Cloning of the S. solfataricus IF2/5B gene
The S. solfataricus IF2/5B gene was amplified by PCR and inserted into an expression vector (pRSETB) that added a tag of six histidines to the N-terminus of the protein and whose transcription was directed by the promoter of phage T7. The correctness of the insertion and of the nucleotide sequence of the PCR fragment was verified by direct sequencing. The construct was then expressed in Escherichia coli BL21(DE3) strain, whose genome carries the RNA polymerase T7 gene under the control of a lac UV5 promoter. Gene expression was induced by isopropyl-β-d-thiogalactopyranoside (IPTG); however, the recombinant protein was not expressed in large amounts, even after long induction times (3 h); concentration was < 5 mg l−1. Possibly, overproduction was precluded because the archaeal factor exerted some toxic effect on E. coli cells; however, the amount of expressed protein was sufficient for subsequent purification.
The aIF2/5B protein was purified from E. coli extracts by a two-step method: a differential thermal denaturation followed by affinity chromatography. Results in show that heat treatment (70°C) of transformed E. coli extracts eliminates most proteins of the host, considerably enriching the lysates with the thermostable archaeal polypeptide. Removal of precipitated proteins by centrifugation resulted in a 10-fold enrichment in the aIF2/5B His-tagged protein (about 70 kDa on SDS-PAGE 12.5%). Final purification of the thermostable factor was accomplished by affinity chromatography of the heat-treated cell extracts on a Ni+ agarose column. An aliquot of the purified recombinant aIF2/5B was used to generate polyclonal antibodies in mouse, which allowed to study the localization of the protein in S. solfataricus cell extracts.
Fig. 1 Purification of recombinant aIF2/5B. Proteins were analysed on a 12.5% SDS-PAGE stained with Coomassie brilliant blue. Lanes: M, molecular weight marker; 1, total extract from E. coli BL21 transformed with pRSETB-His6-aIF2/5B; 2, same extract as in 1 (more ...)
Construction of ‘chimeric’ bacterial–archaeal IF2/5B factors
To investigate the degree of functional similarity existing between the archaeal and the bacterial IF2-like factors, we also constructed ‘chimeric’ proteins by swapping domains between S. solfataricus
aIF2/5B andB. stearothermophilus
IF2. As shown by crystallographic studies (Roll-Mecak et al., 2000
), the IF2-like proteins consist of four principal domains, some of which can be further divided in subdomains (Gualerzi et al., 1991
; Spurio et al., 2000
). The various IF2 domains are known to carry out different tasks. Domain I (or G) contains the guanine nucleotide binding centre and interacts with the GTPase centre of the 50S subunit (Allen et al., 2005
). Domains II and III seem to be involved in the interaction with 30S and 50S ribosomes, respectively; domain II also makes contact with IF1 (Allen et al., 2005
). Domain IV interacts with the large ribosomal subunit (Marzi et al., 2003
; Allen et al., 2005
) and, in the bacterial factor, contains the fMet-tRNAi binding site (Guenneugues et al., 2000
). The hybrid proteins constructed for this study are shown in . The construct termed BaSu1 included the N-terminal domains G, II and III of bacterial IF2 and the C-terminal domain IV from S. solfataricus
aIF2/5B. The construct BaSu2 was similar to BaSu1 except that S. solfataricus
domain IV missed the two C-terminal α-helices characteristic of the archaeal proteins and not found in the bacterial one. Finally, the chimera called SuBa included domains G and II from the archaeal factor and domains III and IV from the bacterial one. This latter protein was poorly soluble and therefore could not be used in the full range of functional tests as the other two (see below).
Fig. 2 Schematic representation of the three IF2 chimeric proteins. (A) BaSu1; (B) BaSu2; (C) SuBa. On the three-dimensional structure of Methanobacterium thermoautotropicum IF2/5B (PDB ID 1G7R; Roll-Mecak et al., 2000) the different domains have been indicated (more ...)
The recombinant ‘chimeric’ proteins were expressed in E. coli and purified as described in Experimental procedures.
aIF2/5B interacts with ribosomes during translational initiation
That aIF2/5B functions as a translation initiation factor was first indicated by experiments where the protein from M. jannaschii
could partially substitute in vivo
for the function of eIF5B, its eukaryal homologue (Lee et al., 1999
). To verify these findings, and to obtain a more detailed insight about the role of the protein, we began with determining the abundance of aIF2/5B and its cellular localization in various experimental conditions.
Preliminarily, we studied the presence of the protein in batch-fractionated S. solfataricus cellular extracts: a whole-cell lysate (S-30), a post-ribosomal supernatant (S-100) and a high-salt ribosome wash (HSRW), prepared by treating S. solfataricus ribosomes with high concentrations of salt (NH4Cl 2 M). Such treatment removes any proteins that are loosely associated with the ribosomes, mainly a pool of translation factors. The various protein preparations were fractionated by SDS-PAGE and subjected to Western blotting with the specific anti-aIF2/5B antibodies.
As shown in (top), the aIF2/5B antiserum (but not the pre-immune control, data not shown) recognized in all preparations a single polypeptide, whose size (about 67 kDa) was that expected for the endogenous aIF2/5B. This result confirmed that the gene for aIF2/5B is actively translated in S. solfataricus; the protein is abundant in the cytoplasmic fraction (S-100), and also present in lesser amount in the ‘ribosome wash’.
Localization of aIF2/5B in cell lysates
More detailed information was obtained by fractionating a sample of S. solfataricus
whole-cell lysate (S-30) on sucrose density gradient. The individual fractions were assayed for the presence of aIF2/5B by Western blotting with the specific antibodies. As shown in (middle), the gradient optical density (260 nm) profile revealed the presence of two main peaks corresponding to the 30S and 50S ribosomal subunits. No peak at 70S was detected, indicating that no stable monomeric ribosomes are present in the ‘resting’ cell lysates; indeed, it is known that the 70S ribosomes of Sulfolobus
and other Crenarchaea readily dissociate into subunits unless engaged in mRNA translation (Londei et al., 1986
). Western blotting showed that most of aIF2/5B localized in the top fractions with low-molecular-weight material, while a smaller amount was visible in the fractions containing the 30S and 50S ribosomal subunits.
The same experiment was then carried out on lysates programmed for translation (Condo et al., 1999
) and incubated for a short time (2 min) at 70°C. The results, illustrated in (bottom), show that a peak of 70S monomers now appeared, while an appreciable amount of aIF2/5B moved from the low-molecular-weight fractions to localize on the ribosomes. Significantly, a fraction of the protein associated with the 70S peak, while the amount of aIF2/5B bound to the subunits, especially the 50S ones, was also increased. On the other hand, when the samples were incubated for longer periods of time (45 min), most of aIF2/5B was again found in the low-molecular-weight fractions (results not shown), indicating that after the initial phases of translation it fell off the 70S ribosomes, most of which were now engaged in the elongation cycle. These results showed that aIF2/5B interacted with the ribosomes during translational initiation and that, like its bacterial homologue, it bound to both ribosomal subunits and also to 70 ribosomes.
Next, we inquired whether purified recombinant aIF2/5B could interact with the archaeal ribosomes independently of other translational components. To this end, purified S. solfataricus ribosomes were incubated with recombinant aIF2/5B for 10 min at 70°C, and the samples were fractionated on density gradients as described above. The fractions were subjected to Western blotting with anti-His antibodies, to detect exclusively the recombinant protein. The latter was found to interact with both ribosomal subunits with a low affinity, as observed for the native factor in the ‘resting’ cell lysates (results not shown). Therefore the archaeal factor, like the bacterial one, could interact by itself with both ribosomal subunits in the absence of other factors or of tRNAi.
Conservation of the ribosome binding sites in archaeal and bacterial IF2 proteins
It is known that the principal ribosome-interacting domains of bacterial IF2 are located in the ‘cup’ of the chalice. Specifically, the N-terminal domain and domain II contact the 30S subunit while domain III (and also in part domain IV) interacts with the 50S subunit (Marzi et al., 2003
; Allen et al., 2005
; Caserta et al., 2006
). The evolutionary conservation of the ribosomal binding sites for the IF2-like proteins in bacteria and archaea was explored by determining the capacity of IF2 and aIF2/5B to interact with the heterologous ribosomes, and by assaying the ribosome-binding capacity of ‘chimeric’ factors, obtained by ‘domain swapping’ between S. solfataricus
aIF2/5B and IF2 of the thermophilic bacterium B. stearothermophilus
As shown in (top), both recombinant aIF2/5B and IF2 bound with low affinity to 30S subunits regardless of their source. Accordingly, the chimeric factors BaSu1 and BaSu2, containing the bacterial ribosomal binding domains, interacted with small subunits from either source.
Fig. 4 Interaction of native and chimeric IF2 factors with archaeal and bacterial ribosomal subunits. The ribosome-bound fraction of the various proteins was determined as described in Experimental procedures. Circle: recombinant S. solfataricus aIF2/5B; diamond: (more ...)
Binding to the 50S subunits, in contrast, appeared to be restricted within the homologous domain ( bottom). Neither aIF2/5B nor IF2 interacted with the heterologous large subunits. As regards the chimeric proteins, BaSu1 and BaSu2 bound to B. stearothermophilus large subunits with an affinity comparable to that of the intact bacterial protein, while being unable to recognize the archaeal particles. Unfortunately, the hybrid protein SuBa, endowed with the archaeal ribosome-binding domains, could not be tested in this assay because of its poor solubility, which led to its appearance in the precipitate after centrifugation even in the absence of ribosomes.
The data on the ribosome-binding capacity of IF2-like proteins were confirmed and extended by performing GTPase assays on the native and chimeric proteins, with both bacterial and archaeal ribosomes (). First, we determined that aIF2/5B has indeed a ribosome-dependent GTPase activity, which was only observed at high temperature (above 60°C) and required only the presence of the recombinant factor and purified S. solfataricus ribosomes ().
GTPase activity of the native and chimeric IF2 factors
As expected from the ribosome binding data, bacterial IF2 was unable to hydrolyse GTP in the presence of the archaeal ribosomes, and the same was true for both BaSu1 and BaSu2 hybrid factors (). Conversely, B. stearothermophilus ribosomes were unable to trigger the GTPase activity of aIF2/5B, but did trigger that of the BaSu1 and BaSu2 chimerae (). Unexpectedly, the protein SuBa, composed of the archaeal domains G and II and of bacterial domains III and IV (), had a ribosome-independent GTPase activity (not shown). The reason for this behaviour is unclear, but could conceivably be ascribed to some folding defect of the G domain in the hybrid factor.
Interaction of IF2 proteins with initiator tRNA
However, eukaryal IF5B does not bind tRNAi by itself and probably intervenes only indirectly in tRNAi/ribosome interaction (Pestova et al., 2000
). It should be noted that aIF2/5B and eIF5B are closer in sequence to each other than either is to IF2 and both lack the critical amino acids in the C-terminal domain which have been shown to be essential for fMet-tRNAi binding (Guenneugues et al., 2000
). On the score of these data, we deemed it important to analyse the interaction of S. solfataricus
aIF2/5B with initiator tRNA.
To this end, we determined whether, and to which extent, the native archaeal and bacterial factors and the hybrid proteins were able to prevent the loss of the amino acid from Met-tRNAi (archaeal) or from fMet- and Met-tRNAi (bacterial). This technique is based on the fact that the interaction between tRNAi and the IF2 factor in solution shields from hydrolysis the ester bond anchoring the methionine to the tRNA.
Unformylated Met-tRNAi of either bacterial or archaeal origin was not protected by any of the factors assayed (not shown). Instead, as illustrated in , fMet-tRNAi (bacterial) was strongly protected by both B. stearothermophilus IF2 and the chimeric protein SuBa, containing the C-terminal domains III and IV of bacterial origin. In contrast, no protection was exerted by aIF2/5B and by the chimerae BaSu1 and BaSu2, containing an archaeal domain IV in a bacterial context.
Fig. 6 Protection of fMet-tRNAi from alkaline hydrolysis in the presence of native and chimeric IF2 proteins. The extent of fMet-tRNAi protection was determined as described in Experimental procedures. Closed circle: recombinant S. solfataricus aIF2/5B; closed (more ...)
On the whole, the results demonstrated that the domain IV of the archaeal and bacterial proteins behaves differently with respect to initiator tRNA binding. Archaeal domain IV is unable to interact stably in solution with either Met- or fMet-tRNAi, both bacterial and archaeal, while bacterial domain IV binds fMet-tRNAi with high affinity but interacts poorly, if at all, with unformylated Met-tRNA. Moreover, it is evident from the data that tRNAi binding is a property of bacterial domain IV itself, independent of the rest of the protein, as already shown by Spurio et al. (2000)
Overexpression of aIF2/5B enhances translation
The next experiments were aimed at investigating the function of aIF2/5B in translation. To this end, we began with determining whether the presence of the factor in excess amounts exerted any effect on the efficiency of in vitro
protein synthesis. These experiments could not be carried out by simply adding increasing amounts of the recombinant factor to samples programmed for in vitro
translation, as the storage buffer in which recombinant aIF2/5B was kept stable inhibited unspecifically the S. solfataricus in vitro
translation system. To circumvent this problem, we obtained another clone of the aIF2/5B gene, including a tract of the 5′ region upstream of the AUG codon, which contained the putative ribosome binding site. The construct included a T7 polymerase promoter which was used to obtain an in vitro
run-off transcript that was translated in the cell-free system yielding a product of the expected size of aIF2/5B (, top). This mRNA was then added in increasing amounts to cell-free systems also programmed with a fixed amount of a reporter mRNA, encoding an housekeeping protein (Condo et al., 1999
). Two different reporter mRNAs were employed: one, encoding a ribosomal protein, was leadered and endowed with a strong SD motif ahead of the AUG initiation codon. Another, encoding an enzyme (α-fucosidase), had a mini-5′ UTR of only nine nucleotides and lacked any SD motif (Cobucci-Ponzano et al., 2006
aIF2/5B enhances in vitro translation of leadered and leaderless mRNAs
As shown in , the in vitro synthesis of both reporter proteins, especially that encoded by the leaderless mRNA, was significantly enhanced in a fashion proportional to the amount of aIF2/5B mRNA added to the system. To rule out the possibility that this effect was due to an unspecific stimulation of protein synthesis exerted by added RNA rather than by aIF2/5B itself, control experiments were carried out where the amount of one reporter mRNA was increased while keeping constant that of aIF2/5B mRNA. This procedure, however, failed to stimulate aIF2/5B translation, showing that RNA does not enhance translation unspecifically (results not shown). On the whole, the results in indicate that aIF2/5B plays an important role in both the ‘leadered’ and ‘leaderless’ pathways for translation initiation, but especially in the latter one.
aIF2/5B stimulates met-tRNAi binding to ribosomes
The results illustrated in the previous paragraph suggested that aIF2/5B acted at some crucial step common to both leadered and leaderless initiation. Previous literature data on the yeast IF2-homologous factor (eIF5B), the closest homologue of the archaeal protein, indicated that it enhanced the interaction of Met-tRNAi with the ribosome (Choi et al., 1998
), suggesting that this could be the function common to all IF2-like proteins. However, eIF5B is apparently non-essential in yeast (albeit very important for a normal growth), in agreement with the fact that the Met-tRNAi binding factor is a different protein (eIF2) in eukaryotes. The same is true for archaea, where the Met-tRNAi-binding protein is a trimeric factor homologous to eIF2 (Yatime et al., 2004
; Pedulla et al., 2005
The archaeal homologue of eIF2 (a/eIF2), similar to aIF2/5B, binds in vitro to both ribosomal subunits; however, ribosome binding of aIF2/5B and of a/eIF2 is mutually exclusive, demonstrating that the factors occupy similar positions on the ribosomes (P. Londei, unpubl. results). Given the ability of aIF2/5B to bind the ribosomes in the absence of other factors, we asked whether it was also able to stimulate to some extent tRNA/ribosome interaction. To this end, S. solfataricus 70S ribosomes were incubated with purified, [35S]-labelled, S. solfataricus Met-tRNAi and increasing amounts of aIF2/5B in the presence of GTP. The ribosomes had been previously treated with high-salt (2 M NH4Cl) to remove any bound extrinsic factors; the absence of aIF2/5B was verified by Western blotting (not shown).
tRNA/ribosome binding was monitored by electrophoresing the incubation mixtures on non-denaturing gels, which allow separation and detection of ribosomes and ribosome–tRNA complexes (Acker et al., 2006
). The results, shown in , revealed that the addition of aIF2/5B to the incubation mixture stimulated appreciably the binding of Met-tRNAi to the ribosomes, specifically to the 30S ribosomal subunits. As expected, bacterial IF2 was unable to stimulate Met-tRNAi interaction with S. solfataricus
Fig. 8 aIF2/5B stimulates the interaction of met-tRNAi with the archaeal 30S subunits. The amount of ribosome-bound met-tRNAi was determined by electrophoresis of the incubation mixtures on non-denaturing gels (see Experimental procedures). The autoradiography (more ...)
We concluded that aIF2/5B could promote Met-tRNAi/ribosome interaction even in the absence of the specific Met-tRNAi binding factor a/eIF2, and also in the absence of a direct interaction in solution between aIF2/5B and Met-tRNAi.
These results were extended and corroborated by assaying the ability of the native and chimeric IF2-like factors to stimulate the binding of both Met-tRNAi and fMet-tRNAi to bacterial ribosomes. As shown in , the interaction of fMet-tRNAi with B. stearothermophilus 30S subunits was stimulated, as expected, by bacterial IF2, but also to a comparable extent by the chimeric protein BaSu1, containing the complete domain IV of archaeal origin which, as seen above, is unable to interact in solution with both fMet- and Met-tRNAi. Interestingly enough, however, the protein BaSu2, which has the same ribosome-binding pattern as BaSu1 () but lacks a C-terminal fragment including the archaeal-specific C-terminal α-helix, was totally inactive in promoting ribosomal binding of either Met- or fMet-tRNAi. These results suggest that the C-terminal segment of archaeal domain IV does interact with tRNAi on the ribosomal surface, and that it is essential in properly adjusting the tRNA in its ribosomal binding site. In agreement with this surmise, BaSu1 also stimulated the binding of Met-tRNAi to bacterial ribosomes, although to a much lesser extent than that of fMet-tRNAi. In contrast, ribosomal binding of Met-tRNAi was not promoted by IF2, in agreement with the observation that the bacterial domain IV recognizes specifically the formyl group on the initiator amino acid.
Interaction of fMet- and met-tRNAi with bacterial ribosomes in the presence of native and chimeric IF2 factors