The position of mRNA on 40S ribosomal subunits in eukaryotic initiation complexes was determined by UV crosslinking using mRNAs containing uniquely positioned 4‐thiouridines. Crosslinking of mRNA positions +11 to ribosomal protein (rp) rpS2(S5p) and rpS3(S3p), and +9–+11 and +8–+9 to h18 and h34 of 18S rRNA, respectively, indicated that mRNA enters the mRNA‐binding channel through the same layers of rRNA and proteins as in prokaryotes. Upstream of the P‐site, the proximity of positions −3/−4 to rpS5(S7p) and h23b, −6/−7 to rpS14(S11p), and −8–−11 to the 3′‐terminus of 18S rRNA (mRNA/rRNA elements forming the bacterial Shine–Dalgarno duplex) also resembles elements of the bacterial mRNA path. In addition to these striking parallels, differences between mRNA paths included the proximity in eukaryotic initiation complexes of positions +7/+8 to the central region of h28, +4/+5 to rpS15(S19p), and −6 and −7/−10 to eukaryote‐specific rpS26 and rpS28, respectively. Moreover, we previously determined that eukaryotic initiation factor2α (eIF2α) contacts position −3, and now report that eIF3 interacts with positions −8–−17, forming an extension of the mRNA‐binding channel that likely contributes to unique aspects of eukaryotic initiation.
During translation, mRNA engages in extensive, functionally important interactions with rRNA and protein components of the small ribosomal subunit, which determine the start site for initiation and ensure the accuracy of tRNA selection during elongation. The path of mRNA on the prokaryotic 30S subunit has been visualized by X‐ray crystallography of ribosomal complexes that correspond to different states in translation (Yusupova et al, 2001, 2006; Jenner et al, 2005, 2007; Selmer et al, 2006; Korostelev et al, 2007). Thus, mRNA is threaded through a channel that wraps around the neck of the 30S subunit, passing through non‐covalently closed tunnels as it enters between the head and shoulder and as it exits between the head and platform, leaving only nts −1–+7 (relative to the +1 A of the AUG initiation codon) exposed on the interface surface. The A‐ (aminoacyl) and P (peptidyl)‐site codons are centred over the axis of helix 44 of 16S rRNA, and a kink between them allows simultaneous pairing with the anticodons of A‐ and P‐site tRNAs without steric clashes. mRNA nts +7–+15 pass through rRNA (h34, h28, the 5′ hairpin loop and the 530 loop) and ribosomal protein ((rp) S3p, S4p and S5p) layers of the narrow entry tunnel. However, the contact between the head and the body formed between h18 in the body and h34 and S3p in the neck, which constitutes part of the mRNA entry tunnel, was observed in 70S ribosomes and in some but not all 30S subunit structures (e.g. Frank et al, 1995; Clemons et al, 1999; Gabashvili et al, 1999; Schluenzen et al, 2000), indicating that it can potentially function as a ‘latch’. It has been suggested that closing of the tunnel could contribute to processivity and directionality, whereas the open conformation could facilitate ribosomal attachment of mRNA during initiation (Schluenzen et al, 2000). The interactions of mRNA with the 30S subunit upstream of the P‐site codon differ at the various stages of translation. In initiation complexes, mRNA nts −5–−12 base pair with 16S rRNA to form the Shine–Dalgarno (SD) duplex located in a cleft formed by S11p, S18p and h20, h28, h37 and the 723 bulge loop of 16S rRNA, and is oriented towards S18p. Nucleotides −1–−4 containing the E‐site codon are positioned in the short exit tunnel between the head and platform and are surrounded by S7p, h45, h28, and the 690 and 790 loops of 16S rRNA. In the post‐initiation state, initial elongation cycles result in lengthening of the SD duplex and its reorientation towards S2p. Further translation leads to melting of the SD duplex and loss of stable interactions of the ribosome with mRNA upstream of position −5. The E‐site nucleotides −1–−3 are positioned on top of a bulge between h44 and h45 and near h24 in initiation and post‐initiation complexes, but have an altered conformation in the latter, adopting an A‐helical shape that can base pair with the E‐site tRNA, and no longer interact with S7p. During the first stage in initiation, mRNA binds prokaryotic ribosomes by forming the SD duplex, which is oriented towards S2p as in post‐initiation complexes, whereas the rest of mRNA is still unbound.
By contrast, much less is known about the mRNA path on eukaryotic ribosomes. There are fundamental mechanistic differences between eukaryotes and prokaryotes in initial ribosomal attachment of mRNA and selection of the initiation codon (Pestova et al, 2007), so that knowledge of the ribosomal position of mRNA and its functional contacts with components of the 40S subunit and with eukaryotic initiation factors (eIFs) at different stages of initiation are of particular interest. Ribosomal attachment to eukaryotic mRNAs is mediated by eIFs instead of by the SD interaction. Moreover, in contrast to 30S subunits that directly attach to the initiation codon region of prokaryotic mRNAs, eukaryotic 43S preinitiation complexes (comprising a 40S subunit, eIF2·GTP/Met‐tRNAMeti, eIF3, eIF1 and eIF1A) first attach to the capped 5′‐proximal region of mRNA and then scan along the 5′‐untranslated region (5′‐UTR) to the initiation codon where they form 48S initiation complexes with established P‐site codon–anticodon base pairing. Initial attachment of 43S complexes is mediated by eIF4F, eIF4A and eIF4B. eIF4F comprises eIF4E (cap‐binding protein), eIF4A (a helicase, whose activity is enhanced by eIF4B) and eIF4G (a scaffold for eIF4E and eIF4A that also binds eIF3). eIFs 4F, 4A and 4B unwind the cap‐proximal region of mRNA and likely promote attachment of 43S complexes through the eIF3–eIF4G interaction. The location of eIF4A/4B/4F on the 40S subunit is unknown, and the molecular mechanism by which mRNA bound to eIF4F enters the mRNA‐binding cleft therefore remains unresolved. Subsequent scanning requires eIF1, which promotes adoption of a scanning‐competent conformation by 43S complexes and enables them to recognize and reject codon–anticodon mismatches (Pestova and Kolupaeva, 2002; Lomakin et al, 2003, 2006). However, the molecular mechanism of scanning, the ribosomal position of mRNA in scanning 43S complexes and its potential contacts with eIFs in these complexes, as well as the mechanism by which eIF4A/4B/4F assist scanning remain unknown. Initiation codon recognition in metazoans is strongly influenced by flanking nucleotides: the sequence GCC(A/G)CCAUGG (in which the initiation codon is underlined) is optimal for initiation, and deviations from it at the −3 and +4 positions (in bold) lead to initiation at the next downstream AUG triplet by leaky scanning (Kozak, 1991). eIF1 is also responsible for the ability of 43S complexes to discriminate against AUG codons in poor context (Pestova and Kolupaeva, 2002; Pisarev et al, 2006). Purines at the +4 and −3 positions interact specifically with AA1818−1819 of 18S rRNA and with the eIF2 α‐subunit, respectively, and it was proposed that these interactions may stabilize conformational changes that likely occur in ribosomal complexes upon initiation codon recognition (Pisarev et al, 2006). After 48S complexes have formed, eIF5 triggers hydrolysis of eIF2‐bound GTP, and eIF5B then mediates displacement of factors from the 40S subunit and joining with a 60S subunit, yielding an elongation‐competent 80S initiation complex.
The position of mRNA on the 40S subunit, as well as its interactions with translation components, except those of nucleotides at positions −3 and +4, is not known for ribosomal complexes at any stage of initiation. RNAse protection studies have shown that, similarly to prokaryotic 70S ribosomes, eukaryotic 80S ribosomes also protect ∼30 nt of mRNA, but that 48S complexes bind an additional 10–20 nt of mRNA on its 5′‐side, likely through eIFs (Kozak, 1977; Lazarowitz and Robertson, 1977). These additional contacts outside the mRNA‐binding cleft could be important for ribosomal attachment to or scanning on mRNA. The ribosomal position of mRNA in initiation complexes upstream of the P‐site is particularly interesting also because eukaryotes lack the SD interaction. To gain insights into the architecture of eukaryotic initiation complexes, we characterized in detail the ribosomal position of mRNA and its specific interactions with eIFs in a stable assembly intermediate, the 48S initiation complex, and in the final assembly product, the 80S initiation complex, by site‐directed zero‐length UV crosslinking using a panel of mRNAs containing 4‐thiouridines (4SU) at unique positions from −26 to +11 relative to the initiation codon. The obtained data are compared with the path of mRNA on the prokaryotic 30S subunit.
Contacts of mRNA with rps and initiation factors in 48S and 80S initiation complexes
To define the path of mRNA on the 40S subunit and to identify specific interactions of mRNA with the components of initiation complexes assembled by scanning‐mediated initiation, a panel of mRNAs was employed that contained uridines at single unique positions from −26 to +11 relative to the A of the initiation codon in addition to the uridine (+2) in the initiation codon (Table I; Figure 1A). Initiation codons were flanked by ∼30‐nt‐long sequences comprising multiple CAA triplets to minimize secondary structure and increase initiation efficiency. mRNAs were transcribed in vitro in the presence of [α‐32P]ATP and 4SU, which can be specifically crosslinked to proteins and nucleic acids by low‐energy (360 nm) irradiation, yielding ‘zero‐length’ crosslinks (Favre et al, 1998) that represent direct contacts of the nucleotide with 48S complex constituents. As discussed elsewhere (Sergiev et al, 2001a, 2001b; Yusupova et al, 2001), interactions of mRNA with prokaryotic ribosomes revealed using this method correlate well with the crystal structures of the mRNA/70S ribosome complex. The presence of 4SU in the 5′‐UTR of an mRNA does not impair ribosomal scanning and the efficiency of 48S complex formation (Pisarev et al, 2006). 48S complexes were assembled from 40S subunits, eIFs 1, 1A, 3, 4A, 4B, 4F and eIF2·GTP·Met‐tRNAMeti. To form 80S initiation complexes, reaction mixtures were supplemented with 60S subunits, eIF5 and eIF5B. Initiation complexes were purified by sucrose density gradient centrifugation and irradiated. After crosslinking, ribosomal complexes were subjected to a second round of sucrose density gradient centrifugation, because a small proportion of them dissociate during the procedure, leading to nonspecific crosslinking of released mRNA with mRNA‐binding components of initiation complexes (e.g. eIF3). To identify crosslinked proteins, purified initiation complexes were treated with RNases and analysed by SDS–PAGE and two‐dimensional (2D) gel electrophoresis. This technique previously allowed us to identify specific interactions in 48S complexes of the nucleotide at position −3 of mRNA with eIF2α and rpS5 (S7p), and of the +4 nucleotide with rpS15 (S19p) (Pisarev et al, 2006, 2007).
The 4SU residue that was present in all mRNAs at position +2 in the initiation codon is base paired with the initiator tRNA anticodon and therefore should not crosslink to other components of initiation complexes. Consistently, no crosslinking with eIFs or rps was observed for control mRNA that contained 4SU only at this position (Figure 1B, lane 12), whereas crosslinking to mRNAs containing 4SU at additional positions yielded specific crosslinking patterns (Figures 1B and C; summarized in Figure 1G, compared to the contacts of mRNA nucleotides with ribosomal proteins in the mRNA/Thermus thermophilus 30S subunit in Figure 1H and mapped onto the crystal structure of the mRNA/T. thermophilus 30S subunit complex in Figure 5). In 48S complexes, nucleotides at positions from −17 to +11 crosslinked to several proteins with molecular weights ranging from ∼6 to ∼32 kDa (Figure 1B, lanes 2–11). The low molecular weights of crosslinked proteins and the fact that the same crosslinking pattern for each individual position was observed in 48S (Figure 1B) and in 80S initiation complexes (Figure 1C) indicated their ribosomal nature. They were attributed to distinct rps by 2D gel electrophoresis taking into consideration the ‘northwest’ shift caused by covalently bound mRNA nucleotides, whose identity was then confirmed by mass‐spectrometry sequencing of tryptic peptides (Supplementary Table 1). To simplify presentation of 2D gel analysis, we assembled 48S complexes on mRNA that simultaneously contained 4SU at positions −14, −8, −4, +5 and +11 (Figure 1D), because crosslinking to 4SU at these positions (Figure 1E) covered the whole spectrum of rps that crosslink to individual mRNA positions from −26 to +11 (see below). Therefore, instead of showing numerous 2D gels that correspond to each individual 4SU position, we have presented a summarizing 2D gel for crosslinking of mRNA containing 4SU at positions −14, −8, −4, +5 and +11 (Figure 1F), and will discuss the crosslinking of individual positions using one‐dimensional gels (Figures 1B and C) as illustrations.
Thus, on the A‐site side of the 40S subunit, mRNA crosslinked to rpS2 (the eukaryotic homologue of S5p), rpS3 (S3p) and rpS15 (S19p). rpS2 and rpS3 crosslinked most strongly to position +11 and much less so to positions +8–+10 (Figure 1B, lanes 10 and 11; Figure 1C, lanes 10–13). The crosslinking intensity of rpS2 (the upper of the two closely migrating bands in lanes 10–13 of Figures 1C and F) was substantially weaker than that of rpS3 (the lower of these bands; Figure 1C, lanes 10–13; Figure 1F). rpS15 crosslinked to position +5 (Figure 1B, lane 9; Figure 1C, lane 8; Figure 1F) and as previously reported, also to position +4 (Pisarev et al, 2006). On the E‐site side of the 40S subunit, mRNA crosslinked to rpS5 (S7p), rpS14 (S11p), rpS28 and rpS26. rpS5 crosslinked to positions from −3 to −5, with maximal efficiency to position −3 (Figure 1B, lane 8; Figure 1C, lanes 4–6; Figure 1F). rpS14 crosslinked to positions −6–−8, with the highest intensity to positions −6/−7 (the upper band in Figure 1C, lanes 1–3, and the upper of the two closely migrating bands in Figure 1B, lanes 6 and 7). rpS28 crosslinked to positions −4–−8, with the highest intensity to position −6 (the lowest band in Figure 1B, lanes 6–8 and Figure 1C, lanes 1–5). rpS26 crosslinked to positions −7–−14, with the highest intensity to positions −7–−10 (Figure 1B, lanes 2–7; Figure 1C, lanes 1 and 2). No rps crosslinked to positions −26, −2 or +7 (Figure 1B, lane 1; Figure 1C, lanes 7 and 9). Figure 1F is a summarizing 2D electrophoresis gel illustrating crosslinking in 48S complexes of all seven rps to mRNA simultaneously containing 4SU at positions −14, −8, −4, +5 and +11. Taking into account the ‘northwest’ shift of crosslinked proteins, it is clearly seen that this mRNA crosslinks to rpS2 (through position +11), rpS3 (through position +11), rpS5 (through position −4), rpS15 (through position +5), rpS14 (through position −8), rpS26 (through positions −8 and −14) and rpS28 (through positions −4 and −8). The radioactive spot that almost co‐migrated with rpS21 (bottom left corner in both panels of Figure 1F) most likely corresponds to crosslinked rpS28 bound to additional undigested nucleotide(s), thus accounting for the greater shift.
UV crosslinking of 48S complexes also revealed that in addition to the previously reported crosslinking of eIF2α to 4SU or 6SG at position −3 (Pisarev et al, 2006), mRNA on the E‐site side of the 40S subunit also interacts extensively with eIF3, yielding a distinctive pattern of radiolabelled eIF3 subunits that crosslink to positions −8–−17 (Figure 1B, lanes 2–6). The most intense crosslinking was between eIF3a and 4SU at position −14 (Figure 1B, lane 3). eIF3a also crosslinked to position −17 albeit less strongly (Figure 1B, lane 2). eIF3d crosslinked to 4SU at positions −8, −10, −11, −14 and −17 with very similar intensities (Figure 1B, lanes 2–6). Crosslinking of eIF3b to positions −14, −17 and −26 was not consistently significantly over the background (Figure 1B, lanes 1–3; data not shown), and we are therefore reluctant to suggest that crosslinking of eIF3b was specific. No labelling of other eIFs by crosslinking to mRNA was detected.
Contacts of mRNA with 18S rRNA in 48S and 80S initiation complexes
Sites of crosslinking in 18S rRNA of 4SU at different positions in mRNA were identified in two stages. First, the approximate region of crosslinking was determined by RNase H mapping (Dontsova et al, 1992). For this, 18S rRNA was extracted from irradiated 48S/80S initiation complexes, hybridized with DNA oligonucleotides complementary to different regions, incubated with RNase H and resolved by gel electrophoresis. Attribution of 32P‐labelled UV crosslinked fragments of 18S rRNA took into account their reduced mobility due to the covalently linked ∼70‐nt‐long mRNA. In the second stage, the exact crosslinked nucleotide in 18S rRNA was identified by primer extension inhibition using a primer chosen on the basis of RNase H mapping. The positions of crosslinked nucleotides in rabbit 18S rRNA are shown in Figure 3A, summarized in Figure 3B and mapped onto the crystal structure of the mRNA/T. thermophilus 30S subunit complex in Figure 6.
As in the case of rps, crosslinking of individual mRNA positions to 18S rRNA was identical in 48S and 80S initiation complexes. Therefore to simplify presentation, only gels that represent crosslinking in 48S complexes are shown (Figure 2). 4SU at position −17 crosslinked to nts 1085–1113 of 18S rRNA (Figure 2A, lane 3). Primer extension analysis identified the crosslinked nucleotide as U1107 of h26 (Figure 2B, lane 1). The control experiment (Figure 2B, lane 2) was done using 48S complexes assembled on mRNA containing 4SU only at position +2 in the initiation codon. Position −14 also crosslinked to nts 1085–1113 (data not shown) but so weakly that precise identification of the crosslinking site by primer extension was impossible, suggesting that position −14 mainly interacts with eIF3a (Figure 1B, lane 3). Positions −11–−8 crosslinked to the 3′‐terminal nts 1857–1863 of 18S rRNA (Figures 2C and D; data not shown). No substantial crosslinking with 18S rRNA was detected for position −7 (data not shown). Position −4 crosslinked to overlapping regions corresponding to nts 912–972 and 947–1052 (Figure 2E, lanes 2 and 3), suggesting that the crosslinking site lay within 26 nt between nts 947–972 of the apical part of h23, but the intensity of crosslinking was too low for unambiguous identification of the exact crosslinking site. Positions +7 and +8 crosslinked to nts 1676–1699 (Figure 2F, lanes 2 and 3); primer extension identified the crosslinked nucleotides as CUUUGUA1684−1690 and A1692 of h28 (Figure 2G, lanes 3 and 4). Crosslinking to CU1684−1685 was most efficient. Very weak crosslinking to these positions was also observed for the flanking +6 and +9 positions (Figure 2F, lanes 1 and 4; Figure 2G, lanes 2 and 5). Medium intensity crosslinking occurred between positions +8/+9 and C1484 of h34 (Figure 2G, lanes 2 and 3). 4SU at +9–+11 positions crosslinked to nts 489–662 (Figure 2H, lanes 2–4), and primer extension analysis identified the crosslinked nucleotides as U617 and A619 of h18 (Figure 2I, lanes 3–5). Position +10 crosslinked more efficiently than at positions +9 and +11, and crosslinking of A619 was more intense than that of U617.(Figures 3A and B)
In conclusion, these results indicate that on the E‐site side of the 40S subunit, mRNA is close to h26, the seven 3′‐proximal nucleotides and the apical region of h23, whereas on the A‐site side of the 40S subunit, mRNA contacts nucleotides in h28, h34 and h18, and as previously reported, in the base of h44 (Pisarev et al, 2006).
Interactions of eIF3 with the 40S subunit
The crosslinking pattern of eIF3 subunits with mRNA positions −8–−17 suggests that eIF3d and eIF3a likely interact with elements of the 40S subunit that form the mRNA‐binding channel. To gain additional insights into the ribosomal position of eIF3, we used the complementary approaches of chemical/enzymatic footprinting and directed hydroxyl radical cleavage of 18S rRNA in 40S/eIF3 complexes. eIF3 binds stably to 40S subunits in the presence of poly(U) (Kolupaeva et al, 2005). Assembled eIF3/poly(U)/40S subunit complexes were separated from unincorporated components by sucrose density gradient centrifugation and subjected to limited digestion with RNase T1 (which cleaves RNA after unpaired G residues) or RNase V1 (which cleaves double‐stranded RNA without base specificity), or treated with N‐cyclohexyl‐N′‐(2‐morpholinoethyl)‐carbodiimide methyl‐p‐toluene sulphonate (CMCT), which modifies unpaired uracil and to a lesser extent guanine residues. Consistent with eIF3's binding to the protein‐rich solvent side of the 40S subunit (Siridechadilok et al, 2005), eIF3 protected very few sites in 18S rRNA from enzymatic cleavage. The strongest protection was in h16. Thus, eIF3 protected G537 from RNase T1 cleavage (Figure 4A, lanes 1 and 2; Figure 4D) and the nearby C539 from RNase V1 cleavage (Figure 4B, lanes 2 and 3; Figure 4D). Protection at the same sites was observed in 43S complexes that contained 40S subunits, eIF3 and eIF2 ternary complex (data not shown). Interestingly, eIF3 did not protect UUU530−532 in the apical loop of h16 (Figure 4A, lanes 3 and 4), which suggests that eIF3 either interacts with the stem but not the apical loop of h16, or that it does not directly interact with h16 but binds closely enough to block access of RNases but not of the smaller CMCT molecule. However, it is also possible that protection from enzymatic cleavage resulted from conformational changes in 40S subunits induced by factor binding rather than from direct interaction between h16 and eIF3.
The footprinting approach was complemented by directed hydroxyl radical probing (Culver and Noller, 2000; Lomakin et al, 2003; Unbehaun et al, 2007), in which locally generated hydroxyl radicals cleave 18S rRNA in the vicinity of Fe(II) tethered to cysteine residues on the surface of eIF3 through the linker 1‐(p‐bromoacetamidobenzyl)‐EDTA (BABE). Native eIF3 contains 65 cysteine residues, some of which might be surface‐exposed and could therefore be conjugated with Fe(II)–BABE. Derivatization of eIF3 did not affect its activity in 48S complex formation (data not shown), and cleavage of 18S rRNA was therefore analysed in 40S/[Fe(II)–BABE]–eIF3/poly(U) and 43S complexes. Medium intensity cleavage occurred at GC537−538 of h16 (Figure 4C, lanes 1 and 2; Figure 4D; data not shown). This site coincides with sites of protection by eIF3 from RNase digestion (Figures 4A and B). Although protection from enzymatic cleavage could theoretically be the result of conformational changes in rRNA induced by factor binding, the fact that h16 was also cleaved by hydroxyl radicals strongly suggests that eIF3 is located at least near this region of the 40S subunit.
The results of UV crosslinking experiments presented here and elsewhere (Pisarev et al, 2006) have revealed striking parallels between the path of mRNA in eukaryotic initiation complexes and on the prokaryotic ribosome, as well as differences, some of which likely reflect unique aspects of eukaryotic initiation. We will compare the data obtained in the present study with the crystallographically visualized mRNA path on the T. thermophilus ribosome containing P‐ and E‐site tRNAs (Yusupova et al, 2001, 2006; Figures 5 and 6), and with previously reported crosslinking data for prokaryotic 70S ribosomal complexes obtained using the same zero‐length crosslinking technique as employed here and for eukaryotic 80S ribosomal complexes phased by cognate tRNA obtained using mid‐range nucleotide derivatives.
Ribosomal position and contacts of mRNA downstream of the P‐site
Efficient UV crosslinking of mRNA at position +11 to rpS2 and rpS3 (homologues of S5p and S3p, respectively) indicates that mRNA enters the mRNA‐binding channel of eukaryotic ribosomes through the same layer of proteins as in prokaryotic ribosomes.
UV crosslinking of positions +9–+11 to h18 (U617 and A619) and of +8/+9 to h34 (C1484) was consistent with the proximity of this region to the 530 loop of h18 and to U1196 of h34 in the T. thermophilus ribosome (Yusupova et al, 2001) and was directly analogous to UV crosslinking of positions +11/+12 to h18 (G530, A532) and of positions +8/+9 to h34 (A1196) in the Escherichia coli ribosome (Dontsova et al, 1992; Sergiev et al, 1997). This consonance suggests that mRNA nucleotides at positions +8–+11 are located very similarly with respect to the elements of h18 and h34 in prokaryotic ribosomes and in eukaryotic initiation complexes.
In eukaryotic initiation complexes, mRNA positions +7/+8, and particularly +7, also crosslinked efficiently to an extensive region of h28 (nts 1684–1690 and nt 1692) with the maximum at CU1684−1685. However, in prokaryotes, position +7 is close only to the equivalent of nt 1689 in the T. thermophilus ribosome (Yusupova et al, 2001) and consistently, crosslinked to the equivalent of nt 1691 in the E. coli ribosome (Dontsova et al, 1992), whereas 16S rRNA nucleotides equivalent to CU1684−1685 of 18S rRNA and mRNA position +7 are too far apart to be crosslinked efficiently (Yusupova et al, 2001). Interestingly, G926 of 16S rRNA (the equivalent of G1203, located on the upper strand of h28 opposite CU1684−1685; Figure 3A) is positioned to interact with the phosphate of the +1 nt of the P‐site codon in the T. thermophilus ribosome (Yusupova et al, 2001). It has therefore been suggested that interaction with G926 of the 3′‐tail of 16S rRNA, which results in mRNA mimicry (Carter et al, 2000), might be important for induction of the active conformation of the 30S subunit's P‐site responsible for mRNA‐independent recruitment of initiator tRNA (Yusupova et al, 2001). The observed crosslinking of mRNA positions +7/+8 to CU1684−1685 in eukaryotic initiation complexes suggests that G1203 would be too distant to interact with the +1 nt of the P‐site codon. Importantly, crosslinking between positions +7–+8 and h28 was no longer observed in eukaryotic 80S complexes that had undergone 2–3 elongation events (Pisarev et al, in preparation), which indicates that in elongation complexes, G1203 might be able to interact with the +1 nt of the P‐site codon. In fact, crosslinking of the +1 nt to the equivalent of rabbit G1203 was observed in phased human 80S ribosomal complexes (Demeshkina et al, 2000, 2003). The possibility that ribosomal contacts of the mRNA P‐site codon differ between initiation and elongation complexes might reflect the specific needs of the eukaryotic initiation codon selection mechanism. Crystallographic studies suggest that the absence of E‐site tRNA induces changes in the 30S subunit that include increased conformational flexibility of h28 (Jenner et al, 2005). The lack of E‐site tRNA in 48S/80S initiation complexes characterized here might thus account for the observed difference. However, although the presence/absence of E‐site tRNA seems to be a likely explanation, it cannot be ruled out that the presence of eIFs in 48S initiation complexes might also influence the conformation of 40S subunits, changing the position of h28 compared to its position in elongating ribosomes, and that such a change could even be preserved in 80S initiation complexes. Thus, it has recently been reported that binding of eIF1 and eIF1A to yeast 40S subunits induces opening of the entry channel ‘latch’ formed between h18 in the body and h34 and rpS3 in the neck, and establishment of a new connection between the head and the body likely mediated by h16 and rpS3 that also results in a slightly different conformation of the beak (Passmore et al, 2007).
The previously reported crosslinking of mRNA at position +4 in the A site codon to C1696 and AA1818−1819 in h44 of 18S rRNA (Demeshkina et al, 2000; Graifer et al, 2004; Pisarev et al, 2006) is also consistent with the proximity of this mRNA position to the equivalent region of 16S rRNA in T. thermophilus ribosomes (Yusupova et al, 2001) and with UV crosslinking of position +4 to the equivalent of C1698 in E. coli ribosomes (Rinke‐Appel et al, 1993). We have suggested that interaction of the +4 nucleotide with AA1818−1819 likely accounts for the +4 nucleotide context rule (Pisarev et al, 2006). Although S12p lies directly under the +5/+6 positions of mRNA (Yusupova et al, 2001) and its eukaryotic homologue rpS23 has a similar location (Spahn et al, 2001a), and although mutations in both S12p and rpS23 affect translational fidelity (Alksne et al, 1993; Anthony and Liebman, 1995; Synetos et al, 1996), neither protein crosslinks to mRNA at this position. However, in contrast to prokaryotic ribosomal complexes, in 48S/80S initiation complexes, mRNA at positions +4/+5 crosslinked efficiently to rpS15, a homologue of S19p (this study; Pisarev et al, 2006). S19p is located in the head of the 30S subunit with its C‐terminal tail pointing towards the interface side but not reaching the A‐site codon (Wimberly et al, 2000). The fact that analogous crosslinking of positions +4–+6 to rpS15 was also observed in 80S complexes phased by cognate P‐site tRNA (Graifer et al, 2004; Bulygin et al, 2005) suggests that crosslinking of rpS15 in initiation complexes might not be the result solely of initiation‐specific conformational changes in 40S subunits (Passmore et al, 2007), but is likely due to N‐ or C‐terminal extensions in rpS15 relative to S19p, whose functional significance is presently unknown.
Taken together, our crosslinking results indicate that the overall ribosomal position of the mRNA downstream of the P‐site codon in eukaryotic 48S/80S initiation complexes is similar to that of mRNA in prokaryotic 70S ribosomes. Differences include the close proximity of mRNA positions +7/+8 to the central portion of the lower strand of h28 and of positions +4/+5 to rpS15.
Ribosomal position and contacts of mRNA upstream of the P‐site
The position of mRNA in eukaryotic initiation complexes upstream of the P‐site is particularly interesting because eukaryotes lack the SD interaction. In eukaryotic 48S/80S initiation complexes, mRNA positions −3/−4 and −6/−7 crosslinked efficiently to rpS5 (the homologue of S7p) and rpS14 (the homologue of S11p), respectively, and position −4 also crosslinked to the apex of h23b (equivalent to the 690 loop of 16S rRNA). These results are consistent with the proximity of this region of mRNA to S7p, S11p and the 690 loop of h23 in T. thermophilus ribosomal complexes (Yusupova et al, 2001) and with crosslinking of positions −2/−3 to the equivalent of rabbit G957 in the apex loop of H23b in phased human 80S ribosomal complexes (Demeshkina et al, 2000, 2003). Some of the other mRNA interactions on the E‐site side of the 40S subunit are also comparable to interactions in prokaryotes. Thus, −8–−11 mRNA residues crosslinked to the 3′‐terminal nts 1857–1863 of 18S rRNA, indicating that they occupy a position in eukaryotic initiation complexes analogous to the mRNA and rRNA elements of the SD duplex in prokaryotic initiation complexes despite the lack of base‐pairing potential (Yusupova et al, 2001).
However, other upstream interactions are unique to eukaryotes. Thus, in 48S initiation complexes, mRNA position −3 also specifically interacts with eIF2α (Pisarev et al, 2006), suggesting that in contrast to prokaryotic initiation complexes, the E‐site in eukaryotic 48S complexes is partially occupied by an initiation factor, eIF2. This interaction is important for 48S complex formation in the presence of eIF1, and accounts for the −3 nucleotide context rule (Pisarev et al, 2006). Positions −6 and −7/−10 also efficiently crosslinked to eukaryote‐specific rpS28 and rpS26, respectively. Crosslinking of S26 to positions −4–−9 was previously observed in phased human 80S ribosomal complexes with RNAs containing mid‐range nucleotide derivatives (Graifer et al, 2004). The identification of rpS26 and rpS28 in contact with mRNA in the immediate vicinity of rpS14 is consistent with observations from cryo‐EM analysis of yeast 40S subunits that additional protein density at the top of the platform surrounds rpS14, leading to a stronger contact of the platform with rpS5 in the head of the 40S subunit, which constitutes part of the mRNA exit channel (Spahn et al, 2001a).
Moreover, UV crosslinking revealed specific interactions between mRNA and two subunits of eIF3 in 48S complexes: the most intense crosslinking occurred between eIF3a and mRNA position −14, whereas eIF3d crosslinked to positions −8–−17 with similar medium intensities. These data are consistent with previous reports that eIF3 can be crosslinked to β‐globin mRNA in 48S complexes, that its 3d subunit is close to 18S rRNA and that binding of eIF3 to the 40S subunit can be stabilized by mRNA (Nygård and Westermann, 1982; Westermann and Nygård, 1984; Unbehaun et al, 2004; Kolupaeva et al, 2005). They also suggest that eIF3d contributes to formation of the mRNA‐binding channel on the 40S subunit itself (as positions −8–−10 crosslink simultaneously to eIF3d and rps), and that eIF3a likely forms a direct extension of it, which together might contribute to the processivity of scanning and stabilize 48S complexes after hydrolysis of eIF2‐bound GTP but prior to subunit joining (Unbehaun et al, 2004). Interaction between mRNA and eIF3 in 48S complexes most likely also accounts for observations that whereas eukaryotic 80S initiation complexes protect ∼30 nt flanking the initiation codon from RNAse cleavage, 48S complexes protect an additional 10–18 upstream nucleotides (e.g. Kozak, 1977; Lazarowitz and Robertson, 1977). The suggested eIF3/mRNA interaction upstream of the E‐site is in good agreement with models derived from cryo‐EM reconstructions that located the five‐lobed eIF3 on the solvent side of the 40S subunit, with its left arm pointing towards the E‐site (Siridechadilok et al, 2005). Our crosslinking data therefore suggest that eIF3d and eIF3a likely contribute to formation of the left arm of eIF3. The specificity of eIF3/mRNA crosslinking is also consistent with the finding that the N‐terminal domain of eIF3a interacts with RPS0, placing it in the vicinity of the mRNA exit channel on the solvent side of the 40S subunit (Valásek et al, 2003), and with the crosslinking reported here of mRNA position −17 to U1107 in the eukaryote‐specific expansion segment ES7 of h26 on the back of the platform (Spahn et al, 2001a).
Enzymatic footprinting and hydroxyl radical cleavage experiments indicate that eIF3 protects G537 and C539 in h16 from RNase T1 and RNase V1 digestion, respectively, and that it also induces cleavage by hydroxyl radicals at GC537−538. Although hydroxyl radical cleavage indicates that eIF3 is indeed located in the vicinity of h16, it is not possible to conclude unambiguously whether protection from RNase cleavage is caused by direct contact of eIF3 with h16 or if it results from conformational changes in the 40S subunit caused by eIF3 binding. Moreover, whereas the eIF3/40S model derived from cryo‐EM reconstructions does not suggest that eIF3 interacts with the shoulder (Siridechadilok et al, 2005), according to another report, the C‐terminal domain of yeast eIF3a (TIF32) can specifically bind to a fragment of 18S rRNA encompassing h16–h18 (Valásek et al, 2003). In contrast to bacterial 30S subunits, in which h16 is folded towards h18, h16 in eukaryotic 40S subunits is rotated towards the back of the 40S subunit and points into the solvent (Spahn et al, 2001a, 2001b; Passmore et al, 2007). Binding of eIF1/eIF1A, the HCV IRES and the CrPV IRES to the 40S subunit all result in opening of the entry ‘latch’ between h18 in the body and h34/rpS5 in the neck and concomitant establishment of a new connection between h16 and rpS3, which suggests that these ribosomal conformational changes constitute general changes that occur during initiation (Spahn et al, 2001b, 2004; Passmore et al, 2007). It is therefore likely that binding of eIF3 to 40S subunits could also result in similar changes, which might account for protection of h16 from RNase digestion.
Taken together, our crosslinking results indicate that the overall ribosomal position of mRNA upstream of the P‐site codon in eukaryotic 48S/80S initiation complexes is similar to that of mRNA in prokaryotic 70S ribosomes. Differences in mRNA contacts include interactions of eukaryotic mRNA with eIF2 in the E‐site, with eukaryote‐specific rpS26 and rpS28 on the platform and with eIF3 on the back of the platform. The interaction of mRNA at position −3 with eIF2α is essential for correct initiation codon selection (Pisarev et al, 2006). Although it is very likely that other eukaryote‐specific contacts of mRNA with 40S subunit and eIF3 that we detected in 48S complexes are employed to accommodate the eukaryote‐specific initiation stages of ribosomal attachment to and scanning on mRNA, we cannot exclude that the contacts between mRNA and initiation complex constituents during attachment and scanning might differ slightly from those in the assembled 48S complexes that are reported here. We did not detect any crosslinking of mRNA with eIFs 4A/4B/4F. Although it is possible that these factors had dissociated from initiation complexes during sucrose density gradient purification, it cannot be strictly excluded that they instead interact with mRNA regions outside those investigated here or that their interactions with mRNA differ during initial attachment of 43S complexes, scanning and formation of 48S complexes.
Materials and methods
UV crosslinking experiments
48S/80S initiation complexes were assembled and purified by sucrose density gradient centrifugation essentially as described (Pisarev et al, 2006). Thus, to form 48S complexes, 100 ng of 32P‐labelled 4‐thioU‐containing mRNAs were incubated with 8 pmol 40S subunits, 10 pmol Met‐tRNAiMet, 5 μg eIF2, 15 μg eIF3, 2.5 μg eIF4A, 0.5 μg eIF4B, 2.5 μg eIF4F, 0.2 μg eIF1A, 0.2 μg eIF1 in 100 μl buffer A (20 mM Tris pH 7.5, 100 mM KAc, 2 mM DTT, 2.5 mM MgAc2, 0.25 mM spermidine) supplemented with 1 mM ATP and 0.4 mM GTP for 10 min at 37°C. To obtain 80S initiation complexes, 48S complexes were further incubated with 10 pmol 60S subunits, 5 μg eIF5 and 5 μg eIF5B for 10 min at 37°C. Assembled ribosomal complexes were purified by centrifugation through 10–30% sucrose density gradients prepared in buffer A in a Beckman SW55 rotor at 53 000 r.p.m. for 75 min. The presence of [32P]mRNA in gradient fractions was monitored by Cherenkov counting. Equal amounts of counts (∼200 000 c.p.m.) of peak fractions were irradiated at 360 nm for 30 min on ice using a UV‐Stratalinker (Stratagene). After UV irradiation, 48S complexes were subjected to a second round of sucrose density gradient centrifugation.
To analyse crosslinked proteins by one‐dimensional electrophoresis, crosslinked ribosomal fractions were treated with RNase A and subjected to electrophoresis in NuPAGE 4–12% Bis‐Tris‐Gel (Invitrogen) followed by autoradiography. UV crosslinked rps were further identified by acidic‐SDS 2D gel electrophoresis exactly as described (Madjar et al, 1979; Pisarev et al, 2006). The identity of rps was confirmed by LC‐nanospray tandem mass spectrometry of peptides derived by in‐gel tryptic digestion at an in‐house facility.
Identification of crosslinked nucleotides in 18S rRNA was done exactly as described (Pisarev et al, 2006). Regions of 18S rRNA crosslinked to 32P‐labelled mRNA were first identified by RNase H digestion of 18S rRNA hybridized with a panel of ∼20‐mer DNA oligonucleotides complementary to different regions of 18S rRNA using a previously described strategy (Dontsova et al, 1992). 18S rRNA fragments were separated by electrophoresis in 12% denaturing gel. Crosslinked fragments were attributed to corresponding regions of 18S rRNA taking into account the reduced mobility of crosslinked rRNA fragments due to the covalently bound mRNA. Precise identification of crosslinked nucleotides in 18S rRNA was done by primer extension inhibition using AMV RT and primers chosen on the basis of RNase H digestion.
Chemical and enzymatic footprinting analysis of eIF3/40S subunit complexes
Binary eIF3/40S complexes were assembled by incubating 30 pmol 40S subunits, 50 pmol eIF3, 3 μg poly(U) RNA in 100 μl buffer A for 10 min at 37°C (Kolupaeva et al, 2005), purified by sucrose density gradient centrifugation and then either enzymatically digested by incubation with RNase V1 (final concentration 0.9 U/ml) for 15 min at 37°C or with RNase T1 (final concentration 0.5 U/ml) for 11 min at 37°C or modified by incubation with CMCT. Cleavage/modification sites in 18S rRNA were identified by primer extension using AMV‐RT.
Directed hydroxyl radical cleavage of 18S rRNA in [Fe(II)–BABE]–eIF3/40S subunit complexes
eIF3 was derivatized with Fe(II)–BABE as described (Culver and Noller, 2000; Lomakin et al, 2003) by incubating 300 pmol eIF3 with 1 mM Fe(II)–BABE in buffer B (80 mM HEPES pH 7.5, 100 mM KCl, 5% glycerol, 2.5 mM MgCl2) for 20 min at 37°C. Fe(II)–BABE‐derivatized eIF3 was separated from unincorporated Fe(II)–BABE on YM‐30 microcons.
Binary [Fe(II)–BABE]–eIF3/40S subunit complexes were assembled by incubating 40 pmol 40S subunits and 50 pmol Fe(II)–BABE‐derivatized eIF3 with 5 μg poly(U) RNA for 10 min at 37°C in 200 μl in buffer B. Complexes were isolated by centrifugation in a Beckman SW55 rotor for 2 h at 4°C and 42 000 r.p.m. in 10–30% linear sucrose density gradients in buffer B adjusted to 4 mM MgCl2. Gradient fractions that contained 40S subunits were analysed for the presence of eIF3 by SDS–PAGE and Coomassie staining (Kolupaeva et al, 2005) and used for directed hydroxyl radical probing. To generate hydroxyl radicals, the reaction mixture was supplemented with 0.025% H2O2 and 5 mM ascorbic acid and incubated on ice for 10 min. Reactions were quenched by addition of 10 mM thiourea. 18S rRNA was analysed by primer extension.
Construction of plasmids and methods for purification of translation components are described in Supplementary data.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
We thank J‐P Bachellerie and A Borovjagin for their generous gifts of 18S rRNA plasmids and A Marintchev for helpful discussion. This study was supported by NIH Grant GM59660 to TVP and HFSP Grant RPG0055/2006‐C to TVP and MMY.
- Copyright © 2008 European Molecular Biology Organization