96
TIBTECH- MARCH 1991 [Vol. 9]
Studying DNA-protein interactions using NMR Irina M. Russu NMR spectroscopy is emerging as a powerful tool in molecular biology and biotechnology; one aspect of which is the use of one- a n d two.dimensional NMR methodologies to investigate the interactions of proteins with DNA. The dynamic a n d structural i n f o r m a t i o n w h i c h NMR can provide, on the changes in conformation a n d molecular flexibility, complements X-ray crystallography d a t a a n d enables mechanistic models of DNA-protein interactions to be
formulated. In the past decade, our understanding of the structural basis of DNAprotein interactions has made great progress, and several mechanisms by which proteins recognize specific base sequences on DNA have been identified. One involves a direct readout of the chemical groups on the DNA bases by protein side chains which are placed in the conformation required for direct contact with the bases by structural motifs in the protein {such as helix-turn-helix and zinc fingers). Alternatively, the protein may recognize conformational features at its recognition site on the DNA. The DNA double helix does not have a monotonous structure, as was once thought, but has sequencespecific variations in local conformation. Different base sequences are also associated with different degrees of molecular flexibility. They too can contribute to the recognition of a given sequence since, for many proteins, the binding induces structural distortions (such as looping, bending or unwinding) in DNA. Two experimental techniques used widely to obtain structural information relevant to DNA-protein interactions are X-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy. NMR complements X-ray crystallography by providing structural and dynamic information on molecular interactions occurring in solution. This review covers several recent applications of L :VI. Russu is at the Department of Melecular Biology and Biochemistry, W~,sleyan University, Middletown, CT 06~57, USA.
NMR spectroscopy in this field. The review does not attempt to be comprehensive, and includes only studies which best illustrate the NMR methodologies currently employed. The investigation of DNA-protein interactions by NMR spectroscopy requires, as a first step, the separate spectroscopic characterization of the two macromolecules of interest {i.e. the DNA and the DNA-binding protein}. As in other NMR studies of biological systems, this characterization must start by assigning NMR resonances to specific nuclei in the molecule. The experiments can then be extended to obtain distances between specific nuclei. At present, NMR is most often applied to determine distances between hydrogen atoms, which can then be used to generate three-dimensional structures of DNA and DNA-binding proteins. A dynamic view of these structures is provided by relaxation.based NMR experiments which can identify regions of the molecule with increased mobility or flexibility. The same experimental approaches can be applied to DNA-protein complexes to characterize their structures and the inter-molecular contacts involved in binding and recognition. NMR spectroscopy in one and two dimensions NMR is a type of absorption spectroscopy which is dependent on the property of 'spin' possessed by the nuclei of certain isotopes - this causes them to behave like small magnets when placed in a magnetic field. Different orientations of nuclear spin have different energies
(~ 1991. Elsevier Science Publishers Ltd (UK) 0167 - 9430/91/$2.00
(in most cases of biological interest, orientation in the direction of the magnetic field corresponds to a lowenergy state; and opposing the direction of the magnetic field, a highenergy state). Absorption of energy at a specific radio frequency will cause the nuclei to 'flip' from a lowenergy state to a high-energy state by a process termed resonance. The frequency required will be proportional to the strength of the magnetic field, and specific to the nucleus observed. At a particular field strength, the resonance frequency of a nucleus will also depend on its chemical environment - these differences are referred to as chemical shifts. In modern NMR spectrometers, the spins of the nuclei of interest (1H, 13C, 31p, 14N o r tSN), are excited using a radio frequency pulse and the resonance signal which is induced is measured in a detecting coil. The induced signal decays with time as a result of relaxation processes as the system returns to its equilibrium state, and the signal is therefore termed the free induction decay (FID). A Fourier transformation (FT) of this signal produces an NMR spectrum with resonance lines at frequencies corresponding to each nucleus observed. An NMR spectrum of a molecule therefore consists of a series of absorption lines (Fig. 1). The dispersion of lines along the frequency axis {chemical shift axis), reflects that each nucleus experiences a different environment. These differences among nuclei are small (of the order of 10 -6 when expressed relative to the strength of the magnetic field). Thus, the frequency axis of an NMR spectrum is usually expressed as parts per million (ppm). When studying proteins or DNA fragments, the number of NMR resonances increases greatly due to the large number of nuclei present in these molecules. As a result, resonances of individual nuclei overlap and the resolution is limited. This severe problem has been overcome by the introduction of two-dimensional NMR {Fig. 2). The two frequency dimensions of two-dimensional NMR come from two time intervals (tl and t2), during which the nuclei can be subjected to different conditions (e.g. radio frequency pulses). A two-dimensional
TIBTECH- MARCH 1991 [Vol. 9]
--Fig. 1
97
--Fig. 2
Preparation l
FT of s(t)
e~(ppm) General scheme for a one-dimensional NMR experiment. The vertical bar represents a radiofrequency pulse. The free induction decay Is(t)] is recorded as a function of time (t). Fourier transformation (IT) of the free induction decay yields the NMR spectrum.
NMR experiment consists, in effect, of a series of one-dimensional experiments in which the time interval tl {'evolution' time), is incremented {from, say, 0-100 ms), and the free induction decay for each experiment is recorded during the time interval t~ {'detection' period), as in onedimensional NMR. The amplitude of the signals detected in t~ is a function of what happened to the nuclei during tl. In many experiments, a further time interval, ~, {'mixing' period), is included to allow interactions between nuclear spins to occur. The two-dimensional NMR spectrum is obtained by two-dimensional Fourier transformation with respect to t~ and t2: after Fourier transformation with respect to t2, a set of spectra modulated in amplitude or in phase as a function of t~ is obtained, and a Fourier transformation of this data set with respect to t~ yields the final two-dlmensional spectrum spread over frequencies ~o~ and (o2. This spectrum is usually most clearly depicted as a contour plot. The diagonal cross peaks {where col is equal to wz} correspond to the one-dimensional spectrum.
Evolution (t,)
General scheme for a two-dimensional NMR experimenL The vertical bars represent radiofrequency pulses. The evolution time, tT, is incremented in separate experiments. For each value of tT, the free induction decay is recorded during the detection time, t2. Two-dimensional Fourier transformation of the data matrix s(tT,t2) yields the twodimensional NMR spectrum.
Mixing (r)
Detection (~)
two-dimensional FT of s (tl, t2)
o)2
O~1
The off-diagonal cross-peaks reveal interactions (connectivities) between nuclei whose resonances are at e.,~ and w2, respectively. The kinds of interactions observed in a two-dimensional NMR experiment depend on the exact pulse sequence used {i.e. the number, length and phases of the pulses within and between the three time periods). Magnetization may be transferred between nuclei by chemical exchange, the nuclear Overhanser effect (NOE) or the spin-spin coupling 0 coupling) mechanism (see Glossary). NOEs originate from through-space interactions between nuclei not necessarily bound to each other. They depend strongly on the distance between two nuclei (r) being, in the first approximation, proportional to 1/r6. NOEs may be observed in two-dimensional 'nuclear Overhauser enhancement spectroscopy' (NOESY) and NOESY cross-peaks reveal the nuclei which are situated in close spatial proximity (
TIBTECH- MARCH 1991 [Vol. 9]
Studying DNA-protein interactions using NMR Irina M. Russu NMR spectroscopy is emerging as a powerful tool in molecular biology and biotechnology; one aspect of which is the use of one- a n d two.dimensional NMR methodologies to investigate the interactions of proteins with DNA. The dynamic a n d structural i n f o r m a t i o n w h i c h NMR can provide, on the changes in conformation a n d molecular flexibility, complements X-ray crystallography d a t a a n d enables mechanistic models of DNA-protein interactions to be
formulated. In the past decade, our understanding of the structural basis of DNAprotein interactions has made great progress, and several mechanisms by which proteins recognize specific base sequences on DNA have been identified. One involves a direct readout of the chemical groups on the DNA bases by protein side chains which are placed in the conformation required for direct contact with the bases by structural motifs in the protein {such as helix-turn-helix and zinc fingers). Alternatively, the protein may recognize conformational features at its recognition site on the DNA. The DNA double helix does not have a monotonous structure, as was once thought, but has sequencespecific variations in local conformation. Different base sequences are also associated with different degrees of molecular flexibility. They too can contribute to the recognition of a given sequence since, for many proteins, the binding induces structural distortions (such as looping, bending or unwinding) in DNA. Two experimental techniques used widely to obtain structural information relevant to DNA-protein interactions are X-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy. NMR complements X-ray crystallography by providing structural and dynamic information on molecular interactions occurring in solution. This review covers several recent applications of L :VI. Russu is at the Department of Melecular Biology and Biochemistry, W~,sleyan University, Middletown, CT 06~57, USA.
NMR spectroscopy in this field. The review does not attempt to be comprehensive, and includes only studies which best illustrate the NMR methodologies currently employed. The investigation of DNA-protein interactions by NMR spectroscopy requires, as a first step, the separate spectroscopic characterization of the two macromolecules of interest {i.e. the DNA and the DNA-binding protein}. As in other NMR studies of biological systems, this characterization must start by assigning NMR resonances to specific nuclei in the molecule. The experiments can then be extended to obtain distances between specific nuclei. At present, NMR is most often applied to determine distances between hydrogen atoms, which can then be used to generate three-dimensional structures of DNA and DNA-binding proteins. A dynamic view of these structures is provided by relaxation.based NMR experiments which can identify regions of the molecule with increased mobility or flexibility. The same experimental approaches can be applied to DNA-protein complexes to characterize their structures and the inter-molecular contacts involved in binding and recognition. NMR spectroscopy in one and two dimensions NMR is a type of absorption spectroscopy which is dependent on the property of 'spin' possessed by the nuclei of certain isotopes - this causes them to behave like small magnets when placed in a magnetic field. Different orientations of nuclear spin have different energies
(~ 1991. Elsevier Science Publishers Ltd (UK) 0167 - 9430/91/$2.00
(in most cases of biological interest, orientation in the direction of the magnetic field corresponds to a lowenergy state; and opposing the direction of the magnetic field, a highenergy state). Absorption of energy at a specific radio frequency will cause the nuclei to 'flip' from a lowenergy state to a high-energy state by a process termed resonance. The frequency required will be proportional to the strength of the magnetic field, and specific to the nucleus observed. At a particular field strength, the resonance frequency of a nucleus will also depend on its chemical environment - these differences are referred to as chemical shifts. In modern NMR spectrometers, the spins of the nuclei of interest (1H, 13C, 31p, 14N o r tSN), are excited using a radio frequency pulse and the resonance signal which is induced is measured in a detecting coil. The induced signal decays with time as a result of relaxation processes as the system returns to its equilibrium state, and the signal is therefore termed the free induction decay (FID). A Fourier transformation (FT) of this signal produces an NMR spectrum with resonance lines at frequencies corresponding to each nucleus observed. An NMR spectrum of a molecule therefore consists of a series of absorption lines (Fig. 1). The dispersion of lines along the frequency axis {chemical shift axis), reflects that each nucleus experiences a different environment. These differences among nuclei are small (of the order of 10 -6 when expressed relative to the strength of the magnetic field). Thus, the frequency axis of an NMR spectrum is usually expressed as parts per million (ppm). When studying proteins or DNA fragments, the number of NMR resonances increases greatly due to the large number of nuclei present in these molecules. As a result, resonances of individual nuclei overlap and the resolution is limited. This severe problem has been overcome by the introduction of two-dimensional NMR {Fig. 2). The two frequency dimensions of two-dimensional NMR come from two time intervals (tl and t2), during which the nuclei can be subjected to different conditions (e.g. radio frequency pulses). A two-dimensional
TIBTECH- MARCH 1991 [Vol. 9]
--Fig. 1
97
--Fig. 2
Preparation l
FT of s(t)
e~(ppm) General scheme for a one-dimensional NMR experiment. The vertical bar represents a radiofrequency pulse. The free induction decay Is(t)] is recorded as a function of time (t). Fourier transformation (IT) of the free induction decay yields the NMR spectrum.
NMR experiment consists, in effect, of a series of one-dimensional experiments in which the time interval tl {'evolution' time), is incremented {from, say, 0-100 ms), and the free induction decay for each experiment is recorded during the time interval t~ {'detection' period), as in onedimensional NMR. The amplitude of the signals detected in t~ is a function of what happened to the nuclei during tl. In many experiments, a further time interval, ~, {'mixing' period), is included to allow interactions between nuclear spins to occur. The two-dimensional NMR spectrum is obtained by two-dimensional Fourier transformation with respect to t~ and t2: after Fourier transformation with respect to t2, a set of spectra modulated in amplitude or in phase as a function of t~ is obtained, and a Fourier transformation of this data set with respect to t~ yields the final two-dlmensional spectrum spread over frequencies ~o~ and (o2. This spectrum is usually most clearly depicted as a contour plot. The diagonal cross peaks {where col is equal to wz} correspond to the one-dimensional spectrum.
Evolution (t,)
General scheme for a two-dimensional NMR experimenL The vertical bars represent radiofrequency pulses. The evolution time, tT, is incremented in separate experiments. For each value of tT, the free induction decay is recorded during the detection time, t2. Two-dimensional Fourier transformation of the data matrix s(tT,t2) yields the twodimensional NMR spectrum.
Mixing (r)
Detection (~)
two-dimensional FT of s (tl, t2)
o)2
O~1
The off-diagonal cross-peaks reveal interactions (connectivities) between nuclei whose resonances are at e.,~ and w2, respectively. The kinds of interactions observed in a two-dimensional NMR experiment depend on the exact pulse sequence used {i.e. the number, length and phases of the pulses within and between the three time periods). Magnetization may be transferred between nuclei by chemical exchange, the nuclear Overhanser effect (NOE) or the spin-spin coupling 0 coupling) mechanism (see Glossary). NOEs originate from through-space interactions between nuclei not necessarily bound to each other. They depend strongly on the distance between two nuclei (r) being, in the first approximation, proportional to 1/r6. NOEs may be observed in two-dimensional 'nuclear Overhauser enhancement spectroscopy' (NOESY) and NOESY cross-peaks reveal the nuclei which are situated in close spatial proximity (
