ABSTRACT
The preparation of a 4-thiouridine phosphoramidite suitable for RNA synthesis and its subsequent
incorporation into oligoribonucleotides is described. The thiol group is
protected with a 2-cyanoethyl group and the 2'-OH with a 1-(2-fluorophenyl)-4-methoxypiperidin-4-yl function. Thiouridine-containing oligoribonucleotides
were used as 350 nm UV crosslinking probes for the photoaffinity labelling of
RNA binding proteins. Specific crosslinking was demonstrated between the Rev
protein of HIV-1 (as a glutathione S-transferase fusion protein) and its RNA target, the Rev-responsive element. It was not possible to generate crosslinks between the RNA bacteriophage MS2 coat protein and
the initiator stem-loop of the replicase gene, to which it binds. These results are
consistent with the structural data available on both systems.
Oligonucleotides substituted with 4-thiopyrimidines can be used to photo-crosslink nucleic acids to proteins (
1
-
3
). Photo-crosslinking allows specific protein-nucleic acid contacts to be mapped since the reacting groups have
to be within a few Ångstroms of one another (
4
). The long wavelength UV light required for photochemical activation of the 4-thiopyrimidines (between 330 and 360 nm) (
4
) does not excite other non-modified chromophores in nucleic acids or proteins, which absorb between
250 and 280 nm. Therefore, minimal background photodamage would be expected
with these thiobases, a tremendous advantage in photoaffinity labelling
experiments. The utility of 4-thiopyrimidines for photo-crosslinking has led to the development of methods for their
incorporation into chemically synthesized oligonucleotides. However, most
previous work has been in the oligodeoxyribonucleotide series using DNA
containing 4-thiothymidine. Our group (
5
) and others (
6
) have used the 2-cyanoethyl group for the critical protection of the thiol group. However,
other protecting groups and convertible base strategies have also been described (
7
-
10
). Less work has been reported with ribonucleotides and 4-thiouridine, mainly because RNA synthesis is more difficult than DNA. This
is largely due to the problem of finding a suitable protecting group for the 2'-OH. Two protecting groups have emerged:
t
-butyldimethylsilyl (tBDMS), removable with fluoride ions (
11
); 1-(2-fluorophenyl)-4-methoxypiperidin-4-yl (Fpmp), cleavable with acid (
12
). Recently the incorporation of 4-thiouridine into RNA using tBDMS protection for the 2'-OH and a 2-cyanoethyl group for the thiol group has been described
(
2
,
13
,
14
). However, there are no reports of the preparation of thiouridine-containing RNA using the Fpmp strategy. This is described in this
publication and, furthermore, the utility of chemically synthesized RNA containing thiouridine in photoaffinity labelling is demonstrated by experiments with the HIV-1 Rev protein and the coat protein of bacteriophage MS2.
The Rev protein of human immunodeficiency virus type 1 (HIV-1) binds to a sequence in the HIV pre-mRNA termed the Rev-responsive element (RRE) (
15
) and viral replication is dependant upon this interaction. The 234 nt RRE is a
complex structure that binds multiple copies of the Rev protein (
16
). A 29 nt fragment of the RRE, termed minSLIIB, has been defined that specifically binds
to a single Rev monomer, though with lower affinity than the intact RRE (
17
). The minSLIIB fragment forms a stem-loop structure with a bubble close to the base of the stem. A number of
nucleotides within this internal loop and adjacent to it have been implicated
in binding Rev (
2
,
17
,
18
). Following substitution with 4-thiouridine, it has been possible to crosslink a residue adjacent to the
bubble to a Rev fusion protein, although the site of interaction on the protein
has not yet been identified (
2
).
The coat protein of the RNA bacteriophage MS2 represses translation of its
replicase gene by binding to a 19 nt stem-loop structure that encompasses the replicase initiation codon (
19
,
20
). The nucleotides involved in binding have been defined by examining the
effects of base substitutions and also using modified bases (
19
,
21
,
22
). The variants included substitution of a stem-loop uridine with 5-bromouridine (
22
), which was subsequently photochemically crosslinked to Tyr85, suggesting an
important role for this residue in binding (
23
). Results from these experiments are largely consistent with the now published
three-dimensional structure of the coat protein bound to a stem-loop variant (
21
,
24
) and to the unpublished wild-type equivalent (Valegård, personal communication).
2'-
O
-1-(2-fluorophenyl)-4-methoxypiperidin-4-yl-5'-
O
-4,4'-dimethoxytrityl-uridine (2'-
O
-Fpmp-5'-
O
-Dmt-uridine; 50 g, 69 mmol) (I, Figure 1) (
12
,
25
) was dissolved in 500 ml anhydrous pyridine and cooled on ice. Methoxyacetic
anhydride (11.6 g, 73 mmol) was added and the reactants were allowed to warm to
room temperature. After 24 h, 25 ml methanol was added and the products
concentrated under reduced pressure. The residual oil was dissolved in 1 l
dichloromethane and washed twice with 250 ml water. The organic layer was
evaporated and the residue was resuspended in 350 ml diethyl ether, solubilized
with 10 ml dichloromethane and precipitated in 4 l pentane to give 2'-
O
-Fpmp-3'-
O
-(methoxyacetyl)-5'-
O
-Dmt-uridine (II) in quantitative yields. This compound was pure by
silica gel TLC (CH
2
Cl
2
/CH
3
OH, 97:3),
R
f
= 0.20.
This compound was dried and dissolved in 550 ml anhydrous dichloromethane, 200
ml anhydrous benzene. 4-Dimethylaminopyridine (1 g, 7.5 mmol), 2,4,6-triisopropylbenzenesulphonyl chloride (39.5 g, 131 mmol) and anhydrous
N
,
N
-diisopropylethylamine (61.18 g, 474 mmol) were added and the reaction mixture stirred
overnight at room temperature. After this time, silica gel TLC (CH
2
Cl
2
/CH
3
OH, 97:3) showed the complete conversion of starting material (
R
f
= 0.20) to product (
R
f
= 0.40). Dichloromethane (400 ml) was added and the reaction mixture extracted
with water (2 * 400 ml). The organic layer was evaporated and the residual gum was
triturated with 800 ml
n
-pentane to give a granular brown solid which was redissolved in 200 ml
dichloromethane. The product was evaporated to a foam and dried under vacuum
for 24 h, yielding 84 g (80 mmol, the greater than 100% yield is due to this
compound not being purified) crude 2'-
O
-Fpmp-3'-
O
-(methoxyacetyl)-5'-
O
-Dmt-4-
O
-(2,4,6-triisopropylbenzenesulphonyl)-uridine (III), which was used without further purification.
This compound (71 g, 68 mmol) was dissolved in 800 ml anhydrous acetonitrile
containing anhydrous
N
,
N
-diisopropylethylamine (38.14 g, 295 mmol) and 3-mercaptopropionitrile (14.2 g, 163 mmol) and stirred overnight. Silica gel
TLC (CH
2
Cl
2
/CH
3
OH, 95:5) showed full conversion of the starting material (
R
f
= 0.70) to product (
R
f
= 0.37). The reaction mixture was concentrated to an oil, redissolved in 800 ml
dichloromethane, and washed twice with 400 ml water. The organic layer was
evaporated and the residue was fractionated by column chromatography on silica
gel using a gradient of dichloromethane to ethanol/dichloromethane (3:97 v/v).
The appropriate fractions were concentrated to give 2'-
O
-Fpmp-3'-
O
-(methoxyacetyl)-5'-
O
-Dmt-4-(
S
-cyanoethyl)-thiouridine (IV). This compound was dissolved in 300 ml 2 M ammonia in methanol and
stirred for 9 h at room temperature. The products were evaporated under reduced pressure, co-evaporated with 2 * 200 ml toluene, then dried. The residue was fractionated by silica gel
chromatography using ethanol/dichloromethane (3:97) as eluant. The appropriate
fractions were concentrated under reduced pressure and dried under vacuum yielding 31.2 g (41 mmol, 59% from II) 2'-
O
-Fpmp-5'-
O
-Dmt-4-(
S
-cyanoethyl)-thiouridine (V). Elemental analysis of V: C, 65.49; H, 5.68; N, 6.71; F, 2.4;
S, 3.4. Expected: C, 65.69, H, 5.71; N, 6.81; F, 2.31; S, 3.89. 1H NMR of V [delta](CDCl
3
): 1.94-2.17 (4H, m, Fpmp H-aliphatic); 2.86 (2H, t, CH
2
CN); 3.02-3.54 (4H, m, Fpmp H-aliphatic); 3.25 (3H, s, Fpmp-CH
3
O); 3.38 (2H, m, SCH
2
); 3.54 (2H, m, H-5'); 3.80 (6H, s, Dmt-CH
3
O); 4.17-4.68 (3H, m, H-2', 3', 4'); 5.76 (1H, d, H-1'); 6.19 (1H, d, H-5); 6.75-7.38 (17H, m,
aromatic-H); 8.24 (1H, d, H-6).
13
C NMR of V [delta] (CDCl
3
): 18.24 (t,
Compound V (31.2 g, 41 mmol) was dissolved in dry dichloromethane (300 ml) containing
N
,
N
-diisopropylammonium tetrazolide (3.19 g, 16 mmol). 2-Cyanoethyl-bis(
N
,
N
-diisopropylamino)phosphoramidite (13.5 g, 45 mmol) (
26
) was added and the mixture stirred at room temperature for 48 h. A further 5.6 g (19 mmol) phosphitylating agent was added and stirring continued for 24 h more. Dichloromethane (200 ml) was added to the reaction mixture,
which was washed with saturated aqueous sodium bicarbonate (2 * 150 ml). The organic layer was evaporated and the residue was dissolved
in 250 ml diethyl ether and precipitated into 2 l
n
-hexane. The crude product was fractionated by silica gel chromatography using 25:75 (v/v) ethyl acetate/dichloromethane containing 2% (v/v) triethylamine. 18 g (18 mmol, 44%) 2'-
O
-Fpmp-3'-
O
-[2-cyanoethyl]-
N
,
N
-diisopropylphosphoramidite-5'-
O
-Dmt-4-(
S
-cyanoethyl)-thiouridine (VI) was obtained.
31
P-NMR [delta] (CDCl
3
, 85% H
3
PO
4
as external standard): 149.59, 150.31.
Oligoribonucleotide synthesis was performed on an Applied Biosystems 381A
synthesizer. The four standard base Fpmp phosphoramidites and ancillary RNA
synthesis reagents were supplied by Cruachem Ltd (Glasgow, UK). The 4-thiouridine Fpmp phosphoramidite, prepared as above, was filtered through 0.5 [mu]m PTFE filters (Millipore) prior to use. The standard 1.0 [mu]mol DNA synthesis cycle was used with the following modifications.
A longer coupling period for the Fpmp phosphoramidites was required and 360 and
480 s were used for the standard and 4-thiouridine phosphoramidites respectively. A capping period of 40 s was
also used. All syntheses were carried out trityl-off and on completion all the protecting groups (other than the Fpmp) were
removed from non-thiouridine-containing oligoribonucleotides by treatment with 1.5 ml 35% aqueous
ammonia at 70oC for 4 h. This step simultaneously removes the oligomers from the solid support. 4-Thiouridine-containing oligoribonucleotides were initially treated with 1 ml 0.3 M 1,8-diazabicyclo[5.4.0]undec- 7-ene (DBU; Aldrich) in dry acetonitrile for 3
h at room temperature to remove the cyanoethyl group that protects the thiol (
5
). The DBU solution was removed by careful decanting and the controlled pore
glass support was extensively rinsed with acetonitrile. The remaining base
protecting groups were removed and the oligomer cleaved from the solid support
using 1.5 ml 35% aqueous ammonia at 25oC for 45 h. Following the removal of these protecting groups the ammonia
was removed by reducing the volume to ~100 [mu]l on a Savant SpeedVac rotary evaporator. The Fpmp protecting groups were removed using the Fpmp deblocking solution (based on acetic acid) supplied by Cruachem and following the
protocol provided. The oligonucleotides were then neutralized with the neutralizing solution provided as part of this kit.
Fully deblocked oligoribonucleotides were purified by two consecutive runs on a
C18 reverse phase HPLC column (Apex-1-ODS, 5 [mu]m particle size, 4.6 * 250 mm; Jones Chromatography, Llanbradach, Wales). Acetonitrile gradients of 5-22% over 25 min (flow rate 1.5 ml/min) and 5-18% over 40 min (flow rate 1 ml/min) were used for the
first and second runs respectively. 0.1 M Triethylammonium acetate, pH 6.5, was
used to buffer the gradients. Both runs were carried out at 55oC. Following the initial HPLC run the oligoribonucleotide fraction was
dried in a Savant SpeedVac and resuspended in ~250 [mu]l sterile water. The evaporation/resuspension cycle was repeated twice.
After the second HPLC purification the samples were desalted using a Filtron 1K
microsep centrifugal concentrator (Flowgen, Sittingbourne, Kent, UK). The
desalted oligoribonucleotides were stored frozen at -20oC.
Purified RNA was labelled at its 5'-terminus using [[gamma]-
32
P]ATP (3000 Ci/mmol; Amersham) and T4 polynucleotide kinase (Pharmacia) and
subsequently purified on 1.5 mm thick 19% (w/v) polyacrylamide gels containing
7 M urea (
27
). The gel slice containing the labelled oligonucleotide was excised and incubated for 12-16 h at 37oC in 0.1 M Tris-HCl, pH 8, 100 mM NaCl, 1 mM EDTA, 0.1% (w/v) SDS. The solution was decanted and the RNA
precipitated by the addition of 3 vol absolute ethanol. The pellet was
collected by centrifugation, resuspended in 50 [mu]l sterile water and stored at -20oC.
The oligoribonucleotides produced were characterized by sequencing, as previously described (
22
,
28
), using RNase T1 (guanosine specific), RNase U
2
(adenosine specific) and RNase A (pyrimidine specific). Base composition analysis was also carried out as described (
29
). Typically, 2.5 [mu]g oligoribonucleotide were digested with snake venom phosphodiesterase (24 [mu]g) and calf intestinal phosphatase (3 U) for ~1 h. The resulting nucleosides were resolved on a C18 reverse phase column, eluting isocratically with 0.1 M
triethylammonium acetate, pH 6.5, for 10 min before applying an acetonitrile gradient of 0-3% over 10 min. A flow rate of 1.5 ml/min and a temperature of 40oC were used.
The GST-Rev fusion protein and MS2 bacteriophage coat protein were purified and their concentrations determined as previously
described (
30
,
31
).
Filter binding assays were performed as described previously (
32
). The MS2 coat protein was serially diluted using 1 mM acetic acid. Protein
concentrations of between 10
-10
and 10
-5
M were incubated with 20 nM RNA in 75 mM Tris-HCl, pH 8, 100 mM NaCl, 10 mM MgCl
2
for 15 min on ice, prior to filtration through 0.45 [mu]m nitrocellulose filters (Whatman). Filters were rapidly rinsed with 250 [mu]l incubation buffer before drying. Filters were added to 5 ml Ecoscint A
(National Diagnostics) and counted in a liquid scintillation counter. The
K
d
was estimated as the protein concentration at which half-maximal binding of the RNA took place.
UV crosslinking of GST-Rev to RRE RNAs was carried out in 10 mM HEPES-NaOH, pH 7.6, 20 mM NaCl, 150 mM KCl, 2 mM MgCl
2
, 0.5 mM EGTA, 10% (v/v) glycerol, 1 mM DTT, 7 [mu]g/ml BSA and 50 [mu]g type V tRNA (Sigma). GST-Rev (6-12 [mu]M) was pre-incubated in the above buffer at room
temperature for 15 min prior to addition of between 44 and 88 nM radiolabelled
oligoribonucleotide. Following incubation on ice for 15 min the samples were
irradiated at 350 nm in a Rayonet RMR-600 photochemical reactor for 5 min. MS2 coat protein crosslinking was carried out in 75 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10 mM MgCl
2
. MS2 coat protein (1 [mu]M) was pre-incubated with between 20 and 50 nM labelled RNA in this buffer for 30 min on ice, before
long wavelength UV irradiation for 15 min at 8-10oC. Samples were exposed on a piece of Parafilm using a hand-held dual wavelength Mineralight UVGL-25 lamp at a distance of 1 cm. A Pyrex plate was used to
filter out light with a wavelength below 310 nm. To examine potential protein-RNA crosslinks generated by irradiation, the samples were then heated for
5 min at 60oC in 150 mM Tris-HCl, pH 6.8, 3% (w/v) SDS and 12% (v/v) glycerol before electrophoresis on a 10% (w/v) SDS-PAGE gel. Alternatively, to examine potential intramolecular crosslinks, the RNA was subjected to limited
alkaline hydrolysis and examined on 0.75 mm thick 19% (w/v) polyacrylamide gels containing 7 M urea, as previously described (
33
)
The preparation of this compound is shown in Figure
1
. The sulphur atom at the 4 position is introduced directly as its protected cyanoethyl derivative by the well documented activation of the
pyrimidine 4-keto oxygen with a 2,4,6-triisopropylbenzenesulphonyl group, followed by displacement with 3-mercaptopropionitrile. This activating agent has previously been used to
produce the 4-(
S
-cyanoethyl) derivatives of uridine and deoxyuridine (
6
,
13
,
34
). The alternative activating agent,
N
-methylimidazole/phosphorus oxychloride (
5
), has furnished the 4-(
S
-cyanoethyl)-protected 4-thiothymidine. All the reactions in Figure
1
proceeded in reasonable yield apart from the final one, the addition of the
phosphoramidite to the 3'-OH of the fully protected ribonucleoside. The time required for this reaction, 96
h, was considerably longer than the 1 h used for analogous reactions (i.e. bis-phosphoramidites and diisopropylammonium tetrazolide as activator) in the
deoxy series (
26
). Long reaction times (18 h) have also been reported for the preparation of
Fpmp phosphoramidites using the more reactive chloro-phosphoramidites (
12
). With deoxynucleosides these reactions are over in minutes. Presumably the
reaction is slow because of steric hindrance caused by the neighbouring 2'-Fpmp group. Although steric hindrance could also account for the
low yields (<50%; cf. the deoxy series where yields are usually near quantitative), it
may simply be due to not having fully optimized the reaction conditions since >90% yields have been reported for the synthesis of Fpmp phosporamidites using chloro-phosphoramidites (
12
). Although these syntheses have been carried out on large (50 g) scales, it is possible to scale down considerably (by factors of 10-20) with little change in yields.
The oligonucleotides that were prepared are shown in Figure
2
. Those synthesized for studies with the Rev protein were based on the minSLIIB
fragment, 31 bases in length. Those used to investigate the interaction with
the MS2 coat protein are based on a stem-loop 20 bases long. The positions at which 4-thiouridine has been substituted for uridine are illustrated in
Figure
2
. The coupling of all the Fpmp phosphoramidites, including that of the protected 4-thiouridine, took place with yields >97%, as monitored by trityl cation release during the synthesis. Following the synthesis
all the protecting groups were removed. It was important to use milder acid
conditions for removal of the Fpmp groups than that originally recommended, as
has recently been suggested (
35
,
36
). A suitable acidic solution is supplied by Cruachem as part of the Fpmp kit.
Following deprotection all the oligoribonucleotides were purified by reverse
phase HPLC. As an example the purification of [
4S
U]GGGCGCAGCGUCAAUGACGCUGACGGUACA is given, but that of all the other oligoribonucleotides was
similar. An initial HPLC separation showed the desired product as the major
peak, which eluted after a multiplicity of peaks which represent failure
sequences (Fig.
3
). This peak absorbs at 340 nm, due to the presence of 4-thiouridine, as well as 260 nm. A second HPLC run, using a shallower
gradient of acetonitrile, resulted in a further purification of the
oligoribonucleotide (Fig.
3
). The final product showed two minor peaks that eluted prior to the desired 4-thiouridine-containing oligoribonucleotide. However, despite clear resolution of
these from the main peak, re-chromatography of the purified main peak always gave a sample that
contained the two minor peaks. Hence these minor peaks could represent alternative secondary structures of the RNA. minSLIIB is known to have a stable secondary structure, reflected in its high
T
m
value of 82oC (
2
). An alternative HPLC purification scheme, based on an initial isolation of Fpmp-substituted product followed by acid treatment and purification of the
fully deblocked RNA, was also tried. This did not give a better yield or a
higher purity than the method described above. With the earliest syntheses
carried out, RNA yields (based on the amount of purified RNA produced from 1 [mu]mol immobilized 3' base) of 0.8-1% were obtained. However, with an increase in expertise and
the change to milder acid deblocking conditions, current syntheses are giving
improved yields of 3-5%. These latter values can be considered typical of the yields
obtainable, in our hands, for oligoribonucleotides 20-31 bases in length using the Fpmp method.
A number of criteria were used to confirm that the correct product had been synthesized and purified for each of the oligoribonucleotides prepared.
All the 4-thiouridine-containing oligoribonucleotides showed an absorbance at 340 nm, due to the presence of this base.
Furthermore, the ratios of absorbances at 340/260 nm were as expected for one
thiouridine per 31 bases (RRE oligoribonucleotides) or per 20 bases (MS2
oligoribonucleotides) (not shown). When labelled at their 5'-termini with
32
P and analysed by denaturing PAGE all the oligoribonucleotides migrated as a single band of the appropriate length (not shown). Base composition analysis of all
the oligoribonucleotides demonstrated a nucleoside ratio consistent with that
expected and showed the presence of a single equivalent of 4-thiouridine, as shown in Figure
4
for [
4S
U]GGGCGCAGCGUCAAUGACGCUGACGGUACA. Sequencing of the thio- and non-thio-containing RRE oligoribonucleotides generated very similar
patterns. Most of the sequences could be read clearly (not shown) and were
consistent with the expected structures. There were a few discrepancies with
the enzymatic sequencing. However, these were artefactual and could be
explained either by the secondary structure of the RNA or by sequence
compressions in G-rich regions.
Binding of each MS2 oligoribonucleotide to the coat protein was observed using
gel shift or filter binding assays (data not shown) and the
K
d
values found were 0.47 [mu]M for the wild-type oligoribonucleotide and 0.37 and 0.36 [mu]M for oligoribonucleotides with 4-thiouridine substitutions at the -5 and -6 positions respectively. These values are in
good agreement with those previously reported (
21
,
22
). However, no crosslinking was observed between the MS2 RNA stem-loop and the coat protein with 4-thiouridine substitutions at either position (data not shown). The
three-dimensional structure of the MS2 coat protein bound to the wild-type operator stem-loop has recently been determined (Valegård
et al.
personal communication) and this allows a rationalization of the negative
crosslinking result. The stem-loop forms a crescent, one leg of which (nucleotides A
-4
to A
-10
) runs adjacent to the protein (
21
). The uridyl base of U
-6
extends from this RNA strand away from the protein, so would not be expected to
form crosslinks with the protein. Nucleotide U
-5
stacks between nucleotide A
-7
and the amino acid Tyr85 of the coat protein. Previously, it has been
demonstrated that the introduction of 5-bromouridine at this -5 position gives rise to photo-crosslinking to Tyr85 (
23
). This contrasts with the negative results reported here. The photochemistry of 5-bromouridine depends on the excitation wavelength (
40
). At 308 nm, the wavelength used to generate crosslinks to Tyr85, a triplet
excited state is produced which reacts preferentially with tyrosine, tryptophan
and histidine side chains. With tyrosine the carbon atom ortho to the phenolic hydroxyl group couples to the 5 position of the uridine ring with HBr elimination. The photochemistry of 4-thiouridine is less well understood (
4
). There appear to be two main reaction pathways. First, an oxidative pathway in
which nucleophiles such as amines can add to the 4 position with loss of the
sulphur. Second, a radical mechanism involving addition to the 6 position with
reduction of the 5=6 double bond. For this pathway the carbon atoms in
compounds such as 2-propanol or 1-aminobutane become attached to the pyrimidine. However, there have
been very few model studies aimed at elucidating the photoreactions of amino
acids with 4-thiouridine and it is not known if tyrosine reacts efficiently with this
base. The structure of the MS2 protein complexed with RNA (
24
; Valegård
et al.
, unpublished results) shows that the interatomic distances between the C5 position of U
-5
and the two carbons of Tyr85 ortho to the phenolic hydroxyl group are 3.63 and
4.09 Å. As 5-bromouridine interacts preferentially with these atoms, photo-crosslinking in this instance is easy to rationalize. The
only protein atoms within 4 Å of the C4 position of U
-5
are three of the aromatic carbon atoms of Tyr85. As the photoreaction of
thiouridine at the C4 position requires a nucleophile, it is unsurprising that
no photo-crosslinking via this mechanism takes place. Only two Tyr85 atoms, the
phenolic carbon and one of the ortho carbons, are nearer than 4 Å to the C6 of U
-5
. Photoreactivity of thiouridine at C6 involves a radical mechanism and hydrogen
abstraction (
4
). No reaction at the phenolic carbon of Tyr85 is therefore expected. Reaction
could take place at the ortho carbon, but we see no sign of this. It may be
that the thiouridyl radical shows poor reactivity with tyrosine or that the
exact juxtapositon of the C6 and ortho carbons is unfavourable for photo-crosslinking.
In addition, both thiouridine-containing MS2 oligoribonucleotides have been examined for the formation of intramolecular RNA-RNA crosslinks using two methods. The RNAs were irradiated with UV
light either in the presence or the absence of the MS2 coat protein. Following
incubation, the RNAs were partially digested using limited alkaline hydrolysis. No significant differences could be observed in the resulting denaturing gel electrophoresis patterns,
regardless of whether or not the coat protein was present (Fig.
7
). Furthermore, identical electrophoretograms were seen in controls in which no
UV light was applied. This suggests that no intramolecular crosslinking had
occurred. Again, structural data can rationalize these results. U
-6
is not near any other bases in the RNA. Substituting 4-thiouridine into the -5 position of the crystal structure suggests an approximate
distance between the sulphur atom and the O1P of nucleotide -6 of 2.7 Å. It is conceivable that the thiol tautomer of
4S
U
-5
could be hydrogen bonded to that nucleotide. Indeed, the crystal structure of
the stem-loop variant (
24
) suggests an analogous hydrogen bond between the exocyclic amino group of C
-5
and O1P of nucleotide -6, which may explain the 6-fold greater affinity that the C
-5
variant has for the MS2 coat protein. However, although the thiobase is within
crosslinking range, phosphodiesters are poorly nucleophilic and so unlikely to
be partners in the photooxidation mechanism. The only other regions of the RNA
that the heterocyclic ring of C
-5
approaches ( <= 4 Å) is the purine ring of A
-7
, against which it stacks. It has been shown that thiopyrimidines can be photo-crosslinked to adjacent pyrimidines (e.g. for dimers Tp
4S
T, Tp
4S
U and
4S
UpT (
4
,
41
). Furthermore, thiouridine is a natural constituent of many tRNAs, occurring at
position 8, and it can be photo-crosslinked to a cytidine present at position 13 (
4
,
42
) and the near location of these two residues has been confirmed by
crystallography (
43
). All these reactions involve photochemically induced coupling of the two
pyrimidine rings (
4
), but reactions of thiopyrimidines with purines are much less likely.
It has been demonstrated that the Fpmp method can be used to incorporate the
photoreactive base 4-thiouridine into chemically synthesized RNA. Furthermore, the
applicability of this base in photo-crosslinking to RNA binding proteins has been shown. It is clear that the
most critical element for high photo-crosslinking yields is the correct juxtaposition of the base with
appropriate amino acid side chains in the protein and therefore failure to
crosslink does not imply that groups are not in close proximity. As the
thiouridine derivative required for chemical synthesis is commercially available, the method should find wide applicability.
This work was supported by grants from the Wellcome Trust to BAC and PGS and by
grants from the UK BBSRC and the University of Leeds to PGS. We thank Dr Chris
Adams and Mr James Murray for their many helpful discussions on the
deprotection and purification of 4-thiouridine-containing oligonucleotides. We are most grateful to Dr Karin Valegård for supplying interatomic distances of the MS2-wild-type RNA structure before publication.
+
Present address: PerSeptive Biosystems Inc., 500 Old Connecticut Path,
Framingham, MA 01701, USA
REFERENCES
Return