Multi-site-specific 16S rRNA methyltransferase RsmF
from Thermus thermophilus
HASAN DEMIRCI,1,3LINE H.G. LARSEN,2,3TRINE HANSEN,2ANETTE RASMUSSEN,2ASHWIN CADAMBI,1
STEVEN T. GREGORY,1FINN KIRPEKAR,2and GERWALD JOGL1
1Department of Molecular Biology, Cell Biology and Biochemistry, Brown University, Providence, Rhode Island 02912, USA
2Department of Biochemistry and Molecular Biology, University of Southern Denmark, 5230 Odense M, Denmark
Cells devote a significant effort toward the production of multiple modified nucleotides in rRNAs, which fine tune the ribosome
function. Here, we report that two methyltransferases, RsmB and RsmF, are responsible for all four 5-methylcytidine (m5C)
modifications in 16S rRNA of Thermus thermophilus. Like Escherichia coli RsmB, T. thermophilus RsmB produces m5C967. In
contrast to E. coli RsmF, which introduces a single m5C1407 modification, T. thermophilus RsmF modifies three positions,
generating m5C1400 and m5C1404 in addition to m5C1407. These three residues are clustered near the decoding site of the
ribosome, but are situated in distinct structural contexts, suggesting a requirement for flexibility in the RsmF active site that is
absent from the E. coli enzyme. Two of these residues, C1400 and C1404, are sufficiently buried in the mature ribosome
structure so as to require extensive unfolding of the rRNA to be accessible to RsmF. In vitro, T. thermophilus RsmF methylates
C1400, C1404, and C1407 in a 30S subunit substrate, but only C1400 and C1404 when naked 16S rRNA is the substrate. The
multispecificity of T. thermophilus RsmF is potentially explained by three crystal structures of the enzyme in a complex with
cofactor S-adenosyl-methionine at up to 1.3 A˚resolution. In addition to confirming the overall structural similarity to E. coli
RsmF, these structures also reveal that key segments in the active site are likely to be dynamic in solution, thereby expanding
substrate recognition by T. thermophilus RsmF.
Keywords: rRNA methyltransferase; 5-methylcytidine; RsmB; RsmF; RNA mass spectrometry
Ribosomal RNAs (rRNAs) are post-transcriptionally mod-
ified in all three domains of life, and many modifications
are phylogenetically conserved. Most modifications are
located in functionally important regions of the ribosome,
where they probably act to fine tune protein synthesis (Agris
2004; Gustilo et al. 2008). Complete modification maps of
bacterial 16S rRNAs have been determined for only a hand-
ful of species, and among these are the enteric bacterium
Escherichia coli and the extremely thermophilic bacterium
Thermus thermophilus (Guymon et al. 2006). Despite the
large phylogenetic divergence of these two organisms, their
ribosome modification patterns are quite similar. Of the 11
E. coli and 14 T. thermophilus 16S rRNA modifications,
eight are identical. This suggests a set of common functional
requirements conserved since divergence from their last
common ancestor, and also suggests common recognition
mechanisms among their modifying enzymes.
For most ribosome modifications, a single enzyme recog-
nizes and modifies a single site. However, there exist nota-
ble exceptions. Among these are dimethylation of two adja-
cent adenosines in 16S rRNA by KsgA (Helser et al. 1972);
pseudouridylation of three adjacent residues in tRNAs by
TruA (Hur and Stroud 2007); pseudouridylation of several
tRNA residues by Pus1 (Motorin et al. 1998), Pus2 (Behm-
Ansmant et al. 2007), or Pus7 (Behm-Ansmant et al. 2003);
or methylation of four tRNA positions by Saccharomyces
cerevisiae Trm4 (Motorin and Grosjean 1999). Even with
these multi-site-specific enzymes, however, homologs from
various species generally modify the same residues.
E. coli 16S rRNA contains two 5-methylcytidine (m5C)
residues, located in or near the highly conserved decoding
3These authors contributed equally to this work.
Abbreviations: rRNA, ribosomal RNA; AdoMet, S-adenosyl-L-methionine;
m5C, 5-methyl-cytidine; MALDI mass spectrometry, matrix-assisted laser
desorption ionization mass spectrometry.
Reprint requests to: Gerwald Jogl, Department of Molecular Biology,
Cell Biology and Biochemistry, Brown University, Box G-E129, Provi-
dence, RI 02912, USA; e-mail: Gerwald_Jogl@brown.edu; fax: (401)
863-6114; or Finn Kirpekar, Department of Biochemistry and Molecular
Biology, University of Southern Denmark, Campusvej 55, 5230 Odense M,
Denmark; e-mail: email@example.com; fax: (+45) 65502467.
Article published online ahead of print. Article and publication date
are at http://www.rnajournal.org/cgi/doi/10.1261/rna.2088310.
RNA (2010), 16:1584–1596. Published by Cold Spring Harbor Laboratory Press. Copyright ? 2010 RNA Society.
center of the 30S subunit (Fig. 1). An m5C967 modification
is produced by RsmB (also called Fmu), while an m5C1407
modification is produced by RsmF, formerly known as
YebU (Andersen and Douthwaite 2006). T. thermophilus
16S rRNA contains m5C967 and m5C1407, as well as two
additional m5C nucleotides, m5C1400 and m5C1404 (E. coli
rRNA numbering used throughout) (Guymon et al. 2006).
While the m5C967 and m5C1407 modifications are pre-
sumably produced by RsmB and RsmF homologs, respec-
tively, the source of the two additional m5C residues has
been unknown. Here we demonstrate that T. thermophilus
RsmF is a multi-site-specific methyltransferase and, in
contrast to the single-site-specific E. coli RsmF, is respon-
sible for the synthesis of three modifications: m5C1407,
m5C1400, and m5C1404. We also demonstrate that RsmB is
responsible for the synthesis of m5C967 in T. thermophilus
as well as is in E. coli, thereby accounting for all four m5C
modifications of 16S rRNA. We present crystal structures
of T. thermophilus RsmF up to 1.3 A˚resolution that reveal
a dynamic region in the active site that is absent from the E.
coli RsmF structure, providing a possible explanation for
the expanded recognition capacity of the T. thermophilus
Identification of T. thermophilus 16S rRNA
With the E. coli RsmB and RsmF protein sequences as
queries, we used conventional BLAST searches (Altschul
et al. 1990) to identify potential homologs encoded by the
T. thermophilus HB8 genome (data not shown). Both RsmB
and RsmF have the highest similarity to the T. thermophilus
protein encoded by TTHA1387 (BLAST scores of 106 and
190, respectively) and second-highest similarity to the pro-
tein encoded by TTHA0851 (BLAST scores of 93 and 81,
respectively). The simplest interpretation of these results is
that TTHA1387 encodes RsmF, responsible for methylation
of C1407, leaving TTHA0851 as the most likely candidate
for the gene encoding RsmB, responsible for methylation of
C967. The similarities of the two E. coli enzymes with other
T. thermophilus proteins were far too low to reveal potential
candidates responsible for methylation of C1400 and C1404.
We next constructed T. thermophilus strains in which
either TTHA0851 or TTHA1387 was inactivated by the
homologous recombination and insertion of a heat stable
kanamycin-resistance gene. 16S rRNA was isolated from
these null mutants and subfragments of z50 nucleotides
(nt) around the regions of interest were further purified,
digested with RNase T1, and analyzed by MALDI mass
spectrometry (Fig. 2). Comparison of the TTHA1387 null
mutant to wild-type T. thermophilus HB8 indicates three
clear differences, each corresponding to the disappearance
of a methyl group (z14.0 Da). The RNase T1 digestion
fragment harboring m5C1407, the nucleotide methylated
by RsmF in E. coli, is absent in the null mutant, indicating
that TTHA1387 is indeed rsmF. The predicted RNase T1
fragment reduced by 14.0 Da is obscured by another RNase
T1 fragment that is present in both the wild-type and
TTHA1387 null mutants (Fig. 2B). Unexpectedly, two addi-
tional RNase T1 fragments are also reduced by 14.0 Da.
One of these contains C1400 while the other contains
C1404. This latter RNase T1 fragment from wild-type
T. thermophilus contains three methyl groups, two on
m4Cm1402 and one on m5C1404 (Guymon et al. 2006),
preventing an unambiguous identification of the missing
methyl group. We therefore performed tandem mass spec-
trometry on the1402CCCG1405RNase T1 fragment with two
methyl groups from the TTHA1387 null mutant and com-
pared it with the triply methylated wild-type RNase T1
fragment (Fig. 2C). The clear w2 ions, as well as the less
intense z3 ions, display a 14.0 Da mass difference between
the two samples, showing that the methylations on
m4Cm1402 were not affected by inactivation of TTHA1387.
Tandem mass spectrometry was also performed on the
RNase T1 fragments appearing as a consequence of the lack
of methylations on C1400 and C1407 (data not shown).
As expected, the C1400-containing fragment revealed no
FIGURE 1. Secondary structure diagram of the 39 minor domain of
16S rRNA indicating the position of the three RsmF substrate
T. thermophilus 16S rRNA methyltransferase RsmF
indications of a methyl group, whereas
the RNase T1 fragment with C1407 ex-
hibited a fragmentation pattern corre-
sponding to the expected mass overlap
with an RNase T1 fragment of a different
sequence. In summary, our data lead us
to conclude that TTHA1387 encodes an
RsmF m5C methyltransferase responsible
for synthesizing m5C1400, m5C1404, and
m5C1407 in 16S rRNA of T. thermophilus.
TTHA0851 as the sole candidate for
the gene encoding the m5C967 methyl-
transferase. An approach conceptually
identical to that described above re-
vealed that disruption of TTHA0851
reduced the relevant RNase T1 fragment
by 14.0 Da (Supplemental Fig. 1A).
Since this fragment is methylated at
G966 and C967 in the wild-type strain
(Guymon et al. 2006), tandem mass
spectrometry was again performed (Sup-
plemental Fig. 1B), showing that only
the methyl group on C967 was absent. In
agreement with the suggested nomen-
clature for rRNA modifying enzymes
(Andersen and Douthwaite 2006), we
hereafter refer to TTHA0851 as rsmB
due to substrate specificity identical to
the originally identified enzyme from
E. coli (Gu et al. 1999).
Effect of temperature on growth
of the rsmF null mutant
One possible explanation for methyla-
tion of multiple sites by RsmF is that
such additional methyl groups improve
ribosomal function at elevated temper-
ature. To address this possibility, we ex-
amined the effect of temperature on
growth of the rsmF null mutant. Wild-
type T. thermophilus and the rsmF null
mutant were cocultured at three tem-
peratures and these cocultures were se-
rially subcultured for seven cycles of
24 h each. The proportion of wild-type
and rsmF null mutant cells in each mixed
culture was determined by spreading di-
lutions onto TEM plates with or without
kanamycin. After seven cycles, no differ-
ence in the relative proportions of the
wild-type and rsmF mutant was exhib-
ited at 70°C. However, at 60°C, the rsmF
null mutant constituted only around 5%
FIGURE 2. (Legend on next page)
Demirci et al.
RNA, Vol. 16, No. 8
of the population, and at 80°C the rsmF null mutant was
unable to grow at all. Thus, methylation by RsmF appears to
facilitate growth at temperatures outside the optimal growth
RsmF substrate preference
The T. thermophilus rsmB and rsmF genes were each cloned
into an E. coli expression plasmid in order to produce
proteins for X-ray crystallography and in vitro methylation
studies. The expression constructs were equipped with
C-terminal histidine6tags to facilitate protein purification.
While we achieved a high expression level of RsmF, we were
unable to do so with RsmB despite a series of optimization
attempts. Consequently, in vitro substrate and structure
analyses were performed exclusively with RsmF.
E. coli RsmF requires the 30S ribosomal subunit as a
substrate when the activity is assayed in vitro (Andersen and
Douthwaite 2006). We assayed 70S ribosomes, 30S ribosomal
subunits, and 16S rRNA for the ability to serve as substrates
for methylation by RsmF in vitro. 16S rRNA subfragments of
z50 nt around the target sites were purified after the in vitro
assay and analyzed by mass spectrometry as described above.
In vitro methylation at 70°C showed an interesting but rather
complex substrate pattern. RsmF completely methylates
C1400 when either 16S rRNA or 30S subunits are used as
a substrate. It methylates C1404 to z35% with 16S rRNA
and completely with 30S subunits, and it produces only trace
amounts of methylation of C1407 with 16S rRNA and z75%
with 30S subunits (Fig. 3). There were no indications of
the 70S ribosome being a substrate in vitro. Curiously,
T. thermophilus RsmF expressed in an E. coli rsmF null
mutant almost completely methylated, in vivo, positions
C1400 and C1404, but not C1407 (data not shown).
X-ray crystal structures of RsmF
We determined the structure of T. thermophilus RsmF (456
amino acids) in three different crystal forms and in a com-
plex with cofactor AdoMet to up to 1.3 A˚resolution (Figs.
4, 5). The structure was solved in space group P43(data set
RsmF1, 1.4 A˚resolution) by molecular replacement using
a search model generated with the program Modeller
(Eswar et al. 2008) from the catalytic domain of the RsmF
homolog YebU from E. coli (Pdb 2FRX) (Hallberg et al.
2006). The structures of the AdoMet-bound form in space
group P2 (RsmF2, 1.82 A˚resolution), of the AdoMet-
bound form (RsmF3, 1.3 A˚resolution), and of the apo-
form (RsmF4, 1.68 A˚resolution) in space group P21212
were subsequently solved by molecular replacement with
the refined RsmF1 model. There are two molecules in the
asymmetric unit in space groups P43and P2 and one mol-
ecule in space group P21212. Electron density is generally
well defined in all crystal forms. The majority of residues
(92.0%, 92.3%, 93.1%, and 92.6%) are in the most favored
region of the Ramachandran plot for RsmF1, RsmF2,
RsmF3, and RsmF4, respectively, and there are no residues
in the disallowed region. The final models consist of
residues 5–178, 194–198, and 201–456 and five additional
residues from the histidine6affinity tag in both chains of
data set RsmF1; residues 2–456 and five affinity-tag
residues in both chains of data set RsmF2; residues 1–456
and six affinity-tag residues in data set RsmF3; and resi-
dues 1–456 and seven affinity-tag residues in RsmF4. The
N-terminal a-amino group was ordered in data sets RsmF3
and RsmF4 and contained additional electron density,
which we interpreted as N-(dihydroxymethyl)-L-methio-
nine, the hydrated form of N-formyl-methionine. Data
collection and refinement statistics are given in Table 1.
The overall structure of RsmF consists of a central
canonical class I methyltransferase catalytic domain with
additional N-terminal and C-terminal
domains (Figs. 4, 5). The catalytic do-
main is formed by a central seven-
stranded b-sheet that is flanked on both
sides by three helices of varying lengths.
An inserted region between strand b7
and helix a11 contains additional heli-
ces a9 and a10, which interact with the
two N-terminal helices a1 and a2. A
second inserted region following strand
b9 includes the short helices a13 and
FIGURE 2. (A) MALDI mass spectra of an RNase T1-digested 16S rRNA subfragment (pos.
1378–1432) from wild-type cells (upper panel) or from the TTHA1387 (putative rsmF) null
mutant (lower panel). Expected digestion products are labeled; fragments affected by the null
mutation are set in italics. (B) Expansion of the signals affected by the TTHA1387 null mu-
tation. The sequence and methylation status of the RNase T1 products are indicated. (C)
MALDI tandem mass spectrometry of the methylated RNase T1 fragment of 16S rRNA (pos.
1404–1407); wild-type cells (upper panel), TTHA1387 null mutant (lower panel). Mass
spectrometric fragments used to deduce the methylation status are labeled. The position of
the backbone fragments (nomenclature according to McLuckey et al. ) in the sequence is
shown. MH+, precursor ion selected for fragmentation; C, cytosine; mC, methylated cytosine;
C>p, cytidine-2´-39-monophosphate; me, methyl group.
FIGURE 3. In vitro methylation with 30S ribosomal subunits or 16S
rRNA from the T. thermophilus rsmF null mutant as a substrate. Effect
on C1400-, C1404-, and C1407-harboring RNase T1 products. In
vitro methylated products are set in italics. *, artifact signal arising
from the enzyme preparation.
T. thermophilus 16S rRNA methyltransferase RsmF
a14, which interact with helix a12. Furthermore, RsmF
contains three additional smaller domains, an N-terminal
domain consisting of a three-stranded b-sheet and two flank-
ing helices (Fig. 5B, colored in blue), and two C-terminal
domains consisting of four-stranded b-sheets and two or one
helix (Fig. 5B, colored in magenta and red).
Cofactor binding, substrate docking,
and conformational flexibility in the active site
The coordination of AdoMet in the T. thermophilus RsmF
active site is similar to that seen in other class I methyl-
transferases. However, the previously published structure of
E. coli RsmF did not contain the cofactor AdoMet in the
active site, precluding a direct comparison. Both the T.
thermophilus and E. coli RsmF cofactor-binding sites reveal
a new variation for the methyltransferase signature motif I
(Malone et al. 1995), with the highly conserved GxGxG
sequence replaced by
three alanines and a proline results in a loop conformation
that is very similar to that observed in other methyltrans-
ferases with a GxGxG motif (e.g., RsmC) (Demirci et al.
2008a). In RsmF, the amide hydrogen atom of the last
glycine residue forms a hydrogen bond with the cofactor
carboxy group (Fig. 6). Other key interactions with AdoMet
are well conserved in RsmF. The cofactor adenine ring is
located in a mainly hydrophobic pocket lined by residues
Val134, Pro160, and Leu211. This pocket is open toward the
solvent. The adenine amino group is not specifically recog-
109AAAPG113. The combination of
nized and interacts with solvent water molecules. The ribose
hydroxyl groups form hydrogen bonds with Glu133 and
Arg138, and the methionine amino group interacts with
Asp177. The AdoMet cofactor is bound in a cleft in the
RsmF active site, which suggests that substrate cytidine bases
are inserted into the active site in an unstacked conforma-
tion. Inspection of the electrostatic charge distribution re-
veals a large positively charged surface region, which would
be consistent with binding to an RNA surface and modifi-
cation of the substrate base in an unstacked orientation (Fig.
6C). To evaluate the possible orientation of a substrate base
in the active site, we performed computational docking
calculations with the program Dock6 (Lang et al. 2009). The
resulting positions of cytosine and m5C in the presence of
AdoMet in data set RsmF3 are highly similar to each other,
with m5C placed into the active site with its phosphate group
toward a positively charged pocket at the entrance of the
active site cleft (Fig. 6E). The position of the phosphate
group is close to a sulfate molecule that we observed in data
set RsmF1, providing further support for the results of the
docking calculation (Fig. 6F).
Interestingly, we observed that three active site segments
were disordered in data set RsmF1. These segments include
residues 179–193 (including helices a9 and a10 and the
catalytic Cys180), residues 199–200, and the N-terminal
residues 1–5, which interact with the first two segments (Fig.
6E,F, colored in green). We observed electron density for the
intervening residues 194–198, which formed a lattice con-
tact with a neighboring molecule. However, the position of
FIGURE 4. Structure-based sequence alignment of RsmF from T. thermophilus and E. coli. Secondary structure elements of T. thermophilus RsmF
are indicated on top. The color scheme for the secondary structure elements is as in Figure 5A. The position of the variant methyltransferase motif
I is marked with a red box. A flexible region observed in the active site is marked with a green box; residues interacting with the cofactor are
marked with orange boxes.
Demirci et al.
RNA, Vol. 16, No. 8
these five residues was not related to their position in the
other three data sets, suggesting that the extended active
site region between residues 179 and 201 can reorient in the
RsmF structure. This observation suggests that this active
site region is dynamic, which may be important for sub-
strate binding at 72°C, the optimum growth temperature of
Substrate recognition mechanisms
We have identified the two enzymes responsible for the
synthesis of the four m5C modifications of T. thermophilus
16S rRNA, and characterized the RsmF methyltransferase
responsible for synthesizing three of these. rRNA modifying
enzymes in bacteria are generally highly specific, with a
one-to-one association between the modifying enzyme and
the modification. A few cases of multitarget ribosome mod-
ifying enzymes have been reported (Helser et al. 1972;
Demirci et al. 2008b), but to our knowledge T. thermophi-
lus RsmF is the first rRNA methyltransferase found to mod-
ify three different nucleotides. Most ribosome modifying
enzymes probably recognize assembly intermediates, and
the data presented here are consistent with that notion.
T. thermophilus RsmF methylates C1400 and C1404 in
vitro using either 16S rRNA or 30S subunits as substrates,
whereas both E. coli (Andersen and Douthwaite 2006) and
T. thermophilus RsmF exclusively utilize 30S subunits as
substrates for methylation of C1407. This may reflect that
C1400 and C1404 methylations do not rely on the asso-
ciation of ribosomal proteins in order to be recognized by
T. thermophilus RsmF. C1407 methylation, in contrast,
depends on both rRNA and the ribosomal protein for the
recognition by RsmF in both T. thermophilus and E. coli.
More puzzling is the observation that T. thermophilus RsmF
does not methylate E. coli ribosomes in vivo on C1407. It
is perhaps worth noting that methylation of C1407 in the
T. thermophilus 30S ribosomal subunit in vitro was less
efficient than methylation of the other two positions,
indicating the need for a particular intermediate assembly
structure or for accessory factors. The only clear in vitro
FIGURE 5. Overall structure of RsmF. (A) Schematic representation of the position of the substrate bases in the 30S ribosomal subunit (Pdb
entry 2WRI) (Gao et al. 2009). C1400, C1404, and C1407 in helix 44 (green) are shown in stick representation with 5-methyl groups as pink
spheres. (B) Schematic stereo representation of the overall structure of RsmF. Secondary structure elements are in orange and yellow for the
catalytic domain, in salmon and blue for additional N-terminal domains, and in magenta and red for C-terminal domains. The flexible region in
the active site including helices a9 and a10 is indicated with an arrow and colored in green. (C) Topology diagram with secondary structure
elements colored as in B.
T. thermophilus 16S rRNA methyltransferase RsmF
methylation observed at 37°C with T. thermophilus RsmF
was on C1400 with 16S rRNA as the substrate (data not
shown), which is not evidently related to the methylation
pattern in vivo in the heterologous system. It seems unlikely
that the aberrant methylation in the heterologous system
reflects species-specific differences in the mature 30S ribo-
somal subunit, given the extreme sequence and structural
conservation of the decoding site. Instead, it may reflect
differences in 30S subunit assembly in the two organisms,
necessary due to the large difference in growth temperature.
The four m5C residues in T. thermophilus 16S rRNA are
clustered in and around the functionally critical decoding
center at or close to sites of contact with tRNA, mRNA, and
EF-G (Ogle et al. 2002; Selmer et al. 2006; Gao et al. 2009).
m5C1400, m5C1404, and m5C1407 are located in the
subunit body, while m5C967 is located in the subunit head,
about 10 A˚from m5C1400 (Wimberly et al. 2000). Exam-
ination of the 30S subunit crystal structure (Wimberly et al.
2000) indicates that the three bases methylated by RsmF are
situated in three distinct structural contexts, but provides
few clues to a common mode of substrate recognition.
C1400 is an unpaired base protruding from a sharply bent
segment of rRNA at the junction of helices 43 and 44, while
C1404 and C1407 are engaged in Watson–Crick pairs within
helix 44 (Fig. 4A). The C5 positions of the latter two bases
are not obviously accessible, such that RsmF would need to
approach them from the major groove side. While a flipping
of C1407 out of the helix via the minor groove could allow
access to this base, such a mechanism would be problematic
for C1404, whose minor groove side is packed against the
rest of the 30S subunit.
While m5C1404 and m5C1407 are about 11 A˚apart
(Selmer et al. 2006), m5C1400 is quite distant from both of
these bases (about 21 and 30 A˚, respectively). The pro-
truded conformation of m5C1400 in the mature 30S sub-
unit is due in part to base-pairing interactions between
adjacent bases and the 1500 region of 16S rRNA (C1399–
G1504 and G1401–C1501). As the 1500 region is one of the
last segments of 16S rRNA to be synthesized, C1400 could
potentially be positioned much closer to C1404 and C1407
in an assembly intermediate, prior to the formation of the
C1399–G1504 and G1401–C1501 base pairs. RsmF could
utilize a single binding mode to then access all three bases,
further facilitated by its flexible active site domain. Meth-
ylation of C1400 in mature 30S subunits would therefore
involve disruption of the adjacent base pairs. A complete
understanding of the recognition mechanism of these
enzymes will require high-resolution structural data on
assembly intermediate-enzyme complexes. Given the large
number of potential subunit assembly intermediates, pre-
cisely defining the physiological substrate for RsmB and
RsmF will be a formidable task.
TABLE 1. Data collection and refinement statistics
a, b, c (A˚)
a, b, g (°)
71.0, 71.0, 186.7
90, 90, 90
66.0, 78.3, 108.1
90, 107.1, 90
89.7, 109.0, 51.0
90, 90, 90
89.8, 109.1, 50.8
90, 90, 90
Number of reflections
Number of atoms
Bond lengths (A˚)
Bond angles (°)
aOne crystal used for each data set.
bThe highest resolution shell is shown in parentheses.
Demirci et al.
RNA, Vol. 16, No. 8
Structural comparison of RsmF with related
A database search with Dali (Holm et al. 2008) confirmed
the structural similarity of the T. thermophilus and E. coli
(PDB 2FRX) (Hallberg et al. 2006) RsmF homologs, which
superimpose with a root-mean-square deviation (RMSD)
of 1.6 A˚for 342 Ca atoms (Fig. 7A) and
are the only two structures in the Pro-
tein Database with this domain organi-
zation. Even so, substantial structural
differences are observed in most of the
loop regions and for the long connect-
ing loop between the methyltransferase
domain and the first C-terminal domain.
While in the active site, the positions
of residues in the cofactor-binding site
and of the two cysteine residues are
conserved (Fig. 7B), there are a number
of positively charged residues (Arg30,
Arg190, Arg194, His195, and Arg203) in
the T. thermophilus structure that are
absent from the E. coli enzyme (Fig. 7B).
Three of these are located in the flexible
region, and the combination of a posi-
tive charge and flexibility close to the
active site is suggestive of a functional
contribution of this region to the mul-
tisite specificity of T. thermophilus RsmF
(Figs. 6C,D, 7B). Methylation of three
rRNA positions may require an increase
in the enzyme’s structural dynamics in
order to accommodate the 30S subunit
in slightly different orientations. Similar
observations have been made for other
multi-site-specific methyltransferases in-
cluding KsgA, which modifies two adja-
cent adenosines in the 30S ribosomal
subunit (O’Farrell et al. 2004; Demirci
et al. 2009), and the PrmA ribosomal
protein methyltransferase, which under-
goes dramatic interdomain movements
to modify multiple lysine residues and the
N-terminal a-amino group on the same
substrate protein (Demirci et al. 2007,
The second C-terminal domain in
RsmF is related to the RNA-binding
PUA (pseudouridine synthase and
archaeosine transglycosylase) domains
(Perez-Arellano et al. 2007). The RlmI
m5C1962 in 23S rRNA (Purta et al.
2008) also contains a PUA domain
(Sunita et al. 2008), although it is
N-terminal to the catalytic methyltransferase domain and
in a different orientation. PUA domains contain six
b-strands, which form a central pseudobarrel closed by
a short 310-helix. A comparison of the C-terminal domain
in RsmF with a typical PUA domain in archaeosine trans-
glycosylase (ArcTGT, Pdb entry 1J2B) reveals that the
central fold is similar (53 Ca atoms align with an RMSD
FIGURE 6. Substrate docking and conformational flexibility. (A) Cofactor-binding site in
RsmF. Bound AdoMet is shown in blue sticks. Hydrogen bonds to coordinating residues are
indicated. (B) Final sA-weighted 2mFO-DFCelectron density map of the cofactor-binding site
(data set RsmF2) contoured at the 1s level. (C,D) Comparison of the electrostatic surface
charge distribution between RsmF from T. thermophilus and E. coli. The location of the
C-terminal domains and of the flexible region (labeled ‘‘FLEX’’) is indicated with circles.
AdoMet and docked m5C are shown as sticks. AdoMet from the T. thermophilus structure is
shown with the E. coli structure for comparison. (E) The modeled position of m5C docked into
the active site of RsmF. m5C and AdoMet are shown in tan and blue sticks, respectively.
Residues in the flexible region are shown in green. (F) The active site region in data set RsmF1.
A sulfate ion is observed close to the position of the m5C phosphate group. Residues 194–198 in
the flexible region interacting with a neighboring RsmF molecule are indicated with an arrow.
T. thermophilus 16S rRNA methyltransferase RsmF
of 1.8 A˚), but that several connecting loop regions are
substantially shorter and two a-helices and one b-strand of
the pseudobarrel are absent. Thus, this PUA-like domain
differs considerably from typical PUA domains. However,
the similarity to RNA-binding PUA domains and the
positive surface charge distribution observed in both RsmF
structures are suggestive for a conserved function of the
PUA-like domain in RNA recognition.
The next most closely related structure is E. coli RsmB
(474 residues, PDB 1SQF) (Foster et al. 2003). A total of
264 Ca atoms can be aligned with an RMSD of 1.5 A˚
between RsmB and RsmF. Both enzymes retain the same
organization for the N-terminal domains and the core
methyltransferase domain. However, an additional 140-
residue N-terminal RNA-binding domain provides sub-
strate specificity to RsmB, whereas the two C-terminal
domains following the core methyltransferase domain (160
residues) are likely to determine substrate recognition by
RsmF. Thus, these two enzymes have evolved substrate
specificity via acquisition of additional,
unrelated RNA recognition domains.
While there are as yet no enzyme–
substrate complexes for rRNA m5C-
catalytic mechanism can be gleaned from
a comparison with the RlmD and TrmA
m5U methyltransferases in covalent in-
termediate complexes with RNA oligonu-
DNA and RNA m5C methyltransferases
to attack the six-position of the pyrimi-
dine base to activate the five-position for
methyl group transfer (Liu and Santi
2000). Cys180 and Cys230 in RsmF are
positioned equivalently to the catalytic
Cys324 and the catalytic base Glu358 of
TrmA. The substrate nucleotides insert
into TrmA and RsmF in unrelated di-
rections, consistent with a lack of struc-
tural homology outside the methyltrans-
ferase domains. Nevertheless, the C5
positions of the pyrimidine rings and of
the 5-methyl carbons are quite similar
with respect to the catalytic cysteine resi-
dues and the AdoMet cofactor. The same
structural homology of the active site
geometry can be observed in comparison
with the RlmD methyltransferase (Sup-
plemental Fig. 2; Lee et al. 2005).
A possible origin of T. thermophilus
T. thermophilus RsmF shows the highest
similarities with proteins from close relatives, namely,
Thermus aquaticus and two Meiothermus species of the
Thermaceae family. Remarkably, the next highest similarities
(BLAST scores between 267 and 359) are with the NOL1/
NOP2/Sun proteins from the Gram-positive Firmicutes
phylum (Supplemental Fig. 3). This similarity, together
with the fact that the Thermaceae family and most members
of the Firmicutes identified in Supplemental Figure 3 are
thermophilic, suggest that rsmF has undergone horizontal
transfer between the Thermaceae family and members of the
Firmicutes phylum. Thus, we speculate that this version of
the RsmF protein, which catalyzes methylation of three
cytidines, may be adaptive for existence in thermally
challenging environments. The effect of the loss of methyl-
ation by RsmF on growth at different temperatures is
consistent with this notion. Our hypothesis of horizontal
transfer of rsmF predicts that RsmF of other members of the
Thermaceae family and members of the Firmicutes phylum
will also be found to introduce multiple m5C modifications.
FIGURE 7. Comparison with other methyltransferases. (A) Differences between the overall
structures of RsmF from T. thermophilus (orange) and E. coli (green). AdoMet bound in the T.
thermophilus structure is shown as blue sticks. (B) Comparison of the active site region in both
enzymes. Residues in the cofactor-binding site and in the flexible region in the T. thermophilus
enzyme are shown as sticks. (C) Comparison between RsmF (orange) and a substrate complex
structure of the TrmA methyltransferase (cyan/blue). (D) Comparison of the active site region
illustrating the differences in the insertion direction of the substrate base in RsmF (m5C in
light orange) and in TrmA (m5U in cyan).
Demirci et al.
RNA, Vol. 16, No. 8
MATERIALS AND METHODS
Cloning of the T. thermophilus rsmB and rsmF genes
The T. thermophilus HB8 loci TTHA0851 (GenBank accession
number BAD70674) and TTHA1387 (GenBank accession number
BAD71210) were PCR amplified from genomic DNA and purified
via the High Pure PCR Template Preparation Kit (Roche). The
100 mL PCRs contained 150 ng DNA, 10 mM of each primer,
10 mM dNTP, 1 unit Phusion DNA polymerase (Finnzymes), and
1x Phusion HF buffer. Primers for rsmB amplification were 59-CC
CTGGACATATGAGGGCCGG-39 and 59-GGCCAAGATCTTGCC
TGAGAG-39, and the temperature cycling was as follows: 98°C/30
sec; 30X (98°C/10 sec, 59°C/20 sec, and 72°C/36 sec); and 72°C/420s.
Primers for rsmF amplification were 59-GCTAGGGTACACATA
TGCTGCCC-39 and 59-GCACGGGGGTGAGATCTAAGCCC-39,
and the temperature cycling was as follows: 98°C/30 sec; 30X
(98°C/10 sec, 62°C /20 sec, 72°C/42 sec); and 72°C/420 sec. The
desired PCR products were purified from agarose gels using the
GFX PCR purification kit (GE Healthcare). The PCR fragments
were digested with NdeI and BglII and inserted into the expression
vector pLJ102 (Andersen and Douthwaite 2006), generating
isopropyl-1-thio-b-D-galactopyranoside (IPTG)-inducible genes
for the recombinant proteins with a C-terminal histidine6tag.
The constructs (designated pLJ102-RsmB and pLJ102-RsmF) were
used to transform an rsmF-deletion derivative of E. coli CP79
(Andersen and Douthwaite 2006).
Deletion of the T. thermophilus rsmB and rsmF genes
Constructs for inactivation of the T. thermophilus rsmB and rsmF
genes were made by inserting the gene for a heat tolerant
kanamycin (HTK) nucleotidyltransferase (Hoseki et al. 1999) into
the methyltransferase parts of either pLJ102-RsmB or pLJ102-
RsmF. The htk part of pUC18-htk (Hashimoto et al. 2001) was
amplified by PCR with primers that introduced an upstream AvrII
site and a downstream SacI site into the product for later disrup-
tion of rsmB. For rsmF disruption, the PCR primers introduced
SacI restriction enzyme sites both upstream of and downstream
from the htk gene. These sites were used to insert the PCR prod-
ucts into pLJ102-RsmB and pLJ102-RsmF to form the plasmids
pLJ102-RsmBThtk and pLJ102-RsmFThtk, which were propa-
gated in the E. coli strain Top10 (Invitrogen). T. thermophilus
HB8 was transformed with pLJ102-RsmBThtk or pLJ102-
RsmFThtk selecting for kanamycin resistance as described by
others (Hashimoto et al. 2001; Cameron et al. 2004). Kanamycin-
resistant transformants were restreaked twice. Gene disruptions
were verified by PCR with primers distal to the interrupted rsmB
or rsmF genes on genomic DNA; resulting PCR products were
characterized by sequencing.
Growth competition assays
Wild-type and rsmF null mutant liquid cultures were grown at
70°C to saturation, then equal numbers of cells from each were
mixed and incubated in 5 mL TEM medium at 60°C, 70°C, or
80°C. After growth for 24 h, 100 mL of the 60°C culture, 10 mL of
the 70°C culture, and 1000 mL of the 80°C culture were trans-
ferred to a fresh 5-mL medium and incubated at the respective
temperatures for another 24 h. This was repeated in independent
triplicates for seven cycles. Samples of 1 mL were collected at each
dilution and half was plated on TEM plates without antibiotic and
the other half was plated on TEM plates with 30 mg/mL
kanamycin. The plates were incubated at 70°C.
Purification of T. thermophilus ribosomal subunits
T. thermophilus culture (1 L) was grown in TEM media (contain-
ing 30 mg/mL of kanamycin when appropriate) with shaking at
70°C to an OD600= 0.6. Cells were harvested and washed once
with 100 mL of buffer A (10 mM NH4Cl, 20 mM MgCl2, 100 mM
KCl, and 10 mM Tris-HCl [pH 7.5]), then suspended in 10 mL of
buffer A, and disrupted by sonication. The lysate was cleared by
centrifugation twice in a Beckman JA20 rotor at 16,000 rpm for
10 min at 4°C. Crude ribosomes were collected by centrifugation
in a Beckman Ti50 rotor at 19,000 rpm for 19 h at 4°C, and
dissolved in buffer A. 70S ribosomes were obtained by centrifu-
gation of 100 A260units of crude ribosomes through a 10%–40%
sucrose gradient (200 mM NH4Cl, 20 mM MgCl2, 20 mM Tris-
HCl [pH 7.5]) in a Beckman SW28 rotor at 20,000 rpm for 18 h at
4°C. Fractions containing intact 70S ribosomes were pooled and
concentrated by centrifugation in a Beckman Ti50 rotor at 40,000
rpm for 22 h at 4°C, washed, and dissolved in buffer A, and stored
50S and 30S ribosomal subunits were obtained by adjusting
100 A260units of crude ribosomes (10 mM NH4Cl, 2 mM MgCl2,
100 mM KCl, and 10 mM Tris-HCl [pH 7.5]) and passing
through a 5%–20% sucrose gradient (200 mM NH4Cl, 1 mM
MgCl2and 20 mM Tris-HCl [pH 7.5]) in a Sorvall AH-629 rotor
at 20,000 rpm for 18 h at 4°C. After pooling of the relevant
fractions, the subunits were adjusted to 10 mM MgCl2 and
pelleted by centrifugation in a Beckman Ti50 rotor at 40,000
rpm for 22 h at 4°C, washed with and dissolved in 10 mM NH4Cl,
10 mM MgCl2, 100 mM KCl, and 10 mM Tris-HCl (pH 7.5), and
stored at ?80°C.
Isolation of 16S rRNA and subfragments
from T. thermophilus and E. coli
Water (400 mL) was added to 100 mL of 30S ribosomal subunits
and the rRNA was extracted with 500 mL phenol, phenol/
chloroform, and chloroform. rRNA was ethanol precipitated
and dissolved in water. Purification of 16S rRNA subfragments
was performed as previously described (Andersen et al. 2004).
Briefly, 16S rRNA was hybridized to an excess of oligodeoxynu-
cleotide complementary to either the region 944–990 or the region
1378–1432. Single-stranded nucleic acids were digested with
Mung Bean Nuclease and RNase A. The resulting mixture was
separated on a polyacrylamide gel. Bands were visualized by ethid-
ium bromide staining, excised, and eluted.
E. coli CP79 with the endogenous rsmF inactivated, but com-
plemented with the T. thermophilus homolog on the plasmid
pLJ102-RsmF, were grown at 37°C to an OD450= 0.45 in 200 mL
of LB medium containing 100 mg/L of ampicillin. RsmF expres-
sion was induced by addition of IPTG to 1 mM, and incubation
for another 3 h. Cells were harvested by centrifugation at 4°C,
washed in 100 mL TMN buffer (50 mM Tris-HCl [pH 7.8],
10 mM magnesium acetate, 100 mM NH4Cl), and resuspended in
2 mL TMN buffer prior to lysis by sonication (7 3 30 sec on ice)
T. thermophilus 16S rRNA methyltransferase RsmF
and removal of debris by centrifugation (10 min/14,000 rpm/4°C/
microcentrifuge). Total RNA was recovered from the supernatant
by phenol extraction and ethanol precipitation. A 16S rRNA
subfragment was isolated as described above using an oligodeoxy-
nucleotide complementary to the region 1378–1432.
In vitro methylation
Reactions contained 50 pmol of 16S rRNA, 30S subunits, or 70S
ribosomes from the T. thermophilus TTHA1387 null mutant as the
substrate in a total volume of 100 mL (containing 100 mM NH4Cl,
10 mM MgCl2, 40 mM Hepes [pH 7.5]), 6 mM b-mercaptoetha-
nol, and 10% glycerol (prepared as a two times concentrated stock
solution), 1.5 mM S-adenosyl methionine, and 2 mg of recombi-
nantly expressed RsmF (see below).
For the reaction at 70°C, water and stock buffer were mixed and
left at room temperature for 15 min. Then a substrate, an enzyme
and S-adenosyl methionine were added and incubated at 70°C for
1 h. The 37°C reaction was started by mixing water and buffer
followed by 15 min at room temperature; the substrate was added
and the mixture transferred to 50°C for 5 min. After cooling to
37°C, S-adenosyl methionine and an enzyme were added and the
incubation continued for 1 h. Reactions were stopped by phenol/
chloroform extraction and the rRNA was recovered by ethanol
precipitation before purification of 16S rRNA subfragments as
described above. Control reactions without enzyme or S-adenosyl
methionine were carried out in all instances.
RNase T1 digestion and mass spectrometry
A purified 16S rRNA subfragment (1–2 pmol) was incubated with
2 units RNase T1 (Roche) and 50 mM 3-hydroxypicollinic acid
(3-HPA) in a total volume of 2 mL for 4 h at 37°C. MALDI mass
spectrometry was performed either on an ABI voyager STR in-
strument or a Waters Q-TOF MALDI instrument; MALDI tan-
dem mass spectrometry was done on a Waters Q-TOF MALDI
instrument. All spectra were recorded in positive ion mode using
3-HPA as the matrix. Experimental details were as previously de-
scribed (Douthwaite and Kirpekar 2007).
Protein expression and purification for crystallization
E. coli BL21 (DE3) (Invitrogen) containing pLJ102-RsmF was
grown to midlog phase in LB media at 37°C in the presence of
200 mg/mL ampicillin. Protein expression was induced at 20°C
with 400 mM IPTG. Cells were pelleted after 18 h by centrifuga-
tion at 4000 rpm for 20 min at 4°C and lysed by ultrasonication
on ice in a buffer containing 20 mM Tris-HCl (pH 8.5), 300 mM
NaCl, 5 mM b-mercaptoethanol, 0.1% Triton X-100, and 5%
glycerol. Cell debris and membranes were pelleted by centrifuga-
tion at 11,000 rpm for 30 min at 4°C. The soluble E. coli proteins
were precipitated by heat treatment at 65°C for 30 min and
pelleted by centrifugation at 11,000 rpm at 4°C for 30 min.
Soluble C-terminally hexahistidine-tagged T. thermophilus RsmF
was further purified by affinity chromatography with nickel-
nitrilotriacetic acid resin (Qiagen). Untagged proteins were re-
moved with buffer containing 20 mM Tris-HCl (pH 8.5), 250 mM
NaCl, and 1 mM imidazole (pH 8.5). Recombinant RsmF was
then eluted with the same buffer containing 150 mM imidazole.
The protein was then purified by cation exchange chromatogra-
phy (SP) (GE Healthcare) at pH 8.5, using a linear gradient of
10 mM to 1 M NaCl concentration. RsmF fractions were pooled
and concentrated and applied to a size-exclusion S200 column (GE
Healthcare) equilibrated with a buffer containing 20 mM Tris-HCl
(pH 8.5) and 200 mM NaCl. Purified RsmF was concentrated to
13 mg/mL for crystallization trials. The C-terminal hexahistidine
tag was not removed for crystallization. For the production of
selenomethionyl proteins, the expression construct was trans-
formed into B834 (DE3) cells (Novagen). The bacterial growth
was carried out in defined LeMaster medium (Hendrickson et al.
1990), and the protein was purified using the same protocol as for
the unmodified protein. To form the RsmF-AdoMet complex,
purified RsmF was mixed with 4 mM AdoMet incubated at 60°C
for 15 min and slowly cooled to room temperature before
performing crystallization experiments.
Crystallization of RsmF
All crystals were obtained using the microbatch technique under
oil at 4°C. To obtain the RsmF1 crystal form, 1 mL of protein
solution was mixed with the reservoir solution containing 20%
(w/v) PEG3350 and 200 mM sodium sulfate decahydrate (pH
6.6). Initial crystals grew over the course of 1–2 wk with maxi-
mum dimensions of 0.3 3 0.3 3 0.2 mm. To obtain the RsmF2
crystal form, 1 mL of the RsmF–AdoMet complex was mixed with
the reservoir solution containing 200 mM NaCl, 12% w/v
PEG8000 and 100 mM HEPES-KOH (pH7.5). Initial crystals
grew over the course of 2–3 wk with maximum dimensions of
0.1 3 0.4 3 0.4 mm. To obtain the RsmF3 crystal form, 1 mL of
the RsmF–AdoMet complex was mixed with the reservoir solution
containing 10% w/v PEG1000, 200 mM NaCl, and 100 mM Tris-
HCl (pH 8.5). The initial crystals grew over the course of 1–2 wk
with maximum dimensions of 0.05 3 0.3 3 0.4 mm. To obtain
the RsmF4 crystal form, 1 mL of the RsmF–AdoMet complex
solution was mixed with a reservoir solution containing 160 mM
magnesium chloride hexahydrate, 80 mM Tris-HCl (pH 8.5), and
24% w/v PEG4000. Initial crystals grew over the course of 1–2 wk
with maximum dimensions of 0.05 3 0.3 3 0.3 mm. RsmF1
crystals were gradually dehydrated by increasing the PEG3350 to
30% w/v and then cryoprotected in a mother liquor supplemented
with 25% v/v glycerol and then flash-frozen by being plunged into
liquid nitrogen. RsmF2 crystals were cryoprotected in a mother
liquor supplemented with 20% v/v ethylene glycol and then flash-
frozen by being plunged into liquid nitrogen. RsmF3 crystals
were cryoprotected by gradually increasing the concentration of
PEG1000 to 30% and then flash-frozen by being plunged into
liquid nitrogen. RsmF4 crystals were cryoprotected in a mother
liquor supplemented with 20% glycerol and then flash-frozen by
being plunged into liquid nitrogen.
X-ray diffraction data for RsmF1, RsmF2, and RsmF4 crystals
were collected on a MAR CCD detector at the X4C beamline of
the National Synchrotron Light Source in Brookhaven at a wave-
length of 0.979 A˚and ?180°C. Diffraction data for RsmF3 crystals
were collected on an ADSC CCD detector at the X4A beamline of
the National Synchrotron Light Source in Brookhaven at a wave-
length of 0.979 A˚and ?180°C. Diffraction data for RsmF1 in
space group P43 were collected to 1.4 A˚resolution with cell
Demirci et al.
RNA, Vol. 16, No. 8
dimensions a = 71.0 A˚, b = 71.0 A˚, and c = 186.7 A˚. Diffraction
data to 1.82 A˚for RsmF2 were collected in space group P2 with
cell dimensions a = 66.0 A˚, b = 78.3 A˚, and c = 108.1 A˚.
Diffraction data to 1.29 A˚for RsmF3 were collected in space group
P21212 with cell dimensions a = 89.7 A˚, b = 109.0 A˚, and c = 51.0
A˚. Diffraction data to 1.68 A˚for RsmF4 were collected in space
group P21212 with cell dimensions a = 89.8 A˚, b = 109.1 A˚, and c =
50.8 A˚. A single crystal was used for each data set. The diffraction
images were processed and scaled with the HKL2000 package
(Otwinowski and Minor 1997). The data processing statistics are
summarized in Table 1.
Structure determination and refinement
The RsmF structure was solved by molecular replacement with the
program Phaser (McCoy et al. 2007) from the CCP4 program
suite (Bailey 1994) in space group P43to 1.4 A˚resolution (data set
RsmF1). The initial search model was built with the program
Modeller (Eswar et al. 2008) from the catalytic domain of E. coli
YebU (Pdb code 2FRX). After the placement of two RsmF
catalytic domains in the asymmetric unit and the initial re-
finement with Refmac (Murshudov et al. 1997), the model was
further rebuilt with ARP/wARP (Langer et al. 2008). The resulting
model was 90% complete and manually checked and completed
with Coot (Emsley and Cowtan 2004). Final crystallographic re-
finement was performed with the program Phenix (Adams et al.
2002). The other crystal forms were subsequently solved by
molecular replacement. The atomic coordinates from the RsmF4
model were then used for initial refinement of the RsmF–AdoMet
complex structure in space group P21212 (RsmF3). There are two
molecules in the asymmetric unit in data sets RsmF1 and RsmF2,
and one molecule in RsmF3 and RsmF4. The crystallographic
R/Rfreefactors are 0.17/0.19, 0.16/0.19, 0.18/0.19, and 0.17/0.19
for the four data sets: RsmF1, RsmF2, RsmF3, and RsmF4,
respectively. The stereochemical quality of the model was assessed
with Procheck (Laskowski et al. 1993). The Ramachandran sta-
tistics (most favored/additionally allowed/generously allowed/
disallowed) are 91.9%/8.1%/0.0%/0.0% for RsmF1, 91.9%/8.1%/
0.0%/0.0% for RsmF2, 93.6%/6.4%/0.0%/0.0% for RsmF3, and
92.5%/7.5%/0.0%/0.0% for RsmF4. The refinement statistics are
summarized in Table 1. Figures were generated using Pymol
Coordinates and structure factors have been deposited in the
Protein Data Bank with accession codes 3M6U, 3M6V, 3M6W,
and 3M6X for data sets RsmF1, RsmF2, RsmF3, and RsmF4,
Supplemental material can be found at http://www.rnajournal.org.
We thank John Schwanof and Randy Abramowitz for access to the
X4A and X4C beamlines at the National Synchrotron Light Source.
This work was supported by grants GM19756 and GM19756-37S1
from the National Institutes of Health.
Received January 14, 2010; accepted April 26, 2010.
Adams PD, Grosse-Kunstleve RW, Hung LW, Ioerger TR, McCoy AJ,
Moriarty NW, Read RJ, Sacchettini JC, Sauter NK, Terwilliger TC.
2002. PHENIX: Building new software for automated crystallo-
graphic structure determination. Acta Crystallogr D Biol Crystallogr
Agris PF. 2004. Decoding the genome: A modified view. Nucleic Acids
Res 32: 223–238.
Alian A, Lee TT, Griner SL, Stroud RM, Finer-Moore J. 2008.
Structure of a TrmA–RNA complex: A consensus RNA fold con-
tributes to substrate selectivity and catalysis in m5U methyltrans-
ferases. Proc Natl Acad Sci 105: 6876–6881.
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic
local alignment search tool. J Mol Biol 215: 403–410.
Andersen NM, Douthwaite S. 2006. YebU is a m5C methyltransferase
specific for 16 S rRNA nucleotide 1407. J Mol Biol 359: 777–786.
Andersen TE, Porse BT, Kirpekar F. 2004. A novel partial modifica-
tion at C2501 in Escherichia coli 23S ribosomal RNA. RNA 10:
Bailey S. 1994. The CCP4 Suite: Programs for protein crystallography.
Acta Crystallogr D Biol Crystallogr 50: 760–763.
Behm-Ansmant I, Urban A, Ma X, Yu YT, Motorin Y, Branlant C.
2003. The Saccharomyces cerevisiae U2 snRNA:pseudouridine-
synthase Pus7p is a novel multisite-multisubstrate RNA:C-
synthase also acting on tRNAs. RNA 9: 1371–1382.
Behm-Ansmant I, Branlant C, Motorin Y. 2007. The Saccharomyces
cerevisiae Pus2 protein encoded by YGL063w ORF is a mitochon-
drial tRNA:C27/28-synthase. RNA 13: 1641–1647.
Cameron DM, Gregory ST, Thompson J, Suh MJ, Limbach PA,
Dahlberg AE. 2004. Thermus thermophilus L11 methyltransferase,
PrmA, is dispensable for growth and preferentially modifies free
ribosomal protein L11 prior to ribosome assembly. J Bacteriol 186:
DeLano WL. 2002. The PyMol molecular graphics system . DeLano
Scientific, San Carlos, CA.
Demirci H, Gregory ST, Dahlberg AE, Jogl G. 2007. Recognition of
ribosomal protein L11 by the protein trimethyltransferase PrmA.
EMBO J 26: 567–577.
Demirci H, Gregory ST, Dahlberg AE, Jogl G. 2008a. Crystal structure
of the Thermus thermophilus 16 S rRNA methyltransferase RsmC
in complex with cofactor and substrate guanosine. J Biol Chem
Demirci H, Gregory ST, Dahlberg AE, Jogl G. 2008b. Multiple-site
trimethylation of ribosomal protein L11 by the PrmA methyl-
transferase. Structure 16: 1059–1066.
Demirci H, Belardinelli R, Seri E, Gregory ST, Gualerzi C, Dahlberg
AE, Jogl G. 2009. Structural rearrangements in the active site of the
Thermus thermophilus 16S rRNA methyltransferase KsgA in a bi-
nary complex with 59-methylthioadenosine. J Mol Biol 388: 271–
Douthwaite S, Kirpekar F. 2007. Identifying modifications in RNA by
MALDI mass spectrometry. Methods Enzymol 425: 1–20.
Emsley P, Cowtan K. 2004. Coot: Model-building tools for molecular
graphics. Acta Crystallogr D Biol Crystallogr 60: 2126–2132.
Eswar N, Eramian D, Webb B, Shen MY, Sali A. 2008. Protein
structure modeling with MODELLER. Methods Mol Biol 426: 145–
Foster PG, Nunes CR, Greene P, Moustakas D, Stroud RM. 2003. The
first structure of an RNA m5C methyltransferase, Fmu, provides
insight into catalytic mechanism and specific binding of RNA
substrate. Structure 11: 1609–1620.
Gao YG, Selmer M, Dunham CM, Weixlbaumer A, Kelley AC,
Ramakrishnan V. 2009. The structure of the ribosome with
elongation factor G trapped in the post-translocational state.
Science 326: 694–699.
T. thermophilus 16S rRNA methyltransferase RsmF
Gustilo EM, Vendeix FA, Agris PF. 2008. tRNA’s modifications bring
order to gene expression. Curr Opin Microbiol 11: 134–140.
Gu XR, Gustafsson C, Ku J, Yu M, Santi DV. 1999. Identification of
the 16S rRNA m5C967 methyltransferase from Escherichia coli.
Biochemistry 38: 4053–4057.
Guymon R, Pomerantz SC, Crain PF, McCloskey JA. 2006. Influence
of phylogeny on posttranscriptional modification of rRNA in
thermophilic prokaryotes: The complete modification map of 16S
rRNA of Thermus thermophilus. Biochemistry 45: 4888–4899.
Hallberg BM, Ericsson UB, Johnson KA, Andersen NM, Douthwaite
S, Nordlund P, Beuscher AE 4th, Erlandsen H. 2006. The structure
of the RNA m5C methyltransferase YebU from Escherichia coli
reveals a C-terminal RNA-recruiting PUA domain. J Mol Biol 360:
Hashimoto Y, Yano T, Kuramitsua S, Kagamiyama H. 2001. Disrup-
tion of Thermus thermophilus genes by homologous recombination
using a thermostable kanamycin-resistant marker. FEBS Lett 506:
Helser TL, Davies JE, Dahlberg JE. 1972. Mechanism of kasugamycin
resistance in Escherichia coli. Nat New Biol 235: 6–9.
Hendrickson WA, Horton JR, LeMaster DM. 1990. Selenomethionyl
proteins produced for analysis by multiwavelength anomalous
diffraction (MAD): A vehicle for direct determination of three-
dimensional structure. EMBO J 9: 1665–1672.
Holm L, Ka ¨a ¨ria ¨inen S, Rosenstro ¨m P, Schenkel A. 2008. Searching
protein structure databases with DaliLite v.3. Bioinformatics 24:
Hoseki J, Yano T, Koyama Y, Kuramitsu S, Kagamiyama H. 1999.
Directed evolution of thermostable kanamycin-resistance gene: A
convenient selection marker for Thermus thermophilus. J Biochem
Hur S, Stroud RM. 2007. How U38, 39, and 40 of many tRNAs
become the targets for pseudouridylation by TruA. Mol Cell 26:
Lang PT, Brozell SR, Mukherjee S, Pettersen EF, Meng EC, Thomas V,
Rizzo RC, Case DA, James TL, Kuntz ID. 2009. DOCK 6:
Combining techniques to model RNA-small molecule complexes.
RNA 15: 1219–1230.
Langer G, Cohen SX, Lamzin VS, Perrakis A. 2008. Automated
macromolecular model building for X-ray crystallography using
ARP/wARP version 7. Nat Protoc 3: 1171–1179.
Laskowski RA, MacArthur MW, Moss DS, Thornton JM. 1993.
PROCHECK: A program to check the stereochemical quality of
protein structures. J Appl Crystallogr 26: 283–291.
Lee TT, Agarwalla S, Stroud RM. 2005. A unique RNA fold in the
RumA-RNA-cofactor ternary complex contributes to substrate
selectivity and enzymatic function. Cell 120: 599–611.
Liu Y, Santi DV. 2000. m5C RNA and m5C DNA methyl transferases
use different cysteine residues as catalysts. Proc Natl Acad Sci 97:
Malone T, Blumenthal RM, Cheng X. 1995. Structure-guided analysis
reveals nine sequence motifs conserved among DNA amino-
methyltransferases, and suggests a catalytic mechanism for these
enzymes. J Mol Biol 253: 618–632.
McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC,
Read RJ. 2007. Phaser crystallographic software. J Appl Crystallogr
McLuckey SA, Van Berkel GJ, Glish GL. 1992. Tandem mass
spectrometry of small multiply charged oligonucleotides. J Am
Soc Mass Spectrom 3: 60–70.
Motorin Y, Grosjean H. 1999. Multisite-specific tRNA:m5C-methyl-
transferase (Trm4) in yeast Saccharomyces cerevisiae: Identification
of the gene and substrate specificity of the enzyme. RNA 5: 1105–
Motorin Y, Keith G, Simon C, Foiret D, Simos G, Hurt E, Grosjean H.
1998. The yeast tRNA:pseudouridine synthase Pus1p displays
a multisite substrate specificity. RNA 4: 856–869.
Murshudov GN, Vagin AA, Dodson EJ. 1997. Refinement of macro-
molecular structures by the maximum-likelihood method. Acta
Crystallogr D Biol Crystallogr 53: 240–255.
O’Farrell HC, Scarsdale JN, Rife JP. 2004. Crystal structure of KsgA,
a universally conserved rRNA adenine dimethyltransferase in
Escherichia coli. J Mol Biol 339: 337–353.
Ogle JM, Murphy FV, Tarry MJ, Ramakrishnan V. 2002. Selection of
tRNA by the ribosome requires a transition from an open to
a closed form. Cell 111: 721–732.
Otwinowski Z, Minor W. 1997. Processing of X-ray diffraction data
collected in oscillation mode. Methods Enzymol 276: 307–326.
Pe ´rez-Arellano I, Gallego J, Cervera J. 2007. The PUA domain—a
structural and functional overview. FEBS J 274: 4972–4984.
Purta E, O’Connor M, Bujnicki JM, Douthwaite S. 2008. YccW is
the m5C methyltransferase specific for 23S rRNA nucleotide 1962.
J Mol Biol 383: 641–651.
Selmer M, Dunham CM, Murphy FV 4th, Weixlbaumer A, Petry S,
Kelley AC, Weir JR, Ramakrishnan V. 2006. Structure of the 70S
ribosome complexed with mRNA and tRNA. Science 313: 1935–
Sunita S, Tkaczuk KL, Purta E, Kasprzak JM, Douthwaite S, Bujnicki
JM, Sivaraman J. 2008. Crystal structure of the Escherichia coli 23S
rRNA:m5C methyltransferase RlmI (YccW) reveals evolutionary
links between RNA modification enzymes. J Mol Biol 383: 652–666.
Wimberly BT, Brodersen DE, Clemons WM Jr, Morgan-Warren RJ,
Carter AP, Vonrhein C, Hartsch T, Ramakrishnan V. 2000.
Structure of the 30S ribosomal subunit. Nature 407: 327–339.
Demirci et al.
RNA, Vol. 16, No. 8