Cell, Vol. 68, 119-131,

January

10, 1992, Copyright

0 1992 by Cell Press

Prokaryotic-like Cis Elements in the Cap-Independent Internal Initiation of Translation on Picornavirus RNA Evgeny V. Pilipenko,’ Anatoly P. Gmyl,’ Svetlana V. Maslova,’ Yuri V. Svitkin,’ Alexander N. Sinyakov,’ and Vadim I. Agol’t *Institute of Poliomyelitis and Viral Encephalitides The USSR Academy of Medical Sciences Moscow Region 142782 USSR tMoscow State University Moscow 119899 USSR

Summary Initiation of translation on picornavirus RNAs is accomplished through internal binding of ribosomes to a complex cis-acting element. Here we show that efficient function of this element involves two appropriately spaced smaller elements: UUUCC and an AUG. This conclusion emerged from analysis of the genome structures of spontaneous revertants of mutant polioviruses with extended insertions between the UUUCC and AUG motifs. It was confirmed by the results obtained with specially designed constructs. A similarity to the prokaryotlc translation initiation mechanism, which involves the Shine-Dalgarno sequence, is emphasized, but in the picornavirus system the position of the UUUCC must be strictly fixed relative to upstream cis-acting elements, and the AUG may not necessarily serve as an initiation codon. Introduction Initiation of translation on the majority of eukaryotic mRNA templates is accomplished by a cap- and 5’ end-dependent mechanism involving scanning of 5’ untranslated regions (UTRs) by small ribosomal subunits (Sonenberg, 1988; Kozak, 1990). Aset of eukaryotic mRNAs, however, appear to exploit a different initiation mode based on a poorly understood cap- and 5’ end-independent internal ribosome binding. The most-studied example of such dissident templates is represented by the genomes of picornaviruses. These animal viruses include several genera, Enterovirus (e.g., poliovirus), Rhinovirus, Cardiovirus (encephalomyocarditis virus), Aphthovirus (foot-and-mouth disease virus), and some unclassified representatives (Rueckert, 1985). They contain a 7500-8000 nucleotide single-strand RNA of mRNA-like polarity. This RNA has no cap and possesses a long (800-1200 nucleotides) UTR with multiple cryptic AUGs. The initiation of translation on the picornaviral RNAs appears to involve binding of the small ribosomal subunit to a complex internal cis-acting element several hundred nucleotides long (Pelletier and Sonenberg; 1988; Trono et al., 1988a; BienkowskaSzewczyk and Ehrenfeld, 1988; Jang et al., 1988, 1989; Belsham and Brangwyn, 1990; Kuhn et al., 1990; for reviews, see Sonenberg, 1990; Jackson et al., 1990; Agol,

1991). Two types of conserved secondary structure models for this element have been proposed, one for enteroand rhinoviruses (ERV) (Blinov et al., 1988; Rivera et al., 1988; Pilipenko et al., 1989a; Skinner et al., 1989) and the other for cardio- and aphthoviruses (CAV) (Pilipenko et al., 1989a). Little is known about the details of ribosome-template recognition in these systems, though several potentially important positions within the cis element have attracted considerable interest. Mutations in the 472-481 region of poliovirus RNA affect the efficiency of tralrslation both in vitro (Svitkin et al., 1985, 1990; Muzychenko et al., 1991) and in vivo (La Monica and Racaniello, 1989). There are several conserved oligonucleotide stretches within the element (Toyoda et al., 1984; Kuge and Nomoto, 1987; Rivera et al., 1988; Pilipenko et al., 1989a, 1989b; Jackson et al., 1990) including a pyrimidine-rich tract present in both the ERV and CAV genomes, albeit at different locations. Mutations in this tract may or may not affect the viability and phenotype of the viruses (Kuge and Nomoto, 1987; lizuka et al., 1989) as well as the in vitro template activity of the respective RNAs (Jang and Wimmer, 1990; Kuhn et al., 1990; Pestova et al., 1991). Noteworthy, not far downstream of the pyrimidine-rich stretch, there is invariably an AUG triplet, either cryptic (in ERV) or initiator (in CAV). In the former case, mutations of this AUG impair the translational activity of the poliovirus RNA(Pelletier et al., 1988a). Although there is evidence that at least some of these elements are involved in interaction with translation factors (Svitkin et al., 1988; Meerovitch et al., 1989; Del Angel et al., 1989; Jang and Wimmer, 1990) their exact function remains obscure. To approach this problem, we constructed poliovirus genomes with extended insertions into a potentially important region of the cis element, within the pyrimidine-rich tract that lies between secondary structure domains D and E (Figure 1A). The inserts corresponded to imperfect direct or inverted repeats of neighboring sequences; they were expected either to perturb the secondary structure of the preceding domain without changing its primary structure, or to generate artificial secondary structures within the insert. Most of the engineered RNA species, while being relatively poor templates for translation, proved to retain infectivity, though the mutants appeared to be disabled to some extent. However, they readily produced, upon cultivation, a second generation of “spontaneous” mutants showing a partial restoration of both growth potential and RNA template activity. Amazingly different changes in the insert were found in the pseudorevertant genomes. Nevertheless, without exception, either a portion of the insert (sometimes with adjoining bases) had been lost or a new AUG triplet had been generated within, or close to, the insert. These and other observations allow us to propose that the oligopyrimidine tract UUUCC (considered to be an analog of the Shine-Dalgarno sequence because of its complementarity to a segment at the 3’ end of the 18s rRNA)

Cell 120

(A)

(B)

$C$

600

U-A A-U A-U domain E

domain

z:t

D

l

0 + -

G-U U-A domain

E

El:

680 -

-

Phy

M

C-G

.a

u

u

?c -u .a

I

Sl cv 0. cer.

%-!

*a -II

5,mg

DMS

G G A u

‘GGCCA 3’ I

ACA

ACA-so0 C-G U-A A-U A-U domain

-

600

C-G U-A A-U domain

E :I:

E “,I! G-U II-&

A-U G-G

U-A

G-c

U-s G-c G-c U-s

560 -

680 - C-G

C-G GG-c

‘-J@3”\ QU\ d-l’

G A

A

A

U

5’mF”

” 563

Ach

Figure

5’

1. The Structure

~,:,I

_

u

G G

A - 625

A -

3’

3’

625

of the Inserts

(A) Schematic representation of the position of the locus (nucleotides 563-569) into which the insertions were made. For the primary structure of these regions, see Pilipenko et al. (19898). (B-D) The proposed secondary structures of the relevant segments of the p23P, p39P, and p39M transcripts. The inserted nucleotides are printed in lowercase letters. Boxes A and B are placed within stippled boxes. The positions where point mutations were observed are boxed. For the conditions of RNA treatment with nucleases Sl, CV, and Phy M and from 9. cereus (B. cer.) as well as with dimethyl sulfate, see Experimental Procedures.

and the appropriately spaced downstream AUG are involved in internal ribosome binding to picornaviral UTRs. The proposal was checked and confirmed with the aid of specially designed mutant viral RNAs. Thus, this eukary-

otic system appears to share some important properties with prokaryotic ones. However, the picornavirus system is peculiar in at least two respects: the position of the oligopyrimidine tract relative to upstream cis-acting ele-

Internal 121

Table

Initiation

of Translation

1. Constructed

Name

pPV1 Mah pPvlIo3 pPVllO3lA4 pPV1103/A6 pPV11031A7 pPVl103lAE pPV1/03/AlO pPVll03123M pPV1/03/23P pPv1/03/39P pPVll03139M pPV1103/48 pPV1/03/24P

Plasmids

and Some

of Their

Properties

Sequence

Infectivity

Plaque Size

Template Activity

Box A-AUG Distance

+ qi +

8.0 4.0 6.5 3.0