ABSTRACT
NMR methods were used to investigate a series of mutants of the pseudoknot
within the gene
32
messenger RNA of bacteriophage T2, for the purpose of investigating the range
of sequences, stem and loop lengths that can form a similar pseudoknot
structure. This information is of particular relevance since the T2 pseudoknot
has been considered a representative of a large family of RNA pseudoknots
related by a common structural motif, previously referred to as `common
pseudoknot motif 1' or CPK1. In the work presented here, a mutated sequence
with the potential to form a pseudoknot with a 6 bp stem2 was shown to adopt a
pseudoknot structure similar to that of the wild-type sequence. This result is significant in that it demonstrates that
pseudoknots with 6 bp in stem2 and a single nucleotide in loop1 are indeed
feasible. Mutated sequences with the potential to form pseudoknots with either
5 or 8 bp in stem2 yielded NMR spectra that could not confirm the formation of
a pseudoknot structure. Replacing the adenosine nucleotide in loop1 of the wild-type pseudoknot with any one of G, C or U did not significantly alter the
pseudoknot structure. Taken together, the results of this study provide support
for the existence of a family of similarly structured pseudoknots with two
coaxially stacked stems, either 6 or 7 bp in stem2, and a single nucleotide in
loop1. This family includes many of the pseudoknots predicted to occur
downstream of the frameshift or readthrough sites in a significant number of
viral RNAs.
A pseudoknot is a structural element of RNA formed when a stretch of nucleotides
within a single-stranded loop region base pairs with a complementary sequence outside that
loop (for reviews, see
1
-
5
). Although 14 types of topologically distinct pseudoknots are possible
according to this broad definition (
3
), most of the pseudoknots documented today are of the so-called H(airpin)-type, in which a stretch of nucleotides in the loop of a hairpin and
adjacent to the stem region base pairs with a complementary region outside of
that hairpin. Since the studies by Pleij and co-workers proposing the presence of an H-type pseudoknot at the 3' end of turnip yellow mosaic virus (TYMV) RNA (
6
-
7
), a large body of evidence has been accumulated indicating that pseudoknots
within messenger RNAs play an important role in a variety of critical
biological processes, such as regulation of protein expression in viral systems
by ribosomal frameshifting or readthrough. The widespread occurrence of the H-type pseudoknots suggests that they may have common structural features
which minimize the energy of their folding, thus forming a basis for their
frequent occurrence in natural systems. Coaxial stacking of the two separate
helical stems to form a single pseudo-continuous double helix has been proposed as a stabilizing feature in
pseudoknots (
7
). This type of tertiary interaction has been confirmed by NMR studies on a
model oligoribonucleotide (
8
) and more recently on a natural sequence RNA (
9
). It appears likely, therefore, that coaxial stacking of the two stem regions
represents a general recurring theme in the folding of the H-type pseudoknots.
A profound consequence of the coaxial stacking of the two helical stems (S1 and
S2) is the generation of two unequal connecting loops (L1 and L2), with L1
crossing the deep major groove of stem S2 and L2 crossing the shallow minor
groove of stem S1 (
7
). This imposes different constraints on the length of each of the two loops. A
study of the loop size requirements on a series of model oligoribonucleotides
with the potential to form a pseudoknot with a 3 bp stem1 (S1) and 5 bp stem2
(S2) showed that the minimum loop lengths were 3 nt for L1 and 4 nt for L2 (
10
). However, the minimum lengths of the loops are dependent on the number of base
pairs in each of the stems. In our recent NMR study of the pseudoknot from the
gene
32
mRNA of bacteriophage T2, we found that loop1 consists of only a single
nucleotide, which crosses the major groove of the 7 bp stem2 (
9
). An analysis of reported pseudoknot-forming sequences revealed that several features of the bacteriophage T2
pseudoknot are frequently repeated in naturally occurring pseudoknots: while
the lengths of stem1 and loop2 are varied, the length of stem2 is often 6 or 7
bp, and loop1 often contains only a single adenosine nucleotide (
9
). It is particularly interesting that in an ideal A-form RNA helix, the distance between two phosphate groups across the deep
major groove reaches a minimum when 6 or 7 bp are bridged (Fig.
1
) (
7
). This `coincidence' immediately suggests a general theme for the tertiary
folding of naturally occurring H-type pseudoknots in which a minimal number of nucleotide(s) is used to
span the deep major groove of 6 or 7 bp helical stem.
RNA molecules with the sequences shown in Figure
2
were transcribed using T7 RNA polymerase and synthetic DNA templates, as
previously described (
9
). The DNA templates consisted of a double-stranded 18 bp T7 promoter sequence and a single-stranded coding sequence. The RNA was then separated from
transcripts of incorrect size by electrophoresis on 20% polyacrylamide gels
under denaturing conditions (8 M urea). RNA was visualized by UV shadowing, and
removed from the gel using a BioRad model 422 electroeluter. The RNA was
further purified by repeated ethanol precipitation, and finally passed through
a Sephadex G25 gel filtration column in 1 mM phosphate buffer and lyophilized.
A typical yield was 1 mg of purified RNA/10 ml transcription volume.
NMR spectra of the RNA sequences were collected at 500 MHz using a Varian Unity-Inova spectrometer. Samples typically contained 10-15 mg of RNA dissolved in 550 [mu]l of 10 mM Na/K phosphate buffer in 90% H
2
O/10% D
2
O at pH 6.8. Two-dimensional nuclear Overhauser effect (NOE) spectra were acquired using
the jump and return method (
16
) with a mixing time of 280 ms. Two-dimensional NMR spectra were acquired in the phase-sensitive mode, typically acquired with 512 blocks of 1024 complex
points, using a sweep width of 12 204 Hz in each dimension. Two-dimensional data sets in 90% H
2
O/10% D
2
O solvent were acquired at 10oC. Spectra were referenced to the solvent water resonance at 4.90 p.p.m. at
10oC.
Variants of the natural sequence of the bacteriophage T2 pseudoknot were
prepared and investigated, with sequences illustrated in Figure
2
. Nucleotides are numbered to be consistent with the wild-type sequence, and the mutants are named as indicated in the figure. In
each case, `dangling' nucleotides were included on the 3' and 5' ends of the pseudoknots. We have found that these nucleotides
stabilize the pseudoknots, most likely through base stacking interactions, and
should be considered as an integral part of the pseudoknot structure (
9
,
11
). Results for each of the RNAs will be discussed in turn.
A sequence with the potential to form a pseudoknot with 6 bp in stem2 was
prepared by deleting 1 bp from the center of the seven base pair stem2 of the
wild-type bacteriophage T2 gene
32
mRNA pseudoknot. This potential pseudoknot-forming sequence was termed PK-STEM6 (Fig.
2
). As was the case for the wild-type pseudoknot, this mutant yielded high-quality NMR spectra with sharp and disperse resonances (Fig.
3
). Assignment of the imino proton resonances was straightforward using the two-dimensional NOE spectrum (Fig.
4
a). Sequential imino-to-imino NOE cross peaks were observed for the imino protons of the
Watson-Crick base pairs in the helical stems, including the closing base pair of
stem2. Of particular significance is the NOE cross peak across the junction of
the two stems (labeled G16-U28). This NOE cross peak, along with the cross peaks between imino
protons and amino, ribose H1', adenosine H2 and pyrimidine H5 protons (data not shown), indicates that
the two helical stems are stacked coaxially to form a single pseudo-continuous A-form helix. The striking feature that loop1 consists of only a
single nucleotide, A8, is particularly clear since the adjacent nucleotides (C7
and G9) are involved in stable G-C base pairs that are well defined by the NMR data. Although the chemical
shifts of some of the protons (especially those belonging to the residues
immediately adjacent to the deleted base pair) in the PK-STEM6 sequence are somewhat different from those of the wild-type sequence, the overall NOE connectivities indicate that the PK-STEM6 sequence adopts a pseudoknot structure which is
remarkably similar to the wild-type bacteriophage T2 pseudoknot. These NMR results provide the first
direct structural evidence for the CPK1 pseudoknots with a 6 bp stem2.
Two additional mutants of the wild-type bacteriophage T2 pseudoknot were constructed to further investigate
the influence of the length of stem2 on the tertiary folding of the pseudoknot.
In mutant PK-STEM5, 2 bp were deleted from the wild-type pseudoknot, and in mutant PK-STEM8, an extra G-C base pair was inserted between U13-A30 and A12-U31 (Fig.
2
). The imino proton region of the one-dimensional NMR spectra of each mutant contained resonances with some
chemical shifts similar to resonances of stem1 and stem2 of the wild-type pseudoknot (Fig.
3
). However, the two-dimensional NOE spectra could not confirm the formation of a pseudoknot
structure, and can perhaps be best accounted for by the presence of more than
one RNA conformation in solution. Therefore, we are unable to provide clear
evidence that a stable pseudoknot can be formed with 5 or 8 bp in stem2 while
maintaining a single nucleotide in loop1. This is consistent with the results
of previous studies of pseudoknot-forming RNA sequences, in which a minimum of 2 or 3 nt were shown to be
required in loop1 when stem2 contained 5 bp (
10
).
In our previous work proposing the existence of the CPK1 family of pseudoknots,
we noted that a substantial majority of the predicted CPK1 pseudoknots had a
single adenosine nucleotide in loop1 (
9
). This observation raised the possibility that the loop1 adenosine may be
conserved for structural or functional reasons. Structurally, an adenosine
residue may be favored as the loop1 nucleotide if it participates in a specific
tertiary interaction with the residues of the helical stem2. To investigate
this possibility, mutants of the wild-type bacteriophage T2 pseudoknot were prepared where the loop1 adenosine
was replaced by G, C or U (designated PK-G8, PK-C8 and PK-U8).
The one-dimensional imino proton spectra of the PK-G8, PK-C8 and PK-U8 mutant sequences are similar to the wild-type pseudoknot (Fig.
3
, noted that the spectrum of PK-C8 is shown as a representative) and this observation immediately suggests
that each of these mutants is similar in structure to the wild-type pseudoknot. The base pairings within the mutant pseudoknots were
confirmed using two-dimensional NOE spectra, where the now familiar patterns of sequential
imino-to-imino proton NOEs are observed through both stems, and across the
junction of the stems (Fig.
4
b, only the spectrum of PK-C8 is shown). As was the case in the wild-type and PK-STEM6 pseudoknots, the NOE cross peak between the imino
protons of nucleotides G16 and U28 (Fig.
4
b), as well as cross peaks to amino, ribose H1', adenosine H2 and pyrimidine H5 protons, provides strong evidence that
the two stems are coaxially stacked. These NMR data indicate that substitution
of the loop1 adenosine by any one of the other three nucleotides (G, C or U) is
tolerated, without disruption of the overall tertiary folding of the
pseudoknot.
The present study provides further evidence for the existence of a family of
similarly structured RNA pseudoknots, containing coaxially stacked stems and a
1 nt loop1 that spans the major groove of a 6 or 7 bp stem2. While
pseudoknotted structures could not be unambiguously detected for the PK-STEM5 and PK-STEM8 RNAs, the PK-STEM6 sequence was shown to adopt a structure similar to that
of the wild-type T2 bacteriophage pseudoknot. This result is significant in that it
demonstrates that the naturally occurring pseudoknots predicted to have a base
pairing arrangement of 6 bp in stem2 are indeed feasible. Many of these
naturally occurring pseudoknots are of critical importance since they are
associated with ribosomal frameshifting processes in retroviruses and other RNA
viruses (Fig.
5
). One such pseudoknot, located downstream of the
gag-pro
frameshift site in simian retrovirus type-1 (SRV-1), has been proposed to have 6 bp in each of the two stems and a
single adenosine nucleotide in loop1 (
9
,
12
), a model supported by the results of the present mutational-structural studies, as well as recent NMR studies conducted in our
laboratory (manuscript in preparation).
Regarding the apparent common occurrence of an adenosine nucleotide in loop1 of
many of the predicted CPK1 pseudoknots (
9
and Fig.
5
herein), the present studies show that any 1 of the other 3 nt (G, C or U) can
serve as the loop1 nucleotide while preserving the pseudoknot structure. This
result argues against the loop1 nucleotide being involved in a tertiary
interaction essential for pseudoknot formation; merely embedding the relatively
hydrophobic base(s) of the loop1 nucleotide(s) in the major groove of stem2 may
provide a substantial stabilizing force for the pseudoknot folding. The present
results are consistent with a recent study investigating the role of the
pseudoknot component in regulating the SRV-1
gag-pro
ribosomal frameshifting efficiency, where ten Dam and co-workers found that substitution of the adenosine residue in loop1 of the
SRV-1 pseudoknot by any 1 of the other 3 nt (G, C or U) had little effect on
the frameshifting efficiency (
13
). In addition, the naturally occurring frameshift-associated pseudoknot of the coronavirus avian infectious bronchitis virus
(IBV) (
14
-
15
) is predicted to contain a single guanosine as the loop1 nucleotide (
12
and Fig.
5
herein). In light of these previous studies and the present structural
investigations, it seems that there is little structural or functional basis
for the conservation of adenosine as the loop1 nucleotide in the CPK1
pseudoknots. Two possible explanations that may account for the previously
observed `conservation' of the loop1 adenosine are: (i) lack of sampling; or
(ii) the stronger tendency for an adenosine residue to appear in the single-stranded regions of folded RNA molecules. Of course, it can not be ruled
out that an adenosine residue might be favored due to thermodynamic or folding
considerations. A systematic investigation of the influence of loop1 nucleotide
identity on the pseudoknot stability and unfolding pathway is in progress. We
have previously noted, however, that the unfolding of the wild-type bacteriophage gene
32
mRNA pseudoknot is quite complex (
9
,
11
).
The combined structural and phylogenetic evidence suggests that the CPK1 motif
represents a naturally preferred common structural theme for the tertiary
folding of RNA pseudoknots (
9
and Fig.
5
herein). It is therefore relevant to further explore the rationale underlying
the apparent popularity of this specific motif. The preference of 6 or 7 bp in
stem2 most likely has its basis in the fact that the distance across the major
groove of the helical stem reaches a minimum when 6 or 7 bp are bridged (Fig.
1
). However, this minimum distance only explains why it is
possible
for a single nucleotide to span the major groove of a 6 or 7 bp stem, it does
not automatically explain why a 1 nt connection appears to be
preferred
. Perhaps it is the limited space within the major groove of stem2 that
restricts the number of nucleotides that are allowed in loop1; the presence of
additional loop1 nucleotides may force the relatively hydrophobic bases to be
exposed to the solvent. It is also noted that when stem2 contains 6 or 7 bp,
corresponding to slightly more than one-half helical turn, the loop1 nucleotide is in a optimum position to embed
its base within the deep major groove with the least possibility of interfering
with the elements (bases, riboses and phosphates) of stem2 (Fig.
1
).
The present query into the underlying rationale of the CPK1 family of
pseudoknots reveals that several specific structural features of the A-form RNA helix may work in concert to give rise to the specific folding of
this motif. With its distinctive structural features and apparent biological
relevance, the CPK1 motif is a significant entry into the rapidly growing
structural database of RNA molecules and will provide valuable structural
information for related biophysical and biochemical studies in the future.
We thank Dr David P. Giedroc for many helpful discussions. This work was
supported by NIH grant R01-AI40187 (to D.W.H. and D.P.G.), and is in partial fulfillment of the
requirements for the Ph.D. degree at the University of Texas at Austin (to
Z.D.).
*To whom correspondence should be addressed. Tel: +1 512 471 7859; Fax: +1 512
471 8696; Email: dave@noddy.cm.utexas.edu
REFERENCES
Return