Requirements for iron-regulated degradation of the RNA binding ...

The EMBO Journal vol.14 no.21 pp.5350-5357, 1995

Requirements for iron-regulated degradation of theRNA binding protein, iron regulatory protein 2

Kazuhiro Iwai, Richard D.Klausner andTracey A.Rouault1Cell Biology and Metabolism Branch, National Institute of ChildHealth and Human Development, Bethesda, MD 20892, USA

'Corresponding author

Iron regulatory proteins (IRPs) regulate the expressionof genes involved in iron metabolism whose transcriptscontain RNA stem-loop motifs known as iron-responsive elements (IREs). When iron concentrationsare low, IRPs bind to IREs in the 5' untranslated region(UTR) of transcripts where they repress translation, orthe 3' UTR of transcripts where they inhibit degrada-tion. The RNA binding activities of the homologousproteins IRP1 and IRP2 are both regulated post-translationally. The binding activity of IRP2 is regu-lated by the degradation of the protein when cells areiron-replete. Here, we demonstrate that a 73 aminoacid sequence that corresponds to a unique exon inIRP2 contains a sequence required for rapid degrada-tion in iron-replete cells. The deletion of this sequenceeliminates the rapid turnover of IRP2, whereas thetransfer of this sequence to the corresponding positionin the homologous protein IRP1 confers the capacity foriron-dependent degradation upon IRP1. Site-directedmutagenesis has demonstrated that specific cysteineswithin the IRP2 exon are required for iron-dependentdegradation. The degradation of IRP2 appears to bemediated by the proteasome in iron-replete cells. Whendegradation is prevented, the RNA binding activity ofIRP2 is not regulated by iron concentration. Thus,degradation is required for the regulation of the RNAbinding activity of IRP2.Keywords: cysteine/iron/iron-responsive element/protea-some/protein degradation

IntroductionThe levels of expression of several proteins involved iniron metabolism are regulated by changes in iron avail-ability. The expression of ferritin, an iron sequestrationprotein, is decreased when cells are depleted of iron,whereas the expression of transferrin receptor (TfR), aniron uptake protein, is increased in cells that are depletedof iron. The expression of the erythroid form of amino-levulinic acid synthase, the rate-limiting step in hemebiosynthesis, is decreased in iron-depleted cells, a regu-latory response that is appropriate because iron is requiredin the final step of heme biosynthesis. Underlying theseseemingly disparate responses is a regulatory system basedon the high-affinity binding of a small family of iron-sensing proteins to RNA stem-loop motifs, known as

iron-responsive elements (IREs). The sensing proteins,known as iron regulatory proteins (IRPs), referred topreviously as IRE binding proteins, iron regulatory factorsand the ferritin repressor protein, have been cloned andcharacterized extensively (reviewed in Klausner et al.,1993; Melefors and Hentze, 1993; Kuhn, 1994). IRPI isan iron-sulfur cluster protein that functions as a cytosolicaconitase when the iron-sulfur cluster is present. In theabsence of the iron-sulfur cluster, IRPI binds IREs withhigh affinity. In cells that express a constitutive IREbinding form of IRPI, the translation of ferritin and thedegradation of the mRNA of TfR are repressed, thusshowing that previously characterized in vitro bindingdata correlate with function in vivo (DeRusso et al., 1995).

Binding studies have shown that a second IRP, IRP2,binds to consensus IREs with a high affinity and specificitysimilar to IRPI (Rouault et al., 1992; Henderson et al.,1993; Guo et al., 1994; Samaniego et al., 1994). Whereasthe RNA binding activity of IRPI is regulated by thereversible assembly and disassembly of an iron-sulfurcluster in an otherwise stable protein, absolute levels ofIRP2 are decreased markedly by iron treatment in severaldifferent cell types (Guo et al., 1994; Samaniego et al.,1994), and the decrease in protein levels can be accountedfor by an increase in the rate of degradation of IRP2(Samaniego et al., 1994).Amino acid sequence homology is high between IRPI

and IRP2 throughout the entire length of the protein, withthe exception of a unique insertion of 73 amino acids inIRP2 relative to IRPI. In an effort to understand themechanism of degradation of IRP2 in iron-replete cells,we have excised the sequence unique to IRP2 and charac-terized the resulting deletion mutant. Here we demonstratethat the IRP2-specific exon is required for the degradationof IRP2 that occurs in iron-replete cells. Moreover, thetransfer of this exon to IRPI confers a similar phenotypeon IRPI, rendering it unstable in iron-replete cells. Thedegradation of IRP2 is inhibited by the treatment of cellswith inhibitors of proteasome function, indicating thatIRP2 may be degraded by the ubiquitin-proteasomedegradation system.

ResultsIRP2 is rapidly degraded in vivo by treatment withironLevels of IRPI and IRP2 were assessed after treatmentwith ferric ammonium citrate (FAC) or desferrioxamineusing two assays: a supershift assay to assess the amountof epitope-tagged protein that could bind IREs, andWestern blots to determine the total amount of immuno-logically detectable recombinant protein, without regardto its RNA binding status. The impact of manipulationsof cellular iron status on levels of recombinant IRPI and

5350

Iron-regulated degradation of IRP2

A IRP1 IRP2 B IRPi IRP2

D F D FD F D F1% 2ME - +- + + + D F D F

Fig. 1. Quantitation of levels of recombinant IRPI and IRP2 after thetreatment of cells with iron chelators or an iron source. RD4 cellswere induced to express recombinant IRPI or IRP2 withdexamethasone. After a 16 h treatment with desferrioxamine (D) or

FAC (F), cells were harvested. (A) Quantitation by a supershift assay.

Lysates (15 jg) were incubated with anti-myc epitope antibody,followed by incubation with radiolabeled IRE in the presence (+) or

absence (-) of 2ME. Mixtures of lysates and radiolabeled probe were

electrophoresed on an 8% native polyacrylamide gel. (B) Quantitationby Western blot analysis. Lysates (30 gg) were separated on a 6%polyacrylamide gel, transferred to nitrocellulose filters and probed withanti-myc antibody.

IRP2 in RD4 cells is illustrated in Figure 1. The supershiftassay on lysates from transfected cells revealed that IREbinding activities of recombinant IRPI and IRP2 were

decreased markedly after the treatment of cells with FAC(Figure IA). It has been established previously that IREbinding activity can be increased or 'recruited' with2-mercaptoethanol (2ME) treatment in lysates in whichIRPI is predominantly in the iron-sulfur cluster-bearingform (Haile et al., 1989; Hentze et al., 1989). Treatmentof the lysates with 1% 2ME recruited the cryptic bindingactivity of IRPl, whereas no binding activity was recruitedby the treatment of lysate from cells expressing recom-

binant IRP2. Western blotting of total recombinant IRP inthe lysates (Figure IA) revealed that levels of IRPIwere unchanged, whereas those of IRP2 were decreasedmarkedly by the treatment of cells with FAC (Figure 1B).To determine the mechanism of loss of IRP2 in iron-

replete cells, pulse-chase experiments were performed.Degradation rates of IRPI and IRP2 were evaluated(Figure 2) when cells were either deprived of iron bytreatment with the iron chelator desferrioxamine, or madeiron-replete by treatment with FAC. IRPI and IRP2 were

both stable over 24 h in cells treated with desferrioxamine.In contrast, IRP2 was degraded by >50% after 3 h incells treated with FAC (Figure 2, lower left panel), whereasIRPl was relatively stable over the 24 h time course (Figure2, lower right panel). A limited amount of degradation ofIRPl was observable in cells treated with FAC, as indicatedby the -50% drop in 35S-labeled IRPI seen at 24 h inFigure 2.

A sequence unique to IRP2 is required for rapidiron-dependent degradationIRP1 and IRP2 are highly homologous proteins with an

overall sequence identity of 58% (Samaniego et al., 1994).A major distinguishing feature between IRPI and IRP2 isthe presence of a 73 amino acid insertion in domain 1

of IRP2 (Figure 3). Cloning of the genomic fragmentcorresponding to this sequence has revealed that the 73amino acids are encoded by a single exon (K.Iwai,unpublished results). When the 73 amino acid insertionwas excised by site-directed mutagenesis from IRP2,levels of the IRP2 exon deletion mutant (IRP2-73) were

no longer decreased markedly after treatment with iron(Figure 4A). The IRE binding affinity remained high, with

an estimated Kd of 10-50 pM, and specificity for theconsensus IRE was unchanged, as judged by competitionassays with unlabeled IREs and unrelated stem-loops (datanot shown). These results indicated that the fundamentaltertiary structure of IRP2 was unchanged by the excisionof this exon sequence.To verify that the IRP2 exon deletion mutant (IRP2-

73) accumulated because of a decrease in the rate ofdegradation, as has been shown for wild-type IRP2(Samaniego et al., 1994; data presented here), pulse-chaseexperiments were performed (Figure 4B). The rate ofdegradation of IRP2 wild-type in iron-treated cells wasmarkedly faster than that of the IRP2-specific exon deletionmutant (t112 of 3-6 versus 24 h), thus supporting thehypothesis that the IRP2-specific exon may containdegradation signals which are activated by iron.

Insertion of the IRP2-specific exon into IRPI resultsin iron-dependent degradation of recombinantIRP1To assess whether the 73 amino acid IRP2-specific exoncontained sufficient information to direct the iron-regulateddegradation of IRP2, the sequence was cloned into IRPIat a homologous position. In Figure 5A, the IRP1/IRP2-specific exon chimeric protein (IRPI +73) showed adecreased IRE binding activity after iron treatment thatwas not increased by treatment with 2ME. When lysatesfrom iron-treated cells were assayed by Western blot, theamount of total mutant protein was decreased significantlyin the presence of iron (Figure SB). When assayed in apulse-chase experiment in the presence of iron, the IRPI/IRP2-specific exon chimeric protein (IRPI +73) was muchmore rapidly degraded than wild-type IRPI, and thedegradation rate was comparable with that of wild-typeIRP2 (t112 -3-6 h; Figure SC). Thus, the IRP2-specificexon conferred the capacity to be degraded at a ratesimilar to that seen in intact IRP2 upon IRPI.

The iron-dependent degradation signal of IRP2requires the participation of cysteines in theIRP2-specific exonCysteines are known to be critical in the ligation of the[4Fe-4S] cluster of IRPI. In that setting, the iron-sulfurcluster is ligated by three cysteines, C437, C503 andC506, and it is the presence of a fully assembled [4Fe-4S] cluster that prevents RNA binding (Hirling et al.,1994; Philpott et al., 1994). In addition, the mutation ofC437 to serine (C437S) or of all three cysteines (437, 503and 506) to serines results in the expression of a mutantprotein in which IRE binding is no longer regulated andthe mutant protein is permanently fixed in the IRE bindingmode (Philpott et al., 1994). To assess whether thecorresponding cysteines were required for the iron-dependent degradation of IRP2 or the IRPI/IRP2-specificexon chimeric protein (IRPI+73), a series of chimericproteins containing cysteine mutations were assessed. Asshown in Figure 6, lanes 5-8, the mutation of one orseveral of the cysteines required for the cluster ligationof IRP1 did not interfere with the iron-induced degradationof chimeric IRPl/IRP2-specific exon chimeric constructs.Thus, the iron-sensing capability of the IRP2 insert doesnot depend on the participation of previously identifiedcysteine ligands of the iron-sulfur cluster.

5351

K.iwai, R.D.Klausner and T.A.Rouault

IRP2

0 3 6 9 12 24(hr)

200kD-

98kD-68kD-

43kD-

200kD-

98kD-68kD-

43kD-

IRP1

0 3 6 9 12 24(hr)

200kD-D D

-0 98kD- __468kD-

43kD-

200kD-F F

-00- 98kD- ___---

68kD-

43kD-

Fig. 2. Comparison of degradation rates of recombinant wild-type IRPI and IRP2 in cells treated with an iron chelator or iron source. RD4 cellswere induced to express recombinant IRPI or IRP2 with dexamethasone. After a 16 h pretreatment with desferrioxamine (D) or FAC (F), a 1 h pulsewith Trans35[S]Label was followed by harvesting of the radiolabeled cells at the indicated time points. Lysates were immunoprecipitated with anti-myc antibody. Arrows point to IRPI or IRP2 signals.

The presence of five cysteines in the IRP2-specificexon, with the spacing C(X30)C(X5)C(X3)C(X23)C, offeredanother set of cysteines which could be assessed for theirrole in iron sensing. The close physical grouping of thethree middle cysteines of the exon is noteworthy, and wequestioned whether the cysteines might be acting inconcert to sense iron levels. To test whether cysteinesfrom the IRP2-specific exon were important in degradation,the degradation rate of the chimeric protein IRP1I/IRP2-specific exon (IRPI +73) was compared with that of asimilar construct, which differed only in that the middlethree cysteines (C168, C174 and C178) of the IRP2-specific domain were simultaneously converted to serinesby site-directed mutagenesis (exon 3C-S). When the threecysteines were mutated to serines, the IRP2-specific exoncould no longer transfer the capacity for rapid iron-dependent degradation to IRPI, as shown in Figure 6,lanes 9 and 10. Levels of expression of the differentconstructs differed markedly in stable cell lines, and alonger exposure in which non-specific bands are moreprominent was required to visualize the exon 3C-S mutantin lanes 9 and 10. A pulse-chase experiment revealed thatthe exon 3C-S mutant was stable, with an estimated tj/2of 24 h (data not shown). Gel retardation assays of allmutant chimeras indicated that IRE binding activity wasintact and that the tertiary structure was therefore funda-mentally intact (data not shown). Thus it appears thatcysteines are critical for the presentation of the signal foriron-dependent rapid degradation, possibly because thecysteines may be involved in the direct ligation of iron oran iron-sulfur cluster.

domain 1 domain2 domaln3 domain4IRPI

C CC

IRP2

IRP1ARP2specific exonchimera p777,

I47 \ 56% 64% / \ 65%

\ hinge-linkerIrzz. 1 73 amino acid domainc ccc c

7.77777777777777.

(IRPI +73)

IRP2 cccccspecific exondeleted

(IRP2-73)

Fig. 3. Schematic comparison of IRPI and IRP2 showing proteinscreated by the excision of the IRP2-specific exon (IRP2-73) and thecreation of an IRPI-IRP2 chimera by the insertion of the IRP2-specific exon (IRP1+73) Percentage sequence identities between thedomains of IRPI and IRP2 are indicated. The position of the threecysteines that ligate the iron-sulfur cluster in IRPI are indicated, asare the five cysteines present in the IRP2-specific 73 amino acid exon.

Inhibition ofproteasome function in vivo preventsiron-dependent degradation of IRP2To gain insight into the mechanism of degradation, iron-replete cells were treated with a variety of reagents knownto interfere with various modes of proteolysis in cells.Reagents shown previously to interfere with lysosomalfunction, including E64 and leupeptin (Palombella et al.,1994), did not interfere with the degradation of IRP2, as

5352

u i


IRP2 IRP2B

mli IRP2

D F D FIR2IRP2 _ _ _-73

0 3 6 9 (hr) IRP1+ 73C437S

IRP1 ~~C503SIRP1 C437S C506SD F D F D F D F

Fig. 4. Deletion of the IRP2-specific exon from IRP2 (IRP2-73)eliminates rapid degradation. RD4 cells were induced to expressrecombinant proteins with dexamethasone. After 16 h of treatmentwith desferrioxamine (D) or FAC (F), cells were harvested.(A) Quantitation by Western blot analysis. Lysates (30 jg) wereseparated on a 6% polyacrylamide gel, transferred to nitrocellulosefilters and probed with anti-myc antibody. (B) Evaluation by pulse-chase experiments of the stability of IRP2 in FAC compared withthe IRP2-73 deletion in FAC. After a 16 h pretreatment withdesferrioxamine (D) or FAC (F), a 1 h pulse with Trans35[S]Label wasfollowed by harvesting of the radiolabeled cells at the indicated timepoints. Lysates were immunoprecipitated with anti-myc antibody.

A IRP1 IRP1+73

D F D F1% 2ME -

+ - + - + - +

1 2 3 4 5 6 7 8 9 10

Fig. 6. The mutation of cysteines in the IRP2-specific exon eliminatesthe ability of the exon to confer rapid iron-dependent degradation toIRPI, whereas the mutation of three other IRPI cysteines does notinterfere with iron-dependent degradation. After the treatment of cellswith either desferrioxamine (D) or FAC (F), lysate waselectrophoresed, transferred and probed as described in Materials andmethods. Levels of chimeric IRP1+73 are reduced markedly aftertreatment with an iron source (lane 4), whereas the levels of the IRPIparent construct are relatively unaffected by treatment with FAC. Themutation of the cysteines in the IRPI portion of the chimeric protein,including C437S, and a construct in which cysteines 437, 503 and 506are all mutated to serines does not interfere with rapid iron-dependentdegradation (lanes 5-8). A chimeric construct in which cysteines 168,174 and 178 are mutated to serines is stable after iron treatment (lanes9 and 10).

0 3 6 9 (hr)

___m

_ IRP1---m --. .......+73

Fig. 5. Ligation of the 73 amino acid IRP2-specific exon to IRPI(IRP1+73) transfers the rapid degradation phenotype to IRP1. Aftertreatment with desferrioxamine and FAC, as described in the legend toFigure 4, RD4 cells were induced to express recombinant proteinswith dexamethasone. (A) Quantitation by a supershift assay.

(B) Quantitation by a Western blot analysis. (C) Quantitation in apulse-chase experiment as described in the legend to Figure 4.

shown in Figure 7. Similarly, degradation was unaffectedby treatment with calpain inhibitor II, an inhibitor ofthe cytosolic neutral Ca-protease, calpain. The peptidealdehyde MG132 (carbobenzoxyl-leucinyl, leucinal-leucinal H; kindly provided by Myogenics Inc.,Cambridge, MA), a potent inhibitor of the 20S subunit ofthe proteasome, interfered with the degradation of IRP2when cells were treated with iron. After the treatment ofcells with the above described proteolysis inhibitors, an

assay ofIRE binding activity of IRP2 (Figure 7A) revealedthat in the presence of FAC, IRP2 binding was absent,except in those cells treated with MG132 where IREbinding activity remained high. In Figure 7B, Westernblotting of the same lysates used in the supershift assaysin Figure 7A showed that IRP2 levels remained high inthe presence of MG 132, but were degraded in the presenceof other proteolysis inhibitors. MG132 has been shownpreviously to block the function of calpain and cathepsinB (Rock et al., 1994), as well as proteasome function.However, here we have shown that the inhibition of

calpain or lysosomal proteases did not inhibit the processof iron-dependent degradation of IRP2, thus showing thatthe effect ofMG132 in this setting most probably dependedon its ability to inhibit proteasome function.To confirm further the role of the proteasome in the

degradation of IRP2, a more specific inhibitor of protea-some function, lactacystin (kindly provided by Dr SatoshiOmura, Kitasato Institute, Tokyo, Japan), was used. Lacta-cystin inhibits proteasome function by covalently bindingto the N-terminal threonine of the mammalian proteasomesubunit X. The effects of lactacystin in the cell are morespecific for the proteasome than are those of MG132(Fenteany et al., 1995), and we found that lactacystininhibits the degradation of IRP2 at a level comparablewith MG132 (Figure 7B). To confirm that a decrease indegradation rather than a change in the level of synthesiswas the cause of increased protein levels, pulse-chaseexperiments were performed which confirmed thatdegradation was inhibited (data not shown). Thus, thedegradation of IRP2 is most probably mediated by theproteasome.

IRE binding activity of IRP2 is not regulated in RD4cells when degradation is inhibitedWhereas there was almost no binding activity of wild-type IRP2 in the supershift assay depicted in Figure 8after the treatment of cells with FAC, the binding activitywas high in cells expressing the IRP2 exon deletion mutant(IRP2-73) after treatment with FAC. Previous studies haveshown that certain cell types, including HeLa cells, lackthe capacity to degrade IRP2, but are still able to modulatethe IRE binding activity of IRP2 in relation to ironconcentrations (Samaniego et al., 1994). When thedegradation of IRP2 was inhibited by the deletion of theIRP2-specific exon from recombinant IRP2 in RD4 cells,we thought it possible that IRE binding would be decreasedby the treatment of cells with iron without a change inprotein half-life, reminiscent of the regulation observed

5353

AEXON3C-SD F

F3 IRPi IRP1 C

IRP1+73D F DF IRPi

K.iwai, R.D.Klausner and T.A.Rouault

o._

cm .c

wtir C.)

_La Q'etI n

IRP2 IRP2-73D F D F

1% 2ME - + + - + - +

., ~ ~~~~,, ,J0

0

+ + + + + Fig. 8. Deletion of the IRP2-specific exon from IRP2 (IRP2-73)

C D F F F F F F eliminates rapid degradation, and IRE binding activity of the stabledeletion mutant is no longer regulated by iron. RD4 cells were1% 2ME + + + - + + - + - + - + induced to express recombinant proteins with dexamethasone. After

16 h of treatment with desferrioxamine (D) or FAC (F), cells wereharvested and quantitated by a supershift assay. Lysates (15 jg) were

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 incubated with anti-myc antibody, followed by incubation withradiolabeled IRE in the presence (+) or absence (-) of 2ME. Mixturesof lysates and radiolabeled probe were electrophoresed on an 8%native polyacrylamide gel.

B L-

.0

-

CL >- r-

f)a0.) co

W 2 j(.

+ + + + + +

D FF FF FF F

.. !..:., _ _ W."

2 3 4 5 6 7 8

Fig. 7. Iron-dependent degradation of IRP2 in vivo is blocked byproteasome inhibitors. RD4 cells were induced to express recombinantIRPI or IRP2 with dexamethasone. Cells were grown in normalmedium (C) or treated with desferrioxamine (D) or FAC (F) for 12 h.(A) Quantitation by a supershift assay. Lysates (15 jg) were incubatedwith anti-myc antibody, followed by incubation with radiolabeled IREin the presence (+) or absence (-) of 2ME. Mixtures of lysates andradiolabeled probe were electrophoresed on a 8% nativepolyacrylamide gel. In the presence of FAC, cells were either nottreated with protease inhibitors (lanes 5 and 6) or treated withleupeptin (500 ,ug/ml; lanes 7 and 8), E64 (100 gM; lanes 9 and 10),MG132 (50 jM; lanes 11 and 12), calpain inhibitor 11 (100 ,uM; lanes13 and 14) or dimethylsulfoxide (0.25%) as a solvent control (lanes 15and 16). Subsequent to these treatments, cells were harvested andlysates were prepared as described in Materials and methods.(B) Quantitation by a Western blot analysis. Lysates (30 jig) were

separated on a 6% SDS-polyacrylamide gel, transferred tonitrocellulose filters and probed with anti-myc antibody. Treatments areas described above, with the addition that cells in lane 6 were treatedwith lactacystin (10 jiM) for 12 h.

in HeLa cells. However, as seen in Figure 8, IRE bindingactivity remained high when the cis-acting degradationsignal was removed. Similarly, when cells were treatedwith the proteasome inhibitors MG132 (Figure 7A) orlactacystin (data not shown), binding remained high even

though cells were treated with iron. Thus it appears thatIRP2 binding activity is strictly regulated by degradationin these cells, and that these cells are unable to imposean observable IRPi-like mode of regulation when thedegradation process is inhibited.

DiscussionThe coexistence of two IRPs in mammalian cells raisesinteresting questions about the role of these two proteins.Both proteins bind consensus IREs with equal affinity and

specificity. Previous studies have shown that while IRPIis a stable protein (Tang et al., 1992; Pantopoulos et al.,1995), its associated iron-sulfur cluster is labile. IRPI isreversibly modified by the post-translational assembly ofan iron-sulfur cluster, and the status of the iron-sulfurcluster determines the function; IRPI is a functionalenzyme (cytosolic aconitase) when the iron-sulfur clusteris intact, or an RNA binding protein when the cluster isabsent (Haile et al., 1992a,b). The regulation of IRP2differs from that of IRPI in that IRP2 is rapidly degradedin most cell types when cells are iron-replete. Cell typesin which the non-reversible iron-dependent loss of IRP2has been described include RD4 cells, a liver cell line,FT02B cells (Guo et al., 1994), Ltk- cells (Hendersonet al., 1993) and B6 cells (Pantopoulos et al., 1995). Herewe have established that a unique sequence of IRP2, anadditional 73 amino acid domain, is the major determinantfor rapid degradation in the presence of iron. Deletion ofthis exon turns IRP2 into a relatively stable form of theprotein, and the transfer of this domain to IRPI is sufficientto turn IRPI into a rapidly degraded protein in iron-repletecells (t1/2 of 3-6 versus 24 h).We have shown that the degradation ofIRP2 is prevented

by the inhibition of proteasome function. IRP2 may belongto a growing list of cellular proteins that are targeted forproteasomal degradation by ubiquitination, although wehave not as yet been able to detect higher molecular weightproducts that could represent ubiquitinated intermediates inlysates of cells treated with inhibitors of proteasomefunction. The number of specific examples in whichthe ubiquitin-dependent pathway has been shown to beimportant is growing and includes the processing of IBcBa(Palombella et al., 1994), the generation of antigenicpeptides for major histocompatibility complex class Ipresentation (Rock et al., 1994), the turnover of p53(Scheffner et al., 1994), the degradation of cyclins (Glotzeret al., 1991) and the degradation of MATa2 repressor inyeast (Chen et al., 1993). In fact, proteasome-mediatedproteolysis has been implicated recently in the degradationof the bulk of short- and long-lived proteins (Rock et al.,1994). Interestingly, some proteins also appear to bedegraded by the proteasome in the absence of priorubiquitination (Murakami et al., 1992).Our results indicate that the IRP2-specific exon can

function as a transferable iron-sensitive 'degradationdomain'. Exons often encode independent stable protein

5354

A

_CL

-j

*NW

1


folding domains (Sharp, 1994), and the fact that thesequence outside the IRP2-specific exon otherwise closelyaligns with the sequences of IRPI, porcine mitochondrialaconitase and the aconitase of Escherichia coli (Rouaultet al., 1992) makes it likely that the 73 amino acid insertfunctions as an independent domain. The degradationdomain of IRP2 offers an interesting candidate for thefurther study of regulated degradation because the IRP2-specific exon predisposes to rapid degradation only whencells are iron-replete. Starvation for iron results in thestabilization of IRP2, analogous to the stabilization ofGCN4 observed when yeasts are subjected to amino acidstarvation (Komitzer et al., 1994).A further distinctive feature of the degradation domain

is that it contains five cysteines, one of which is precededby a proline (P167, C168) and one of which is followedby a proline (C201, P202), a common finding in cysteinesinvolved in the ligation of iron-sulfur clusters (Matsubaraand Saeki, 1992). Here we have shown that the mutationof three cysteines in the IRP2-specific exon eliminatesthe ability of this sequence to transfer iron-dependentdegradation to IRPI. Cysteines may be key to the abilityto register changes in cellular iron concentrations becauseof the inherent capacity of the thiolate anion to bind ironin various forms. One or more of the cysteines of theIRP2-specific exon is clearly required for degradation,whereas mutagenesis in the chimeric constructs showsthat the three cysteines that ligate the iron-sulfur clusterof IRPI (C437, C503 and C506) are not required forexpression of the degradation signal. While other cysteinesthat are conserved between IRPI and IRP2 may alsocontribute to the degradation signal, cysteines of the IRP2-specific exon are indispensable.

Characterization of the clearly distinct mechanism ofiron sensing by IRP2 demonstrated by cysteine muta-genesis has enabled us to observe further important distinc-tions about the cellular control of the RNA binding activityof IRPI and IRP2. If IRP2 was able to ligate an iron-sulfur cluster comparable with the iron-sulfur cluster ofIRP1, we would predict that the stable exon deletionmutant of IRP2 (IRP2-73) would be capable of regulatingRNA binding in the absence of protein degradation,analogous to IRP1. However, we found that while theremoval of the IRP2-specific exon resulted in a loss ofthe ability to degrade the protein in the presence of iron,as expected, the stable IRP2 deletion mutant (IRP2-73)was unable to inhibit RNA binding in iron-replete cells,as indicated in Figure 8. Similarly, the inhibition ofproteasome function in RD4 cells revealed that there wasno regulation of RNA binding when the degradation ofIRP2 was inhibited. These results raise the possibility thatIRP2 is regulated via protein degradation in most cellsprecisely because the sensing of iron concentrations byIRP2 is not coupled to the regulation of RNA binding. Incontrast to IRPI, the iron-sensing process of IRP2 doesnot appear to impinge directly on the RNA binding site,and iron sensing and IRE binding are therefore notmutually exclusive activities. These differences in theiron-sensing process between IRPI and IRP2 may provideclues as to why two distinct IRE binding proteins existin cells.

While the effects of iron sensing on RNA bindingare different in IRPI and IRP2, it is interesting to note

that the RNA binding sites also differ in severaldefinable ways. The RNA binding site of IRP2, althoughcapable of binding consensus IREs with equal affinity,can bind alternative ligands that are not bound by IRPI;conversely, IRPI can bind ligands that are not boundby IRP2 (Henderson et al., 1994; J.Butt, H.Y.Kim,J.B.Basilion, R.D.Klausner and T.A.Rouault, manuscriptin preparation). Furthermore, the impact of the mutationof residues homologous to the RNA binding residuesof IRPI (Philpott et al., 1994) on the binding of RNAligands to IRP2 is clearly distinguishable (Butt et al.,manuscript in preparation). Thus, not only is the iron-sensing mechanism different, but the RNA binding sitesare distinguishable.The simplest hypothesis to explain the iron sensing of

IRP2 is to postulate that the cysteines within the IRP2-specific exon bind iron directly, in either the ferric orferrous form, in an iron-sulfur cluster or some othercomplex form. The direct binding of iron can lead tometal-catalyzed oxidation reactions in the vicinity ofthe bound metal, and the proteins that contain residuesdamaged by oxidation are targeted for degradation by theproteasome (Stadtman and Oliver, 1991; Grune et al.,1995). In addition, the integrity of iron-sulfur clusters hasbeen established as an important key to the stabilityof several proteins. One such protein, the glutamineamidophosphoribosyl transferase of Bacillus subtilis, con-tains an [4Fe-4S] cluster in the C-terminus in a regionthat is important in interactions between subunits in thetetrameric enzyme (Smith et al., 1994). The cluster islabile to oxidative stress, and cluster destruction appearsto result in changes in the tertiary structure of the enzymethat unmask degradation signals which are otherwisesequestered by the tertiary structure (Switzer, 1989).Similarly, the [4Fe-4S] cluster of endonuclease III appearsto stabilize a DNA binding motif and contribute to thearchitectural integrity of the protein (Kuo et al., 1992). Inthe case of the E.coli transcription factor fumarate nitratereduction, the integrity of an associated oxidation-sensitiveiron-sulfur cluster appears to be required for maintenanceof the active dimeric form of the protein, providing yetanother example in which an iron-sulfur cluster determinesstructure and function (Khoroshilova et al., 1995). Futurestudies of IRP2 will be directed towards furthercharacterizing the iron degradation domain. An importantfocus will be to determine the metal binding statusof IRP2 and the role of the cysteines of the IRP2degradation domain.

Materials and methodsConstruction of mutantsDeletion of the IRP2-specific sequence (IRP2-73). Nucleotides whichencode a sequence of 73 amino acids that corresponds to an entire exonin the genomic sequence were deleted by PCR. The deleted amino acidsequence was CAIQNAPNPGGGDLQKAGKLSPLKVQPKKLPCRG-QTTCRGSCDSGELGRNSGTFSSQIENTPILCPFHLQPVP. The aminoacid sequence after deletion was D131FSK134E210PETVL215.Insertion of the IRP2-specific exon into IRPI (IRPJ+73). The exonsequence above was cloned into IRP1 between R134 and R135 of theIRPI sequence (Rouault et al., 1992) by a two-step PCR (Philpott et al.,1994). The resulting constructs were cloned into the episomal vectorp220 under the control of the GRE5 promoter, as described previously(Samaniego et al., 1994), and stable transformants of RD4 cells wereestablished.

5355

K.Iwai, R.D.Klausner and T.A.Rouault

Site-directed mutagenesis of cysteines in the IRP2-specific exon.Cysteines 168, 174 and 178 (Rouault et al., 1992) were converted toserines using a two-step PCR and ligated to myc epitope-tagged IRPI(Kaptain et al., 1991) between R134 and R135, as described above. Theresulting constructs were cloned into the episomal vector p220 underthe control of the GRE5 promoter as described previously (Samaniegoet al., 1994).

Cell linesThe human rhabdomyosarcoma cell line RD4 was maintained inDulbecco's modified Eagle's medium (DMEM) supplemented with 10%fetal bovine serum (FBS; Biofluids, Rockville, MD), 100 IU/ml penicillinG and 100 gg/ml streptomycin. Stable transformants of RD4 cells wereestablished by the transfection of expression constructs of IRPI andIRP2 and their mutants using the calcium phosphate method followedby selection in the presence of hygromycin B (250 gg/ml; Calbiochem).Clonally isolated transformants were established.

Biosynthetic labelingBiosynthetic labeling was performed as described previously (Samaniegoet al., 1994). RD4 cells stably transformed with myc epitope-taggedIRPs and their mutants were induced with dexamethasone (20 nM) toexpress recombinant proteins for 32 h, and then treated with eitherdesferrioxamine (100 jiM) or FAC (100 gg/ml) for 16 h in the continuouspresence of dexamethasone. Cells were metabolically radiolabeled with0.1 mCi Trans35[S]Label (ICN) in cystine- and methionine-free DMEMsupplemented with 5% dialyzed FBS for 1 h in the continued presenceof dexamethasone and either desferrioxamine or FAC. After labeling,cells were washed twice with complete medium, and chased in thepresence of either desferrioxamine or FAC in the absence of dexametha-sone. Cells were harvested at 0, 3, 6, 9, 12 and 24 h time points, andlysates were immunoprecipitated with anti-myc antibody and separatedusing 8% SDS-PAGE. Quantitation was performed using a Phosphor-Imager (Molecular Dynamics, Sunnyvale, CA.)

Western blotsWestern blots of recombinant proteins were performed as describedpreviously (Samaniego et al., 1994). Briefly, lysates (30 jg) from 48 hdexamethasone-induced RD4 stable transformants were separated ona 6% SDS-PAGE gel and transferred to nitrocellulose membranes.Membranes were blocked and incubated with anti-myc antibody (9E10),followed by incubation with rabbit anti-mouse IgG (Cappel), andvisualized with 125I-labeled donkey anti-rabbit Ig (Amersham).

Gel retardation and supershift assaysGel retardation and supershift assays were performed as describedpreviously (Philpott et al., 1994). In supershift assays, lysate (15 jig)from stable transfectants of RD4 cells was incubated with anti-mycantibody on ice for 30 min. Radiolabeled IRE was added and the mixturewas separated on an 8% native polyacrylamide gel.

ReferencesChen,P., Johnson,P., Sommer,T., Jentsch,S. and Hochstrasser,M. (1993)

Multiple ubiquitin-conjugating enzymes participate in the in vivodegradation of the yeast MATa2 repressor. Cell, 74, 357-369.

DeRusso,P.A., Philpott,C.C., Iwai,K., Mostowski,H.S., Klausner,R.D. andRouault,T.A. (1995) Expression of a constitutive mutant of ironregulatory protein 1 abolishes iron homeostasis in mammalian cells. J.Biol. Chem., 270, 15451-15454.

Fenteany,G., Standaert,R.F., Lane,W.S., Choi,S., Corey,E.J. andSchreiber,S.L. (1995) Inhibition of proteasome activities and subunit-specific amino-terminal threonine modification by lactacystin. Science,268,726-731.

Glotzer,M., MurrayA.W. and Kirschner,M.W. (1991) Cyclin is degradedby the ubiquitin pathway. Nature, 349, 132-138.

Grune,T., Reinheckel,T., Joshi,M. and Davies,K.J. (1995) Proteolysis incultured liver epithelial cells during oxidative stress. Role of themulticatalytic proteinase complex, proteasome. J. Biol. Chem., 270,2344-2351.

Guo,B., Yu,Y and Leibold,E.A. (1994) Iron regulates cytoplasmic levelsof a novel iron-responsive element-binding protein without aconitaseactivity.v. Biol. Chem., 268,24252-24260.

Haile,D.J., Hentze,M.W., Rouault,T.A., Harford,J.B. and Klausner,R.D.(I1989) Regulation of interaction of the iron-responsive element binding

protein with iron-responsive RNA elements. Mol. Cell. Biol., 9, 5055-5061.

Haile,D.J., Rouault,T.A., Harford,J.B., Kennedy,M.C., Blondin,G.A.,Beinert,H. and Klausner,R.D. (1992a) Cellular regulation of the iron-responsive element binding protein: disassembly of the cubane iron-sulfur cluster results in high affinity RNA binding. Proc. Natl Acad. Sci.USA, 89, 11735-11739.

Haile,D.J., Rouault,T.A., Tang,C.K., Chin,J., Harford,J.B. andKlausner,R.D. (1992b) Reciprocal control ofRNA binding and aconitaseactivity in the regulation of the iron responsive element binding protein:role ofthe iron-sulfur cluster. Proc. NatlAcad. Sci. USA, 89, 7536-7540.

Henderson,B.R., Seiser,C. and Kuhn,L.C. (1993) Characterization of asecond RNA-binding protein in rodents with specificity for iron-responsive elements. J. Biol. Chem., 268, 27327-27334.

Henderson,B.R., Menotti,E., Bonnard,C. and Kuhn,L.C. (1994) Optimalsequence and structure of iron-responsive elements- selection ofRNAstem-loops with high-affinity for iron regulatory factor. J. Biol. Chem.,269, 17481-17489.

Hentze,M.W., Rouault,T.A., Harford,J.B. and Klausner,R.D. (1989)Oxidation-reduction and the molecular mechanism of a regulatoryRNA-protein interaction. Science, 244, 357-359.

Hirling,H., Henderson,B.R. and Kuhn,L.C. (1994) Mutational analysis ofthe [4Fe-4S]-cluster converting iron regulatory factor from its RNA-binding form to cytoplasmic aconitase. EMBO J., 13, 453-461.

Kaptain,S., Downey,W.E., Tang,C., Philpott,C.C., Haile,D.J., Orloff,D.G.,Harford,J.B., Rouault,T.A. and Klausner,R.D. (1991) A regulated RNAbinding protein also possesses aconitase activity. Proc. Natl Acad. Sci.USA, 88, 10109-10113.

Khoroshilova,N., Beinert,H. and Kiley,P.J. (1995) Association of apolynuclear iron-sulfur center with a mutant FNR protein enhancesDNA binding. Proc. Natl Acad. Sci. USA, 92, 2499-2503.

Klausner,R.D., Rouault,T.A. and Harford,J.B. (1993) Regulating the fateof mRNA: the control of cellular iron metabolism. Cell, 72, 19-28.

Kornitzer,D., Raboy,B., Kulka,R.G. and Fink,G.R. (1994) Regulateddegradation of the transcription factorGCN4. EMBO J., 13, 6021-6030.

Kuhn,L.C. (1994) Molecular regulation of iron proteins. Baillieres Clin.Haematol., 7, 763-785.

Kuo,C.F., McRee,D.E., Fisher,C.L., O'Handley,S.F., Cunningham,R.P.and Tainer,J.A. (1992) Atomic structure of the DNA repair [4Fe-4S]enzyme endonuclease. Science, 258,434-440.

Matsubara,H. and Saeki,K. (1992) Structural and functional diversity offerredoxins and related proteins. Adv. Inorg. Chem., 38, 223-280.

Melefors,O. and Hentze,M.W. (1993) Iron regulatory factor - theconductor of cellular iron regulation. Blood Rev., 7, 251-258.

Murakami,Y., Matsufuji,S., Kameji,T., Hayashi,S., Igarashi,K., Tamura,T.,Tanaka,K. and Ichihara,A. (1992) Ornithine decarboxylase is degradedby the 26S proteasome without ubiquitination. Nature, 360, 597-599.

Palombella,V.J., Rando,O.J., Goldberg,A.L. and Maniatis,T. (1994) Theubiquitin-proteasome pathway is required for processing the NF-kappaBI precursor protein and the activation of NF-kappa B. Cell, 78, 773-785.

Pantopoulos,K., Gray,N.K. and Hentze,M.W. (1995) Differentialregulation of two related RNA-binding proteins, iron regulatory protein(IRP) and IRPB. RNA, 1, 155-163.

Philpott,C.C., Klausner,R.D. and Rouault,T.A. (1994) The bifunctionaliron-responsive element binding protein/cytosolic aconitase: the role ofactive-site residues in ligand binding and regulation. Proc. Natl Acad.Sci. USA, 91, 7321-7325.

Rock,K.L., Gramm,C., Rothstein,L., Clark,K., Stein,R., Dick,L.,Hwang,D. and Goldberg,A.L. (1994) Inhibitors ofthe proteasome blockthe degradation of most cell proteins and the generation of peptidespresented on MHC class I molecules. Cell, 78, 761-771.

Rouault,T.A. et al. (1992) An iron-sulfur cluster plays a novel regulatoryrole in the iron-responsive element binding protein. BioMetals, 5,131-140.

Samaniego,F., Chin,J., Iwai,K., Rouault,T.A. and Klausner,R.D. (1994)Molecular characterization of a second iron responsive element bindingprotein, iron regulatory protein 2 (IRP2): structure, function and post-translational regulation. J. Biol. Chem., 269, 30904-30910.

Scheffner,M., Huibregtse,J.M. and Howley,P.M. (1994) Identification of ahuman ubiquitin-conjugating enzyme that mediates the E6-AP-dependent ubiquitination of p53. Proc. Natl Acad. Sci. USA, 91, 8797-8801.

Sharp,P.A. (1994) Split genes and RNA splicing. Cell, 77, 805-815.Smith,J.L., Zaluzec,E.J., Wery,J.P., Niu,L., Switzer,R.L., Zalkin,H. and

Satow,Y.(1994) Structure of the allosteric regulatory enzyme of purinebiosynthesis. Science, 264, 1427-1433.

5356


Stadtman,E.R. and Oliver,C.N. (1991) Metal-catalyzed oxidation ofproteins. J. Biol. Chem, 266, 2005-2008.

Switzer,R.L. (1989) Non-redox roles for iron-sulfur clusters in proteins.BioFactors, 2, 77-86.

Tang,C.K., Chin,J., Harford,J.B., Klausner,R.D. and Rouault,T.A. (1992)Iron regulates the activity ofthe iron-responsive element binding proteinwithout changing its rate of synthesis or degradation. J. Bio. Chem.,267,24466-24470.

Received on July 20, 1995

5357

Requirements for iron-regulated degradation of the RNA binding ...

Documents