Nucleic Acids Research, Vol. 19, No. 1 155

Ribosome-messenger recognition: mRNA target sites for ribosomal protein S1 Irina V.Boni*, Dilara M.Isaeva, Maxim L.Musychenko and Nina V.Tzareva M.M.Shemyakin Institute of Bioorganic Chemistry, USSR Academy of Sciences, UI.Miklucho-Maklaya, 16/10, 117871 GSP Moscow, USSR Received June 26, 1990; Revised and Accepted November 29, 1990

ABSTRACT Ribosomal protein S1 is known to play an important role in translational initiation, being directly involved in recognition and binding of mRNAs by 30S ribosomal particles. Using a specially developed procedure based on efficient crosslinking of Si to mRNA induced by UV irradiation, we have identified Si binding sites on several phage RNAs in preinitiation complexes. Targets for Si on Q,B and fr RNAs are localized upstream from the coat protein gene and contain oligo(U)-sequences. In the case of Qf RNA, this Si binding site overlaps the S-site for Q,B replicase and the site for Si binding within a binary complex. It is reasonable that similar U-rich sequences represent S1 binding sites on bacterial mRNAs. To test this idea we have used E. coil ssb mRNA prepared in vitro with the T7 promoter/RNA polymerase system. By the methods of toeprinting, enzymatic footprinting, and UV crosslinking we have shown that binding of the ssb mRNA to 30S ribosomes is S1 -dependent. The oligo(U)-sequence preceding the SD domain was found to be the target for S1. We propose that S1 binding sites, represented by pyrimidine-rich sequences upstream from the SD region, serve as determinants involved in recognition of mRNA by the ribosome. INTRODUCTION In the regulation of gene expression at the translational level the efficiency by which 30S ribosomal subunits recognize and bind translational initiation regions of mRNAs is a major factor that determines the level of protein synthesis (1,2). The mechanism for specific ribosome-messenger binding is still debated. There are two ways by which a ribosome as a ribonucleoprotein might recognize a nucleotide sequence of mRNA while searching for a translational start: RNA-RNA and RNA-protein interactions. Both kinds of interactions are known to be involved in initiation of translation (1-4). Interaction between the 3'-terminal part of 16S RNA and a purine-rich sequence preceding the initiation codon [the Shine-Dalgarno (SD) interaction] has been proven in many experiments (1-3,5,6). The vast majority of bacterial *

To whom correspondence should be addressed

and phage mRNAs use the SD interaction during translational initiation. But the presence of the 'good' SD domain situated at the proper distance from the initiation codon is not a guarantee that a sequence will be bound by the ribosome with high efficiency. To explain a wide variation in rates of translation of different genes, including different rates for different cistrons within one polycistronic mRNA (7), other factors must be taken into account. These could be a secondary structure which masks or on the contrary exposes the initiation signals (2,8-10), or other primary structure recognition determinants. Two additional sites of complementarity between mRNA and 16S-RNA were recently proposed as extra recognition signals (11-12). In our work we have concentrated on the second type of ribosome-messenger interactions: interactions of mRNA with ribosomal proteins. The possibility that site-specific mRNAprotein interactions take place during translational initiation cannot be excluded and the best candidate to have special targets on mRNAs is ribosomal protein S1. This protein has prominent RNA-binding features with high affinity to pyrimidine-rich sequences (13). Numerous observations suggest that S1 is directly involved in the process of mRNA recognition and binding (13-16). S1 is associated with 30S subunits via its N-terminal globular domain by means of protein-protein interactions (17,18), while its elongated C-terminal part [containing RNA-binding center(s)] lies free within 30S particles in such a manner as to provide binding of mRNA to the ribosome. Our aim in this work was to localize S1 binding sites on different mRNAs. We have developed a procedure that allowed us to find target sites for S1 on phage RNAs (Q,B and fr) bound to 30S subunits in preinitiation complexes. These sites are situated upstream from the coat protein genes and contain oligo(U)-tracts. To test the idea that similar sequences represent targets for S1 on bacterial mRNAs we have chosen the ssb gene of E. coli which contains an oligo(U)-sequence upstream from the SD-domain (19,20). We have shown that in vitro initiation complex formation between the ssb mRNA and 30S subunits is dependent on S1: free S1 activates the process of mRNA binding by 30S particles lacking S1, and acts as a translational repressor if added in a molar excess over the 30S particles. The target for S1 on the ssb mRNA has been found to be the oligo(U) sequence in the leader of the messenger.

156 Nucleic Acids Research, Vol. 19, No. I

MATERIALS AND METHODS 30S subunits, protein SI, antibodies against SI (aintlS1 IgG) E. coli MRE 600 30S ribosomal subunits prepared according to (21) were generously supplied by V.I. Machno and S.V. Kirillow. 30S particles lacking protein Si were prepared using poly(U)-Sepharose (13). Protein Si was isolated and purified as described previously (18). Rabbit antiserum against Si was obtained by a routine procedure (22). Purified IgG fraction was obtained by affinity chromatography on protein A-Sepharose CL-4B (Pharmacia). Immunobldtting of total 30S ribosomal proteins [fractionated according to Laemmli (23)] indicated that our preparation of anti-Si IgG reacted only with SI. Electrotransfer of proteins onto Schleicher and Schuell BA85 nitrocellulose sheets and development of immunoblots were performed according to Towbin et al. (24). Messenger RNAs RNA-containing bacteriophages were gifts of A.B. Chetverin (phage Q(3) and V.M. Berzin (phage fr). Phage RNAs were prepared by the SDS-phenol procedure (4). The plasmid pJA45 bearing the E. coli ssb gene was kindly provided by J.A. Brandsma (19). The PstI-EcoRI fragment containing the ssb gene and its leader part was ligated into the corresponding sites of pGEM 2 (Promega) underihe 1coto of T7 promoter. The resulting plasmid pGS 2.3 was linearized at the EcoRI site and used for mRNA preparaion in vitro. RNA synthesis and purification were performed using the reagents and protocols of the Riboprobe System II (Promega). Preparation of 30S-phage RNA complexes 30S subunits (40 A260 units, 3 nmol) were incubated with the phage RNA Q,B or fr (about 40 A260 units) at 37°C for 10 min in 1 ml of binding buffer (20 mM Tris-HCl,pH 7,6110 mM MgCl2/ 100 mM NH4Cl/ 6 mM 2-mercaptoethanol) and then cooled on ice. A portion of the mixture was stored as a control, while another part was irradiated at 0°C with stirring by the full light of three low pressure mercury lamps at a dose of 15 qiuanta per nucleotide (18). The incident light intensity was 2x1017 quanta/cm2/min (X=254 nm). Irradiated and control samples were supplemented with SDS and EDTA to give final concentrations of 1% and 0.02 M rtively. The m s were then fractionated on 10-30% linear sucrose grad ih 10mM Tris-HCl, pH 7.6/ 0.1% SDS/ 0.15 M LiCl/ 10 mM EDTA by centrifugation in Beckman SW-27 rotor at 120C for 20 h at 26000 rpm. Fractions containing phage RNA were pooled, precipitated with ethanol, pelleted, dissolved in binding buffer and stored at -200C. Isolation and identification of phage RNA agments bound to Si To isolate phage RNA crosslinked to protein Si a specific immunoadsorbent was prepared. 500 1l of protein A-Sepharose CL-4B gel swollen in water was shaken with 3.4 mg of purified anti-Si IgG in binding buffer at 20°C for I1 h, cnilectod by centrifugation and washed with the same bufl%r. Theb`ing of IgG to the adsorbent was about 100%. Phage RNA isol from irradiated and control samples (about 3-4 A260 units of each) was incubated overnight at 4°C in binding buffer with 40 IAI of the immunoadsorbent gel. Subsequently 20U of RNase Ti were added and samples were kept at 37°C for 30 min to yield partial digestion, sedimented, washed several imes, collectd by centrifugation and resuspended in kinase buffer (10 mM Tris-

HCl,pH 8.0/10 mM Mg acetate/15 mM 2-mercaptoethanol). To label the immobilized fragments, 100- pCi of -y-32P-ATP (Ariersham) and 5U of T4 polynucleotide kinase (Boehringer Mannheim) were added and samples were -tat 370C for 30 min, washed with water, supplemented wi201 of 10% SDS and 40 Al of IOM urea, incubated at 370C for 2h witi shaing, and sedimented. Supernatants were loaded on a 8% actylamide gel containing 0.1% SDS and 8M urea. The gels were run at 40 mA until the xylene cyanol reached te bottom. Labeled fragments crosslinked to SI were visualizd by -autoradiography, eluted from the gel overnight in a solution contaning 0.3 M NaCl/O. 1% SDS and 10 mg/ml proteinase K (Merck), extracted with phenol, precipitated with ethanol in the prmence of carrier tRNA, dissolved in 1OM urea and factdonated on a 15% sequencing gel. Individual fragments v-isualized by autoradiography were eluted with 12% NaCl04 at room temperature overnight, precipitated with 5 vol. of cold acetone in the presence of 100 Ag of Dextran 110000 (Fluka),collected by centrifuigation,dried in vacuo, dissolved in water, and subjected to sequence analysis by the enzymatic method of Donis-Keller et al. (25) using RNA sequencing kits and the corresponding protocol of P.L.-Pharmacia.

mRNA and its 36s initiation complex formation with thi ssi

depndence on S1 Initiation complex formation between 30S particles (standard or lacking SI), the prepared in vitro and purified usb mRNA and was tested uncharged initiator tRNA (Boehringer M u toeprinting techniques according to Harftz et al. (26) with mior modifications. A 17-base primer 5'-GCC.4ATGCGCC.. ACCAT-3' used in primer extension reactions had complementarity to nucleotides +76 to +92 of te ssb mRNA coding sequence (20). Usually each reaction contgined in 10 A1 of binding buffer 2.5 pmol of 30S subunits,Y25 pruol of tRNAet and 0.5 pmol of the ssb RNA annealed with the 5'4lbeled primer. In experiments on Sl dependence of initiatin complex formation,

30S*IF3* m

RNAL30S(UF3N. RNA

(1)

0-C h(254nm) cross-Linked complex

dissociation

II(

EDTA sucmse dient li)

Ams

16S

1 40*

20011Tri4iH7.

1

EDTA lOoNM LUCL '4ASOS

05N

30%

10% isolation of

m- RNA from ais-Lld complex

.132PlATP RNsosT1 %M

M9214 --b ,i enzymatic sequence

(EX}

._ preteifts K

(3

Figure 1. A scheme for the isolation and analysis of phage RNA fragments crosslinked to protein SI by UV irradiation.

GU

free SI in binding buffer or an equal volume of the buffer was added to mRNA prior to addition of 30S subunits. Probes were preincubated with mRNA at 37°C for 10 min, and then incubated for 10 min more with 30S subunits and initiator tRNA. For primer extension AMV reverse transcriptase (USSR preparation, Omutninsk) was added followed by incubation at 37°C for 30 min.

RNase protection experiments Partial digestions of the ssb RNA and its complex with S1 were performed with RNases U2 (A-specific) and T2 (non specific). The ssb mRNA (0.8 pmol per reaction) was preincubated at 37°C for 10 min in binding buffer with increasing amounts of S1 or equal volumes of binding buffer. Subsequently, 3 Ag of carrier tRNA and 3 'sequence units' of RNase U2 (Pharmacia sequence grade) or 5 x 10-3 U of RNase T2 (Sankyo) were added and digestion allowed to proceed at 37°C for 3 min (U2) or 2 min (T2). Reaction volume was 15 M1. Digestions were stopped by the addition of 15 A1 of 0.4 M Na-acetate (pH 4.8)/20 mM EDTA containing 0.5 mg/ml of carrier tRNA, followed by phenol extraction and precipitation with ethanol. Probes were analyzed by primer extension.

UV

+

.

Nucleic Acids Research, Vol. 19, No. 1 157

UV-induced crosslinking of Si to the ssb mRNA Each reaction contained in 10 /Al of binding buffer 6 pmol of the ssb mRNA, 0,15 or 30 pmol of SI and 1U of RNasin (Promega). 5 Al of each probe was irradiated with UV-light (incident light intensity was 3 x 1017 quanta/min/cm2) for 8 min at 0°C. Subsequently each irradiated and non-irradiated sample was analyzed by primer extension.

RESULTS AND DISCUSSION Site-specific interaction of protein Si with RNAs from RNAcontaining bacteriophages It is well known that on phage RNAs there is only one 'entry' site for the ribosome-the initiation region of the coat protein gene (27). 30S ribosomal subunits are able to recognize this site even in the absence of an initiator tRNA, but protein S I appears to be strongly needed for the recognition process (13-16, 27-29). To identify phage RNA regions involved in interaction with SI during initiation complex formation we have developed a special procedure which is schematized in Fig. 1. The approach is based on the ability of the RNA-binding center of SI to C A

-E L G + + A G

-

-

SI-..

03 02

-

G

U

+G

G +

+-E(02±03)

u u

u

--

C

U

A

01I--,I

C A

A

G

r

A

_

_U1

a

C

C

xc-

A

Uu

A

U

8%

_

C i

u

15%

s

I

A G.U LJ

__.m. 1.

4m_

MM

C U

1 2 4 7-A

A

B

C

Figure 2. UV crosslinking of protein SI to Qfl RNA within 30S-Q,B RNA preinitiation complex. A. Isolation of labelled fragments crosslinked to SI (step VII on Fig. 1). Autoradiogram of a 8% acrylamide gel containing 0.1% SDS and 8M urea. UV + and - indicates the lanes with irradiated and nonirradiated control samples. Exposure time-5 min. This step in the procedure gives the same picture for all phage RNAs tested. The position of the labeled material corresponds to that of S1 in the same gel as was revealed by immunoblotting (indicates by an arrow). B. Separation of Q,B RNA fragments specific for SI after removing of linked protein by proteinase K treatment (step VIII on Fig. 1). Autoradiogram of a 15% sequencing gel. XC-xylene cyanol. Exposure time- 20 min. QI-the major S1 bound fragment, Q2,Q3-the minor ones which were revealed with longer exposure. C. UV crosslinking of protein S1 to QjS RNA is sequence specific. Enzymatic sequencing of Ql fragment (B). Lanes: E-control (- enzyme), L-a ladder prepared by nonspecific cleavage, G-digestion by RNase TI, U+C- Bacillus cereus pyrimidine specific RNase, A+U-RNase Phy M, A-RNase U2. All RNases were from RNA sequencing kit (P.L.-Pharmacia). D. Comparison of sequence patterns for major (lane 1) and minor (lanes 2-6) S1 linked Q,B RNA fragments. A gap in digestion patterns indicated by an arrow might point the region where residual crosslinked peptides are located.

158 Nucleic Acids Research, Vol. 19, No. I covalently crosslink with bound polynucleotides under UVirradiation of SI-polynucleotide complexes (18). This property is observed not only in model systems: SI is the most prominent protein crosslinked to mRNAs under UV-irradiation of intact E. coli cells (30). We have investigated where S1 binds to phage RNAs within 30S preinitiation complexes formed in the absence of initiator tRNA in conditions described by Van Duin et al. (28). 30S subunits used in this experiments contained almost equimolar amounts of SI and no less than 0.2 mole of IF3 per mole of ribosomes. After irradiation with UV-light (X= 254 an) at doses which don't abolish ribosomal activity (18), the complexes were dissociated by addition of SDS and EDTA and phage RNA was separated from 16S RNA and nonlinked proteins on sucrose gradients (step El[, Fig. 1). Full length phage RNA isolated from the gradient fractions was immobilized on the specific immunoadsorbent, protein A-Sepharose contining anti-SI IgG, and partially digested with RNase TI in the presence of magnesium ions to conserve the elements of a secondary structure. All protein-free fragments were washed out and as a result only oligonucleotides covalently bound to Si remained attached. They were labeled at 5'-ends with 32P, eluted from the immunoadsorbent and purified in a SDS and urea containing acrylamide gel. The labeled material was present in the gel only in the case of irradiated complex, indicating that only SI-linked fragments could be isolated by this procedure (Fig. 2A). The products from the gel were blotted onto nitroceliose sheets and visualized with anti-SI IgG using the procedure of Towbin et al. (24) (data not shown). This was an additional proof that labeled material in the gel was specific for Si-linked fragments. The SI-bound oligonucleotides eluted from the gel were treated by proteinase K to digest bound protein and fractionated on 15% sequencing gel (Fig. 2B). The autoradiogram revealed one major and some minor fragments, which were isolated and sequenced using base-specific ribonucleases (Fig. 2C and D).

Results of sequencing allow us to determine witiin the phage RNA primary structure the location of SI-bound fagments. All the fragments, the major and minor ones, belong to a single region of phage RNA sequence which is located upstream from the coat protein gene: in the case of Q0 RNA this region comprises nucleotides 1247-1322,with A of the coat protein initiation codon being at 1344 (31 and personal communication of M. Billeter); in the case of fr RNA nucleotides 1226-1297 and 1336, correspondingly (32) (for fr RNA se4encing gels are not shown). The common features of these Ofi and fr regions are the presence of an oligo(U)-tracts which apsr to be the sites of SI-crosslinking (Fig. 3A,B). The site of ctoslinking can be identified on sequence lanes of several frag s due to a gap in the RNase cleavage patterns which can be accunted by the presence of residual crosslinked peptides (Fig. 2D). It is of special interest that in the case of Q8 RNA the region direcly involved in S I crosslinking (1292- 1306} coincides with the site bound to SI in a binary complex SI Q0 RNA (33). Moreover, the region protected by covalently bound Si against digestion with RNase TI (the main Q,3 fragment identified in our experiments) completely overlaps with the so called S-site of Q(3 replicase binding (31) (Fig. 3 C). SI is known to be an essential component of phage replicases and is responsible for recognition and binding of phage RNA (31,34). Apparently, Si participates in recognition of phage RNAs by ribosomes and by replicases by similar ways: in both cases SI recognizes and binds the same region of the phage RNA and thus provides the proper positioning of the RNA on ribosome or replicase surfaces. This 39St-1 ) + t RNAM'

_,G

_O

4

_

A. AUCUUGAUACUACCUUUAGUUCGUUUAAACACGUUCUUGAUAGUAUCUUUUUAUUAACCCAACGCG

~ ~~~~~~~I ~ 1290 1310

I

1247

B. AACGGGCGUAUACGUUAAGUCACCUUUCUCGAUGGUCCAUACCUUAGAUGCCUUGGCACUUUUCAG

1291

1226 C.

U

U C G U G U

_G _C A _ A Z

U|O |

1°|t 1 | 3 | 5 I S I ju M NWNWW_M

S 4.

G~ _U_

=

S

A-U

U-A

U

CU-ACACGU

AGUUC

C

y

U-A C-G A-U U-A A-U

AA C-G

+15

C-GUUUA A-UUCAA A-U C-G C-G C-G

AUCUUG-CUUUUUAUUAA-UCAAUUUGAUCAUG ............

... 1247

---______ site of S1 crossl inking

fMet (coat protein)

Figure 3. Nucleotide sequences within QO (A) and fr (B) RNAs involved in fomaton. The interaction with protein SI during translational initiation sites most likely participating in UV crosslinkdng are undelined. On the QO RNA sequence the site involved in binding to S1 witiin the binary coplex (33) is indicated by arrows. C-a secondary strzcture model proposed for S site of QOB replicase binding (31). QI fragment is indicated by arrows, the site of S1 crosslinking is underlined. complx

4

5

6

Fpe 4. Extension inhibition analysis (toepnntin of die b mRNA. S1 atdvates and then inhibits toeprints with 30S subunits lacking S1 [30S(-Sl)]. Extension inhibition reactions were performed as described in MateriaLs and Methds. Lane 1: primer extension without addition of 30S subunits. Lanes 2-6: toeprin in the presence of 0.25 AM of 30S(-S1) ad 2.5 M -of tRNAfMe. Concentraions of SI in reactions are indicated over las. S9 atO.6pM stmuae ribosome binding, but a 4-fold excess of SI over 30S(-SI) (lane 4) is already inhibitory for 30S initiation complex formation.

Nucleic Acids Research, Vol. 19, No. 1 159 oligo(U) sequences in Q,B and fr RNAs serve as recognition signals for 30S ribosomes during initiation complex formation.

SI-dependent binding underlies the repressor function of replicase on coat protein synthesis (27,3 1). During initiation of translation SI-messenger interaction not only enhances the affinity of the ribosome to phage RNA [30S subunits lacking SI cannot bind phage RNAs (16)] but also might direct the initiation codon of the coat protein gene to the ribosomal decoding center (see 16). At the same time, SI is known to be a translational repressor of the coat protein synthesis in vitro when added in molar excess over the ribosome (13,35). From our data this effect can be explained as competition between free SI (see 33) and SI as a part of the 30S subunit for the same site on the messenger. Although this SI-binding site does not overlap the ribosome binding site of the coat protein gene determined in RNase protection experiments (3, 36), the secondary structure of the region (Fig. 3C) is such that SI site and the initiation codon are rather close to each other. We propose that the sites containing

Table 1.

Site-specific interaction of Si with bacterial messengers Being an obligatory component of the translational machinery of E. coli and other Gram-negative bacteria, protein SI is involved in initiation of translation of almost all cellular mRNA and plays a general regulatory role in protein biosynthesis (13,37). We have proposed that interaction of SI with bacterial messengers during translational initiation might also be site-specific, as in the case of phage RNAs, with high affinity to U-rich sequences. According to Sherer et al. (38) and Dreyfus (39), the optimal ribosome binding site contains a U-rich region preceeding the SD domain. There exist also experimental evidences that U-rich sequences located upstream from the SD signal provide high efficiency of translational initiation both in natural context

Examples of Eacher ichia coli genes which contain oligo(U)-stretches in 5'-untranslated regions.

Genes..AUG ara ara ara atp bio deo

BAD FGH C E A

C

dna A eco Rl methilase fol A gal E

ACCCGUUUUUUUUGGAUGGAGUGAAACC AUG UCAAUUCAGUUUUUGCCCUACACAAAACGACACUAAAGCUGGAGAGACC AUG UGCAAUAUGGACAAUUGUUUCUUCUCUGAAUGGCGGGAGUAUGAAAAGC AUG UUUACCAACACUACUACGUUUUAACUGAAACAAACUGGAGACUGUC AUG ACCUAAAUCUUUUCAAUUUGGUUUACAAGUCGAUU AUG AAACUCGCAAGGUGAAUUUUAUUGGCGACAAGCCAGGAGAAUGAI AUG -GGGUCUUUUCGACGUACGUCAACAAUC AUG UAUUUUUUAUUUUAAUAAGGUUUUAAUUA AUG .AAUUUUUUUUAUCGGGAAIUCUCA AUG ,AUUUCtUACCAUAAGCCUAAUGGAGCGAAUL AUG (P2T

his S met S

(P1)

GGGUUCAUUUUUAUAUUCAGAAAGAGAAUAAA(GUG TIAACr'ririU7rCiArUtAIM.lTIAAr.AArUAAUr.C!C!UrCrUACt

fItH

AAUAUUCGGCACAUAAUUGCUGUUUGUUUUUUAAUCAAGGUAUCAUGA AUG

omp A

AUAUUCAUGGCGUAUUUUGGAUGAUAACGAGGCGCAAAAAAUG

pho A

GUUUUUAUUUUUUAAUGUAUUUGUACAUGGAGAAAAUAAAGUG GUUUUACAGUUCGAUUCAAUUACAGGAAGUCUACCAGAGAUG GUACAUCAUUUUCuUUUUUUACAGGGUGCAUUUACGCCqAUG

rnh RF 1 RF2 rpl rps rps rps rps ssb tar tra

J B G J M

D

CAUUUUGACUAGCUCAAUAAAAGAAAUCAGAC AUG GAAGUGAGUUCCAGAGAUUUUCUCUGGCAAACAUCCAGGAGCAAAGCU AUG

CAACUUUUAUAUAGAGGUUUUAAU1AUG UUUUGGACAAUCCUGAAUUAACAACGGAGUAUUUC UG UAAUCAUUUUCGUUUAUAAAAUAAUUGGAGCUCUGGUCUC AUG UUUUAUUACGUGUUUACOAAGCAAAACCAGGAGCUAUUU UG GUUUACACUUAUUCAGAACGAUUUUUUUCAGGAGACACGAA UG GAGAUCAACUUAUUUUCAGGAAGGUGCCU UG

UUUUUUUCUUCGGAAUAUCAUCAUCAUGAGUUUUAACGCAAAGGA

UG

160 Nucleic Acids Research, Vol. 19, No. I Table 2

5'-Untranslated regions of operons regulated by attenuation mechanism.

Genes.G his leu phe thr

G A A A trp E pyr B

UUUUUAUUGCGCGGUUGAUAACGGUUCAGACAGGUUUAAAGAGGAAUAACAA JUG UUUUUUAUGCCCGAAGCGAGGCGCUCUAAAAGAGACAAGGACCCAAACC AUG CCUUUUUUAUUGAUAACAAAAAGGCAACAC AUG UUUUUUUUUCGACCAAAGGUAACGAGGUAACAAC AUG UUUUUUUGAACAAAAUUAGAGAAUAACP AUG CUUUUUUUUGCCCAGGCGUCAGGAGAUAAAA AUG

(40-42) and when placed before other genes (9,42-44). We believe that this U-rich enhancer sequences serve as additional recognition signals and are targets for ribosonal protein Si during initiation of translation. We have noticed that 5'-untranslated regions of many E. coli genes contain oligo(U) strethes: among these there are many highly expressed in a cell-those of ribosomal proteins, translational factors etc. (Table 1). Prmoter proximal genes of operons controlled by an attenuion mechanism, where oligo(U) sequences are parts of P-' nt terminators, can also be regarded as containing such (Table 2). Here it should be added, that the SI reqient for the synthesis of anthranilate synthetase, encoded by two first genes of the E. coli trp operon, was shown in direct biochemical experiments (45), and crosslinling of Sl to the trp mRNA was observed as a result of UV-irradiation of intact E. coIl cels (30). To test the idea that oligo(U)-stretches in leader parts of bacterial genes represent targets for S1, we have studied the ssb gene (see Table 1). 30S initiation complex formation and its dependence on Si were examined in toeprint (extension inhibition) experiments (Fig. 4). As was shown for odher mRNAs, the 30S particle bound to the messenger in the pe of initiator tRNA inhibits reverse transcription from a primr ainealed downstream on the template that leads to apperance of a toqint signal at + 15 position from the first base of the inin codon (2,26). When 30S subunits completely devoided of Si werie tested by this method for their ability to bind the ssb mRNA, rnotrint signal was observed (Fig. 4). The addition of free SI produced a significant stimulation of the mRNA binding ability of the 30S subunit that led to appearance of the toeprint at + 15 position. Increasing of SI concentration in reactions resulted in disappearance of the toeprint, reflecting an inhibition of initiaion complex formation. We have obtained similar effects with other messengers (will be published elsewhere). or' and sr, Thus SI can act both as a translaonal ' e depending on its molar ratio to 30S subunits. It is r to ask how an excess of free S1 prevents binding of 30S sbunits to the ssb mRNA. As in the case of Q13 RNA (see abovie), this effect is expected if free Sl binding to the mRNA occurs at the same site as does binding of S1 when it is a part of 30S subunit. This simple competition mechanism can explain the function of SI as gene-nonspecific translational repressor in vitro. According to this model, in order to identify the target for Si witiin

0 r Pi

=

D ki .eo iz Krdase

ast".

-.. -

m I.-

w- ,

U

G

A

C

^A 27 -

_-A +0

i.4

:1!i

.'

__ U.-."

e2 ACuUAA]J.

'

"

! Jl.

'I.

st6-

'Al1ACACG

A

A

GC

A

C

G

G

C

_7-

Fle 5. Footpinting of protein SI bound to the ssb mRNA. Lanes 1,2,3-partial RNase U2 digestion of the ssb mRNA (0.8 pmol) incubated without (lane 1) or with increaing amount of SI: 8 pmol (lane 2) and 16 pmol (lane 3). Lanes 4,5-partial digestion of the ssb mRNA with RNase T2 without Si (lane 4) and in the presence of 4 itM SI (lane 5). Below-a part of the leader sequence of the ssb mRNA with indication of nucleotides protected by SI against RNase U2 (small arrows) and RNase T2 (large arrow).

translation initiation complexes, it is sufficient to find the SI binding site on the messenger within a complex between mRNA atnd pure S1. To identify the SI binding site on the ssb mRNA we have used two methods: enzymatic footprinting and UV-induced crosslinldng. Partial digestions of the free ssb RNA and its complex with protein SI with RNases U2 and T2 folowed by primer extension revealed that S I protected a region comprising an oligo(U)-tract preceding the SD domain and some upstream nucleotides (A-22,A-25,26). No pronounced protection of other mRNA parts within translational initiation region was observed

Nucleic Acids Research, Vol. 19, No. 1 161

(jiM ) Si

1.51

3 UV-i- +|- | U G C A 0

1~ ~

~

~

*0

..

_1

__so

S .

414 *

*

'

--22

--13

Figure 6. UV induced crosslinking of SI to the ssb mRNA. Primer extension analysis of the ssb mRNA irradiated by UV light (X= 254 nm) with or without SI, as indicated over lanes. The position of strong UV- and SI-dependent signal is indicated by a large arrow, that of UV-independent but SI-dependent signal-by a thin arrow. UGCA-sequence lanes.

(Fig. 5). Some reducing in intensity of the band A+6 (lane 5) is not typical for all RNase T2 protection experiments (data not shown). Changes in the bands A+9 and A+ 11 (A+ II decreases while A +9 increases in intensity) in the RNase U2 reactions reflect rather minor conformational changes than S1 binding at this site. Thus, SI protects only one extended region on the ssb mRNA that reflects specificity of S1 binding to oligo(U)-tract. The size of this protected region (9-12 nucleotides) corresponds well to the average binding site for S1 (47). The interaction of S1 with the oligo(U)-stretch was also tested in UV-crosslinking experiments (Fig. 6). The complex of S1 with the ssb mRNA was irradiated with UV-light and was analyzed by primer extension. The stop of reverse transcription within the oligo(U) sequence in the control experiment without addition of S1 appears most likely due to the UV-induced modification of U-residues (uridine photodimerization). Irradiation of the ssb mRNA in the presence of S1 with the same dose of light causes intensification and broadening of this signal proportionally to the amount of SI added, suggesting covalent bond formation between SI and U-residues of the messenger. Covalent cross-links cause termination of cDNA synthesis at the site of crosslinking. In the presence of SI the signal is a result of two independent photoreactions: photomodification of uracil residues and RNAprotein crosslinking. Broadening of the signal probably means that all U-residues in a cluster can participate in crosslinking while for UV-induced photodimerization certain positions are more favorable. Upstream from the oligo(U)-sequence an UV-independent but S 1-dependent stop is also observed (Fig. 6). At the present

Figure 7. Structural elements within the translational initiation region of mRNA and corresponding components of translational machinery involved in specific interactions. SI, attached to the 30S subunit via its N-terminal globular domain connected with elongated RNA-binding domain by flexible hinge, serves as an mRNA catching arm of the ribosome (see 14).

moment, we do not know the real origin of this signal. It may be a toeprint from the same or another SI molecule bound to the leader part of the mRNA; or it may be simply a cut in the mRNA chain caused by RNase traces in the S1 preparations. This part of the ssb RNA is very sensitive to RNase attacks (Fig. 5). More work is required to characterize the mode of SI binding with mRNAs. Taken together, our results on localization of the SI target sites on the ssb mRNA and phage RNAs Q,B and fr suggest that clusters of uracil residues are important elements involved in ribosomemessenger recognition and binding through the interaction with protein S1. The results also suggest that unstructured regions upstream from the SD domain could be reactive in the initiation process not simply because the absence of secondary structure would expose the initiation signals as was proposed in (10), but because they might serve as recognition signals bound by components of translational machinery such as ribosomal protein S1.

CONCLUDING REMARKS Examples of the genes containing in their 5'-untranslated regions U-clusters are presented in the Tables 1 and 2. As one can see these clusters are situated at different distances from the initiation codon thus being a 'drifting' element of mRNA primary structure. Our proposal that these elements work in a process of ribosomemRNA recognition and binding via interaction with SI requires some comments. Long distances between the SD domain and S1 binding site may be explained in some cases by the presence of hairpin-loop structures as in the case of Qfl RNA. If stable secondary structure is absent, flexibility of unstructured RNA chain can also allow its folding between two points of attachment to the ribosome: RNA-binding site of S1 and 3'-terminus of 16S RNA, involved in the SD interactions. We cannot also exclude that mRNA-protein S1 and SD interactions occur in some cases not simultaneously but in a consecutive order. The variable spacing from the SD of the proposed S1 binding site can explain why the distribution of bases more than 20 nucleotides upstream from the initiation codon is random. Additionally, the S1 binding site cannot be revealed by computer approaches because the affinity of SI to polynucleotides is not limited only to U-residues. Poly(A), poly(C) and in the general case (-G) sequences [such as isolated 3'-terminal fragment of 16S RNA (47)] are also good

162 Nucleic Acids Research, Vol. 19, No. I ligands for Si (13). We cannot exclude that in bacterial genes such (-G) sequences (and not only oligo(U)-tracts) might also be targets for SI during translational initiation. This could explain the universality of SI function in protein synthesis. It is possible that a hierarchy of SI binding sites with respect to their affinity to SI exists in a cell. In our scheme of arrangemnt of mRNA sequence elements involved in recogtion by components of translational machinery (Fig. 7), the S1 binding site is designated as 'enhancer' to reflect its enhancing effect on efficiency of translation and its nonfixed position with respect to the initiation codon. This model suggests that a wide variety in rates of translation of different bacterial genes can be stipulated not only by using strong or weak initiation codons or the SD domains, but also by variation in strength of S1-mRNA interactions. In a wide range of various combinations of RNA-RNA and RNA-protein interactions there exist extreme situations when only one type of interactions works during translational initiation (here we don't regard codon-anticodon interactions). This can be illustrated by the following examples. The A-protein genes of RNA phages have long SD domains and can be efficiendy tanslated by ribosomes of Gram-positive bacteria which don't contain S1 (48). 30S particles lacking the 3'terminal fragment of 16S RNA (antiSD sequence) can recognize the coat protein gene of the MS2 RNA (49) but in the absence of SI binding 30S to M$2 RNA is negligible (13,15,16). This indicates that the SD interactions play not very important role in initiation of coat protein s is. 5'-Untranslated regions of several plant virus RNAs (alfalfa mosaic virus RNA4, TMV RNA etc.) being placed before reporter genes can provide efficient traslation by E. coli ribosomes despite of lacking the SD domain (50,51). RNAprotein interactions are likely to work in these cases and S1 is the best candidate to bind (-G) leader sequences of plant virus RNAs thus providing their affinity to E. coli ribosomes.

ACKNOWLEDGEMENTS The authors tank Drs. V. Machno and S. Kirillow for providing 30S subunits of E. coli ribosomes, Dr. A. Chetverin for phage Qi, Dr. V. Berzin for phage fr, Dr. J. Brandsma for the gift of the plasmid pJA45, Dr. M.Billeter for providing data on Q(3 RNA primary structure, Dr. V. Rostapshov for synthesis of the oligonucleotide primer. We tiank Prof. L. Gold for his encouragement and support during the writing of this paper. We also thank Kathy Piekarski and Doug Irvine for their help in preparing the manuscript. REFERENCES 1. Gold, L., Pribnow, D., Schneider, T., Shinedling, S., Singer, B.S., and Stormo, G. (1981) Annu. Rev. Microbiol. 35, 365-403 2. Gold, L. (1988) Annu. Rev. Biochem. 57, 199-233. 3. Steitz, J.A., Sprague, K.U., Steege, D.A., Yuan, R.C., Laughrea, M., Moore, P.B., and Wahba, A.J. (1977) in Nucleic Acid Protein Recognition, ed. Vogel, H.J. (Academic Press, New York), pp. 493-508. 4. Chang, C., and Craven, G.R. (1977) J. Mol. Biol. 117, 401-418. 5. Hui, A., and de Boer, H.A. (1987) Proc. Nail. Acad. Sci. USA 84,

4762-4766. 6. Jacob, W.F., Santer, M., and Dahlberg, A.E. (1987) Proc. Natl. Acad. Sci. USA 84, 4757-4761. 7. Sampson, L.L., Hendrix, R.W., Huang, W.M., and Casjens, S.R. (1988) Proc. Nati. Acad. Sci. USA 85, 5439-5443. 8. Looman, A.C., Bodlander, J., de Gniyter, M., Vogelaar, A., and Van Knippenberg, P.H. (1986) Nucleic Acids Res. 14, 5481-5497. 9. Coleman, J., Inouye, M., and Nakamura, K. (1985) J. Mol. Biol. 181, 139-143.

10. Schauder, B., and McCarthy, J.E.G. (1989) Gene 78, 59-72. 11. Petersen, G.B., Stockwell, P.A., and Hill, D.F. (1988) EMBO J. 7, 3957-3962. 12. Thanaraj, T.A., and Pandit, M.W. (1989) Nucleic Acids Res. 17, 2973-2985. 13. Subramanian, A.R. (1983) Progr. Nucleic Acid Re. Mol, Biol. 28, 101-142. 14. Subramanian, A.R. (1984) Trends Biochem. Sci. 9, 491-494. 15. Suryanarayana, T., and Subramanian, A.R. (1983) Biochem. 22, 2715-2719. 16. Suryanarayana, T., and Subramanian, A.R. (1984) Biochem. 23, 1047-1051. 17. Georginis, S., and Subramanian, A.R. (1980) J. Mol. Biol. 141, 393-408. 18. Boni, I.V., Zlatkin, I.V., and Budowsky, E.I. (1982) Eur. J. Biochem 121, 371 -376. 19. Brandsma, J.A., Bosch, D., de Ruyter, M., and Van de Putte, P. (1985) Nucleic Acids Res. 13, 5095-5109. 20. Sangar, A., Williams, K.R., Chase, J.W., and Rupp, W.D. (1981) Proc. Natl. Acad. Sci. USA 78, 4274-4278. 21. Kirillov, S.V., Machno, V.I., and Semenkov, Yu. P. (1980) Nucleic Acids Res. 8, 183-196. 22. Van Dieijen, G., Zipori, P., Van Prooijen, W., and Van Duin, J. (1978) Eur. J. Biochem. 90, 571-580. 23. Laemmli, U.K. (1970) Nature (London) 227, 680-685. 24. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350-4354. 25. Donis-Keller, H., Maxam, A.M., and Gilbert, W. (1977) Nucleic Acids Res. 4, 2527-2538. 26. Hartz, D., McPheeters, D.S., Traut, R., and Gold, L. (1988) Methods Enzymol. 164, 419-425. 27. Fiers, W. (1979) Comp. Virology 13, 69-204. 28. Van Duin, J., Overbeek, G. P., and Backendorf, C. (1980) Eur. J. Biochem. 110, 593-597. 29. Backendorf, C., Overbeek, G.P., Van Boom, J.H., Van der Marel, G., Veeneman, G., and Van Duin, J. (1980) Eur. J. Biochem. 110, 599-604. 30. Schouten, J.P. (1985) J. Biol. Chem. 260, 9916-9928. 31. Meyer, F., Weber, H., and Weissman, C.H. (1981) J. Mol. Biol. 153, 631-660. 32. BerLzin, V.M., Avots, A., Jansone, J., Cintere, L., and Tzimanis, A. (1987) Nucleic Acids Res. 15, 6741. 33. Goeltz, S., and Steitz, J.A. (1977) J. Biol. Chem. 232, 5177-5179. 34. Wahba, A.J., Miller, M.J., Niveleau, A., Landers, T.A., Carmichael, G.G., Weber, K., Hawley, D.A., and Slobin, L.I. (1974) J. Biol. Chem. 249, 3314-3316. 35. Van Dieijen, G., Van Knippenberg, P.H., and Van Duin, J. (1976) Eur. J. Biochem. 64, 511-518. 36. Steitz, J.A. (1969) Nature (London) 224, 957-964. 37. Shnier, J., Stoffler, G., and Nishi, K. (1986) J. Biol. Chem. 261, 11866-11871. 38. Scherer, G.F.E., Walkinshaw, M.D., Arnott, S., and Moore, D.J. (1980) Nucleic Acids Res. 8, 3895 -3907. 39. Dreyfus, M. (1988) J. Mol. Biol. 204, 79-94. 40. Boyen, A., Piette, J., Cunin, R., and Glandsdorf, N. (1982) J. Mol. Biol. 162, 715-720. 41. McCarthy, J.E.G., Schairer, H.U., and Sebald, W. (1985) EMBO J. 4, 519-526. 42. Olins, P.O., Devine, C.S., Rangwala, Sh.H., and Kavka, K.S. (1988) Gene

73, 227-235. 43. McCarthy, J.E.G., Sebald, W., Gross, G., and Lammers, R. (1986) Gene 41, 201-206. 44. Schauder, B., Blocker, H., Frank, R., and McCarthy, J.E.G. (1987) Gene 52, 279-283. 45. Van Dieijen, G., Van Knippenberg, P.H., Van Duin, J., Koekman, B., and Pouwels, P.H. (1977) Mol. Gen. Genet. 153, 75-80. 46. Schneider, T.D., Stormo, G.D., Gold, L., and Ehrenfeucht, A. (1986) J. Mol. Biol. 188, 415-431. 47. Yuan, R.C., Steitz, J.A., Moore, P.B., and Crothers, D. M. (1979) Nucleic Acids Res. 7, 2399-2418. 48. Isono, K., and Isono S. (1976) Proc. Nail. Acad. Sci. USA 73, 767-771. 49. Melanon, P., Leclerc, D., Destroimaisons, N., and Brakder-Gingras, L. (1990) Biochem. 29, 3402-3407. 50. Galliei D.R., Sleat, D.E., Watts, J.W., Turnr, Ph.C., and Wilson, T.M.A. (1987) Nucleic Acids Res. 15, 8693 -8711. 51. Galle, D.R. and Kado, C.I. (1989) Proc. Natl. Acad. Sci. USA 86, 129-132.