ABSTRACT
Mass spectrometry-based methods have been used to study post-transcriptional modification in the 1900-1974 nt segment of domain IV in 23S rRNA of
Escherichia coli
, a region which interacts with domain V in forming the three-dimensional structure of the peptidyl transferase center within the
ribosome. A nucleoside constituent of
M
r 258 (U*) which occurs at position 1915, within the highly modified
oligonucleotide sequence 1911-
[Psi]
AACU*A
[Psi]
-1917, was characterized as 3-methylpseudouridine (m
3
[Psi]
). The assignment was confirmed by chemical synthesis of m
3
[Psi]
and comparison with the natural nucleoside by liquid chromatography-mass spectrometry. 3-Methylpseudouridine is previously unknown in nature and is the only
known derivative of the common modified nucleoside pseudouridine thus far found
in bacterial rRNA.
Knowledge of the structures and functions of post-transcriptional modifications in ribosomal RNA is, in a phylogenetic
sense, very limited. Twenty eight different modified nucleosides are presently
known in rRNA (
1
), with most detailed work having been carried out on 16S and 23S rRNA from
Escherichia coli
(citations summarized in
2
-
4
) and 28S rRNA from yeast and man (
5
). In contrast, 79 modifications are known in tRNA from a wider range of
organisms, for which functional roles have been dealt with in much greater
detail (
6
-
9
). In the case of rRNA the lower level of current structural information is due
in part to experimental difficulties associated with assignment of correct
identities of modification (as opposed to sites of modification) in large RNAs.
The concept that post-transcriptionally modified nucleosides in RNA are functionally important (
10
) is based, in part, on their location in nucleotide sequences that are very
highly conserved, single stranded and surface exposed. In 23S rRNA in
particular the importance of accurate structural characterization of
modifications as a prerequisite to the ultimate understanding of the higher
order structure of rRNA and its function in translation (
11
) is especially strong for domains IV (nt 1656-2004) and V (nt 2043-2625). Phylogenetic evidence, based on comparative sequence
analysis, indicates that domain IV (and domain II) interacts with domain V in
forming the three-dimensional structure of the peptidyl transferase center within the
ribosome (
12
). This interaction is supported by chemical footprinting experiments, which
have shown that specific bases in domain IV are protected from chemical
modification by both A-site (C-1941) and P-site (A-1916 and A-1918) bound tRNA (
13
). Cross-linking experiments have revealed tertiary interactions within single-stranded regions of domain IV (nt 1911-1921 and 1964) (
14
) and between single-stranded regions of domains IV (nt 1777-1792) and V (nt 2584-2588) (
15
). Furthermore, inter-RNA cross-linking experiments identify the same region of domain IV (nt 1912-1920) with two regions of 16S rRNA (nt 1408-1411 and 1518-1520;
16
), implying that these regions are exposed at the interface between ribosomal
subunits.
General methods based on mass spectrometry (
2
,
4
) that facilitate the characterization of post-transcriptionally-modified nucleosides at the primary structure level have been
developed and recently applied to the identification of modifications in the
central loop of domain V from
E.coli
23S rRNA (
4
). Here we describe the application of these techniques to the characterization
of post-transcriptionally-modified nucleosides found in the1900-1974 nt region of domain IV in
23S rRNA, resulting in identification at position 1915 of 3-methylpseudouridine (Fig.
1
), m
3
[Psi] (symbols and nomenclature used for modified nucleosides are taken from
1
), a nucleoside not previously known in nature.
Authentic 2'-
O
-methylpseudouridine ([Psi]m) was a gift from Prof. M.J.Robins (Brigham Young University, Provo,
UT).
Pseudouridine ([Psi], 2.44 mg, 10 [mu]mol; Sigma) was dissolved in 2 ml anhydrous methanol.
Trimethylsilyldiazomethane (2 M solution in a mixture of hexanes; Aldrich) was
added in 25 [mu]l aliquots five times (125 [mu]l total) over a period of 3 h, while stirring at room temperature (
17
,
18
). The resulting clear colorless solution was analyzed by reversed phase HPLC-mass spectrometry (LC/MS) (
19
), showing a product ~95% pure: HPLC
t
R
, 11.9 min without internal reference; A
254
/A
280
ratio by HPLC, 2.34; [lambda]
max
(H
2
O, pH ~6), 261 nm, [lambda]
max
(pH 13), 283 nm; molecular mass, 258 (MH
+
,
m
/
z
259), corresponding to an isomer of monomethylated uridine. The main product
was collected and dried using a SpeedVac vacuum centrifuge. The reaction yield
of m
3
[Psi] was estimated by UV absorbance measurement to be 40%. The preparation of m
3
[Psi] has been previously reported (
20
) as one of several products formed by treatment of pseudouridine with
diazomethane, but due to the stated (
20
) tentative nature of the structure assignment, care was taken in the present
study to ensure placement of the methyl group in the synthetic product at N-3, rather than N-1 or in ribose, by mass spectrometry. The collision-induced dissociation mass spectrum of the protonated synthetic
molecule (MH
+
, 259) produced by fast atom bombardment ionization of the synthetic product was
determined using a VG-70SEQ tandem mass spectrometer. Dissociation of the uracil ring occurs by
a retro-Diels Alder type reaction in which N(3), C(4) and O
4
are expelled as HNCO (N-3-unsubstituted) or CH
3
NCO (N-3 methylated) (
21
). The common fragment ion (base + CH
2
)
+
, consisting of the base moiety plus C-1' (
22
), was used to monitor the N(3),C(4),O
4
loss reaction, which occurs in [Psi] as 125
+
-> 82
+
. In the mass spectrum of synthetic m
3
[Psi] this process was observed as loss of 57 mass units (CH
3
NCO), 139
+
(65% rel. int.) -> 82
+
(72% rel. int.), confirming methylation of [Psi] at N-3. An ion corresponding to loss of HNCO was observed (
m
/
z
96, 11% rel. int.), in close accord with studies of the dissociation of
isotopically labeled uracil, in which N(1),C(2),O
2
is lost in an alternate minor pathway (
21
).
50S ribosomal subunits were prepared from frozen
E.coli
MRE 600 cells (1/2 log phase; Grain Processing Corp., Muscatine, IA) (
23
). Ribosomal RNA (23S and 5S) was isolated from 50S subunits by extraction with
phenol and chloroform (
23
). 23S rRNA was separated from 5S rRNA by precipitation of the former in the
presence of 2 M LiCl. Purity of the 23S rRNA isolate was assessed by gel
electrophoresis (1% agarose, 3.5 * 50 mm tube) using an Applied Biosystems Model 230A Micro-Preparative Electrophoresis System.
A deoxyoligonucleotide complementary to the
E.coli
23S rRNA sequence from positions 1900 to 1974 (75mer) was synthesized on a 1.0 [mu]mol scale using an Applied Biosystems Model 380B DNA Synthesizer at the
University of Utah Protein/DNA Core Facility. The DNA 75mer was purified by a
DEAE anion exchange HPLC method, as described (
4
). Following protection of 10 nmol rRNA (residues 1900-1974) by the DNA 75mer unprotected RNA was digested with mung bean
nuclease and the RNA 75mer was isolated as previously detailed (
4
).
An aliquot of the RNA 75mer (0.3 nmol) was completely hydrolyzed to nucleosides
using nuclease P1 (Gibco BRL), venom phosphodiesterase (Sigma) and alkaline
phosphatase (Calbiochem) (
24
). The remaining portion of the RNA 75mer (2.7 nmol) was digested using RNase T
1
(2000 U; Ambion) for 30 min at 37oC.
Anion exchange chromatography (DEAE) as previously described (
4
) was used for purification of the synthetic DNA 75mer, the resulting RNA 75mer
and the oligonucleotides resulting from complete RNase T
1
hydrolysis of the RNA 75mer. The RNA oligonucleotide fractions were further
purified prior to mass spectrometry by reversed phase HPLC with a Supelcosil LC-18S column (4.6 * 250 mm) and pre-column (4.6 * 20 mm) using gradient conditions as described (
2
).
Analysis of nucleosides in enzymatic digests was carried out using a combined
HPLC-mass spectrometer in which nucleosides separated by HPLC were passed
through a UV detector and into a mass spectrometer, as earlier detailed (
19
,
25
). The mass spectrometer consists of a non-commercial quadrupole mass analyzer with thermospray interface (Vestec
Corp, Houston, TX), controlled by a Vector/One data system (Teknivent Corp., St
Louis, MO). Reversed phase HPLC separation of nucleosides was carried out using
a gradient elution system based on buffered NH
4
OAc (
19
). Thermospray ionization mass spectra were acquired by scanning the mass range
110-350 every 1.7 s during the 35 min chromatographic separation.
Identification of nucleosides was based principally on: (i) relative retention
time, compared with values obtained from reference nucleosides acquired under
standardized operating conditions (
19
); (ii) characteristic mass values for the protonated base and protonated
molecule (MH
+
). These values (
19
) for the nucleosides of interest in the present study are as follows [retention
time in min (mass protonated base/mass MH
+
)]: [Psi], 3.5 (base ion absent/245); m
1
[Psi], 8.6 (base ion absent/259); [Psi]m, 11.9 (base ion absent/259); m
5
C, 12.4 (126/258); m
5
U, 15.4 (127/259); Um, 17.0 (113/259); m
3
U, 17.1 (127/259). The procedures for interpretation of data for
characterization of nucleosides in RNA hydrolysates have been described (
19
,
25
).
Oligonucleotides from RNase T
1
hydrolysis of the 75mer fragment from
E.coli
23S rRNA were dissolved in H
2
O and diluted with CH
3
OH to 1:9 (v/v) to a final sample concentration of ~2.7 pmol/[mu]l. This concentration is calculated assuming quantitative recovery from
two chromatographies, which is unlikely; the value therefore represents an
upper limit to the amount analyzed. Samples were continuously infused at a flow
rate of 2.0 [mu]l/min, using a Harvard model 22 syringe pump, into the electrospray ion
source of a Sciex API III+ mass spectrometer (Thornhill, Ontario, Canada).
Negative ion mass spectra were acquired, with the instrument operated with mass
analyzer-3 in r.f.-only mode. Calibration of the mass scale was accomplished using
synthetic oligodeoxynucleotides as mass reference standards (
2
).
An analysis of post-transcriptionally-modified nucleosides present in a total hydrolysate of
E.coli
23S rRNA using combined LC/MS has recently been reported (
4
). During the course of this LC/MS analysis the presence of a previously
unidentified ribonucleoside species of
m
/
z
259 was detected both as a nucleoside and as a component of an adenosine-containing dinucleotide (
4
). We now describe the identification of this novel ribonucleoside, including
confirmation by chemical synthesis, and demonstrate the sequence location of
this nucleoside using a nuclease protection strategy (
4
) in combination with mass spectrometric methodologies developed in our
laboratory (
2
).
An RNA 75mer constituting nt 1900-1974 was prepared as described in Materials and Methods. The yield of the
RNA 75mer (3.0 nmol) was determined by UV absorbance at 260 nm. An aliquot (10%
of the sample) was hydrolyzed to its nucleoside substituents and the resultant
hydrolysate was analyzed by LC/MS to give the chromatogram shown in Figure
2
. In addition to the four major ribonucleosides, three post-transcriptionally-modified species of known structure were found: pseudouridine ([Psi], retention time
t
R
, 3.55 min), 5-methylcytidine (m
5
C,
t
R
, 11.85 min) and 5-methyluridine (m
5
U,
t
R
, 14.95 min). The detection of these modified nucleosides is consistent with the
modifications known to occur in this region: [Psi]-1911 and [Psi]-1917 (
26
,
27
), m
5
U-1939 (
26
,
27
) and m
5
C-1962 (
26
,
28
). Their identification by LC/MS provides rigorous identification of the
structure of each of the three nucleoside species. Interestingly, a fourth post-transcriptionally-modified species was detected, eluting 0.7 min before m
5
C (see Fig.
2
) and corresponding to a
t
R
of 11.7 min, referenced internally to the elution time of adenosine (19.9 min;
19
). The mass spectrum recorded at this elution time consisted principally of an
ion of
m
/
z
259 (data not shown), which was interpreted as the protonated molecular ion (MH
+
) of a monomethylated uridine derivative (
M
r
258), corresponding to a nucleoside previously detected by LC/MS in a total
digest of 23S rRNA (
4
). The absence of a corresponding base fragment ion in the mass spectrum is
characteristic of C-nucleosides, e.g. [Psi] and its derivatives (
19
), and suggests that the unidentified nucleoside is a methylated pseudouridine.
The reconstructed ion chromatogram (RIC) shown (inset of Fig.
2
) confirms that the ion of
m
/
z
259 corresponds to the unknown nucleoside (after minor allowance for a 6 s
offset due to the transit time between the UV detector and mass spectrometer).
Other ions of
m
/
z
259 shown in the RIC correspond to the
13
C isotope peak of m
5
C (MH
+
258) and MH
+
of m
5
U at their characteristic retention times (
19
). Based on their chromatographic retention times (
19
), the following known methylated uridine derivatives were excluded as possible
candidates; 1-methylpseudouridine (m
1
[Psi]), 3-methyluridine (m
3
U), m
5
U and 2'-
O
-methyluridine (Um). The latter three compounds had also been excluded as
the identity of U*-1915 on the basis of HPLC data by Smith
et al
. (
28
).
The RNA 75mer (2.7 nmol) was enzymatically hydrolyzed with RNase T
1
and the resultant oligonucleotide mixture was fractionated using DEAE anion
exchange HPLC (data not shown). The oligonucleotide fractions were dried in a
vacuum centrifuge, dissolved in H
2
O and re-chromatographed by reversed phase HPLC (data not shown) in order to
facilitate subsequent analysis by electrospray mass spectrometry. Three
modification-containing oligonucleotides (listed in Table
1
) were identified based (
29
) on accurate molecular mass measurement. The presence of post-transcriptional modification is revealed by incremental increases in mass
(e.g. net increase of +14 Da for CH
3
) not predicted by the rDNA sequence (
2
). For example, the mass spectrum presented in Figure
3
shows four multiply-charged ions (at
m
/
z
504.0, 588.1, 705.9 and 882.6), corresponding to the loss of seven to four
protons from the neutral oligonucleotide, and from which the
M
r
value was derived as 3534.67. Based on the experimentally determined molecular
mass of this oligonucleotide, the base composition was determined (
29
) to be C
2
U
3
A
5
G>p + CH
2
(>p denotes a 2',3'-cyclic phosphate) (calculated
M
r
3534.16). Similarly, solely from accurate molecular weight measurement, the
base composition and net modification of the two other oligonucleotides (see
Table
1
) were determined to be C
2
UAGp + CH
2
and C
2
U
4
A
3
Gp + CH
2
respectively.
An aliquot (0.27 nmol) of the 11mer (C
2
U
3
A
5
G>p + CH
2
) was enzymatically hydrolyzed to its nucleoside substituents and analyzed by
LC/MS. The resultant chromatogram (data not shown) revealed that this
oligonucleotide was composed of [Psi], C, G, A and the unidentified methylated uridine derivative, eluting at
the same position shown for U* in Figure
2
. Enzymatic conversion of U -> [Psi] is the only post-transcriptional modification not manifested by a concomitant
change in molecular mass and so the presence of [Psi] in the 11mer is not reflected in the composition determined by
oligonucleotide mass measurement (Table
1
). Therefore, the U
3
component of the 11mer established by molecular mass measurement is shown by
LC/MS to be composed of [Psi] and U*, with no unmodified U.
Two additional aliquots (0.27 nmol each) of the oligonucleotide 11mer were
enzymatically hydrolyzed for further LC/MS analysis. In consecutive analyses
one aliquot was co-injected with authentic [Psi]m while the other aliquot was co-injected with authentic m
3
[Psi]. The resultant chromatograms are presented in Figure
4
. Partial chromatographic resolution of U* from authentic [Psi]m demonstrates only that the modified residue is not [Psi]m (Fig.
4
A). The elution time of the isomer m
1
[Psi] is significantly earlier when using the same gradient system, 8.6 min (
19
). Retention times of other monomethyl uridines are: m
3
U, 17.1 min; m
5
U, 15.4 min; Um, 17.0 min (
19
). Co-elution of U* with authentic m
3
[Psi] identifies the modified nucleoside as m
3
[Psi] (Fig.
4
B). A
254
/A
280
absorbance ratios by HPLC: U*, 2.2 +- 0.2; synthetic m
3
[Psi], 2.34.
An aliquot (0.27 nmol) of the 10mer (C
2
U
4
A
3
Gp + CH
2
) was enzymatically hydrolyzed to its nucleoside substituents and analyzed by
LC/MS to verify the base composition determined directly from molecular mass
measurement. The resultant chromatogram (shown in Fig.
5
) revealed that the modified residue was m
5
U. The composition of this oligonucleotide species was found to be C
2.04
U
2.77
G
1.00
A
2.84
+ m
5
U, based on comparison of molar response ratios from UV detection compared with
authentic standards (except for m
5
U). These results are consistent with earlier work in which the nucleotide
sequence was deduced as AAAm
5
UUCCUUG (
27
,
30
), confirmed by the corresponding gene sequence (
31
) to represent 23S rRNA nt 1936-1945.
An aliquot (0.27 nmol) of the 5mer (C
2
UAGp + CH
2
) was enzymatically hydrolyzed to its nucleoside substituents and analyzed by
LC/MS to verify the base composition determined directly from molecular mass
measurement. The resultant chromatogram (shown in Fig.
6
) revealed that the modified residue was m
5
C. The composition of this oligonucleotide species was measured as C
1.06
U
0.99
G
1.00
A
1.04
+ m
5
C based on comparison of molar response ratios from UV detection compared with
authentic standards (except for m
5
C). These data, with the corresponding gene sequence (
31
), corroborate the structure of the oligonucleotide as 1960-ACm
5
CUG-1964, reflecting identification of position 1962 as m
5
C (
28
) and correction (
28
) of earlier incorrect sequences (
26
,
30
) which had been determined prior to availability of the gene sequence.
The case for establishment of a complete modification map of 23S rRNA is 3-fold: first, accumulating evidence which demonstrates a role for 23S rRNA
in translation (
11
,
32
); second, conceptual arguments that modified nucleosides are functionally
important (
10
), principally as borne out by studies of tRNA (
6
,
8
); third, the striking fact that nearly all modified nucleosides in 16S and 23S
rRNA are concentrated near the functional center of the ribosome (
33
,
34
). Correct assignment of positions and, especially, identities of post-transcriptional modifications in rRNA has been problematical (discussed in
1
). As a consequence, in spite of the availability of a large number of rRNA gene
sequences (see for example
35
), not one fully processed 23S RNA sequence is known at the RNA level and only
one 16S structure is probably complete, with the recent placement of [Psi]-516 in
E.coli
(
3
).
Structural characterization of modified nucleosides is particularly difficult
if, as in the present case, the nucleoside is previously unknown. Mass
spectrometry has evolved as a particularly important approach in such work (
36
), because of its relatively high sensitivity to subtle structural features and
the advantages that accrue from direct coupling to HPLC. The fact that U-1915 was probably a methylated uridine derivative was known from earlier
work demonstrating incorporation of [
14
C]methyl groups (
26
,
27
; cf. discussion in
28
). Furthermore, treatment of synthetic 23S RNA transcripts with a partially
purified enzyme fraction from
E.coli
resulted in [Psi]-1915 formation (footnote 2 in
37
). Thus, from earlier work there was reason to believe that U*-1915 was a modified [Psi] which contained a post-transcriptionally added carbon atom derived from a methyl
group. Reverse transcription was strongly inhibited at position 1915 (
34
), indicative of a uracil substitution at a position inhibitory to normal base
pairing. In principle, the N(1)-O
2
group would still be available for hydrogen bond pairing. However, the observed
reverse transcriptase stop is consistent with recent NMR results (
38
) showing that N(1)-H of [Psi] is involved in hydrogen bonding to the phosphate backbone. Thus the
normal
anti
conformation of in a [Psi]-A base pair would be unable to switch to the required
syn
conformation. Interestingly, inhibition of reverse transcription serves to
distinguish m
3
[Psi] from the ribose methylated isomer [Psi]m, a pair which is otherwise differentiated by ~0.25 min difference in HPLC retention time (Fig.
4
A). As in the present case, mass spectrometry distinguishes derivatives of U and
[Psi] by the clear absence of glycosidic bond cleavage in the latter, due to
greater strength of the C-C (versus C-N) bond (
19
,
25
).
Although pseudouridine itself is widely distributed in rRNA (
39
), 3-methylpseudouridine is the only derivative of [Psi] thus far found in bacterial rRNA (
1
). Eukaryotic rRNA additionally possesses m
1
[Psi] (
40
), [Psi]m (
41
,
42
) and the hypermodified derivative m
1
acp
3
[Psi] (
43
,
44
). It is worth noting that
Bacillus subtilis
and
Zea mays
chloroplast LSU rRNA show a strong reverse transcriptase stop at the position
equivalent to m
3
[Psi]-1915 (
39
).
The functional role of m
3
[Psi] is unknown, even though it occurs in a very highly conserved section of
domain IV which is structurally remarkable in containing two pseudouridines, [Psi]-1911 and [Psi]-1917 (
26
,
27
), as shown in Figure
7
. Pseudouridine has been speculated to play a role in the peptidyl transferase
reaction (
45
,
46
) and more recently has been demonstrated by NMR to confer significantly
increased stabilization to A-form helical structure through enhanced base stacking and an increased
population of the C3'-endo conformer (compared with U) (
38
). As pointed out (
38
), these modifications also provide a means for control of regional
hydrophobicity in RNA loops, by a decrease (U -> [Psi]) or increase ([Psi] -> m
3
[Psi] or m
1
[Psi]) in hydrophobic character of the base.
This work was supported by National Institutes of Health grant GM29812. The
University of Utah Protein/DNA Core Facility is supported by grant 5P30CA42014
from the National Cancer Institute. The authors are indebted to Robin Gutell
for helpful discussions and to Morris J.Robins for a generous gift of 2'-
O
-methylpseudouridine.
REFERENCES
Return