Ferritin functions as a proinflammatory cytokine via iron-independent PKC-ζ/NFκB-regulated signalling in rat hepatic stellate cells

Hepatic Stellate Cell Isolation

Studies were performed with institutional Animal Ethics Committee approval and compliance with Australian regulatory guidelines. Rat HSC were isolated from normal male Sprague-Dawley rats (600±50g) by sequential pronase/collagenase perfusion, cultured on plastic 12-well plates and grown in Medium 199 (Invitrogen) supplemented with 10% calf + 10% horse serum, 50μg/mL ascorbic acid and penicillin/streptomycin (100U and 100μg/mL, respectively), to induce an activated phenotype (11, 12). In addition, isolated HSC were cultured on Teflon® tissue culture inserts (Millipore) to maintain cells in a quiescent phenotype (11). 24hr prior to experimentation, HSC were washed and cultured in serum-free M199 medium. HSC were then treated with either the iron-replete tissue ferritins horse spleen or rat liver ferritin (termed “ferritin”) (Sigma-Aldrich), the iron-free recombinant human H-(rHF) or L-subunit (rLF) ferritins, or with de-ironed ferritin (apoferritin). Recombinant ferritins rHF and rLF (13), and apoferritin (7, 8) were prepared as described. Five day culture-activated HSC were routinely used except where indicated.

SDS-Polyacrylamide Gel Electrophoresis

Whole cell protein extracts were prepared by lysis of HSC cultures in 1X SDS sample buffer (62.5mM Tris HCl pH6.8, 2% w/v SDS, 10% glycerol, 50mM DTT, 0.01% w/v bromophenol blue). 25μL cell protein was subjected to SDS-PAGE and western blot using specific antibodies, with detection of antigen using the ECL system (GE), as described (11). Digital images of blots were captured using the FUJIFILM LAS3000 system and analyzed using FUJIFILM Multi Gauge V3.0 software.

Extraction of RNA from HSC

Subconfluent (80-90%) rat HSC were washed with ice-cold PBS and placed on ice prior to total RNA isolation using the RNeasy mini kit (Qiagen) as per manufacturer's instructions. Total RNA was then treated with 1μL of RQ1 RNase free DNase I (1U/μL Promega) per 1 μg of total RNA for 30min at 37°C to ensure complete removal of DNA contamination. Total RNA concentration and quality was estimated by spectrophotometry at 260-280nm.

Quantitative Real-Time RT-PCR

3μg of total RNA was used to generate complementary cDNA template using an oligodT₁₅ primer and SuperScript III™ (Invitrogen). Real time PCR reactions were performed in a final volume of 15μL with forward and reverse oligonucleotide primers (Table 1) used at a final concentration of 500nM, using Platinum® SYBR® Green qPCR SuperMix-UDG (Invitrogen) and the reaction profile as recommended by the manufacturer. All PCR reaction efficiencies were between 0.95-1.10. Melt curve analysis revealed the presence of a single product with a single melt signature. PCR was performed using an RG-3000 thermal cycler (Corbett Research Australia). mRNA quantitation was achieved using the two standard curves method, with the fluorescence measurements of the unknown samples back-referenced to a standard curve relating concentration to fluorescence in arbitrary units. mRNA levels were normalized to expression of the housekeeping gene Basic Transcription Factor 3 (BTF3) and expressed relative to untreated or control samples (presented as relative mRNA expression).

Nitric Oxide Determination

Freshly isolated HSC were cultured with ferritin (100pM-100nM). At days 3, 5 and 7, culture medium was harvested and at days 3 and 5 replaced with fresh medium containing ferritin. The concentration of the stable and non-volatile breakdown product of nitric oxide, (NO₂^- nitrite) in the conditioned medium was then determined using the Griess Reagent System (Promega) as per manufactures instructions.

Immunoprecipitation

Day 5 culture-activated HSC were incubated with rHF, rLF or ferritin for 1hr. Cells were then lyzed for immunoprecipitation studies as per manufacturer's instructions (Roche). 500μL of cell lysate was pre-cleared for 3hr with 25μL of the homogeneous protein-A-agarose suspension before immunoprecipitation with the rabbit anti-phospho-YXXM PI3-kinase target motif (Cell Signaling Technology, 1:100) at 4°C overnight. Immunoprecipitate was washed and the resultant bead-antibody-protein complex resuspended in 1X SDS sample buffer and boiled for 3min. Beads were pelleted by brief centrifugation and the supernatant resolved using 9% SDS PAGE. After transfer to Nitrocellulose, blots were probed with goat anti-mouse antibody to T cell immunoglobulin-domain and mucin-domain 2 (Tim-2) (R&D Systems, 1:1250 in 5% SMP blocking buffer) overnight at 4°C. Primary antibodies were detected using a rabbit anti-goat IgG antibody conjugated to HRP (Sigma, 1:1000 in 5% SMP blocking buffer).

Electrophoretic Mobility Shift Assay (EMSA)

The consensus double stranded NFκB binding site oligonucleotide (5'-AGT TGA GGG GAC TTT CCC AGG C-3' Promega) was 5'-end-labelled with [γ-³²P]ATP (3000 Ci/mmol-1; Amersham Pharmacia Biotech) using T4 polynucleotide kinase (New England Biolabs®). Nuclear extracts were prepared as described (14) and protein content determined using the Lowry method with Peterson's modification (Sigma). EMSA was performed as described (15) with protein/DNA complexes resolved on 8% non-denaturing polyacrylamide gel (37:5:1). For supershift analysis of NFκB subunits, 2μg of anti-p50 and anti-p65 antiserum (Santa Cruz) were added after the initial oligonucleotide probe and nuclear extract reactions, and incubated at room temperature for 1hr.

Statistical Analysis

Data are presented as mean values of 3-4 separate experiments performed in duplicate ± SEM. Groups were compared using One-way ANOVA with Dunnett's multiple comparison post hoc test or the Student t-test, where applicable using Prism 4.0 (GraphPad Software). P<0.05 was considered statistically significant.

Ferritin activates NFκB in activated rat HSC

Given the role NFκB plays in mediating the inflammatory response of HSC, the potential for ferritin to influence this inflammatory response was assessed by investigating its effect on NFκB activity. Ferritin induced a significant dose-dependent increase in IKK α/β phosphorylation (Figure 1A,B). An optimal ferritin concentration of 10nM was chosen, giving maximal effects on IKK phosphorylation, with no effect on cell viability. HSC incubated with 10nM ferritin demonstrated a significant time-dependent, transient increase in IKKα/β phosphorylation with maximal expression of 4.3-fold after 120min (Figure 1C,D). The binding of nuclear extracts from HSC treated with ferritin to consensus NFκB binding sites was analysed by EMSA. NFκB DNA-binding activity was increased within 5min, peaking at 6hr, returning to basal levels by 24hr (Figure 1E-G). EMSA supershift revealed NFκB upregulated by ferritin contained both p50 and p65 subunits (data not shown).

Ferritin induces NFκB-responsive proinflammatory genes

The consequences of ferritin-induced NFκB DNA-binding activity were then assessed. Activated HSC were treated with 0-100nM ferritin for 0-240min. Ferritin induced a significant dose- (Figure 2A) and time-dependent (Figure 2B-E) increase in expression of NFκB-transcriptionally regulated genes known to play a role in mediating hepatic fibrogenesis (15-19); including ICAM1 (5.24±0.72 fold; Figure 2B), IκBα (3.54±0.73 fold; Figure 2C), iNOS (92.42±37.20 fold; Figure 2D), IL-1β (48.10±8.30 fold; Figure 2E) and RANTES (37.84±14.32 fold; results not shown) after 2-4hr. The effect of ferritin was most evident when using 5-100nM ferritin (Figure 2A). The sensitivity of activated HSC to ferritin was reflected in the EC₅₀ values (Table 2). Ferritin had no effect on the expression of HSC activation markers procollagen α₁(I) or αSMA mRNA (results not shown).

Teflon®-cultured (quiescent) HSC were incubated with ferritin (1-100nM) for 3hr. Ferritin had no effect on RANTES, ICAM1, IκBα expression and only 100nM ferritin induced IL-1β (2.5 fold; Figure 2F). αSMA and procollagen α₁(I) expression was also unaffected (data not shown).

To further confirm the role of NFκB in ferritin-induced signaling in activated HSC, IκBα phosphorylation (therefore NFκB activation) was blocked using the specific inhibitor Bay11-7082 (20), and the ferritin-induced expression of IL-1β mRNA assessed. IL-1β transcription stimulated by 10nM ferritin was completely inhibited by Bay11-7082 (Figure 2G), confirming that ferritin induced IL-1β via IκBα phosphorylation and NFκB activation.

Ferritin induces ICAM1, IκBα and Nitric Oxide production

10nM ferritin induced a significant time-dependent increase in ICAM1 protein expression in activated HSC after 24hr (1.52±0.11 fold) and 48hr (1.93±0.15 fold) (Figure 3A,B). IκBα protein levels were also increased by ferritin (Figure 3A). Nitric Oxide produced by ferritin-treated HSC was assessed as a functional consequence of iNOS mRNA induction. Ferritin induced a significant dose-dependent increase in Nitric Oxide production after 3d and 5d with maximal increases of 3.22±0.78 and 4.78±1.36 fold, respectively (Figure 3C). After 7 days Nitric Oxide production returned to basal levels (Figure 3C).

Rat HSC express Tim-2

Recent data suggests that Tim-2 acts as a receptor for H-ferritin endocytosis in B cells (21), although its role in initiating cell signaling is not known. We investigated the expression of Tim-2 in HSC. Western blot revealed two Tim-2 bands in quiescent HSC (Figure 4A). Tim-2 expression (both bands) was unchanged with time as HSC became activated (Figure 4A,B), as evidenced by the upregulation of αSMA, a marker of HSC activation (Figure 4A).

PI3-kinase is activated by ferritin

Contained within the Tim-2 amino acid sequence is a consensus PI3-kinase target phosphorylation motif (YXXM) (21). Thus, we investigated the potential role of Tim-2 and PI3-kinase in mediating the NFκB-transcriptional events triggered by ferritin. Activated HSC were incubated with 10nM ferritin for 0-240min before cell protein extracts were subjected to western blot (Figure 5A). A significant relationship between YXXM phosphorylation (i.e. all proteins with phospho-YXXM-motif, PI3-kinase target site) and ferritin exposure time was observed with maximal increase of 2.87±0.31 fold at 120min (Figure 5B). rHF and rLF induced similar increases in YXXM phosphorylation (data not shown). To assess whether Tim-2 was one of these phosphorylated proteins, we immunoprecipitated all proteins that had undergone YXXM phosphorylation in response to treatment with either ferritin, rHF or rLF, and probed for Tim-2 (Figure 5C). No difference in phospho-Tim-2 expression was observed 1hr after treatment of activated HSC with 10nM ferritin, rHF or rLF. This suggests that Tim-2, acting through PI3-kinase, is not involved in ferritin-induced signaling. To further assess the role of PI3-kinase in ferritin-induced NFκB-gene transcription, the PI3-kinase inhibitor wortmannin was used (22). Wortmannin (100nM) alone had no effect on IL-1β mRNA expression, however, wortmannin significantly inhibited the ferritin-induced expression of IL-1β by 78.4 % (Figure 5D).

Ferritin is a PKCζ activator

PKCζ, a downstream target of PI3-kinase, is intimately involved in NFκB activation (23). As As ferritin activates both PI3-kinase and NFκB, we investigated the role of PKCζ in the ferritin-induced signaling events. Ferritin induced the time-dependent phosphorylation of PKCζ (Figure 6A,B). Significant enhancement of PKCζ phosphorylation (4.19±0.24 fold) by ferritin was evident after 30min, which increased until 120min (5.78±1.0 fold). Using a pseudosubstrate inhibitor of PKCζ (100 μM) we attempted to block the effects of ferritin on IL-1β gene transcription (Figure 6C). Two variants of the PKCζ pseudosubstrate inhibitor were used; the first was myristoylated (M) to allow cell entry, the second non-myristoylated (NM; ie, non-cell permeable), serving as a control (Figure 6C). The M-PKCζ pseudosubstrate inhibitor completely abolished the effects of ferritin on IL-1β transcription; conversely, the NM-PKCζ pseudosubstrate inhibitor control had no effect. To further assess the role of ferritin as a PKC-ζ activator, activated HSC were exposed to ferritin (10fM-100nM) for up to 7d. Long-term treatment with PKC activators has been shown to cause PKC down-regulation in fibroblasts (24). Ferritin induced a dose-dependent decrease in expression of the 80kDa, or phosphorylated form of PKCζ becoming significant at 10pM (51.5±8.0% decrease), with expression undetectable using 100nM ferritin (Figure 6D). This provides further strong support for the role of ferritin as a PKC-ζ activator in HSC biology.

Ferritin induces the phosphorylation of MAP-kinase

MAP-kinase is a downstream target of PKCζ and an activator of NFκB (25), thus we investigated the role of MAP-kinase in ferritin-induced signaling. Treatment of day 5 activated HSC with ferritin (10pM-100nM) for 1hr caused a significant dose-dependent increase in MAP-kinase phosphorylation (Figure 7A,B). Maximal MAP-kinase p42/p44 phosphorylation was observed using 100nM ferritin (7.7±0.62 and 16.9±2.65 fold, respectively). Figures 7C,D demonstrate the time-dependent increase in MAP-kinase p42/p44 phosphorylation following ferritin treatment. Maximal expression was observed after 60min (2.23±0.41 and 2.83±0.35 fold, respectively), returning to basal levels after 4hr (Figure 7C,D). Similar effects were observed using rHF and rLF (data not shown). To further assess the role of MAP-kinase in ferritin-induced signalling, HSC were treated with 10nM ferritin + U0126 (10μM, 30min), a selective inhibitor of MEK1/2 (a known activator of MAP-kinase), and IL-1β expression assessed. Ferritin-induced IL-1β expression was inhibited by 70±7.5% (Figure 7E), implicating both MEK1/2 and MAP-kinase in ferritin-induced signalling.

Ferritin induces NFκB activation and IL-1β expression via an iron-independent pathway

As tissue ferritin stores iron it is possible that the NFκB-transcriptional events elicited by ferritin may be, in part, due to iron-induced oxidant stress. We used de-ironed ferritin (apoferritin) and iron-free rHF and rLF to assess their effect on NFκB DNA-binding activity and IL-1β gene transcription. Treatment of activated HSC with apoferritin, rHF or rLF caused a significant increase in NFκB DNA-binding activity (Figure 8A). rHF and rLF also induced a significant time-dependent increase in IL-1β expression (Figure 8B). rHF induced a 221.7±51.2 fold increase in IL-1β; significantly greater than rLF (69.9±17.7 fold) (Figure 8B). In addition, iron chelators of both Fe²⁺ (Dipyridyl) and Fe³⁺ (Deferoxamine) were used to block the potential effects of iron contained within tissue ferritin. Neither Dipyridyl (Figure 8C) nor Deferoxamine (Figure 8D) had any effect on IL-1β gene expression induced by ferritin. Taken together, these data provide further evidence of a ferritin-induced signalling pathway resulting in NFκB-regulated gene expression that is independent of iron.