Translational control during the acute phase response. Ferritin synthesis in response to interleukin-1.

Interleukin-1 (IL-1 beta) increases the synthesis of both heavy and light (L)-ferritin subunits when added to human hepatoma cells (HepG2) grown in culture. RNase protection and Northern blot analysis with L-ferritin probes revealed that no changes in L-ferritin mRNA levels occur after cytokine stimulation. However, the induction coincides with an increased association of the L-subunit mRNA with polyribosomes. Since the recruitment of stored ferritin mRNA onto polyribosomes is seen when iron enters the cell, the effect of IL-1 beta on iron uptake was tested and was found to be unaffected by the lymphokine. Neither transferrin receptor mRNA levels nor the number of receptors displayed on the cell surface was affected by IL-1 beta. However, the action of the cytokine on ferritin translation is inhibited by the action of the intracellular iron chelator deferoxamine. These data indicate that IL-1 beta induces ferritin gene expression by translational control of its mRNA. The pathway of induction is different from iron-dependent ferritin gene expression whereas regulation requires the background presence of cellular iron.

Interleukin-1 (IL-lj3) increases the synthesis of both heavy and light (L)-ferritin subunits when added to human hepatoma cells (HepG2) grown in culture. RNase protection and Northern blot analysis with Lferritin probes revealed that no changes in L-ferritin mRNA levels occur after cytokine stimulation. However, the induction coincides with an increased association of the L-subunit mRNA with polyribosomes. Since the recruitment of stored ferritin mRNA onto polyribosomes is seen when iron enters the cell, the effect of IL-16 on iron uptake was tested and was found to be unaffected by the lymphokine. Neither transferrin receptor mRNA levels nor the number of receptors displayed on the cell surface was affected by IL-l& However, the action of the cytokine on ferritin translation is inhibited by the action of the intracellular iron chelator deferoxamine.
These data indicate that IL-18 induces ferritin gene expression by translational control of its mRNA. The pathway of induction is different from iron-dependent ferritin gene expression whereas regulation requires the background presence of cellular iron.
Following infection, inflammation, or injury, an acute phase response (APR)' occurs involving the synthesis and release from the liver of a series of proteins (acute phase reactants) such as al-acid glycoprotein, serum amyloid A, al-antitrypsin (alAT), and complement factor B (Baumann et al., 1987;Dente et al., 1985;Frisch and Ruley, 1987;Gehring et al., 1987;Geiger et al., 1987;Morrone et al., 1988;Perlmutter et al., 1986;Sipe et al., 1985). In contrast, the output of other liver-derived proteins such as albumin and transferrin diminishes (Ramadori et al., 1985;reviewed in Dinarello, 1988 altered hepatic transcription of these genes represents an adaptive response to minimize damage during the APR. Activated macrophages invade damaged tissues and release a number of factors into the bloodstream including IL-l& This 17.4-kDa lymphokine reproduces most acute phase changes when administered to rats (Auron et al., 1984;Ramadori et al., 1985). Some of these in uiuo responses are also reproduced by the administration of recombinant IL-l@ to hepatoma cells grown in vitro (Karin et al., 1985). However, purified cytokines do not induce the production and release of all the acute phase proteins from human hepatoma cells (Morrone et al., 1988). As an example, a1AT output is unchanged in hepatoma cells stimulated by IL-l@.
Ferritin is a ubiquitous iron storage protein, the shell of which consists of a mixture of 24 heavy (H, M, 21,000) and light (L, M, 19,000) subunits (Theil, 1987). We studied the capacity of IL-l/3 to stimulate ferritin production by human hepatoma cells (HepG2) because plasma iron levels characteristically fall during the APR (Beissel, 1977). This reduction may result from an increase in liver ferritin synthesis as demonstrated in a rat model (Konijn and Hershko, 1977). Iron does increase the transcription of the L-subunit mRNA 2-3-fold in rat liver and in bullfrog red blood cells (White and Munro, 1987;Dickey et al., 1987). However, most of the ferritin induction seen in cells to which iron is administered occurs at the level of translation of both the H-and L-subunit mRNAs (Aziz and Munro, 1986;Rogers and Munro, 1987;Schull and Theil, 1982;Walden and Thach, 1986;Rouault et al., 1988). Since there is evidence that increased rat liver ferritin synthesis is also controlled at the level of translation during the APR (Konijn et al., 1981;Campbell et al., 1989), we chose HepGP cells to determine how human ferritin gene expression is regulated by the cytokine IL-lp.
We find that IL-lp induces ferritin synthesis in HepG2 cells and that the translational efficiency of the L-subunit mRNA increases in the absence of changes in the steady-state levels of its mRNA. This occurs independently of any changes in iron uptake into the cell. An increase in the ferritin content of hepatocytes would increase the iron storage capacity of the liver, and the increase in iron retention within the organ may afford a protective response during the APR (see "Discussion"; Beissel, 1977;Konijn and Hershko, 1977 (Rogers and Munro, 1987). The immunoprecipitated proteins were applied to 15% polyacrylamide gels containing 6 M urea, 0.1% SDS, 0.1 M sodium nhosnhate (DH 7.2) or onto 15% Laemmli SDS gels (Laemmli, 1970 RNA Blotting-Total RNA from the gradient (20 rg) was denatured in 50% formamide, 2.2 M formaldehyde, 20 mM MOPS, 5 mM sodium acetate, 0.5 mM EDTA (pH 7.4) at 60 "C for 10 min and then fractionated by electrophoresis on 1.5% agarose-formaldehyde gels (Rave et al., 1979). RNA was transferred from gel to nylon bond membranes (Amersham) using standard procedures (Thomas, 1980). The RNA was filter immobilized by a 3-min exposure of the filters to ultraviolet light irradiation. Hybridization-Prehybridization of the filters was carried out for 3 h in a solution consisting of 50% formamide, 50 rg of denatured salmon sperm DNA per ml, 5 X SSC, 0.1% sodium dodecyl sulfate, and 5 x Denhardt's solution.
Overnight hybridization of the filters was carried out in 20 ml of the same buffer at 42 "C with the addition of 50 ng of randomly primed '*P-labeled probes. The filters were washed twice for a total of 1 h in 2 x SSC, 0.2% sodium dodecyl sulfate at room temperature followed by two washes for a total of 1 h in 0.5 X SSC, 0.1% sodium dodecyl sulfate at 55 "C.

RNase
Protection Analysis-Quantitation of L-ferritin and transferrin receptor (TfR) mRNA levels was performed as described elsewhere (Carrazana et al., 1988 Iron Uptake-Labeling of transferrin with 59Fe has been described previously (Klausner et al., 1983 showed that ferritin synthesis increased s-fold in cells exposed to IL-l/3 for 2 and 6 h compared with an equal number (5 X 106) of control cells. Total protein synthesis increased about 25% after 6 h of IL-l@ stimulation as measured by trichloroacetic acid-precipitatable [35S]methionine within the lysates, suggesting that protein synthesis was in general only slightly affected in HepG2 cells within 6 h of IL-l@ exposure. Scanning also revealed a 20-fold increase in ferritin subunit synthesis in HepG2 cells responding to overnight exposure to ILl/3. These values were determined by combining the readings from the data displayed in Fig. IA with those from a duplicate set of labelings and immunoprecipitations. In a separate experiment, duplicate flasks of HepG2 cells were exposed to iron in the form of 1.25 pM Fe2Tf for 2.5 h. Fig. 1B shows a IO-fold increase in ferritin synthesis in ironexposed cells relative to control. No ferritin was immunoprecipitated from lysates of iron-treated HepG2 cells incubated with preimmune rabbit serum (Fig. lB), reflecting the specificity of the antiferritin antibody immunoprecipitations.
L-ferritin mRNA Levels Are Unchanged by IL-10 Stimulation of HepGZ Cells-Total RNA was extracted from control HepGP cells, from cells treated for 2, 6, and 14 h with 0.2 ng/ ml IL-lp, and from cells treated for 2 h with 2.5 PM human FelTf as an iron source. Equal aliquots of these mRNA samples (20 pg/slot) were used in a Northern blot analysis of L-ferritin mRNA by hybridization with a human L-ferritin cDNA insert probe (Fig. 2) (Dorner et al., 1985). Subsequently the filter was rehybridized with a mouse @-actin cDNA probe, and the data in Fig. 2A represent different exposures of the same filter hybridized with each probe separately. Scanning of these autofluorographs showed that the ratio of L-ferritin mRNA to actin mRNA levels within HepGP was not increased after stimulation with either iron or IL-l@ RNase protection of a predicted 380-base L-ferritin cRNA fragment by the same RNA preparations is shown in Fig. 2B. Experiments were performed with a calculated loo-fold excess of cRNA probe to the estimated levels of ferritin mRNA in a total RNA preparation. These data confirm that there is no increase in L-ferritin mRNA levels in HepG2 cells stimulated with IL-1P. Effect of IL-ID and Iron on the Polyribosome Distribution of Ferritin mRNAs-Iron increases polyribosomal association of stored cytoplasmic H-and L-ferritin mRNAs (Aziz and Munro, 1986;Rogers and Munro, 1987). In order to investigate possible changes by IL-lfi of the polyribosome distribution of ferritin mRNAs, equal numbers of HepG2 cells were exposed for 7 and 14 h or to iron as 2.5 PM Fe,Tf for 4 h. The lysates from these cells were separated through 15-50% sucrose gradients. Total polyribosome profiles, analyzed by UV absorbance, were unaffected by IL-lb stimulation of HepG2 cells. Fractions at the top of each gradient were devoid of ribosomes and were designated as RNP; those containing only 18 S ribosome subunits, as the 40 S peak (Aziz and Munro, 1986); monosome fractions, as the 80 S peak; and polyribosome fractions were at the bottom of each gradient. Northern blots were used to assess the L-ferritin mRNA distribution among the fractions. The autoradiographs were scanned, and the ferritin mRNA distributions across the gradients are shown in Fig. 3 Cytoplasmic extracts were fractionated on 15-50% sucrose gradients, and the total RNA was isolated from different fractions of the gradient by phenol-chloroform extraction. RNA from polyribosomes, monosomes (80 S), small ribosomal subunit (40 S), and fractions free of ribosomal subunits (RNP) were identified by the UV absorbance and the rRNA pattern after gel fractionation (Aziz and Munro, 1986). The samples were pooled and Northern blotted onto nylon membranes. The polyribosome distribution of both the L-subunit and @actin mRNAs was determined by hybridization of the filters with labeled probes and densitometry of the resulting autofluorograms. The mRNA detected in each fraction is expressed as a percentage of the total amount of the same mRNA present in all the fractions of the gradient. Polyribosome gradients separate from those described have been presented in abstract form (Rogers et al., 1989).
control gradient with about 15% as RNP, about 30% in the 40 S peak fractions, 30% in the SO S peak fractions, and 17% of L-ferritin mRNA associated with the polyribosomes. The L-ferritin mRNA distribution from either 7-h or 14-h IL-l& stimulated HepG2 lysates was shifted toward the polyribosome fractions. Fig. 3 shows that 50% of L-ferritin mRNA becomes associated with 80 S fractions while 30% is present in polyribosome fractions. Iron also induces increased polyribosome association.
Reprobing of the same blots with labeled H-ferritin cDNA demonstrated a distribution of heavy subunit mRNA within polyribosome gradients similar to that of the L-chain mRNA (data not shown). In contrast, actin and cvlAT (not shown) mRNAs were exclusively associated with the polyribosomes irrespective of growth conditions. These data show that IL-l/3 increases the translational efficiency of ferritin mRNAs without increasing mRNA levels.
The Rate of Iron Uptake Is Unaffected by the Presence of Inter&kin-l&-An increase in iron uptake from transferrin would indirectly induce ferritin synthesis by elevating intracellular iron levels. Therefore, the effect of IL-lp on the uptake of labeled iron from 5gFezTf into HepG2 cells at 37 "C was determined in triplicate over a 6-h time course (Fig. 4). The rate of uptake of radiolabeled iron into cells is unaltered by stimulation of HepG2 cells for up to 6 h with IL-W This indicates that IL-l/3 does not change transferrin receptor activity.
To exemplify the effect of a significant influx of iron on receptor activity, HepG2 cells were pretreated overnight with 1.25 PM FezTf. The rate of labeled iron uptake into these cells was similar to control and IL-I@-treated cells. However, the absolute levels of labeled iron accumulation into iron-treated cells was significantly reduced compared with control or IL-l&treated cells, suggesting that the influx of iron from Fe2Tf reduced the number of transferrin receptors.
To confirm this, transferrin binding to the cell surface at 4 "C was determined. There was no increase in transferrin receptors on the cell surface in HepG2 cells stimulated with IL-l@ for 6 h compared with control cells. Untreated HepG2 cells possess 35,000 receptors/cell whereas after 6 h of IL-10 stimulation the same cells express 30,000 receptors/cell. As expected, the influx of iron into the cells treated with 5 PM FezTf decreased the number to 22,900 receptors/cell.

Intracellular Iron Chelation Prevents IL-lb-induced
Ferritin Synthesis-The chelator deferoxamine binds intracellular iron, making it completely unavailable for metabolic use.  1 and 3). IL-p-induced ferritin synthesis is absent from cells grown in the presence of 100 FM deferoxamine (lane 6), whereas oiAT production is unchanged (lane 5). The synthesis of olAT was determined by immunoprecipitation from the same lysates as ferritin, but unlike ferritin it was not affected by exposure of the cells to deferoxamine, IL-l& or iron over several experiments. Cells (106) were incubated for the indicated times in triplicate with 0.4 pM 59FeZTf in DMEM containing 1% FCS. Cells were either left as control, coincubated with 5 ng of IL-l& or pretreated with 1.25 pM FezTf. The wells were washed three times in phosphate-buffered saline and solubilized in 0.5% SDS, 20 mM nitriloacetate. The intracellular 5gFe incorporated was then counted using a y-counter, and the iron uptake for each time period was calculated from the average of each of the triplicates. Two independent experiments provide similar estimates of iron influx. Cells were labeled with [""Slmethionine for 30 min after prior incubation with IL-1B and/or deferoxamine (Ofi. Equal aliquots from each lysate were immunoprecipitated with either antiferritin antibody or with antibody to cu,AT. The immunoprecipitates were separated using a 15% acrylamide gel (Laemmli, 1970), and autofluorography was performed. Lane I, (u,AT from control enhance the translational efficiency of ferritin mRNAs by changing cellular iron levels (Mullner and Kuhn, 1988;Mullner et al., 1989). However, RNase protection analysis with a vast excess of cRNA probe shows that IL-l/3 does not influence receptor mRNA levels compared with control cells within the same 14-h time period that ferritin translation is induced. The amount of RNase protection of a predicted 420-base and a 300-base TfR fragment is unchanged by RNA isolated from control or IL-l/3-stimulated HepG2 cells (Fig. 6). As expected, TfR mRNA is more abundant after chelation of intracellular iron from HepG2 cells with deferoxamine whereas irontreated cells exhibit TfR levels diminished to a half that seen in controls (Fig. 6B). The unexpected smaller 300-base TfR cRNA is probably the result of a polymorphism between HepG2 TfR mRNA and the cloned TfR (McClelland et al., 1984), which permits RNase digestion at mismatched base(s). Northern blot analysis also shows that TfR mRNA levels are unchanged in hepatoma cells after IL-l@ stimulation, and actin mRNA levels remain unchanged under the same conditions (data not shown).

DISCUSSION
Ferritin gene expression is regulated by iron through a well described translational control mechanism. Our results demonstrate that interleukin-l& a major mediator of inflammation and the APR, stimulates the synthesis of both H-and Lferritin subunits. Detailed investigation of L-subunit mRNA expression revealed that translational control mechanisms regulate L-subunit synthesis in response to IL-l@ in human hepatoma cells. Northern blot and RNase protection analyses show that L-ferritin mRNA levels in HepG2 cells are unaffected by IL-l/3 treatment. The response of ferritin synthesis to IL-lp is accompanied by a redistribution of L-ferritin mRNA toward the polyribosomes consistent with an increase in translational efficiency. This occurs within 2 h of cytokine administration and persists for at least 14 h.
Previous studies are consistent with our data. Rat liver and spleen ferritin synthesis is elevated 3-4-fold 6 h after the onset of an experimentally induced inflammatory response (Konijn and Hershko, 1977;Campbell et al., 1989). Konijn et al. (1981) suggested that increased ferritin synthesis occurs as the result of translational mechanisms since cytoplasmic extracts taken from rat liver reproduced this induction in the absence of nucleii in vitro. More recently L-subunit mRNA was shown to be recruited from mRNPs to polyribosomes in rat liver and spleen cells 12 h after a turpentine-induced inflammation (Campbell et al., 1989). The mRNAs for both H-and L-ferritins are translationally activated within the first 2 h of administering iron to human and rat hepatoma cells (Rogers and Munro, 1987), human erythroleukemia cells (K562) (Rouault et al., 1988), and mouse fibroblast cell lines (Walden and Thach, 1986). In intact animals, a similar induction by iron results in a lo-20-fold increase in liver ferritin synthesis (Aziz and Munro, 1986;Schull and Theil, 1982;White and Munro, 1988).
Since translation of ferritin mRNA is so sensitive to changes in intracellular iron levels we sought to exclude the possibility that IL-l@ acts to stimulate transferrin receptormediated iron uptake into cells. Under such circumstances IL-l@ regulation of ferritin synthesis would be indirect. However, the rate and levels of labeled iron uptake from transferrin into HepG2 cells were not increased by the presence of IL-lp. In contrast, transferrin receptor number and iron uptake were down-regulated in cells preloaded with iron (Fig.  4). Therefore, IL-lp appears not to stimulate ferritin translation by increasing the influx of exogenous iron through either an increase in transferrin receptor number or receptor cycling rate.
Transferrin receptor mRNA levels are mediated by the same iron-sensitive trans-acting factor that controls ferritin translation (Mullner and Kuhn, 1988;Mullner et al., 1989). This protein binds to a conserved 28-base sequence in the 5'untranslated region of ferritin mRNAs (iron regulatory elements, IREs (Aziz and Munro, 1987;Hentze et al., 1987;Leibold and Munro, 1988;Rouault et al., 1988) and to similar regions in the 3'-untranslated region of transferrin receptor mRNA in such a way as to regulate both ferritin translation and transferrin receptor stability in an iron-dependent fashion (Bridges and Cudkowicz, 1984;Klausner and Harford, 1989;Mattia et al., 1984;Mullner and Kuhn, 1988;Rao et al., 1986;Rudolf et al., 1985). The exact mechanism by which iron regulates the binding of this trans-acting factor to IREs is at present undetermined.
Iron may mediate direct conformational changes to the IRE-binding protein . Alternatively, it may increase hemin synthesis, which has been shown to derepress ferritin translation in vitro (Lin et al., 1990). A third possibility is that other factors may be stimulated by iron, which serves to modulate binding of the repressor to IREs.
The absence of any changes in TfR mRNA with IL-lp levels suggests that the cytokine does not redistribute intracellular iron pools in order to stimulate ferritin synthesis (Fig.  6). Such events would change transferrin receptor mRNA levels by mechanisms associated with the binding of the ironregulated factor to receptor mRNA. These data, therefore, indicate that the cytokine acts to stimulate L-ferritin translation in a manner different from iron-induced translation. The action of IL-lfi does depend on the baseline availability of iron since deferoxamine inhibits the action of the lymphokine in HepG2 cells (Fig. 5). The complete absence of iron within cells treated with chelator may serve to "lock" the iron-dependent repressor to the IRE present in the 5'-untranslated region of all ferritin mRNAs. This would prevent a ribosomal association of ferritin mRNAs in any circumstance. It remains to be seen whether IL-@ induction of Lferritin mRNA translation is triggered by intermediate signals that permit altered binding of either the IRE-binding factor or other factors to L-subunit mRNPs. In this regard IL-P may enhance expression of trans-acting factors that modulate binding of repressor to the IRE of ferritin mRNAs but not to TfR mRNA IREs or indeed binding to alternative sites on the ferritin mRNA. The steady-state levels of H-subunit mRNAs are also unchanged, and its translational efficiency is altered in response to IL-l@.' These observations are consistent with the movement of iron between the serum and major tissue storage sites reported by several other studies (Konijn and Hershko, 1977;Konijn et al., 1981;Campbell et al., 1989). Infections associated with fever cause a depression in serum iron levels of human subjects (Beissel, 1977) whereas both endotoxin and IL-l/3 also cause a depression of serum iron levels in chickens (Klasing, 1984). In addition, the depletion of iron from the serum of rats in which inflammation is induced by turpentine correlates with an increased ferritin content in the liver and spleen (Konijn and Hershko, 1977;Konijn et al., 1981;Campbell et al., 1989). Several groups have proposed that such alterations might serve to divert labile intracellular iron to storage sites thereby reducing its availability for release from tissues into the serum (Konijn and Hershko, 1977;Weinberg, 1978Weinberg, ,1985. The liver is the major iron storage tissue although a marked increase in the translation of L-ferritin mRNA also occurs in rat spleen macrophages during inflammation (Campbell et al., 1989).
Reduction in serum iron during the acute phase response *J. Rogers, K. R. Bridges, G. P. Durmowicz, J. Glass, P. E. Auron, and H. N. Munro, unpublished observations. may serve a protective role by withholding iron from the siderophores of opportunistic bacteria. A reduction in the bioavailability of iron may also provide protection against cell injury by hydroxyl radicals that are generated from macrophage-derived superoxide in the presence of serum iron (Babior, 1984;Thomas et al., 1985). Human hepatoma cells do exhibit a marked increase in the steady-state levels of ferritin shells as measured by protein staining on nondenaturing gels.' These data suggest that ferritin protein accumulates in these cells rather than being degraded at an increased rate when stimulated by IL-l@.
Tumor necrosis factor, a peptide released from mature macrophages, increases the level of H-ferritin mRNA 4-6fold after a 48-h exposure of mouse adipocytes and human muscle cells (Torti et al., 1988). This effect probably results from of enhanced transcription of the ferritin H gene. Our experiments focused on the early responses to IL-l@ and have not ruled out subsequent transcriptional responses of ferritin genes to the cytokine. These observations support our conclusions that translational control is exerted on L-ferritin synthesis early in the acute phase response.