Modifications in Thermus thermophilus 23 S Ribosomal RNA Are Centered in Regions of RNA-RNA Contact*

Ribosomal RNA from all organisms contains post-transcriptionally modified nucleotides whose function is far from clear. To gain insight into the molecular interactions of modified nucleotides, we investigated the modification status of Thermus thermophilus 5 S and 23 S ribosomal RNA by mass spectrometry and chemical derivatization/primer extension. A total of eleven modified nucleotides was found in 23 S rRNA, of which eight were singly methylated nucleotides and three were pseudouridines. These modified nucleotides were mapped into the published three-dimensional ribosome structure. Seven of the modified nucleotides located to domain IV, and four modified nucleotides located to domain V of the 23 S rRNA. All posttranscriptionally modified nucleotides map in the center of the ribosome, and none of them are in contact with ribosomal proteins. All except one of the modified nucleotides were found in secondary structure elements of the 23 S ribosomal RNA that contact either 16 S ribosomal RNA or transfer RNA, with five of these nucleotides physically involved in intermolecular RNA-RNA bridges. These findings strongly suggest that the post-transcriptional modifications play a role in modulating intermolecular RNA-RNA contacts, which is the first suggestion on a specific function of endogenous ribosomal RNA modifications.

Ribosomal RNA from all organisms contains post-transcriptionally modified nucleotides whose function is far from clear. To gain insight into the molecular interactions of modified nucleotides, we investigated the modification status of Thermus thermophilus 5 S and 23 S ribosomal RNA by mass spectrometry and chemical derivatization/primer extension. A total of eleven modified nucleotides was found in 23 S rRNA, of which eight were singly methylated nucleotides and three were pseudouridines. These modified nucleotides were mapped into the published three-dimensional ribosome structure. Seven of the modified nucleotides located to domain IV, and four modified nucleotides located to domain V of the 23 S rRNA. All posttranscriptionally modified nucleotides map in the center of the ribosome, and none of them are in contact with ribosomal proteins. All except one of the modified nucleotides were found in secondary structure elements of the 23 S ribosomal RNA that contact either 16 S ribosomal RNA or transfer RNA, with five of these nucleotides physically involved in intermolecular RNA-RNA bridges. These findings strongly suggest that the posttranscriptional modifications play a role in modulating intermolecular RNA-RNA contacts, which is the first suggestion on a specific function of endogenous ribosomal RNA modifications.
All cellular protein synthesis is performed by ribosomes, which are large ribonucleoprotein particles. The prokaryote ribosome consists of two stable and separable entities, a 50 S and a 30 S subunit. The 50 S subunit contains two rRNAs of ϳ3000 and 120 nucleotides (23 S and 5 S rRNA, respectively) and around 35 proteins, whereas the 30 S subunit contains 16 S rRNA of ϳ1600 nucleotides and around 20 proteins; the exact numbers vary with the species. Eukaryotic ribosomes are larger, but structural features are remarkably conserved between the domains of life.
rRNAs are post-transcriptionally modified at specific nucleotides, but the number of modified nucleotides varies greatly. The large ribosomal RNA in mitochondria contains just a few modified nucleotides (1), Escherichia coli 23 S rRNA has 25 (2,3), whereas over 100 are found in vertebrate cytoplasmic 28 S rRNA (4). The function of post-transcriptional rRNA modifications is far from clear, although they have been implicated in various processes. Specific rRNA methylation is used by numerous antibiotics-producing microorganisms as a means of autoprotection (see e.g. Refs. 5 and 6), but only a small fraction of post-transcriptional modifications can be assigned to this well defined function. E. coli 23 S rRNA modifications cluster in functionally principal parts of the ribosome such as the peptidyl transferase center and inter-subunit bridges when modeled into the structure of the Haloarcula marismortui large ribosomal subunit (7). The modified nucleotides in the central domains of 23 S rRNA from H. marismortui itself are all located in regions of intra-or intermolecular RNA-RNA contact (8) suggesting structure stabilization.
The post-transcriptionally modified rRNAs are essential for assembly of functional E. coli ribosomal subunits (9,10). This is in contrast to Bacillus stearothermophilus, where functional ribosomes can be assembled with in vitro transcribed 23 S rRNA, albeit less efficiently than with authentic 23 S rRNA (11). Inactivation of the modifying enzymes in E. coli has been reported to impair protein synthesis and cell growth (12)(13)(14), but recent data suggest that these effects may in some cases be ascribed to (unknown) secondary functions of these enzymes, not to the rRNA modification itself (15,16). A ribose methylation of G2270 (E. coli numbering 2251) is required for assembly of the large and small ribosomal subunits in yeast mitochondria (1), and various 25 S rRNA pseudouridines (⌿) 3 are required for optimal translation in yeast (17). On the other hand, systematic elimination of single ribose methylations in yeast rRNA does not affect cell growth (18). In summary, there is no coherent picture of the significance and function of post-transcriptional rRNA modifications.
Because it is now well established that the 23 S rRNA constitutes the peptidyl transferase activity with numerous modified nucleotides but no proteins located in the active site (19,20), the spatial localization and molecular interactions of modified * This work was supported by the Danish Biotechnology Instrument Centre and the Danish Natural Science Research Council. 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. rRNA nucleotides become increasingly important to analyze. The availability of moderate to high resolution x-ray crystallography structures of ribosomes and ribosomal subunits from various prokaryotes (19 -23) yields a unique opportunity to view post-transcriptional rRNA modifications in their threedimensional context and, thereby, gain insight on their function. We searched for modified nucleotides in the 23 S rRNA from Thermus thermophilus, since the three-dimensional structure of the entire ribosome was first published for this organism (19). Hence, all modified rRNA nucleotides can be viewed in their three-dimensional surroundings. Most of the modifications were found directly by mass spectrometry (MS), but the mass-silent modification pseudouridine was located by chemical modification with a carbodiimide reagent followed by reverse transcriptase primer extension (24). A total of 11 modified nucleotides were found and were fitted into the threedimensional structure of the T. thermophilus ribosome (19). The function of the modified nucleotides is discussed in light of their placement.

EXPERIMENTAL PROCEDURES
Isolation of rRNA-T. thermophilus (strain HB8) was grown as batch culture in a 3-liter BTS-05 bioreactor (Applicon, B.V., Schiedam, the Netherlands) at 72°C and pH-controlled at 7.5. The growth medium contained 4 g/liter yeast extract, 8 g/liter polypeptone, and 2 g/liter NaCl with maize oil as antifoam. Ribosomal subunits from 30-g dry weight of harvested cells were prepared via sucrose gradients as described (25). The fractions containing the large ribosomal subunit were pooled, and the RNA-protein complex was precipitated with 2.5 volumes of ethanol. The RNA was purified by three rounds of extraction with a 1:1-mixture of phenol/chloroform and recovered by ethanol precipitation. 5 S rRNA was separated from 23 S rRNA by ultrafiltration through a Microcon YM-100 membrane filter (Millipore) as described by the manufacturer. The flow-through contained the 5 S rRNA. 23 S rRNA was isolated from 5 S rRNA by gel filtration through a Microspin S400 HR column (Amersham Biosciences) as recommended by the manufacturer.
Details on the site-directed mung bean nuclease protection for isolation of rRNA subfragments can be found in a previous study (2). Briefly, oligodeoxynucleotides complementary to T. thermophilus 23 S rRNA position 2505-2568 (E. coli 2492-2557) or 2240 -2289 (E. coli 2229 -2278) were annealed to the 23 S rRNA followed by digestion with a combination of mung bean nuclease and RNase A to degrade single-stranded nucleic acids. The digestion mixture was separated on a denaturing polyacrylamide gel and stained with ethidium bromide. The band corresponding to the protected 23 S rRNA subfragment was cut out and eluted.
RNase Digestion-6 pmol of 5 S rRNA was digested with either 0.4 g of RNase A (Sigma) or 10 units of RNase T1 (U. S. Biochemical Corp.), and 2 pmol of 23 S rRNA was digested with either 3 g of RNase A or 100 units of RNase T1 as described (26). 0.5 volume of 0.5 M HCl was added to the RNase T1 digestions to hydrolyze 2Ј,3Ј-cyclic phosphates with incubation at room temperature for 30 min. The HCl was removed under vacuum after addition of 4 volumes of H 2 O. The 23 S rRNA digestions were size-separated by in-house fabricated micro reverse-phase columns (5) with minor adjustments. Poros 50R3 (Perseptive Biosystems) was used as column material, and columns were equilibrated and washed with 10 mM triethylammonium acetate (TEAA) pH 7.0. RNase A digestions were eluted in two steps using 8% acetonitrile (MeCN)/10 mM TEAA and 15% MeCN/10 mM TEAA, respectively. The first elution buffer contained 6% MeCN/10 mM TEAA for the RNase T1 digestions. The eluates were dried and dissolved in water to a final concentration of to 2 pmol of intact rRNA per microliter.
Mass Spectrometry and Data Analysis-MALDI mass spectrometry was performed as previously described (26) using 3-hydroxypicolinic acid as matrix on either a Bruker Reflex IV (Bruker-Daltonics, Bremen, Germany) or a Perseptive Voyager STR (PerSeptive Biosystems, Framingham, MA) instrument in positive ion reflector mode, using delayed ion extraction. The spectra were processed using the "m/z" free software (Proteomics Ltd., New York, NY) with three-point internal calibration on predicted unmodified digestion fragments.
Theoretical digestions of rRNA were done using a "Mongo Oligo Mass Calculator" (www-medstat.med.utah.edu/massspec/ mongo.htm). The calculated compositions from observed masses were obtained by using an "Oligo Composition Calculator" (medstat.med.utah.edu/masspec/compo.htm) with the following parameters: a mass accuracy of Ϯ0.005%, one to four methyl groups, and one to three guanosine nucleotides for RNase T1 digestions or one to three pyrimidine nucleotides for RNase A digestions. Matching of nucleotide compositions to the T. thermophilus 23 S rRNA sequence was done with the GPMAW software (Lighthouse Data, Odense, Denmark).
MALDI quadrupole time-of-flight (MicroMass, Manchester, UK) tandem mass spectrometry on positive ions was performed as reported (27). Nanoelectrospray ionization on a Bruker Esquire electrospray ion trap instrument (Bruker-Daltonics, Bremen, Germany) was performed in negative ion mode. These samples were dissolved in 50% MeCN/10 mM TEAA to a concentration of 1-2 pmol/l. The nanoelectrospray needles (Proxeon, Odense, Denmark) were washed with 80% formic acid prior to sample loading to reduce cation salt formation.
Sequencing of 23 S/5 S rRNA Operons-T. thermophilus has two operons coding for 23 S and 5 S rRNA. The two operons have identical 5Ј-promoter regions but different 3Ј-downstream regions (28). Two PCRs were performed with the common promoter-specific primer 5Ј-AGCGGGGGATCTTGAAAGCTGG-GGT-3Ј (Ϫ54 to Ϫ30 upstream of the coding region) and the two different downstream primers 5Ј-TGTCTAGAAGCTTCACCT-TATCCCCCTTTGC-3Ј (complementary to position 3270 -3300 in the region termed "675" in Ref. 28) or 5Ј-ACGCGCCCACCT-GGAGGTGT-3Ј (complementary to position 3273-3292 in the region termed "700" in Ref. 28) to amplify the two operons. The DNA template was obtained from a T. thermophilus colony that was suspended in water and boiled. The desired PCR products were purified via agarose gels and sequenced using a "Big Dye" terminator cycle sequencing kit (Applied Biosystems). Regions of 23 S rDNA that did not correspond to the sequence published by Hopfl et al. (29) (GenBank TM accession number X12612) were sequenced on both strands. GenBank TM accession numbers for the two T. thermophilus 23 S/5S operons are AY379548 and AY379549, respectively.
Structure Modulation of T. thermophilus Ribosome-The coordinates for the structure of the T. thermophilus ribosome reported in a previous study (19) can be found in the Protein Data Bank (accession 1GIY). The MOLMOL software (30) was used to investigate the ribosome structure around the methylated nucleotides and to graphically highlight these nucleotides.

RESULTS
Sequencing of T. thermophilus 23 S/5 S rDNA Genes-Our strategy for mapping post-transcriptional modifications is based on complete digestion of the RNA with a nucleotide-specific RNase and subsequent analysis of the digestion mixture by MALDI-TOF mass spectrometry. Fragments harboring posttranscriptional modifications are identified by deviations between the observed fragment masses and the masses calculated from the gene sequence (26). We therefore needed the exact sequence of the T. thermophilus large ribosomal subunit RNA genes. T. thermophilus has two operons each encompassing genes for 23 S rRNA, 5 S rRNA, and tRNA Gly . After PCR amplification, we sequenced each of the 23 S/5 S rDNA operons and found them identical. Compared with the previously published 23 S rDNA (29) (GenBank TM accession number X12612), G445 and G1180 were absent and an extra Gly at position 519 was present in our version of T. thermophilus strain HB8. The 5 S rDNA was identical to the previously published sequences (28).

Detection of 23 S rRNA Post-transcriptional Modifications by Mass
Spectrometry-rRNA from the large ribosomal subunit was purified via isolation of the ribosomes followed by separation of the ribosomal subunits by centrifugation through sucrose density gradients. The 23 S/5 S rRNA was recovered by phenol/chloroform extraction, and the two rRNA species were size-separated by gel filtration or ultrafiltration.
The 5 S rRNA was digested to completion with either RNase A or RNase T1, and the digestions were mass analyzed (data not shown). We did not detect any post-transcriptional modifications in 5 S rRNA, which is in accordance with absence of post-transcriptional modifications in all other bacterial 5 S rRNA investigated. The only unexpected observation was the presence of a mixed population of 5 S rRNAs with either a single or two A-nucleotides at the 5Ј-end.
T. thermophilus 23 S rRNA was likewise digested to completion with either RNase A or RNase T1. The digestions were additionally separated into two fractions by reverse phase chromatography. This was necessary mainly to reduce the amount of mono-and dinucleotides that otherwise tend to form gas-phase clusters appearing in the m/z regions of less abundant digestion fragments of larger size. Fig. 1A shows the low mass region of the RNase A digestion. Four fragments, at m/z 990.14, 1671.30, 1718.24, and 1992.33 did not correspond to calculated digestion fragments, and are therefore likely to harbor a post-transcriptional modification. No further potentially modified fragments were identified in the higher m/z range after RNase A digestion (data not shown). Fig. 1B shows the high mass region of the RNase T1 digestion; here a single candidate for a post-transcriptionally modified fragment  Table 1).
Determination of Modified Nucleotides-McCloskey and co-workers have reported how the nucleotide composition and methylation level of RNase T1 digestion products can be predicted as a function of mass accuracy (31). Methylation is by far the most common mass-increasing modification in rRNA. We therefore calculated the possible nucleotide composition of all the above fragments potentially harboring post-transcriptionally modified nucleotides assuming a mass accuracy of Ϯ0.005%, allowing one to four methyl groups and up to two sites where cleavage after the normal target nucleotides is prevented due to methylation. The calculations gave unique composition for all digestion fragments, but for the two longest RNase T1 fragments, which yielded three and four possible compositions, respectively. We then compared potential compositions of all the modified digestion fragments to the T. thermophilus 23 S rRNA sequence, also taking the enzyme specificities into account. An RNase substrate nucleotide was assumed in the 3Ј-end of the fragment, and methylations were assumed on internal, uncleaved substrate nucleotides. Only the largest modified RNase T1 fragment (m/z 3565.51) could be located on the T. thermophilus 23 S by this strategy, here to position 1932-1942. All the remaining smaller   in superscript), and NϩMe indicates a not precisely located methyl group. b The fragment is assumed to be preceded by, and end in, an RNase substrate nucleotide in the 23 S rRNA sequence. c Modified nucleotide assigned after site-directed mung bean nuclease purification of a 23 S rRNA subfragments; see text for details. d This RNase A fragment is assumed to overlap with the RNase T1 fragment at position 1932-1942. modified RNase fragments could have originated from numerous positions in the 23 S rRNA, and an exact location could therefore not be assigned.
We subsequently performed tandem mass spectrometry on the post-transcriptionally modified (i.e. methylated) fragments to obtain a sequence and information on the exact location of the modification; the predicted compositions were taken into account in tandem mass spectrum interpretations. Fig. 2A shows an electrospray ionization-ion trap MS/MS mass spectrum of the doubly deprotonated RNase T1 digestion fragment with the predicted composition (A, C 2 , G, U, and Me). The sequence 5Ј-AC[CϩMe]UG-3Ј can be read from the c-and y-ion series (nomenclature according to Ref. 32), and the methyl group must be on the nucleobase, because a distinct signal corresponding to neutral loss of a methylated cytosine is present at m/z 748.0. An RNase T1 digestion fragment with the deduced sequence 5Ј-AC-CUG-3Ј preceded by a G-nucleotide is present only once in T. thermophilus 23 S rRNA, and we can therefore assign the methylated C to position 1983. An example of MS 3 to disclose a methyl group on the ribose is shown in Fig. 2B. The MS/MS results of the RNase A digestion fragment (G, U 2 , and Me) revealed that the methyl group was present on a tandem MS fragment that harbored the first two nucleotides (a c 2 -fragment; not shown). Using this fragment as precursor for MS 3 resulted in signals corresponding to an unmodified G-nucleotide (x 1 ) and the loss of an unmodified uracil-base (M-UH) Ϫ , leaving the ribose of the uridine-nucleotide as the only possible site for methylation.
We obtained better tandem MS results on the larger digestion products with a MALDI-quadrupole-TOF instrument, exemplified in Fig. 3 with the m/z 3213.38 species from the RNase T1 digestion as precursor ion. A series of backbone fragment ions (dominated by y-ions) plus a series of y 8 -ions that have lost consecutive nucleotides from the 3Ј-end allowed complete reading of the sequence. The two methyl groups were assigned to U1960 and C1963, respectively. By a similar tandem MS approach, we assigned the two methyl groups in the m/z 3565.51 RNase T1 digestion fragment to U1937 and C1941. The C1941-methylation is on the ribose, deduced from tandem MS of the unique overlapping RNase A fragment at m/z 1992. 33 Three of the methylated nucleotides could not be assigned to a specific position in the 23 S rRNA, because their sequence contexts were present in numerous copies. These were [Um]GU, [mA]UG, and G[Gm]GGC. However, comparison with modification patterns in other organisms gives suggestions to the location of these modifications. E. coli and a number of other species harbor ribose-methylated uridine at position 2552 (E. coli numbering) in the sequence context [Um]GU, identical to the above (1, 3,4,7,33). Likewise, A2503 (E. coli numbering) in the sequence AUG is methylated in E. coli (3) and in Bacillus megaterium. 4 The two equivalent positions in T. thermophilus are U2563 and A2514, respectively. We therefore isolated the T. thermophilus position 2505-2568 (E. coli. 2494 -2557) 23 S rRNA subfragment by site-directed mung bean nuclease protection (2). Parallel RNase A and RNase T1 digestions followed by mass spectrometry revealed the expected [mA]UG and [Um]GU fragments (data not shown), demonstrating that the T. thermophilus 23 S rRNA harbors a basemethylated A2514 (E. coli 2503) and a ribose-methylated U2563 (E. coli 2552). Using the above logic (1, 3, 33) and approach, the ribose methylation in G[Gm]GGC was located to position 2263 (E. coli 2252). The modifications and means of identification are summarized in Table 1.
Detection of Pseudouridines-Pseudouridine is a structural isomer of uridine and is therefore a mass-neutral post-transcriptional modification. We screened T. thermophilus 23 S 4 B. T. Porse and F. Kirpekar, unpublished observation. rRNA for ⌿s by the CMCT modification method. CMCT reacts with the N3 position in uracil, the N1 position in guanine, and both the N1 and N3 positions in ⌿. Subsequent mild alkaline treatment of the RNA strips off the CMCT from all positions but N3 in ⌿, which yields a distinct stop at the 3Ј-adjacent nucleotide by primer extension analysis of the RNA (24). More than 98% of the T. thermophilus 23 S rRNA was screened for ⌿s by this method. Unambiguous CMCT-dependent primer extension stops were observed at positions U1932 (E. coli 1911), U1938 (E. coli 1917) (Fig. 4), and U2616 (E. coli 2605); all three positions are also ⌿s in E. coli. Fig. 4 also reveals a strong primer extension stop at position 1936 (E. coli 1915), where T. thermophilus contains a methylated uridine and E. coli harbors an m 3 ⌿. It was, however, not possible to discern whether the stop was CMCT-dependent or not.
Less intense CMCT-dependent primer extension stops were observed at positions 918, 2201, and 2210 (E. coli 871, 2180, and 2189, respectively). It was recently shown in studies on H. marismortui 23 S rRNA that weak CMCT-dependent stops are often false positives with regard to ⌿ (8, 34). We therefore investigated these putative ⌿s using an independent method. rRNA subfragments were isolated by site-directed mung bean nuclease protection, and these subfragments were cyanoethylated by treatment with acrylonitrile, which specifically modifies N1 on ⌿s, resulting in a 53.0-Da mass increment (27). The subfragments were digested with RNase A and/or RNase T1, and the digests were subjected to MALDI mass spectrometry, but none of the positions 918, 2201, or 2210 (E. coli 871, 2180, and 2189, respectively) could be verified as ⌿s, because none of the relevant digestion products exhibited the 53.0-Da mass shift (data not shown). These nucleotides are therefore considered unmodified. The data on ⌿s are summarized in Table 2.

DISCUSSION
A combination of mass spectrometric and biochemical approaches identified 11 post-transcriptionally modified nucleotides in 23 S from T. thermophilus (see Tables 1 and 2). Eight modifications were single methylations, and three were ⌿s.
A relevant issue is the exhaustiveness of our detection/identification of post-transcriptional modifications. The search for ⌿s interrogated nearly the entire 23 S rRNA and can essentially be considered complete. Three ⌿s being found is comparable to the number in H. marismortui and Deinococcus radiodurans (34) but significantly less than the eight ⌿s present in E. coli (3). The MALDI-TOF technique makes detection of a single modified digestion fragment hard if it is located among a large molar excess of other ions. In particular, post-transcriptional modifications that remain in mono-or dinucleotides both after RNase A and RNase T1 digestions are difficult to identify ab initio, because this m/z range is dominated by matrix signals, a large excess of unmodified digestion products, and various adducts of both. 184 nucleotides or around 6% of the 23 S rRNA nucleotides remain in mono-or dinucleotides regardless of the RNase used. Furthermore, modified RNase fragments that coincide in mass with an unmodified fragment or various adducts of these will remain undetected, although the former scenario is rare as demonstrated by earlier calculations (31). We do not have an estimate on the number of missed modifications, but there are most likely more post-transcriptional modifications in T. thermophilus 23 S rRNA to be identified.
Seven of the modified nucleotides mapped to domain IV of the 23 S rRNA secondary structure, and four were found in domain V (Fig. 5A). No modifications were detected in domain II of T. thermophilus 23 S rRNA, which is noteworthy because most other organisms harbor post-transcriptional modifications in this domain. Our data strongly suggest that the hypermodified helix 35 of E. coli containing m 1 G745, ⌿746, and m 5 U747 has no modified counterparts in T. thermophilus. This is surprising because an m 1 G is found at either position 745 or    Table 3 gives an overview of which other organisms (potentially) have modified nucleotides in secondary-structure positions equivalent to those found in T. thermophilus; the table is not complete, but phylogenetically representative in view of the existing literature. It should also be noted that many of the modifications in eukaryotes and archaea are putative, because only the small nucleolar RNAs specific for given modifications have been identified, whereas the modifications themselves have not been experimentally confirmed. Some of the modification positions listed in Table 3 coli 1920, 1939, 1942, and 1962) are only found in a single or very few other organisms. Given that the function of ribosomal RNA modifications is largely unknown, and that some appear dispensable (15,16), it is intriguing that small nucleolar RNAs directing ribose methylation were identified in Pyrococcus species (36,37) and Arabidopsis thaliana (38) at positions equivalent to T. thermophilus C1941 (E. coli 1920) and C1983 (E. coli 1962), respectively. Because the two former organisms are completely unrelated to T. thermophilus and synthesize rRNA modifications by a very different mechanism, it is likely that these modifications are specifically important in the named organisms. Targeted gene disruption experiments may shed light on the function of these post-transcriptional modifications.

S rRNA Modifications in T. thermophilus
Several reports indicate that certain nucleotide modifications play a role in thermal stabilization of RNA structure. The number of ribose-methylated nucleotides increases with increasing growth temperature for Sulfolobus solfataricus rRNA (39) and Bacillus stearothermophilus (40), and hyperthermophilic Methanococci have a higher content of tRNA ribose-methylated nucleotides than mesophilic ditto (41). Ribose methylation (42), 5-methylation of U (43) and C (44), as well as ⌿s (45) raise the thermal stability of RNA structures.
Although the present study generally does not position the post-transcriptional modifications at the atomic level, the six completely characterized modifications are of the thermo-stabilizing nature: three ribose methylations and three ⌿s. The base methylated U1960 and C1983 are tentatively also thermostabilizing methylations as counterparts to the homologous m 5 U1939 and m 5 C1962 in E. coli (Table 1). These types of posttranscriptional modifications are in good agreement with the fact that T. thermophilus has optimal growth at 75°C.
We mapped the modified nucleotides in the three-dimensional structure of the ribosome (19). The structure is at 5.5-Å resolution, allowing tracing of the phosphodiester backbone of rRNAs (and polypeptide backbone of proteins) but not of individual nucleobases. Careful analysis of the available x-ray coordinates revealed that none of the modified nucleotides appear in contact with ribosomal proteins, suggesting that post-transcriptional rRNA modifications generally do not have a function in intermolecular interactions between rRNA and protein. Fig. 5B shows a slice of the ribosome with a thickness of less than tRNAs in both the A-and P-sites. Remarkably, all eleven modified nucleotides locate within this slice and are thus densely packed around the center of the ribosome. The following modified domain IV nucleotides are involved in contact with the 16 S rRNA of the small ribosomal subunit: ⌿1932 (E. coli 1911), mU1936 (E. coli 1915), ⌿1938 (E. coli 1917), and Cm1941 (E. coli 1920); these are all located in helix 69 of 23 S rRNA that forms contact to helices 44 and 45 of 16 S rRNA, whereas mC1983 (E. coli 1962) is at the basis of helix 71 in 23 S rRNA and contacts 16 S rRNA at helix 44. The three first modifications are highly conserved as either ⌿ or m 3 ⌿ among the domains of life (34,46), but the high resolution structure of the E. coli ribosome (23) does not implicate a function of these modifications. The domain IV-methylated nucleotides also contact tRNA. mU1936 (E. coli 1915) interacts with the D-stem of tRNA in the A-site, whereas the methylated C1963 (E. coli 1942) is part of a conserved loop and contacts the acceptor stem of tRNA in the A-site. mU1960 (E. coli 1939) is also a part of the latter loop, but this nucleotide is not in direct contact with the A-site tRNA.
Three of the four modified domain V nucleotides are implicated in tRNA associations. Gm2262 (E. coli 2251) is part of the P-loop of 23 S rRNA and interacts with the tRNA acceptor tails

Modification/position, T. thermophilus Position, E. coli
Nucleotide also modified in the following organisms