Evolution of translational initiation: new insights from the archaea

FEMS Microbiology Reviews 29 (2005) 185–200 www.fems-microbiology.org

Evolution of translational initiation: new insights from the archaea Paola Londei

q

*

Dipartimento di Biochimica Medica e Biologia Medica (DIBIM), Universita´ di Bari – Piazzale Giulio Cesare, 70124 Bari, Italy Received 31 July 2004; received in revised form 22 September 2004; accepted 28 October 2004 First published online 20 November 2004

Abstract Recent in silico and experimental data have shed new light on the mechanism and components of translational initiation in archaea. The available data about the structure of archaeal mRNAs, mRNA/ribosome interaction and archaeal translation initiation factors are reviewed and analyzed in the conceptual framework of the evolution of translational initiation. A model of the initiation step of translation in the Last Universal Common Ancestor of extant cells is presented and discussed. Ó 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. Keywords: Archaea; Leaderless mRNA; Translation initiation; Translation initiation factors; Last Universal Common Ancestor

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . mRNAs in the archaea: ‘‘Leadered’’ and ‘‘Leaderless’’ messages. . . . . . . . . . . . . . . . . . . . . . . . . . . . Translation initiation factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Translation initiation factors in the primary domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. The IF2-like factors: universally conserved proteins with divergent functions?. . . . . . . . . . . . . . 3.3. a/eIF2: the met-tRNAi binding factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. aIF1A and aSUI1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. aIF5A and aIF6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding remarks: towards a reconstruction of the ancestral translation initiation mechanism . . . . . 4.1. Structure of the mRNAs and initiation mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Initiation factors in ancestral translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Models of ancestral translational initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.

q

Edited by Mike Dyall-Smith. Address for correspondence: Dpt. Biotecnologie Cellulari ed Ematologia, Universita´di Roma La Sapienza – Policlinico Umberto I, Viale Regina Elena 324 – 00161 Roma, Italy. Tel.: +39 06 4940463; fax: +39 06 4462891. E-mail address: [email protected]. *

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

185 186 188 188 189 191 192 194 195 195 195 197 197 197

1. Introduction Translation, the last phase of the gene expression pathway, is highly conserved in evolution as regards its basic mechanism and components. However, initiation,

0168-6445/$22.00 Ó 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.femsre.2004.10.002

186

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

the first (and rate-limiting) step of translation, is known to have incurred an extensive divergence in the bacteria and the eukaryotes. Bacterial initiation entails direct recognition of the translation initiation region on the mRNA by the small ribosomal subunit (30S). This interaction is guided by the pairing of complementary RNA sequences, the purine-rich Shine–Dalgarno (SD) motif on the mRNA and the pyrimidine-rich anti-Shine– Dalgarno (anti-SD) sequence on the 16S rRNA [1,2]. Ribosomal recognition of the translation start site is modulated by three accessory factors, the monomeric proteins known as IF1, IF2 and IF3 [3,4]. By contrast, translation initiation in eukaryotes follows a mechanism of considerable intricacy and complexity. mRNA/ribosome recognition does not rely on a direct mRNA/rRNA interaction and requires the help of many different factors. The current consensus model holds that the 40S ribosomal subunit, in a complex with numerous accessory proteins, binds in the vicinity of the capped 5 0 end of the mRNA and then slides in a 3 0 direction until the initiation codon – normally the first AUG in any frame – is encountered [2,5,6]. Eukaryotic ribosomes do not seem to be able to land directly on the mRNA unless the latter is equipped with special ‘‘internal ribosome entry sequences’’ (IRES) [7]. The factors assisting eukaryotic initiation are about 10, many of which complex multimeric proteins [8–10]. Several of them are required for the unwinding of secondary structures in the mRNA, a necessary preliminary step for ribosomal binding to the message. The picture is further complicated by the peculiar structure of eukaryotic mRNAs, which appear to adopt a circular shape owing to the interaction between the poly (A) binding proteins at the 3 0 end with the complex of factors recognizing the cap at the 5 0 end [11,12]. The canonical textbook explanation for such divergent translation initiation mechanisms is that they reflect the different organization of the gene expression apparatus in prokaryotes and eukaryotes. In the bacteria, transcription and translation are contemporaneous and mRNAs are usually polycistronic. Therefore, direct mRNA/ribosome recognition via the SD/anti-SD interaction is necessary for speeding up the decoding of the more distal open reading frames (ORFs) in a polycistronic message. On the other hand, a large complement of initiation factors is superfluous since simple unicellular organisms can dispense with an elaborate translational regulation as that enjoyed by the more sophisticated and differentiated eukaryotic cells. This neat scenario was challenged when the archaea came into the picture. The expectation was that the archaea, being prokaryotes, had a translation initiation apparatus similar to the bacterial one. There was no structural or physiological reason to expect anything different. Indeed, early data coming from translational studies in the methanogens and the halophiles indicated

that the archaea did possess polycistronic mRNAs, often – but not always – endowed with clearly identifiable SD motifs ahead of the initiation codons [13–15 and references therein]. The discovery came therefore as a big surprise, following the sequencing of the first complete archaeal genome, that the archaea are endowed with over 10 genes encoding proteins homologous to eukaryotic initiation factors [16,17]. To date, the list of putative translation initiation factors found in all completely sequenced archaeal genomes (21 at the time of this writing) includes homologues of most of the known eukaryotic factors, excepting only those involved in cap recognition (Table 1). Such data elicited, of course, many questions and comments. Were these proteins actually involved in translational initiation and, if so, why such a wealth of translation initiation factors in a prokaryotic domain? Most comments favoured the idea that initiation in archaea had a somewhat hybrid character, resembling bacteria in the structure of mRNAs (polycistronic organization and SD motifs) and eukarya in the complement of initiation factors [16,17]. This, in a sense, was a matter of fact: but what does it mean in terms of the actual mechanism of initiation in archaea? Addressing these questions required experimental data; some of which, although slowly, have been forthcoming in the last few years. In the following paragraphs, I shall try to review what is currently known about initiation in archaea and what the emerging picture can teach us about the broader topic of the evolution of the mechanism and apparatus for translational initiation.

2. mRNAs in the archaea: ‘‘Leadered’’ and ‘‘Leaderless’’ messages The complete sequencing of numerous archaeal genomes in the last few years has yielded a wealth of information on the nature and the organization of genetic information in the third kingdom of life. Like bacterial ones, archaeal genomes are at most a few megabases in size, and contain little non-coding DNA. Several introns have been detected in RNA-coding genes [18], while to date only one example of an interrupted protein-coding gene has been found [19]. Some proteins, however, contain inteins [20]. As in bacteria, genes are tightly clustered and may form polycistronic transcription units, some of which resemble closely in structure the corresponding bacterial unit. Gene clusters tend to be larger and more bacterialike in methanogens and halophiles, while the same genes may be unlinked in the thermoacidophiles [14 and references therein]. The presence of polycistronic transcripts translated into several polypeptides has been experimentally veri-

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

187

Table 1 Initiation factors in the primary domains Bacteria

Eukarya

Archaea

Universally conserved factors IF1 Stimulates IF2 function IF2 Binds f-met-tRNAi YCiH (only some phyla) Function unknown EFP Stimulates first peptide bond

eIF1A Assists scanning eIF5B Promotes subunit association eIF1/SUI1 Fidelity factor eIF5A Stimulates first peptide bond

aIF1A Function unknown aIF2 Promotes met-tRNAi/ribosome interaction aSUI1 Function unknown aIF5A Function unknown

–

–

eIF4F (trimer) Binds cap and unwinds mRNA eIF3 (octamer) Interacts with eIF4F and ribosome eIF2 (trimer) Binds met-tRNAi eIF2B (pentamer) Guanine nucleotide exchange factor eIF5 GTPase of eIF2 eIF6 Inihibits subunit association

–

Domain-specific factors IF3 Fidelity factor – – – – – –

fied in a number of cases [13 and references therein]. Sometimes, mechanisms of translational regulation similar to the bacterial ones have been detected, as for instance autogenous regulation by ribosomal proteins on their encoding mRNAs [21]. As regards the mechanism of initiation, the presence of SD sequences in archaeal mRNAs (and that of the corresponding anti-SD sequence in the 16S ribosomal RNA) has been known for many years [15 and references therein], although the experimental verification of their function has been obtained only recently [22]. However, already two decades ago some unusual mRNAs had been found in halophilic archaea, which lacked entirely, or almost so, a 5 0 untranslated region [23]. In these ‘‘leaderless’’ mRNAs, the initiation codon is located right at the 5 0 end or just a few nucleotides downstream; SD sequences are obviously missing. To explain how the leaderless mRNAs of halophiles could be translated it was originally proposed that they were endowed with ‘‘cryptic’’ SD sequences, located a few bases downstream of the initiation codon; these sequences were recognized by the small ribosomal subunit with the customary RNA/RNA interaction mechanism [23]. A similar mechanism was proposed to account for the translation of the few leaderless mRNAs existing in bacteria and bacteriophages, such as the transcript of the k repressor protein gene. In this latter case, it was suggested that a special sequence downstream of the initiation codon, termed the ‘‘downstream box’’, mediated direct mRNA/ribosome interaction [24]. However, the

– a/eIF2 (trimer) Binds met-tRNAi only a, b, d subunits function unknown – aIF6 Function unknown

‘‘downstream box’’ sequence would not base pair with the anti-SD motif, but with another tract of the 16S ribosomal RNA spanning nucleotides 1469–1483 of 16S rRNA (E. coli numbering) [24,25]. The ‘‘downstream box’’ concept could be extended to the translation of any prokaryotic leaderless mRNA lacking ‘‘cryptic’’ SD motifs. In summary, until recently mRNA/ribosome interaction in archaea was believed to be based on an mRNA/ rRNA interaction as in bacteria, the only oddity being the presence of a sizeable amount of leaderless mRNAs whose translation, however, also relied on a mechanism essentially similar to the canonical SD/anti-SD pairing. This view has been first challenged by in silico investigations of the putative structure of the transcripts in the thermophilic archaea Sulfolobus solfataricus and Pyrobaculum aerophilum [26–28]. From the position of the presumptive promoters, the authors deduced that a large proportion of mRNAs were leaderless in these species. However, plausible ‘‘downstream boxes’’or ‘‘cryptic’’ SD sequences were seldom identifiable. In the case of polycistronic mRNAs, the first cistron was very often leaderless, while the ‘‘internal’’ ones usually did possess well-defined SD motifs ahead of the initiation codons. On the basis of these data, it was proposed that two distinct mechanisms for initiating translation existed in archaea: one of bacterial type, based on SD motifs and operating on the internal cistrons of polycistronic mRNAs, and another of unknown nature employed for initiating translation of leaderless cistrons [27,29].

188

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

The first set of experimental data supporting the notion of two distinct mRNA types and translation mechanism in archaea was obtained using S. solfataricus as the model organism. With the aid of a cell-free translation system, it was demonstrated that SD motifs, when present, are essential for translational initiation [22]. In fact, the disruption of SD motifs by site-directed mutagenesis was found to completely inhibit in vitro translation, a situation more extreme than that found in bacteria where disruption of SD sequences usually reduces but not abolishes protein production. However, similar studies recently conducted in vivo on Halobacterium salinarium seem to indicate that in halophiles the disruption of SD motifs has less extreme consequences than in Sulfolobus, leading to a reduction of translational efficiency but not to a total inhibition of protein synthesis [30]. The most remarkable finding, however, was that the in vitro translation of the mRNAs whose SD motifs had been disrupted could be rescued by deleting entirely the 5 0 untranslated region (UTR), namely by rendering the mRNA leaderless [22]. This result argued strongly against the presence of a cryptic SD motif or a ‘‘downstream box’’, as it appeared very unlikely that a normal leadered mRNA should possess a masked ribosome binding sequence just in case it would somehow lose its 5 0 UTR. It suggested, instead, that when the SDbased mechanism was no longer operative, even an artificial leaderless mRNA could be translated using the alternative route, existing to operate on natural leaderless messages. The first hints as to the features of the alternative initiation mechanism came from studies performed on bacterial leaderless mRNAs. The ‘‘downstream box’’ notion was beginning to be challenged in the bacterial domain by studies indicating that no mRNA/16S rRNA interactions were involved in the decoding of leaderless mRNAs in E. coli [31,32]. A study performed in vivo using mRNAs endowed with 5 0 terminal as well as internal initiation codons [33] demonstrated that the presence of an excess of initiation factor 2 (IF2) favoured ‘‘leaderless’’ translation over leadered, SD-based, one. As IF2 in bacteria is the factor that promotes f-met-tRNA entry in the ribosomal P site, these results suggested that 30S ribosomal subunits pre-loaded with initiator tRNA were a prerequisite for leaderless translation. This meant that the recognition of a 5 0 terminal AUG codon obligatorily required the interaction with the anticodon of the ribosome-bound f-met-tRNA, without any previous mRNA/ribosome interaction. Such a mechanism is reminiscent of eukaryotic initiation. However, perhaps the most interesting aspect of leaderless mRNAs is that, regardless of their source, they appear to be universally translatable by bacterial, eukaryal and archaeal ribosomes [33]. Therefore, ‘‘leaderless’’ initiation can be considered neither ‘‘bacterial’’ nor ‘‘eukaryal’’, but might be

a very old mechanism still operative in the three domains of life (see discussion). Further information and support for the existence of two different mechanisms for translational initiation in archaea came from studies addressing the mechanism of ribosome/mRNA interaction in S. solfataricus [34]. Briefly, it was found that, in the presence of SD motifs, Sulfolobus small ribosomal subunits interacted directly and strongly in vitro with leadered mRNAs in the absence of any other translational component. Unlike their bacterial counterparts, the binary 30S/mRNA complexes thus formed were very stable. Leaderless mRNAs, by contrast, were unable to form binary complexes with 30S subunits. A 30S subunit/leaderless mRNA interaction could be detected only in the presence of met-tRNAi [34] supporting the idea that codon–anticodon pairing was required for initiation site recognition, as previously observed for leaderless mRNA translation in E. coli [33]. Of course, the presence of two different mechanisms for ribosome/mRNA recognition in the archaea raises a host of questions. Why two mechanisms? Are there any physiological reasons why ‘‘leaderless’’ translation is seemingly so widespread in archaea, especially in the hyperthermophiles, despite being rather exceptional in bacteria and eukarya? Or is leaderless translation a molecular fossil, carried over from the Last Universal Common Ancestor (LUCA) and retained preferentially in the archaea because of their slow evolutionary rate? Far more is necessary to know in order to begin to answer some of these questions. A very important task is to understand the function of the initiation factors in the two types of initiation.

3. Translation initiation factors 3.1. Translation initiation factors in the primary domains As it is well known from a wealth of genetic and biochemical studies, the bacteria and eukarya have widely different sets of translation initiation factors. Bacteria, which are always very parsimonious in their endowment of cellular components, possess only three translation initiation factors (IF). The pivotal function of adapting f-met-tRNA in the ribosomal P site is performed by IF2, a 90 kDa guanine-nucleotide-binding (G) protein. IF1, a small polypeptide of 71 amino acids, stimulates the activity of IF2 and IF3 in the formation of the 30S ternary complex and modulates the interaction of IF2 with the ribosome. IF3 acts a ‘‘fidelity factor’’, discouraging initiation at non-optimal codons such as AUU [4]. Eukaryotes by contrast, are endowed with a much more elaborate set of IFs [8,10]. Several factors, absent in bacteria, are required to unfold the mRNA (which

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

comes from the nucleus tightly packed with certain proteins) so as to allow its interaction with the ribosome. The cap-binding factor (termed eIF4F) is a complex of three polypeptides [8]. The actual cap-binding function is performed by eIF4E, a relatively small protein of 25 kDa. The other two members of the eIF4F complex are eIF4A, an RNA helicase, and eIF4G, a large modular protein that interacts with several other components, including the mRNA and the multimeric factor eIF3 [35]. No SD–anti SD interaction exists in eukarya, and ribosome/mRNA recognition is assisted by a complex set of factors. To ‘‘scan’’ the mRNA in search of the initiation codon the eukaryotic 40S subunits must be loaded with initiator tRNA (met-tRNAi). Such loading is accomplished by the trimeric factor eIF2, a G protein, which interacts with met-tRNAi only in the active, GTP-bound form. Following the successful interaction of met-tRNAi with the 40S subunit, GTP is hydrolyzed to GDP and eIF2 falls off the ribosome [36]. Despite the similar name and function, eukaryal IF2 has no homology with bacterial IF2; also, unlike the latter, it requires a GDP/GTP exchange factor to be reactivated after GTP hydrolysis. The exchange factor is a pentameric protein called eIF2B. Moreover, unlike IF2, eIF2 has no intrinsic GTPase activity: the hydrolysis of the eIF2-bound GTP is triggered by an accessory GTPase, called eIF5 [36]. Other factors required for a successful ‘‘scanning’’ are eIF1, eIF1A, and eIF3; the latter interacts with eIF4F, the cap-binding complex [10]. Once the initiator AUG codon has been identified a further factor, eIF5B, is required to stimulate the joining of the 60S ribosomal subunit to the 40S initiation complex, thereby forming the complete 80S ribosome [37]. Like eIF2, eIF5B is a G protein: upon 60S subunit joining, GTP must be hydrolyzed to GDP to dissociate the factor from the newly assembled 80S ribosome. However, the formation of the 80S ribosomes per se does not require GTP hydrolysis [38]. The most interesting feature of eIF5B is that it has homology with bacterial IF2, as will be discussed later in better detail. Still another factor, eIF5A, is required immediately after the assembly of the 80S ribosome to stimulate the formation of the first peptide bond [39]. To complete this brief survey, the recently characterized factor eIF6 must also be mentioned. This protein seems to be a very important regulator of eukaryotic translation: it interacts with the 60S ribosomal subunit preventing its joining with the 40S particle. Therefore, 80S ribosomes cannot form efficiently unless eIF6 is released from the 60S subunit. Such release requires the phosphorylation of the factor, which is controlled by means of several regulatory kinase cascades [40]. How do archaea compare with the other two domains in their complement of translation initiation fac-

189

tors? As noted previously, the recent sequencing of many archaeal genomes has overturned the expectations of most investigators, i.e. that the archaea, being prokaryotes, should possess a restricted set of initiation factors similar to the bacterial one. To the contrary, archaeal genomes appear to have a decidedly ‘‘eukaryotic’’ complement of (putative) translation initiation factors. As shown in Table 1, all archaeal genomes sequenced so far include genes encoding for homologues of most of the eukaryotic initiation factors. The only factors apparently restricted to the eukaryotic domain are the members of the eIF4F cap-binding complex, eIF3 and the GTPase eIF5. Of course, homology in primary sequence does not necessarily mean homology in function. Once the candidate factors are identified by in silico studies, that they indeed have a role in translation must be demonstrated experimentally. Despite their scarcity, the available experimental studies of archaeal putative initiation factors have begun to delineate a very interesting scenario that will be presented in the next paragraphs. 3.2. The IF2-like factors: universally conserved proteins with divergent functions? An essential function performed by translation initiation factors is initiator tRNA selection and adaptation in the ribosomal P site. As briefly described above, different and unrelated proteins, IF2 and eIF2, perform this task in bacteria and eukarya respectively. Thus, until recently it was generally believed that the initiator tRNA binding factor had evolved independently in the bacterial and eukaryal domains. This belief was, however, repeatedly challenged in the last few years. First, genome sequencing in archaea revealed the presence of genes encoding proteins homologous to both bacterial IF2 and the three subunits of eukaryal IF2. Almost contemporaneously, a homologue of bacterial IF2 was recognized in the completely sequenced yeast genome. The first experimental study addressing the function of the yeast IF2 homologue revealed that was indeed a translation factor, unable to bind directly met-tRNAi but apparently capable of stimulating its binding to the ribosomes [41]. However, the factor was not essential, since yeast mutants lacking it were viable albeit very slow-growing. Later, the function of the rodent IF2-like protein was unravelled in a series of elegant experiments that made use of in vitro systems with purified proteins and ribosomes. These studies established that the rodent IF2-like factor is required primarily to promote the joining of the 60S subunits to the 48S initiation complex, thereby facilitating the formation of 80S ribosomes [37]. The factor was then dubbed eIF5B, a designation which, although sensible in the context of eukaryal initiation, somewhat obscures the very special evolutionary status of this protein. In

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

fact, being represented in bacteria, archaea and eukarya, eIF5B/IF2 is one of the few universal translation initiation factors [42]. This means that the IF2-like proteins were already present in the LUCA; their conservation in all modern members of the three primary domains strongly suggests that they still share an important common function. Yet, as explained above, the available experimental information on bacterial IF2 and eukaryal eIF5B indicates that these factors have incurred a marked functional divergence. To understand something more about the hypothetical shared function of the IF2-like factors, and to get some insight into the evolution of the tRNAi binding factors, it becomes very important to unravel the function of the archaeal homologues of IF2/eIF5B and of eIF2. Why should archaea possess a eukaryotic set of these proteins? Have the archaeal proteins the same functions as those of their eukaryal counterparts? To date, the only published experimental data about the function of the archaeal IF2-like proteins show that M. jannaschii IF2 homologue can partially rescue yeast mutants lacking eIF5B [43]. These results suggest at least some degree of functional homology between the eukaryal and archaeal factors. Yet, as will be explained below, more-detailed studies suggest some caution before concluding that the archaeal and eukaryal IF2-like proteins have entirely overlapping functions. For this reason I shall henceforth refer to the archaeal factor as aIF2 rather than as aIF5B as some investigators do. In contrast with the paucity of functional data, there is much detailed information about the structure of aIF2. Crystallographic studies on the M. thermoautotrophicum aIF2 have revealed that it is characteristically shaped as a chalice [44]. The globular ‘‘cup’’ of the chalice (N-terminal region) includes three domains, the first of which is the guanine-nucleotide-binding domain. The ‘‘stem’’ of the chalice is a long alpha-helix, while the globular ‘‘base’’ (domain IV) corresponds to the C-terminal domain known to bind f-met-tRNA in bacterial IF2 [45]. It must be pointed out that the aIF2 proteins are generally smaller than their eukaryal and bacterial homologues, since they lack the long and poorly conserved N-terminal tracts of uncertain function present in both IF2 and eIF5B. Our laboratory is currently studying the function in archaeal translation of S. solfataricus aIF2. The factor has been obtained in recombinant form by standard cloning and over-expression procedures, and it has been tested in a number of in vitro assays to unravel its role in translational initiation. Moreover, natural aIF2 has been produced by translation of its mRNA in a cell-free system for protein synthesis prepared from S. solfataricus cell lysates [22]. The assays performed with the recombinant protein [77] have revealed that aIF2 is a ribosome-dependent, thermophilic GTPase that interacts preferentially with

50S subunits and 70S ribosomes. Like its eukaryal homologue, aIF2 does not bind met-tRNAi or any other tRNA, including bacterial f-met-tRNA. Nevertheless, when the recombinant protein is added to a purified system containing only ribosomes and met-tRNAi it stimulates strongly the binding of the latter to the ribosomes, a function reminiscent of that of its bacterial counterpart (Fig. 1). Moreover, when aIF2 is overexpressed in vitro by translating its mRNA in a cell-free system for protein synthesis, it displays a most remarkable stimulatory effect on the translation of other mRNAs, both leadered and leaderless, suggesting that aIF2 function is common to both initiation mechanisms (Fig. 2; Benelli, D., Maone, E., Londei, P., unpublished results). Stimulation of met-tRNAi/ribosome interaction is a good candidate for such function. Stabilization of met-tRNAi binding to the ribosomes was also the function originally proposed for the yeast IF2 homologue, following the observation that slowgrowing yeast strains lacking eIF5B could be partially rescued by over-expressing initiator tRNA [41]. These data have been integrated with those obtained in higher eukarya [37] in the proposal that eukaryal eIF5B stabilizes met-tRNAi binding to the ribosome while stimulating the formation of the 80S ribosome [46]. However, despite all our efforts, so far we have been unable to produce any evidence that archaeal IF2 has a role in facilitating ribosomal subunit joining. This, of course, may only be due to technical problems, but it must be observed that also in the bacterial domain 80000 70000 60000 counts per min

190

50000 40000 30000 20000 10000 0 0

10

20

30

40

50

60

70

80

pmol aIF2 Fig. 1. aIF2 stimulates met-tRNAi binding to ribosomes. About 30 pmol of purified S. solfataricus 70S ribosomes were incubated for 10 min at 60 °C with an excess of tRNA charged with 35S methionine and with increasing amounts (10–100 pmol) of purified recombinant S. solfataricus aIF2. Ribosome-bound met-tRNA was separated from unbound one by electrophoresis on non-denaturing gels. The amount of ribosome-bound tRNA was determined by autoradiography of the gels using an Istant Imager apparatus. Full squares: aIF2; open squares: control (BSA).

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

Fig. 2. aIF2 stimulates in vitro translation. S. solfataricus cell-free protein synthesizing systems were prepared as described by [22]. Control samples (C) only contained endogenous mRNAs; aIF2 samples were supplemented with about 3 lg of S. solfataricus aIF2 mRNA obtained by in vitro transcription of the cloned gene. The samples were incubated for 40 min at 73 °C and analyzed by SDS– PAGE. The arrow indicates the newly synthesized aIF2 protein (about 70 kDa); other bands resulting from the aIF2-stimulated translation of endogenous mRNAs are indicated with unlabelled arrows.

sound proof that IF2 has a relevant role in subunit joining is still lacking. Therefore, the existence of a complete functional overlap between eukaryal IF5B and archaeal or bacterial IF2 cannot be taken for granted as yet. In conclusion, although a final word on the function of aIF2 cannot yet be told, the available results seem to indicate that its principal role in translational initiation consists in stabilizing the interaction of met-tRNAi with the ribosome. Future work will tell whether or not it is also involved in subunit joining. Also, it must be kept in mind that aIF2 most probably works in co-operation with other factors, and that its role may be fully understood only when examined in the context of a completely reconstructed archaeal initiation system. For instance, a factor that is probably relevant for aIF2 function is the IF1 homologue, aIF1A, of which more shall be said in a subsequent section. 3.3. a/eIF2: the met-tRNAi binding factor In eukarya, it is well established that the fundamental function of adapting met-tRNAi into the ribosomal P site is carried out by the trimeric factor eIF2, whose component subunits are termed a, b and c [8]. Ensuring an efficient met-tRNAi/ribosome interaction is particularly crucial in the eukaryal domain as the 40S subunit must carry met-tRNAi to successfully identify the initiation codon during the mRNA scanning process. Therefore, eIF2 is a preferential target of mechanisms

191

regulating eukaryotic translation initiation. Inhibition of eIF2 activity is obtained through the phosphorylation of its a subunit; this is triggered by a number of stress signals which activate certain kinase cascades [47]. When the a subunit is phosphorylated, the factor becomes unable to interact with the guanine nucleotide exchange factor eIF2B. As a result, eIF2 remains locked in the inactive GDP-bound form and translation is severely inhibited or entirely shut off [36]. The other two subunits of eIF2, b and c, perform different tasks. The c subunit contains the guanine nucleotide binding site and part, at least, of the tRNA binding site [48], while the b subunit, besides being probably involved in tRNA binding, includes the domains for the interaction with the accessory factors, i.e. the GTP/ GDP exchange protein eIF2B and the GTPase eIF5 [49,50]. In summary, eIF2 is a very complex factor, which lies at the centre of an elaborate network of interactions regulating its activity. Until recently, the prevalent rationale for the divergence of the tRNAi binding factors in bacteria and eukarya was that the latter had to evolve a brand-new protein to achieve a more finely tuned and sophisticated regulation of translational initiation than that required by the simple prokaryotic cells. Needless to say, this view has been challenged by the discovery that all archaeal genomes so far sequenced include homologues of the three component subunits of eukaryal eIF2. That these proteins associate to form a trimeric complex has been demonstrated in a couple of recent works on P. abyssi a/eIF2 [51,52]. Trimeric a/eIF2 is decidedly smaller in size than its eukaryal counterpart, owing principally to the much-reduced length of its b subunit. In all archaea, the c subunit is the largest protein (about 45 kDa) followed by the a subunit (about 30 kDa). The b subunit of only 15 kDa comes a far third, while in the eukarya it is generally about the same size as the c subunit (about 50 kDa). The archaeal b polypeptide is so much smaller than its eukaryal counterpart because it lacks the domains, which, in the eukaryal subunit, interact with eIF2B (the guanine nucleotide exchange factor) and eIF5 (the GTPase). This agrees with the observation that all archaea lack a homologue of eIF5 as well as a complete eIF2B. Archaeal genomes do include homologues of the a, b and d subunits of eIF2B, which in the eukaryotic factor have a regulatory role. However, no counterparts of the catalytic subunits c and e have been detected. Therefore, it is probable that the archaeal homologues of eIF2B a, b and d subunits have a function unrelated to guanine nucleotide exchange [53]. The crystal structure of the IF2B a subunit of P. horikoshii has been solved and it has been proposed that it exists in a complex with the b and d homologues [54]. The small archaeal b subunit is almost entirely occupied by a conserved domain containing a zinc-finger

192

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

motif whose function in the eukaryal polypeptide is still uncertain, although it has been proposed to bind mRNA [55]. Has a/eIF2 the same function as its eukaryal homologue? Recently, it has been demonstrated that the factor from Pyrococcus abyssi, reconstituted in vitro from the cloned recombinant subunits, is able to interact preferentially with met-tRNAi. Apparently, the dimer formed by the a and c subunit is sufficient for a stable binding of met-tRNAi; however, the c subunit by itself is also able to interact with the tRNA, although in a much less stable fashion [52]. The recent determination of the crystal structure of the isolated c subunit of P. abyssi a/eIF2 [51] has also shown that it contains a complete tRNA binding domain similar to that found in the translational elongation factor Tu, another G protein showing homology with the c subunit of eIF2 and a/ eIF2. It seems therefore established that a/eIF2 is the initiator tRNA binding factor in the archaea, although the details of its function remain to be worked out. Of the many open questions, one regards the nature of the tRNA binding site: the work of Yatime et al. [52], as well as data from our laboratory suggests that a ac dimer is necessary and sufficient for a stable interaction with met-tRNAi (Fig. 3). This differs from the situation in eukarya, where the c subunit also has the largest share in met-tRNAi binding [48], but its principal helper seems to be the b subunit [56], while the a polypeptide has been implicated mainly in the regulation of the fac-

30

pmol met-tRNA bound

25

20

15

10

5

0

alfa

beta

gamma

a-b

a-g

b-g

trimer

Fig. 3. The ac dimer of a/eIF2 is necessary and sufficient for mettRNAi binding. About 50 pmol each of purified recombinant a, b and c subunits of S. solfataricus a/eIF2 were incubated, individually and in all possible combinations, with an excess of S. solfataricus tRNA charged with 35S met. The samples were incubated for 15 min at 65 °C and the amount of met-tRNA bound to the proteins was determined by filtration through nitrocellulose disks (0.22 lm).

tor activity. However, the eukaryal b subunit has evidently incurred an extensive evolutionary rearrangement, increasing its size by the addition of new domains. Therefore, it is quite possible that it has also achieved a role in tRNAi binding, absent in its archaeal homologue. Another question regards the regulation of guanine nucleotide binding to a/eIF2. As explained previously, the complex of the eukaryal factor with GDP is several orders of magnitude more stable that that with GTP. As a consequence, following GTP hydrolysis, the eIF5B factor is required to catalyze GDP/GTP exchange in order to reactivate eIF2. As a complete IF5B is missing in the archaea, it is reasonable to suppose that a/eIF2 can dispense of a GTP/GDP exchange factor. If so, no eukaryal-type functional regulation of a/eIF2 based on the inhibition of guanine nucleotide exchange should exist in archaea. However, it has been reported recently that the a subunit of P. horikoshii a/eIF2 is phosphorylated by a specific protein kinase [57]. If these results are confirmed, determining the function of a subunit phosphorylation of archaeal a/eIF2 will be a quite interesting task, which may help to clarify why the archaea, unlike bacteria but like eukarya, have adopted a trimeric mettRNAi binding factor. The role of the a/eIF2 b subunits remains at the moment elusive. In our laboratory, we have so far failed to detect any involvement of this protein in any of the factorÕs functions, although it must be remembered that these are in vitro experiments. Further work is necessary to identify the conserved role of the a/eIF2 b subunit along the archaeal/eukaryal evolutionary line. Another problem that remains as yet unsolved is whether a/eIF2 possesses an intrinsic GTPase activity or whether it requires an accessory GTPase as its eukaryal counterpart. As discussed earlier, GTP hydrolysis on eukaryal eIF2 is triggered by the helper factor eIF5, which has no recognizable homologue in archaea. Therefore, a/eIF2 either has an autonomous GTPase activity or is helped by a new and still unidentified GTPase. Some years ago Kyrpides and Woese [53] suggested that the companion GTPase of a/eIF2 might be aIF2. As said previously, aIF2 does possess an intrinsic ribosome-dependent GTPase activity, and it is just possible that, upon interacting with the ribosome, it would promote the hydrolysis of its own bound GTP and of that bound to a/eIF2. This hypothesis is currently being investigated in our laboratory. 3.4. aIF1A and aSUI1 As shown in Table 1, the list of putative archaeal initiation factors homologous to eukaryal proteins includes two polypeptides known as eIF1A and eIF1 (also termed SUI1 in yeast). The most interesting feature of these proteins is that they have homologues also in the

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

bacterial domain and therefore, like IF2, are universally conserved initiation factors [42]. Bacterial IF1 and archaeal/eukaryal IF1A constitute the IF1-like factor family. Bacterial IF1 is a small polypeptide of 71 amino acids essential for cell survival [58]. It has been shown to interact with the A site on the 30S ribosomal subunits [59]. Its reported functions include: (1) the enhancement of the rate of 70S ribosome dissociation and subunit association, (2) the stimulation of the activity of IF2 and IF3 in the formation of the 30S ternary complex, (3) the modulation of the interaction of IF2 with the ribosome, increasing its affinity for the 30S ribosomal subunit when IF1 is bound and indirectly favouring its release when IF1 is ejected [4]. The eukaryal IF1-like protein, aIF1A, is a 17-kDa polypeptide that is essential for cell survival as its counterpart in bacteria. The factor has been shown to stimulate the binding of met-tRNAi in the P site [60] and, in collaboration with eIF1 (SUI1), to promote ribosome scanning and correct initiation codon selection [61]. Interestingly, it has been shown that eIF1A interacts with eIF5B (the IF2 homologue), forming a stable complex even outside the ribosome [62,63]. As some earlier data suggested that bacterial IF1 and IF2 may also interact [64], the proposal has been put forward that a complex between the IF1-like and the IF2-like proteins is a universal feature of translation initiation in all domains [46]. The function of such a complex would be that of occupying the ribosomal A site during the early stages of initiation, thereby stimulating the correct access of tRNAi in the P site and promoting the joining of the large subunit to the initiation complex. However, this model is based mainly on the properties of the eukaryal proteins and requires more data on their homologues in the other two domains to be validated. For instance, a convincing demonstration that bacterial IF1 and IF2 interact is still lacking; in any event, a putative complex of these proteins would not be stable in a free state and would form only on the surface of the ribosome [63]. Our laboratory is currently investigating the role of aIF1A in archaeal translation initiation. The protein of S. solfataricus has been obtained by PCR cloning and over-expression in E. coli. In vitro assays with the recombinant S. solfataricus aIF1A have shown that it enhances the activity of aIF2 in promoting the entry of met-tRNAi in the ribosomal P site, suggesting an interaction between the two proteins. Indeed, when mixed in solution in equimolar amounts under appropriate conditions, in the absence of ribosomes or other translational components, the two factors can be coimmunoprecipitated, indicating that they form a sufficiently stable complex in vitro (Fig. 4). These results, albeit still preliminary, are in agreement with the hypothesis that a complex of the IF1-

193

Fig. 4. aIF2 and aIF1A interact in vitro in the absence of ribosomes. About 50 pmol of purified recombinant S. solfataricus aIF2 and aIF1A were mixed and incubated at 65 °C for 15 min in the presence or in the absence of 1 mM GTP. Immunoprecipitation was carried out using polyclonal antibodies specific for aIF2. The precipitated proteins were separated by SDS–PAGE and revealed by Western blotting with the anti His-tag antibodies that stain both recombinant proteins. The position of the recombinant proteins and that of the antibodies heavy and light chains (Ab h and Ab l) are indicated by the arrows. Lane 1: immunoprecipitation with 1 mM GTP; 2 = same without GTP; 3 = control recombinant aIF2; 4 = control recombinant aIF1A; 5 = precipitation with pre-immune serum.

like and the IF2-like factors is a common feature of the three primary domains. More experimental data are required, however, to work out the structural and functional features of the archaeal aIF1A/aIF2 complex, and especially to determine whether it has any role in promoting subunit joining, supposedly an important aspect of its universally conserved function [46]. Another interesting protein, essential for eukaryotic translation and present in all archaea so far examined is the factor known as eIF1 (in higher eukaryotes) or SUI1 (in yeast). In eukarya, it seems to have an important role in ensuring the accuracy of initiation codon selection and in controlling translational fidelity during elongation [65]. In bacteria, however, it is well established that the ‘‘fidelity factor’’ for initiation codon selection is IF3, a strictly bacterial protein absent in both archaea and eukarya (Table 1) and unrelated to eIF1/SUI1. Certain bacterial species, including E. coli, do possess a homologue of eIF1/SUI1, whose function is still unknown [66]. Therefore, eIF1/SUI1 is a third universally conserved factor, which is, however, present only in a minority of bacterial phyla while being found in all archaea and eukarya [42]. The peculiar distribution in the bacteria of the SUI1like factor (known as YciH in E. coli) is amenable to two different interpretations. The protein may have arisen in the archaeal/eukaryal branch and may have been acquired later by some bacterial phyla by horizontal gene transfer. Alternatively, it may be a truly ancestral factor, i.e. already present in the LUCA and later lost in most bacterial phyla during the separate evolution of the

194

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

bacterial domain. Phylogenetic trees drawn by comparing the available sequences of SUI1-lke proteins in the three domains favour the latter interpretation. In fact, bacterial, archaeal and eukaryal SUI1 homologues group into clearly distinct clades (Fig. 5). If the bacteria had acquired the protein by horizontal gene transfer from the archaea or the eukarya, bacterial SUI1 homologues would have clustered either with their archaeal or with their bacterial counterparts. The S. solfataricus SUI1 homologue (aSUI1) has been cloned and purified in our laboratory. The most relevant result so far obtained is that the protein is tightly bound to the small ribosomal subunits, where it remains during both translational initiation and elongation (Fig. 6). Currently, we are investigating whether the over-expression of aSUI1 in a cell-free protein synthesizing system exerts any effect on the translation of leadered or leaderless mRNAs endowed with AUG or other initiation codons.

Fig. 6. S. solfataricus SUI1 is localized on the 30S ribosomal subunits. About 10 ll of S. solfataricus cell extract were loaded on a 10–30% sucrose density gradient made in 30 mM KCl, 20 mM Tris/HCl pH 7, 10 mM MgCl2, which was run at 36,000 rpm for 4 h in a Beckman SW41 rotor. The gradient was fractionated and the position of the ribosomal subunits was determined by optical density measurements at 260 nm (top graph). The fractions were then probed for the presence of SUI1 with polyclonal antibodies raised against recombinant S. solfataricus SUI1 (bottom strip).

3.5. aIF5A and aIF6

Fig. 5. Distance tree showing the phylogenetic distribution of the available sequences of SUI1-like proteins. Numbers on the branches represent bootstrap values. The high values at the branches (92%) indicate that the separation of the bacterial sequences from the archaeal/eukaryal ones is quite reliable. The scale bar gives the number of amino acid substitutions per site. The three bacterial phyla (cProteobacteria, Cyanobacteria, Bacteroides Cytophaga) containing the SUI1-like factor are evidenced. A = archaeal sequences; E = eukaryal sequences.

Scarce or no information is available about these archaeal factors. IF5A is actually an interesting protein: it belongs to the universal set of factors (Table 1) and appears to have a conserved function in bacteria and eukarya, namely that of promoting the formation of the first peptide bond [39]. This makes it more similar to a specialized elongation factor than to an initiation factor, and in fact, the protein is termed EF (i.e. elongation factor) P in the bacteria (see Table 1). The function of archaeal IF5A has not been investigated experimentally, although it is likely to be similar to that of its bacterial and eukaryal counterparts. Curiously, the presence of this protein in the archaeal lineage has been known for a long time since, like its eukaryal counterpart, it contains a unique hyper-modified form of lysine known as hypusine [67]. The structure of aIF5A from M. jannaschii has been solved by crystallographic techniques [68]. Studying the function of the archaeal IF6 homologue should be high on the agenda of investigators working in the translation initiation field. As explained previously, eukaryal IF6 inhibits the association of the ribosomal subunits when bound to the 60S particle in its non-phosphorylated form. Phosphorylation of eIF6 triggers its release from the 60S subunits thereby allowing the assembly of 80S ribosomes and stimulating translation [40]. It will be very interesting to see whether a similar kind of regulation also operates in the archaea.

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

4. Concluding remarks: towards a reconstruction of the ancestral translation initiation mechanism As will be clear from the above overview of the state of the art, the study of translation initiation in archaea is still in its infancy and either mechanism or components of the process are far from being fully understood. Nevertheless, the information available from the archaea has greatly changed our perception of the evolution of translation initiation. Although it is too early to reconstruct in a meaningful way the evolutionary history of translation initiation, some speculations can be made that will also help to shape future research in the field. 4.1. Structure of the mRNAs and initiation mechanism As described in a previous paragraph, there is enough evidence to state with some certainty that the archaea employ two different initiation mechanisms to translate leadered and leaderless mRNAs [34]. Since there is no apparent physiological reason that may explain this fact, a possibility is that ‘‘leaderless’’ and ‘‘leadered’’ initiation have different evolutionary histories. According to this view, ‘‘leaderless’’ initiation would be the ancestral mechanism, carried over from the LUCA and retained preferentially in the archaea because of their slower evolutionary rate [33,34]. Several observations argue in favour of the hypothesis that ‘‘leaderless’’ initiation is the ancestral mechanism. The most compelling evidence is that leaderless mRNAs are universally translatable (at least in vitro) by archaeal, bacterial and eukaryotic ribosomes [33]. Since ‘‘normal’’ eukaryotic mRNAs are poorly, if at all, translated in bacterial systems (and vice versa) this is a remarkable fact that argues for a common conserved mechanism underlying leaderless translation. The recent observation that in the protozoan Giardia lamblia most mRNAs are leaderless [69] may also be taken as supporting the ancestral nature of such mRNAs, depending on whether or not G. lamblia is classified as a primitive eukaryote. It is known that Giardia occupies a deep branch of the eukaryal evolutionary tree, but it is still doubtful whether this indicates true primitivity or is an artefact of evolutionary analysis, due to an abnormally fast ‘‘ticking’’ of the molecular clock, a frequent occurrence in parasitic organisms like G. lamblia. If leaderless mRNAs are indeed the ancestral type of message, what about the SD-motif based, direct mRNA/ ribosome interaction also found in present-day archaea? The logical surmise is that this mechanism has arisen to allow an efficient decoding of polycistronic messages, especially of the cistrons located internally in a long mRNA. Since extant bacteria also possess polycistronic mRNAs and SD motifs, a possibility is that the evolution of the SD-based mechanism of ribosome/mRNA interaction has preceded the branching of the bacterial

195

domain off the root of tree of life, i.e. has occurred while still at the LUCA stage of cellular evolution. If so, one may envisage that the LUCA underwent several steps in the evolution of translation initiation, starting, in a more primitive stage, with monocistronic leaderless mRNAs to achieve, in a later stage, also polycistronic mRNAs endowed with SD motifs in the internal cistrons but still leaderless as far as the initial cistron was concerned. This is the situation still prevalent in the Archeal domain, especially in the Crenarcheota. The alternative scenario would be the independent evolution of polycistronic mRNAs and SD/anti-SD interaction in bacteria and archaea, which, although possible, seems less likely, considering that the order and number of genes in several operons are conserved between bacteria and archaea [14]. If polycistronic mRNAs and the SD/anti-SD based mRNA/ribosome interaction were already present in the LUCA, it follows that they have been lost during the evolution of Eukaryotes. The evolutionary pressure underlying such loss has been, in all probability, the separation between nucleus and cytoplasm leading in turn to the separation between transcription and translation [70]. On the other hand, the ‘‘ribosome scanning’’ mechanism of present-day Eukaryotes may be envisaged as an elaborate version of leaderless initiation: it is based on ribosome binding at the 5 0 end of the message, with the added functions of RNA unfolding and directional ribosome movement to hit the correct initiation AUG codon, which is recognized mainly by means of basepairing with the anticodon of the met-tRNAi. Notwithstanding the added sophistications, the basic mechanism has been conserved to the extent that modern eukaryotic ribosomes are still able to translate leaderless mRNAs from bacterial sources [33]. To gain a better understanding of the evolutionary status of ‘‘leaderless’’ initiation we also need to determine which initiation factors it requires. In bacteria, translation of leaderless mRNAs has been shown to be stimulated by an excess of IF2, perhaps because this factor promotes the loading of initiator tRNA on the 30S subunits [33]. Recent studies have suggested that bacterial leaderless mRNAs may be read by 70S ribosomes, even without the aid of any factors [71,72]. On the other hand, the protozoan Giardia, whose mRNAs are all leaderless, seems to be endowed with a normal eukaryotic complement of initiation factors [69]. It is hoped that the study of archaea may shed light on these seemingly contradictory evidences. 4.2. Initiation factors in ancestral translation The previous survey of the available information about archaeal initiation factors will have made clear that we need to know much more about the function of these proteins. A glance to Table 1 will show that

196

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

the majority of the archaeal IFs (4 over 6) are in fact universally conserved proteins. The other two factors (a/eIF2 and aIF6) are shared only with the eukaryotes and therefore arose in the common ancestor of archaea and eukarya, after the branching of the bacterial domain. As yet, no archaeal-specific translation initiation factor has been identified, but this is not very surprising since the search for initiation factors has been carried out looking for homologies with the bacterial or eukaryal proteins. Obviously, any archaeal-specific IF would have escaped detection: therefore, it is entirely possible that archaeal translation initiation makes use of unique factors still to be discovered. Needless to say, it would be of the utmost interest to unravel the function of the universally conserved factors in ancestral initiation. At present, however, only a few speculations are possible. As shown in Table 1, the set of ancestral factors includes four proteins: the IF1-like factors, the IF2-like factors, the SUI1- like factors and the EFP/IF5A proteins. Three of these (IF1, IF2, IF5A/EF-P) are ubiquitous in the three primary domains. The SUI1-like factors are not truly universal, since they are missing in the majority of bacterial phyla, indicating that they are probably not essential in modern bacterial translation. However, the reconstruction of the evolutionary tree of the SUI1-like factors (see Fig. 5) suggests that they, too, were present in the LUCA and have later been lost in most bacterial species. It is not known whether bacterial SUI1 is functional in the species that still possess it; likewise, we are still ignorant of the function in translation of archaeal SUI1. At present, it is therefore impossible to speculate meaningfully on the hypothetical ancestral function of this factor. As to the truly universal factors, the EF-P/IF5A proteins seem to have the same function in bacteria and eukarya, namely stimulating the formation of the first peptide bond [39]. Even if the function of archaeal IF5A has not been studied yet, it is safe enough to assume that it will turn out to be similar to that of its eukaryal and bacterial counterparts. Therefore, the EF-P/IF5A factors can be looked at as a sort of specialized elongation factors that catalyzed the formation of the first peptide bond in ancestral and modern cells alike. As regards the IF2-like proteins, the available data on their present function in the primary domains suggest that their ancestral role consisted in promoting the interaction of tRNAi with the ribosomal P site. The other function detected in eukaryotic cells, stimulating ribosomal subunit joining, requires further data in the bacteria and the archaea to be validated as universal. It is debatable whether the ancestral IF2-like factor was capable of direct tRNA binding. In present-day cells, only bacterial IF2 has this function, while its eukaryal and archaeal counterparts do not bind met-

tRNAi. Therefore, the tRNA-binding capacity may have been acquired during the evolution of bacterial IF2, or may have been lost by the archaeal/eukaryal IF2 homologues. At present, it is impossible to decide between these possibilities; however, it must be pointed out that bacterial IF2 is specialized for interacting with f-met-tRNAi, and it is known that the formyl group plays an important role in this interaction [45]. Since formylation of tRNA-bound methionine is a unique bacterial feature, it is perhaps easier to imagine that the ancestral IF2 had no special tRNA-binding capacity, but could indirectly favour the entry and stabilization of tRNA in the P site by interacting with the ribosome in correspondence of the A site. In ancestral translation (as well as in modern one), the IF2-like factor worked in close collaboration with the IF1-like protein, also a universally conserved factor. As we have seen, the available experimental evidence suggests that the IF1-like and the IF2-like factors interact in all cells to form a complex designed to occupy the ribosomal A site during translational initiation [46,63]. Further work on the bacterial and archaeal IF2-like and IF1-like proteins is needed to confirm this surmise and to work out the structural and functional features of the hypothetical universal complex. To complete the picture, it is worthwhile to spend a few words on the possible evolutionary history of the eukaryal/archaeal met-tRNAi binding factor, the trimeric protein eIF2. The c subunit of this factor belongs to a family of tRNA-binding G proteins that includes the elongation factor Tu/1-A and its specialized version SELB [73]. The latter is the factor that selects the selenocysteine tRNA during the synthesis of selenoproteins in all primary domains [74–76]. It is interesting to note that the c subunit of eukaryal and archaeal IF2 has the best homology with SELB [73]. This indicates that a/eIF2 c arose by gene duplication from the elongation factor Tu at about the same time as SELB; each of the two proteins became specialized for interacting with a single aminoacyl-tRNA species, while the elongation factor Tu remained capable of binding any aminoacyl-tRNA. Probably, a/eIF2 acquired a selective affinity for met-tRNAi during the evolution of the common ancestor of archaea and eukarya. It is probable that in the eukaryal/archaeal line the c subunit of IF2 initially bound met-tRNAi as a monomer. This is indicated by the fact that the c polypeptide hosts the tRNA binding domain and can still interact weakly with met-tRNAi in isolation [52]. The later addition of the a and b subunits improved the met-tRNAi binding capacity of the factor and added regulatory functions, at least in the eukaryal lineage. However, a full understanding of the evolution of this factor must await a better knowledge of its structure and function in the archaea.

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

197

Fig. 7. The rectangles indicated with the letters A, P, E indicate the respective ribosomal tRNA binding sites. The blue triangle is met-tRNAi. The ovals marked 1A and 2 symbolize the IF1-like and the IF2-like factors respectively. See text for the discussion.

4.3. Models of ancestral translational initiation In Fig. 7, the above considerations are combined in a hypothetical model of ancestral translation initiation. Our assumptions are that ancestral mRNAs were leaderless, and, at least in early LUCA, monocistronic. The first stages of initiation were assisted only by the IF1like and the IF2-like factors. An IF1/IF2 complex, probably assembled on the surface of the ribosome, occupied the ribosomal A site thereby guiding the tRNAi in the P site. Once tRNAi was stably bound in the P site, the ribosome could interact with the leaderless mRNA by codon–anticodon pairing. In early LUCA, ribosome/ mRNA interaction possibly required complete 70S ribosomes rather than isolated small subunits as it happens in modern initiation. Initiation complexes formed by isolated small subunits may have evolved later, together with the ‘‘invention’’ of the SD–anti-SD interaction. Two slightly different scenarios are conceivable depending on the assumptions made about the interaction of ancestral IF2 with tRNAi. If the factor was capable of direct tRNA binding, f-met-tRNA was probably already used as the initiator tRNA species. Therefore, the bacterial branch retained the streamlined ancestral situation, while the archaeal/eukaryal line evolved another factor for the binding of met-tRNAi and lost the tRNA binding domain on IF2/IF5B. Assuming that the ancestral IF2 had no tRNA binding-capacity, it is probable that the tRNAi chose the ribosomal P site primarily by negative default, the A site being occupied by the IF1/IF2 complex. Perhaps at this stage any tRNA could have been used as the initiator species, depending on the codon present at the 5 0 end of the leaderless mRNA. Later, bacterial IF2 acquired the capacity of selective f-met tRNA binding, while the archaeal/eukaryal branch evolved a different met-

tRNAi binding factor. Much of the uncertainty in drawing these models is due to the fact that we still do not really understand why different initiator tRNAs should be used in the bacteria and in the archaea/eukarya, or why (and when) AUG was chosen as the preferential initiation codon. Some aspects of the above models are, however, amenable to experimental testing. No doubt an exciting future awaits investigators interested in the evolution of translation.

Acknowledgements Thanks are due to the people in my lab who expertly and enthusiastically worked on archaeal initiation in the last few years: Alessandra Berardi, Dario Benelli, Enzo Maone, Nadia Pedulla´, Rocco Palermo, Laura Peri and Alessandra Barbazza. I especially thank Prof. Piero Cammarano for many helpful discussions and for working out the SUI1 phylogenetic tree. The work in the authorÕs laboratory was supported by MURST projects PRIN 2000 ‘‘Translational initiation of leadered and leaderless mRNAs in bacteria and archaea: comparative analysis of the functions of IF2, an essential and universally conserved translation initiation factor’’ and PRIN 2002 ‘‘Evolution of the gene expression machinery: characterization of a set of universally conserved factors modulating translational initiation’’.

References [1] Kozak, M. (1983) Comparison of initiation of protein synthesis in procaryotes, eucaryotes and organelles. Microbiol. Rev. 47, 1–45.

198

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

[2] Kozak, M. (1999) Initiation of translation in prokaryotes and eukaryotes. Gene 234, 187–208. [3] Gualerzi, C.O. and Pon, C.L. (1990) Initiation of mRNA translation in prokaryotes. Biochemistry 29, 5881–5889. [4] Gualerzi, C.O., Brandi, L., Caserta, E., La Teana, A., Spurio, R., Tomsic, J. and Pon, C. (2000) Translation initiation in bacteria In: The Ribosome: Structure, Function, Antibiotics and Cellular Interactions (Garrett, R.A., Douthwaite, S.R., Liljas, A., Matheson, A.T., Moore, P.B. and Noller, H.F., Eds.), pp. 477–494. ASM Press, Washington DC. [5] Kozak, M. (1989) The scanning model for translation: an update. J. Cell. Biol. 108, 224–229. [6] Kozak, M. (2002) Pushing the limits of the scanning mechanism for initiation of translation. Gene 299, 1–34. [7] Hellen, C.U. and Sarnow, P. (2001) Internal ribosome entry sites in eukaryotic mRNA molecules. Genes Dev. 15, 1593–1612. [8] Merrick, W.C. (1992) Mechanism and regulation of eukaryotic protein synthesis. Microbiol. Rev. 56, 291–315. [9] Pestova, T.V. and Hellen, C.U.T. (2001) Functions of eukaryotic factors in initiation of translation. Cold Spring Harb. Symp. Quant. Biol. 66, 389–396. [10] Pestova, T.V. and Kolupaeva, V.G. (2002) The roles of individual eukaryotic translation initiation factors in ribosomal scanning and initiation codon selection. Genes Dev. 15, 2906–2922. [11] Gallie, D.R. (1998) A tale of two termini: a functional interaction between the termini of an mRNA is a prerequisite for efficient translational initiation. Gene 216, 1–11. [12] Wells, S.E., Hillner, P.E., Vale, R.D. and Sachs, A.B. (1998) Circularization of mRNA by eukaryotic translation initiation factors. Mol. Cell 2, 135–140. [13] Amils, R., Cammarano, P. and Londei, P. (1993) Translation in archaea In: Biochemistry of Archaea (Kates, M., Kushner, D. and Matheson, A., Eds.), New Comprehensive Biochemistry series, pp. 393–437. Elsevier, Amsterdam. [14] Ramirez, C., Kopke, A.K.E., Yang, D-C., Boeckh, T. and Matheson, A.T. (1993) The structure, function and evolution of archaeal ribosomes In: Biochemistry of Archaea (Kates, M., Kushner, D. and Matheson, A., Eds.), New Comprehensive Biochemistry series, pp. 439–466. Elsevier, Amsterdam. [15] Dalgaard, J.Z. and Garrett, R.A. (1993) Archaeal hyperthermophile genes In: Biochemistry of Archaea (Kates, M., Kushner, D. and Matheson, A., Eds.), New Comprehensive Biochemistry series, pp. 535–563. Elsevier, Amsterdam. [16] Dennis, P.P. (1997) Ancient ciphers: translation in archaea. Cell 89, 1007–1010. [17] Bell, S.D. and Jackson, S.P. (1998) Transcription and translation in archaea: a mosaic of eukaryal and bacterial features. Trends Microbiol. 6, 222–227. [18] Lykke-Andersen, J., Aagard, C., Semionenkov, M. and Garrett, R.A. (1997) Archaeal introns: splicing, intercellular mobility and evolution. Trends Biochem. Sci. 22, 326–331. [19] Watanabe, Y., Yokobori, S., Inaba, T., Yamagishi, A., Oshima, T., Kawarabayasi, Y., Kikuchi, H. and Kita, K. (2002) Introns in protein-coding genes in Archaea. FEBS Lett. 510, 27–30. [20] Gogarten, J.P., Senejani, A.G., Zhaxybayeva, O., Olendzenski, L. and Hilario, E. (2002) Inteins: structure, function and evolution. Annu. Rev. Microbiol. 56, 263–287. [21] Kraft, A., Lutz, C., Lingenhel, A., Grobner, P. and Piendl, W. (1999) Control of ribosomal protein L1 synthesis in mesophilic and thermophilic archaea. Genetics 152, 1363–1372. [22] CondoÕ, I., Ciammaruconi, A., Benelli, D., Ruggero, D. and Londei, P. (1999) Cis-acting signals controlling translational initiation in the thermophilic archaeon Sulfolobus solfataricus. Mol. Microbiol. 34, 377–384. [23] Betlach, M., Friedman, J., Boyer, H.W. and Pfeifer, F. (1984) Characterization of a halobacterial gene affecting bacterio-opsin gene expression. Nucl. Acids Res. 20, 7949–7959.

[24] Shean, C.S. and Gottesman, M. (1992) Translation of the prophage k CI transcript. Cell 70, 513–522. [25] Etchegaray, J.P. and Inouye, M. (1999) Translational enhancement by an element downstream of the initiation codon in Escherichia coli. J. Biol. Chem. 274, 10079– 10085. [26] Sensen, C.W., Klenk, H.P., Singh, R.K., Allard, G., Chan, C.C., Liu, Q.Y., Penny, S.L., Young, F., Schenk, M.E., Gaasterland, T., Doolittle, W.F., Ragan, M.A. and Charlebois, R.L. (1996) Organizational characteristics and information content of an archaeal genome: 156 kb of sequence from Sulfolobus solfataricus P2. Mol. Microbiol. 1, 175–191. [27] Slupska, M.M., King, A.G., Fitz-Gibbon, S., Besemer, J., Borodovsky, M. and Miller, J.H. (2001) Leaderless transcripts of the crenarchaeal hyperthermophile Pyrobaculum aerophilum. J. Mol. Biol. 309, 347–360. [28] She, Q. et al. (2001) The complete genome of the crenarchaeon Sulfolobus solfataricus P2. Proc. Natl. Acad. Sci. USA 98, 7835– 7840. [29] Tolstrup, N., Sensen, C.W. and Garrett, R.A. (2000) Two different and highly organized mechanisms of translation initiation in the archaeon Sulfolobus solfataricus. Extremophiles 4, 175– 179. [30] Sartorius-Neef, S. and Pfeifer, F. (2004) In vivo studies on putative Shine–Dalgarno sequences of the halophilic archaeon Halobacterium salinarium. Mol. Microbiol. 51, 579–588. [31] OÕConnor, M., Asai, T., Squires, C.L. and Dahlberg, A.E. (1999) Enhancement of translation by the downstream box does not involve base pairing of mRNA with the penultimate stem sequence of 16S rRNA. Proc. Natl. Acad. Sci. USA 96, 8973– 8978. [32] La Teana, A., Brandi, L., OÕConnor, M., Freddi, S. and Pon, C.L. (2000) Translation during cold adaptation does not involve mRNA–rRNA base pairing through the downstream box. RNA 6, 1393–1402. [33] Grill, S., Gualerzi, C.O., Londei, P. and Bla¨si, U. (2000) Selective stimulation of translation of leaderless mRNA by IF2: evolutionary implications for translation. EMBO J. 19, 4101–4110. [34] Benelli, D., Maone, E. and Londei, P. (2003) Two different mechanisms for ribosome/mRNA interaction in archaeal translation initiation. Mol. Microbiol. 50, 635–643. [35] Gross, J.D., Moerke, N.J., von der Haar, T., Lugovskoy, A.A., Sachs, A.B., McCarthy, J.E. and Wagner, G. (2003) Ribosome loading onto the mRNA cap is driven by conformational coupling between eIF4G and eIF4E. Cell 115, 739–750. [36] Kimball, S.R. (1999) Eukaryotic initiation factor eIF2. Int. J. Biochem. Cell Biol. 31, 25–29. [37] Pestova, T.V., Lomakin, I.B., Lee, J.H., Choi, S.K., Dever, T.E. and Hellen, C.U. (2000) The joining of ribosomal subunits in eukaryotes requires eIF5B. Nature 403, 332–335. [38] Shin, B.S., Maag, D., Roll-Mecak, A., Arefin, M.S., Burley, S.K., Lorsch, J.R. and Dever, T.E. (2002) Uncoupling of initiation factor eIF5B/IF2 GTPase and translational activities by mutations that lower ribosome affinity. Cell 111, 1015–1025. [39] Aoki, H., Dekany, K., Adams, S.L. and Ganoza, M.C. (1997) The gene encoding the elongation factor P protein is essential for viability and is required for protein synthesis. J. Biol. Chem. 272, 32254–32259. [40] Ceci, M., Gaviraghi, C., Gorrini, C., Sala, L.A., Offenhauser, N., Marchisio, P.C. and Biffo, S. (2003) Release of eIF6 (p27BBP) from the 60S subunit allows 80S ribosome assembly. Nature 426, 579–584. [41] Choi, S.K., Lee, J.H., Zoll, W.L., Merrick, W.C. and Dever, T.E. (1998) Promotion of met-tRNAi Met binding to ribosomes by yIF2, a bacterial IF2 homolog in yeast. Science 280, 1757–1760.

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200 [42] Kyrpides, N.C. and Woese, C.R. (1998) Universally conserved translation initiation factors. Proc. Natl. Acad. Sci. USA 95, 224– 228. [43] Lee, J.H., Choi, S.K., Roll-Mecak, A., Burley, S.K. and Dever, T.E. (1999) Universal conservation in translation initiation revealed by human and archaeal homologs of bacterial translation initiation factor IF2. Proc. Natl. Acad. Sci. USA 96, 4342–4347. [44] Roll-Mecak, A., Cao, C., Dever, T.E. and Burley, S.K. (2000) Xray structures of the universal translation initiation factor IF2/ eIF5B: conformational changes on GDP and GTP binding. Cell 103, 781–792. [45] Guenneugues, M., Caserta, E., Brandi, L., Spurio, R., Meunier, S., Pon, C.L., Boelens, R. and Gualerzi, C.O. (2000) Mapping the fMet-tRNA f-met binding site of initiation factor IF2. EMBO J. 19, 5233–5240. [46] Roll-Mecak, A., Shin, B.-S., Dever, T.E. and Burley, S.K. (2001) Engaging the ribosome: universal IFs of translation. TIBS 26, 705–709. [47] Colthurst, D.R., Campbell, D.G. and Proud, C.G. (1987) Structure and regulation of eukaryotic initiation factor eIF2: sequence of the site in the a subunit phosphorylated by the haem-controlled repressor and by the doublestranded RNA-activated inhibitor. Eur. J. Biochem. 166, 357–363. [48] Erickson, F.L. and Hannig, M. (1996) Ligand interaction with eukaryotic translation initiation factor 2: role of the gamma subunit. EMBO J. 15, 6311–6320. [49] Das, S., Maiti, T., Das, K. and Maitra, U. (1997) Specific interaction of eukaryotic translation initiation factor 5 (eIF5) with the b subunit of eIF2. J. Biol. Chem. 272, 31712–31718. [50] Das, S. and Maitra, U. (2000) Mutational analysis of mammalian translation initiation factor 5 (eIF5): role of the interaction between the beta subunit of eIF2 and eIF5 in eIF5 function in vivo and in vitro. Mol. Cell. Biol. 11, 3942– 3950. [51] Schmitt, E., Blanquet, S. and Mechulam, Y. (2002) The large subunit of initiation factor aIF2 is a close structural homologue of elongation factor Tu. EMBO J. 21, 1821–1832. [52] Yatime, L., Schmitt, E., Blanquet, S. and Mechulam, Y. (2004) Functional molecular mapping of archaeal translation initiation factor 2. J. Biol. Chem. 279, 15984–15993. [53] Kyrpides, N.C. and Woese, C.R. (1998) Archaeal translation initiation revisited: the initiation factor 2 and eukaryotic initiation factor 2B a-b-d subunit families. Proc. Natl. Acad. Sci. USA 95, 3726–3730. [54] Kakuta, Y., Tahara, M., Maetani, S., Yao, M., Tanaka, I. and Kimura, M. (2004) Crystal structure of the regulatory subunit of archaeal initiation factor 2B (aIF2B) from hyperthermophilic archaeon Pyrococcus horikoshii OT3: a proposed structure of the regulatory subcomplex of eukaryotic IF2B. Biochem. Biophys. Res. Commun. 319, 725–732. [55] Laurino, J.P., Thompson, G.M., Pacheco, E. and Castilho, B.A. (1999) The beta subunit of eukaryotic translation initiation factor 2 binds mRNA through the lysine repeats and a region comprising the C2–C2 motif. Mol. Cell Biol. 19, 173–181. [56] Das, A., Bagchi, M.K., Ghosh-Dastidar, P. and Gupta, N.K. (1982) Protein synthesis in rabbit retticulocytes. A study of peptide chain initiation using native and b-subunit depleted eukaryotic initiation factor 2. J. Biol. Chem. 257, 1282–1288. [57] Tahara, M., Ohsawa, A., Saito, S. and Kimura, M. (2004) In vitro phosphorylation of initiation factor 2 alpha (aIF2 alpha) from hyperthermophilic archaeon Pyrococcus horikoshii OT3. J. Biochem. (Tokyo) 135, 479–485.

199

[58] Cummings, H.F. and Hershey, J.W.B. (1994) Translation initiation factor IF1 is essential for cell viability in E. coli. J. Bacteriol. 176, 198–205. [59] Carter, A.P., Clemons Jr., W.M., Brodersen, D.E., Morgan-Warren, R.J., Hartsch, T., Wimberly, B.T. and Ramakrishnan, V. (2001) Crystal structure of an initiation factor bound to the 30S ribosomal subunit. Science 291, 498–501. [60] Chaudhuri, J., Si, K. and Maitra, U. (1997) Function of eukaryotic translation initiation factor 1A (eIF1A) in initiation of protein synthesis. J. Biol. Chem. 272, 7883–7891. [61] Pestova, T.V., Borukhov, S.I. and Hellen, C.U.T. (1998) Eukaryotic ribosomes require initiation factors 1 and 1A to locate initiation codons. Nature 394, 854–859. [62] Choi, S.K., Olsen, D.S., Roll-Mecak, A., Martung, A., Remo, K.L., Burley, S.K., Hinnebusch, A.G. and Dever, T.E. (2000) Physical and functional interaction between the eukaryotic orthologs of initiation factors IF1 and IF2. Mol. Cell. Biol. 20, 7183–7191. [63] Marintchev, A., Kolupaeva, V.G., Pestova, T.V. and Wagner, G. (2003) Mapping the binding interface between human eukaryotic initiation factors 1A and 5B. Proc. Natl. Acad. Sci. USA 100, 1535–1540. [64] Boileau, G., Butler, P., Hershey, J.W. and Traut, R.R. (1983) Direct cross-link between initiation factors 1,2,3 and ribosomal proteins promoted by 2-iminothiolane. Biochemistry 22, 3162– 3170. [65] Cui, Y., Dinman, J.D., Kinzy, T.G. and Peltz, S.W. (1998) The Mof2/Sui1 protein is a general monitor of translational accuracy. Mol. Cell. Biol. 18, 1506–1516. [66] Cort, J.R., Koonin, E.V., Bash, P.A. and Kennedy, M.A. (1999) A phylogenetic approach to target selection for structural genomics: solution structure of YciH. Nucl. Acids Res. 27, 4018–4027. [67] Schumann, H. and Klink, F. (1989) Archaebacterial protein contains hypusine, a unique amino-acid characteristic for eukaryotic translation initiation factor-4D. Syst. Appl. Microbiol. 11, 103–107. [68] Kim, K.K., Hung, L.W., Yokota, H., Kim, R. and Kim, S.H. (1998) Crystal structures of eukaryotic translation initiation factor 5A from Methanococcus jannaschii at 1.8 A resolution. Proc. Natl. Acad. Sci. USA 95, 10419– 10424. [69] Li, L. and Wang, C.C. (2004) Capped mRNA with a single nucleotide leader is optimally translated in a primitive eukaryote, Giardia lamblia. J. Biol. Chem. 279, 14656–14664. [70] Londei, P. (1998) A hypothesis on the mechanism of translational initiation. Biochim. Biophys. Acta 1396, 169–178. [71] Udagawa, T., Shimizu, Y. and Ueda, T. (2004) Evidence for the translation initiation of leaderless mRNAs by the intact 70S ribosome without its dissociation into subunits in eubacteria. J. Biol. Chem. 279, 8539–8546. [72] Moll, I., Hirokawa, G., Kiel, M.C., Kaji, A. and Blaesi, U. (2004) Translation initiation with 70S ribosomes: an alternative pathway for leaderless mRNAs. Nucl. Acid. Res. 32, 3354– 3363. [73] Keeling, P.J., Fast, N.M. and McFadden, G.I. (1998) Evolutionary relationships between translation initiation factor eIF2 gamma and selenocysteine-specific elongation factor SELB: change of function in translation factors. J. Mol. Evol. 47, 649– 655. [74] Ringquist, S., Schneider, D., Gibson, T., Baron, C., Bock, A. and Gold, L. (1994) Recognition of the mRNA selenocysteine insertion sequence by the specialized translation elongation factor SELB. Genes Dev. 8, 376–385. [75] Fagegaltier, D., Hubert, N., Yamada, K., Mizutani, T., Carbon, P. and Krol, A. (2000) Characterization of mSelB, a novel

200

P. Londei / FEMS Microbiology Reviews 29 (2005) 185–200

mammalian elongation factor for selenoprotein translation. EMBO J. 19, 4796–4805. [76] Rother, M., Wilting, R., Commans, S. and Bock, A. (2000) Identification and characterisation of the selenocysteine-specific

translation factor SelB from the archaeon Methanococcus jannaschii. J. Mol. Biol. 299, 351–358. [77] Berardi, A. (2002) PhD thesis, University of Rome, La Sapienza.