We have previously employed RINm5F cells ectopically expressing LEPRb to characterise leptin signal transduction in an insulin-producing cell line 18. In order to identify genes regulated by leptin in pancreatic β-cells, we compared gene expression in leptin-stimulated and non-stimulated RINm5F cells by differential hybridisation of high-density filter arrays containing 27,648 non-redundant rat cDNAs. From 33 transcripts that showed a strong differential signal after 16 h of leptin treatment, we selected 13 for further expression analysis by Northern blot hybridisation. Table 1 lists 8 transcripts that were confirmed to be upregulated by leptin. Five transcripts were identified as false positives (see Additional file 1 for the complete list).

In this study we focus on those genes potentially related to inflammatory processes. Four transcripts encode secreted acute phase proteins: fibrinogen-β, tissue-type plasminogen activator (tPA), pancreatitis-associated protein (PAP1), and lipocalin-2 20. Preprotachykinin-1 is the precursor of substance P, a mediator of neurogenic inflammation 21. In addition, the mitochondrial enzyme manganese-dependent superoxide dismutase (MnSOD) has also been classified as an "acute phase protein" because it is upregulated in many inflammatory conditions 22.

Considering that interleukin-1β (IL-1β) is an important inflammatory mediator in pancreatic β-cells, we asked whether leptin and IL-1β had overlapping effects on the expression of inflammation-related genes. Therefore, RINm5F cells were stimulated with IL-1β, leptin, or a combination of leptin and IL-1β for different time periods, and mRNA levels of specific transcripts were analysed by Northern blotting (Figure 1). As controls, we used probes for iNOS, which is known to be induced by IL-1β, and ribosomal protein L4 (RPL4) as a loading control.

According to the different time courses of leptin-induced upregulation, Northern blot analysis allowed us to distinguish immediate early (fibrinogen-β), delayed early (PAP1) and late target genes (lipocalin-2, preprotachykinin, tPA, MnSOD, and phosphatidate phosphohydrolase 2a/PPAP2A), which had their expression maxima at 1–3 h, 7 h and 16 h of leptin stimulation, respectively. Concerning the effect of IL-1β, mRNA levels for fibrinogen-β, tPA, lipocalin2, MnSOD and PPAP2A were found to be upregulated by IL-1β alone. Among them, lipocalin2 and MnSOD were more strongly induced by IL1β than by leptin. As expected, iNOS mRNA was induced by IL-1β but not by leptin. Costimulation with IL1β and leptin revealed synergistic effects of both cytokines on the expression of fibrinogen-β, lipocalin-2 and MnSOD, i.e. the effect of costimulation exceeded the added effects of each cytokine.

It is well known that IL-1β and IFNγ synergistically regulate inflammation-related genes in pancreatic β-cells 23. Since both leptin and IFNγ modulate gene expression by activating the JAK/STAT pathway, we compared the effects of these cytokines on the activation of STAT factors in the RINm5F cell line. As we have previously reported 16, leptin induced tyrosyl phosphorylation of both STAT3 and STAT1 (Figure 2). In contrast, IFNγ rather selectively activated STAT1. Although in this direct comparison IFNγ elicited a stronger tyrosyl phosphorylation of STAT1, this result hints to the possibility that both cytokines have partially overlapping effects on gene expression in RINm5F cells.

Next we analysed the effects of leptin and IFNγ on the expression of the inflammation-related transcripts identified in the present study. RINm5F cells were stimulated with leptin or IFNγ either alone or in combination with IL-1β, and transcript levels were determined by Northern blot hybridisation. A probe for IFNγ-regulated factor 1 (IRF1) was included as a positive control for the action of IFNγ. As shown in Figure 3, IFNγ alone did not upregulate the transcripts that were induced by leptin. On the other hand, IRF1 mRNA levels were not affected by leptin. These results show that leptin and IFNγ differentially regulate the expression of inflammation-related genes in RINm5F insulinoma cells, probably because the cytokines act via different STAT factors. Interestingly, similar to leptin, IFNγ enhanced the IL-1β-induced upregulation of the mRNAs for tPA, lipocalin-2 and MnSOD.

Next, we asked which signalling pathway might be involved in the induction of the inflammation-related genes by leptin. To determine the role of the three intracellular tyrosine residues (Tyr985, Tyr1077, Tyr1138) in LEPRb-mediated activation of gene expression, we generated RINm5F cell lines that stably express LEPRb point mutants in which phenylalanines replace two of the three tyrosines, or the triple mutant. Leptin binding assays verified that all LEPRb constructs were expressed at the cell surface and differed by less than 2.2-fold in leptin binding (surface binding sites of mutant receptors relative to the wild type receptor: FFY, 1.10 ± 0.05; FYF 1.4 ± 0.06; YFF, 2.01 ± 0.13; FFF, 2.13 ± 0.47; for designations of the LEPRb mutants see Figure 4).

To characterise the signalling potential of these cells, we analyzed the leptin-induced phosphorylation of STAT factors and of ERK1/2 by Western blotting with activation state-specific antibodies (Figure 4). The results show that the full activation of STAT1 and STAT3 required the presence of Tyr1138, whereas the phosphorylation of STAT5 was mediated equally well by Tyr1077 and Tyr1138. The major part of ERK phosphorylation was dependent on the membrane-proximal tyrosine residue (Tyr985). These findings are in agreement with results that we have previously obtained in transiently transfected HIT-T15 insulinoma cells 18. Intriguingly, a residual phosphorylation of STAT factors and of ERK1/2 was detectable even when all intracellular LEPRb tyrosine residues had been exchanged for phenylalanine. This cannot be explained by the action of endogenous LEPRb since neither the parental RINm5F cells nor control cells that were infected with the empty vector showed detectable effects of leptin (data not shown).

These cell lines were now used to study the capacity of the LEPRb point mutants to upregulate mRNA levels of the leptin target genes identified in this work. As shown in Figure 5, induction by leptin depended on the presence of Tyr1138, suggesting that the transcription of these genes is regulated by STAT3 and/or STAT1. Consistent with the tyrosine-independent activation of STAT3 detected by Western blotting (Figure 4), weak leptin effects were also observed in the cell lines expressing LEPRb-YFF or LEPRb-FYF. However, the degree of induction by these constructs did not exceed that of the triple mutant (FFF). Interestingly, transcript levels were always higher in cells expressing LEPRb-FFY compared to wild type LEPRb, consistent with a role for Tyr985 and Tyr1077 in the negative regulation of leptin signalling 2428.

To substantiate the hypothesis that leptin controls the expression of inflammation-related genes by way of transcriptional regulation, we decided to determine the effect of leptin on promoter activity. We selected two rat genes whose promoters had not previously been analysed for regulation by STAT factors, Lcn2 (encoding lipocalin-2) and Tac1 (preprotachykinin-1). As in our previous studies 18, we used HIT-T15 hamster insulinoma cells for the promoter assays because RINm5F cells were only poorly transfected in transient assays. The results of the luciferase assays are presented in Figure 6 and show clearly that both the Lcn2 and the Tac1 promoter are regulated by leptin. Consistent with the Northern blot results, promoter induction depended on the presence of Tyr1138 in LEPRb, supporting the hypothesis that STAT1 and/or STAT3 are involved. Also in agreement with the Northern blot data, Lcn2 promoter activity was very low in the absence of leptin and was strongly induced by leptin, whereas the Tac1 promoter showed considerable basal activity in the absence of a stimulus. These data indicate that leptin upregulates expression of lipocalin-2 and preprotachykinin-1 by control of transcription.

To address the question whether the regulation of these promoters by leptin was a specific feature of insulinoma cells, we took advantage of a LEPRb-expressing PC-12 pheochromocytoma cell line that was established in a previous study 25. In this cell line, the effect of leptin on target gene promoters was found to be enhanced by simultaneous treatment with forskolin. Forskolin is a nonspecific activator of adenylate cyclases and enhances intracellular levels of cAMP. Similar to the results obtained in HIT-T15 cells transfected with wild type LEPRb, leptin induced both the Tac1 and the Lcn2 promoter in the PC-12 cell line (Figure 7). The weaker effect of leptin as compared to the HIT-T15 cells is likely due to the fact that the PC-12 cells were not kept serum-free during treatment with leptin and/or forskolin, in order to avoid differentiation of the cells. Treatment with forskolin further enhanced the effect of leptin on both promoters, as previously observed for other leptin-regulated promoters in this cell line (proopiomelanocortin, neuropeptide Y and PAP 26).

To assess a possible cytotoxic effect of leptin on insulin-producing cells, we determined the effect of leptin treatment on the enzymatic activity of caspase 3 (and caspase 3-like proteases) as a key player in apoptosis induction. Exposure of RINm5F-LEPRb cells to leptin led to caspase 3 activation in a dose-dependent manner (Figure 8). Co-administration of leptin with IL-1β revealed an additive effect of the cytokines. Leptin did not alter caspase activity in untransfected RINm5F cells (data not shown), indicating that the effect depended on the presence of LEPRb. As a positive control, the cells were also exposed to a mix of IL-1β and IFNγ which is known to be proapoptotic for RINm5F cells 27. In this direct comparison, the combination of IL-1β (50 u/ml) and IFNγ (50 ng/ml) resulted in a stronger induction of caspase activity than IL-1β (50 u/ml) plus leptin (100 ng/ml).

Rat RINm5F insulinoma cells were grown in RPMI 1640 medium with L-glutamine, 10% (v/v) fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin at 37°C and 5% CO₂. HIT-T15 hamster insulinoma cells were cultivated under the same conditions except that 5% horse serum was added to the medium. Rat PC-12 pheochromocytoma cells stably expressing wild type LEPRb 25 were grown in DMEM supplemented with 5% fetal calf serum and 10% horse serum on collagen-coated culture dishes. To obtain single cells for plating, cells were passed through a syringe with a 22-gauge needle. HIT-T15 and PC-12 cells were transiently transfected by the polyethylenimine method (Jet-PEI reagent; Polyplus-Transfection, Illkirch, France). The RINm5F cell line stably expressing wild type LEPRb has been described previously 18. For stable expression of LEPRb tyrosine/phenylalanine mutants, the mutant cDNAs were subcloned from the pMET7 vector constructs 28 into the retroviral vector pWZLNeo. These mutant constructs and the corresponding wild type receptor used in the experiments shown in Figures 4, 5, 6 contain a C-terminal myc-tag. Confluent RINm5F cells were infected with pWZL-Neo viruses, and infected cells were selected with geneticine sulphate (G418; 1 mg/ml) as described before 18. Cell surface expression of the different LEPRb constructs on RINm5F cells was measured using a mouse leptin-secreted alkaline phosphatase chimeric protein as described 28. Murine leptin was purchased from PeproTech (London, UK), human IL-1β from Roche Diagnostics (Mannheim, Germany), and rat IFNγ from Serotec (Düsseldorf, Germany).

RINm5F-LepRb cells grown to 80% confluence were incubated in serum-free medium for 20 h before they were stimulated with murine leptin (100 ng/ml; PeproTech, London, UK) or left unstimulated for 16 h. Total RNA was isolated using the RNeasy Midi Kit (Qiagen, Hilden, Germany), and poly-A⁺-RNA was prepared using oligo-(dT)₃₀-coupled paramagnetic beads (Chemagen, Baesweiler, Germany). Complex cDNA probe was generated from 500 ng of poly-A⁺-RNA by reverse transcription in the presence of 7.5 MBq [α-³³P]-dCTP (111 TBq/mmol Hartmann Analytics, Braunschweig, Germany) according to the protocol suggested by the manufacturer of the cDNA array (RZPD Berlin, Germany). The ³³P-labelled single-stranded cDNA was hybridised to high density filter arrays that contained double-spotted, PCR-amplified cDNA inserts of the rat UniGene cDNA library (library 953, RZPD Berlin). This library comprises a set of 27,648 non-redundant rat cDNA clones provided by Bento Soares (University of Iowa). The filter were washed at high stringency (final wash in 150 mM NaCl/15 mM Na-citrate pH 7.0/0.1% SDS at 65°C) and exposed for 72 hours to a phosphor storage screen (Fujifilm Imaging Plate BAS-MS2325, Fuji Photo Film, Japan). For probe removal, the filter was doused with boiling 0.1% SDS, 5 mM phosphate buffer (ph 7.6) and left at room temperature under continuous agitation. This procedure was repeated and probe removal confirmed by exposure of an phosphor storage screen before rehybridization.

Signal intensities were evaluated using the AIDA™ software (AIDA Advanced Image Data Analyzer and AIDA Array Compare V 3.0; Raytest, Straubenhardt, Germany) as follows: integrated intensity values were determined for each spot and means of the two spots containing the same cDNA probe were calculated. For background correction, the signal intensity of the space between the spots was subtracted. To normalise for global intensity differences between the data sets, intensity values were divided by the intensity value of the spots containing β-actin cDNA. mRNA levels of β-actin were confirmed by Northern blot analysis to be unaffected by leptin treatment of RINm5F cells (data not shown). Only transcripts with an absolute signal intensity ≥ 1.0 that showed a ≥ 2.5-fold increase were considered relevant. The data presented in Table 1 and and Additional file 1 were obtained by successive hybridisation of the same filter array with the two complex cDNA probes derived from one experiment. Confirmatory Northern blot hybridisations (last column of table in Additional file 1) were done with RNA samples from independent experiments. Potentially interesting transcripts were arbitrarily selected for further analysis.

Samples (10 μg) of total RNA were separated by denaturing formaldehyde electrophoresis on 1% (w/v) agarose gels and transferred by capillary blot onto positively charged nylon membranes (Hybond N+; Amersham, Freiburg, Germany). Labelled probes were generated from the inserts of cDNA clones listed in Table 1 by random oligonucleotide priming. Murine cDNA clones were used to generate the probes for iNOS (inducible nitric oxide synthase; IMAGE clone 3583251, GenBank acc. No. BE372884) and IRF-1 (interferon regulatory factor 1, IMAGE clone 3600525, GenBank acc. No. BC003821). Blots were re-hybridised with other cDNA probes after probe removal by incubation with 0.5% SDS at 100°C.

Stably transfected RINm5F cells were incubated in serum-free medium for 18–22 h before cytokines (100 ng/ml leptin, 50 u/ml IL-1β, or 25 ng/ml IFNγ) were added for 15 min. Cells were washed with phosphate buffered saline and lysed in 1% SDS, 20 mM Tris-Cl pH 7.4 in a boiling water bath for 5 min. Total cellular lysates were separated by SDS-PAGE (8% gels), blotted on-to nitrocellulose, and phosphorylation of specific proteins was analysed using the following antibodies (Cell Signaling Technology, Beverly, MA): anti-p(Y701)-STAT1, anti-p(Y705)-STAT3, anti-p(Y694/699)-STAT5A/B, and anti-pT/pY-ERK1/2. Horseradish peroxidase-labelled anti-rabbit IgG (IgG-POD from Pierce Chemical Co., Rockford, IL) was used as the second antibody. Blots were re-used after stripping the primary antibody by incubation at room temperature for 20–30 min in 2% SDS, 50mM Tris-Cl, 150 mM NaCl, pH7.4 in the presence of 100 mM 2 mercaptoethanol.

A luciferase reporter construct containing the promoter region (-865 to +447; pSUB201-LPPT) of the rat preprotachykinin A gene (Tac1) was kindly provided by John Quinn, University of Liverpool, UK 5556. A 976-bp fragment of the rat lipocalin-2 (Lcn2) promoter region (-930 to +46, the first nucleotide of the cDNA sequence with GenBank acc. No. X13296 was defined as +1) was amplified with primers rLcn2-930for (5'-ATA

GGTACC

CCATGG

HIT-T15 cells were seeded in six-well plates (3.3–4.0 × 10⁵ cells per well) and transfected with 0.5 μg of pSUB201-LPPT or pGL3- rLcn2(-930), along with 1.0 μg of the β-galactosidase reporter control plasmid pSVβ-gal (Promega) and 0.3 μg of the pMET7-LEPRb expression plasmids 28. 24 h or 48 h after transfection, the cells were stimulated with 100 ng/ml leptin for 22 hours in serum-free medium. Luciferase activities were determined from duplicate wells with the help of a commercial kit (Promega) and data were normalised to β-galactosidase activities. For assays of PC-12-L8 cells, cells from one 80 cm²-flask were seeded into 12 wells of collagen-coated six-well plates and transfected with reporter gene plasmids as above. Fresh serum-containing medium was added the day after transfection, and cells were treated with leptin (100 ng/ml) and/or forskolin (10 μM) 24 h later. To avoid effects of differentiation, care was taken that the cells never reached confluence.

RINm5F cells were seeded in a white, non-translucent 96-well plate (10,000 cells per well). The following day, cells were supplied with fresh medium and stimulated with cytokines as indicated. After incubation for 72 h, DEVD-directed protease activity was determined from triplicate wells with the help of the Caspase-Glo™ 3/7 Assay (Promega) which detects caspase activity based on the sequence-specific cleavage of a proluminescent caspase-3/7 DEVD-aminoluciferin substrate. The data were evaluated for statistical significance with the two-tailed Student's t test, and the differences were considered significant at P < 0.05.