Probing the conformation of DNA structures - 4-way junctions and bulges - with fluorescence

Gyorgy Vámosi*, Christoph Gohlke* #, Alastair I H Murchie+, David M J Lilley+, and Robert M Clegg*

*   Department of Molecular Biology, Max Planck Institute for Biophysical Chemistry, Göttingen, F.R.G.
#   Department of Molecular Biology, Institute for Molecular Biotechnology e.V., Jena, F.R.G.
+   CRC Nucl. Acid Struc. Group, Department of Biochemistry, University Dundee, U.K.

Poster presented at the 11th International Biophysics Congress, July 25-30, 1993, Budapest, Hungary


Fluorescence summary, anisotropy and FRET
FRET measurements on DNA duplexes
The bending of DNA and RNA by bulged nucleotides observed by FRET
Melting of the 18bp An-bulged DNA duplexes
Single strand binding to the bulge
Melting of the three-stranded A9dT10 structure
Melting of linear DNA duplexes monitored via the fluorescence intensity and anisotropy of rhodamine
Salt induced conformational transition of the four-way DNA junction monitored by the temperature dependence of rhodamine anisotropy


Fluorescence studies, especially fluorescence resonance energy transfer (FRET), have aided in determining conformations of complex molecular structures of nucleic acids. Oligonucleotides are synthesized, covalently labeled with dyes, and assembled to form specific structures. For instance, we have applied sensitive fluorescence methods to probe the helical structure of simple specifically labeled DNA duplexes and to investigate more complex structures, such as four-way DNA junctions (a model in solution, of the Holliday genetic recombination junction) and bulged DNA molecules (duplex structures with extra nucleotides in one of the strands). FRET, fluorescence anisotropy and intensity determinations, and kinetic measurements of helix-coil transitions provide us with molecular scale information regarding the conformations of these structures, and the conformational changes ensuing upon perturbations of the molecular environment. Examples will be given and the techniques will be discussed.


Fluorescence techniques provide detailed physical information concerning the molecular environment of a fluorophore. By attaching fluorophores covalently to macromolecular structures, knowledge of the immediate molecular environment of the fluorophore can be determined, thereby providing valuable information concerning the structural state of the macromolecule in the immediate vicinity of the fluorescent label. The fluorescence parameters can also report on structural and dynamic aspects of the macromolecular structure on a more global scale; in addition, inter- as well as intramolecular interactions and conformations of macromolecules often affect significantly the spectroscopic properties of associated fluorophores. Because fluorescence is a very sensitive technique experiments can be carried out over a large range of concentrations, including low concentrations into the nanomolar domain. The large variety of kinetic and steady state experiments that can be conducted in very diverse experimental situations makes this spectroscopic technique a method of choice for many biochemical systems, and fluorescence methods now play a central role in the analytical methods routinely used in molecular biology and biochemistry.

In the last several years, the synthesis of defined sequences of nucleic acids and the attachment of a large variety of fluorescent labels at selected positions of the oligonucleotides (both at internal nucleotide sequence positions and on the strand ends) has become convenient and straightforward. This spectacular advance has opened up many possibilities for making detailed physical measurements on nucleic acid structures. The results of fluorescence studies on well defined oligonucleotide structures have furnished valuable information about the structure of even complicated nucleic acid complexes (such as branched structures, bulges and multistranded complexes). Because fluorescence can be observed conveniently under a variety of different experimental circumstances, these results on well defined model systems can be applied advantageously to aid in the interpretation of subsequent fluorescence measurements made on more complex molecular structures and biological systems involving nucleic acids (such as protein-DNA complexes and large complex RNA structures).

We present here an excerpt of our fluorescence studies dealing with the fluorescence of labeled DNA duplex nucleic acid structures, and bulged DNA and RNA molecules. We record a variety of spectroscopic parameters: fluorescence quantum yield and anisotropy, and fluorescence resonance energy transfer (FRET), in both steady state and time resolved modes. These measurements provide unique information about the nucleic acid structures in solution, and of associated conformational changes. Examples are given below.

Fluorescence summary, anisotropy and FRET

Figure 1

FRET measurements on DNA duplexes

Figure 2FRET measurements on a series of duplex DNA molecules with different numbers of base pairs: The 5´ terminus of each complementary strand is labeled with a fluorophore, either fluorescein (Fl) or rhodamine (Rh). The helical structure of the DNA is clearly observable in the plot of the FRET efficiency in Figure 2. This result is not only a clear indication of the helical structure of DNA in solution, but also shows that the fluorescence measurements can be interpreted in terms of the Förster theory of fluorescence resonance energy transfer (see Fig. 1).

Figure 3
Conclusions from our spectroscopic and thermodynamic experiments concerning spectroscopic and molecular parameters of the fluorescent dyes used in these studies: Figure 3 depicts our estimates for the locations of the Fl and Rh dye molecules relative to the double-stranded DNA structure. These estimates have been made from a quantitative analysis of the FRET experiments (Fig. 2), and anisotropy measurements. Rh interacts more strongly with the DNA molecule than the Fl. This is consistent with the known aggregation properties of the dyes in aqueous solutions. In addition, the overall charge of the Fl is negative and is therefore repulsed from the negatively charged DNA polyelectrolyte, whereas Rh is neutral and does not experience a strong electrostatic interaction with the DNA. The fluorescence anisotropy and the "apparent" quantum yields of the fluorophores are sensitive monitors of the double- and single-stranded state of the DNA in the solution.

The bending of DNA and RNA by bulged nucleotides observed by FRET

Figure 10The efficiency of fluorescence resonance energy transfer (FRET) was measured between fluorescein and rhodamine. The dyes were covalently attached to the 5'-termini of a series of double stranded DNA and RNA molecules containing bulge loops of length ranging from 1 to 9 adenosine nucleotides (An, Figure 10).
Figure 11

Bulged molecules show decreased electrophoretic mobility with increasing numbers of bulged nucleotides (Figure 11). This can be interpreted in terms of increased bending.

Figure 12
Experimental FRET efficiencies determined from the enhanced acceptor fluorescence are shown in Figure 12. The FRET efficiencies increase as the number of bulged nucleotides is increased from 1 to 7. This is consistent with an increased bending of the DNA and RNA molecules at the bulge site. The A9 molecules have lower FRET efficiencies than the A7 molecules, indicating a change in the structural progression within each series. This is not evident in the gelelectrophoretic results of Fig. 11. FRET efficiencies calculated from either the donor anisotropy or the mixed donor-acceptor anisotropy yield similar results.

Bending angles were estimated by simulating the experimental FRET efficiencies with a geometric model of kinked DNA (Fig. 10); the positions of fluorescein and rhodamine relative to the DNA helix used for these calculations are depicted in Fig. 3. According to this model, 7 bulged adenosine nucleotides bend the DNA helix by 90±10°.

Melting of the 18bp An-bulged DNA duplexes

Figure 13The fluorescence quantum yields of fluorescein and rhodamine are strongly dependent on their molecular environment. We can selectively observe the state (double or single stranded) of both ends of the DNA molecule during thermal denaturation (Figure 13, compare Fig. 6). RNA molecules show similar results.

Conclusions from these measurements are:
Figure 14
1. bulged bases destabilize the molecules - Tm decreases with n.

2. for higher bulges the two stems melt approximately independently - the 7bp fluorescein stem melts with lower Tm than the 11bp rhodamine stem (Figure 14).

Single strand binding to the bulge

Figure 15Titration results of an 18bp A9-bulged DNA duplex with (dT)10 are shown in Figure 15. Addition of (dT)10 at low temperature strongly decreases the FRET efficiency (left panel - the acceptor fluorescence is linearly dependent on the FRET efficiency). The fluorescence of fluorescein increases (right panel) and its anisotropy decreases. In contrast, the rhodamine fluorescence parameters remain constant.

Figure 16Proposed model for the binding of (dT)10 to the A9 bulge at low temperature (Figure 16): the binding leads to a destabilization of the neighboring stems of the bulged molecule. In this special case the shorter fluorescein stem completely melts. Similar loop single strand interactions are found in RNA pseudoknots.

Melting of the three-stranded A9dT10 structure

Figure 17Fluorescence parameters of the 18bp A9-bulged DNA molecule are compared in Figure 17 at different temperatures with and without added (dT)10: The normalized acceptor fluorescence (top panel) and melting curves determined from the fluorescein/donor fluorescence (bottom panel) differ only at lower temperatures. Melting curves determined from the directly excited rhodamine fluorescence are identical throughout the whole temperature region (bottom panel).

Figure 18

Figure 18 is a proposed model, including strand interchange and subsequent complete strand dissociation, in order to explain the data in Fig. 17. The order of thermal denaturation is:

linear duplex -> bulge + single-strand -> single strands

Melting of linear DNA duplexes monitored via the fluorescence intensity and anisotropy of rhodamine

Figure 4 and 5A series of DNA duplexes (12, 14, 16, 18, 20 basepair length) was prepared. One of the strands was labeled at the 5' end with rhodamine (via a C6 linker). The fluorescence intensity (F) and anisotropy (r) of rhodamine were measured over a wide temperature range (Figures 4 and 5). These fluorescence parameters are sensitive indicators of the double and single stranded states of DNA.

Assuming a two-state (all-or-none) model for the dissociation of the strands, F and r of the reaction mixture are (Equation 1)
Equation 1
alpha(T) is the extent of melting (0<alpha<1); F ,Fss and Fds are to be understood as molar values. The ss and ds indexes of F and r refer to the single and double stranded species (no index refer to the total reaction mixture). All quantities are temperature dependent.

Figure 6 The above equations provide two different ways to determine the extent of melting vs. temperature. In order to extract the extent of melting (Figure 6) from the F(T) and r(T) curves (Figs. 4,5), the low and high temperature baselines i.e. the fluorescence and the anisotropy of the purely single stranded and purely double stranded species must be known in the temperature region of the transition. The low temperature baseline of anisotropy, rds was calculated from a linear extrapolation of the 1/r vs. T/viscosity function; it is approximately linear for many molecular shapes, including cylindrical molecules. Fds was determined by an extrapolation of the double stranded region of the F(T) curve. (Alternatively, the fluorescence data of higher melting molecules can also be taken as Fds). The high temperature baselines are taken from measurements with single stranded DNA (or from linear extrapolation).

The apparent enthalpy change of the melting transition was determined from the melting curves by means of a van't Hoff's analysis (Equation 2)
Equation 2
Figure 7
The apparent transition enthalpies determined from our fluorescence data (Figure 7) deviate from those calculated from the nearest neighbour interaction enthalpies determined by calorimetry (Breslauer et al. 1986), especially for the shorter duplexes. The observed fluorescence parameters are sensitive to the state of the local environment of the dye, e.g. fraying effects at the ends preceding the melting of the whole macromolecule. The determination of the baselines at the low temperatures is a possible source of error. 

Salt induced conformational transition of the four-way DNA junction monitored by the temperature dependence of rhodamine anisotropy

Figure 8The four-way DNA junction is the confluence of four DNA strands forming a double stranded structure with four helical arms. According to gel electrophoresis and FRET experiments (data not presented) at low salt concentration the junction has an open, extended square structure (Figure 8, left), whereas at higher salt concentration the molecule adopts a compact, folded, stacked, antiparallel X-structure (Fig. 8, right). This behavior is consistent with a strong electrostatic repulsion between the phosphate charges, especially at the point of strand exchange, partially screened by metal ions.

Four-way junction molecules were prepared from four 34 nucleotide length DNA single strands. One strand was labeled at the 5' end with rhodamine (via a C6 linker). A 17 basepair duplex was similarly prepared. The fluorescence anisotropy (r) of the duplex and the junction molecules are compared over an extended temperature range (below the melting temperatures) at different salt concentrations.

The temperature dependence of r is described by the Perrin equation (Equation 3):
Equation 3
r0, anisotropy at T=0 K; k, Boltzmann's constant; T, absolute temperature; tau, fluorescence lifetime; eta, viscosity; V is a geometrical factor (for a spherical molecule its the volume)

Figure 9 The slope of the Perrin plot (1/r vs. T/viscosity) is determined by the geometrical/hydrodynamical parameters of the rotating dye-DNA complex and the fluorescence lifetime of the dye. Experiments were made in parallel with the four-way junction and the duplex molecules in order to facilitate a comparison of the rotational motions of the two macromolecules, and to compensate for local segmental motions that are similar for the two dye-DNA complexes.
The ratio of the slopes of the Perrin plots of the junction and duplex molecules (Figure 9) shows that the relative rotational mobility of the junction molecule compared to the duplex increases as the concentration of NaCl is increased from 0.2 mM to 1.3 M. This change is consistent with the folding of the junction molecule into the more compact form (see Fig. 8).

Copyright © 1995-2017, Christoph Gohlke