Phosphorylation and activation of both iron regulatory proteins 1 and 2 in HL-60 cells.

Iron regulatory proteins (IRPs) are RNA-binding proteins that post-transcriptionally regulate synthesis of iron uptake (transferrin receptor) and storage (ferritin) proteins. Our previous work demonstrating that IRP1 is phosphorylated by protein kinase C supported the hypothesis that factors in addition to iron modulate IRP function. We have investigated changes in activity and expression of both IRP1 and IRP2 during phorbol 12-myristate 13-acetate (PMA)-induced differentiation of HL-60 cells. In contrast to IRP1, IRP2 was highly phosphorylated in untreated cells. PMA stimulated phosphorylation of IRP1 and IRP2 by at least 2-3-fold without affecting incorporation of [35S]methionine into the proteins. IRP1 and IRP2 isolated from PMA-treated cells displayed different phosphopeptides. Phosphorylation of IRPs was associated with a 2-fold increase in high affinity RNA binding activity without altering KD, and this was accompanied by a 50% increase in transferrin receptor mRNA abundance. PMA acted on a latent pool of binding activity that is present in a nonaconitase oxidized form and is largely composed of a stable but inactive species of IRP2. Desferal and hemin modulated iron-responsive element binding activity in HL-60 cells without affecting the phosphorylation state of IRP1. Hemin appeared to reduce the abundance of phosphorylated IRP2. Thus, multiple factors affect the function of both IRPs and indicate that extracellular agents may program changes in cellular iron metabolism by altering the phosphorylation state of these regulatory RNA-binding proteins.

Iron regulatory proteins (IRPs) are RNA-binding proteins that post-transcriptionally regulate synthesis of iron uptake (transferrin receptor) and storage (ferritin) proteins. Our previous work demonstrating that IRP1 is phosphorylated by protein kinase C supported the hypothesis that factors in addition to iron modulate IRP function. We have investigated changes in activity and expression of both IRP1 and IRP2 during phorbol 12myristate 13-acetate (PMA)-induced differentiation of HL-60 cells. In contrast to IRP1, IRP2 was highly phosphorylated in untreated cells. PMA stimulated phosphorylation of IRP1 and IRP2 by at least 2-3-fold without affecting incorporation of [ 35 S]methionine into the proteins. IRP1 and IRP2 isolated from PMA-treated cells displayed different phosphopeptides. Phosphorylation of IRPs was associated with a 2-fold increase in high affinity RNA binding activity without altering K D , and this was accompanied by a 50% increase in transferrin receptor mRNA abundance. PMA acted on a latent pool of binding activity that is present in a nonaconitase oxidized form and is largely composed of a stable but inactive species of IRP2. Desferal and hemin modulated iron-responsive element binding activity in HL-60 cells without affecting the phosphorylation state of IRP1. Hemin appeared to reduce the abundance of phosphorylated IRP2. Thus, multiple factors affect the function of both IRPs and indicate that extracellular agents may program changes in cellular iron metabolism by altering the phosphorylation state of these regulatory RNA-binding proteins.
Iron is an essential nutrient for the growth and optimal health of virtually all organisms (1)(2)(3)(4). However, excessive levels of iron can be toxic due to the increased potential for generation of reactive oxygen metabolites. Consequently, regulatory mechanisms have evolved that allow cells to maintain iron homeostasis in response to variations in the availability of and/or requirement for iron. Two proteins that contribute to the maintenance of iron homeostasis are ferritin, an intracellular iron storage protein, and transferrin receptor (TfR), 1 a membrane protein that transports iron into the cell. In nearly all cell types of higher eucaryotes, ferritin and TfR synthesis is coordinated in a divergent manner in response to the level of intracellular iron (1)(2)(3)(4). In the presence of excess intracellular iron, ferritin synthesis is elevated to accommodate the need for additional iron storage capacity, whereas TfR synthesis is diminished to minimize the potentially toxic effects of intracellular iron overload. The opposite response is exhibited in iron-depleted cells; TfR synthesis is increased, whereas ferritin synthesis is reduced.
Iron-dependent changes in ferritin and TfR synthesis are regulated primarily at the post-transcriptional level through specific regulated protein-RNA interactions (1)(2)(3)(4). The translation and stability of ferritin and TfR mRNAs, respectively, is modulated by cytosolic proteins (iron regulatory proteins (IRPs)). IRPs influence mRNA utilization by binding to specific stem-loop motifs (iron-responsive elements (IRE)) present in the 5Ј-or 3Ј-untranslated regions of ferritin and TfR mRNAs, respectively (5)(6)(7)(8). In addition, the mRNA encoding erythroid 5-aminolevulinate synthase, the rate-limiting enzyme in formation of heme, contains an IRE in its 5Ј-untranslated region and appears to be translationally regulated by iron (9 -11). Thus, IRPs are sensory and regulatory components of a homeostatic system that coordinates the synthesis of proteins involved in storage, uptake, and metabolic utilization of iron.
A fundamental question regarding the regulation of ferritin and TfR synthesis that has received much attention concerns how changes in cytoplasmic iron levels are transmitted into alterations in IRE RNA binding activity. For IRP1 much evidence has accumulated in support of the "aconitase model" as one mechanism by which iron regulates RNA binding activity of IRP1 without variation in abundance of the protein (12)(13)(14)(15). However, iron-dependent changes in degradation of IRP1 appears to represent another means for affecting function of this binding protein at least in some cell types (16). For IRP2, iron appears to largely act by influencing the steady state level of the protein since in some, but not all, iron-loaded cells the rate of degradation of IRP2 is enhanced as compared with the rate observed in iron-depleted cells (17)(18)(19). In some cell types iron appears to modulate IRP2 RNA binding activity without affecting the steady state level of the protein (18). However, the biochemical nature of this stably inactive pool of IRP2 has not been addressed.
Because IRPs are central regulators of iron utilization, they serve as likely focal points at which extracellular agents could act to influence the rate of uptake or metabolic fate of iron. Several physiological and pathological situations are associated with alterations in the uptake, storage, or utilization of iron, and it is apparent that multiple effectors can influence IRP function (20 -23). These include iron-independent increases in TfR expression and iron uptake that occur during cellular proliferation such as after lymphocyte activation (24 -26). In addition, certain cell types exhibit specialized requirements for iron during or after they differentiate. For example, when monocytes differentiate into macrophages or during erythropoiesis, stimulation of iron uptake and alteration in iron storage occur and agents other than iron appear to be the effectors (26,27). Under such circumstances alterations of IRP function through the action of extracellular agents could provide a means for inducing changes in cellular iron utilization.
We have recently begun to describe a novel mechanism by which a variety of extracellular agents may influence cellular iron metabolism as a result of phosphorylation of IRP1 (23). In HL-60 cells, PMA-induced phosphorylation of IRP1 was associated with a rapid stimulation of IRE binding activity; however, the potential role of IRP2 was not examined nor were the biochemical characteristics of the latent pool of binding activity on which PMA acted. In the present paper we describe experiments that define the forms of IRPs present in untreated HL-60 cells and the effects that PMA and changes in iron status have on these forms of the binding proteins. We show that unstimulated HL-60 cells contain a significant level of latent IRE binding activity, which is largely IRP2, that is almost exclusively in a nonaconitase oxidized form. Furthermore, we show that PMA or the iron chelator desferal act on the latent forms(s) of IRP to convert them from an inactive or perhaps low affinity state to a high affinity state with respect to RNA binding with concurrent increases in the abundance of TfR mRNA. Our results further demonstrate a direct relationship between increases in phosphorylation of both IRPs and activation of their RNA binding function by compounds that activate protein kinase C and other kinases. Our results support the hypothesis that phosphoregulation and iron regulation represent different mechanisms for affecting IRP function.

MATERIALS AND METHODS
Cell Culture-HL-60 cells were grown and induced to differentiate with PMA as described (23). Following treatment with PMA (0.2 M), freshly prepared hemin (20 M) or desferal (100 M) cells were lysed as described (23), and lysate supernatants were stored at Ϫ70°C. Lysate protein concentrations were determined by Bradford assay kit (Pierce) with bovine serum albumin as a standard. C58 rat lymphoma cells were cultured in high glucose Dulbecco's modified Eagle's medium plus 10% heat-inactivated horse serum and penicillin/streptomycin.
Gel Retardation Analysis-The IRE-containing 73-nucleotide RNA used for gel retardation assays was synthesized using [␣-32 P]UTP, and cell lysates were analyzed for RNA binding activity as described previously (23). Cell culture experiments were done independently three to five times for each treatment group, and gel retardation assays were performed in triplicate for each independent culture sample. For supershift assays, 2.8 g of lysate protein was incubated with 10 g of anti-IRP2 IgG for 10 min at 4°C. Then [ 32 P]IRE RNA and subsequently heparin were added in a manner identical to the standard gel shift assay. For supershift assays, gels were run at 190 V for 2 h at 4°C. For alkylations studies, lysates were incubated with 1 mM N-ethylmaleimide (NEM) for 30 min in ice before the addition of 2-mercaptoethanol (2-ME) where indicated.
Metabolic Labeling and Immunoprecipitation-HL-60 cells were cultured in the presence of [ 35 S]methionine or [ 32 P]orthophosphate to assess the relative rate of synthesis and the extent of phosphate incorporation into IRPs as described (23,28). Immunoprecipitations for IRP1 used rabbit polyclonal IgG against purified rat liver IRP1 (23). These antibodies did not react with IRP2 (results not shown). Polyclonal IgG against a 73-amino acid region present in IRP2 but not IRP1 were generously provided by Betty Leibold (University of Utah) (17). Densitometry values refer to the integrated density of the entire band and not just a "slice" through the band.
Phosphopeptide Mapping-IRP1 and IRP2 were separately immunoprecipitated from PMA-treated (IRP1 and IRP2) or untreated (IRP2 only) cells and separated by SDS-polyacrylamide gel electrophoresis. Radioactive bands for each protein were cut from dried gels and twodimensional tryptic phosphopeptide mapping was performed as described (29). Radioactive peptide fragments were detected by autora-diography or, in later experiments, using a PhosphorImager. We thank P. T. Tuazon and J. A. Traugh (University of California, Riverside) for performing the phosphopeptide mapping.
Analysis of TfR mRNA Abundance-Cell cultures were incubated with PMA, hemin, desferal, or Me 2 SO (diluent for PMA) as described below. After 5 h, total RNA was extracted using TRISOLV (Biotecx Laboratories, Inc.) (30) for RNase protection assays (Ambion). Total RNA samples (30 g) were hybridized overnight at 45°C with both antisense human TfR and human glyceraldehyde phosphate dehydrogenase 32 P-labeled RNA probes (500,000 cpm each). Single-stranded RNA was digested at 37°C for 30 min with RNase A (0.17 units) and T1 (0.67 units). Protected fragments were resolved using a 5% acrylamide/8 M urea gel. Human TfR cDNA, pCDTR1 (kindly provided by F. Ruddle, Yale University), in which the encoded poly(A) tail was deleted (pTfR⌬1) was used for RNase protection assays. After digestion with StyI, a 429-nucleotide 32 P-labeled antisense RNA probe was generated by in vitro transcription. Human glyceraldehyde dehydrogenase cDNA (Ambion) was used to produce a 403-nucleotide 32 P-labeled antisense RNA. Probes were purified through a 5% acrylamide/8 M urea gel. The TfR and glyceraldehyde phosphate dehydrogenase antisense RNAs protected fragments of 374 and 316 nucleotides, respectively (results not shown).

PMA Rapidly
Stimulates RNA Binding Activity of IRP1 and IRP2-To assess the role of phosphorylation in the regulation of IRP function, we examined the relationship between IRE RNA binding activity, phosphorylation state, and synthesis rate in HL-60 cells acutely treated with PMA. The addition of 0.2 M PMA to HL-60 cells rapidly stimulated IRE RNA binding activity (Fig. 1A). Within 30 min of addition of PMA, IRE RNA binding activity was stimulated by 73% (Fig. 1A). After 4 h in the presence of PMA, IRE RNA binding activity remained almost 2-fold greater than the activity present in uninduced cells. In human cells, IRP1 and IRP2 co-migrate during gel shift analysis (31). To distinguish the relative contributions of the two IRPs to the binding activity present in control and PMA-treated HL-60 cells, we performed a "supershift" assay with an antibody directed against a region found only in IRP2 (kindly provided by E. A. Leibold (17)). First, we determined the level of antibody required to supershift IRP2 activity in rodent cells where the two IRE-binding proteins are easily separated by gel shift analysis (31,32). In C58 rat lymphoma cells as little as 4 g of IgG completely eliminated the IRP2 band and resulted in the appearance of supershifted bands (Fig. 1B, compare lanes 1 and 2, note band d). IRP1 was not affected even when as much as 20 g of IgG was used (Fig. 1B, lane 3). Preimmune IgG was without effect (not shown). Second, we observed that in control HL-60 cells anti-IRP2 IgG supershifted a significant portion of the RNA binding activity (Fig. 1C, compare lanes 4 and 6). Saturating concentrations of IgG reduced total IRE binding activity (0.344 Ϯ 0.015 pmol/mg protein) by 74%, indicating that the majority of the binding activity in control cells was due to IRP2 and the remainder (26%; 0.092 Ϯ 0.011 pmol/mg protein) was due to IRP1 (Fig. 1C, lanes 4 and 6). Third, in HL-60 cells treated with PMA for 30 min, total IRE binding activity increased by 2-fold (0.696 Ϯ 0.026 pmol/mg protein), and after the addition of anti-IRP2 IgG, 81% of the binding activity was removed, with 19% (0.143 Ϯ 0.026 pmol/mg protein) not being affected by the antibody (Fig. 1C, lanes 5 and 7). Fourth, PMA treatment of HL-60 cells resulted in a 2.6-fold increase in the abundance of the supershifted species (Fig. 1C, lanes 6 and 7, band d). Taken together, PMA treatment increased IRP1 RNA binding activity by 60%, whereas IRP2 activity increased 2.2-2.6-fold.
PMA-dependent stimulation of IRE RNA binding activity could represent changes in the affinity of interaction of the binding proteins with RNA (K D ) and/or a change in the amount of protein in the high affinity state (B max ). Lysates from HL-60 cells treated with PMA for 30 min exhibited nearly a 2-fold increase in B max for total IRE binding activity compared with uninduced cells (Fig. 2) (Table I). In contrast, K D was unaffected by PMA treatment (Table I). Although a specific IRE binding activity of lower affinity (K D ) was detected by RNA saturation analysis with low specific activity RNA, it was not affected by PMA treatment of the cells. 2 Our results support the hypothesis that PMA treatment induces conversion of the IRPs, particularly IRP2, from an inactive form to an active form and/or that PMA treatment stabilizes the binding protein(s) in an active form for RNA binding.
PMA Acts on a Redox-sensitive Pool of IRE Binding Activity in HL-60 Cells-To determine if PMA affected total IRE binding activity through changes in cysteine oxidation state or through changes in abundance of the binding proteins, we treated extracts from untreated and PMA-treated cells with 2-ME. Examination of the level of 2-ME-inducible RNA binding activity present in extracts obtained from PMA-treated as well as uninduced cells indicated that the level of 2-ME-inducible IRE binding activity remained essentially unchanged throughout the 4-h time course, irrespective of the presence or the absence of PMA (Fig. 3). Furthermore, we observed that the level of activation by PMA treatment of cells was similar to the level achieved in vitro by treating extracts of untreated cells with 2% 2-ME (compare Figs. 1A and 3). Thus, PMA treatment resulted in an increase in the percentage of IRPs that bound RNA with high affinity from 50 to 93% of the total 2-MEinducible activity and a concomitant decrease in the percentage of binding protein that was in the inactive pool (from 50 to 7%). Taken together with our supershift results, it appears that in untreated cells IRP1 and particularly IRP2 are present in inactive form(s) apparently due, at least in part, to the presence of oxidized cysteines in the proteins.
The Redox-sensitive Pool of IRE Binding Activity in HL-60 Cells Is in a Nonaconitase Form-Several studies have used high levels (ϳ300 mM) of 2-ME to convert holo-IRP1 (cytosolic aconitase) to a high affinity state for RNA binding as a means for measuring the total amount of IRP1 in cell extracts (12,14,15). Substrates for aconitase such as citrate completely block the action of 2-ME on holo-IRP1. In the case of IRP2, the extent to which an inactive form of the protein stably exists in cells is unclear because iron accelerates turnover of the protein; how-ever, in some cells it does appear that an inactive form of IRP2 can stably exist (18,(31)(32)(33). To characterize the inactive pool of IRE binding activity in HL-60 cells, we determined the optimal level of 2-ME required for activation of RNA binding, the effect of aconitase substrates on the response to 2-ME, and the extent to which sulfhydryl modifying agents such as NEM affected the RNA binding activity.
Maximal activation of IRE RNA binding activity occurred with 10 -50 mM 2-ME (Fig. 4A), a level of 2-ME that would not   Table I. be sufficient to significantly activate RNA binding by holo-IRP1. 3 Furthermore, previous addition of citrate reduced the action of 2-ME slightly, which also indicated that holo-IRP1 was not a significant component of the latent pool present in these extracts (Fig. 4A). Although the small effect of citrate on the action of 2-ME was not statistically significant, it was consistently observed, suggesting that approximately 10 -20% of the 2-ME-inducible pool of RNA binding activity was due to cytosolic aconitase.
Because the latent pool of IRE binding activity was redox regulated, we tested the effect of NEM on IRE binding activity in untreated and PMA-treated cells. The addition of 1 mM NEM completely inactivated RNA binding in untreated and PMAtreated cells when measured without any added 2-ME (Fig. 4B,  lane 3). To test whether the form(s) of IRP present in the latent pool contained disulfide or other forms of Cys that would be nonreactive with NEM, we added gradually increasing levels of 2-ME to extracts previously treated with NEM to determine if any IRE binding activity could be recovered. When we added back a molar excess of 2-ME (over NEM) to the NEM-treated extracts, we found that untreated extracts now contained more 2-ME-inducible IRE binding activity than was present in extracts from PMA-treated cells. Thus, untreated HL-60 cells contain a latent pool of IRE binding activity, mostly IRP2, in which oxidation of one or more cysteines is associated with inactivation of RNA binding. We have found that the "redox" state of IRP2 varies in a cell type-specific manner. 3 PMA Stimulates Phosphorylation of IRP1 and IRP2 in HL-60 Cells-To further assess the extent to which the effect of PMA on IRP function was due to post-translational events, we determined the relationship between protein synthesis, protein phosphorylation, and the increase in IRE RNA binding activity. We observed no change in the incorporation of [ 35 S]methionine into IRP1 (Fig. 5A) or IRP2 (Fig. 5B) as a function of PMA treatment, indicating that the relative rate of synthesis of IRPs was not altered under these conditions. In contrast to the lack of effect of PMA on synthesis rates of IRP1 and IRP2, we did observe marked changes in phosphorylation state of both proteins after the addition of PMA. IRP1 was not phosphorylated in untreated HL-60 cells incubated with [ 32 P]phosphate for 4, 8 (Fig. 6A, lanes 1 and 6), or 24 h (not shown). However, within 30 min after the addition of PMA, IRP1 became phosphoryl-ated, and it remained phosphorylated for at least 4 h (Fig. 6A,  lanes 2-5). Densitometric analysis revealed that untreated cells contained no detectable phosphorylated IRP1 when measured at the beginning or end of the experimental period (Fig.  6A, compare lanes 1 and 6). IRP2 was also found to be a phosphoprotein, but when compared with IRP1 it exhibited different phosphorylation characteristics. IRP2 was highly phosphorylated in untreated HL-60 cells that had been incubated with [ 32 P]phosphate for 4 or 8 h (Fig. 6B, lanes 1 and 6,  respectively), and PMA treatment stimulated further incorporation of phosphate into the protein within 60 min by 2-3-fold (Fig. 6B, compare lanes 1 and 3). The level of phosphorylation of IRP2 in untreated HL60 cells increased somewhat at the 4-h time point (Fig. 6B, compare lanes 6 and 1), but the level of phosphorylation of the binding protein remained 2-3-fold higher in PMA-treated cells (Fig. 6B, compare lanes 5 and 6). IRP2 is phosphorylated in several other cell types (results not shown).
To determine the number of sites phosphorylated in IRP1 and IRP2 and whether or not individual sites displayed different kinetics of phosphorylation after PMA treatment, we performed phosphopeptide mapping of the proteins isolated from untreated and/or PMA treated HL-60 cells. IRP1 isolated from PMA-treated cells had one major site of phosphorylation (Fig.  7A, spot a) that comigrated with one of the sites identified when the purified protein was phosphorylated by protein kinase C (results not shown) (23). In contrast, IRP 2 displayed one site in untreated cells (results not shown) that co-migrated with spot b in PMA-treated cells (Fig. 7, B and C). PhosphorImager analysis revealed a second site in PMA-treated cells (Fig. 7C,  spot c). Both of these sites present in IRP2 differed in migration from the site observed for IRP1, indicating that they are likely to represent phosphorylation of different residues in the two related proteins. The major phosphopeptide observed in IRP2 increased in intensity by about 2-fold in PMA-treated cells (Fig.  7, B and C) as compared with untreated cells (results not shown). Given the low intensity of the second phosphorylation site observed for IRP2 in PMA-treated cells, it remains to be determined if, as compared with untreated cells, it represents a new phosphorylation site or increased phosphorylation of a site that was partially phosphorylated before the addition of PMA. Taken together, both IRP1 and IRP2 are phosphorylated in response to PMA treatment of HL60 cells but that IRP2 has a much higher basal level of phosphorylation in untreated cells.

Contrasting Effect of PMA and Desferal Versus Hemin on Transferrin Receptor mRNA Abundance in HL-60 Cells-
Monocyte macrophage differentiation and macrophage activation are associated with complex changes in iron uptake and storage that are mediated at the transcriptional and posttranscriptional levels (26, 34 -36). To evaluate how alteration in IRP function affects transferrin receptor expression, we determined the effects of PMA treatment on TfR mRNA levels by RNase protection assay. In HL-60 cells treated with PMA for 5 h, the TfR mRNA level was increased (Fig. 8, lane 2). Hemin treatment decreased TfR mRNA abundance relative to controls, whereas desferal increased TfR mRNA abundance (Fig. 8,  lanes 3 and 4, respectively). In a series of experiments in which TfR mRNA levels were measured relative to glyceraldehyde phosphate dehydrogenase levels, PMA increased TfR mRNA levels 50 Ϯ 5% (n ϭ 4), hemin decreased TfR mRNA by 68 Ϯ 7% of control values (n ϭ 4), and desferal increased TfR mRNA levels by 17 Ϯ 23% (n ϭ 3). We found the effect of desferal on TfR mRNA levels to be somewhat variable. In several experiments where the effect of desferal was specifically examined, we found that TfR mRNA levels increased relative to controls by 142 Ϯ 24% (n ϭ 5), with the range being from no stimulation in one experiment to a 220% stimulation in a separate experiment (results not shown).
Comparison of the Effects of Hemin and Desferal Versus PMA on IRP Function-It is well documented that alterations in intracellular iron level affect IRE binding activity and rates of synthesis of ferritin and TfR. Treatment of a variety of cell types with hemin as an iron source decreases IRE binding activity, whereas the iron chelator desferal stimulates binding activity. In many of the cell types examined, these changes appear to occur without alterations in the rate of synthesis or abundance of IRP1, but it does appear, in certain cases, to be mediated by changes in degradation of this binding protein (16,19,37). For IRP2, iron status influences the turnover rate of the protein in some but apparently not all cells (17,18). Treatment of HL-60 cells with hemin for 4 h decreased IRE RNA binding activity by 30 -40% relative to untreated cells (Fig. 9,  A and B). The addition of desferal stimulated IRE RNA binding to a level similar to that observed for PMA treatment of the cells (Fig. 9, A and B). The changes in spontaneous IRE binding activity were directly related to the changes we observed in abundance of TfR mRNA.
Differences were noted concerning the regulation of the function of IRPs by hemin and desferal as compared with PMA treatment. First, neither hemin or desferal affected total phosphorylation state of IRP1 (Fig. 10A) or the relative synthesis rates of the protein (Fig. 10B). For IRP2, hemin reduced the amount of phosphorylated protein by 30 -40% (Fig. 10C)  out changing synthesis of this binding protein (Fig. 10D). It remains to be seen if iron status differentially affects phosphorylation of specific sites but not the total level of phosphate in IRPs. Second, hemin reduced IRE binding activity as measured in both the absence or the presence of 2-ME (Fig. 9C), suggesting that hemin reduced the level of IRE-binding protein, presumably IRP2, present in HL-60 cells. The reduction in the level of binding activity was similar to the extent to which hemin reduced the level of phosphorylated IRP2. In contrast, the total level of 2-ME-inducible IRE binding activity was essentially identical in extracts from untreated, PMA-treated, and desferal-treated cells. Taken together, these results support the hypothesis that phosphorylation of IRP1 and IRP2 represents a means for influencing cellular iron homeostasis by factors other than iron. DISCUSSION For IRPs to represent true central regulators of iron homeostasis, it seems likely that they should respond to multiple effectors. In this regard both intracellular and extracellular effectors of IRP1 have been identified. Iron is clearly an intracellular effector of IRP1 function and could be considered an allosteric regulator of the binding protein. A number of other effectors of IRP1 function have been identified including nitric oxide, oxidative stress, and, in our laboratory, phosphorylation (20 -23). Modulation of IRP1 by these agents provide mechanisms to affect the uptake and metabolic fate of iron by extracellular agents. IRP2 appears to be modulated by iron as well as by changes in proliferation status of some cells. In the latter case specific agents that act on IRP2 have not been identified. Our current work extends our observations concerning the mechanisms by which phosphorylation affects the functions of IRP1 and also demonstrate that IRP2 is phosphoregulated in response to extracellular stimuli. Our results support the hypothesis that IRP1 and IRP2 are iron-based and phosphorylation-based regulators of TfR mRNA accumulation and presumably ferritin and erythroid 5-aminolevulinate synthase synthesis.
Our observations provide further insight into the mechanisms by which factors other than iron affect IRP function. We showed that extracts of HL-60 cells possess an inactive form of IRP(s) that is largely if not exclusively in a nonaconitase form and is mostly composed of IRP2. PMA appears to act on an inactive form of IRP1 and particularly IRP2 and induces their conversion to high affinity RNA-binding proteins. This action of PMA was directly associated with increased phosphorylation of the proteins. Our conclusion is based on the amount of 2-ME (10 -50 mM) required to maximally activate the latent pool of binding activity in extracts from untreated HL-60 cells and that inclusion of aconitase substrates does not substantially block the 2-ME effect on RNA binding. Several pieces of evidence indicate that this latent form of IRP is functionally important. First, we showed that PMA treatment of HL-60 cells rapidly stimulated IRE binding activity, and this was found to be associated with a significant increase in TfR mRNA level. Second, we have found that the size of the latent pool of IRP is affected by cell growth state and the level of serum in which HL-60 cells are cultured (results not shown) and TfR expression is affected by serum in these cells (38). Third, supershift analysis demonstrated that the inactive pool of binding protein in HL-60 cells is largely IRP2, and we (results not shown) and others (31)(32)(33) have observed inactive IRP2 in rodent cell lines, finding that it is not restricted to the HL60 cell system. These observations indicate that the latent form of IRP2 can be recruited into the functional high affinity pool of binding protein.
To more clearly determine the mechanism(s) by which PMA affected IRP function, particularly IRP2, we examined the extent to which post-translational events affected RNA binding activity.
Our results indicated that RNA binding by IRP2 can be FIG. 7. Tryptic phosphopeptide mapping of IRP1 and IRP2 isolated from PMA-treated HL-60 cells. HL-60 cells were treated with PMA for 60 -120 min to induce a high level of phosphorylation of IRP1 and IRP2. IRP1 and IRP2 were separately immunoprecipitated and fractionated by SDS-polyacrylamide gel electrophoresis as indicated under "Materials and Methods." Bands corresponding to IRP1 and IRP2 were excised from the dried gel, eluted, subjected to tryptic digestions overnight, and then separated by chromatography and electrophoresis as described. A is an autoradiogram of an IRP1 phosphopeptide map, and B is an autoradiogram of a IRP2 phosphopeptide map; both are from PMA-treated cells. C is a PhosphorImager analysis of a IRP2 phosphopeptide map that reveals the presence of a second, less radioactive spot for IRP2 in PMA-treated cells. The origin is indicated by the dot. One spot was seen for IRP1 (A, band a), and two spots were seen for IRP2 (B and C, spots b and c). For IRP2 only spot b was observed in untreated cells (results not shown) but at a lower intensity than seen in PMA-treated cells. Total RNA was extracted and isolated from the cells for analysis of TfR mRNA levels using a ribonuclease protection assay according to the manufacturer's instructions. Antisense TfR and glyceraldehyde phosphate dehydrogenase (GAPDH) probes (500,000 cpm each) were hybridized at 45°C for 16 h with 30 g of total RNA from each treated cell lysate digested with a mixture of RNase A/T1 and separated on a 5% acrylamide/8 M urea gel. TfR and glyceraldehyde dehydrogenase probes resulted in protected fragments of 374 and 316 nucleotides, respectively. Undigested TfR and glyceraldehyde phosphate dehydrogenase probes migrated at 429 and 403 nucleotides, respectively (lane 5). The total area of each band was quantitated by computerized densitometry as described under "Materials and Methods." modulated without changes in the amount of binding protein.
We obtained several lines of evidence indicating that PMA treatment influenced the amount of IRP2 in the high affinity state by altering the proportion of the protein that was stably reduced. First, PMA stimulates IRE binding activity in HL-60 cells, as measured in the absence of dithiothreitol or 2-ME, without affecting the total amount of binding protein, as evaluated by the 2-ME induction assay. Further support for this conclusion came from the observation that the ability of PMA to stimulate IRE binding activity occurred without changes in the synthesis rates of IRPs. Thus, oxidation or reduction of cysteine(s) appears to represent a mechanism by which IRP2 RNA binding activity can be regulated in these cells. Second, similar results were obtained when 1 mM dithiothreitol was included in the cell lysis buffer, although the magnitude of the difference between untreated and PMA-treated cells was lessened (results not shown). Thus, phosphorylation may increase the proportion of IRP2, which binds RNA with high affinity by inducing a structural change in the binding protein, which causes the reduced form of the protein to be favored, and/or by stabilizing the protein in the reduced state during cell lysis. As noted above, the redox state of IRP2 varies in a cell typespecific manner. In HL-60 cells (Figs. 1 and 3) and C58 cells, only a portion of IRP2 is stably reduced. 4 However, FTO-2B rat hepatoma cells contain IRP2 in a fully reduced state, even when extracts are prepared without dithiothreitol or 2-ME (Ref. 17 and results not shown). Thus, RNA binding by IRP2 can be regulated in a cell type-specific manner without large changes in the levels of the protein through an oxidation or reduction mechanism.
Is there a role of phosphorylation in affecting the redox state of IRP2 as there appears to be for another nucleic acid-binding protein (39)? We note that within the 73-amino acid region found in human and rat IRP2, but not IRP1, there are two Cys residues with an adjacent Arg residue. Based on studies with model peptides (40), such Cys residues are predicted to be especially redox active and may well contribute to the redox and/or iron regulation of IRP2. Furthermore, some of the Ser and Thr residues in the 73-amino acid region fit the consensus sequences for protein kinase C or mitogen-activated protein kinases, both of which can be activated directly or indirectly, respectively, by PMA. Thus, phosphorylation within this region might affect the redox properties of nearby Cys residues. Our results support the hypothesis that oxidation or reduction of one or more Cys in IRP2 represents a physiological mechanism for regulation of its RNA binding function.
To evaluate the role of cysteine oxidation state, we made use of our observation that NEM completely inactivated spontaneous RNA binding in extracts of untreated and PMA-treated cells and used it as a means to assess the percentage of binding . The total area of bands for IRP1 and IRP2 were quantitated using a computerized densitometer as described under "Materials and Methods." activity that was in a reduced state. We found, as others have recently demonstrated (32), that NEM can inactivate spontaneous binding activity for IRP1 and IRP2. Our results on the effect of NEM on IRP2, like those of Kuhn and Henderson (32), differ from those of Kim et al. (41) and further suggest that the cellular form of IRP2 differs from the recombinant form of the binding protein. We also demonstrated that after inactivation of stably reduced IRPs with NEM we could still activate the latent pool with low levels of 2-ME, further indicating that oxidation or reduction of one or more critical Cys in the binding protein(s) contributed to the regulation of its RNA binding function. NEM treatment followed by the addition of 2-ME permitted us to demonstrate that PMA treatment of HL-60 cells reduced the size of the latent pool of IRP though a change in cysteine oxidation state. Also, we have found that IRP2 is reversibly inactivated by diamide (results not shown) and recently a similar phenomenon was observed in other cell types (32). Thus, it is apparent that post-translational regulation of IRP2 can occur as a result of changes in the cysteine oxidation state without alterations in the abundance of the binding protein.
To what extent do phosphorylation-dependent and iron-dependent mechanisms overlap to regulate IRP2 function? Roles for oxidation or reduction, involving the intra-and intermolecular formation or destruction of disulfide bonds, in influencing protein function have been demonstrated in several systems (42)(43)(44)(45)(46)(47)(48). Oxidized species of proteins such as HMG-CoA reductase (42), the hemin-controlled protein kinase (43), and other nucleic acid binding proteins (42)(43)(44)(45)(46)(47)(48) exist in cells, even in the presence of the cellular glutathione redox system. Furthermore, IRP1, FNR, and soxR provide striking examples of how changes in ligation state of Cys residues, as a result of formation of Fe-S clusters, greatly influence protein function (13,49,50). However, in the case of IRP2 our results presented here as well as those of other investigators (17)(18)(19) indicate that irrespective of whether an Fe-S cluster forms in this binding protein, iron decreases the abundance of IRP2. Along these lines, it appears that iron reduces the abundance of the phosphorylated form of IRP2, suggesting that this form of the binding protein may be preferentially degraded when iron is in excess. However, it remains to be shown whether the phosphorylated form of IRP2 is the preferred substrate for degradation. In contrast to the effect of iron status in affecting the steady state level of IRP2, PMA affects the relative distribution of the binding protein between active (reduced) and inactive (oxidized) pools without altering the total level of the protein. Because desferal treatment affects RNA binding in HL-60 cells without apparent changes in the total level of binding protein, it appears that iron depletion may affect the abundance of the latent form of IRP2 in cells, but this remains to be investigated further. Overall, our results demonstrate that IRP2 can stably accumulate in an inactive form in cells and that multiple pathways exist for modulating the RNA binding activity of the protein.
We demonstrated that phosphorylation of IRP1 and IRP2 was associated with activation of their RNA binding activity. Because IRP2 apparently cannot act as an aconitase even if it formed an Fe-S cluster (17), it is likely that phosphorylation affects other functions. This could include an effect on the relative abundance of the oxidized and reduced forms of IRP2 as indicated above or on the action of iron in affecting RNA binding function perhaps through mechanisms involving altered degradation of the protein. Given the observations that RNA binding by IRP2 is stimulated during liver regeneration and lymphocyte activation, it will then be of interest to define the extent to which phosphorylation affects the function of this binding protein during cell proliferation (24 -26, 51). With regard to IRP1, we also found that increased phosphorylation was associated with a stimulation of its binding activity. These results support the concept that within cells the Fe-S cluster in IRP1 may be labile and that phosphorylation may affect assembly (or reassembly) of cluster in intact cells (1,23). A direct effect of phosphorylation on cluster assembly/disassembly remains to be more directly demonstrated. However, given our observations 5 that the iron-free form of IRP1 is the preferred substrate for phosphorylation, phosphoregulation may indirectly activate RNA binding by predominantly acting as a determinant of whether or not IRP1 functions as an aconitase or RNA-binding protein. Taken together, our results provide support for the hypothesis that alterations in phosphorylation state of IRP1 or IRP2 provide a means by which extracellular agents affect iron homeostasis.