Iron chaperones PCBP1 and PCBP2 mediate the metallation of the dinuclear iron enzyme deoxyhypusine hydroxylase

A potential obstacle in the analysis of endogenously expressed, nonheme iron enzymes is the propensity for their Fe(II) cofactors to oxidize and/or dissociate from active site ligands after cells are disrupted and the contents diluted. Thus, we sought to examine a dinuclear iron enzyme activity that could be measured in living cells. DOHH catalyzes the second step of the posttranslational modification of a conserved lysine in eIF5A to Hpu (Fig. 1A) (24). Step one of the hypusination of human eIF5A is the transfer of an aminobutyl moiety from cellular spermidine to the ε-amino group of lysine 50 to form Dhp, a reaction that is catalyzed by Dhp synthase. In the second step, DOHH hydroxylates the Dhp to form Hpu. DOHH is a HEAT-repeat enzyme that contains a peroxo-linked, dinuclear iron center in which the iron ions are coordinated by two sets of conserved histidine and glutamate residues (23, 25, 26). Hpu synthesis occurs only on a single lysine residue on a single protein, eIF5A. However, this modification is conserved among all eukaryotes and archaea examined and is essential for the function of eIF5A, which facilitates translation of polyproline sequences at the ribosome (27). Yeast and mice that lack eIF5A or the hypusination machinery are not viable (28, 29) because of this defect in translation of polyproline-containing proteins. Polyproline-containing proteins are encoded by ∼10% of the yeast genome (27) and, in humans, include Huntingtin and SH3 domain- and WW domain-binding proteins.

We measured the activity of DOHH in intact HEK293 cells by incubating them with [³H]spermidine and analyzing the labeled amino acids by ion-exchange chromatography. DOHH activity can be inhibited in cells by treatment with desferrioxamine (Dfo), an iron chelator (30). In untreated cells (0 Dfo), nearly all of the labeled amino acid eluted in fractions that comigrated with Hpu (Fig. 1B). However, when cells were incubated with increasing concentrations of Dfo, a peak of labeled amino acid corresponding to Dhp appeared. Dhp increased and Hpu decreased with increasing concentrations of Dfo, indicating that cellular iron deficiency could lead to partial or full inactivation of DOHH. We then examined the effects of PCBP depletion on DOHH activity (Fig. 1 C and D). Because depletion of PCBP1 or PCBP2 in iron-replete cells had little effect on the synthesis of Hpu, we treated all cells with a low concentration of Dfo (20 μM) for 2 h before labeling. Cells treated with no siRNA or a nontargeting control siRNA exhibited a very low level of Dhp accumulation (18–20% Dhp; 80–82% Hpu), indicative of a low level of DOHH inhibition. In contrast, cells treated with siRNA against PCBP1, PCBP2, or both PCBP1 and -2 exhibited a marked accumulation of Dhp (81–83%) and a corresponding reduction in Hpu (17–19%), indicative of a marked inhibition of DOHH activity. By Western blotting, levels of DOHH protein were similar in control and siRNA-treated cells, and PCBPs were efficiently depleted to <10% of control levels (Fig. S1A). Thus, the loss of DOHH activity could not be attributed to an absence of the DOHH protein in PCBP-depleted cells. Deletion of both PCBPs simultaneously produced an inhibition of DOHH that was similar to the inhibition produced by deleting either PCBP individually. This suggested that PCBP1 and PCBP2 were not functionally redundant but instead functioned cooperatively to activate DOHH. We confirmed that the observed inhibition of DOHH activity was not attributable to off-target effects by demonstrating inhibition of DOHH in cells depleted of PCBPs using siRNAs of alternative, unrelated sequences (Fig. S1B).

We hypothesized that the loss of DOHH activity in PCBP-depleted cells was attributable to loss of the iron cofactor from the active site, and we examined DOHH from cells depleted of PCBPs for evidence of impaired iron cofactor incorporation. Recombinant DOHH is purified from Escherichia coli in two forms, one that migrates slowly in native gels and contains no iron and no enzymatic activity (apo-DOHH) and one that migrates more rapidly and contains both iron and enzymatic activity (holo-DOHH) (23, 31). Furthermore, only Fe(II), and not other divalent metals, can convert apo-DOHH to the more compact holo form, which contains a peroxo-iron bridge, in vitro (23, 26). We confirmed that the faster-migrating, holo form of DOHH corresponds to the iron- and oxygen-bound enzyme by noting its appearance when both iron and air are present but its absence when iron is added in the absence of oxygen (Fig. S2A). Endogenous DOHH in HEK293 cell extracts also migrated as multiple species by native-gel electrophoresis and immunoblotting (Fig. 2A). Cells made iron deficient by treatment with a high concentration of Dfo (100 μM) contained a single species of DOHH that exhibited the slower migration characteristic of the apo form. This complete conversion to the apoenzyme was consistent with the complete loss of DOHH activity observed under this condition in Fig. 1B. DOHH from cells treated with low or no Dfo migrated as three species that corresponded to the apo form, the holo form, and an intermediate form that likely corresponded to partially metallated or incompletely folded DOHH. Addition of iron to cells did not quantitatively convert all DOHH to the holo form, suggesting these forms are interconverted in cells (Fig. S2B). Again, no significant difference in the percentage of DOHH migrating as the holoenzyme was observed between cells treated with a control siRNA vs. PCBP1 or PCBP2 siRNAs, when cells were not treated with Dfo (Fig. 2 A and B). However, addition of very low concentrations of Dfo (5 and 10 μM) was accompanied by a marked decrease in the percentage of holo-DOHH in cells lacking PCBP1 or PCBP2. Control cells treated with 5 μM Dfo exhibited a 27% decrease in holo-DOHH, whereas PCBP1- and PCBP2-depleted cells exhibited 83% and 86% decreases in holo-DOHH, respectively. Cells treated with 10 μM Dfo exhibited proportionally smaller amounts of holo-DOHH and PCBP1-depleted cells lost both holo and intermediate forms of DOHH. The presence of DOHH protein and efficient depletion of PCBP1 and PCBP2 were confirmed by denaturing gel electrophoresis and immunoblotting (Fig. 2A, lower gels). These data indicated that cells lacking PCBP1 or PCBP2 exhibited a specific loss of the iron-bound form of DOHH.

We examined the conversion of apo-DOHH to holo-DOHH using in vitro synthesis and iron-center metallation. We found that DOHH synthesized in wheat germ extracts, which contain low amounts of iron and no PCBP1, migrated exclusively as apo-DOHH by native-gel electrophoresis and could be converted to holo-DOHH in vitro in the presence of increasing concentrations of Fe(II) (Fig. 3A). Using the conversion of apo- to holo-DOHH as an indicator of iron binding, we determined that, at an iron concentration of 1.3 ± 0.2 μM, approximately half of the DOHH bound iron and converted to the holo form. Because iron binding is coupled to oxygen binding and the formation of a very stable complex, this value does not represent an equilibrium dissociation constant. Nevertheless, this concentration is similar to the affinities of PCBP1 and PCBP2 for iron (17).

We then determined whether cell lysates containing PCBP1 and PCBP2 could convert apo-DOHH to holo-DOHH. Labeled apo-DOHH was synthesized by in vitro translation, HEK293 cell lysates were added, and conversion to holo-DOHH was monitored by native gels. Addition of lysate (25 or 50 μg) from cells treated with control siRNA resulted in the dose-dependent conversion of apo- to holo-DOHH (Fig. 3 B and C). In contrast, lysates from cells treated with siRNAs against PCBP1 or PCBP2 were much less efficient in converting apo- to holo-DOHH. We considered that the reduced capacity of these lysates to metallate DOHH might be attributable to reduced levels of iron in these cells. In previous studies, uptake of iron in Huh7 cells depleted of PCBP1 was not different from control cells (17). Here, we used inductively coupled plasma mass spectrometry (ICP-MS) to measure the iron content of cells and found no difference between control cells and cells depleted of PCBP1 or PCBP2 (Fig. S3A). These studies indicated that lysates containing both PCBP1 and PCBP2 could more efficiently donate iron to the active site of DOHH than lysates lacking PCBP1 or PCBP2.

Our studies indicated that cellular PCBP1 and PCBP2 were involved in the incorporation of iron into the active site of DOHH. Both PCBP1 and PCBP2 have been found to exhibit iron-mediated binding to ferritin in vivo, when cells are treated with iron, and in vitro, when the PCBPs are loaded with iron (17, 32). Both PHD2 and FIH1 were found in a complex with PCBP1 in cells, but only the FIH1 interaction exhibited a requirement for iron (19). We subjected HEK293 cells to iron, Dfo, or no treatment, then isolated endogenous PCBP1 by immunoprecipitation, and examined immune complexes for the presence of DOHH. DOHH was undetectable in complexes from Dfo or untreated cells but was present in complexes from iron-treated cells (Fig. 4A). No DOHH was detected in immunoprecipitates using bulk IgY as a negative control, demonstrating the specificity of the interaction. Similar amounts of PCBP1 and DOHH were present in each lysate and similar amounts of PCBP1 were isolated in each condition. We then tested whether PCBP1 was detectable in immunoprecipitates of DOHH. We constructed stable cell lines that contained either the empty vector or inducibly expressed, Flag-epitope-tagged DOHH. In this cell line, Flag-DOHH is expressed at levels three- to fourfold higher than the endogenous DOHH (Fig. 4B, Lower Left). PCBP1 was detected in immunoprecipitates of Flag-DOHH from iron-treated cells but was not detected in immunoprecipitates from the control cell line that did not express Flag-DOHH (Fig. 4B, Upper Right), although both cell lines contained similar amounts of PCBP1. These data suggested that PCBP1 and DOHH physically interacted in cells and that this interaction was more stable in the presence of elevated intracellular iron.

We previously demonstrated the roles of PCBP1 and PCBP2 in delivering iron to ferritin and the mononuclear iron enzymes of the 2-oxoglutarate dioxygenase family. We have presented evidence here of the role of PCBP1 and PCBP2 in the metallation of the dinuclear iron enzyme DOHH. In addition to its incorporation into nonheme iron enzymes, cellular iron can be used for the assembly of Fe-S clusters or the synthesis of heme. Because the subcellular localization of PCBPs is restricted to the cytosol and the nucleus (13), they cannot be directly involved in delivering iron to the mitochondrial Fe-S cluster or heme biosynthetic machinery. They could, however, be involved in the delivery of iron to mitochondria or be involved in the assembly of Fe-S clusters in the cytosol. We examined the activity of two cytosolic Fe-S cluster-containing enzymes, aconitase and xanthine oxidase, in cells that lacked PCBP1 and PCBP2. We depleted PCBPs and then treated cells with Dfo for 0, 4, or 18 h to produce progressive degrees of iron deficiency. Cells were lysed, cytosol was separated from mitochondria, and cytosolic aconitase activity was measured. As expected, Dfo treatment led to progressive loss of aconitase activity (Fig. 5A). Depletion of PCBP1 or PCBP2 also led to a loss of aconitase activity compared with control siRNA-treated cells with no (0 Dfo) or mild (4-h Dfo) iron deficiency, with only the 4 h Dfo treatment reaching statistical significance. Loss of aconitase activity was not attributable to loss of the aconitase protein, however, because immunoblotting revealed that the protein was present at similar levels (Fig. 5A, Lower). In contrast, depletion of PCBP1 or PCBP2 in Huh7 cells had no effect on xanthine oxidase activity, although activity was diminished by 18 h Dfo treatment (Fig. 5B). We confirmed the separation of cytosol and mitochondria by immunoblotting with antibodies against tubulin and voltage-dependent anion channel (VDAC), respectively (Fig. S4 A and B). We also confirmed the efficient depletion of PCBP1 and PCBP2 (Fig. S4C).

Although xanthine oxidase and cytosolic aconitase are both Fe-S cluster enzymes, they differ in important ways. Xanthine oxidase contains two [2Fe-2S] clusters in domains separate from the active site, which contains a molybdopterin cofactor (33). Cytosolic aconitase contains a single, labile [4Fe-4S] cluster in the active site (34). Under specific cellular conditions, this cluster can partially disassemble into a [3Fe-4S] form or it can completely disassemble (35, 36). Partial disassembly is associated with loss of enzyme activity, and complete disassembly is associated with both loss of enzyme activity and acquisition of RNA-binding activity. Cytosolic aconitase, also known as iron regulatory protein (IRP)1, is a bifunctional protein. In its iron-free form, it can bind to specific stem-loop structures [called iron-regulatory elements (IREs)] in mRNAs and alter their stability or translation (37). If the loss of cytosolic aconitase activity in the PCBP-depleted cells were attributable to complete absence of the [4Fe-4S] cluster, these cells would exhibit a reciprocal gain in IRE-binding activity. We measured IRE binding in PCBP-depleted cells by mixing lysates with radiolabeled IRE-containing oligonucleotides and separating IRP1/IRE complexes from free IREs by gel electrophoresis. In cells treated with control siRNA, the increasing iron deficiency produced by Dfo treatment resulted in increasing amounts of IRP1/IRE complexes (Fig. 5C, Upper). In contrast, cells depleted of PCBP1 or PCBP2 exhibited lower levels of IRE-binding activity for each level of Dfo treatment compared with control siRNA cells. Addition of high concentrations of the sulfhydryl reductant 2-mercaptoethanol (2ME) to lysates can convert non–IRE-binding forms of IRP1 to the IRE-binding form, thereby providing a measure of the total amount of IRP1-mediated, IRE-binding activity in the lysate (38). Addition of 2ME indicated that levels of IRE-binding activity were similar in all cells (Fig. 5C, Lower). IRP2, which does not contain an Fe-S cluster but does bind IREs, migrates as a separate species under the electrophoretic conditions used in these assays. No IRP2/IRE complexes were detected, likely because of the very low levels of IRP2 protein in these cells. These data indicated that, although depletion of PCBPs impaired aconitase activity, it did not promote the acquisition of IRE-binding activity and likely did not produce complete loss of the [4Fe-4S] cluster in cytosolic aconitase/IRP1.

We examined the role of PCBPs in the delivery of iron to mitochondria using two methods. First, we isolated mitochondria from cells depleted of PCBPs and measured the iron content by ICP-MS (Fig. S5A). Although cells depleted of PCBP2 exhibited a slightly reduced mitochondrial iron content, the difference was small, and no effect was seen upon PCBP1 depletion. Depletion of PCBPs and separation of intact mitochondria from cytosol was confirmed by immunoblotting (Fig. S5B). Second, we measured the biosynthesis of heme in cells depleted of PCBPs and found that the synthesis of heme was not different from control cells (Fig. 5D). These measurements indicated that PCBPs did not play a major role in the delivery of iron to mitochondria.

Huh7, HEK293, and HEK293T cell propagation is described in SI Materials and Methods. PCBP1 and PCBP2 were depleted by single (Huh7) or two sequential (HEK293 and HEK293T) transfections of RNAi (Invitrogen) (19). A nontargeting, scrambled sequence siRNA pool was used as a control. Cells were harvested 4 d after the first transfection.

Immunoblotting was performed as described in SI Materials and Methods.

Detailed protocols appear in SI Materials and Methods. For measurement of DOHH activity, cells were labeled with [1,8-³H]spermidine, and radiolabeled Hpu and Dhp were separated by ion-exchange chromatography (30); c-aconitase activity was measured using the coupled aconitase/isocitrate dehydrogenase assay (42, 43). Xanthine oxidase activity was determined using the Amplex Red assay kit (Invitrogen).

DOHH apo- and holoenzyme were separated using native PAGE (23). For in vitro metallation, ³⁵S-labeled apo-DOHH was incubated with Fe(II) or with HEK293T lysate. Detailed protocols appear in SI Materials and Methods.

Immunoprecipitations of PCBP1 were performed as described (19). A stable cell line expressing Flag-DOHH under the doxycycline-inducible CMV promoter was constructed, and Flag-DOHH was isolated (16). Detailed protocols appear in SI Materials and Methods.

To measure IRP1 binding to [³²P]IRE, HEK293 cells were depleted of PCBPs and assayed by electrophoretic mobility shift (38, 44). Total cellular and mitochondrial iron content were analyzed by ICP-MS (45, 46). Detailed protocols appear in SI Materials and Methods.