemiluminescence reaction (Amersham).
Secretion Assay-Confluent monolayers of hepatoma cells in 24-well culture plates were incubated for 24 h with 400 l of treatment medium. To determine the amount of secreted APPs, 15-70 l of the culture supernatants were directly applied to immunoelectrophoresis plates (33). The stained precipitation peaks were scanned with a Hewlett-Packard desk top scanner and the areas integrated by using the NIH image program version 1.60. The values are normalized to equal number of cells (2.5 ϫ 10 4 ) and given in arbitrary immunoelectrophoretic units. Lower detection limit of the assay is Ͻ1 unit. Quantitation by this method is highly reproducible and is linear up to 70 units. Medium samples yielding values Ͼ50 units were routinely reanalyzed using 5-fold serial dilutions.

Generating Hepatoma Cells Stably Expressing OB-R L -Our
previous studies on transiently transfected rat and human hepatoma cells suggested that OB-R L but not OB-R S mediated gene induction similar to that of IL-6 type cytokine receptors (10,15). Since treatment of non-transfected cells with leptin failed to elicit an IL-6-like response, we assumed these cells did not express biologically relevant amounts of signaling-component OB-R L . Therefore, we constructed a retroviral vector in which the MSCV long terminal repeat controlled the expression of the human OB-R L (18,19). Recombinant retroviruses based on MSCV have previously been shown to efficiently transduce hepatoma cells and to direct long-term stable expression of exogenous genes (25).
Recombinant vector supernatants were applied to H-35 and HepG2 cell monolayers. Approximately 10 -20% of the cells in the vector-treated cultures of each line gave rise to vigorously growing G418-resistant cell populations, termed H-35(OB-R) and HepG2(OB-R). Northern analysis of pooled clones and individual clones (examples in Fig. 1, B and C) indicated that each contained the expected ϳ8-kb full-length human OB-Rneo fusion mRNA. All clonal lines showed two additional, but less prominent mRNA forms at ϳ6 and ϳ4 kb. Human OB-R cDNA faintly hybridized to parental HepG2 cell RNA at ϳ6 and ϳ4 kb (Fig. 1C, marked with stars), but did not hybridize detectably to rodent mRNA. However, cDNA to mouse OB-R L reacted with RNA from parental H-35 cells that appeared on Northern blots as not well resolved mRNA species in the range of ϳ4 to ϳ6 kb (Fig. 1A). RNA from adult male mouse liver showed a similar ill resolved fraction at ϳ4-ϳ7 kb but also contained a prominent form at ϳ2 kb. This short form was not observed in any of the tested rodent and human hepatoma cells and lines (Fig. 1, and data not shown).
Screening of the G418-selected hepatoma clones for leptininducible cell response revealed an interesting distinction between H-35 and HepG2 cells. All H-35 clones, regardless of the relative OB-R mRNA level, exhibited evidence of leptin-mediated activation of STAT proteins and induction of APP genes (see below). In contrast, essentially all HepG2 cells were nonresponding with the exception of those clones with highest OB-R mRNA signal (e.g. clone 7 in Fig. 1C) which showed traces of STAT protein regulation and APP gene induction (see Fig. 9 below). Therefore, for further characterization of OB-R signaling, we focussed our attention first on H-35(OB-R) clones.
Leptin Binding-RNA analysis indicated the individual H-35(OB-R) clones contained different levels of OB-R mRNA ranging from high (clone 13) to low (clone 18) (Fig. 1B). To correlate mRNA level with receptor expression, we determined 125 I-leptin binding activities in a representative set of clones ( Fig. 2; Table I). The parental H-35 cells showed a leptin binding activity with a K d of 0.7 nM that was comparable to that identified for the cloned OB-R (6). The number of ϳ5000 binding sites per cell was strikingly similar to the number determined by Cohen et al. (14) for HepG2 cells. K d values for leptin binding to the different H-35(OB-R) clones were similar to those for the parental line. However, the number of binding sites was significantly higher in those clones in which a prominent OB-R mRNA signal could be detected. The maximal level of ϳ9000 leptin-binding sites was measured for clone 13 suggesting that in this clone ϳ4000 binding sites represent the OB-R L . The assay also indicated that due to the high background of leptin binding to H-35 cells, the contribution of OB-R L to the overall number of binding sites could not be detected in clones with low OB-R mRNA expression such as in H-35(OB-R) clone 18 (Table I).
Leptin Regulates STAT Proteins-Signaling by OB-R in the H-35 cell clones was detectable by the activation of the SIE binding activity of STAT1 and STAT3 (Fig. 3A). The STAT response of the individual clones was roughly proportional to the level of human OB-R mRNA or leptin binding, although deviations were also noted (see low response of clone 1 in Fig.  3A but high leptin binding in Table I). Clonal variations were also evident in the STAT activation by LIF and IL-6. The lower responses of the clones than the parental cells were not a consequence of selection for OB-R expression since similar clonal phenotypes were detected in lines selected for ectopic expression of other gene products (data not shown).
OB-R signaling in transiently transfected hepatoma cells indicated that OB-R L has capability to activate STAT5, as   detected by binding to TB-2 (10,15). Analysis of the H-35(OB-R) clones showed that leptin induced TB-2 binding activity only in clone 13. However, as illustrated in Fig. 3B, this TB-2 binding activity was, based on antibody supershift analysis, attributable entirely to a STAT1/STAT 3 heterodimer (STAT3 homodimer does not bind to TB-2). The treatment of the same cells with growth hormone resulted in STAT5 activation identifiable by the supershift technique (Fig. 3B). Thus, these data suggest that signaling by OB-R L primarily involves STAT1 and STAT3, and STAT5 only when this factor is present at relatively high concentrations.
Induction of APP Genes by Leptin-Hepatoma cells respond to IL-6 treatment not only by activation of STAT1 and STAT3 but also by induction of APP genes. Since leptin activation of STAT proteins was qualitatively similar to that by IL-6 ( Fig. 3), we wanted to determine if this activity also extended to the regulation of APP genes. In the example of H-35(OB-R) clone 13, the time courses of leptin and IL-6 effects were similar for STAT activities and for induction of the mRNA for the representative APPs, hemopexin, and thiostatin (Fig. 4). The immediate rise and subsequent fall in STAT1 and STAT3 activity followed by a resumption of elevated STAT3 activity is characteristic for the action of IL-6-type cytokines on liver cells (25,28,34). Another similar characteristic was the temporally de-layed accumulation of the APP mRNAs (35). These results also show that factor-induced APP mRNA levels were not strictly correlated with the STAT protein activity in the same cells.
Stimulation of APP expression provides a simple assay for determining leptin action whereby the amounts of immunodetectable hemopexin and thiostatin in the culture medium represented a test for OB-R L function (Fig. 5). Whereas leptin was ineffective on parental H-35 cells, the H-35(OB-R) clones produced a dose-dependent increase of APP. The magnitude of this stimulation was synergistically enhanced by dexamethasone, a regulatory process previously described for IL-6 (16, 35). Moreover, the leptin response appeared to be correlated with the level of OB-R expression (Table I). Approximately 1 ng/ml (or 0.06 nM) leptin was required to attain half-maximal APP production in the presence of dexamethasone, a value one-tenth the K d of leptin binding (Fig. 2). However, in the absence of dexamethasone, a severalfold higher dose of leptin was needed to reach half-maximal stimulation. An identical dexamethasonedependent shift in sensitivity was detected with IL-6 (Fig. 5).
The activity of IL-6 type cytokines is characterized not only by the induction of type 2 APP genes, such as thiostatin, but also by the synergism with IL-1 or TNF␣ on type 1 APP genes, such as ␣ 1 -acid glycoprotein (36). Experiments were therefore performed to determine if the same synergistic regulatory ac- tivity could also be detected for leptin (Fig. 6). Although leptin alone, like IL-6 (36), was unable to induce the ␣ 1 -acid glycoprotein gene, it enhanced the IL-1-or TNF␣-stimulated expression of this gene 3-5-fold. Half-maximal stimulation under these conditions required only 0.1 ng/ml leptin.
Comparison of Leptin and Insulin Action-Cohen et al. (14) have reported that leptin opposes in part the action of insulin in normal HepG2 and H4-II-E rat hepatoma cells. Since both H-35 cells and H4-II-E cells are highly responsive to insulin and show considerable cross-talk between signaling by insulin and IL-6 (37), we reasoned that the parental and established H-35(OB-R) clones were exquisitely suited to assess the precise contribution of endogenous leptin binding activity and of the introduced OB-R L .
Parental H-35 cells, as well as H-35(OB-R) clones 13 and 18, responded to insulin but not to leptin by prominent phosphorylation of insulin receptor ␤ subunit and insulin receptor substrates (IRS) (Fig. 7A, upper panel). These experiments also show that both IRS-1 and IRS-2 are abundant in these cells (Fig. 7A, bottom panel) and that each undergoes a prominent electrophoretic shift toward larger size following insulin treatment, probably due to extensive phosphorylation (38) (Fig. 7, A  and B). Although no significant changes in the patterns of the tyrosine-phosphorylated cell proteins were evident following leptin treatment (Fig. 7A, upper panel), reaction of the Western blots with antiphosphotyrosine-STAT3 revealed the expected leptin effect on STAT3 in the H-35(OB-R) clones but not the parental cells (Fig. 7A, center panel). Thus, activity of leptin appears to correlate well with both the relative binding activity of OB-R (Table I) and leptin effect on STAT protein (Fig. 3).
Immunoprecipitation of IRS-1 or IRS-2 from control or leptin-treated cell lysates with their respective antibodies and immunoblot analysis with antiphosphotyrosine did not detect any phosphorylated IRS isoform in parental or OB-R L -expressing cells. In contrast, both insulin and insulin plus leptin treatments yielded a similarly strong tyrosine phosphorylation of IRS-1 and IRS-2. Similar results were obtained when we used antiphosphotyrosine antibodies for immunoprecipitation and anti-IRS antibodies for immunoblotting (example IRS-1, Fig.  7B, second lowest panel).
The anti-IRS-1 and anti-IRS-2 immunoprecipitates were also analyzed for coprecipitated PI3K. Precipitates from control and leptin-treated parental H-35 cells did not show any detectable PI3K signals, whereas those from insulin-or insulin plus leptin-treated cells were clearly positive. In contrast, the IRS-2 precipitates from leptin-treated H-35(OB-R) clones (e.g. clone 13 in Fig. 7B, data for others not shown) produced a detectable PI3K signal suggesting that leptin, through the action of OB-R L , induced an association of PI3K with IRS-2.
To ascertain whether leptin-responsiveness of each H-35(OB-R) clone was stably maintained, each clonal line was subjected to subcloning after 8 weeks in continuous culture. Twenty-four randomly selected clones were then analyzed. Each clone exhibited regulatory responses similar to those of the initial clones (e.g. see subclone 19 of H-35(OB-R) clone in 13; Fig. 8, below).
Leptin Effect on PI3K Is Shared by LIF-The leptin-stimulated recruitment of PI3K to IRS was reproducibly observed in OB-R L cells but never in parental cells. This OB-R L action probably involves low level phosphorylation of IRS-1 and IRS-2 (undetectable in our assay system in Fig. 7), thus providing a binding site for PI3K. Recent reports have demonstrated that treatment of cells with LIF (39), but not IL-6 (37) can activate IRS-1 or IRS-2 with subsequent recruitment of PI3K. Therefore, it is not unexpected that OB-R L , that is structurally and functionally related to LIFR, will similarly act on IRS. Indeed, treatment of H-35 cells with LIF promotes an association of PI3K with IRS-1 and IRS-2 (Fig. 8, left panel). The LIF effect, like the leptin effect, was more prominently manifested through IRS-2. Treatment of H-35(OB-R) clones with LIF (e.g. subclone 19 of clone 13; Fig. 8, right panel) caused similar recruitment of PI3K to IRS-2. Surprisingly, LIF was more effective than leptin at P13K recruitment to IRS-2 even though these cytokines had inverse effects on STAT proteins (Fig. 3A).
Leptin Response of HepG2 Cells-The parental HepG2 cells used for our OB-R studies bound leptin with an apparent K d of ϳ1 nM (Table I), a value similar to that determined by Cohen et al. (14). However, the number of leptin-binding sites per cell (ϳ1300) that we measured was approximately 3 to 4 times lower. Clone 7 of the OB-R-transfected HepG2 cells displayed the highest OB-R mRNA signal (Fig. 1C) but showed only minimally increased leptin binding activity (Table I). These cells responded to leptin by a minor activation of STAT3 that was clearly detectable by EMSA (Fig. 9A) but barely visible by antiphosphotyrosine STAT3 Western (Fig. 9C). Moreover, leptin treatment produced a 2-fold induction of haptoglobin and antichymotrypsin, as compared with the 26-and 18-fold stimulation, mediated by IL-6 (data not shown). In the HepG2(OB-R) clone 7, and in the parental HepG2 line, leptin failed to reduce the insulin-stimulated phosphorylation of IRS-1 or to enhance recruitment of PI3K to IRS-1 (Fig. 9B). Furthermore, leptin was unable to produce a detectable PI3K recruitment through IRS-2 (Fig. 9C), a response OB-R L prominently achieved in H-35 cells (Fig. 7B). Treatment of the HepG2(OB-R) with OSM, as expected, caused marked activation of STAT proteins (Fig. 9A), prominent phosphorylation of STAT3 (Fig. 9C, upper panel), and recruitment of PI3K to IRS-2 (Fig. 9C, lower panel). DISCUSSION In this study, we have described signaling by OB-R L that is stably expressed in cultured hepatoma cells. Our cell system yielded levels of OB-R L mRNA that exceeded the total OB-R mRNA signal detectable in normal liver as well as in the parental hepatoma cells (Fig. 1). The relatively high leptin binding by hepatoma cells (Table I, Ref. 14) has been tentatively attributed to OB-R S (9,14). If OB-R signals as homooligomer-receptor complex, as recently suggested (15), the possibility exists that the endogenous hepatoma OB-R forms may act in a transdominant negative manner. The H-35(OB-R) clones with different levels of OB-R L expression and leptin binding activities (Table I) permitted us to assess the presumed inhibitory influence of the endogenous receptors. The unusually low OB-R signaling in HepG2 cells (Fig. 9) may in part be due to such an inhibition. However, H-35 cells which have a higher level of endogenous leptin binding activity than HepG2 cells (Table I), still show a prominent signaling introduced by OB-R L even when the expression of OB-R L was relatively low (e.g. clone 18 in Figs. 3 and 5). These results suggest that the endogenous leptin binding activity may not competitively associate in the same manner with exogenous OB-R L in the two cell lines.
Still uncertain is whether expression of OB-R L in liver is physiologically significant and whether our cell system reproduces biologically relevant conditions. The study of Vaisse et al. (12) have shown that ob/ob, but not db/db, mice injected with leptin respond by prominent activation of STAT3 in the hypothalamus, attesting to the action of the OB-R L in this tissue. In contrast, the analysis of several peripheral organs of ob/ob mice failed to show an appreciable STAT3 response. Nevertheless, the liver appeared to show a minor increase of STAT3 activity (Fig. 1C in Ref. 12). Considering that in the liver similar STAT activation is mediated by various inflammatory cytokines, some of which could conceivably arise during treatment, the assessment of the leptin response by in vivo measurement of STAT3 induction may be somewhat ambiguous. Ex vivo studies of primary liver cell cultures are even more prone to artifacts (40). The use of established hepatoma cell lines alleviates many complications associated with primary liver cell cultures, such as heterogeneity of the cell population, deregulated expression of cytokines, and unstable phenotype of the parenchymal liver cells (40). Yet, the deviation of the hepatoma phenotype from the phenotype of normal liver cells, e.g. expression of OB-R isoforms, must be acknowledged and considered when using hepatoma cells as an experimental system.
The results of our studies have provided important informa- tion regarding OB-R L . This receptor proved to be functionally equivalent to IL-6 type cytokine receptors. This functional similarity was first suggested by the data from transient transfection experiments (10, 15) but has now been confirmed with stable OB-R L transductants. OB-R L prominently activates STAT3 (Fig. 3) and STAT3 in turn has been defined by us and others as one of the mediators of induction of APP genes (30,34,(41)(42)(43). The finding that leptin triggers a similar STAT3 activation in the hypothalamus (14) suggests that in these cells STAT3 may also act on STAT3-sensitive gene(s). However, whether in hypothalamic cells the interaction of STAT3 with hypothalamic gene sequences induces altered patterns of transcription remains to be shown.
The prominent induction of APP genes in H-35(OB-R) clones (Fig. 4) provides a simple and sensitive bioassay for leptin action. Data in Fig. 5 indicate that the endogenous leptin binding activity does not significantly change the ligand sensitivity of the OB-R L , since the half-maximal stimulation was achieved with approximately the same leptin concentration independent of the level of receptor expression. The leptin concentrations utilized ranged from 0.1 to 10 ng/ml depending upon the type of treatment and reporter gene product analyzed (Figs. 5 and 6). Importantly, these values are well within the range of leptin concentrations detected in the circulation of normal lean individuals and much below those of obese individuals (44,45). Thus, we would expect that signaling through OB-R L in peripheral organs, including liver, would either be constitutive or desensitized. The much lower leptin concentration reported for cerebrospinal fluid (45) suggests that it might be more suitable for the regulated action of OB-R L assuming that the receptor in hypothalamic cells displays the same dose responsiveness as the receptor in hepatoma cells.
Our studies of nontransfected hepatoma cells (Figs. 8 and 9) could not reproduce the leptin response reported by Cohen et al. (14) for HepG2 cells. However, these authors have described an action of leptin in their HepG2 cell cultures that was strikingly similar to the one we observed in H-35 cells which were engineered to express OB-R L (Fig. 8). The leptin binding activities determined in our parental hepatoma cell cultures were commarker were reacted with anti-PI3K. The antiphosphotyrosine immunoprecipitates were reacted with anti-IRS-1 (fourth panel). To demonstrate leptin and insulin action on STAT3 phosphorylation, aliquots of the original cell lysates were analyzed on Western blot with antiphosphotyrosine-STAT3 (bottom panel). parable to the values for HepG2 cells determined by Cohen et al. (14), but these binding activities proved insufficient to elicit a response. We believe this suggests that the endogenous, presumably OB-R S , in our cell lines are signaling incompetent and that those in the HepG2 cell line used by Cohen et al. (14) may include OB-R isoform(s) with signaling capabilities equivalent to the OB-R L . Analysis of the leptin effect on STAT protein activities and APP expression in those HepG2 cells would provide more evidence in support of this hypothesis.
Our results showing that signaling by OB-R L is similar to that of IL-6 type cytokine receptors would also explain the data obtained for IRS-1 and IRS-2. Several studies have found that hematopoietin receptors, including LIFR/OSMR, are able to increase tyrosine phosphorylation of IRS (38,39,46), probably through JAKs (39,47) and that this modification likely creates a docking site for PI3K (38,48). Since our studies could not detect a leptin-mediated reduction of insulin action on IRS phosphorylation, the cellular mechanism by which leptin exerts its anti-insulin effects in the HepG2 cell system of Cohen et al. (14) remains to be defined.
In conclusion, our work demonstrates that signaling functions could only be detected for OB-R L . These activities appear to be comparable to those IL-6-type cytokine receptors. These experimental culture systems provide possible insight into the OB-R signaling pathway(s). This information may help guide the analyses of leptin action in other cell types, including the identification of precise intracellular signaling pathway activated by OB-R L in hypothalamic cells important for regulating mammalian body weight homeostasis.