Purification and Properties of a Protein That Binds to the C-terminal Coding Region of Human c-myc mRNA*

The short half-life of c-myc mRNA is influenced by sequences in the 3‘-untranslated region and the C-termi- nal part of the coding region. In cell-free extracts, a poly-soma1 protein binds to RNA corresponding to the coding region stability determinant. This and other observations suggest that the protein is bound to polysome-as- sociated c-myc mRNA and protects the mRNA from a ribosome-associated endoribonuclease (Bernstein, P. L., Herrick, D. J., Prokipcak, R. D., and Ross, J. (1992) Genes & Dev. 6,6424354). Here, we describe a four-step purification of the binding protein: solubilization from ribo-somes, ammonium sulfate precipitation, RNA affinity chromatography, and reverse-phase high performance liquid chromatography. The 70-kDa protein can be renatured from solutions containing sodium dodecyl sulfate or organic solvents, greatly facilitating its purification. Protein binding to c-myc coding region RNA is blocked by diamide and N-ethylmaleimide, indicating a require- ment for sulfhydryl groups. The protein also binds to N-myc coding region RNA but with approximately 5-fold lower affinity than to the comparable c-myc region. Excess c-myc competitor RNA induces &fold destabilization of c-myc mRNA in cell-free

The short half-life of c-myc mRNA is influenced by sequences in the 3'-untranslated region and the C-terminal part of the coding region. In cell-free extracts, a poly-soma1 protein binds to RNA corresponding to the coding region stability determinant. This and other observations suggest that the protein is bound to polysome-associated c-myc mRNA and protects the mRNA from a ribosome-associated endoribonuclease ( Here, we describe a four-step purification of the binding protein: solubilization from ribosomes, ammonium sulfate precipitation, RNA affinity chromatography, and reverse-phase high performance liquid chromatography. The 70-kDa protein can be renatured from solutions containing sodium dodecyl sulfate or organic solvents, greatly facilitating its purification. Protein binding to c-myc coding region RNA is blocked by diamide and N-ethylmaleimide, indicating a requirement for sulfhydryl groups. The protein also binds to N-myc coding region RNA but with approximately 5-fold lower affinity than to the comparable c-myc region. Excess c-myc competitor RNA induces &fold destabilization of c-myc mRNA in cell-free mRNA decay extracts. In contrast, N-myc coding region competitor RNA has no effect on c-myc mRNAhalf-life. Therefore, the protein we have purified probably affects c-myc mRNA metabolism with high specificity. Protein production in mammalian cells is influenced by the half-life of the mRNA, which, in turn, is determined by mRNA structure and truns-acting regulatory factors (reviewed in Refs. [1][2][3][4]. mRNAs encoding abundant proteins like globin or constitutively expressed enzymes like glyceraldehyde phosphate dehydrogenase are often encoded by very stable mRNAs with half-lives of 6 h or more (reviewed in Refs. 5 and 6). This arrangement saves energy by permitting each mRNA to be translated multiple times. It is unknown whether these mRNAs are long-lived by default or because they contain stabilizing sequences. Other mRNAs whose levels fluctuate rapidly are, of necessity, short-lived and are of special interest for several reasons. First, many of them encode proteins involved * This work was supported by Grants CA23076, CA07175, and CA63676 from the National Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This with 18 U.S.C. Section 1734 solely to indicate this fact. article must therefore be hereby marked "advertisement" in accordance in cell replication or in acute responses to growth factors and inflammatory stimuli (7,8). Second, the half-lives of these mRNAs fluctuate in response to specific trans-acting factors.
Two observations illustrate the important links among mRNA half-life regulation, protein production, and cell growth control. (i) Many labile mRNAs encode labile proteins required for homeostasis and for proper cell growth and development (5, 6,9). Rapid turnover of both the mRNAs and the proteins permits rapid responses to various stimuli. (ii) Changes in the decay rate of certain mRNAs can initiate and/or promote carcinogenesis. For example, mutations in proto-oncogenes can lead to the production of mRNAs containing modified 5'or 3"untranslated regions and a wild-type coding region (10,11). The half-lives of the altered mRNAs are sometimes prolonged, so that the mRNA and protein are overproduced, leading in some cases to neoplastic transformation (12)(13)(14)(15)(16)(17)(18)(19).
Studies of unstable mRNAs have focused primarily on three questions. What sequences in these mRNAs determine their lability? What factors regulate their half-lives? What enzymes degrade them? Sequences in the 3"untranslated region and the coding region influence the cytoplasmic stability of c-fos and c-myc mRNAs (20-311, perhaps by their binding to cytosolic factors and forming mRNA-protein complexes that either shield the mRNAs from or expose them to mRNases (1,(32)(33)(34)(35). We have focused on the regulation of c-myc mRNA stability. c-myc mRNA and protein are inherently short-lived, but the mRNA half-life can vary from 15 to 120 min. I t is most stable in growing cells and in cells exposed to translational inhibitors and is least stable in resting cells (30,(36)(37)(38). Although an 8-fold change in mRNA half-life might seem modest, it apparently results in overproduction of sufficient c-myc protein to promote drastic changes in cell phenotype (see above). Therefore, the trans-acting factors that regulate c-myc mRNA stability might influence cell growth and neoplastic transformation through their effects on c-myc protein production. Several putative regulatory factors have been identified in vitro using mRNA decay and RNA-protein binding assays. One factor, a cytosolic destabilizer, reduces the c-myc mRNA half-life by 3-to 6-fold (38). This destabilizer is not detected in extracts from cells treated with cycloheximide or puromycin, suggesting that translation inhibitors stabilize c-myc mRNA in cells by blocking synthesis of a trans-acting factor, not by blocking c-my mRNA translation per se (see also Ref. 39).
A different putative regulatory protein co-sediments with polysomes through a sucrose cushion, can bind to sequences from the c-myc mRNA coding region, and thereby appears to stabilize the mRNA (1). This coding region determinant-binding protein (CRD-BP)' was identified using RNA-protein gel binding protein; RSW, ribosomal salt wash; DTT, dithiothreitol; PCR, The abbreviations used are: CRD-BP, coding region determinantpolymerase chain reaction; nt, nucleotidefs); PAGE, polyacrylamide gel electrophoresis; HPLC, high performance liquid chromatography; NEM, N-ethylmaleimide.

9261
shift and U V cross-linking assays and binds with considerable specificity to sequences located in the C-terminal 180 nucleotides of the coding region (coding region determinant (CRD)). Two observations suggest, but do not prove, that the CRD-BP functions to regulate c-myc mRNA turnover in cells. (i) The sequence to which it binds, the CRD, influences c-myc mRNA stability in intact cells (29-31, 40). Moreover, the CRD can function on its own as a destabilizer, independent of other cmyc sequences, when inserted within the coding region of a stable mRNA. (ii) The CRD-BP seems to protect c-myc mRNA from endonucleolytic attack (1). In cell-free mRNA decay assays using polysomes from cultured cells, the half-life of polysomeassociated c-myc mRNA was approximately 40 min and the mRNA was degraded in a 3' to 5' direction (41). When the reaction mixtures were supplemented with an excess of sense strand RNA corresponding to the c-myc CRD, the mRNA halflife fell to less than 5 min, and the mRNA was cleaved endonucleolytically within or close to its CRD (1). These results suggested that the exogenous RNA functioned as a competitor, titrating the CRD-BP off of c-myc mRNA and thereby exposing an endonuclease cleavage site. Consistent with this interpretation, other exogenous RNAs, including other regions of c-myc mRNA, had no effect on c-myc mRNA stability. Conversely, the c-myc CRD competitor RNA had no effect on the stability of some other mRNAs.
Here, we describe a four-step procedure for purifying the c-myc CRD-BP. Several properties of the purified protein have been determined, and its binding specificity has been assessed.
Preparation of RNAs by in Vitro Danscription-c-myc CRD cDNA was synthesized by polymerase chain reaction (PCR) using a 5' primer containing the phage SP6 RNA polymerase promoter plus c-myc cDNA bases 1705-1722 and a 3' antisense primer from c-myc cDNA bases 1869-1886 containing anEcoRI site. The c-myc portion of the 191 nucleotide (nt) RNAincluded c-myc n t 1705-1886, which encodes amino acids 381439. Nucleotide 1886 is the third base of amino acid 439, which is the C-terminal residue. N-myc cDNA template was synthesized by PCR with a 5' primer containing the SP6 promoter plus N-myc cDNA bases 1509-1525 and a 3' antisense primer from N-myc cDNA bases 1673-1690 plus an EcoRI site. The 190-nt RNA from this cDNA contained N-myc n t 1509-1690, with 1690 being the third base of the codon for the C-terminal amino acid, number 464. Polyadenylated c-myc CRD RNA used for the RNA-affinity column was transcribed from PCR-generated SP6-c-myc 1705-1886 DNA that had been cloned into pBS-pA.1, a plasmid containing a poly(dA).poly(dT) tract in pBS-KS.Ml3-(Stratagene). The RNAcontained c-myc n t 1705-1886,22 nt of plasmid sequence, and -80 3"terminal adenylate residues. @-Globin RNA was synthesized from a n AuaII-digested human P-globin cDNA clone, pSPkPC, and contained 20 n t from the plasmid followed by the first 163 n t of the 5'noncoding and coding regions of p-globin. pGEM vector RNAs were transcribed from pGEM4 DNA (Promega) digested with Hue11 (162 nt) or from pGEM3Z DNA (Promega) digested with NarI (251 nt). Transcriptions were carried out in either of two ways. The method of Melton et al.  ford (55), except where noted.

B ).
Culturing of HeLa cells and construction and transfection of the Glob-Myc-Glob gene were performed as described (1). Briefly, the Glob-Myc-Glob gene is driven by a cytomegalovirus immediate-early promoter and expresses a 1006-nt mRNAcontaining 112 n t of cytomegalovirus exon 1, 13 nt from a Hind111 linker, 419 n t from the P-globin cap site to the EcoRI site, 249 n t from the terminal coding region of c-myc mRNA (amino acids 357 through 439), 6 n t from an EcoRI linker, encoding Gln-Phe, and 207 n t from the P-globin EcoRI site to the mRNA 3' terminus (see Fig. 9). The c-myc segment (black box, Fig. 9) is in-frame, so the mRNA should utilize the normal P-globin translation initiation and termination signals, and the c-myc segment should be translated. Cell-free mRNA decay reactions contained 0.7 A,,, units of polysomes (-25 pg of RNA) and were incubated as described previously (1,43). Where indicated, 2 pg of competitor RNA was added at the start of the reactions (1). Total RNA prepared by phenol extraction was electrophoresed and blotted to a Zeta-Probe (Bio-Rad) membrane (11, which was hybridized with P-globin [32PlcDNA prepared by random priming (47). Gel Shift and W Cross-linking Analy~es-~~P-Labeled c-myc CRD RNA (5 ng) was incubated with protein samples in 10-30-pl reactions Containing 5 m~ Tris-C1, pH 7.6,2.5 m~ EDTA, 2 m~ DTT, 5% glycerol, 0.2 xm spermine, 0.5 pg/pl tRNA, with the protein contributing 10-150 m~ NaCl to the final reaction (1). (NaCl concentrations of up to 200 m~ had no effect on binding.) [32PlRNA was added last. When fractions of purified CRD-BP containing less than 0.5 mg/ml protein were assayed, carrier protein (bovine serum albumin (Boehringer Mannheim) or ovalbumin (Sigma)) was added to a final concentration of 2 mglml prior to adding the CRD-BP. Reactions were incubated for 10 min at 30 "C. RNase T l ( 1 unit, Sigma) was added, and incubation was continued for 10 min a t 30 "C. Heparin was then added to a final concentration of 5 mg/ml, and the reactions were incubated for 10 more min a t 30 "C. RNA-protein complexes were separated from free RNA fragments in a 6% nondenaturing polyacrylamide gel, visualized by autoradiography (11, and quantitated with a PhosphorImager (Molecular Dynamics). Where indicated, protein samples were pretreated with the sulfhydryl-modifying reagents N-ethylmaleimide (Sigma) or diamide (Sigma) for 10 min at 30 "C before incubation with the [3ZPlRNA probe (see figure legends). For assays using unlabeled competitor RNA, the RNA was first heated to 65 "C for 2 min, transferred to 50 "C for 5 min, and brought to room temperature for 10 min (48). The competitor was then added on ice to reaction mixtures containing all components except [32P]RNA. The probe was then added, and the reactions were transferred to 30 "C and processed as above. For W cross-linking analyses, binding reactions were carried out as for the gel shift assays. The reaction mixtures were then exposed to W light, treated with RNase A, and electrophoresed in a sodium dodecyl sulfate (SDS1-containing 8% polyacrylamide gel (1). Electrophoresis in Sodium Dodecyl Sulfate-containing Polyacrylamide Gels and Protein Renaturation-SDS polyacrylamide gel electrophoresis (SDS-PAGE) was performed under reducing conditions (49). Gels were fixed with methanol and acetic acid and stained with Coomassie Brilliant Blue or silver (Bio-Rad). To elute proteins following SDS-PAGE, gels electrophoresed with Bio-Rad standards were not fixed but were stained with 0.3 M copper chloride (50) and sliced by hand. Each slice was destained by soaking in 0.25 M Tris-C1, pH 9.0, 0.25 M EDTA (50). Destaining solution was discarded, and proteins were eluted in 50 m~ Tris-C1, pH 7.6,O.l xm EDTA, 150 m~ NaCI, 0.1% SDS, 5  Coding region determinant (CRD) RNA was identical to the ["PIRNA probe, P-globin RNA contained 183 nt.
(511, resuspended in 50-100 1. 11 of 6 M guanidine HCI in 50 mM Tris-Cl, pH 8.0,l mM EDTA, incubated a t room temperature for 15 min, brought to 10% glycerol, and dialyzed overnight a t 4 "C against a minimum 400-fold excess of 50 mM Tris-CI, pH 7.6,l mM EDTA, 100 mM NaCl,lO% glycerol, 1 mM reduced glutathione, 0.1 mM oxidized glutathione (52). Protein Purification-The following describes preparative purification of CRD-BP from -1 x 1 O ' O K562 cells (see Table I). Unless otherwise noted, all procedures were performed a t 4 "C. RSW (3-5 mg/ml protein) was diluted 1:lO with 20 mM triethanolamine, pH 7.6, 1 m~ EDTA. Crystalline ammonium sulfate was added slowly with gentle mixing to a final concentration of 40%. and the suspension was mixed for 1 h and then centrifuged a t 10,000 x g for 15 min. The supernatant was brought to 60% ammonium sulfate, and the resulting precipitate, which contained the CRD-BP, was pelleted by centrifugation and resuspended in 20 m~ triethanolamine, pH 7.6,2.5 mM EDTA, 100 m~ NaCl, 2 m~ D m , 10% glycerol (1 x RS buffer) a t a final protein concentration of -5 mg/ml.
For RNAaffinity chromatography (53), -200 pg/ml of polyadenylated CRD RNA (-500 pg of RNA for material from 1 x 1O'O cells) was incubated in solution in 1 x RS buffer, ammonium sulfate-purified protein (final concentration of 2-3 mg/ml protein), tRNA (0.25 mg/ml), and RNasin (Promega; 800 unitdml) for 15 min a t room temperature. Heparin was then added to a final concentration of 1 mg/ml, and the sample was incubated for an additional 15 min at room temperature. The reaction mix was brought to 250 mM NaCl and was applied to a poly(U)-Sepharose (Pharmacia LKB Biotechnology Inc.) column pre-equilbrated at 4 "C with TEGD buffer (20 mM triethanolamine, pH 7.6.1 mM EDTA, 10% glycerol, 2 mM D m ) plus 250 mM NaCI. All subsequent steps were carried out a t 4 "C. After applying the material, column flow was stopped for 15 min, to facilitate poly(A)-poly(U) annealing. The column was then rinsed with 5 column volumes of TEGD containing 350 mM NaCI. Bound proteins were eluted with 1-2 column volumes of TEGD with 1.5 M NaCI. Cytochrome c (Sigma) was added to the high salt eluate from the RNA affinity column to a final concentration of 75 pg/ml. The solution was adjusted to 0.1% trifluoroacetic acid, filtered through a 0.45-pm membrane (Millipore), and applied to a C4 silica based column (Vydac, 4.6 mm x 25 cm analytical) equilibrated at 56 "C with H20, 0.1% trifluoroacetic acid and connected to two Waters model 510 pumps with an automated gradient maker (54). The column was then rinsed with 1-2 column volumes of H20, 0.1% trifluoroacetic acid. Proteins were eluted at 56 "C with a gradient of acetonitrile in H,O, 0.1% trifluoroacetic acid. Cytochrome c carrier (50 pg) was added to each fraction. The fractions were lyophilized, resuspended in 6 M guanidine HCI in 50 m~ Tris-C1, pH 8.0, 1 mM EDTA and renatured as described above.
Protein concentrations were determined as per Bradford (55).

RESULTS
Purification ofthe CRD-BP-The CRD-BP co-sediments with polysomes but can be solubilized with high salt and assayed by RNA-protein gel shift, using [32P]RNA corresponding to the c-myc CRD (1). Optimal solubilization of the CRD-BP was achieved by incubating polysomes in 1.0 M NaCI, pelleting out the polysomes, and harvesting the supernatant RSW. The binding activity in the RSW showed specificity for the c-myc CRD sequence, since neither P-globin RNA (Fig. 1) nor pGEM4 vector RNA (data not shown) competed for binding as effectively as CRD RNA. Binding was consistently more efficient with RSW than with polysomes (Fig. 1, compare lunes 2 and 31, perhaps because endogenous polysomal RNAs, including c-myc mRNA, competed for the CRD [32P]RNA probe. We have not investigated further why increased radiolabeled free [32P]RNA failed t o appear in reactions containing CRD competitor RNA (lanes 5-7), but reciprocal changes in bound uersus free RNA were observed in reactions with purified CRD-BP (Fig. 8, lunes 3-7).
To determine the actual size of the CRD-BP (in the absence of cross-linked RNA) and to assess its renaturability, RSW protein was separated by SDS-PAGE, and proteins with molecular masses of -53-87 kDa were eluted, renatured, and analyzed by gel shift. We focused on this size range, because the previously characterized RNACRD-BP complex was -75 kDa (1). Most of the binding activity was observed in one slice corresponding to a M, -70 (Fig. 2). Although recovery of active renaturated protein was less than 100% in this and other experiments, this result provided confidence that the protein was renaturable.
For this reason, we devised a four-step purification strategy (Table I). Polysomes were chosen as the starting material, since RSW prepared from the polysomes contained a t least 80% of total cell CRD-BP but accounted for only 5-10% of total cytoplasmic protein of K562 cells (data not shown). For the RNA affinity step, optimal binding and recovery was achieved by incubating protein plus polyadenylated CRD RNA in solution, binding the protein-RNA complexes to a prepoured poly(U)-Sepharose column, and eluting the protein by dissociating it from the RNAwith 1.5 M NaCl (Fig. 3). only 5 1 0 % of the input protein, was present in the high salt eluate ( Fig. 3 and Table I).
High performance liquid chromatography (HPLC) using a C4 reverse-phase column was chosen as the final step, rather than SDS-PAGE, because more protein could be fractionated with very high resolution, and CRD binding activity could be renatured from acetonitrile-trifluoroacetic acid. Preparative fractionation was achieved with a shallow gradient, and approximately 90% of the recovered RNA binding activity was detected in fractions corresponding to 37% acetonitrile (Fig. 4). Approximately 7 0 4 0 % of the protein in this fraction migrated in a single band of the expected molecular mass of 70 kDa (Fig. 5A,  lane 6).
To determine whether this purified CRD-BP formed an RNAprotein complex identical to that of the RSW, fractions from each purification step were incubated with CRD 32P-RNA. Protein-RNA complexes were covalently linked by exposure to W light, treated with RNase A, and analyzed by SDS-PAGE. An identical cross-linked complex of M , 75 was observed with each sample (Fig. 6A). To confirm that the purified protein recognized c-myc CRD RNA specifically, a competition assay was performed, in which CRD-BP was incubated with CRD [32PlRNA plus 500 ng of unlabeled c-myc CRD RNA or pGEM RNA. Only CRD competitor RNA eliminated the cross-linked complex, confirming the binding specificity of the purified protein (Fig. 6B).
In summary, considerable purification of the CRD-BP was achieved with only four steps. Specific activities and precise recoveries were not obtained, because the gel shift assay for the CRD-BP is only semi-quantitative (data not shown). However, an estimate of protein recovery was made by electrophoresing HPLC-purified CRD-BP in an SDS-polyacrylamide gel along with two protein quantitation standards, staining the gel with Coomassie Blue, and comparing band intensities (Fig. 5B). Based on this assay, approximately 30-60 pg (2.5-5 x l O I 4 molecules) of CRD-BP was recovered from 1 x 10" cells (Table  I). Assuming 50% recovery of the protein, there are 50,000-100,000 molecules/cell of CRD-BP. c-myc mRNA abundance has not been quantitated carefully in cultured cells, in part because it varies, depending on growth conditions. Nevertheless, c-myc mRNA is not highly abundant, and its steady-state level can be estimated, as follows. If the average K562 cell contains 5 x lo5 mRNA molecules [20 pg/cell of total RNA x 2.5% mRNA, aver- The acetonitrile concentration was brought to 30% over 10 min and was maintained a t 30% until the cytochrome c was eluted. The column was developed with a 3650% linear acetonitrile gradient (0.5Wmin change in acetonitrile concentration; flow rate, 1 ml/min). Fractions were pooled to represent a l%/sample change in acetonitrile concentration, collected into tubes containing 50 pg of cytochrome c, and renatured. Each fraction was assayed by gel shift with c-myc CRD ""P-RNA. Lane 1, RNA probe alone, no added cellular protein. Lane 2, high salt eluate from the CRD RNA affinity column. Lanes 3-14, fractions numbered 1-12 from the reverse-phase column. These fractions eluted between 32 and 43% acetonitrile, as diagrammed above. Each lane represents a 1% increase in acetonitrile concentration from left to right. Fractions eluting before and after the 3 2 4 3 % acetonitrile range had no RNA binding activity (data not shown).
of the CRD-BP and to its potential interactions with other mRNAs besides c-myc (see below).
Sulfhydryl Requirement of Purified CRD-BP-Some RNAbinding proteins that regulate mRNA translation and metabolism are affected by divalent cations (AU binding factor; Ref. 58) or redox potential (AU binding factor and iron-responsive element-binding protein; Refs. 58-60). Magnesium was not required for binding of CRD-BP to the CRD, because gel-shifted bands were indistinguishable when reactions contained either magnesium (1.5 mM) or EDTA (2 mM) and no magnesium ( Fig.  1 and data not shown). To assess the role of sulfhydryl groups, HPLC-purified CRD-BP was preincubated with the sulfhydryl alkylating agent N-ethylmaleimide (NEM) or with the sulfhydryl oxidizing reagent diamide. The protein was then incubated with CRD ["PIRNA, and binding was assayed by gel shift. Both NEM and diamide blocked RNA binding in a concentration-dependent manner (Fig. 7). The blocking effect was reversed by excess DTT, and binding was consistently enhanced by DTT (Fig. 7A, compare lanes 3 and 8; Fig. 7B, compare lanes  3 and 7). RNA binding activity in unfractionated RSW was also sensitive to NEM and diamide (data not shown). These results indicate a stringent requirement for reducing agents to maintain CRD-BP function and suggest one possible mechanism for regulating CRD-BP activity in cells (see "Discussion").
Specificity of Purified CRD-BP: Binding to c-myc and N-myc RNAs and Effect on mRNA Stability in Vitro-Although we have not proven that the CRD-BP protects c-myc mRNA from an endoribonuclease in cells, two observations illustrate the considerable specificity with which the CRD-BP might interact with c-myc mRNA. (i) The CRD-BP does not bind to prokaryotic vector RNAs ( Ref. 1; Fig. 6B) or to globin RNA (Fig. 1). (ii) Exogenous c-myc CRD competitor RNA destabilizes c-myc mRNA in vitro but not polysome-associated c-myb or AP4 mRNAs, even though AP4 mRNA encodes a basic helix-loophelix leucine zipper motif similar to that encoded by the c-myc CRD (1). These data imply a limited function for the CRD-BP, perhaps directed specifically at c-myc mRNA. On the other hand, the quantity per cell of CRD-BP might be in excess of the amount of c-myc mRNA by as much as 1500-fold (see above). If this is so, three important questions arise. (i) Is the CRD-BP actually bound to polysome-associated c-myc mRNA? (ii) If so, does more than one CRD-BP molecule bind to each c-myc mRNA? (iii) Are any excess CRD-BP molecules bound to other mRNAs or to ribosomes? Data base searches for mRNA sequences related to the c-myc CRD revealed significant homology with the 3' end of the human N-myc mRNA coding region. This segment of both mRNAs encodes a carboxyl-terminal loop-helix and leucine zipper peptide. (The 5' end of the c-myc CRD, located at nt 1705, begins in the loop region of a helix-loop-helix motif.) The c-myc and Nmyc CRD regions are 68 and 53% identical in nucleotide and amino acid sequence, respectively, and their amino acid sequences are 74% similar when conservative amino acid changes are considered. Thus, these regions share both sequence and topological homology, and they are probably functionally similar as well. K562 cells contain little or no N-myc mRNA (61). Nevertheless, having found the region of homology, it seemed important to assess whether the N-myc "CRD" RNA segment could bind to the purified c-myc CRD-BP and, if so, could destabilize polysomal c-myc mRNA.
c-myc CRD [32P]RNA was incubated with affinity-purified CRD-BP in the presence of unlabeled c-myc or N-myc competitor RNAs. Both competitors were derived from the C-terminal 180-190 nt of their respective mRNA coding regions. The Nmyc competitor reduced binding to an extent intermediate between that of the c-myc and P-globin competitors ( Fig. 8; see also Fig. 1). To estimate the relative binding affinities of each RNA, the radioactivity in each RNA-protein complex was quantitated and plotted against the log of the competitor concentration. A 50% reduction was observed with = 80 ng of c-myc CRD competitor and with -400 ng of N-myc competitor. In contrast, 400 ng of globin competitor RNA had no effect on binding, and 800 ng of globin RNA reduced binding by 2040% (Fig. 1 and data not shown). Based on these results, the c-myc CRD-BP binds approximately 5-fold more tightly to the c-myc CRD than to the comparable N-myc sequence. The possibility that other proteins in the affinity-purified fraction interfered with the N-myc competitor seems unlikely, because the major protein in this fraction is the CRD-BP (Fig. 5A), and only one UV-crosslinked protein is observed (Fig. 6A).
To investigate whether binding of the c-myc CRD-BP to Nmyc mRNA was functionally significant, the effect of N-myc an 8% SDS-polyacrylamide gel, which was stained with silver. The protein amounts indicated below represent the percentage of total protein recovered at each step that was loaded onto each lane (see also Table  I). Lane 1, protein molecular weight standards; sizes in kDa are indicated on the left. Lane 2, 2.5 pg of polysomal protein (0.002% of total). Lane 3, 2.5 pg of ribosomal salt wash ( R S W ) protein (0.005% of total). Lane 4, 2.5 pg of 4040% ammonium sulfate-precipitated protein (0.02% of total). Lane 5,0.5 pg of CRD RNA affinity-purified material (0.05% of total). Lane 6, 3.5 pl of material purified by reverse-phase HPLC (0.4% of total). B , estimation of CRD-BP yield. Aliquots (1, 2, 5, and 10 pl of 950 pl total) of CRD-BP purified by reverse-phase HPLC were separated by SDS-PAGE in a minigel. Protein standards (fructose-&phosphate dehydrogenase (85 kDa) and glutamate dehydrogenase (56 kDa)) were co-electrophoresed, and the gel was stained with Coomassie Blue.
competitor RNA was assessed in cell-free mRNA decay reactions. Assay mixtures containing polysomes were supplemented with no exogenous RNAor with vector (pGEM), N-myc, or c-myc competitor RNAs. If the c-myc CRD-BP in polysomes were capable of binding tightly enough to the N-myc competitor RNA, then a mRNA containing the c-myc CRD should be destabilized when N-myc competitor is present. The polysomes were derived from a HeLa cell line transfected with a chimeric gene, Glob-Myc-Glob, containing 249 nt of the c-myc mRNA CRD (encoding amino acids 357-4391 embedded in frame within the coding region of human globin mRNA (Fig. 9). Glob-Myc-Glob mRNA has been used to investigate how the c-myc CRD influences mRNA half-life in cells and in vitro (1,40).
Consistent with previous experiments (11, Glob-Myc-Glob mRNA was destabilized approximately %fold in reactions containing c-myc CRD competitor RNA (Fig. 9). Moreover, a prominent endonuclease decay product was generated (asterisk). This product is thought to arise by a cleavage within the c-myc CRD following titration of the CRD-BP from the mRNA. Destabilization was not observed with pGEM or, more importantly, with N-myc competitor RNA. Endogenous c-myc mRNA was also destabilized in the presence of c-myc but not N-myc competitor RNA (data not shown). To control for the possibility that the N-myc competitor might block c-myc mRNA decay in these assays, reactions were supplemented with varying amounts of a 1:l mixture of c-and N-myc competitor RNAs. Under these conditions, Glob-Myc-Glob mRNA was degraded solely in proportion to the amount of c-myc competitor present (data not shown). The N-myc RNA in the mixture had no effect on mRNA destabilization induced by the c-myc competitor. Moreover, the N-myc competitor did not inhibit the endoribonuclease, because the endonucleolytic decay product was observed in reactions containing the c-myc plus N-myc mixture c-myc Coding Region Binding Protein 9267 (data not shown). Therefore, N-myc RNA binds to the c-myc CRD-BP in a gel shift assay (Fig. 8) but fails to induce c-myc mRNA destabilization in vitro (Fig. 9). Since the competitors were present a t a minimum 1000-fold excess over polysomal c-myc mRNA in these reactions, sufficient N-myc competitor RNA should have been available to induce a functional interaction with the CRD-BP, had one occurred. The simplest explanation for these results is that the binding affinity of N-myc RNA for the CRD-BP is too low to titrate the protein away from polysome-associated c-myc mRNA.

DISCUSSION
The c-myc CRD-BP was initially identified by its capacity to bind to the CRD of c-myc mRNA and to protect the mRNA from endonucleolytic attack in vitro. The half-lives of other mRNAs, including c-fos and p-tubulin, are also influenced by sequences in their coding regions (26)(27)(28)(62)(63)(64)(65), and several proteins bind specifically to the c-fos coding region determinant (33). Therefore, the c-myc CRD-BP might be a paradigm for RNAbinding proteins that regulate mRNA translation andor stability by interacting with the mRNA coding region. For this reason, the CRD-BP was purified with the goals of characterizing its properties and binding specificity, determining if and when it binds to c-myc mRNA in cells and understanding how it might protect and stabilize the mRNA.
Purification of the CRD-BP was facilitated by its specificity for c-myc CRD RNA and by its renaturability. Other sequenceor nucleotide-specific RNA-binding proteins, including the yeast poly(A)-binding protein (66) and the iron-responsive element-binding protein (67), can also be renatured following SDS-PAGE. These results suggest that RNA-binding proteins might be renatured more readily than most enzymes. If so, the high resolution afforded by gel electrophoresis and reversephase HPLC should facilitate their purification. The four-step procedure for human c-myc CRD-BP (Table I) yields highly purified material containing one major 70-kDa protein (Fig. 5).
Purified CRD-BP and anti-CRD-BP antibodies will be required to determine the intracellular function of the protein, but several observations already suggest that the CRD-BP affects c-myc mRNA stability. (i) c-myc mRNA is destabilized in vitro under conditions likely to dissociate the protein from the mRNA (1). (ii) The c-myc CRD affects mRNA half-life in cells (29,30,40). On the other hand, unequivocal links between the CRD-BP and mRNA metabolism have not been established. We do not know if the CRD-BP is associated with c-myc mRNA or whether the CRD-BP affects other aspects of c-myc mRNA metabolism or function (transport, translation, ek.). Whatever its functions may be, the effect of sulfhydryl group modification suggests one possible mechanism for regulating its binding properties in cells (Fig. 7). Other nucleic acid-binding proteins, including transcription factors (USF43 (681, NF-KB (69), and fos-jun (70)) and RNA-binding proteins (AU binding factor (58); iron-responsive element-binding protein (59-60)), are also affected by redox potential.
Multiple and perhaps even redundant control processes are exploited by mammalian cells to maintain c-myc protein levels within certain limits compatible with cell viability (9,(71)(72)(73)(74)(75)(76)(77). If the CRD and CRD-BP are part of this regulatory network, they might function only under special limited conditions, for ex-  (1). Where noted, decay reactions were supplemented with 2 pg of the following competitor RNAs: c-myc CRD, N-myc CRD, or pGEM3Z. Reactions were incubated for the indicated times, and total RNA was isolated, electrophoresed, transferred by blotting to a membrane, and hybridized with a :'2PP-labeled 6-globin probe. The asterisk indicates the 5' mRNA decay product resulting from cleavage of Glob-Myc-Glob mRNA within its CRD. Diagram: 1006-nt mRNA (without poly(A)) generated from the cytomegalovirus promoter-driven Glob-Myc-Glob gene. Thin and thick rectangles indicate untranslated and coding regions, respectively. A, indicates poly (A). The stippled region indicates a 112-nt 5"untranslated segment from cytomegalovirus. Unfilled and black rectangles indicate human 6-globin and c-myc sequences, respectively. The c-myc segment was derived entirely from the region encoding amino acids 357-439. The reading frame is maintained, so that translation begins and ends at the indicated AUG and UAA sites.
ample, during particular stages of cell differentiation or in response to particular environmental factors. It would not be unusual for an essential mRNA like c-myc to be regulated by a RNA-binding protein with a unique target sequence. The metabolism and/or function of other RNAs is influenced by RNAbinding proteins with specific functions (the iron responsive element-binding protein; proteins involved in RNA splicing (reviewed in Ref. 78); viral RNA-binding proteins (Refs. 79-82; reviewed in Ref. 83)). Ribosome formation itself requires sequential binding of proteins to specific sites in rRNAs (reviewed in Ref. 84), and some RNA-binding proteins direct mRNAs to specific intracellular locations (85). Two observations highlight the specific and perhaps unique functional interaction of the c-myc CRD-BP with the c-myc CRD. (i) The CRD-BP can bind with reduced affinity to N-myc mRNA sequences (Fig. 8), but N-myc competitor RNA does not induce c-myc mRNA destabilization in vitro (Fig. 9). (ii) Although the coding region determinants of c-myc and c-fos mRNAs are both purine-rich and readily interact with cytoplasmic proteins, these proteins are not identical and differ in size, binding specificity, and other properties (33). Moreover, poly(A) and poly(G) compete with c-fos RNA for binding to fos-binding proteins (33) but do not compete with myc CRD RNA for binding to purified myc CRD-BP (data not shown). Therefore, no evidence exists that the c-myc CRD-BP affects other mRNAs besides c-myc, in spite of estimates indicating a large excess of CRD-BP relative to c-myc mRNA in cells (see "Results"). If this apparent excess CRD-BP is confirmed by more quantitative assays, it will be important to determine whether multiple CRD-BP molecules bind to each c-myc CRD and whether as yet unidentified mRNAs interact with the CRD-BP.