Journal of Neurruhernist~yLippincott—Ravcn Publishers, Philadelphia

© 1996 International Society for Neurochemistry

Demonstration and Characterization of the Iron RegulatoryProtein in Human Brain

Jing Hu and James R. Connor

George M. Leader Family Laboratory for Alzheimer’s Disease Research, Department of Neuroscience and Anatomy,Penn State University College of Medicine, M. S. Hershey Medical Center, Hershey, Pennsylvania, U.S.A.

Abstract: Iron regulatory proteins (IRPs) are cytoplasmicRNA binding proteins that regulate expression of ferritin,erythroid 5-aminolevulinic acid synthase, and transferrinreceptor through interaction with conserved RNA stem—loop structures called iron-responsive elements (IREs).Two lRPs (IRP1 and IRP2) have been reported. In thepresent study we provide evidence for and initial charac-terization of the IRPs in human brain. Two RNA—proteincomplexes were obtained by RNA band shift assay oncytoplasmic extracts from human brain. Competitionstudies indicate that the formations of the RNA—proteincomplexes are specific to the IRE structure. UV cross-linking of brain cytoplasmic extracts with ferritin IRE RNAtranscripts revealed a single RNA—protein complex witha molecular mass of 110 kDa, A single band at 100 kDawas obtained with IRP1 antiserum on western blot analy-sis of brain cytoplasmic extracts, and a supershift in theRNA—protein complexes was observed with an IRP1 anti-serum. Two cDNA clones were isolated from a humanbrain cDNA library with IRP1 cDNA probes, and both ofthese cDNA probes recognized a single mRNA species(4.0 kb) from human astrocytoma cells. Purified humanbrain lAP protein has a molecular mass of ~-~100kDa andis capable of forming two RNA—protein complexes withferritin IRE RNA and reacts strongly with IRP1 antiserum.These data indicate that APi is predominant in the adulthuman brain and, in this tissue, is capable of forming adouble IRE/IRP complex. This latter observation sug-gests the brain IRP undergoes posttranslational modifi-cation, the result of which may influence the stability ofthe IRE/IRP complex. Key Words: Brain—Iron—Iron-responsive element— Iron regulatory protein—Ferritin—Transferrin receptor.J. Neurochem. 67, 838—844 (1996).

do all cells for many aspects of their cell physiology,including electron transport, NADPH reductase activ-ity, myelination of axons, and as a cofactor for severalenzymes involved in neurotransmitter synthesis (Lar-kin and Rao, 1990). An imbalance in brain iron andhence a dysfunction in iron-related metabolism are sus-pected in some neurological disorders, including Par-kinson’s disease and Alzheimer’s disease (Ehmann etal., 1986; Connor et al., 1992; Good et al., 1992).

Because of the blood—brain barrier, cells in the braindo not have direct access to circulating iron. Iron mustbe transported across the blood—brain barrier andstored in a bioavailable form. The same elegant systeminvolving the transferrin receptor and ferritin that existsin systemic organ exists in brain cells for regulatingiron concentration and availability (Connor et al.,1990; Connor and Menzies, 1995). By controlling thelevel of expression of these latter two proteins, thecells can determine the amount of iron acquired andsequestered. The coordinated control of ferritin andtransferrin receptor by cellular iron occurs at the post-transcriptional level and is mediated by cytoplasmicRNA binding proteins, known as the iron regulatoryproteins (IRPs) (Klausner et al., 1993).

The IRPs are trans-acting cytoplasmic proteins thatbind to highly conserved RNA structures identified asiron-responsive elements (IREs) (Theil, 1994). Fern-tin mRNA contains a single IRE in the 5’ untranslatedregion (UTR) (Aziz and Munro, 1987; Hentze et a!.,1987), where binding of the IRP prevents binding ofthe 43S translation preinitiation complex to the mRNA,thereby blocking initiation of translation (Gray andHentze, 1994). In contrast, transferrin receptor mRNA

The importance of iron for normal neurologicalfunction is well established (Beard et al., 1993). Ironlevels are high in the brain at birth, and iron uptakeinto the brain is a continual process throughout life(Roskams and Connor, 1994). The concentration ofiron in the brain is region dependent, and in the basalganglia iron is present at a concentration equal to thatfound in the liver (Hallgren and Sourander, 1958; Con-nor et a!., 1992). Neurons and glia require iron as

Received October 10. 1995; final revised manuscript receivedMarch 22, 1996; accepted March 28, 1996.

Address correspondence and reprint requests to Dr. J. R. Connorat George M. Leader Family Laboratory for Alzheimer’s DiseaseResearch, Department of Neuroscience and Anatomy, Penn StateUniversity College of Medicine, M. S. Hershey Medical Center,Hershey, PA 17033, U.S.A.

Abbreviations used: IRE, iron-responsive element; IRP, iron regu-latory protein; SDS-PAGE, sodium dodecyl sulfale—polyacrylamidegel electrophoresis; UTR, untranslated region.

838

IRP IN HUMAN BRAIN 839

contains five IREs in the 3’ UTR, and IRP binding isbelieved to increase stability of the mRNA (Casey etal., 1989; Mullner et al., 1989). In addition, IREs havebeen reported in the 5’ UTRs of erythroid 5-aminolev-tilinic acid synthase (Cox et al., 1991; Dandekaret al.,1991) and mitochondrial aconitase (Zheng et al.,1992) mRNAs, implicating the IRP in regulation ofcrythroid protoporphyrin biosynthesis and the citricacid cycle.

~I’woIRPs have been identified that are now termedIRPI and IRP2 (Guo et al., 1994; Samaniego et al.,1994). IRP I, which has also been reported as the ironregulatory factor, fernitin repressor protein, and IREbinding protein, is conserved during evolution. IRP2,also termed IRE-BP B2 and IRFB, has been identifiedIll rodent tissues and human HeLa cells (Henderson etal., 1993; Guo et a!., 1994, 1995a,b; Samaniego etal.. 1994). IRP2 bears an amino acid sequence 57%identical and 75% similar to that of IRP1 (Rouault etal., 1990). The gene for IRP2 has been mapped tochromosome 15, whereas thegene for IRP1 is localizedon chromosome 9 (Rouault et al., 1990). The func-tional distinctions between these two proteins are Un-known, but they appear to be differentially regulatedby iron (Guo et al., 1994, 1995a,b; Samaniego et al.,1994). In this study we provide the initial demonstra-lion and identification of the IRPs in human brain.

MATERIALS AND METHODS

liasmid constructs and preparation of RNAtranscripts

‘l’he pGEM 7Zf(+)-5 ‘L containing the complete5’ UTRof rat liver L chain ferritin eDNA (Bettany et a!., 1992)was provided by R. S. Eisenstein, University of Wisconsin—Madison. The pTfR-IREs was generatedby subeloning 680hp corresponding to 3,370—4,048 in the published sequence(Schneider et al., 1984) from a human transferrin receptorcI )NA (pCDTR- I; purchased from the American Type Cu!-lure Collection) into pGEM 3Z (Promega). The pTfR-IREspoly (A) was generated from pTfR-IREs by replacing a Hin-cll/HindlII fragment with a modified XbaI/BamHI fragmentfront the 3’ end of the human TfR eDNA clone pCDTR-leulllaining 270 bp of polyadenylated sequences (Neupert ettI., 1990). The ferritin IRE RNA transcripts were preparedluau S’mal-digested pGEM 7Zf(+)-5’L containing 100bases of 5’ UTR and 40 bases from the vector. The TfRIRFs RNA transcripts, 207 bases prepared from Dral-di-gcslcd pTfR-IREs, contained 167 bases, including IRE Aand B sequences, and 40 bases from the vector. The TfR1REs-poly(A) RNA transcripts were generated from pTfR-IRNs poly(A) with HindlIl digestion. RNA transcripts weresynlhesized from 1 jig of linearized plasmid DNA in thepresence of 100 jiCi of [a-32P1CTP (800 Ci/mmol; Du-Poul), 12 pM unlabeled CTP, the other three unlabeled nu-denudes at 0.5 mM, 100 mM dithiothreitol, 20 units ofRNase inhibitor, and 20 units of T7 RNA polymerase (Pro-niega) in a 20-pl reaction volume. The reaction was carriedonl for I h at 37°C,and the yield of transcript synthesizedwas determined by trichloroacetic acid precipitation. Underlhcsc conditions, the specific activity was 5 x l0~cpmlpgol RNA. Full-length RNA transcripts were obtained by gel

purification on a 5% polyacrylamide/8 M urea gel. Large-scale transcription (Mega RNA transcription kit; Ambion)was performed according to the manufacturer’s instructionsincluding 5 jig of linearized plasmid DNA and 7.5 mMunlabeled ribonucleotides. 5S-CTP (I p1; Du Pont) wasadded to determine theyield of RNA. Unincorporated nude-otides were removed by aG-50 Micro Cotumn (Pharmacia).

Preparation of human brain cytoplasmic extractsHuman brain tissue was collected at autopsy, frozen, and

stored at —70°C.Brain tissue was dissected free of leptomen-inges and blood vessels, dissected into regions of interest,andthen washed in phosphate-buffered saline. Subsequently.the tissue was homogenized in 3 volumes of 10 mM HEPES(pH 7.6), 3 mM MgC1

2, 40 mM KC1, 5% glycerol, 1 mMdithiothreitol, and 0.2% Nonidet P-40 supplemented with 1mM phenylmethylsulfonyl fluoride and 5 pg/mI of leupeptin(Sigma) by an Ultra-Turrax T 25 (IKA Labortechnik) for10 s. Nuclei and debris were removed by centrifugation at10,000 g for 30 mm. The supernatant was collected andfurther centrifuged at 100,000 g for 60 mm, andthe resultingsupernatant was stored at —70°C.Protein concentration wasdetermined by the Bio-Rad assay.

RNA band shift assayBinding reactions were carried out as described by Leibold

and Munro (1988). Human brain cytoplasmic extract (10jig) was incubated with radiolabeled RNA transcripts (0.2nM) in 10 p1 of 10 mM HEPES (pH 7.6), 3 mM MgCl2,40 mM KCI, and 5% glycerol for 30 mm at room tempera-ture. RNase Ti (I unit; Boehringer Mannheim) was addedfor 10 mm, followed by heparin (Sigma) at 5 mg/mi for anadditional 10 mm. RNA—protein complexes were separatedon a 4% native polyacrylamide gel (acrylamide/bisacry!-amideratio, 60:1) for 2 h at 150 V (following preelectropho-resis for 30 mm at the same voltage). The gel was dried,and the RNA—protein complex formation was visualized byautoradiography.

UV cross-linking of the RNA — protein complexHuman brain cytoplasmic extract (60 jig) was incubated

with radiolabeled fernitin IRE RNA transcript (2 nM) in a20-pI volume for 30 mm at room temperature, followedby treatment with RNase TI and heparin. The sample wastransferred to a 96-well microplate on ice, covered by plasticmembrane, and UV-irradiated at a distance of 4 cm for 45mm (Stratalinker 1800; Stratagene).

Western blot analysisProteins from human brain cytoplasmic extracts (60 jig)

were separated on 7.5% sodium dodecyl sulfate—poly-acrylamide gel electrophoresis (SDS-PAGE). The proteinwas transferred to nitrocellulose membrane (Schleicher &Schue!!) and then incubated overnight at 4°Cwith eitherIRPI antiserum (against amino acids 408—421), which wasprovided by T. A. Rouault (National Institutes of Health),or IRP2 antiserum (Guo et al., 1994), which was providedby E. A. Leibold (University of Utah). Then the membraneswere incubated with alkaline phosphatase-conjugated goatanti-rabbit IgG (Bio-Rad), and the resulting complex wasvisualized using bromochloroindolyl phosphate and nitroblue tetrazolium.

Isolation of IRP1 cDNA clones from human braincDNA library

Two eDNA probes—one near the 5’ coding region corre-sponding to positions 220—501 and one near the 3’ coding

J. Neuroche,n., Vol. 67, No. 2, 1996

840 J. HU AND J. R. CON1VOR

FIG. 1. Equal amounts of cytoplasmic protein extracts (10 jig)from normal human brain were incubated with either radiolabeledferritin IRE RNA (Ferritin-IRE) orTfR-IREs RNA[TfR -IREs(A,B)],followed by treatment with heparin alone (H) or RNase Ti andheparin (Ti/H). The RNA—protein complexes were separated ona 4% native polyacrylamide gel. Arrows indicate the two RNA—protein complexes that are formed. The free RNA probes areindicated.

region corresponding to positions 1,226—1,872 (Rouault etal., 1990) —were generated from a chimeric human IRP1eDNA clone [pCDLSR-IRE-BP(Kaptain et al., 1991)]. Theprobes were gel-purified and labeled with [32P]dCTP (DuPont) using a random primed DNA labeling kit from Boeh-ringer Mannheim. Unincorporated nucleotideswere removedby a G-50 Micro Column (Pharmacia). The specific activitywas -~3—5x l0~cpm/jig. A human brain eDNA library(Stratagene) in X ZAP 11 vector was screened with botheDNA probes. Positive plaques were selected andresereenedthree times. The phages were excised by Helper Phage(Stratagene). The eDNA inserts in Bluescript plasmid wereanalyzed with restriction endonucleases.

Northern blot analysisThe northern blot analysis was performed on a human

astrocytoma cell line (SW 1088) obtained from the Ameri-can Type Culture Collection (HTB 12). The cell line waschosen because it represents both human and brain-derivedtissue. Total RNA (30 jig) was isolated from astrocytomacells by lysis with 4 M guanidine thiocyanate (Chomczynskiand Sacehi, 1987) and separated on a 1% agarose gel. Afterelectrophoresis, the RNA was transferred to a nylon hybrid-ization membrane (Amersham) by capillary elution over-night, and RNA was UV-cross-linked to the membrane in aUV Stratalinker (Stratagene). The resulting blots wereprobed with two eDNA fragments derived from the eDNAclones that were isolated from a human brain eDNA librarywith the IRP1 eDNA probes, which had been labeled withF’2P]dCTP (Du Pont) using a random primed DNA labelingkit from Boehringer Mannheim.

Purification of IRP from human brainThe brain IRP was isolated as described according to the

method of Neupert et al. (1990) with some modification.Frozen human brain tissue (300 g) was extensively washedwith cold phosphate-buffered saline and homogenized, anda supernatant wasgenerated as described above. Theproteins

(3 g) in the resulting supernatant were reduced by 2% ~-

mercaptoethanol and incubated with equilibrated heparin-Sepharose beads (Pharmacia) at a ratio of 0.1 g/ 1 ml ofbeads under constant shaking for 3 h at room temperature.The heparin-Sepharose beads were collected, washed, andpacked into a column. Afterwashing in 10 mM HEPES (pH7.5), 40mM KCI, 3mM MgCI

2, and 5% glycerol, the boundproteins (16 mg) were eluted with 250 mM KCI. To preparethe RNA affinity column, TfR-IREs poly(A) RNA tran-scripts (200 jig) were synthesized and loaded onto a poly-(U)-Sepharose (Sigma) column at room temperature. Afterthe concentration of KCI in the elutes from the heparin-Sepharose column was adjusted, RNase inhibitor (30 units/ml), heparin (5 mg/ml), and yeast tRNA (35 jig/mI) wereadded. The samples were loaded onto the RNA affinity col-umn and recycled three times. The bound proteins wereeluted with 25 mM HEPES (pH 7.5) and 1 M KC1. Theeluted proteins (100 jig) were concentrated and equilibratedin 10 mM HEPES (pH 7.6), 3 mM MgC12, 40 mM KCI,and 5% glycerol with a Centrieon-30 mieroconeentrator(Amieon). Purified proteins were separated by 7.5% SDS-PAGE, stained overnight in 0.2% Coomassie Brilliant BlueR (Sigma), and destained in 30% methanol and 10% aceticacid. RNA binding activity was analyzed by RNA band shiftassay.

RESULTS

Formation of RNA — protein complexFormation of RNA—protein complex in human brain

extracts was demonstrated using RNA band shiftassays. Two RNA transcripts containing IRE se-quences were used. The ferritin IRE RNA transcriptscontain 100 bases of 5’ UTR of rat ferritin L chainmRNA, and the TfR-IREs RNA transcripts includetwo IRE sequences (IRE A and B). Cytoplasmic ex-tracts isolated from frozen human brain tissue wereincubated with radiolabeled fernitin IRE RNA tran-scripts or TfR-IREs RNA transcripts. In the presenceof RNase Tl and hepanin, two closely migrating RNA—protein complexes were observed that formed distinctbands (Fig. I). If human liver, placenta, or kidneytissue, prepared under conditions identical to those forbrain, was used, only a single RNA—protein complexwas observed (Fig. 2). A human astrocytoma cell linederived from brain tissue also yielded a doublet IRE!IRP complex (Fig. 2). The differences in migrationalpatterns are due to differences in RNA length. The

FIG. 2. Cytoplasmic protein ex-tracts (30 jig) from human brain(Br.), liver (L), placenta (P),astrocytoma cells (SW; 5 jig),and kidney (K) were incubatedwith

32P-labeled ferritin IRE tran-script and then separated on a4% native polyacrylamide gel.Two bands are seen in humanbrain and astrocytoma cells (SW1088); a single band was de-tected in human liver, placenta,and kidney.

J. Neuro(he,n., Vol. 67, No. 2, 1996

IRP IN HUMAN BRAIN 841

FIG. 3. Equal amounts of human brain cytoplasmic extract (10

1ig) were incubated with radiolabeledferritin IRE RNAtranscripts(0.2 nM) in the presence of increasing concentrations of eitherunlabeled TfR-IREs RNAtranscripts or /3-actin RNAtranscripts(two-, 10-, 100-, and 500-fold molar excess). The leftmost lanein both panels represents the control (no competing transcript).

length of RNA fragments of the brain doublet complexin the slower band is 40 nucleotides, whereas the fasterhand contains 60 nucleotides. The length of RNA frag-ments in the single complex from the nonbrain tissuesis 28 nucleotides (data not shown).

Competition studiesTo determine the specificity of the formation of two

RNA—protein complexes in human brain extracts,competition studies were performed. Brain cyto-

I)I~usIT1ieextracts were incubated with radiolabeled fer-ruin IRE RNA transcripts in thepresence of increasingconcentration of unlabeled TfR-IRE RNA transcripts(specific competitor) or human ~-actin RNA tran-scripts (nonspecific competitor). The intensity of bothRNA—protein complexes decreased at twofold molarexcess and disappeared at 500-fold molar excess ofunlabeled TfR-IREs RNA transcripts. The intensity oflhc two RNA—protein complexes was not affected byadding molar excess of unlabeled human ~-actin RNAtranscripts even at 500-fold higher concentration (Fig.3). Because the sequence of ferritin IRE RNA haslittle homology to that of TfR-IREs RNA (Leiboldand Munro, 1987; Casey et al., 1989), these data sug-gest that the two RNA—protein complexes are specifi-cally formed by brain cytoplasmic protein binding tothe !RE structure.

I)etermination of the molecular mass of theRNA —protein complex

Brain cytoplasmic extracts were incubated with ra-cliolaheled ferritin IRE RNA transcripts and cross-linked by UV irradiation. A single RNA—protein corn-

FIG. 4. Human brain extracts (60 jig) were incu-bated with radiolabeled ferritin IRE RNAtranscripts,followed by treatment with RNase Ti and heparin.The sample was UV-irradiated in a Stratalinker 1800for 45 mm and incubated with the mixture of RNaseTi and RNase A. The RNA—protein complex wasresolved by 7.5% SDS-PAGE. Prestained molecularmarkers are shown in kDa. A single radiolabeledband is visible at —.110 kDa.

FIG. 5. Equal amounts of protein ex-tracts from normal brain (60 jig) wereseparated by 7.5% SDS-PAGE andtransferred to nitrocellulose mem-branes. The membranes were immu-noblotted with either IRP1 antiserumat 1:1,000 dilution (IRP1, againstamino acids 408—421) or IRP2 anti-serum at 1:250 dilution (IRP2). Themolecular weights (xiO

3) fromprestained markers are indicated.

plex at -.~l10 kDa was seen on 7.5% SDS-PAGE (Fig.4). This observation suggests that protein in humanbrain extracts directly interacts with fernitin IRE RNAtranscripts and that the UV-cross-linked RNA—proteincomplex has a molecular mass of ~-.110 kDa.

Detection of IRP in human brain extractsWestern blot analysis was performed using antisera

for IRPI and IRP2 on human brain extracts (Fig. 5).Two bands at ‘-.100 and 69 kDa were recognized byIRPI antiserum (against amino acids 408—421). IRP2antiserum recognized two bands at -.~75and 55 kDa.

A human brain eDNA library (Stratagene) wasscreened with two eDNA probes generated from a hu-man IRPI eDNA clone. When RNA from human astro-cytoma cells was hybridized to the two eDNA probesgenerated from thehuman brain eDNA library, a singlemRNA species of ‘-“4 kb was observed (Fig. 6), simi-lar to that reported for other human tissue (Hirling eta!., 1992).

Analyses of partially purified human brain IRPRNA affinity purification yielded —‘100 jig of pro-

tein that separated on SDS-PAGE as one band near100 kDa and a second band around 40 kDa (Fig. 7A).The l00-kDa band was recognized by an IRPI antise-rum (against amino acids 408—421) but not by IRP2antiserum. Neither antiserum recognized the 40-kDa

FIG. 6. Total RNA (30 jig) isolatedfrom human astrocytoma cells (SW1088) was separated on a 1% agar-ose gel and transferred to a nylonmembrane. The membranes were hy-bridized to two cDNA probes derivedfrom cDNA clones: BPi-5’ and BPi-3’. These cDNA clones were isolatedfrom a human brain cDNA library withhuman IRPi cDNA probes. The num-bers indicate the size of the RNA.

J. Neurochem., Vol. 67, No. 2, 1996

842 J. HU AND J. R. CONNOR

FIG. 7. RNA affinity purification was used to purify the brain IRP.A: Human brain extracts (iO jig; left lane) and partially purifiedbrain IRP (1, 3, and 5 jig) were analyzed on 7.5% SDS-PAGEand stained with Coomassie Brilliant Blue. Molecular weightstandards were included (right lane). B: Equal amounts of puri-fied brain lAP (1 jig) were separated by 7.5% SOS-PAGE andtransferred to nitrocellulose membranes. The membranes wereimmunoblotted with either anti-IRP1 antiserum (lRPi, providedby T. A. Rouault) or anti-IRP2 antiserum (IRP2, provided by E. A.Leibold). Molecular weights (xiO

3) are indicated. C: RNA bind-ing activity of the purified brain lAP was analyzed by RNA bandshift assay. Before incubation with radiolabeled ferritin IRE RNAtranscripts, purified brain lAP (20 ng) was treated with ~-mer-captoethanol (lane 2) or diamide (lane 3). Lane 1 represents thecontrol. Following RNase Ti and heparin treatment, the RNA—protein complexes were analyzed on a 4% native polyacrylamidegel.

band (Fig. 7B). The partially purified proteins resultedin the appearance of two RNA—protein complexeswhen incubated with radiolabeled ferritin IRE RNAtranscripts (Fig. 7C). Pretreatment with 2% ~-mercap-toethanol enhanced the intensity of the slower-migrat-ing RNA—protein complex, whereas the intensity ofboth RNA—protein complexes was decreased by pre-treatment with I mM diamide. The formation or activ-ity of the two RNA—protein complexes was not af-t’ected by removal of the protein at 40 kDa by aCentri-con-SO microconcentrator (data not shown).

A supershift analysis was performed on the purifiedprotein complex (Fig. 8). RNA—protein complexeswere completely supershifted by an IRP1 antiserum

Iagainst amino acids 130—151 (Eisenstein et al.,1993)]. Other IRP1 antisera [against amino acids707—721 (Eisenstein et al., 1993) or against aminoacids 408—421] did not cause supershifting, nor didaddition of IRP2 antiserum result in a supershift of theRNA—protein complex.

DISCUSSION

These data demonstrate that IRPI is the predominantIRP in adult human brain. The presence of an IRP inhuman brain is not surprisingly given the presence offerritin and the transfèrrin receptor in this tissue. Thenovel observations in this study are the predominanceof IRPI in human brain and the formation of a doubleIRE/IRP complex. Previous reports of IRP in rodentbrain tissue had indicated that IRP2 is abundant in

brain, especially in fetal tissue (Samaniego et al.,1994). Thus, our different results may reflect differ-ences in age as well as difference in species. There arespecies differences in the amount of ferritin in rodentversus human brain. There is as much as 10 times moreferritin per unit of protein in human brain than in rodentbrain (Roskams and Connor, 1994; Connor et al.,1995), and the cellular distribution of ferritin in brainmay also differ between mice and humans (Dickinsonand Connor, 1995). Both of these observations maybe relevant to the type of IRP that is expressed.

Only a single RNA—protein complex has been de-tected in studies of human liver and placenta extracts(Rouau!t et al., 1989; Neupert et al., 1990), and ourstudy confirmed those findings. However, in brain, adouble IRE!IRP complex is found. Two RNA—proteincomplexes have been detected in rodent tissue extracts(Leibold and Munro, 1988; Henderson et al., 1993),but the two RNA—protein complexes are due to twodistinct IRE binding proteins (Henderson et al., 1993;Guo et al., 1994), which are now referred to as IRPIand IRP2 (Samaniego et al., 1994). In the presentstudy, the RNA band shift analysis of the purified brainIRP indicated that IRP1 is capable of forming both ofthe complexes seen on the RNA band shift assay withthe IRE. Furthermore, the supershift data in this reportindicate that IRPI is involved in the formation of boththe fast- and the slow-migrating IRE/IRP complexes.

The partially purified IRP from human brain showsbiochemical characteristics and a molecular weightsimilar to those of IRPI. Also, the size of the singlespecies of mRNA detected in the human astrocytomacell line by two IRPI eDNA probes is ~-~4kb, which

FIG. 8. Equal amounts of purifiedbrain IRP (50 ng) were incubatedwith rabbit serum (lane 1; 1 jil),IRP1 antiserum against aminoacids 408—421 (lane 2; 1 p1), IRP1antiserum against amino acids130—151 (lane 3; 1 jil), IRP1 anti-serum against amino acids 707—721 (lane 4; i jil), and IRP2 antise-rum (lane 5; 1 jil) for 30 mm in thepresenceof RNase inhibitor (i unit,5’ prime 3’ prime) followed by ad-dition of radiolabeled ferritin IRERNA transcripts. After treatmentwith RNase Ti and heparin, theRNA—protein complexes wereseparated on a 4% native poly-acrylamide gel.

.1. Neurochem., Vol. 67, No. 2, 1996

JRP IN HUMAN BRAIN 843

is consistent with a previous report (Hirling et al.,1992) for IRPI mRNA. The RNA binding activity ofIRPI can be activated by reducing agents and inacti-vated by oxidizing agents (Hentze et al., 1989; Mullneret al., 1992). The RNA binding activity of purifiedbrain IRP is also activated by reducing the proteinwith /3-mercaptoethanol and inhibited by the oxidizingreagent diamide. These data provide further evidencethat the RNA affinity-purified protein from brain is

predominantly IRPI and that IRP1 is involved in theformation of both RNA—protein complexes.

The sizes of the RNA fragments in the doublet IRE!IRP complex from brain are different from each otherand from that in the single IRE/IRP complex seen inthe human liver. These size differences contribute tothe dil’ference in migration on the band shift assaysand indicate that IRPI can protect different-size RNAfragments. The process by which different-size RNAfragments can be protected may involve posttransla-tional modification of IRP I, which either could directlyaffect binding or could affect the conformation of the

protein. Several explanations are possible as to whyIhe IRP would be modified. One explanation is thattuodifications of the protein may cause a difference in

lhe conformation of the IRE/IRP complex, resultingin differing sensitivity to RNase digestion, and thusaffect the stability of the complex. Indeed, the flankinglegion of the IRE has shown to be important for theinteraction between the IRE and the IRP (Harrell etal., 1991; Theil, 1994). Preliminary evidence indicatesthat there are differences in the stability of the twoIRE/IRP complexes (authors’ unpublished data).

‘l’he presence of IRP2 in human brain was not dem-onstrated in this study. IPR2 antisera did reveal twobands on an immunoblot of human brain cytoplasmicextract, hut these bands were at —‘75 and 55 kDa,which is below that reported for the estimated molecu-lar mass of IRP2 (Samaniego et al., 1994; Guo et al.,1994). The reason for this discrepancy in molecularmass is not known. Possibly IRP2 in brain is not pres-ent at sufficient levels to be detected by western blotanalysis, or the75- and 55-kDa protein bands representthe protcolytic fragments of IRP2. Whether IRP2 existsat all in human tissue is still under study. A secondIRP (IRP2) has been reported in rodent tissue includ-ing brain (Henderson et al., 1993). The existing evi-dence for an IRP2 in human tissue is the isolation ofa eDNA for IRP2 from a human T cell eDNA library(Rouault et al., 1990) and a human fetal brain eDNAlibrary (Guo et al., l995a,b). It is possible that IRP2is a developmentally regulated protein and thus is not

present in the adult tissue used in this study. IRP2uRNA has been reported to be excess of IRPI mRNA

in the fetal mouse brain (Samaniego et al., 1994).In conclusion, IRP1 is predominant in adult human

bt’ain and forms a doublet complex with IRE-con-taining RNA. The doublet complex indicates that post-translational modification of IRP occurs in brain,which may regulate the sensitivity of RNA to diges-

tion. The doublet complex and posttranslational modi-fication of the IRP could reflect the heterogeneous pop-ulation of cells in the brain and their different ironrequirements as demonstrated by differences in stain-able iron and in the cellular distribution of ferritinsubunits (Connor and Menzies, 1995). Because irondysregu!ation has been demonstrated in several neuro-degenerative diseases, future studies will examine IRPexpression in these disease states.

Acknowledgment: The human brain tissue used in thisstudy was supplied by the Brain Tissue Resource Center.McLean Hospital, Belmont, MA, U.S.A. The research wassupported by grant IIRG-94-l22 from the Alzheimer’s Asso-ciation and funds from the Jane B. Barsumian Trust.

REFERENCES

Aziz N. and Munro H. N. (1987) Iron regulates ferritin mRNAtranslation through a segment of its 5’ untranslated region. Proc.Null. Acad. Sci. USA 84, 8478—8482.

Beard J. L., Connor J. R., and Jones B. C. (1993) Iron in the brain.Nutr. Rev. 51, 157—170.

Bettany A. J., Eisenstein R. S., and Munro H. N. (1992) Mutagenesisof the iron-regulatory element further defines a rote for RNAsecondary structure in the regulation of ferritin and transferrinreceptor expression. J. Biol. Cheat. 267, 1653 1—16537.

Casey J. L., Koeller D. M., Ramin V. C., Klausner R. D., andHarford J. B. (1989) Iron regulation of transferrin receptormRNA levels requires iron-responsive elements and a rapidturnover determinant in the 3’ untranslated region of themRNA. EMBO 1. 8, 3693-3699.

Chomczynski P. and Sacchi N. (1987) Single-step method of RNAisolation by acid guanidinium thiocyanate-phenol-chloroformextraction. Anal. Bioche,n. 162, 156—159.

Connor J. R. and Menzies S. L. (1995) Cellular managenlent ofiron in the brain. J. Neural. Sci. 134 (Suppl.). 33—44.

Connor J. R., Menzies S. L., St. Martin S.. and Mufson E. J. (1990)Cellular distribution of transferrin, ferritin. and iron in normaland aged human brains. J. Neurosci. Rex. 27, 595—598.

Connor J. R., Menzies S. L., St. Martin S. M., and Mufson E. J.(1992) A histochemical study of iron. transferrin, and ferritinin Alzheimer’s diseased brains. J. Neurosci. Rex. 31, 75—83,

Connor J. R., Snyder B. S., Arosio P., Locftler t). A., and LeWittP. (1995) A quantitative analysis of isoferritins in select regionsof aged, parkinsonian, and Alzheimers diseased brains. .1. Neu-rochem. 65, 717—724.

Cox T. C., Bawden M. J., Martin A., and May B. K. (1991) Humanerythroid 5-aminolevulinate synthase: promoter analysis andidentification of an iron-responsive element in the mRNA.EMBO i. 10, 1891—1902.

Dandekar T., Stripecke R., Gray N. K., Goossen B., Constable A.,Johansson H. E., and Hentze M. W. (1991) Identification of anovel iron-responsive element in murine and human erythroiddelta-aniinolevulinic acid synthase niRNA. EMBOJ. 10, 903—1909.

Dickinson T. K. and Connor J. R. (1995) Cellular distribution ofiron, transferrin, and ferritin in the hypotransferrinemic (Hp)mouse brain. J. Camp. Neural. 355, 67—81).

Ehmann W. D., Markesbery W. R., Alauddin M., Hossain T. I., andBruhaker E. H. (1986) Brain trace elements in Alzheimer’sdisease. Neurotoxicologv 7, 195—206.

Eisenstein R. S., Tuazon P. T., Schalinske K. L., Anderson S. A., andTraugh J. A. (1993) Iron-responsive element-binding protein.Phosphorylation by protein kinasc C. J. Biol. Chem. 268,27363—27370.

Good P. F., Olanow C. W., and Perl D. P. (1992) Neuromelanin-containing neurons of the substantia nigra aceumulate. BrainRex. 593, 343—346.

J. Neu,’ochenr, Vol. 67, ivo. 2, 1996

844 J. HU AND J. R. CONNOR

Gray N. K. and Hentze M. W. (1994) Iron regulatory protein pre-vents binding tf the 43S translation pre-initiation complex toferritin and eALAS mRNA5. EMBO J. 13, 3882—3891.

Guo B., Yu Y., and Leibold F. A. (1994) Iron regulates cytoplasmiclevels of a novel iron-responsive element-binding protein with-out aconitase activity. J. Biol. Chem. 269, 24252—24260.

Guo B., Brown F. M., Phillips J. D., Yu Y., and Leibold F. A.(1995a) Characterization and expression of iron regulatory pro-tein 2 (IRP2). Presence of multiple IRP2 transcripts regulatedby intracellular iron levels. I. Biol. C/tern. 270, 16529— 16535.

Guo B., Phillips J. D., Yu Y., and Leibold E. A. (1995b) Ironregulates the intracellular degradation of iron regulatory protein2 by the proteasome. J. Biol. Chem. 270, 21645—2165 1.

Hallgren B. and Sourander P. (1958) The effect of age on non-haemiron in the human brain. J. Neurochem. 3,41—51.

Harrell C. M., McKenzie A. R., Patino M. M., Walden W. E.,and Theil E. C. (1991) Ferritin mRNA: interactions of ironregulatory element with translational regulator protein P-90 andthe effect on base-paired flanking regions. Proc. Nail. Acad.Sci. USA 88, 4166—4170.

Henderson B. R., Seiser C., and Kuhn L. C. (1993) Characterizationof a second RNA-binding protein in rodents with specificity foriron-responsive elements. J. Biol. Chem. 268, 27327—27334.

Hentze M. W., Caughman S. W., Rouault T. A., Barriocanal J. G.,Dancis A., Harford J. B., and Klausner R. D. (1987) Identifica-tion of the iron-responsive element for the translational regula-tion of human ferritin mRNA. Science 238, 1570—1573.

Hentze M. W., Rouault T. A., Harford J. B., and Klausner R. D.(1989) Oxidation—reduction and the molecular mechanism ofa regulatory RNA—protein interaction. Science 244, 357—359.

Hirling H., Emery-Goodman A., Thompson N., Neupert B., SeiserC., and Kuhn L. C. (1992) Expression of active iron regulatoryfactor from a full-length human eDNA by in vitro transcription!translation. Nucleic Acids Res. 20, 33—39.

Kaptain S., Downey W. E., Tang C., Philpott C., Haile D., OrloffD. G., Harford J. B., Rouault T. A., and Klausner R. D. (1991)A regulated RNA binding protein also possesses aconitase activ-ity. Proc. Nail. Acad. Sci. USA 88, 10109—10113.

Klausner R. D., Rouault T. A., and Harford J. B. (t993) Regulatingthe fate of mRNA: the control of cellular iron metabolism. Cell72, 19—28.

Larkin F. C. and Rao A. (1990) Importance of fetal and neonataliron: adequacy for normal development of the central nervoussystem, in Brain, Behaviour, and Iron in the infant Diet, pp.43—62. Springer-Verlag, New York.

Leibold E. A. and Munro H. N. (1987) Characterization and evolu-tion of the expressed rat ferritin light subunit gene and its pseu-dogene family. Conservation of sequences within noncodingregions of ferritin genes. J. Biol. C/tern. 262, 7335—7341.

Leibold E. A. and Munro H. N. (1988) Cytoplasmic protein bindsin vitro to a highly conserved sequence in the 5’ untranslatedregion of ferritin heavy- and light-subunit mRNAs. Proc. Nail.Acad. Sci. USA 85, 2171—2175.

Mullner E. W., Neupert B., and Kuhn L. C. (1989) A specificmRNA binding factor regulates the iron-dependent stability ofcytoplasmic transferrin receptor mRNA. Cell 58, 373—382.

Mullner E. W., Rothenberger S., Muller A. M., and Kuhn L. C.(1992) In vivo and in vitro modulation of the mRNA-bindingactivity of iron-regulatory factor. Tissue distribution and effectsofcell proliferation, iron levels and redox state. Eur. J. Biochem.208, 597—605.

Neupert B., Thompson N. A., Meyer C., and Kuhn L. C. (1990) Ahigh yield affinity purification method for specific RNA-bindingproteins: isolation of the iron regulatory factor from humanplacenta. Nucleic Acids Rex. 18, 51—55.

Roskams A. J. I. and Connor J. R. (1994) Iron, transt’errin, andferritin in the rat brain during development and aging. J. Neuro-chem. 63, 709—7 16.

Rouault T. A., Hentze M. W., Haile D. J., Harford J. B.. and KlausnerR. D. (1989) The iron-responsive element binding protein: amethod for the affinity purification of a regulatory RNA-bindingprotein. Proc. NatI. Acad. Sci. USA 86, 5768—5772.

Rouault T. A., Tang C. K., Kaptain S., Burgess W. H.. Haile D. J.,Samaniego F., McBride 0. W., Harford J. B., and KlausnerR. D. (1990) Cloning of the eDNA encoding an RNA regula-tory protein—the human iron-responsive element-binding pro-tein. Proc. Natl. Acad. Sci. USA 87, 7958—7962.

Samaniego F., Chin J., Iwai K., Rouault T. A., and Klausner R. D.(1994) Molecular characterization of a second iron-responsiveelement binding protein, iron regulatory protein 2. Structure,function, and post-translational regulation. I. Blot. Chern. 269,30904—309 10.

Schneider C., Owen M. J., Banville D., and Williams J. G. (1984)Primary structure of human transferrin receptor deduced fromthe mRNA sequence. Nature 311, 675—678.

Theil E. C. (1994) Iron regulatory elements (IREs): a family ofmRNA non-coding sequences. Biochem. J. 304, I —11.

Zheng L., Kennedy M. C., Blondin G. A., Beinert H., and ZalkinH. (1992) Binding of cytosolic aconitase to the iron responsiveelement of porcine mitochondrial aconitase mRNA. Arch. Bio-chem. Biophys. 299, 356—360.

J. Neurochem.. Vol. 67, No. 2. 1996