Structure of 18 S ribosomal RNA in native 40 S ribosomal subunits.

We have analyzed the structure of 18 S rRNA in native 40 S subunits using chemical modification followed by primer extension. The native subunits were modified using the single-stranded specific reagents dimethyl sulfate and 1-cyclohexyl-3-(2-morpholinoethyl)carbodiimide metho-p-toluenesulfonate. The modification pattern of the 18 S rRNA was compared to that obtained from derived 40 S subunits prepared by dissociation of unprogrammed 80 S ribosomes. Eighteen nucleotides showed different accessibility to the chemical probes in derived and native subunits. Half of these nucleotides were found in the central domain of the rRNA between the 1060 loop and the central pseudoknot. The remaining nucleotides were located in two clusters in the 5′- and 3′-domains of the 18 S rRNA. Derived 40 S subunits are free from non-ribosomal proteins. In contrast, native subunits are intermediates in protein synthesis initiation and contain stoichiometric amounts of initiation factor 3 (Sundkvist, I. C., and Staehelin, T. (1975) J. Mol. Biol. 99, 401-418). The possible role of this factor in altering the structure of 18 S rRNA in the native 40 S subunits is discussed.

Protein synthesis initiation in eukaryotes involves a number of different initiation factors (reviewed in Ref. 1). Initiation factor eIF-3 1 binds to 40 S subunits and prevents formation of unprogrammed 80 S ribosomes by inhibiting association of the 40 S and 60 S ribosomal subunits in the absence of mRNA. Initiation factor eIF-2 selects the specific initiator tRNA (Met-tRNA f ) and brings it to the 40 S subunit. The resulting 43 S pre-initiation complex binds mRNA with the help of a series of initiation factors. The 60 S subunit now joins the mRNA containing 48 S initiation complex in a reaction that requires an additional initiation factor (eIF-5) and is associated with the hydrolysis of GTP.
Several of the initiation factors are found to be associated with the so-called native 40 S ribosomal subunits (40 S N ) in vivo (2). Most of these factors are present on the 40 S N particles in small quantities, but eIF-3 is present in stoichiometric amounts (2). Initiation factor 3 is a huge multisubunit protein with a total mass of approximately 0.7 MDa (3). The factor displays RNA binding properties, and one of its subunits can be cross-linked to 18 S rRNA in the 40 S⅐eIF-3 complex (4). This suggests that rRNA may, at least in part, be responsible for binding the factor to the small ribosomal subunit. However, the location of the eIF-3 interaction site in 18 S rRNA is not known.
The ribosomal RNA is considered to be involved in various ribosomal functions such as A-and P-site-related activities and peptide bond formation (for a review see Ref. 5). In prokaryotes the rRNA is directly involved in the binding of initiation factors and mRNA during protein synthesis initiation (6 -9). Less is known about the functional role of rRNA in the eukaryotic ribosome, but studies using chemical cross-linking and chemical and enzymatic footprinting have indicated that the rRNA is involved in mRNA binding, subunit interaction, and binding of elongation factors (10 -12).
We have previously studied the structure of 18 S rRNA in derived 40 S subunits prepared by dissociation of isolated 80 S ribosomes (11,13). In contrast to the native subunits, derived particles are free from additional non-ribosomal proteins. In this report, we have compared the structures of 18 S rRNA in native and derived 40 S subunits using chemical modification. The two types of 18 S rRNAs showed distinct but limited structural differences. The role of the non-ribosomal proteins in altering the structure of the 18 S rRNA in the 40 S N particles is discussed.
Modification of rRNA-Chemical modification of the 18 S rRNA in derived or native 40 S subunits was performed as described previously using the single strand specific reagents DMS and CMCT (13). DMS modifies single strand adenines and cytosines, whereas CMCT modifies unpaired uridines and guanines (16). CMCT also modifies single strand cytosines at the pH used here (13). The ribosomal subunits were incubated for 5 min at 37°C in Buffer A containing 5 mM MgCl 2 and modifying reagent (DMS or CMCT) as indicated. Control samples were treated identically with the exception that no modifying reagent was added.
RNA Extraction-The 18 S rRNA was extracted from the derived and native 40 S subunits using phenol (17). The extracted RNA was precip-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Identification of Modified Sites-End labeling of cDNA primers, primer extension, RNA sequencing, and gel electrophoresis were as described previously (18). The gels were exposed to x-ray films, and the autoradiograms were analyzed using a microcomputer-assisted image analysis system (18).

RESULTS
We have used the two single strand specific reagents DMS and CMCT to study the structure of 18 S rRNA in native 40 S subunits. The footprinting pattern generated by the two reagents was compared to that of derived 40 S subunits prepared from isolated 80 S ribosomes. The native subunits are intermediates in protein synthesis initiation and contain additional non-ribosomal proteins (2). Due to these proteins, the 40 S N particles are unable to associate with 60 S subunits in the absence of proper initiation. The derived subunits lack the additional non-ribosomal proteins and can spontaneously form unprogrammed 80 S ribosomes in the presence of 60 S subunits. The 40 S D particles also have a tendency to dimerize. However, no such dimerization was found under the conditions used in these experiments (not shown).
A comparison of the footprinting patterns obtained from the derived and native 40 S subunits showed that the reactivity of some of the bases in 18 S rRNA was different in the two types of particles ( Figs. 1 and 2). The affected nucleotides were concentrated in three regions. The first region, located in the 5Ј-domain, contained 3 bases that showed reduced accessibility to chemical modification in the native subunits. One of the bases, U 44 , was found in the interhelical sequence connecting helices 4 and 5. This was the only nucleotide in this part of the rRNA that was accessible to chemical modification in the derived 40 S subunits. The remaining 2 affected bases (U 63 and U 76 ) were located in hairpin 6. The latter base was only moderately exposed to chemical modification in the 40 S D particles.
The second affected region was found in the central domain of the 18 S rRNA. Hairpin 25 contained four nucleotides that showed altered reactivity in the 40 S N particles. Two of the nucleotides, A 1044 and A 1060 , became less exposed to chemical modification in the native particles. A 1044 is involved in a canonic Watson-Crick base pair in the helical stem, while A 1060 is found in the apical loop of hairpin 25. The apical loop also contained 2 bases, A 1064 and A 1065 , that showed increased reactivity in the 40 S N particles. These were the only nucleotides in the 18 S rRNA that were more accessible to chemical modification in native than in derived subunits. The adjacent helix 27 contained 2 bases, U 1117 and C 1134 , that were protected against modification in the 40 S N particles. These bases were located in the apical and internal loops of the helix, respectively. The internal loop also contained 1 base, G 1133 , that served as a natural stop for the reverse transcriptase in the derived subunits. This natural stop was almost absent in 18 S rRNA from the native subunits, indicating that this site was only available for limited nucleolytic attack in the derived 40 S subunits.
The hinge region between the three domains in 18 S rRNA contained 2 bases U 1195 and A 1198 that were very exposed in 40 S D subunits but became almost completely inaccessible to modification in the 40 S N particles. One additional protected base, C 1182 , was found in the apical loop of the adjacent hairpin 29. The latter base was less exposed to chemical modification in the 40 S D particles than the previous 2 bases. However, this base was also almost completely protected against modification in the native particles.
Our structural analysis also covered the whole 3Ј-domain with the exception of the 39 bases located at the 3Ј-end of the 18 S rRNA. The analysis showed that the domain contained a cluster of 6 bases that were less accessible for chemical modification in 40 S N particles than in derived subunits. The affected bases (C 1554 , A 1557 , C 1563 , C 1569 , C 1573 , and C 1575 ) were located in hairpin 44, where 3 of these bases (A 1557 , C 1563 , and C 1573 ) were involved in putative Watson-Crick base pairs. Most of the affected bases were only moderately accessible for chemical modification in the 40 S D particles. DISCUSSION We have analyzed and compared the structure of 18 S rRNA in derived and native 40 S ribosomal subunits. Eighteen of the nucleotides in 18 S rRNA were found to react differently to the chemical probes CMCT and DMS in the two types of ribosomal subunits. As the 40 S N particles contain additional non-ribosomal proteins it seems reasonable to assume that these additional proteins cause the structural differences seen in the modification pattern of the 18 S rRNAs obtained from the derived and native subunits. The effect could be due to a direct interaction of these proteins with the rRNA or caused by indi-rect structural rearrangements induced by the extra proteins. The footprinting technique cannot distinguish between these two possibilities. The increased exposure of A 1064 and C 1065 must clearly depend on structural alterations in the rRNA, but the cause of the protections is less obvious.
Data based on protein synthesis experiments show that the 40 S N particles contain initiation factors eIF-3, eIF-2, eIF-4A, eIF-4B, and eIF-5 (2). Most of these factors are present in less than stoichiometric amounts on the 40 S N particle, but eIF-3 is present in a close to 1:1 complex with the 40 S subunit. Furthermore, ribosome-bound eIF-3 can be cross-linked to 18 S rRNA (4). Thus, it seems likely that the structural differences seen between 18 S rRNA in 40 S D and 40 S N are caused by the presence of eIF-3 on the native particles.
The interaction between IF-3 (the prokaryotic homolog to eIF-3 (19)) with 16 S rRNA in the 30 S subunit has been studied by site directed mutagenesis, footprinting, and direct crosslinking (7)(8)(9)20). These techniques show the importance of the central domain for ribosomal binding of IF-3. Ribosome-bound IF-3 alters the accessibility of nucleotides in hairpins 23, 24, and 25 (mouse numbering) from modification by CMCT, kethoxal, and RNase V1 (8,9). The involvement of the latter loop in IF-3 binding was also shown by mutagenesis (20). Furthermore, nucleotides in the adjacent region between helices 25 and 20 have been directly cross-linked to IF-3, and the ribosomebound factor increases the susceptibility of the phosphodiester bond in this region for attack by RNase V1 (7, 8). Interestingly, the footprinting pattern of the central domain differed in derived and native 40 S subunits. Nucleotides with altered reac- tivities were found in hairpin 25 and in the sequence between the helices 25 and 20. However, no footprints were detected in helices 23 and 24 from 40 S N particles. Instead, marked differences in the footprinting pattern between derived and native subunits were seen in the sequence preceding the central pseudoknot. Although the homologous region in prokaryotes has not been cross-linked or footprinted by components of the initiation machinery, the central pseudoknot region seems to be linked to the initiation process. The pseudoknot undergoes conformational changes during the transition from inactive to active 30 S subunits (21), a transition that can be induced by initiation factors (22,23). Mutations that disturb the central pseudoknot prevent polysome formation presumably by interfering with the initiation process (24,25). Two bases in the apical loop of hairpin 25 (A 1064 and C 1065 ) showed increased exposure to chemical modification in the native subunits. The homologous sites in 16 S rRNA interact with P-site-bound tRNA (26). Thus, it is possible that this site is open for interaction with the initiator tRNA in the native particles.
Half of the nucleotides that displayed different chemical reactivity in native and derived subunits were found outside of the central domain in regions that have not been linked to any initiation-dependent ribosomal function. Could these nucleotides be protected from chemical modification by ribosomebound eIF-3? eIF-3 is considerably larger than IF-3, and although the two proteins are homologs, they have similar but not identical functions in protein synthesis (19). The difference in size and function may suggest that the two proteins do not bind to the respective ribosomes at completely identical sites. One other explanation for the additional sites could be the differences in salt concentration used in the various footprinting studies. Here we have used a salt concentration optimal for in vitro protein synthesis (3,27) to avoid salt-induced destabilization of the native 40 S subunits during the modification experiments. Thus, the Mg 2ϩ concentration used here is considerably lower than that used during the footprinting of IF-3 on the 16 S rRNA (8,9). Variations in the Mg 2ϩ concentration affect the accessibility of the nucleotides in 18 S rRNA for chemical modification. Such effects are seen between helices 4 and 7. 2 It is of course also possible, although less likely, that the additional reactivity changes were caused by the non-ribosomal proteins present in substoichiometric amounts on the 40 S N particles.
Native subunits are prevented from premature association with the 60 S subunit by eIF-3 (1). Emanuilov et al. (28) found that the binding site of eIF-3 on 40 S N subunits was partly overlapping the interface region of the particle, suggesting that the ribosome-bound factor directly interferes with subunit joining. However, others have found that eIF-3 binds to the 40 S subunit without interfering with the interface (29, 30) (Fig.  3C). Cross-linking experiments have shown that eIF-3 can be cross-linked to a number of ribosomal proteins depending on the length of the reagent used (31,32). Some of these proteins are clearly interface proteins (14,33), suggesting that part of the binding site for eIF-3 overlaps the subunit interface.
Where are the affected rRNA sites located on the native 40 S subunit? Unfortunately, nothing is known about the folding of 18 S rRNA in the small ribosomal subunit. However, Brimacombe (34,35) has suggested a model for the three-dimensional folding of 16 S rRNA in the prokaryotic 30 S subunit. The general structures of the 16 S-like rRNAs and the basic topology of the 30 S and 40 S subunits are similar (36 -38). Thus, it seems reasonable to assume that the homologous helices have the same location in the 30 S and 40 S particles. If so, the affected sites in the central domain would be located at the protuberance side of the 40 S subunit close to the rRNA sites affected by subunit-subunit interaction (11), while the affected sites in the 5Ј-and 3Ј-domains would be positioned in the middle of the body and in the head, respectively. The co-localization of rRNA structures involved in subunit-subunit interaction and in the binding of eIF-3 to the protuberance indicates that structural alterations in this region of the rRNA could be involved in preventing premature association of the 40 S N particle with the large ribosomal subunit.