Nucleic Acids Chemistry and Biology: The 5th Cambridge Symposium
Single-molecule studies of DNA and RNA four-way
junctions
S.A. McKinney*, E. Tan*, T.J. Wilson†, M.K. Nahas*, A.-C. Déclais†, R.M. Clegg*, D.M.J. Lilley† and T. Ha*1
*Department of Physics, University of Illinois, Urbana-Champaign, Urbana, IL 61801, U.S.A., and †Cancer Research UK Nucleic Acid Structure Research Group,
Department of Biochemistry, The University of Dundee, Dundee DD1 5EH, U.K.
Abstract
Branched helical junctions are common in nucleic acids. In DNA, the four-way junction (Holliday junction)
is an essential intermediate in homologous recombination and is a highly dynamic structure, capable of
stacking conformer transitions and branch migration. Our single-molecule fluorescence studies provide
unique insight into the energy landscape of Holliday junctions by visualizing these processes directly. In the
hairpin ribozyme, an RNA four-way junction is an important structural element that enhances active-site
formation by several orders of magnitude. Our single-molecule studies suggest a plausible mechanism for
how the junction achieves this remarkable feat; the structural dynamics of the four-way junction bring about
frequent contacts between the loops that are needed to form the active site. The most definitive evidence
for this is the observation of three-state folding in single-hairpin ribozymes, the intermediate state of which
is populated due to the intrinsic properties of the junction.
Introduction
Branched helical structures of nucleic acids are ubiquitous in
Nature, and there are many known examples in which their
structural properties are essential for biological functions
[1]. For instance, a DNA four-way junction called a
Holliday junction is formed as a key intermediate in DNA
recombination [2], and many enzymes have developed that
specifically target and process the Holliday junction [3,4].
In a small, self-cleaving RNA enzyme called the hairpin
ribozyme, an RNA four-way junction found in the natural
form of the RNA allows the molecule to fold and function
efficiently under physiological conditions. In contrast, the
minimal form of the hairpin ribozyme that lacks the junction
requires two or three orders of magnitude higher Mg2+
concentrations to achieve optimal activity. We have begun
single-molecule studies of the dynamic properties of the
Holliday junction and hairpin ribozyme with the aim of
understanding the functional consequences of their structural
dynamics.
Single-molecule FRET (fluorescence
resonance energy transfer)
Single-molecule fluorescence imaging [5] is a powerful tool
for probing biological events directly without the temporal
and population averaging of conventional ensemble studies
[6,7]. In particular, single-molecule FRET methods [8,9]
provide a powerful means of observing the dynamic structural
changes of biomolecules, as well as of subpopulations in a
Key words: branch migration, fluorescence resonance energy transfer (FRET), hairpin ribozyme,
Holliday junction, RNA folding, single-molecule spectroscopy.
Abbreviation used: FRET, fluorescence resonance energy transfer.
1
To whom correspondence should be addressed (e-mail [email protected]).
heterogeneous mixture [10–18]. In a typical single-molecule
FRET experiment, a biological macromolecule is labelled
with donor and acceptor fluorophores at different positions.
The internal motion of the molecule can bring the two
fluorophores closer together or further apart. The FRET
efficiency E, defined as the fraction of donor excitation events
that result in the excitation of the acceptor, is a strong function
of the distance R between the two fluorophores, according to
the equation:
E = 1/[1 + (R/R0 )6 ]
R0 is the distance at which E is 50%, and is typically
within the range 3–6 nm, an ideal length scale for biological
macromolecules. Often, the molecular motions are very
difficult to synchronize or are too rare to detect using
ensemble FRET. In these cases, single-molecule FRET
opens up new opportunities to probe structural changes of
biological molecules in real time.
In order to observe single-molecule conformational
changes over millisecond or second time scales, the molecules
need to be immobilized. This is achieved most easily by
a specific tether, such as biotin–streptavidin binding to a
glass surface. Although this is an unnatural situation, the
biological activities of DNA and RNA molecules remain
faithfully preserved, as comparisons with measurements in
bulk solution have shown [10–12,16].
The Holliday junction in DNA
DNA is often considered as a static molecule whose
main role is in information storage. However, it must
participate in functional processes, including transcription,
replication, repair and recombination. The Holliday junction
C 2004
Biochemical Society
41
42
Biochemical Society Transactions (2004) Volume 32, part 1
Figure 1 Structural dynamics of single Holliday junctions
(a) Cartoons illustrating the structures of four-way DNA junctions. The ‘open structure’ shows four single-stranded DNA
molecules hybridizing to form a junction of four helices (A, B, C and D). In the presence of Mg2+ ions, two neighbouring
helices can undergo co-axial stacking to form the stacked X-structure. Two stacking conformers are possible, depending on
the stacking partners. The open structure is considered a common intermediate for transitions between conformers and
branch migration. (b) Single-molecule FRET time records of four-way junctions at three different Mg2+ concentrations. Donor
and acceptor fluorophores are attached to the ends of helices such as to give higher FRET for conformer I than for conformer
II. (c) Conformer transition rates as a function of Mg2+ concentration. Both forward and backward rates decrease with
increasing Mg2+ concentration while remaining similar to each other for any given condition.
is formed as a key intermediate in DNA recombination
by combining two homologous DNA molecules [3]. To
understand how different enzymes recognize and process
the Holliday junction during recombination, its intrinsic
structural properties first need to be known. The Holliday
junction is a highly dynamic structure, capable of undergoing
two major structural changes: stacking conformer transitions
and branch migration, which are strongly dependent on the
types and concentrations of metal ions in solution. In
the absence of bivalent metal ions such as Mg2+ , the Holliday
junction takes on a so-called ‘open structure’, in which each
of the four helices points to the corner of a square [19].
In the presence of Mg2+ , a more compact structure called
C 2004
Biochemical Society
the ‘stacked-X structure’ is formed [20,21]. Neighbouring
helices are co-axially stacked in pairs, and the axes are rotated
to subtend an acute angle between them, hence the name
stacked-X structure (Figure 1a). In this structure, two of
the DNA strands run continuously through a given pair
of stacked helices in an antiparallel direction, while the other
two are exchanged between helical pairs. By symmetry, there
are two alternative stacking partners for each helix, and
hence there are two different conformations. If we label the
four helices A–D consecutively around the junction, then
conformer I forms with A stacked on D and C stacked on B,
while conformer II has A stacked on B and C stacked on D
[22,23].
Nucleic Acids Chemistry and Biology: The 5th Cambridge Symposium
Figure 2 Dynamics of the hairpin ribozyme and its RNA four-way junction
(a) The hairpin ribozyme used in FRET studies. The secondary structure comprises two internal loops in adjacent arms (A and
B) of a four-way junction. Ribozyme folding brings the donor and acceptor fluorophores closer to each other, increasing FRET.
The cleavage site is marked by an arrow; however, cleavage activity is prevented by substitution of a deoxyribonucleotide
at the −1 position. (b) Single-molecule FRET time records of the hairpin ribozyme fluctuating between the folded and
unfolded states (100 ms time resolution). (c) Single-molecule FRET time records of the hairpin ribozyme within the unfolded
state (4 ms time resolution). Rapid fluctuations between the proximal and distal states are apparent. (d) The sum of forward
[distal (D) to proximal (P)] and backward (proximal to distal) rates of fluctuations measured from the RNA four-way junction
(i.e. lacking the loops) as a function of Mg2+ concentration.
We attached fluorescent probes to the Holliday junction
so that one conformer had a higher FRET efficiency than
the other, and thus observed individual Holliday junctions
fluctuating between the two stacking conformers [12]
(Figure 1b). The fluctuation rate decreased substantially with
increasing Mg2+ concentration, but the relative populations
remained constant (Figures 1b and 1c). This is in marked
contrast with the folding of an RNA junction (see below),
where an increase in Mg2+ concentration increases the
population in the folded state. Therefore, unlike in RNA
folding, the rates of stacking conformer transitions cannot
be measured using conventional stopped-flow techniques,
making them truly non-synchronizable dynamics. Before our
studies, the only estimate for stacking conformer transition
rates came from NMR linewidth studies in one junction [24].
Why does adding more Mg2+ slow the dynamics of the
DNA junction? Crystal structures show that four negatively
charged phosphate groups are within 6 Å of each other [25–
27], leading to unfavourably strong electrostatic repulsion
that can only be screened by metal ions. Stacking conformer
transitions require unstacking of neighbouring helices, with
a likely intermediate resembling the open structure, and this
becomes less populated if the stacked structure is stabilized
further by increasing Mg2+ concentration. Branch migration
has also been shown to slow down by 1000-fold upon Mg2+
addition [28], and it has been proposed that the Mg2+ -free
open structure is the common intermediate for branch
migration and stacking conformer transitions [12].
Almost all structural studies of the Holliday junction have
used junctions where the DNA sequence does not allow
branch migration. Will the stacking conformer transitions
occur in the context of a mobile junction? How are conformer
transitions related to branch migration? How do the complex
structural dynamics of the Holliday junction influence
the activities of junction-processing enzymes that catalyse
branch migration or junction resolution? Single-molecule
fluorescence techniques appear to be well poised to address
these interesting questions.
The hairpin ribozyme in RNA
Ribozymes are cellular RNA molecules that catalyse chemical
reactions. They have fundamental implications for the
evolution of life on the planet, and provide insight into
biocatalysis in general [29]. Like protein enzymes, they must
fold into a conformation that provides a local environment
C 2004
Biochemical Society
43
44
Biochemical Society Transactions (2004) Volume 32, part 1
where catalysis can proceed. The single-molecule FRET
method has been used extensively to probe the folding
properties of various ribozymes both large [10,30,31] and
small [11,16,32], as well as of a small RNA junction involved
in ribosomal protein binding [33,34].
The hairpin ribozyme is one of the nucleolytic ribozymes
that bring about a site-specific cleavage of the RNA backbone
by means of a trans-esterification reaction. The natural form
of the hairpin ribozyme comprises two major structural
elements: a four-way RNA junction and two internal loops
carried by adjacent A and B arms of the junction (Figure 2a).
The loop–loop interaction is essential for ribozyme activity,
generating a 105 -fold acceleration of site-specific cleavage or
ligation reactions. While the minimal form without arms
C and D can still catalyse the cleavage reaction, and has
yielded much insight into the catalytic mechanism [35], its
folding requires two to three orders of magnitude higher
Mg2+ concentrations [36–38], and the internal equilibrium
between cleavage and ligation is shifted compared with the
natural form [39].
In order to gain more insight into the role of the RNA
four-way junction in the function of the hairpin ribozyme,
we attached fluorescent labels to the ends of A and B arms of
the ribozyme, such that folding brings the donor and acceptor
in close proximity, yielding high FRET. As expected, the
ribozyme spends a progressively longer time in the folded
state (high FRET state in Figure 2b) and a shorter time in
the unfolded state (low FRET state in Figure 2b) as the
Mg2+ concentration is increased. Surprisingly, examination
of the unfolded state with improved time resolution (4 ms
compared with 100 ms) revealed that the unfolded state is
made up of at least two states, termed the proximal and
distal states (Figure 2c) [16]. Fluctuations between the distal
and proximal states have also been observed in the four-way
junction that lacks the loops; hence they reflect the intrinsic
dynamics of the four-way junction. The rate of fluctuation
between the proximal and distal states decreased 1000-fold
when the Mg2+ concentration was varied from 0.1 mM to
50 mM (Figure 2d), but their relative populations remained
relatively constant, except for a bias towards the proximal
state that developed above 20 mM Mg2+ .
Since the proximal state brings the two loop-carrying arms
in close proximity, we hypothesized that the proximal state
is an obligatory intermediate in a sequential folding pathway
(distal ↔ proximal ↔ folded; Figure 3b). This hypothesis was
strongly supported by the data on a sequence variant (C25 U)
that shows impaired folding [40] (Figure 3a) [16]. The crystal
structure [41] reveals that G+1 is extruded from loop A and
is inserted into a pocket within loop B, where it forms a
Watson–Crick base pair with C25 of loop B. This pairing is
perturbed in the C25 U variant, destabilizing the folded state
and allowing us to observe unfolded ribozymes at millimolar
Mg2+ concentrations, where the fluctuations between the
distal and proximal states are slow enough to be resolved
directly. Single-molecule time records clearly demonstrate
that direct transitions between the distal and folded states
are rare (Figure 3a). The proportion of apparently direct
C 2004
Biochemical Society
Figure 3 The hairpin ribozyme folds in a three-state process
(a) Single-molecule FRET time record of a C25 U variant hairpin
ribozyme fluctuating between the distal, proximal and folded states. (b)
Proposed model for the stepwise folding of the hairpin ribozyme. The
ribozyme undergoes folding into the active conformation in two stages,
corresponding to the properties of the junction and the interactions
between the loops. The structural dynamics of the junction promote rapid
active-site formation due to the spontaneous fluctuations between the
two states of the junction, i.e. distal and proximal. The distal state
comprises a number of distinct conformations that cannot be distinguished on the basis of our current data. Single-molecule spectroscopy
uniquely provides information on the previous and subsequent
history of a given state, allowing us to conclude that the proximal state
is an obligatory intermediate in the formation of the ultimate folded
state.
conversions between the distal and folded states (typically
less than 10%) was fully explicable in terms of the lifetime
of the proximal state and the time resolution of the data.
Therefore it is highly probable that the proximal state is an
obligatory intermediate in ribozyme folding.
These results provide new insight into the mechanism
of hairpin ribozyme folding. The rapid conformational
exchange within the unfolded state repeatedly brings the two
loop elements into proximity, increasing the probability of
interaction between them and thus of folding. Loop–loop
interaction involves numerous specific contacts [41], and the
local conformation of the loops in the folded ribozyme is
significantly altered from that in the isolated loops [42,43]. It
will therefore require multiple conformational adjustments to
achieve the active state. The junction dynamics occur at a rate
of ∼100 s−1 at 0.5 mM Mg2+ , whereas the rate of formation
of the folded state is significantly lower (3 s−1 ) [16]. Folding
under these conditions must be limited by the conformational
changes involved in loop–loop interactions rather than by the
junction dynamics.
The folding rate of 3 s−1 observed for the natural form is
almost three orders of magnitude higher than the folding
rate of the minimal form even at much higher Mg2+
concentrations [11]. In addition, the natural form displayed
large heterogeneities in both folding and unfolding rates [16]
that originate from the loops, in contrast with the minimal
Nucleic Acids Chemistry and Biology: The 5th Cambridge Symposium
form, which showed heterogeneities only in unfolding rate
[11]. We suggest that, in the absence of the junction, the
encounter between the loops is rate limiting for folding, and
thus the effects of the heterogeneity are observable only in
the unfolding reaction.
The presence of the junction accelerates the folding of the
hairpin ribozyme by nearly three orders of magnitude, as well
as substantially reducing the required Mg2+ ion concentration
for folding into the physiological range. The junction should
therefore be regarded as an integral part of the ribozyme
that ensures efficient folding – a kind of ‘folding enhancer’.
A similar situation has been found for the hammerhead
ribozyme, where the Mg2+ concentration required for activity
is greatly reduced (again into the physiological range) by
inclusion of the loops on helices I and II [44]; it is likely
that interaction between the loops promotes folding into the
active conformation. Such auxiliary folding elements may be
common in small autonomously folding RNA species.
Conclusions
Our single-molecule studies have shown that the four-way
junctions of DNA and RNA are highly dynamic structures
capable of undergoing global changes in millisecond and
second time scales. These fluctuations are intrinsic properties
of the junction structures, and are likely to have important
functional consequences for many branched RNA molecules
and for proteins and enzymes that recognize and process the
junction structures.
Funding for these studies was provided by the National Institutes of
Health and the National Science Foundation (T.H.), and by Cancer
Research UK (D.M.J.L.).
References
1
2
3
4
5
6
7
8
9
10
11
12
Lilley, D.M.J. (2000) Q. Rev. Biophys. 33, 109–159
Holliday, R. (1964) Genet. Res. 5, 282–304
Lilley, D.M.J. and White, M.F. (2001) Nat. Rev. Mol. Cell Biol. 2, 433–443
West, S.C. (1997) Annu. Rev. Genet. 31, 213–244
Moerner, W.E. and Orrit, M. (1999) Science 283, 1670–1676
Weiss, S. (1999) Science 283, 1676–1683
Xie, X.S. and Lu, H.P. (1999) J. Biol. Chem. 274, 15967–15970
Ha, T., Enderle, T., Ogletree, D.F., Chemla, D.S., Selvin, P.R. and Weiss, S.
(1996) Proc. Natl. Acad. Sci. U.S.A. 93, 6264–6268
Ha, T. (2001) Methods 25, 78–86
Zhuang, X.W., Bartley, L.E., Babcock, H.P., Russell, R., Ha, T.J., Herschlag,
D. and Chu, S. (2000) Science 288, 2048–2051
Zhuang, X.W., Kim, H., Pereira, M.J.B., Babcock, H.P., Walter, N.G. and
Chu, S. (2002) Science 296, 1473–1476
McKinney, S.A., Declais, A.C., Lilley, D.M.J. and Ha, T. (2003) Nat. Struct.
Biol. 10, 93–97
13 Deniz, A.A., Laurence, T.A., Beligere, G.S., Dahan, M., Martin, A.B.,
Chemla, D.S., Dawson, P.E., Schultz, P.G. and Weiss, S. (2000) Proc. Natl.
Acad. Sci. U.S.A. 97, 5179–5184
14 Schuler, B., Lipman, E.A. and Eaton, W.A. (2002) Nature (London) 419,
743–747
15 Rothwell, P.J., Berger, S., Kensch, O., Felekyan, S., Antonik, M., Wohrl,
B.M., Restle, T., Goody, R.S. and Seidel, C.A. (2003) Proc. Natl. Acad. Sci.
U.S.A. 100, 1655–1660
16 Tan, E., Wilson, T.J., Nahas, M.K., Clegg, R.M., Lilley, D.M.J. and Ha, T.
(2003) Proc. Natl. Acad. Sci. U.S.A. 100, 9308–9313
17 Ha, T., Rasnik, I., Cheng, W., Babcock, H.P., Gauss, G., Lohman, T.M. and
Chu, S. (2002) Nature (London) 419, 638–641
18 Brasselet, S., Peterman, E.J.G., Miyawaki, A. and Moerner, W.E. (2000)
J. Phys. Chem. B 104, 3676–3682
19 Clegg, R.M., Murchie, A.I.H. and Lilley, D.M.J. (1994) Biophys. J. 66,
99–109
20 Duckett, D.R., Murchie, A.I., Diekmann, S., von Kitzing, E., Kemper, B. and
Lilley, D.M. (1988) Cell 55, 79–89
21 Murchie, A.I., Clegg, R.M., von Kitzing, E., Duckett, D.R., Diekmann, S. and
Lilley, D.M. (1989) Nature (London) 341, 763–766
22 Miick, S.M., Fee, R.S., Millar, D.P. and Chazin, W.J. (1997) Proc. Natl.
Acad. Sci. U.S.A. 94, 9080–9084
23 Grainger, R.J., Murchie, A.I.H. and Lilley, D.M.J. (1998) Biochemistry 37,
23–32
24 Overmars, F.J.J. and Altona, C. (1997) J. Mol. Biol. 273, 519–524
25 Nowakowski, J., Shim, P.J., Prasad, G.S., Stout, C.D. and Joyce, G.F. (1999)
Nat. Struct. Biol. 6, 151–156
26 Ortiz-Lombardia, M., Gonzalez, A., Eritja, R., Aymami, J., Azorin, F. and
Coll, M. (1999) Nat. Struct. Biol. 6, 913–917
27 Eichman, B.F., Vargason, J.M., Mooers, B.H.M. and Ho, P.S. (2000)
Proc. Natl. Acad. Sci. U.S.A. 97, 3971–3976
28 Panyutin, I.G. and Hsieh, P. (1994) Proc. Natl. Acad. Sci. U.S.A. 91,
2021–2025
29 Doudna, J.A. and Cech, T.R. (2002) Nature (London) 418, 222–228
30 Russell, R., Zhuang, X., Babcock, H.P., Millett, I.S., Doniach, S., Chu, S. and
Herschlag, D. (2002) Proc. Natl. Acad. Sci. U.S.A. 99, 155–160
31 Bartley, L.E., Zhuang, X., Das, R., Chu, S. and Herschlag, D. (2003) J. Mol.
Biol. 328, 1011–1026
32 Bokinsky, G., Rueda, D., Misra, V.K., Rhodes, M.M., Gordus, A., Babcock,
H.P., Walter, N.G. and Zhuang, X. (2003) Proc. Natl. Acad. Sci. U.S.A. 100,
9302–9307
33 Ha, T., Zhuang, X.W., Kim, H.D., Orr, J.W., Williamson, J.R. and Chu, S.
(1999) Proc. Natl. Acad. Sci. U.S.A. 96, 9077–9082
34 Kim, H.D., Nienhaus, G.U., Ha, T., Orr, J.W., Williamson, J.R. and Chu, S.
(2002) Proc. Natl. Acad. Sci. U.S.A. 99, 4284–4289
35 Berzal-Herranz, A., Joseph, S., Chowrira, B.M., Butcher, S.E. and Burke,
J.M. (1993) EMBO J. 12, 2567–2573
36 Murchie, A.I.H., Thomson, J.B., Walter, F. and Lilley, D.M.J. (1998)
Mol. Cell 1, 873–881
37 Walter, N.G., Burke, J.M. and Millar, D.P. (1999) Nat. Struct. Biol. 6,
544–549
38 Zhao, Z.Y., Wilson, T.J., Maxwell, K. and Lilley, D.M.J. (2000) RNA 6,
1833–1846
39 Fedor, M.J. (1999) Biochemistry 38, 11040–11050
40 Wilson, T.J., Zhao, Z.Y., Maxwell, K., Kontogiannis, L. and Lilley, D.M.J.
(2001) Biochemistry 40, 2291–2302
41 Rupert, P.B. and Ferre-D’Amare, A.R. (2001) Nature (London) 410,
780–786
42 Cai, Z. and Tinoco, Jr, I. (1996) Biochemistry 35, 6026–6036
43 Butcher, S.E., Allain, F.H. and Feigon, J. (1999) Nat. Struct. Biol. 6,
212–216
44 Khvorova, A., Lescoute, A., Westhof, E. and Jayasena, S.D. (2003)
Nat. Struct. Biol. 10, 1–5
Received 3 September 2003
C 2004
Biochemical Society
45