In Vivo Association of the Stability Control Protein {alpha}CP with Actively Translating mRNAs

Posttranscriptional controls play a major role in eucaryoticgene expression. These controls are mediated by sequence-specificinteractions of cis-acting determinants in target mRNAs withone or more protein factors. The positioning of a subset ofthese mRNA-protein (RNP) complexes within the 3' untranslatedregion (3' UTR) may allow them to remain associated with themRNA during active translation. Robust expression of human ${alpha}$ -globinmRNA during erythroid differentiation has been linked to formationof a binary complex between a KH-domain protein, ${alpha}$ CP, and a 3'UTR C-rich motif. Detection of this " ${alpha}$ -complex" has been limitedto in vitro studies, and the functional state of the ${alpha}$ -globinmRNA targeted by ${alpha}$ CP has not been defined. In the present studywe demonstrate that a significant fraction of ${alpha}$ CP is associatedwith polysomal mRNA. Targeted analysis of the polysomal RNPcomplexes revealed that ${alpha}$ CP is specifically bound to activelytranslating ${alpha}$ -globin mRNA. The bound ${alpha}$ CP is restricted to thepoly(C)-rich 3' UTR motif and is dislodged when ribosomes areallowed to enter this region. These data validate the generalimportance of the 3' UTR as a sheltered site for RNP complexesand support a specific model in which the stabilizing functionof ${alpha}$ CP is mediated on actively translating target mRNAs.

INTRODUCTION

Top

Introduction

Posttranscriptional regulation of gene expression plays a pivotalrole in developmental processes and in cellular differentiation(16). A prominent component of this control is exerted at thelevel of mRNA stability (35). The half-lives of mammalian mRNAscan be as short as 15 min or as long as several days. Generallyspeaking, mRNAs with short half-lives encode proteins that aretightly controlled over narrow time frames, while mRNAs withlong half-lives often encode proteins expressed at high levelsin terminally differentiated cells (3, 39). The stability ofindividual mRNAs reflects the interaction of general determinants,such as the 5' m⁷Gppp cap and a 3' terminal poly(A) tail, withtranscript-specific regulatory elements. The latter determinantsare often restricted to the 3' untranslated region (UTR) (4)and are recognized by sequence- and structure-specific RNA bindingproteins (23). The net contributions of general and specificstability determinants generate expression profiles unique toeach mRNA species. Identifying the structures and compositionsof these RNP complexes and determining how they mediate controlover mRNA stability presents a major challenge to current researchefforts.

Globin mRNAs accumulate to 95% of the total cellular mRNA duringthe 2- to 3-day window of terminal erythroid differentiation(36). This is a period at which these cells have already undergoneglobal transcriptional arrest. Hence, the shift in mRNA complexityduring terminal differentiation of the erythroblast reflectsstabilization of globin mRNAs and selective destabilizationof most nonglobin mRNAs (reviewed in reference 36). Due to thehigh-level stability of globin mRNAs and the numerous geneticand experimental model systems that are available for theirstudy, these mRNAs constitute an ideal model for the study ofstabilization mechanisms.

Initial insight into the mechanism of globin mRNA stabilizationwas gained from the study of ${alpha}$ -Constant Spring ( ${alpha}$ ^CS) thalassemia.The ${alpha}$ ^CS mutation, found at a high frequency in Southeast Asia,is a single base substitution at the translation stop codon(UAA $->$ CAA) of the major ${alpha}$ -globin gene ( ${alpha}$ 2) (8, 29). This mutationresults in ribosome read-through into the 3' UTR with terminationat an in-frame UAA within the poly(A) addition site AAUAAA.The ${alpha}$ ^CS mutation results in greater than 95% loss of ${alpha}$ -globinmRNA expression from the affected locus and a correspondingdecrease in overall ${alpha}$ -globin synthesis (8, 29). The dramaticloss of ${alpha}$ -globin expression reflects instability of ${alpha}$ ^CS mRNA (24).Studies further revealed that this destabilization is triggeredby ribosome entry into the 3' UTR (30, 44). The model that emergedfrom these studies was that one or more determinants withinthe 3' UTR of human ${alpha}$ -globin (h ${alpha}$ -globin) mRNA are involved inmaintenance of its stability; interference with the structureor function of this complex(es) by the elongating ribosome triggersaccelerated mRNA decay.

Studies to define the determinants of h ${alpha}$ -globin mRNA stabilitysubsequently confirmed certain aspects of this model. Threediscontinuous C-rich elements within the 3' UTR of h ${alpha}$ -globinmRNA were linked to its high-level accumulation in erythroidcells (44, 45). A corresponding set of high-affinity bindingproteins was identified by in vitro analyses (42). Biochemicalpurification of these proteins (20) revealed that they compriseda set of proteins sharing a characteristic triple repeat ofthe 50-amino-acid KH domain RNA binding motif (37). Based ontheir binding to h ${alpha}$ -globin mRNA and their poly(C) specificity,these proteins were named ${alpha}$ -globin poly(C) binding proteins,or ${alpha}$ CPs (20, 26). { ${alpha}$ CPs are also referred to as PCBPs [poly(C)-bindingproteins] and hnRNP Es (14, 22, 31).} Interaction of ${alpha}$ CP withthe C-rich 3' UTR determinants forms an RNP " ${alpha}$ -complex." Mutationswithin the 3' UTR that destroyed the ability of ${alpha}$ CPs to formthis complex in vitro resulted in a corresponding loss of mRNAaccumulation in vivo (42). Studies carried out in transgenicmice and using in vitro decay systems further suggested thatthe ${alpha}$ -complex protects the poly(A) tail from rapid shortening(30, 43). Using in vitro approaches it was possible to identify ${alpha}$ -complexes on the 3' UTRs of three additional stable mRNAs.These findings led to the hypothesis that the ${alpha}$ -complex mightconstitute a general determinant of high-level mRNA stability(18).

While the model of ${alpha}$ -complex-mediated mRNA stabilization is supportedby experimental data, certain aspects remain speculative. Demonstrationof the ${alpha}$ -complex formation on ${alpha}$ -globin mRNA has been limited toin vitro binding studies, and these studies have been restrictedto the analysis of 3' UTR sequences. Whether the ${alpha}$ -complex formsin vivo and whether it is limited to the 3'UTR and/or bindsto other regions of the target mRNA is not known. In addition,the provisional model assumes that ${alpha}$ CP is normally present onactively translating mRNAs and that an antiterminated ribosome,as occurs in the ${alpha}$ ^CS mutation, has the capacity to disrupt thisRNP complex (45). Alternatively, the ${alpha}$ -complex may be limitedto mRNAs that are translationally inert or have cycled out oftranslation due to stochastic or age-related phenomena.

To further define the in vivo pathways of mRNA stabilizationand the role(s) of ${alpha}$ CP binding in this process, we tested a numberof assumptions intrinsic to ${alpha}$ CP binding and function. Significantly,these studies and others (S. Waggoner and S. A. Liebhaber, unpublisheddata) demonstrate a stable and selective in vivo associationof ${alpha}$ CP with ${alpha}$ -globin mRNA. These studies also localize the bindingsite exclusively to the C-rich motif in the 3' UTR. This associationis dependent on the 3' UTR positioning and is compatible withefficient polysome loading and active translation. These datasupport a model in which the RNP ${alpha}$ -complex can exert a stabilizingeffect on actively translating mRNAs.

MATERIALS AND METHODS

Top

Materials and Methods

Cell culture. Human erythroleukemia (K562) cells were grown in RPMI 1640 mediumand mouse erythroleukemia (MEL) cells were cultured in minimalessential medium (MEM) supplemented with 10% fetal bovine serumcontaining 100 U of penicillin/ml and 100 µg of streptomycinsulfate/ml in a 37°C, 5% CO₂ incubator. MEL cells stablytransfected with the tet transactivator (MEL/tTA) (20a) wereused for conditional expression of the h ${alpha}$ -globin mRNA. For transfections,MEL/tTA cells supplemented with 500 ng of tetracycline (TET)/mlwere split 1 day before electroporation, washed with cold phosphate-bufferedsaline (PBS), pelleted by centrifugation at 4°C, and resuspendedin cold serum-free MEM at a concentration of 5 x 10⁷/ml (0.5ml of cells). Two micrograms of plasmid and 18 µg of vectorcarrier were added to the cell suspension. Electroporation wasperformed with a BRL Cell-Porator system at 250 V, 1,180 µF,and low resistance. The cells were then plated in prewarmedcomplete MEM and maintained in the incubator for 24 h priorto analysis.

Expression plasmids. The full-length h ${alpha}$ 2-globin gene cloned into pTet-splice vector(Gibco-BRL) was designated pTet-WT. pTet-CS was generated byreplacing the termination codon of ${alpha}$ 2-globin with CS mutation,and pTet-Neu was generated by replacing the 42-bp protectedregion of the WT gene with a 42-bp neutral sequence which cannotform the ${alpha}$ -complex (detailed by Kong et al. [20a]).

Preparation of cell extracts and polysome purification. MEL and K562 cells grown under standard conditions (see above)were washed three times with cold PBS and resuspended in proteinaseinhibitor buffer (1 mM potassium acetate, 1.5 mM magnesium acetate,2 mM dithiothreitol [DTT], 10 mM Tris-Cl [pH 7.4], 100 mg ofphenylmethylsulfonyl fluoride [PMSF]/ml, 2 mg of aprotinin/ml,2 mg of pepstatin A/ml) at a density of 2 x 10⁷/ml. The cellswere lysed with 25 strokes in a Dounce homogenizer. The lysatewas clarified by centrifugation at 4°C for 10 min at 10,000x g. The supernatant was layered onto a 30% sucrose cushion(in 1 mM potassium acetate, 1.5 mM magnesium acetate, 2 mM DTT,10 mM Tris-Cl [pH 7.4]) and centrifuged for 2.5 h at 130,000x g (32,500 rpm; SW41 rotor). The supernatant (S130) was harvestedand the pellet (polysomes) was resuspended in protease inhibitorbuffer (1 mM potassium acetate, 1.5 mM magnesium acetate, 2mM DTT, 10 mM Tris-Cl [pH 7.4], 100 mg of PMSF/ml, 2 mg of aprotinin/ml,2 mg of pepstatin A/ml). All fractions were stored at -80°C.

Sucrose gradients. Ten-to-50% linear sucrose gradients containing 100 mM KCl, 5mM MgCl₂, 2 mM DTT, 20 mM HEPES (pH 7.4) were prepared in Beckmanultracentrifuge tubes by using a two-chamber gradient mixer.K562 cells were incubated with cycloheximide (100 µg/ml;freshly prepared in ethanol) for 15 min prior to harvesting.The cells were then transferred to a 50-ml tube containing 20ml of frozen crushed PBS and 100 µg of cycloheximide/mland centrifuged at 1,000 x g for 5 min at 4°C. The cellpellet was washed with ice-cold PBS twice and was lysed by repeatedpipetting in 500 µl of ice-cold TMK100 lysis buffer (10mM Tris-HCl [pH 7.4], 5 mM MgCl₂, 100 mM KCl, 2 mM DTT, 1% TritonX-100, and RNase inhibitor [100 U/ml; Promega], in diethyl pyrocarbonate-treatedwater) for 5 min. The nuclei were cleared at 10,000 x g for10 min at 4°C, and the supernatants (S10) were removed andlayered onto the prepared gradients. The SW41 rotor and theultracentrifuge were both precooled to 4°C, and the gradientswere centrifuged at 40,000 rpm for 85 min at 4°C. The gradientswere collected in 16 fractions (650 µl per fraction) fromthe top to the bottom by displacing them upwards with 60% sucrose,and the gradient profile was monitored via UV absorbance at254 nm with an ISCO UA-5 detector (Lincoln, Nebr.). The fractionswere collected into 1.5-ml microcentrifuge tubes containing70 µl of 10% sodium dodecyl sulfate (SDS). Each samplewas digested with 8 µl of protease K (20 mg/ml) solutionat 37°C for 30 min. Samples were stored at -80°C priorto RNA extraction.

For ${alpha}$ CP-polysome interaction studies, 25 optical density at 260nm (OD₂₆₀) units of redissolved polysomes was used as a startingmaterial. For EDTA treatment, a polysome aliquot was broughtto 20 mM EDTA and incubated at 4°C for 20 min before loadingonto the gradient supplemented with 20 mM EDTA. For nucleasetreatment, RNase A was added to a polysome aliquot to a finalconcentration of 100 µg/ml (9) and incubated for 15 minin a cold room before loading. For the salt sensitivity study,a polysome aliquot was adjusted to the desired final concentrationof KCl (0.5 or 0.8 M) and loaded onto a gradient containingthe same concentration of salt. After loading, each gradientwas centrifuged at 40,000 rpm in a Beckman L8-m ultracentrifugewith an SW41 rotor for 150 min at 4°C. Fractions were collectedas described above, and proteins were precipitated with a fourfoldexcess of cold acetone prior to SDS-polyacrylamide gel electrophoresis(PAGE) and Western blotting.

Antibodies. Primary antibodies to the ${alpha}$ CP isoforms have been previously detailedand characterized (7). PK antibody (Biodesign International)was used at 1:3,000. Polyclonal anti-L7a rabbit serum was raisedagainst epitope residues 253 to 266 in the human L7a sequenceas originally described by Ziemiecki et al. (47) and used withoutfurther purification at 1:1,000. Secondary antibodies were asfollows: donkey anti-rabbit immunoglobulin G (IgG)-horseradishperoxidase (HRP) (used at 1:5,000; Amersham) and donkey anti-goatIgG-HRP (used at 1:5,000; Santa Cruz).

SDS-PAGE and analysis. For Western and Northwestern analysis, proteins were separatedon SDS-12.5% PAGE and electroblotted to nitrocellulose membranes(Protran BA 85; Schleicher & Schuell) for 1 h at 150 mAin transfer buffer (20 mM Tris, 150 mM glycine, 20% methanol)using a Semi-phor transfer apparatus (Hoefer). For Western analysis,the membranes were blocked in 3% nonfat milk in 1x PBS for 1h at room temperature, followed by an additional hour with appropriateantisera. Signals were developed by incubation with HRP-labeledsecondary antibodies (Amersham) as detailed by the supplier(ECL reagents; Boehringer Mannheim).

RPA. Probes used for the RNase protection assay (RPA) were generatedby in vitro transcription of plasmids containing cDNA insertsfor h ${alpha}$ -globin and h ${zeta}$ -globin (25), human glyceraldehyde-3-phosphatedehydrogenase (GAPDH) and mouse GAPDH (Ambion, Austin, Tex.),or direct transcription from a human ${gamma}$ -globin genomic DNA PCRtemplate containing a contiguous region extending from intron1 through intron 2. This 320-nucleotide (nt) antisense ${gamma}$ -globinRNA probe generated from an SP6 promoter element in the 5' amplificationprimer yields a 223-nt protected fragment corresponding to exon2 of the ${gamma}$ -globin mRNA (a kind gift from J. E. Russell, Universityof Pennsylvania). In vitro transcriptions were carried out inthe presence of ${alpha}$ [³²P]CTP (400 Ci/mmol, 10 mCi/ml; Amersham,Arlington Heights, Ill.) using a Maxiscript SP6 kit under conditionsrecommended by the manufacturer (Ambion). Final concentrationsof ATP, GTP, and UTP were 0.5 mM, and that of CTP was 0.06 mM.

RNA was phenol extracted from sucrose fractions and ethanolprecipitated with 1 µl of glycogen (15 µg/µl)as a carrier. The precipitated RNA was redissolved in 21 µlof hybridization buffer (40 mM piperazine-N,N'-bis(2-ethanesulfonicacid) [pH 6.4], 1 mM EDTA [pH 8.0], 0.4 M NaCl, 80% formamide)supplemented with probes. Samples were heated at 84°C for15 min, incubated overnight at 50°C, and digested for 15min at room temperature in 200 µl of RNase assay buffer(300 mM NaCl, 10 mM Tris Cl [pH 7.5], 5 mM EDTA [pH 8.0]) containing1 µl of RNase cocktail (500 mg of RNase A/ml, 20,000 Uof RNase T₁/ml; Ambion). Digestions were terminated by additionof 18 µl of an SDS-protease K (10%:2 mg/ml) mixture toeach sample followed by incubation for 20 min at 37°C. RNAwas extracted, precipitated, dissolved in loading buffer, andresolved onto a 6% acrylamide-8 M urea gel. Radioactivity inbands of interest was quantified by PhosphorImager analysis(Storm 840; Molecular Dynamics).

Immunoprecipitation. Ten microliters of anti-CP rabbit serum or preimmune rabbitserum was incubated with 200 µl of a 50% protein A-Sepharoseslurry in 600 µl of PBS for 1 h at 4°C with gentlerocking. S130 and the polysome fraction (150 µl) werediluted in 600 µl of IP binding buffer (20 mM HEPES [pH7.9], 150 mM NaCl, 0.05% Triton X-100, 100 U of RNase inhibitor/ml).The diluted S130 or polysome fraction was incubated with antibody-coupledprotein A-Sepharose for 2 h at 4°C. The immunoprecipitatedcomplexes were pelleted by brief low-speed centrifugation andwashed five times with IP binding buffer. A 400-µl aliquotof IP elution buffer (0.1 M Tris-HCl [pH 7.5], 12.5 mM EDTA,0.15 M NaCl, 1% SDS) was added to the protein A-Sepharose-RNPcomplex pellet and the complex was disrupted by boiling for3 min. RNA was then extracted and ethanol precipitated priorto RPA analysis.

RESULTS

Top

Results

Globin gene expression is restricted to erythroid cells andis developmentally specified. The human-derived K562 cell lineexpresses abundant levels of a fetal or adult ${alpha}$ -globin, as wellas its embryonic precursor, ${zeta}$ -globin. This phenotype makes thesecells an optimal model for analysis of h ${alpha}$ -globin gene expression.To characterize in vivo interactions between h ${alpha}$ -globin mRNA and ${alpha}$ CP proteins and relate these interactions to ${alpha}$ -complex function,we asked whether h ${alpha}$ -globin mRNA and ${alpha}$ CPs colocalize on ribosomes.

h ${alpha}$ -globin mRNA is quantitatively loaded on polysomes. The distribution of ${alpha}$ -globin mRNA in K562 cells was studied bycell fractionation and sucrose gradient analyses. In the initialexperiment, a clarified K562 cytoplasmic extract was fractionatedinto preribosomal and ribosomal fractions by sedimentation at130,000 x g through a 30% sucrose cushion. The content of h ${alpha}$ -globinmRNA in these fractions was assessed by RPA (Fig. 1). The datarevealed that ${alpha}$ -globin mRNA quantitatively fractionated withthe polysomal pellet. To confirm and extend this observation,the distribution of h ${alpha}$ -globin mRNA was assessed across a polysomegradient. A clarified K562 cytosolic extract was applied toa linear 10-to-50% sucrose gradient (Fig. 2A). RNA extractedfrom each fraction was resolved on an agarose gel (Fig. 2B),and the distributions of h ${alpha}$ -globin and ${zeta}$ -globin mRNAs were determinedacross the gradient by RPA (Fig. 2C). h ${alpha}$ -globin and ${zeta}$ -globin mRNAsboth have 142-codon open reading frames. Consistent with thedata in Fig. 1, ${alpha}$ -globin mRNA was quantitatively incorporatedin the polysomal fractions. The distributions of h ${alpha}$ - and ${zeta}$ -globinmRNAs peaked at the 4-5 polysome region (Fig. 2D). Higher-molecular-weightpolysomes are involved in translation of various nonglobin mRNAsin the K562 cells. The positioning of the ${zeta}$ -globin mRNA peakslightly to the right of the ${alpha}$ -globin peak was reproducible andmay indicate a slightly higher translation efficiency of ${zeta}$ -globinmRNA in this embryonic erythroid cell environment. We concludefrom these data that h ${alpha}$ -globin is quantitatively incorporatedon polysomes in K562 cells.

View larger version (32K):
[in this window]
[in a new window]
FIG. 1. h ${alpha}$ -globin mRNA localizes to the polysome fraction of K562 cells. A clarified (S20) cytoplasmic extract from log-growth K562 was pelleted through a 30% sucrose cushion to separate prepolysome supernatant (S130) and polysomal (P) pellet fractions. h ${alpha}$ -globin and GAPDH mRNAs were detected by RPA. The positions of the protected bands corresponding to GAPDH and ${alpha}$ -globin are noted. The band above ${alpha}$ -globin corresponds in size to the primary ${alpha}$ -globin transcript and may be detecting contaminant genomic DNA. A 25-nt DNA ladder is shown (M).

View larger version (51K):
[in this window]
[in a new window]
FIG. 2. Distribution of h ${alpha}$ - and ${zeta}$ -globin mRNAs across a K562 cell polysome gradient. (A) Sucrose gradient profile of K562 cytosolic extract. The absorbance profile (OD₂₅₄) of the gradient is shown. The top of the gradient is to the left; the positions of absorbance peaks corresponding to preribosomal RNPs, 40S, 60S, and 80S, and polysomes (2-, 3-, 4-, 5-, 6-, 7-, and 8-somes) are indicated. The 18 fractions (Fx #) collected for subsequent analysis are identified below the tracing. (B) Agarose gel electrophoresis of RNA extracted from the polysome gradient fractions. A 2-µg RNA sample from each fraction (in panel A) was applied to the gel and electrophoresed, and the abundant 28S, 18S, and 5S rRNAs were directly visualized by ethidium bromide staining. The distributions of these RNAs were consistent with the OD peak assignment (in panel A). (C) Detection of globin mRNAs across the K562 polysome gradient. Each gradient fraction was assessed for h ${alpha}$ -globin and h ${zeta}$ -globin mRNAs by RPA with corresponding ³²P-labeled probes. h ${alpha}$ -globin and h ${zeta}$ -globin mRNAs protected probe fragments of 132 and 150 bp, respectively. (D) Distribution of globin mRNAs across the K562 polysome gradient. The contents of h ${alpha}$ -globin and ${zeta}$ -globin mRNAs in each fraction (in panel C) were quantified by PhosphorImager analysis. The amount of each mRNA species in each fraction is depicted as a percentage (ordinate) of the total for the corresponding species across the gradient.

A fraction of ${alpha}$ CP proteins is polysome associated. Direct association of ${alpha}$ CP with polysomes was tested. Equal aliquotsof S20 and S130 and a fivefold-concentrated aliquot of polysomeproteins were separated by SDS-PAGE and probed with a seriesof antisera (Fig. 3). As expected, the large ribosome subunitprotein L7a was restricted to the polysome pellet. In contrast,the ${alpha}$ CP2 isoforms were present in both the supernatant and polysomalfractions. On the basis of relative intensity of signals andthe relative loading on each lane, it was estimated that 10to 20% of the total cytoplasmic ${alpha}$ CP2 isoforms was localized tothe polysome pellet.

View larger version (23K):
[in this window]
[in a new window]
FIG. 3. ${alpha}$ CP proteins are ribosome associated. A clarified K562 cell cytoplasmic extract (S20) was fractionated into prepolysomal (S130) and polysomal (P) fractions by sedimentation through a 30% sucrose cushion. Each of the three preparations was resolved on an SDS-PAGE gel; equal aliquots of S20 and S130 fractions and fivefold-concentrated aliquots of the polysome fraction were separated by SDS-PAGE. ${alpha}$ CP2/KL and ribosomal L7a proteins were detected by Western blotting with corresponding antisera. The band at 37 kDa represents an isoform of ${alpha}$ CP2 ( ${alpha}$ CP2KL). The positions of the molecular mass markers (not shown) are indicated on the left.

The observation that a fraction of cytoplasmic ${alpha}$ CPs was enrichedin the polysome pellet suggested that these proteins might bepolysome associated. To test this possibility, polysome gradientswere probed for ${alpha}$ CP. An initial study was carried out on a nativepolysome preparation (Fig. 4A). The distribution of L7a proteinwas appropriately limited to the 60S, 80S, and polysome regionsof the gradient. In contrast, the ${alpha}$ CP2 signal extended from thepre-40S region through the entire polysome region. Since itis possible that a certain amount of the prepolysomal ${alpha}$ CP2 signalmay result from dissociation during handling, the ${alpha}$ CP contentin the polysome-associated fraction of the gradient is consideredas a minimal estimate. These data, in combination with thosein Fig. 3, demonstrate that a substantial population of cytoplasmic ${alpha}$ CP is associated with polysomes.

View larger version (50K):
[in this window]
[in a new window]
FIG. 4. Characterization of the ${alpha}$ CP-polysome interaction. Polysome aliquots (as in Fig. 3) were treated in the indicated manners and analyzed on 10-to-50% sucrose gradients. (A) No additional treatment. The OD₂₅₄ is indicated (upper panel; Polysome Profile). Proteins in each fraction were precipitated, separated by SDS-12% PAGE, transferred to membranes, and probed with the indicated antibodies (Western blots). See the legend to Fig. 3 for details. (B) EDTA treatment. The polysome sample was resuspended in 20 mM EDTA (final concentration) prior to sucrose gradient fractionation. This treatment dissociates polysomes into 40S and 60S ribosome subunits. The splitting of the ${alpha}$ CP signal seen in this Western blot is occasionally observed. (C) Treatment with 0.5 M KCl. The polysome fraction was brought to 0.5 M KCl (final concentration) prior to sucrose gradient fractionation. This treatment removes proteins loosely associated with the polysomes. (D) Treatment with 0.8 M KCl. The polysome fraction was supplemented with 0.8 M KCl (final concentration) prior to sucrose gradient fractionation. This treatment removes almost all proteins from the polysomes that are not intrinsic ribosomal proteins.

The interaction of ${alpha}$ CP with polysomes was further characterized.Ribosome stability is dependent on the presence of Mg²⁺ (5).Treatment of the K562 polysome preparation with EDTA causedthe predicted collapse of the polysome profile to 40S and 60Speaks (Fig. 4B). L7a was appropriately localized to the 60Speak. The release of ${alpha}$ CPs from EDTA-dissociated polysomes tothe top of the gradient (Fig. 4B, lower panel) indicated thatthe ${alpha}$ CPs, although associated with polysomes, are not directlybound to either of the two ribosomal subunits.

The basis for the ${alpha}$ CP-polysome association was assessed for saltresistance. An aliquot of the polysome preparation was broughtto a final KCl concentration of 0.5 or 0.8 M. Under these highsalt conditions, most protein-protein interactions are interruptedwhile the structure of the core ribosomal complex remains intact(46). As expected, the polysome profile and the distributionof L7a were maintained under the high-salt treatment (Fig. 4Cand D, upper panels). The distribution of ${alpha}$ CPs was only minimallyaltered at the 0.5 M KCl concentration (Fig. 4C) but changedsignificantly at 0.8 M KCl, with a shift to the lighter, prepolysomefractions (Fig. 4D). This level of salt resistance clearly distinguishes ${alpha}$ CP from intrinsic ribosomal proteins (e.g., L7a) and defines ${alpha}$ CP as a tightly bound polysome accessory protein.

Association of ${alpha}$ CP with polysomes is RNA dependent. The tight association of ${alpha}$ CP with polysomes might reflect a directinteraction of ${alpha}$ CP with the 80S ribosome per se. Alternatively,this association might be mediated indirectly via binding toactively translating mRNAs. To distinguish between these twomodels, an aliquot of the K562 polysomes was treated with RNaseA prior to sucrose gradient fractionation. RNase A will digestexposed regions of the mRNA and in so doing will sever linkagesbetween ribosomes in the polysome complex. RNase digestion resultedin the expected collapse of the polysome profile to a singlemonosome peak (80S) (Fig. 5, top panel). A second distinct peak,generated at the top of the gradient, most likely representedmRNP fragments released by the RNase digestion. L7a proteinappropriately localized to the 80S peak (Fig. 5, bottom panel).In contrast, ${alpha}$ CP was restricted to the preribosomal mRNP peakat the top of the gradient (Fig. 5, middle panel). Of particularnote, ${alpha}$ CP was not represented in the monosome region. These datademonstrate that the association of ${alpha}$ CPs with polysomes is indirectand RNA dependent.

View larger version (66K):
[in this window]
[in a new window]
FIG. 5. Association of ${alpha}$ CP with polysomes is RNA dependent. The polysome fraction was prepared as detailed in the legend for Fig. 3, and an aliquot was treated with RNase A prior to sucrose gradient fractionation. Details of the analysis are as in the legend to Fig. 4.

${alpha}$ CP binds to actively translating ${alpha}$ -globin mRNA. The partial overlap in the distributions of ${alpha}$ CP and h ${alpha}$ -globinmRNA on the polysome gradient (Fig. 2 and 4) and the RNA dependenceof ${alpha}$ CP association with ribosomes (Fig. 5) suggested that ${alpha}$ CPmight be directly bound to actively translating h ${alpha}$ -globin mRNA.To test this model, ${alpha}$ CP-containing complexes were immunoprecipitatedfrom S130 and polysome fractions of a K562 cytosolic extract;RNAs isolated from the immunoprecipitates were assayed for h ${alpha}$ -globinmRNA and for two mRNAs that do not associate with ${alpha}$ CP, GAPDHand ${gamma}$ -globin (Fig. 6). The input polysomal RNA preparation containedequivalent amounts of all three mRNAs, while ${alpha}$ CP2 complexes containedabundant amounts of h ${alpha}$ -globin mRNA with only trace levels ofthe two control mRNAs. These data demonstrate a specific invivo association of ${alpha}$ CPs with actively translating, polysomalh ${alpha}$ -globin mRNA.

View larger version (44K):
[in this window]
[in a new window]
FIG. 6. In vivo association of ${alpha}$ CP with polysome-bound h ${alpha}$ -globin mRNA. Equal amounts of the prepolysomal (S130) and polysomal (P) fractions of K562 cytoplasmic extracts were individually immunoprecipitated with an antibody specific to the ${alpha}$ CP2 and ${alpha}$ CP2-KL isoforms. RNA extracted from the immunoprecipitates was assessed for specific mRNA content by RPA. This study was carried out independently four times with consistent results. (A) RPA using probes to h ${alpha}$ -globin and GAPDH mRNAs. (B) RPA using probes to h ${alpha}$ -globin and ${gamma}$ -globin mRNAs.

In vivo binding of ${alpha}$ CP to h ${alpha}$ -globin mRNA is restricted to the 3' UTR. Prior work from our laboratory has defined a C-rich motif withinthe 3' UTR of h ${alpha}$ -globin mRNA that serves as a high-affinity bindingsite for ${alpha}$ CP (7). Of note, these studies relied on in vitro bindinganalyses and were limited to the study of 3' UTR target sequences.The immunoprecipitation studies detailed in the present reportdemonstrate in vivo binding of ${alpha}$ CP to polysomal h ${alpha}$ -globin mRNA,but they did not define the site of this interaction on thetarget mRNA. This binding may be uniquely mediated by the previouslydefined 3' UTR poly(C)-rich motifs. Alternatively, ${alpha}$ CP couldbind to one or more additional sites elsewhere in the full-lengthGC-rich h ${alpha}$ -globin mRNA. To resolve this issue, MEL cells expressinga tet-transactivator fusion protein (MEL/tTA cells) were transfectedwith wild-type h ${alpha}$ -globin mRNA ( ${alpha}$ ^WT) and with a derivative RNAin which the 3' UTR ${alpha}$ CP binding site has been replaced by a comparablysized "neutral" sequence ( ${alpha}$ ^Neu) (Fig. 7A). Both constructs wereunder the control of the tet operator and cytomegalovirus promoter.The neutral sequence substitution in ${alpha}$ ^Neu destroys ${alpha}$ CP bindingto the 3' UTR when assayed in vitro and destabilizes the mRNAwhen assessed in vivo (20a). Twenty-four hours posttransfectionin TET-deficient medium, the MEL cells were harvested and polysomeswere isolated. Immunoprecipitation of ${alpha}$ CP complexes from thepolysome (P) and prepolysomal (S130) fractions was performedalong with a control preimmune immunoprecipitation (Fig. 7B).Analysis of the two fractions prior to immunoprecipitation confirmedrobust representation of both h ${alpha}$ -globin mRNAs in the polysomefraction. Immunoprecipitation of ${alpha}$ CP-containing polysomal RNPsresulted in selective isolation of the wild-type h ${alpha}$ -globin mRNAbut failed to bring down the mutant ( ${alpha}$ ^Neu) mRNA. These data demonstratedthat the ${alpha}$ CP binding site in the 3' UTR is the unique in vivobinding site for ${alpha}$ CP on translating h ${alpha}$ -globin mRNA.

View larger version (82K):
[in this window]
[in a new window]
FIG. 7. Interaction of ${alpha}$ CP2 with ${alpha}$ -globin mRNA is uniquely dependent on the 3' UTR poly(C)-rich region and is sensitive to displacement by an antiterminated 80S. (A) Expression of h ${alpha}$ -globin mRNAs with distinct 3' UTRs (WT, Neu, and CS) in MEL/tTA cells. The structures of the three encoded mRNAs under the control of a tet promoter are shown. The positions of the translation start site (AUG), termination site (UAA), ${alpha}$ CP binding site (protected region [PR]), and the antitermination mutation in the ${alpha}$ CS (UAA $->$ CAA) are shown. The cross-hatched box represents the substitution of a neutral sequence for the PR motif. (B) ${alpha}$ CP binding on h ${alpha}$ -globin mRNA is restricted to the C-rich 3' UTR stability motif. Each of the indicated ${alpha}$ -globin mRNAs was expressed in transfected cells from a corresponding plasmid. After 24 h of induction of expression in TET-deficient medium, the cells were lysed and the clarified cytoplasmic (S20) extracts were layered onto a 30% sucrose cushion. The isolated prepolysomal (S130) and polysomal (P) fractions were separately immunoprecipitated with antibody to ${alpha}$ CP2 and ${alpha}$ CP2-KL or with preimmune serum (PI). RNA was extracted from the starting material and from each immunoprecipitate. h ${alpha}$ -globin and GAPDH mRNAs were detected by RPA. The origin of each sample is indicated above its respective lane. The positions of the RPA probes are indicated to the right of the gel. (C) Selective dissociation of ${alpha}$ CP from the antiterminated ${alpha}$ CS mRNA. MEL/tTA cells were separately transfected with pTet-WT and pTet-CS plasmids. The transfected genes were transcriptionally induced for 24 h in TET-deficient medium. TET was then added back to the medium at a concentration of 500 ng/ml for an additional 2 h. The cells were subsequently lysed and clarified (S10), and RNP complexes were immunoprecipitated with anti- ${alpha}$ CP2 sera. mRNA content in the precipitate was analyzed by RPA as described for panel B. These studies were carried out independently three times with consistent results.

The ${alpha}$ -complex is displaced by ribosome extension into the 3' UTR. Analysis of ${alpha}$ ^CS mRNA was carried out to further characterizethe association of ${alpha}$ CP with h ${alpha}$ -globin mRNA. ${alpha}$ ^CS mRNA has a singlebase substitution at the translation termination codon (UAA $->$ CAA) that allows the translating ribosome to enter the 3' UTRand translate 31 additional amino acids. This translationalextension traverses the ${alpha}$ CP binding site (Fig. 7A). MEL/tTA cellswere transfected with the ${alpha}$ ^WT and ${alpha}$ ^CS expression vectors, andtranscription was induced for 24 h by transferring the cellsto TET-deficient medium. ${alpha}$ CP complexes were immunoprecipitatedfrom the respective polysome pellets (Fig. 7B). RPA revealedthat ${alpha}$ ^WT and ${alpha}$ ^CS mRNAs were both present in the ${alpha}$ -complexes. Ofnote, the efficiency of ${alpha}$ ^CS mRNA recovery in the ${alpha}$ -complex precipitatewas somewhat lower than that of the ${alpha}$ ^WT mRNA (Fig. 7B). Thiswould be consistent with displacement of ${alpha}$ CP by the antiterminatedribosome. However, the continued association of a substantialamount of ${alpha}$ ^CS mRNA with ${alpha}$ CP suggested that this displacement waseither not efficient or that, once displaced, the ${alpha}$ CP could readilyreassociate with the mRNA. A third possibility was that the ${alpha}$ CP-bound ${alpha}$ ^CS mRNA represented the subset of newly synthesizedmRNA that had not been fully translated. To decide among thesemodels, the ${alpha}$ -complex immunoprecipitation study was repeatedon cells pulsed with ${alpha}$ ^WT or ${alpha}$ ^CS mRNAs followed by a 2-h chasein which the expression of the transfected genes was shut offby transfer of the cells into TET-containing medium. The contentof ${alpha}$ -globin mRNA in the ${alpha}$ CP-containing RNPs from these pulse-chasedcells was determined by RPA (Fig. 7C). Consistent with the knowninstability of ${alpha}$ ^CS mRNA, the total level of ${alpha}$ ^CS mRNA in thesecells at the end of the chase was lower than that of ${alpha}$ ^WT mRNA.Significantly, ${alpha}$ ^CS mRNA could no longer be detected in the ${alpha}$ -complexpellet. These results suggest that the antiterminated ribosomecan efficiently and permanently displace ${alpha}$ CP from the 3' UTRof ${alpha}$ -globin mRNA.

DISCUSSION

Top

Discussion

The present report addresses questions central to the formationand function of 3' UTR RNP complexes. The particular complexbeing studied comprises a single molecule of ${alpha}$ CP bound to a C-richmotif (7). The resultant 3' UTR ${alpha}$ -complex is a major determinantof h ${alpha}$ -globin mRNA expression. The impact of ${alpha}$ CP binding on h ${alpha}$ -globinmRNA expression is most likely mediated via mRNA stabilization(44). The data in the present report demonstrate that a fractionof cytosolic ${alpha}$ CP is polysome associated (Fig. 3 and 4) and thatthis association is RNA dependent (Fig. 5). Coimmunoprecipitationof polysomal mRNPs from cell extracts directly demonstratesspecific binding of ${alpha}$ CP to h ${alpha}$ -globin mRNA (Fig. 6). The in vivoassociation of ${alpha}$ CP with h ${alpha}$ -globin mRNA was demonstrated to berestricted to the 3' UTR C-rich motif and was subject to displacementby antiterminated ribosomes that read through the ${alpha}$ CP bindingsite (Fig. 7). We conclude from these data that ${alpha}$ CP is boundto actively translating mRNAs in vivo and that the maintenanceof the ${alpha}$ -complex is dependent on its positioning in the 3' UTR.

Lines of evidence, both in vivo and in vitro, support a rolefor the ${alpha}$ -complex in maintaining the high-level expression ofh ${alpha}$ -globin mRNA (see the introduction). Based on the presenceof poly(C)-rich cis elements in the 3' UTRs of additional highlystable mRNAs and the ability of these elements to form high-affinitymRNP complexes containing ${alpha}$ CP, it has been further suggestedthat the ${alpha}$ -complex may constitute a general determinant of mRNAstabilization (18, 34, 38). However, formation of these RNPcomplexes in vivo has not been tested, and the mode of ${alpha}$ -complexaction remains to be fully defined. Here, we show that a fractionof the cytoplasmic ${alpha}$ CPs is polysome associated and is bound toan actively translating target mRNA. These data extend our understandingof ${alpha}$ CP action and can be used to refine models of ${alpha}$ -complex-mediatedalterations in mRNA stability and/or function.

While a subpopulation of ${alpha}$ CPs in K562 cells are polysome associated,approximately 80% are prepolysomal (Fig. 3). A similar estimatehas been reported for ${alpha}$ CPs in HeLa cells (1, 15). These datasuggest that ${alpha}$ CPs may be involved in functional interactionswith nontranslating as well as translating mRNAs. This suggestionis consistent with prior studies demonstrating that the ${alpha}$ -complexcontributes to the stability of translationally blocked as wellas translationally active h ${alpha}$ -globin mRNAs (44, 45).

The relationship of subcellular ${alpha}$ CP localization with its functionscan be inferred from studies of 15-lipoxygenase (LOX) mRNA.LOX mRNAs are maintained in a stable and translationally silentstate over a several-day span of erythroid differentiation (17,40). This stored LOX mRNA is present on free cytoplasmic mRNPparticles (41). Translational silencing of LOX mRNA has beenlinked to binding of ${alpha}$ CP, as well as hnRNP K (28), to a pyrimidine-richdifferentiation control element determinant in its 3' UTR (31-33).Based upon the function of ${alpha}$ -complexes in a subset of stablemRNAs, it has been proposed that the same ${alpha}$ CP-containing complexthat mediates the translational silencing of LOX mRNA may alsomediate its stable storage in erythroid cells (18). Therefore,the observation in the present report that ${alpha}$ CP is present inboth polysomal and nonpolysomal cytosolic fractions is consistentwith studies implicating the ${alpha}$ -complex, or ${alpha}$ -like complexes, incontrol over the stability and function of both translationallyactive and translationally silenced mRNAs.

The association of ${alpha}$ CP with polysomes occurs via binding to mRNAs.This was indicated by the release of ${alpha}$ CP from the polysomes afterRNase treatment (Fig. 5) and confirmed by coimmunoprecipitationstudies (Fig. 6). The stability of this interaction at 0.5 MKCl (Fig. 4C) suggested that this interaction is quite stableat physiologic salt concentrations (0.15 M NaCl). In comparison,association of the closely related KH domain FMRP with polysomesis significantly more salt sensitive (9). This tight associationof ${alpha}$ CP with polysomal mRNAs is consistent with the high bindingaffinity that has been previously determined for the interactionof ${alpha}$ CP with its target h ${alpha}$ -globin mRNA (K_d_(app), 0.5 nM [7]).

Several members of the KH domain family of RNA binding proteinshave been shown to be associated with polysomes. One of thebest-characterized is FMRP. This protein contains three RNAbinding elements: two KH domains and an RGG box. Silencing ofthe X-linked FMR1 gene by pathological expansion of a 5' UTRtrinucleotide CGG repeat results in the highly prevalent fragileX mental retardation (FMR) syndrome (reviewed in reference 19).Early studies aimed at defining a role for the FMR protein demonstratedthat it was polysome associated and that it bound a subset ofbrain RNAs (2, 10). More recent studies have defined a restrictedsubset of FMR-targeted mRNAs and further demonstrated that lossof FMR expression results in a selective shift in the polysomedistribution of a subset of these binding targets (6). Thus,the FMR protein appears to control as-yet-undefined aspectsof translation. Whether FMR associates with nontranslating mRNAsis not known, although some FMRP is prepolysomal (12). SCP160p,a 160-kDa protein containing 14 KH domains with an undefinedrole, may target a subset of polysomal mRNAs (13, 21). CRD-BP,a four-KH domain protein that binds to the c-myc mRNA codingregion stability determinant is also ribosome associated (11).Thus, a number of KH proteins have been implicated in variousaspects of posttranscriptional control. In each case, the mechanismsand pathways are now under study but appear in at least a subsetof their actions to overlap in targeting polysomal mRNAs.

While the present study demonstrates an association of ${alpha}$ CP with ${alpha}$ -globin mRNA, the data do not address where and when the ${alpha}$ -complexforms in vivo. Some insight into the question of where ${alpha}$ -complexassembly can occur may be gleaned from the analysis of ${alpha}$ ^CS mRNA.Newly synthesized ${alpha}$ ^CS mRNA is bound by ${alpha}$ CP, while ${alpha}$ ^CS mRNAs thathave spent several hours in the cytoplasm are excluded fromthe ${alpha}$ -complex (Fig. 7). The data suggest that the majority of ${alpha}$ CP loading occurs in the nucleus or at least prior to cytoplasmictranslation; once displaced by the translating ribosome, the ${alpha}$ -complex cannot efficiently reform in the cytoplasmic compartment.The dislocation of ${alpha}$ CP from the ${alpha}$ ^CS mRNA in the cytoplasm maybe sufficient to expose these mRNAs to rate-limiting steps inmRNA decay and contribute to its dramatic destabilization.

Although ${alpha}$ CP is present on the same set of polysomes as h ${alpha}$ -globinmRNA, ${alpha}$ CP is present elsewhere in the polysome profile as welland its distribution clearly extends on polysomes too largeto accommodate ${alpha}$ -globin mRNA (Fig. 4A). This extensive distributionof ${alpha}$ CP throughout the polysome gradient and the RNA dependenceof the entire ${alpha}$ CP polysome profile (Fig. 5) suggest that ${alpha}$ CP isbinding to a wide range of cellular mRNAs. This broad rangeof target mRNAs is consistent with the suggested role of the ${alpha}$ -complex as a general determinant of mRNA stabilization (18)as well as contributing to translational controls (reviewedin reference 27). A broad-based approach to the identificationof the full spectrum of target mRNAs, based upon the high affinityand structural specificity of ${alpha}$ CP-mRNA interactions, may affordfurther insight into the role(s) that this family of RNA bindingproteins plays in gene regulation.

ACKNOWLEDGMENTS

This work was supported by NIH grants HL 65449 and CA72765 toS.A.L. and by the generous support of the Doris Duke Foundation.

FOOTNOTES

* Corresponding author. Mailing address: Departments of Genetics and Medicine, University of Pennsylvania School of Medicine, Room 428 CRB, 415 Curie Blvd., Philadelphia, PA 19104. Phone: (215) 898-7834. Fax: (215) 573-5157. E-mail: liebhabe{at}mail.med.upenn.edu.

REFERENCES

Top