MOLECULAR AND CELLULAR BIOLOGY, JUlY 1990, p. 3441-3455 0270-7306/90/073441-15$02.00/0 Copyright C 1990, American Society for Microbiology

mRNA Poly(A) Tail,

a

Vol. 10, No. 7

3' Enhancer of Translational Initiation

DAVID MUNROEt AND ALLAN JACOBSON* Department of Molecular Genetics and Microbiology, University of Massachusetts Medical School, Worcester, Massachusetts 01655 Received 2 February 1990/Accepted 4 April 1990

To evaluate the hypothesis that the 3' poly(A) tract of mRNA plays a role in translational initiation, we constructed derivatives of pSP65 which direct the in vitro synthesis of mRNAs with different poly(A) tail lengths and compared, in reticulocyte extracts, the relative efficiencies with which such mRNAs were translated, degraded, recruited into polysomes, and assembled into messenger ribonucleoproteins or intermediates in the translational initiation pathway. Relative to mRNAs which were polyadenylated, we found that nonpolyadenylated [poly(A)-] mRNAs had a reduced translational capacity which was not due to an increase in their decay rates, but was attributable to a reduction in their efficiency of recruitment into polysomes. The defect in poly(A)- mRNAs affected a late step in translational initiation, was distinct from the phenotype associated with cap-deficient mRNAs, and resulted in a reduced ability to form 80S initiation complexes. Moreover, poly(A) added in trans inhibited translation from capped polyadenylated mRNAs but stimulated translation from capped poly(A)- mRNAs. We suggest that the presence of a 3' poly(A) tail may facilitate the binding of an initiation factor or ribosomal subunit at the mRNA 5' end.

Most eucaryotic mRNAs have a sequence of polyadenylic acid [poly(A)] at their 3' termini (14, 19, 39, 48, 50). These poly(A) "tails" are added post-transcriptionally in the nucleus, with an initial length, in mammalian cells, of approximately 200 to 250 adenylate residues (9). Following transport of mRNA to the cytoplasm, poly(A) tracts are gradually shortened so that, in steady state, poly(A) tail lengths are heterogeneous, ranging from 50 to 70 A's (2, 8, 10, 38, 59, 80). Although it has been almost two decades since the discovery of these poly(A) tracts, their function(s) has yet to be clarified. Earlier results from our laboratory (37, 53, 54, 61, 79) led us to propose that poly(A) has a role in translation. More specifically, we proposed that an interaction of the cytoplasmic poly(A)-binding protein (PABP) with a critical minimum length of poly(A) facilitates the initiation of translation of polyadenylated [poly(A)+] but not nonpolyadenylated [poly(A)-] mRNAs. The results of several different experimental approaches have provided evidence which indirectly supports this hypothesis. These results include (i) the correlation of specific changes in mRNA poly(A) tail length with translational efficiency following fertilization in Spisula oocytes (72), during early development in Dictyostelium discoideum (61), oocyte maturation in Xenopus laevis (34, 55), oocyte activation in mice (33, 84, 86), salivary gland development in Drosophila melanogaster (67), and in response to physiologic stimuli in the rat hypothalamus (11, 69, 89); (ii) the higher translational activity of poly(A)+ as opposed to poly(A)- mRNAs in vitro, as demonstrated in systems as diverse as cell extracts derived from rabbit reticulocytes (16, 73), Ehrlich ascites tumor cells (31), and wheat germ (73), as well as in microinjected Xenopus oocytes (15, 21, 34) and electroporated tobacco protoplasts (22); (iii) a correlation between the abundance and stability of PABPs and the rate of translational initiation in developing or heat-shocked Dictyostelium discoideum amoebae (53,

54); (iv) the demonstration that exogenous poly(A) is a potent and specific inhibitor of the in vitro translation of poly(A)+ but not poly(A)- mRNAs in rabbit reticulocyte (4, 28, 37), wheat germ (4), L-cell (49), and pea seed (81) extracts; and (v) the demonstration that purified PABP can stimulate translation in vitro (81). The experiments in this study were designed to provide a direct test of our model. We have constructed a set of plasmids which direct the in vitro synthesis of a set of mRNAs differing only in their respective poly(A) tail lengths and compared the efficiencies with which such synthetic mRNAs were translated, degraded, recruited into polysomes, and assembled into messenger ribonucleoproteins (mRNPs) or intermediates in the translational initiation pathway in a reticulocyte cell-free translation system. In addition, we have reevaluated the effects of exogenous competitor poly(A) in vitro by using mRNAs that differ either in cap or poly(A) tail status or both. Our results demonstrate that poly(A) is involved in a late step in the initiation of translation and lead us to suggest that poly(A) is the formal equivalent of a transcriptional enhancer, i.e., that PABP bound at the 3' end of mRNA may facilitate the binding of an initiation factor or ribosomal subunit at the mRNA 5' end. MATERIALS AND METHODS Plasmid constructions. A series of vectors for the in vitro transcription of polyadenylated RNAs, designated pSP65An, were derived by insertion of different lengths of poly(dA:dT) within the polylinker of pSP65 (56). These vectors were constructed by conventional techniques (52) as follows (Fig. 1). pSP65 DNA was linearized by digestion with PstI, tailed to various extents with dATP and terminal transferase (Ratliff Biochemicals), digested with HindIII, and separated from the low-molecular-weight products of digestion by preparative electrophoresis in low-melting-point agarose. A 7-base oligonucleotide, including part of a Hindlll recognition site (5'-AGCTTTT-3'), was phosphorylated at its 5' terminus with ATP and polynucleotide kinase, tailed to various extents with dTTP and terminal transferase, purified

* Corresponding author. t Present address: Center for Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02139.

3441

3442

.

MOL. CELL. BIOL.

MUNROE AND JACOBSON

B1 2 3 4 5 6

A

5. CTGCAGCCC;AGCIT .....r 3

5'

.GACGTCGGGTTCGAA

Pa_ _.ll

bull

C

2

3

4

5 6

-7

M * 235

PsI

.....CTGCA

GCCCAAGCTT.

ACGCGGGTTCGAA ......

* 55 d(A) 'a limg

* 34,

.....CTGCAAAAAAAAA() G .

9

GCCCAAGCTT

* 25

(n)M.AAAAAAAOACGTCGGGTTCGAA ........ Hinc

gof

* IC'r

pu-fica1si' D

CTGCMAAAA.AAA )

G

AGCTT.. A..

ep

0*A 5 -A GCLZT3

d(7T) lailing pp

AGCTTTTTITLTLL .....

CTGCAAAA.AAAA tn)

A(n)AGCTT

.....G TTTTTTTTTTTTTTCGA

|

A....

DNA lgalion. transforma!cr

P

FIG. 1. Construction and characterization of pSP65A vectors and their transcripts. (A) Summary of construction scheme for pSP65A vectors (see Materials and Methods for details). (B) Size of dA:dT inserts. Inserts of (dA:dT)" within the polylinker of pSP65A plasmids (and their RBG derivatives) were characterized by restriction digests (EcoRI-HindIII) which bracketed the A/T region. Digestion products were fractionated on a 5% polyacrylamide gel (in TBE) and stained with ethidium bromide. Lane 1, pRBG with 0 A's; lane 2, pRBG with 23 A's; lane 3, pRBG with 32 A's; lane 4, pRBG with 68 A's; lane 5, Hinfl-cut pBR322; lane 6, an example of a pRBG plasmid preparation with a heterogeneous A/T region arising from deletions accumulated during replication in E. coli HB101. (C) Poly(A) tail lengths of in vitro transcripts. RNA transcribed in vitro from plasmids containing different lengths of dA:dT insert (and labeled with [a-32P]ATP) were digested with either RNases A and T1 or RNases A, TI, and T2. Digestion products were purified on oligo(dT)-cellulose and analyzed on a 5% polyacrylamide gel (in 8 M urea and TBE). Odd-numbered lanes were digested with RNases A, Tl, and T2; even-numbered lanes were digested with RNases A and

VOL. 10, 1990

by DEAE-cellulose chromatography, and ligated to the gel-purified vector molecules. Templates for transcription of polyadenylated rabbit globin mRNAs were constructed by using pGEMlRPG and pfGl (gifts from Argiris Efstratiadis). The 0.34-kilobase-pair (kbp) AccI-BglI fragment from p,G1 (20) was gel purified and ligated with the gel-purified 3.1-kbp AccI-BamHI fragment from pGEMlRfiG to produce pDMR3G. The 0.95-kbp SphI-SmaI fragment from pDMRf3G was then gel purified and ligated with the gel-purified 3-kbp SphI-SmaI fragment from selected pSP65A vectors. After linearization with HindIll, transcripts from these templates include 7 nucleotides of 5' noncoding sequence derived from the vector, 10 nucleotides of rabbit f-globin 5' untranscribed flanking sequence, the entire rabbit P-globin 5' untranslated region and coding region, 9 out of 95 nucleotides of 3' untranslated region, 11 nucleotides derived from the vector, and a poly(A) tail of specified length. Templates for transcription of polyadenylated vesicular stomatitis virus N (VSV.N) mRNAs were constructed by using plasmid JS223 (82) (a gift from Jack Rose). The 1.3-kbp XhoI fragment from JS223, containing the VSV.N cDNA, was gel purified and ligated into the Sall site of pSP65, producing AS65N. AS65N was digested with SmaI and NdeI, fragment ends were made blunt by filling in with DNA polymerase, and the 1.3-kbp fragment was gel purified and ligated into SmaI-cut, phosphatase-treated pSP65A vectors. After linearization with HindIll, transcipts from these templates include 16 nucleotides of 5' noncoding sequence derived from the vector, the entire VSV.N 5' untranslated region, coding region, and 3' untranslated region, 15 nucleotides derived from the vector, and a poly(A) tail of specified length. Poly(A) tail length determinations. All plasmid templates were initially screened to determine the length of the poly(dA:dT) insert, located just 5' of the HindIII site in the polylinker, by digestion with HindIll and EcoRI (pSP65A plasmids and rabbit 3-globin constructs) or HindIII and BglI (VSV.N gene constructs). Restriction digests of plasmids containing poly(dA:dT) inserts were directly compared with otherwise identical plasmids without such inserts on 5% polyacrylamide gels (in TBE [52]). Because replication in bacterial hosts often caused deletions within the poly(dA: dT) insert [resulting in plasmid preparations with heterogeneous poly(dA:dT) lengths; for example, see Fig. 1B], plasmids were prepared in a mutant host (HB1O1A) which allows plasmid replication without deletion (D. Munroe, Ph.D. thesis, University of Massachusetts Medical School, Worcester, Mass., 1989). To measure poly(A) tail lengths, RNA was transcribed in vitro (with [a-32P]ATP as the label), and 5 x 105 cpm of each transcript was digested either with RNases A and T1 (100 ,ug/ml each) or T2 (5.0 U/ml) in 2x SSC (0.3 M NaCl, 0.03 M sodium citrate) at 37°C for 30 to 60 min. Digestion products were purified by an oligo(dT)-cellulose batch procedure (36) and analyzed on 5% polyacrylamide gels run in the presence of 8 M urea and TBE (52). In vitro transcription and capping of synthetic mRNAs. In vitro transcription and cotranscriptional capping, directed

POLY(A) ENHANCES TRANSLATIONAL INITIATION

3443

by SP6 RNA polymerase (Boehringer), were done as described previously (43, 56). After transcription, DNA templates were removed by digestion with RQ1 DNase (1 U/,ug of DNA; Promega). Transcripts were further purified by two phenol-CHCl3 (1:1) extractions, CHC13 extraction, Sephedex G-50 spin chromatography, and ethanol precipitation. The integrity of all RNA samples was verified by electrophoresis in 4 to 6% polyacrylamide (acrylamide-bisacrylamide, 20:1) gels containing 8 M urea and buffered with TBE (52). Analysis of RNA 5' termini. Transcripts labeled in vitro with [oa-32P]ATP (106 cpm) were digested with RNases A (100 ,ug/ml), T1 (100 ,ug/ml), and T2 (5.0 U/ml) in 2x SSC at 37°C for 30 to 60 min. Reaction products were separated by two-dimensional thin-layer chromatography on cellulose plates (Analtech) in isobutyric acid-concentrated NH40HH20 (pH 4.3) (577:38:385) in the first dimension and saturated (NH4)2SO4-1 M sodium acetate-isopropanol (80:18:2) in the second dimension as described previously (43). In vitro protein synthesis. mRNA-dependent translation extracts were prepared from commercial rabbit reticulocyte lysate (Promega) as described previously (59). Reaction

conditions for in vitro protein synthesis and procedures for analyzing the translation products of individual mRNAs have been described (37, 59). Measurement of RNA decay rates in reticulocyte lysates. Synthetic RNAs (