INTRODUCTION

Termination of protein synthesis happens through stop codon recognition by polypeptide chain release factors RF1 or RF2 (1, 2). The factors are codon-specific, RF1 recognizing UAG and UAA and RF2 recognizing UAA and UGA (3). A third factor, RF3, has been shown to amplify RF1 and RF2 termination (4-6) although interpretations differ as to whether this is a general effect (7, 8) or whether it might be specific to less stable termination complexes (9, 10). Prokaryotic and eukaryotic release factors have been identified by using a simplified assay. Charged fMet-tRNA_f^Met is bound to the ribosomal P-site via an AUG triplet. fMet-tRNA_f^Met·AUG·ribosome complexes are then exposed to release factors RF1 or RF2 in the presence of free stop triplets (11). Ribosomal peptidyltransferase catalyzes hydrolysis of the fMet pseudopeptide from P-site fMet-tRNA (12). The mechanism of termination induction by stop codon recognition is still unknown (13).

In first termination models, release factors RF1 and RF2 were believed to recognize stop codons in a tRNA-like manner (12). The idea was supported by release factors and suppressor tRNAs competing in recognition of stop codons (14). Studies by immunomicroscopy (15) questioned the idea of release factors entering the ribosomal A-site like a tRNA. New models proposed stop signal recognition by base pairing with 16 S ribosomal RNA (13, 16, 17).

Cross-linking experiments between a minimessenger and RF2 (17) on the ribosome showed that, at least for stop signal recognition by release factors, the stop codon needs to be exposed at the ribosomal A-site. The authors pointed out that there is no evidence for a conformational change when the termination complex forms. Also, functional analysis of the release reaction points to a tRNA-like action of release factors in termination. Early biochemical studies showed that tRNA can induce release activity of peptidyltransferase in acetone (18). Stop codon occupation of the A-site was not necessary. In ethanol, an excess of bulk tRNA induced fMet-ethyl ester formation. Furthermore, the study of the functional sites of release factors (19, 20) indicates that RF1 and RF2 might have a tRNA-like shape, spanning the codon recognition site and peptidyltransferase (21). Two studies have led to the proposal of a specific interaction between the terminal tRNA in the P-site and release factor RF1 (22) or RF2 (23), as if the release factor is localized near the P-site tRNA.

Recent publication of the structure of the ternary complex between EF-Tu·GTP·tRNA¹ (24) and its striking similarity to the structure of elongation factor EF-G (25, 26) showed the first example of a protein domain (EF-G domains 3, 4, and 5) mimicking an RNA (tRNA moiety of the ternary complex). A tRNA-like shape for release factors has repeatedly been proposed (21, 24, 27).

However, no one has yet reported a defined in vitro release assay that is dependent on stop codons in the ribosomal A-site. In 1983, Tate et al. (28) reported termination by release factor in the presence of a sense codon-occupied A-site, using free stop triplet. In 1987, Buckingham et al. (29) presented results with an AUGUAA mini-mRNA. No release induction by release factors could be observed, and again, free stop triplet activated the reaction. These results brought up the question of whether a first step in stop codon recognition might occur before entry of this codon into the ribosomal A-site (13).

Here we show release induction by AUGUAA and other chemically synthesized minimessengers. In all cases, stop codons were localized to the ribosomal A-site when binding the minimessage by fMet-tRNA_f^Met via an AUG codon in the P-site. Release induction of the different complexes by release factors RF1 or RF2 shows that exposure of a stop codon in the ribosomal A-site is necessary and sufficient for polypeptide chain termination. In several cases we observed partial release activation by free stop triplet, as previously reported (28, 29). We find about 10 times higher activity of release factors RF1 and RF2 when a stop codon in the minimessage is exposed at the P-site compared with the use of a free stop triplet. The presence of a sense codon in the A-site inhibits release induction by free stop triplet. Release factor codon specificity and discrimination of out-of-frame stop signals are dependent on reaction time, competition with A-site-binding aminoacyl-tRNA, and ion concentration.

MATERIALS AND METHODS

RNA oligonucleotides were synthesized on an ABI synthesizer. Protection groups were eliminated by gel filtration on Sephadex G-25 (Pharmacia Biotech Inc.). Purity and sequence were controlled by mass spectroscopy (30). Release factors have been purified from wild type strains (31) or from overexpressed strains (32). Tight couple ribosomes were prepared (33). fMet-tRNA_f^Met (Subriden) was charged and formylated as described (31). Ribosomal binding of f-[³⁵S]Met-tRNA_f^Met fixes the AUG codon of a given RNA oligonucleotide to the P-site of the ribosome (34). Ribosomes and tRNA were kept in an equimolar ratio (50 pmol/50 µl of complex mix). Free AUG triplets or AUG codon-containing minimessengers were used at 2.5 nmol in 50 µl of complex mix.

0.5 pmol of fMet-tRNA_f^Met·messenger·ribosome complex were incubated together with release factors in the presence or absence of free stop triplet (ionic conditions are stated for each series individually). Hydrolyzed formylmethionine was extracted with ethyl acetate at pH 1. Maximal release under these conditions was between 50 and 70% of the total f-[³⁵S]Met present in the reaction mix. Assays were repeated two to five times with independently prepared complexes, and the results were averaged. Amounts of free fMet at zero time of each reaction series (3-8%) were subtracted.

RESULTS

Release Activity with Minimessengers Is Dependent on Stop Signals Located at the Ribosomal A-site (Table I)

Different RNA oligonucleotides were fixed by charged f-[³⁵S]Met-tRNA_f^Met to 70 S ribosomes. The complexes were incubated at different concentrations of release factors RF1 or RF2 and in the absence or presence of free stop triplet. In complexes where the A-site was not occupied by a stop codon, fMet release was dependent on free stop triplet addition and high release factor concentration (lines 1 and 2). Oligonucleotides presenting a stop codon to the A-site allowed termination in the presence of RF1 or RF2 at relatively low concentration and independent of free stop triplet (lines 7-9). Addition of 5-UUC to a given messenger, presumably in the ribosomal E-site, did not change the release efficiency to a significant extent (lines 8 and 9). This was not changed in the presence of excess deacyl-tRNA^Phe (data not shown). Free stop triplet release induction is dependent on the accessibility of the A-site and is efficiently inhibited when a sense codon occupies the A-site (lines 3-6) although an increase of release activity could be observed when a free stop triplet was added (lines 4, 6, 8, and 9), indicating a certain degree of breathing of the system in the present conditions. When the message was elongated 3 from the A-site (line 7), free stop triplet slightly inhibited termination. We conclude that the system is not fully independent of effects due to free stop triplet in high excess. Release factors might recognize triplets in competition with the minimessenger before or after having been bound to the ribosomal complex. In addition, mRNA 3 from the A-site might sterically inhibit access of free stop triplet to the A-site.

Table I. Stop signal recognition on minimessengers E, P, A stand for ribosomal exit, peptidyl, and aminoacyl sites. Oligonucleotides were fixed via fMet-tRNA to the P-site. Where indicated, free stop triplet (UGA, UAA) was added at 0.1 mM. Release reactions were incubated for 15 min at 30 °C in 30 mM MgCl2, 75 mM NH4Cl, and 50 mM Tris, pH 7.5 (34). E, P, A stand for ribosomal exit, peptidyl, and aminoacyl sites. Oligonucleotides were fixed via fMet-tRNA to the P-site. Where indicated, free stop triplet (UGA, UAA) was added at 0.1 mM. Release reactions were incubated for 15 min at 30 °C in 30 mM MgCl2, 75 mM NH4Cl, and 50 mM Tris, pH 7.5 (34).     E  P  A RF2 (5 picounits) RF2 (50 picounits) RF2 (5 picounits) UGA RF2 (50 picounits) UGA RF1 (15 picounits)    RF1 (15 picounits) UAA fmol 1)    AUG 10 15 40 230 10 135 2) UUCAUG 26 22 45 203 3)    AUGUUC 2 8 0 0 10 20 4)    AUGUUCUAA 16 15 33 50 10 20 5)    AUGUUCUGAGCCCC 0 0 4 8 6) UUCAUGUUUUUCUAA 0 5 0 20 7) UUCAUGUAAGCCCC 224 214 180 169 8) UUCAUGUAA 200 215 250 280 220 9)    AUGUAA 179 193 217 280

Specificity of in Vitro Peptide Release Using Minimessengers

To reproduce release factor codon specificity and stop signal frame specificity, minimessages with release factor specific stop codons or bearing incremental insertions between the AUG (Met) codon and the stop codon were synthesized.

In classical in vitro termination conditions established for the use of free stop triplet, our assay was only marginally specific against a +1 frame stop codon (line 3), and no specificity of RF2 against UAG recognition could be observed, even at a relatively low concentration of the factor (line 8). +2 or 2 out-of-frame stops were not recognized (lines 2 and 4). A stop signal that was localized an entire codon downstream from the A-site (Table II, line 1, and Table I, lines 4 and 5) was also not recognized.

Table II. Fidelity of stop signal recognition E, P, A stands for ribosomal exit, peptidyl, and aminoacyl sites. Oligonucleotides were fixed via fMet-tRNA to the P-site. Release specificity was performed under classical conditions (34): 15 min of incubation at 30 °C (30 mM MgCl2, 75 mM NH4Cl, 50 mM Tris, pH 7.5). E, P, A stands for ribosomal exit, peptidyl, and aminoacyl sites. Oligonucleotides were fixed via fMet-tRNA to the P-site. Release specificity was performed under classical conditions (34): 15 min of incubation at 30 °C (30 mM MgCl2, 75 mM NH4Cl, 50 mM Tris, pH 7.5).     E  P  A RF2 (5 picounits) RF2 (50 picounits) RF1 (15 picounits) fmol 1) UUCAUGUUUUAA 5 15 2) UUCAUGUUUAA 15 10 15 3) UUCAUGUUAA 85 135 80 4) UUCAUGA 5 0 10 5) UUCAUGUAA 200 215 220 6)    AUGUAA 179 193 7) UUCAUGUGA 80 140 25 8) UUCAUGUAG 90 195 95

Mg²⁺ has been shown to play a role in coordinating secondary structures of nucleic acids (e.g. Refs. 35 and 36) but also in coupling of ribosomal subunits and binding of nucleic acids (mRNA, tRNA) to ribosomes (33). The optimal Mg²⁺ concentration for in vitro termination (30 mM) has been established empirically (37) and is somewhat high compared with common Mg²⁺ concentrations (5-10 mM) used in translation in vitro assays (33). 30 mM is optimal for release reactions using free stop triplet, where the reaction is the third order (fMet-tRNA_f^Met·AUG·ribosome intermediate, stop triplet, and release factor have to form a complex allowing fMet-tRNA_f^Met hydrolysis). In our assay, a stop codon is already part of the fMet-tRNA_f^Met·messenger·ribosome intermediate. It was therefore reasonable to ask whether a lower Mg²⁺ concentration could be sufficient for a reaction of the second order, thereby allowing efficient discrimination of non-cognate stop codons. Release factor concentration was kept below saturation. By diminishing reaction time, we were able to show differences in activity of release factors for different stop codons. Fig. 1 shows that indeed stop codon specificity of RF1 and RF2 is directly dependent on incubation time and Mg²⁺ concentration. At 8 mM [Mg²⁺], a clear difference in reaction velocity between specific and nonspecific reactions appeared. Codon-specific reactions were mostly completed within the first 30 s. Nevertheless, RF2 showed less codon specificity than RF1, which showed codon specificity over the entire range of Mg²⁺ concentrations tested.

We then tested the effect of incubation time on frame specificity in stop codon recognition (Fig. 2). UUC AUG UAA GCC CC was used as a positive control for in-frame stop codon recognition. The +1 stop (UUC AUG UUAA) message, which had turned out to have poor discrimination (Table II), was tested together with +2 (UUC AUG UUUAA) and 2 (UUC AUGA) oligonucleotides. Recognition of out-of-frame stops decreased by half when the reaction time was reduced from 4 min to 30 s.

We wanted to see whether a tRNA recognizing the cognate A-site codon would be able to compete with a release factor. In that case, the presence of a cognate tRNA in the A-site should diminish out-of-frame stop codon recognition by a release factor. It has been shown previously that it is possible to occupy a ribosomal A-site with a cognate aminoacyl-tRNA in the absence of EF-Tu when this site is programmed by a message (38). In Fig. 3 we show that, under our conditions, release factor recognition of a +1 frame stop signal on an UUC AUG UUA A is diminished 2-fold when the ribosomal A-site has previously been occupied by a UUA that decodes Leu-tRNA^Leu. This 2-fold inhibition was constant at 30 s and 5 min of incubation time. Use of uncharged tRNA^Leu did not result in enhanced out-of-frame stop signal discrimination.

Activity of RF1 and RF2 with Stop Codon-containing Messengers Versus Free Triplets

Our results (Table I) suggested a higher affinity of release factors for messages containing an A-site stop codon versus complexes dependent on free stop triplet addition. We therefore studied the kinetics of termination reactions dependent on release factor concentration, comparing the two systems (Fig. 4). Conditions of kinetics were the same as for Caskey et al. (34). As previously shown, at a low concentration of release factors, fMet release is linear during the first 5 min, and the velocity is dependent on release factor concentration when using free stop triplet. 30 mM Mg²⁺ is optimal for efficient fMet release with free stop triplet.

At the same release factor concentration, this reaction was completed after 1.5 min when using a stop signal on a minimessage. We therefore studied the kinetics of our assay at intervals of <1 min (Fig. 5). More than half of the reaction was completed within 10 s when using RF1, and within the limits of our experiments, we could not observe a difference in velocity at two different release factor concentrations. RF2 seems to react at least three times slower in comparable conditions. In both cases, a plateau is reached after <2 min. We conclude that this plateau corresponds to the actual amount of active release factor. In this case, 5 picounits of release factor activity (34) corresponds to 50 fmol of active protein. Continuation of the RF2 reaction after 2 min at a significantly lower level might be due to slow release of RF2 from previous termination complexes.

DISCUSSION

Recognition of stop codons depends on participation of termination factors, which was initially suggested by Ganoza (39). Capecchi (40) used the mRNA of bacteriophage R17 to direct in vitro protein synthesis programmed by a mutant with a stop codon at the sixth amino acid position of the coat protein. Release of the hexapeptide was shown to be dependent on the presence of protein factors (2). A defined termination assay established by Caskey et al. (11) measured release of fMet from an fMet-tRNA_f^Met·AUG·ribosome intermediate. The work described here presents an important improvement of the defined in vitro termination assay. We show that a minimessenger bound to the ribosome through interaction with an fMet-tRNA_f^Met is able to program termination by release factor recognition. A stop codon located at the A-site is necessary and sufficient for release factor-induced termination. UAG (with RF1) and UGA (with RF2) were less efficiently recognized than UAA with both factors. This is consistent with observations of Caskey et al. (34) in their system. In our initial experiments, RF2 recognized UAG as well as UGA. Reducing reaction time and Mg²⁺ concentration resulted in specific recognition of UAG and UAA by RF1 and UAA and UGA by RF2, although the fidelity of RF1 was higher than RF2 under tested conditions.

In a previous publication (29), termination induction by release factors was reported when a stop codon was located one or two stop codons 3 from the ribosomal A-site. We did not observe termination induction by release factors with stop signals located outside of the A-site. 2 and +2 out-of-frame stop signals were efficiently discriminated. However, a +1 out-of-frame stop codon was fairly active. Again, specific discrimination of this recognition was dependent on incubation time. Furthermore, the presence of a cognate aminoacyl-tRNA in the A-site competed with +1 out-of-frame stop signal recognition.

Our results and conclusions are supported by kinetic comparison between free stop triplet and entire minimessenger in in vitro termination. The activity of release factors with the fMet-tRNA_f^Met·messenger·ribosome complex is strikingly enhanced, resulting in an increase in velocity of 2 orders of magnitude. RF2-related termination appears to occur at least three times slower than termination by RF1. We do not know if this is due to lower affinity or a lower k_cat of the factor. A lower activity rate of RF2 might be a reason for the observation that there are about five times more molecules of RF2 than RF1 present in an Escherichia coli cell (32). Additionally, a lower velocity in RF2-related termination might compensate for lower specificity in stop codon recognition.

The new assay could provide insights in other aspects of translational termination. RRF, ribosomal recycling factor, has been shown to release mRNA from ribosomes after peptide release has occurred. Using an entire minimessage instead of separate triplets, our new assay might be able to determine factor requirements for ribosomal recycling after polypeptide chain termination.