THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 266, No. 5, Issue of February 15, pp. 3172-3177,1991 Printed in U.S.A.

The Adenosine-Uridine Binding Factor Recognizes the AU-rich Elements of Cytokine, Lymphokine, and Oncogene mRNAs*

(Received for publication, September 10,1990)

Paul Gillis and James S . MalterS From the Department of Pathology, Tulane University School of Medicine, New Orleans, Louisiana 701 12

Selective mRNA degradation is an important control point in the transient expression of a variety of mRNAs coding for growth regulators. A variety of labile mRNAs coding for lymphokines, cytokines, and onco- genes contain within their 3”untranslated region an AU-rich region shown to destabilize these messages. We recently identified a cytosolic protein, adenosine- uridine binding factor (AUBF), which complexes with four tandem AUUUA reiterations of a synthetic RNA transcript. We now show that AUBF forms RNase T1- resistant band-shifted complexes with a variety of in vitro transcribed mRNAs including granulocyte-mac- rophage colony-stimulating factor, interferon-y, inter- leukin-3, c-fos, and v-myc. Formation of complexes was specifically inhibited by AUUUA containing RNA, but not by irrelevant RNA. After brief ultraviolet light-induced cross-linking, AUBF-RNA complexes with the exception of c-fos comigrated on sodium do- decyl sulfate-polyacrylamide gel electrophoresis. Mu- tations within the AUUUA motifs demonstrate that both nucleotide sequence and secondary structure are important in AUBF-AUUUA RNA complex formation. Based upon these data, we suggest AUBF may interact with a variety of labile mRNAs with multiple AUUUA reiterations or single reiterations within an AU-rich 3”untranslated region.

In eukaryotic cells, cellular growth, differentiation, and response to environmental stimuli are associated with differ- ential mRNA stability (1). Transiently expressed messages coding for lymphokines, cytokines, oncogenes, and transcrip- tional activators are typically unstable with half-times of 10- 30 min (2, 3). These ephemeral mRNAs have an AU-rich sequence in the 3”untranslated region (UTR)’ which confers cytoplasmic instability (2, 3). In particular, these messages contain the pentanucleotide AUUUA either singly or in mul- tiple reiterations. When this region of GM-CSF mRNA was inserted into the normally stable @-globin mRNA, the chi- meric message was rapidly degraded (2). When the corre- sponding region was removed from c-fos, the messages pro- duced from transfected constructs acquired greater cyto- plasmic longevity and the ability to transform cells (4). In

* This work was supported by National Institutes of Health Grant CA-01427 (to J. S. ,M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

’$ To whom correspondence should be addressed. ’ The abbreviations used are: UTR, untranslated region; GM-CSF,

granulocyte-macrophage colony-stimulating factor; IL-3, interleukin- 3; AUBF, adenosine-uridine-binding factor; INF-y, interferon-y; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; SDS- PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.

addition, IL-3 was overexpressed and enhanced the tumorgen- icity of a mast cell line when truncated IL-3 constructs missing their AU-rich 3’-UTR were transfected ( 5 ) . Clearly, regulatory mechanisms must exist which can initiate and modulate the degradation of subclasses of mRNAs.

We recently identified a cytoplasmic mRNA-binding pro- tein, adenosine-uridine binding factor (AUBF), which com- plexes with AUUUA elements in synthetic RNA transcripts. Based on the specificity and kinetics of this interaction, we proposed AUBF might be involved in the turnover of labile mRNAs (6). Phorbol ester, previously shown to stabilize GM- CSF and IL-3 mRNAs in a mast cell line ( 5 ) and c-fos in a monocyte line (7), dramatically increased AUBF activity in resting peripheral blood mononuclear cells (8). These data suggested that AUBF may participate in 12-0-tetradecanoyl- phorbol-13-acetate-mediatedstabilizationofAUUUAmRNAs. Before AUBF can be considered a stabilizing factor of labile mRNAs, it is pertinent to demonstrate its ability to bind a variety of in vitro transcribed messages containing the AUUUA motif. Therefore, in this report, we have evaluated whether lymphokine, cytokine, and oncogene mRNAs can bind AUBF. In addition, we have explored the nucleotide sequence and secondary structure requirements for AUUUA RNAs to complex with AUBF.

MATERIALS AND METHODS

Cell Culture and Lysate Preparation-Jurkat cells were maintained in RPMI 1640 with 10% fetal calf serum a t a density of 0.8-2 X lo5 cells/ml. Cells were >95% viable as assessed by trypan blue exclusion. Log-phase cells were pelleted at 800 X g for 10 min, washed with phosphate-buffered saline without calcium or magnesium, resus- pended in 500 pl of phosphate-buffered saline, and transferred to a microfuge tube. After a brief sedimentation (5-10 s), the cell pellet was resuspended in 25 mM Tris, pH 7.9, and 0.6% Nonidet P-40 on ice for 5 min prior to centrifugation at 10,000 X g to remove intact cells, nuclei, and mitochondria. The supernatant was transferred to a fresh tube and snap frozen a t -70 “C.

cDNA Constructs-The production of pT3/T7-AUUUA (pAU) has been described (6). In the construction of this vector, recombinant plasmids were identified with multiple copies of the AUUUA coding oligonucleotides. In pT3/T7-AUUUA-M (pAU-M), three copies were present, which if transcribed by T7 RNA polymerase produced RNA containing 4 AUUUA and 8 UAAAU motifs. Conversely, transcrip- tion with T3 RNA polymerase generated RNA with 8 AUUUA and 4 UAAAU motifs. pT3/T7-AUUUA was shortened on the T7 promoter side by digestion with PstI alone or PstI and SphI, blunting with T4 DNA polymerase, and blunt-end religated to produce pAU -4 or -10 bases upstream of the AUUUA coding oligonucleotide. pT3-AU was constructed by blunt-end ligation into the SmaI site of pT3/T7a-19 with 28 base, AUUUA coding oligonucleotides as in the construction of pAU except that AUUUA coding oligomers are in the reverse orientation. Run-off transcripts from this construct after lineariza- tion with BamHI, SalI, or PvuII produced transcripts of 55, 65, and 268 bases, respectively. Complementary oligonucleotides (28-mers in all cases) coding for alterations of the AUUUA motif were produced on a Milligen/Biosearch Cyclone synthesizer by phosphoramidite chemistry. After purification by standard methods, the 5’ termini

3172

A UBF Binds Labile RNAs in Vitro 3173

were phosphorylated by T4 DNA kinase, and oligomers mixed in equimolar amounts, boiled in 100 mM NaCI, and slowly cooled to room temperature to anneal. Duplex oligomers were then blunt-end ligated into SmaI digested dephosphorylated pT3/T7 a19 (Bethesda Research Laboratories, Gaithersburg, MD). Recombinant clones were identified by Southern blotting with end-labeled single-stranded oli- gomers. The orientation of the insert was determined by dideoxynu- cleotide sequencing with Sequenase (US. Biochemicals, Cleveland, OH) as described by the manufacturer. GM-CSF cDNA was obtained from the Genetics Institute, Cambridge, MA. Interferon-y (INF-y) cDNA was a kind gift of Dr. Giorgio Trinchieri, Wistar Institute. IL- 3 cDNA was obtained from the ATCC. c-fos genomic DNA was provided by Dr. A-B. Shyu, Harvard University. The PstI-EcoRI v- myc cDNA fragment was obtained from Dr. Laura Levy, Tulane University. Using standard methods, the cDNAs were excised, iso- lated, and excised from agarose gels prior to ligation into the poly- linker of pT3/T7 a19. IL-3 cDNA was cloned into the KpnI and BamHI sites, INF-y cDNA at the PstI site, c-fos genomic clone at the BamHI and EcoRI sites, GM-CSF cDNA at the KpnI site, and v- myc cDNA at the Hind111 and EcoRI sites. pT3/T7 al9-fos was further truncated by digestion with Not1 and NcoI which removed approximately 2.5 kilobases of intronic and coding sequences. After digestion, the ends were filled in with Klenow and blunt-end ligated. The orientation of the inserts was determined by restriction mapping. Full length (AUUUA+) sense transcripts were produced by T3 RNA polymerase (GM-CSF, 825 bases; fos, 2690 bases; myc, 970 bases; and INF-y, 915 bases) and T7 RNA polymerase (IL-3, 900 bases). AUUUA- transcripts were produced from EcoRI-digested IL-3 cDNA (455 bases); NcoI-digested GM-CSF cDNA (570 bases); SspI-digested v-myc (630 bases), and INF cDNA (370) and PuuII- and SspI-digested c-fos DNA (680 bases). AUUUA- irrelevant RNAs were produced from linearized pT3/T7 a19 which lacked insert.

In Vitro Transcription-Radiolabeled or unlabeled RNA was pro- duced essentially as described (6) except that reactions were per- formed at 37 "C for 30 min prior to treatment for 15 min with 1 unit of RNase-free DNase (Promega, Madison, WI). Following phenol/ chloroform extraction, transcription mixtures were passed through a Sephadex G-50 minispin column prior to use. Specific activity with typically 107-108 cpm/Fg of RNA.

Band-shift Assays-4 pg of Jurkat cytoplasmic lysate (lo5 cells) was incubated with 5 X lo3 cpm of AUUUA RNA (denoted S80 in the text) in 15 mM Hepes, pH 7.9, 10 mM KCl, 10% glycerol, 5 mM MgCI,, 0.2 mM dithiothreitol, and 2 pg Escherichia coli tRNA in a total volume of 10 pl. The mixture was then incubated for 10 min at 30 "C, 20 units of RNase T1 (BRL) was added and incubation continued for 30 min at 37 "C. Samples were electrophoresed on 7% native polyacrylamide gels in 0.25 X TBE (Tris-borate-EDTA) buffer, dried, and autoradiographed overnight with Kodak X-AR film and two intensifying screens. For competition experiments, unlabeled RNAs at the amounts shown were preincubated for 10 min at 30 "C with cytoplasmic lysate in the buffer described above prior to the addition of radiolabeled RNA. For UV cross-linking studies, after the RNase T1 digest, reaction mixtures were UV irradiated in open microfuge tubes in a Stratolinker (Stratagene) for 5 min on the automatic setting prior to the addition of 2 X Laemmli buffer without 2-mercaptoethanol. The samples were boiled for 3 min and electro-

phoresed on 15% denaturing SDS-PAGE with molecular size markers (Rainbow Markers, Amersham, Chicago, IL).

RESULTS

If AUBF is involved in the post-transcriptional regulation of labile mRNAs by annealing to the AUUUA elements, it must bind such messages in an in vitro system. The AUUUA motif is a highly conserved sequence in the 3'-UTR of a variety of labile messages occurring between 1 and 8 times in GM-CSF, INF-7, IL-3, c-fos, and v-myc mRNAs (Table I). As a first step, we subcloned the cDNAs coding these labile messages into an RNA transcription vector. After lineariza- tion with the appropriate restriction endonuclease, AUUUA+ or AUUUA- run-off transcripts were produced and equimolar amounts incubated with AUBF containing Jurkat lysate. RNase T1 was added and resistant complexes were resolved by electrophoresis on 7% low ionic strength polyacrylamide gels and detected by autoradiography. As shown in Fig. 1, the 80-base RNA (S80) with 4 adjacent AUUUA reiterations (denoted pA U in Fig. 1) formed a band-shifted complex with AUBF which roughly comigrated with band-shifted complexes from all AUUUA+ mRNA species examined. Messages lack- ing the AUUUA elements failed to produce band-shifted complexes. Preincubation of the lysate with Proteinase K abolished complex formation. This data demonstrates that labile mRNAs form an RNase T1-resistant complex with a protein component of the Jurkat cell lysate.

The specificity of the protein-RNA interactions was as- sessed by competition experiments. Lysate was preincubated with increasing concentrations of unlabeled AUUUA RNA (S80) or 80-base irrelevant RNA transcribed from plasmid polylinker prior to addition of labeled probe, RNase digestion, and band-shift assay. The negative control RNA has no homology to the labile mRNAs investigated, while S80 shares only the AUUUA elements. As shown (Fig. 2), complex for- mation using full-length AUUUA INF-7, 11-3, c-fos, and v- myc mRNA probes was progressively inhibited by increasing amounts of unlabeled AUUUA RNA but unaffected by similar concentrations of control RNA. Labile RNA-protein com- plexes were not competed by similar concentrations of poly(A) or poly(U) (not shown). Analogous competition data was observed for radiolabeled GM-CSF RNA-protein complexes (not shown). The c-fos message had two complexes specifically inhibited by AUUUA RNA and two that were not. Presum- ably noncompeted complexes reflect protein interactions with c-fos RNA at different sites from the AUUUA motifs.

Proteins capable of forming stable complexes with AUUUA containing mRNAs were assessed on nonreducing SDS- PAGE to determine their molecular weight. Reaction mix-

TABLE I AUUUA motif is found in the 3'-untranslated region of GM-CSF,

interferon-y, v-myc, c-fos, and IL-3 mRNAs The numbers denote the nucleotide positions in the respective cDNAs. The sequence of the S80 AUUUA probe

is also shown for comDarison. The AUUUA motifs are underlined.

GM-CSF

IL-3

INF

TATTTATATATTTATATTTTT~T ATT~ATrTATTTATTTATTTA AGTTCATATTCCAT- 638 703 TATTTATTTATGTATGTATGTATTTATTTATTTA TTGCCTGGAGTGTGAACTGTATTTATTTTAGC 736 810 ATCTATTTATTAATATTATTTATATGGGGAATATATTTTTAG . . . . . G T ~ T A A T A G 648 727

c-fos TTAATTTATTTATTAAGATGATTCTCAGAT-TTTTTATTTTATTTTT~CTA 3970 4028

v-myc ACATTCTTCATGCTTGGGGATG~CTCTTCAACTTTTTTCTTTr~TTTTGTATTTAAGGCAT

S80

~ . "

1550 TGCCTGCAGGTCGACTCTAGAGGATCCATTTATTTATTTATTTA AGCTTGGATCCGGGTACCGAG 18 80

1614

3174 A URF Rinds Labile RNAs in Vitro

GM-CSF I L - 3 INF PAU AU+ AU- AU* AU- AU+ AU- "- Fos MYC

AU* AU- AU' AU-

1 I "

0

"

FIG. 1. AUUUA containing RNAs form band-shifted com- plexes with protein from Jurkat cytoplasmic lysate. liadiola- heled in vitro transcrihed RNAs containing (AIJ+) or larking (ACJ-) t h e AUUIIA motifs were incuhated with AUHF containing Jurkat lysate for 10 min prior to RNase T1 digestion and electrophoresis on 7% native low-ionic strength polyacrylamide gels. After drying, the gels were exposed overnight to Kodak X-AR film with two intensi- fving screens. The position of the hand-shifted complex is shown hy the upper-left hrockct, while free RNA is shown hy the hottom left hrarhrt. pAU, hand-shifted complex with SRO AUUUA RNA.

A 0

0 SO 100 IS0 SO 100 I50 0 250 500 750 250 500 750 eALMJAcpn l rp l "

". I

C D

0 30 60 90 10 60 90 PAVWA"

0 250 500 750 250 500 750 "-

c I

FIG. 2. AUUUA containing cytokine, lymphokine, and on- cogene RNA-protein band-shifted complexes are specifically competed by unlabeled AUUUA RNA. Rand-shift assavs with .lurkat lysate were performed as descrihed in the legend to Fig. 1 except that. unlaheled S8O RNA (denoted pAUUIJA) or irrelevant RNA (denoted control and transcrihed from pT3lT.i nlR polylinker, 58 hases in length) at the amounts shown were preincuhated with Jurkat lysate for I O min at 30 "C prior to the addition of radiolabeled AIJUUA containing RNAs: c-fos (Pond A ) , v-myc (Pnnrl A ) , I I J (Pond C), and INF-7 ( P n n d D ) . Position of the hand-shifted com- plexes are shown hy the n r r o d s ) on the left, while free RNA is shown I)v the hrnckrt.

tures containing lysate and AUUUA+ RNAs shown in Fig. 1 were incuhated as ahove and then exposed to UV light for 5 min hefore electrophoresis on SDS-PAGE. We have previ- ously demonstrated that AUBF. AUUUA complexes migrate at a molecular mass of 28-46 kDa in SDS-PAGE (6). These data were obtained for RNase A-protected AURF.AUUUA complexes. When RNase T1 was substituted for RNase A, AUHF.AUUUA RNA complexes migrated as two bands of approximately 44 and 65 kDa (Fig. 3, lane p A U ) . Both hands

200 + 97+

45- gr

I 1

30 + 21 +

FIG. 3. AURF-labile RNA complexes comigrate on SDS- PAGE. Radiolaheled AUIllJA HSAs as shown nt the top were incuhated with Jurkat lysate as descrihed in the legend to Fig. 2 except that after RNase T l digestion, reactions were I ' V cross-linked in open microfuge tuhes for 5 rnin. 1,aernmli hnffer without 2-merrap- toethanol was added and samples hoiled for 1% min prior to S1)S- I'A(;E. Molecular masses (kI>a) are shown along the left, while the hrackd on the right denotes the 44-kDa AI'HF-AI'III:A RNA com- plexes.

FIG. 4. Multiple AUUUA reiterations are necessary for AURF binding. HO-hase HNAs with altered A I J I ' I ' A motifs as shown along thca top were in v i f ro trnnsrrihed and screened hy the hand-shift assay as descrihed in the legend to Fig. 1 for hinding to AUI3F. T h e nrrnru on the left denotes the AI'l3F-RNA complexes while free RNA is shown hy the hrnrkct on the left.

were competed with a 100-fold excess of unlaheled S80 RNA (not shown). With the exception of c-fos which migrated as A

dominant hand of 60 kDa, all other protein-RNA complexes displayed strong hands at 44 and 65 kDa. Complexes at these molecular weights were competed hy unlaheled AUUUA RNA (not shown). The 60-kDa c-fos-protein complex was similarly competed (not shown). For GM-CSF, INF-y, and v-myc RNAs, additional higher molecular weight complexes were also present. AURF.AUUUA RNA complexes that received shorter periods of irradiation or were electrophoresed imme- diately following RNase T1 digestion demonstrated similar data (not shown). When UV cross-linking was extended he-

AUBF Binds Labile RNAs in Vitro

A A B C D E F G H

w ” ” e w D q

+

3175

FIG. 5. The AUBF-AUUUA RNA interaction is dependent on RNA of a minimum length. pAU was short- ened as described under “Materials and Methods” and in uitro transcribed with T 7 RNA polymerase to produce 76- (lane B ) or 70-base (lane C) AUUUA RNA. pT3-AU was linearized with BamHI, SalI, or PuuII and transcribed to produce 55- (lane D), 65- (lane E ) , or 268-base ( l a n e F ) transcripts, respectively. S80 AUUUA RNA is shown in lane A. S80 AUUUA RNA was digested with RNase T1 and used in the band-shift assay with (lane G ) or without (lane If) additional RNase T1 digestion following incubation with the cytoplasmic lysate. Panel A shows the band-shifted complexes (ar- row) and free probe (bracket) with these AUUUA RNAs while the lengths and orientation of the AUUUA motifs are shown in Panel B.

1 2

6 A 5’ a u u u a u u u a u u u a u u u ~ 3 ~ (80 bases) B 5’ a u u u a u u u a u u u a u u u ~ 3 ’ (76 bases) c 5’ a u u u a u u u a u u u a u u u a 3 ’ (70 bases) D 5’ auuuauuuauuuauuua 3’ (55 bases) E 5’ auuuauuuauuuauuu~3’ (65 bases) F 5‘ auuuauuuauuuauuua 1-3’ (268 bases) G 5‘ auuuauuuauuuauuua 3’ (23 bases) H 5’ auuuauuuauuuauuua 3’ (23 bases) e

3

0

FIG. 6. Secondary structure of AUUUA motifs does not af- fect AUBF binding. RNAs were transcribed from pAU-M as de- scribed under “Materials and Methods” with T 7 RNA polymerase to produce RNA with 4 AUUUA and 8 UAAAU motifs (lane 2 ) or T3 RNA polymerase to generate 8 AUUUA and 4 UAAAU motifs (lane 3 ) and used for band-shift assay as described in the legend to Fig. 2. Band-shifted AUBF-RNA complex with S80 is shown in lane 1 for comparison. Only the band-shifted complex is shown.

yond 5 min, nonspecific complex formation was enhanced (not shown). Cumulatively, these data demonstrate that AUUUA containing labile RNAs bind in uitro with AUBF.

As a variety of labile RNAs with different numbers and configurations of AUUUA elements bound to AUBF, we investigated the primary sequence requirements necessary for complex formation. A comparison of AUBF affinities as as- sessed by band-shift assays for a variety of similar sequences and reiterations is shown in Fig. 4. The S80 probe with 4 adjacent AUUUA reiterations gives a characteristically large amount of complex. However, when guanosine was substi- tuted for the middle uridine (pyrimidine to purine) in the middle of each AUUUA reiteration, AUBF recognition was entirely abolished. When the middle uridine was replaced with cytosine (pyrimidine to pyrimidine) complex formation was still observed but at one-fifth of that observed with AUUUA RNA as quantitated by scanning densitometry. The substitution of the second and fourth adenosine with a uridine

25 50 7.5 I 00 bases

to produce two adjacent AUUUUUUUA motifs bound AUBF as well as AUUUA RNA. Interestingly, individual AUUA and AUUUA motifs failed to support binding. These data suggest that AUBF can recognize AU-rich elements somewhat differ- ent from the canonical AUUUA motif. In addition, AUBF apparently has a requirement for multiple AUUUA reitera- tions.

In addition to nucleotide sequence requirements, we eval- uated the effects of the secondary structure on AUBF recog- nition and binding. To address this issue, we produced a variety of RNA probes of different lengths that all contained four adjacent reiterations of the AUUUA motif. Reduction of the 5’ leader upstream of the AUUUA elements from 35 (as in S80, lane A, Fig. 5) to 31 (lane B, Fig. 5) and 25 bases (lane C, Fig. 5) had no appreciable effect on AUBF complex for- mation. A very short probe was produced by digestion of S80 with RNase T1 to produce a 23-base fragment with a 4-base 5’ leader, and 3-base 3’-tail surrounding the 4 AUUUA motifs (S23). If S23 was used in the band-shift assay without sub- sequent digestion with RNase T I , band-shifted complex for- mation was reduced by 65% (lane H, Fig. 5). If RNase T1 digestion was included after incubation of S23 with cyto- plasmic lysate, complex formation was further reduced (lane G, Fig. 5, 17% of S80). The position of the AUUUA elements relative to the 3’ end of the mRNA was altered by producing progressively longer transcripts. As shown in lane F, Fig. 5, binding activity was not affected by the addition of several hundred bases to the 3’ end. No differences in complex formation were seen for transcripts of 450 bases as well (not shown). Thus, it appears that AUUUA ligands must be of a minimum length to complex with AUBF, suggesting that secondary structure must play a t least a partial role in rec- ognition.

Several important RNA -binding proteins recognize stem- loop structures. Computer modeling with Zucker’s folding program (9) of the AUUUA RNAs shown in Fig. 5 failed to demonstrate a consistent secondary structure (not shown). In order to further assess the effects of secondary structure, transcription vectors were produced with multimers of the

3176 A UBF Binds Labile RNAs in Vitro

AUUUA coding oligonucleotides in opposite orientations. When transcribed with T 7 RNA polymerase, RNA probes with 4 reiterations of the AUUUA motif and 8 reiterations of UAAAU were produced. When T3 RNA polymerase was used, RNAs with 8 AUUUA and 4 UAAAU reiterations were made. Based upon computer modeling, these probes are expected to form stem-loop structures with the AUUUA motifs base pair- ing with complementary UAAAU regions (not shown). As seen in Fig. 6, the AUUUA motifs are bound by AUBF even in the context of stem-loops. The slower migration of the AUBF-AUUUA RNA complexes presumably represents larger protected RNA fragments.

DISCUSSION

Conserved 3' AU-rich elements render a variety of onco- gene, lymphokine, and cytokine mRNAs susceptible to rapid degradation (1-3,5). Although the mechanisms which control the rapid turnover of important mRNAs remains enigmatic, recent studies have shown that phorbol esters, calcium iono- phore, or mitogenic antibodies enhance the half-lives of these AU-rich mRNAs (2, 5, 17). In a cell-free system, the initial endonuclease cleavage of c-myc RNA was immediately 3' of the AUUUA motifs (15). These studies support the hypothesis that cytoplasmic trans factors interact at or near the AUUUA destabilizing motifs. Protein binding could modulate the deg- radation of these messages (5, 16) accounting for variable message stability (5). In this report, we have investigated if AUBF, a recently described cytosolic AUUUA-specific mRNA-binding protein can interact in uitro with a variety of unstable AUUUA containing RNAs. In addition, we have examined the primary and secondary sequence requirements for RNA to bind to AUBF.

Band-shift assays were performed with AUBF containing Jurkat cytoplasmic lysate and a variety of in uitro transcribed labile RNAs. Band shifted complexes were observed with full- length AUUUA containing RNAs but not truncated AUUUA- negative RNAs. Complex formation was abolished if the lysate was omitted or pretreated with proteinase K, demonstrating that a cytosolic protein was part of the complex. The inter- action site of the RNA with protein was finely mapped by competition studies. An 80-base RNA whose only homology with the labile RNAs assayed here were AUUUA motifs effectively competed v-myc, c-fos, GM-CSF, IL-3, and INF-7 RNAs for binding while an AUUUA- irrelevant RNA of the same length as S80 had little or no effect. Finally, SDS-PAGE of the protein-RNA complexes with or without UV cross- linking demonstrated 44- and 65-kDa species in all cases with the exception of c-fos. These complexes were also sensitive to competition by unlabeled AUUUA RNA. Despite its unusual migration (60 kDa), the c-fos-protein complex was also com- peted by AUUUA RNA. Cumulatively, these data show that a cytosolic protein forms stable RNase T1-resistant com- plexes with labile RNAs which are mediated by the AUUUA elements. As the protein activity reported here demonstrated identical specificity, size, and biochemical characteristics with AUBF, we believe they are the same protein.

The differential sensitivity to competition of the AUBF- RNA complexes screened here suggests that AUBF may have different affinities for individual AUUUA ligands. GM-CSF, IL-3, and fos RNAs could be effectively competed by 25-50 ng of S80 RNA while myc and interferon required approxi- mately 10-fold more. The number of AUUUA motifs does not correlate with ease of competition as GM-CSF RNA contains 8 AUUUA motifs to only 1 for v-myc RNA. These data suggest that other characteristics of the RNA such as secondary or tertiary structure may influence AUBF binding. The iron

response element-binding protein recognizes a stem-loop structure located in the 5'-UTR of ferritin and 3'-UTR of transferrin mRNAs (10,11). Binding is dependent on primary sequence as well as an internal bulge and terminal loop structure (10). We have not detected significant differences in AUBF binding to AUUUA RNAs with imposed secondary structure. However, extremely short AUUUA transcripts with minimal flanking sequences were poor ligands for AUBF and are consistent with the need for a defined secondary structure.

The data shown in this report demonstrate that AUBF binds RNAs with multiple reiterations of the AUUUA motifs. Stable AUBF. AUUUA RNA complexes were seen with 80- base RNAs with 4 AUUUA or 2 AU7A elements. A single AUUA or AUUUA motif did not support binding. As AUBF bound v-myc RNA which contains AU-rich regions as well as a single AUUUA motif, a combination of both structural features is apparently also sufficient for binding. In addition, separated AUUUA motifs can also bind AUBF as INF-7, with 3 spaced elements forming stable complexes. Therefore bind- ing activity is conferred by between 2 and 4 AUUUA elements or combinations of 1 AUUUA motif and AU-rich regions which can be spaced or adjacent. Included in these mRNAs are many messages coding for lymphokines (IL-1, IL-2, IL-4, IL-5, IL-7, and IL-8), cytokines (P40, monocyte-derived neu- trophil activating factor, platelet factor 4, and tumor necrosis factor a), and oncogenes (sis, jun, ets, and raf).

We explored the effect of alterations in the AUUUA motif on AUBF binding. Substitutions of the middle uridine of the AUUUA motif with guanosine completely abolished binding while a large reduction was observed with replacement by cytosine. These data suggest that the middle uridine may be the critical base for AUBF binding. These data also show that single-base alterations of the basic AUUUA motif are suffi- cient to markedly affect AUBF recognition and/or binding. Similarly rigid primary sequence requirements have been reported for a variety of DNA-binding proteins (12-14). In- terestingly, the conversion of the second and fourth adenosine to uridine (AU7A) had no effect on binding. These data suggest that AUBF can interact with uridine-rich sequences bounded by terminal adenosines. The requirement for flanking aden- osines is underscored by the fact that poly(U) did not effec- tively compete AUBF-labile RNA binding. We do not as yet know if guanosine can substitute for adenosine at either or both termini nor the maximal length the uridine stretch can be. This is of particular interest as many labile mRNAs lack the canonical AUUUA motif but do contain AU,A-rich re- gions. Thus it is quite possible that these mRNAs are also bound by AUBF. These data also predict that a small number of mutations in the 3'-UTR of AU-rich mRNAs with conver- sion of U to C or G would prevent an interaction with AUBF. Such a change could have a major effect on the regulated turnover and accumulation of these crucial mRNAs.

Acknowledgments-We are grateful to Mike Wilson for technical assistance, Ellis Diaz for expert photography, and Bea Delucca for secretarial assistance. We also appreciate M. Gerber and P. Supakar for valuable comments.

REFERENCES 1. Raghow, R. (1987) Trends Biol. Sci. 12,358-360 2. Shaw, G., and Kamen, R. (1986) Cell 46,659-667 3. Jones, T., and Cole, M. (1987) Mol. Cell. Biol. 7, 4513-4521 4. Meijlink, F., Curran, T., Miller, A. D., and Verma, I. (1985) Proc.

5. Wodnar-Filipowicz, A., and Moroni, C. (1990) Proc. Natl. Acad.

6. Malter, J. S. (1989) Science 246, 664-666

Natl. Acad. Sci. U. S. A . 8 2 , 4987-4991

Sci. U. S. A. 87, 777-781

AUBF Binds Labile RNAs in Vitro 3177

7. Mitchell, R. L., Zokas, L., Schreiber, R. D., and Verma, I. (1985) 12. Zhang, X.-Y., Ehrlich, K. C., Wang, R., and Ehrlich, M. (1986)

8. Malter, J. S., and Yin, H. (1991) J. Biol. Chem. 266, 3167-3171 13. Gierl, A., Lutticke, S., and Saedler, H., (1988) EMBO J. 7,4045- 9. Turner, D. H., Sugimoto, N., Jaeger, J . A., Longfellow, C. E., 4053

Freier, S. M., and Kierzek, R. (1987) CoZdSpring Harbor Symp. 14. Lewis, C . D., Clark, S. P., Felsenfeld, G., and Gould, H. (1988) Quant. Biol. 52,123-133 Genes & Deu. 2,863-873

Cell 40,209-217 Nucleic Acids Res. 14,8387-8397

10. Miillner, E. W., Neupert, B., and Kiihn, L. C. (1989) Cell 58, 15. Brewer, G., and Ross, J . (1988) Mol. Cell. Biol. 8, 1697-1708

11. Leibold, E., and Munro, H. (1988) Proc. Nutl. Acad. Sci. U. S. A. 17. Lindsten, T., June, C. H., Ledbetter, J. A., Stella, G., Thompson, 373-382 16. Schuler, G. D., and Cole, M. (1988) Cell 55,1115-1122

85,2171-2175 C. B. (1989) Science 244 , 339-343