Crystallographic study on oligonucleotide coiled-coils

Crystallographic study on oligonucleotide coiled-coils Thesis submitted for the Degree of Doctor of Philosophy Daniela De Luchi Barcelona, 2008 Departament d’Enginyeria Química Escola Tècnica Superior d’Enginyeria Industrial de Barcelona Universitat Politècnica de Catalunya Crystallographic study on oligonucleotide coiled-coils Memoria presentada por Daniela De Luchi para acceder al Grado de Doctor en Ciencias. Trabajo realizado en el Departamento d’Enginyeria Química de la ETSEIB-UPC, dirigido por el Dr. Juan A. Subirana Torrent y codirigido por la Dra. J. Lourdes Campos. Barcelona, Junio 2008 ACTA DE QUALIFICACIÓ DE LA TESI DOCTORAL Reunit el tribunal integrat pels sota signants per jutjar la tesi doctoral: Títol de la tesi: ............................................................................................................... Autor de la tesi: .............................................................................................................. Acorda atorgar la qualificació de: No apte Aprovat Notable Excel·lent Excel·lent Cum Laude Barcelona, …………… de/d’….................…………….. de ..........…. El President El Secretari ............................................. ............................................ (nom i cognoms) (nom i cognoms) El vocal El vocal El vocal ............................................. ............................................ ..................................... (nom i cognoms) (nom i cognoms) (nom i cognoms) Ai miei genitori e a Martina iv Agradecimientos En primer lugar quiero darle las gracias a mi Director de tesis, Prof. Juan A. Subirana, por haberme dado la posibilidad de realizar una tesis doctoral, por haber sido una fuente de conocimientos, experiencia, recursos y continuas ideas. Le quiero agradecer la ayuda en la realización y redacción de este trabajo, en sus varias fases, y especialmente por la ayuda en la interpretación de los diagramas de difracción aquí presentados. Quiero darle mis más sinceros agradecimientos a la Dra. Valentina Tereshko, en primer lugar por la ayuda profesional, los consejos y el apoyo y, no menos importante, por haber hecho tan agradable mi estancia en Chicago. A la Dra. Lourdes Campos, por su amabilidad, la infinita paciencia y la continua disponibilidad y ayuda en más de una ocasión. A la Dra. Isabel Usón, por el continuo interés y disponibilidad; también le quiero dar las gracias al Prof. G. M. Sheldrick por hacer posibles mis estancias en Goettingen, donde la formación de los estudiantes resulta verdaderamente importante. Le quiero dar las gracias a mi hermana Martina por el continuo apoyo y las muchas imágenes que me ha ayudado a realizar y que ahora están en esta tesis (grazie Architetto Marty!). Por las correcciones de este manuscrito un “grazie” especial a la Dra. Marianna Biadene (¡y sin olvidar las muchas cenas juntas!). Le quiero dar las gracias a la Dra. Nuria Valls por su amistad y por haberme ayudado y escuchado tantas veces, dentro y fuera del trabajo; también quiero agradecerle a la Dra. Carme Cáceres su disponibilidad y su continuo interés. A todos los compañeros del departamento, por hacer más agradables las horas pasadas juntos: a las ya doctoras Montse Vera y Meritxell Palau, a Sebastià Gestì, Emma Botines, Laura y Elena, Elsa, Gina y Mireya, a las Dras. Nuria Saperas, Lourdes Urpì, Lourdes Franco, Maria Teresa Casas y a todos los profesores, doctorandos y proyectistas que han estado y están en el departamento. v A las compañeras de piso, Rosa, Dominique y (actualmente) a Ma. Carmen y Antonia, gracias por haberme hecho sentir como en casa desde el primer día y por acogerme cada vez que vuelvo. Un “gracias” especial a Jordi, por todo lo que hemos compartido, los viajes, las excursiones y mucho más. Un immenso grazie ai miei genitori, che nonostante la distanza mi sono stati vicini in tutti questi anni. Grazie per aver appoggiato, cercato di capire o solo accettato le mie scelte, non ce l’avrei fatta senza di voi! Esta tesis ha sido realizada gracias a una beca del proyecto europeo HPRN-CT2000-00009 y a la beca pre-doctoral AP2003-2309 otorgada por el Ministerio de Educación y Ciencia de España. vi Abstract The crystallographic study of the coiled-coils generated by DNA oligonucleotides is the main subject of this thesis. When the straight axis of a simple helix (minor coil) follows itself a helical path, then the structure is called a coiled-coil (major coil). The parameters that define a DNA coiled-coil are: the inclination of the oligonucleotides axis with respect to the major coil axis; the number of oligonucleotides per turn; the kink angle and the torsion angle between consecutive oligonucleotides. Previous works show that the DNA sequences d(AT)6 (Campos et al., 2005) and d(AT)5 generate coiled-coils with very different geometrical characteristics. In order to better understand the properties of these structures, fourteen oligonucleotides with sticky-ended sequences have been crystallized. The presence of a sticky end determines the coiled-coil properties. The sequences studied are (CG)n(AT)m and (AT)m(CG)n, and some other very similar to those. The majority of them have and , so that the sticky end is usually represented by the sequence d(CG). The geometrical characteristics of the coiled-coils have been studied, and the relation between the aforementioned parameters ( , , and ) has been calculated. Due to the intrinsically difficult crystallization of such sequences, in several cases only poor diffracting crystals have been obtained and the determination of their atomic structures has not been possible. Despite this, the structure of the sequence dCG(AT)5 could be determined at 3.1 Å resolution showing unambiguously that the (AT)5 fragment generates a double helix with Hoogsteen base pairs. It is not clear whether the Hoogsteen base pairing influences or not the geometry of the super-coils and only some hypothesis could be formulated. Recently, a crystal of the short fragment dCG(AT)2 has been obtained and its diffraction pattern has been measured up to 2.6 Å resolution. At the present time, the structure has not been solved yet and only some preliminary considerations are shown in this work. vii As a complementary study, the melting temperatures (Tm) of AT-rich oligonucleotides have been determined, and a simple equation for their prediction is shown (see Appendix III.1). Finally, the structure of the complex of d(UBrAGG) with an anthraquinone derivative, previously solved in our laboratory, has been refined (see Appendix III.3). viii List of abbreviations A Adenine C C y t osine CC Correlation coefficient CCD Coupled charge device CSD Cambridge structural database DNA Deoxyribonucleic acid G Guanine HPLC High performance liquid chromatography MAD Multiwavelength anomalous diffraction M IR Multiple isomorphous replacement MPD 2-Methyl-2-pentanediol MR Molecular Replacement NDB Nucleic acid database NMR Nuclear magnetic resonance PDB Protein databank PEG Polyethylene glycol RMSD Root mean square deviation RNA Ribonucleic acid R-WC Reverse Watson-Crick SAD Single wavelength anomalous diffraction SF Structure factor S IR Single isomorphous replacement SR Synchrotron radiation T Thymine TMAO Trimethylamine n-oxide U Uracil W-C Watson-Crick 4 Nucleic acids commonly form helical secondary structures using two or more strands. The most common DNA conformation in vivo is represented by the B-form. The B-DNA can be described by the well known Watson - Crick Model of the double helix, derived in 1953 from the X-ray diffraction pattern of a DNA fiber. The base of this model is the specific recognition between a purine and a pyrimidine base: adenine with thymine (uracil in the RNA) and guanine with cytosine. Standard Waston-Crick hydrogen bonds are shown in Figure I.1.3. These combinations lead to virtually identical base geometries. This identity was the basis of the realization that it is possible to build a regular double helix with an arbitrary sequence and it was also the basis for understanding the replication of the genetic code. As shown in Figure I.1.3, A-T pairs are maintained by two hydrogen bonds, while C-G pairs present three hydrogen bonds. Reverse Watson-Crick base pairing is also possible. This binding mode often occurs in parallel-stranded DNA. In this case, only two hydrogen bonds are found between guanine and cytosine (Figure I.1.4). The most important pairing alternative to the Watson-Crick is probably the Hoogsteen scheme (see Figure I.1.4). It was observed for the first time in the crystal structure of a complex of adenine and thymine bases methylated in the position linking to the sugars in the nucleosides (Hoogsteen, 1959). In this mode of binding the purines are rotated 180° around the glycosidic bond and form two hydrogen bonds to the pyrimidines through N7 and N6 (AT base pair) or O6 (CG base pair). Protonation of cytosine is a prerequisite for CG Hoogsteen base-pairs to occur. Figure I.1.3. The canonical Watson-Crick basepairs, stabilized by two H-bonds in the AT base pair and by three H-bonds in the CG base pair. 5 Figure I.1.4. Reverse Watson-Crick, Hoogsteen and Reverse Hoogsteen hydrogen bonds. CG Hoogsteen pairing (not shown) is also possible and commonly found in nucleic acids triple helices; protonation of cytosine at N3 is necessary for it to occur. Homo-purinic R:R and homo-pyrimidinic Y:Y base pairings are also possible, in particular guanine-guanine base pairs are easily found in G-rich sequences (see Figure I.1.5). Figure I.1.5. Homo-purinic guanine-guanine symmetric hydrogen bonds. 6 Guanines rich sequences tend to form four-stranded (or quadruplex) structures called G tetrads, in which the guanines interact through their Watson-Crick and Hoogsteen edges (Figure I.1.6). Figure I.1.6. G-quadruplex hydrogen bonding scheme: the tetrads are normally stabilized by a monovalent cation (as K+ or Na+) in the centre of the structure. The Hoogsteen hydrogen bondings between adjacent guanines are shown I.1.2 DNA conformation Torsion angles Nucleic acids chains are highly flexible due to the torsion angles in the sugarphosphate backbone (Figure I.1.7). The backbone torsion angles are as follows: α (PO5’), β (O5’-C5’), γ (C5’-C4’), δ (C4’-C3’), ε (C3’-O3’) and ζ (O3’-P). Since the sugar forms a ring, the intracyclic torsion angles of the sugar rings, ν0-ν4, are dependent on one another. Deviations from planarity forces one of the ring atoms out of the plane, the net effect is termed the “sugar puckering”. The face of the sugar ring that is toward the glycosidic bond is termed the “endo” face, the face that is away is termed the “exo” face. C2’-endo is typical of B-DNA while C3’-endo is the most common sugar conformation in A-form RNAs and DNAs. 7 The glycosidic bond goes from C1’ to N9 for purines and from C1’ to N1 for pyrimidines. The permitted angles of rotation χ exist in two regions: anti and syn. The anti conformation corresponds to , the syn conformation corresponds to . The latter value is typical for the purines involved in Hoogsteen base pairing. Figure I.1.7. Torsion angles in the nucleic acid backbone. Base pairs, dimer step and helical parameters A common point of reference is needed to describe the three-dimensional arrangements of bases and base pairs in nucleic acid structures. The program 3DNA (Lu & Olson, 2003) used in this work implements the parameter sets derived from the Tsukuba Workshop on Nucleic Acid Structure and Interactions held in 1999 and recommended by the NDB (Berman et al., 1992). There are two sets of local parameters commonly in use in nucleic acid conformational analysis: step parameters which show the stacking geometry between neighbor base-pairs, and helical parameters which demonstrate the position and orientation of a base-pair relative to the helical axis (Figure I.1.8). The values of local vs helical rise and twist from these two sets of parameters can be quite different in DNAs which deviate significantly from B-form DNA. 8 Although very useful for standard Watson-Crick hydrogen bonds, in the case of Hoogsteen basepairs, the purine rotation into the syn conformation and the different hydrogen bonding scheme makes most of the aforementioned parameters unusuable at least for comparison with standard values. Figure I.1.8. Standard reference frame for the description of nucleic acid base-pair geometry as recommended in the NDB (Berman et al., 1992). 9 I.2 X-Ray Diffraction and Macromolecular Crystallography I.2.1 Macromolecular crystallization Obtaining crystals in many cases remains a trial and error process and frequently represents the rate limiting step in the determination of macromolecular structures. Crystals are grown by slow, controlled decrease of the solubility of the macromolecule, usually dissolved in an aqueous solution. The basic requirements for a crystallization experiment are the purity and homogeneity of the sample. All the oligonucleotides studied in this thesis have been synthesized on an automatic synthesizer by the phosphoramidite method and purified by gel filtration and reverse phase HPLC at the Institut Pasteur of Paris. A large number of variables have an influence on the macromolecule crystallization and there is no or very limited a priori information about which one must be modified. Many variables can influence the crystallization, among them: pH, ionic strength, DNA (protein) concentration, nature and concentration of the precipitant (MPD, PEG, spermine, etc), temperature, nature and concentration of ions and additives. A typical macromolecular phase diagram is shown in Figure I.2.1. Crystals dissolve in the undersaturated region, where the concentration is below the macromolecule solubility, and grow in the supersaturated region. A relatively large supersaturation is required to overcome the activation energy barrier which exists when forming the crystal (Asherie, 2004). If the supersaturation is too large, then disordered structures, such as aggregates or precipitates, may form. There are three stages of crystallization common to all systems: nucleation, growth and cessation of growth. 10 Nucleation is the process by which molecules or non-crystalline aggregates which are free in solution come together in such a way to produce a thermodynamically stable aggregate with a repeating lattice, the first semblance of the solid state. The nuclei can be formed in the labile zone of the supersaturation area of the phase diagram. The degree to which nucleation occurs is determined by the degree of the supersaturation of the solute. Crystal growth generally starts at solute concentration sufficient for nucleation to occur, and continues at concentration below the nucleation threshold. After the formation of the first nuclei the concentration of the solute is slightly decreased, the nucleation stops and the crystals formed can grow to a bigger size without the competition of new nuclei. Cessation of growth can occur for different reasons. The most obvious is the decrease in concentration of the crystallizing solute to the point where the solid and solution phases reach exchange equilibrium. Figure I.2.1. A schematic phase diagram showing the solubility of a macromolecule in solution as a function of the precipitant concentration (Mc Pherson, 1999). 11 Crystallization techniques Vapor diffusion is the most common method of crystallization. A schematic representation of a hanging drop is shown in Figure I.2.2. A small amount of oligonucleotide solution is mixed with the precipitant (typical volumes range from 1 to 10 μL) and placed on a siliconised cover slip. The drop is then suspended and sealed over the well solution. The difference in precipitant concentration between the drop and the well solution is the driving force which causes water to evaporate from the drop until the concentration of the precipitant in the drop equals that of the well solution. The sitting drop is a variation of the method which allows bigger volumes to equilibrate. Dialysis The sample containing the macromolecule is placed inside a dialysis cell. The cell is placed inside a solution containing the crystallization agents, which then diffuse through the membrane into the dialysis cell reducing the solubility of the macromolecule. Batch method In the batch method, concentrated protein is mixed with concentrate precipitant solution to produce a final concentration which is supersaturated in terms of the solute macromolecule and therefore leads to crystallization. This can be done with up to ml amounts of solution and typically results in larger crystals due to the larger volumes of solute present and the lower chance of impurities diffusing to the face of the crystal. This technique is by far the most expensive in terms of consumption of the solute macromolecule. Liquid-liquid diffusion The protein and precipitant solutions are layered on top of each other allowing a slow equilibration. Nucleation and crystal growth generally occurs at the interface between the two layers, at which both the concentrations are at their highest values. 12 Figure I.2.2. Vapor diffusion from a hanging drop. The macromolecular concentration in the drop increases over time. DNA (and protein) crystals are characterized by relatively high water content, typically ranging from 30% to 80% of the volume. The typical volume per base pair, , varies from about 1300 Å3 per base pair, for medium/high resolution structures, to about 2000 Å3 per base pair, for low resolution structures. Real crystals are mosaics of many submicroscopic arrays in rough alignment with each other; this phenomenon is much more pronounced in macromolecular crystals than in crystals of rigid organic or inorganic molecules. 13 I.2.2 Crystals and symmetry Crystals are made up of identical parallelepiped-shaped blocks called unit cells that constitute a three dimensional translation lattice (Figure I.2.3). The cell is defined by the vectors a, b and c; they define the length a, b, c, and the angles α, β, γ which characterize the unit cell. The volume V of the unit cell can be calculated as follows: . The unit cell is the smallest unit that can generate the entire crystal by translation operations alone. The content of the unit cell is obtained by repetition of a single object through the symmetry elements. This part of the unit cell is called the asymmetric unit. Within the cell there can be several symmetry related asymmetric units with identical contents, but in general in different orientations. The cell is always chosen so that the symmetry elements are positioned in accord with volume A of the International Tables for Crystallography. To define the planes in the crystal the Miller indices (h, k, l) have been introduced. The h, k, l terms define parallel planes with intercepts a/h, b/k, c/l on the three a, b, c axes of the unit cell with h, k, l small integer numbers. For example, the (234) planes, shown in Figure I.2.4, cut the unit cell edges a into two parts, b into three parts and c into four parts. Figure I.2.3. Diagram of the lattice created by the translation of the unit cell; the vectors a, b and c and the angles α, β and γ are indicated. 20 The readout deadtime is the time during which the X-ray beam must be shuttered off in order to read out the integrated signal in the detector. Image Plate The heart of the Image Plate is a storage phosphor screen. When the storage phosphor is exposed to X-rays, secondary electrons are trapped in color centers, whose number is proportional to the X-ray energy. After the exposure, these metastable centers can be excited by a red laser to release visible photons in a process known as photostimulation or bleaching. To completely erase the remaining, unbleached centers after the plate is read, it is exposed to an intense, broadband light source for some tens of seconds. This process brings the phosphor back to the ground state. The biggest advantage of this scheme is that it allows a relatively large active area (up to 345 mm diameter). Perhaps, their biggest disadvantage is their relatively long readout time since it typically takes several tens of microseconds to bleach each pixel on the image plate (Nave, 1999). The total readout time for the entire plate is typically on the order of 1 to 2 minutes. This long readout time is a serious disadvantage in experiments at synchrotron beamlines. The other principle disadvantage of the Image Plate is its relatively low sensitivity. Chargerd Couple Device CCD-based detectors were developed in order to address the long readout times and the low sensitivity of the Image Plates. A schematic representation of a CCD detector is given in Figure I.2.8. The X-rays excite a scintillator screen to produce visible photons, a fiber optic taper transfers the light photons to the CCD chip, in which the photons induce a charge generation. The charges are then transferred and detected. The CCD reads out the photon almost instantaneously and immediately after is ready for a new exposure. Figure I.2.8. X-rays excite a phosphor screen, producing visible photons which are phocused onto a CCD imager using a fiber optic taper. 21 I.2.5 Principles of X-ray diffraction X-rays are electromagnetic waves characterized by a wavelength λ in the range of 0.1-100 Å. The interaction between the traveling waves and the electrons in the crystal gives rise to scattering. Two kinds of scattering take place: Thomson, also called elastic or coherent scattering (no exchange of energy with the molecules takes place), and Compton or inelastic scattering. In X-ray crystallography, coherent scattering gives rise to diffraction. In 1913, W. L. Bragg and his son showed that diffraction could be regarded as if it were reflection from sets of equivalent, parallel planes of atoms in a crystal. The planes are designated by a set of three numbers called lattice or Miller indices, hkl. The index h gives the number of parts into which the set of planes cut the edge a of each cell; the indexes k and l respectively give the number of parts into which the set of planes cut the edges b and c (Figure I.2.4). Each set of parallel planes is treated as an independent diffractor and produces a single reflection. W. L. Bragg showed that a set of parallel planes with indexes hkl and interplanar spacing dhkl produces a diffracted beam only when the angle of incidence of the X-rays of wavelength λ meets the following condition (the so called Bragg’s law): (I.1) The geometric construction and the equations in Figure I.2.9 show the necessary conditions for producing a strong diffracted ray. Figure I.2.9. Conditions that produce strong diffracted rays. If the additional distance traveled by the more deeply penetrating ray R2 is an integer multiple of λ, then rays R 1 and R2 interfere constructively (Rhodes, 1993). 22 If this difference in path length for rays reflected from successive planes is equal to an integral number of wavelengths (that is, if ), then the rays reflected from successive planes emerge from the crystal in phase with each other, interfering constructively to produce a strong diffracted beam. The hkl planes can be described through a vector normal to the hkl plane and of length . The points at the end of these vectors form the reciprocal lattice. The reciprocal lattice is spatially linked to the crystal because of the way the lattice points are defined, so if the crystal is rotated, the reciprocal lattice rotates with it. Each reciprocal lattice point must be arranged with respect to the X-ray beam in order to satisfy Bragg’s law and produce a reflection from the crystal. The Bragg’s law of diffraction is illustrated in three dimensions by the Ewald sphere (shown in Figure I.2.10). The radiation of wavelength λ is represented by a sphere of radius . The crystal is represented by the reciprocal lattice with its origin at the point O on the Ewald sphere where the beam leaves it. If the reciprocal lattice point P lies on the surface of the Ewald sphere, the length of the vector , perpendicular to the reflecting plane hkl, is , that is the Bragg’s law. Figure I.2.10. The Ewald construction. When a reciprocal lattice point with indices hkl lies on the surface of the Ewald sphere, the interference condition for that particular reflection is fulfilled (Dauter, 1999). 23 X-rays are significantly scattered basically only by electrons. The amplitude of scattering for an atom is known as the atomic scattering factor and is described by the following equation: (I.2) where is the electron density of an atom, at position r, and is the scattering vector. The atomic scattering factor depends on the length of (since ) but is independent of the direction of the vector . The bigger the angle (thus higher the resolution) the smaller is the scattering factor. Since the atoms scattering the X-rays are not fixed in their position, but vibrate around an equilibrium position their scattering factor is affected. The motion is dependent on the temperature. The scattering factor diminishes because of thermal vibration especially at high diffraction angles. In order to account for atomic and molecular vibrations, the atomic scattering factor must be corrected as follows: (I.3) where . For structure with low resolution only the isotropic temperature factor can be refined while for high (near to atomic) resolution structures the anisotropic B factor can be refined (with several, usually six, atomic displacement parameters). The X-ray radiation scattered by one unit cell is known as the structure factor and symbolized by F or F(hkl). It is the Fourier transform of the scattering density (electrons in the molecule) sampled at the reciprocal lattice point hkl. The intensity of the scattered radiation is proportional to the square of the amplitude, |F|2. The structure factor is represented by: (I.4) with representing the amplitude of the scattered wave, and its phase relative to the origin of the unit cell. can also be written as the sum of contributions from each volume element of electron density in the unit cell: (I.5) 24 The structure factor is the Fourier transform of the electron density and vice versa, therefore the electron density can be written as follows: (I.6) While structure amplitudes are directly obtained from measured reflection intensities, the phases are lost. This is known as the crystallographic phase problem, methods for obtaining an initial set of experimental phases will be described in paragraph I.2.8. 25 I.2.6 The Patterson function If the Fourier transform, used to calculate the electron density map, is written with all the phase angles , the so called Patterson function is obtained: (I.7) where u, v, w are the coordinates of the Patterson cell. (where ) can be considered as the convolution of the electron density with itself. The result is that the high values of P happens at positions u corresponding to an interatomic distance vector. A simple example of construction of Patterson map is shown in Figure I.2.11 (Rhodes, 1993). The number of peaks in a Patterson map is , reduced to because of them are located at the origin. The Patterson map represents an important tool for structure determination as it is the essence of the Molecular Replacement, and it is the basis for finding the heavy atoms positions in the Isomorphous Replacement or Multiple Anomalous Dispersion Methods. Figure I.2.11. Construction of a Patterson map. (a) Structure of unit cell containing three atoms. Two of the six interatomic vectors are shown. (b) Patterson map is constructed by moving all interatomic vectors to the origin. Patterson “atoms” (peaks in the contour map) occur at the head of each vector. (c) Complete Patterson map, containing all peaks from (b) in all unit cells. Peak at origin results from self-vectors. Image of original structure is present (origin and two darkened peaks) among other peaks. (Rhodes, 1993). 26 I.2.7 Data collection Diffraction from a crystal is obtained when Bragg’s law is fulfilled. Ewald sphere illustrates Bragg’s law of diffraction in three dimensions. When a reciprocal lattice point lies at the surface of the Ewald sphere, the interference condition for that particular reflection is fulfilled and it gives rise to a diffracted beam. When the crystal is not rotated during the X-ray exposure the diffraction pattern (called a “still” photograph) will consist of spots arranged in a set of concentric ellipses. As shown in Figure I.2.12, in the rotation method, the start and end orientations of the diffracting plane form two intersecting ellipses with all reflections recorded between them in the form of a lune. Due to the crystal mosaicity (crystals are composed of small blocks slightly misoriented with respect to one another) and the beam divergence (the incident radiation is not directed precisely along one line), the diffraction corresponding to a particular reflection is spread over a range of crystal rotation, some reflections come into the diffracting position during one exposure and finish during the next (Dauter, 1999). Figure I.2.12. The rotation method and the “lunes”. When the crystal is rotated, reflections from the same plane in the reciprocal lattice form a lune, limited by two ellipses corresponding to the start and end positions (Dauter, 1999). 27 Several variables must be taken into account for a successful data collection, among them: the rotation range, the crystal-to-detector distance, the blind region, the total rotation range. Rotation range The maximum permitted rotation range to avoid overlap of neighboring lunes can be estimated with this formula: , where depends on the mosaicity and beam divergence, is the high-resolution limit and is the length of the primitive unit cell dimension along the direction of the X-ray beam. The best orientation of the crystal is with the longest unit cell axis along the spindle axis of crystal rotation, in this way the longest edge will never lie parallel to the beam and reflections will not overlap. Crystal-to-detector distance The distance should be adjusted to match the maximum resolution of the diffraction. If one unit cell dimension is so large that setting the distance to maximum resolution leads to overlap of reflections, it is better to sacrifice the resolution and set the distance so that reflection profiles separate. Blind region The reciprocal lattice points lying close to the rotation axis will never cross the Ewald sphere, and will never diffract. This part of the reciprocal lattice, on both sides of the spindle axis, is called the “blind region”. It is narrow at low resolution and wide at high resolution. If the crystal is triclinic there is no way to avoid loss of completeness due to the blind region, but if the crystal has symmetry axes, it is possible at least to collect symmetry equivalent reflections to those in the blind region. Total rotation range The total rotation range affects the completeness of the data set. Due to limited available time at the synchrotron, it is often impossible to collect 180° or 360°; it is thus important to quickly index the data and, on the basis of the symmetry and orientation of the crystal, decide the total rotation range necessary for the data collection. The goal of data collection is a set of consistently measured, indexed intensities. Several computer programs have been developed in order to process diffraction data. 28 Although it is not a proper statistical quantifier, the data quality is usually judged by the global Rmerge factor, it is given by: where is the individual intensity measurement and is the average intensity for this reflection (Dauter, 1999). This value is highly influenced by the redundancy of the data and is always higher for data in high symmetry space group than those in low symmetry. A good quality indicator is the ration of intensities to their uncertainties, , the accepted resolution limit is where the falls below about 2.0. Well scaled data should have their as close as possible to 1, its value is given by: (I.8) 29 I.2.8 The Phase Problem The main problem in crystallography after having recorded good data is the lack of phase information. The possible ways to obtain phases are:  Molecular Replacement (MR): if a similar structure is known, it can be used to calculate initial phases.  Multiple Isomorphous Replacement (MIR) or Single Isomorphous Replacement (SIR).  Anomalous Diffraction (MAD, SAD).  Direct Methods. Molecular Replacement Molecular Replacement (MR) exploits the existence of a known model structure to solve the phase problem of an unknown structure. The first requirement is the similarity between the unknown and the known structure. Placement of the molecule in the target unit cell requires its proper orientation and precise position, which involves rotation and translation. The principles of the MR method are based on the Patterson function of a crystal structure. The Patterson function represents a vector map in which interatomic distance vectors are represented by peaks of positive density. When the interatomic distances are between atoms inside the molecule they are called self-Patterson vectors while when between atoms belonging to different molecules in the unit cell they are called crossPatterson vectors. The distinction between self and cross vectors is fundamental to MR since similar or identical molecules will give similar or equal Patterson map respectively, apart from a rotational and a translational term. This is because the selfvector give a representation of the molecule itself, while cross-vectors are useful in finding the position of the model in the unit cell because they are related to intermolecular atomic distances. The principle of separating the Patterson vectors into these two groups can be used for orientation and translation determination. 36 measured intensities that were not included in the refinement, whereas R measures how well the current model predicts the entire data set that produced the model. A strong deviation between R and Rfree indicates overfitting of the model. In crystallographic refinement two functions are commonly used. They are the least-square residual and the maximum likelihood. The least-square refinement has been used for many years (Konnert, 1976; Konnert, et al., 1980); the function to be minimized is: (I.18) The summation is over all crystallographically independent reflections and is the weight given to an observation. The main limitation is the possibility of getting trapped in local minima, when the model is not very good or not complete. To overcome this problem different approaches have been chosen. The program CNS (Brünger et al., 1998) includes a molecular dynamics algorithm which exploits a simulated annealing technique. In the simulation the temperature is increased and the atoms are allowed to move freely from their original position, the temperature is then slowly cooled down allowing the structure to rearrange and eventually find a global minimum (Brünger et al., 1998). In the maximum-likelihood method, implemented in the program REFMAC (Murshudov et al., 1997), given the model, the probability function that a set of data would be observed is calculated: (I.19) where is the probability distribution of the structure factor given the model structure factor (Pannu, et al. 1996). TLS refinement (Howlin et al., 1993; Schomaker et al., 1993; Murshudov et al., 1999). Normally four parameters per atom are refined for macromolecular structures: 37 the three co-ordinates and a temperature factor. The temperature factor is a measure of the mean displacement of an atom. Since data sets from proteins or DNA are generally not highly overdetermined, the temperature factor is considered as isotropic; only for very high resolution data it can be refined anisotropically. In 1968, Schomaker & Trueblood described a parameterization that allows the description of anisotropic motion with many fewer parameters than an independent anisotropic B factor for each atom. This parameterization is called TLS (translation, libration and screw). In this system the motion of a group of atoms is described by three matrices. The explicit assumption of TLS-B-factors is that the group of atoms moves as a rigid group. In the TLS formalism, 20 parameters are used to describe the motion of the entire group of atoms. Since the anisotropic B-factor of one atom requires six parameters, any TLS group composed of more than three atoms results in a decrease in the total number of parameters. The TLS refinement has been implemented in the program REFMAC 5 (Murshudov et al., 1997). I.2.10 Validation and deposition Structure deposition at the Nucleic Acid Data Base (NDB) or at the Protein Data Bank (PDB) is the final and fundamental step once a structure is refined. The importance of structure validation before deposition is of course fundamental. In this work the program 3DNA (Lu & Olson, 2003) has been used. This program implements the parameter set derived from the Tsukuba Workshop on Nucleic Acid Structure and Interactions of 1999 and recommended by the NDB (Berman et al., 1992). It must be taken into account that the geometrical parameters calculated by the program 3DNA are optimized for the structures of complementary Watson-Crick basepairs, the presence of Hoogsteen hydrogen bonds, makes some of the parameters calculated by the 3DNA unusable at least for comparison with standard values. 38 I.2.11 Introduction to fiber diffraction Although in this work no fiber diffraction experiment has been performed, many of the crystals obtained were characterized by a fibrous nature. It appears therefore relevant to briefly introduce the concepts of fiber diffraction. Many biological macromolecules are of a fibrous nature. Sometimes the orientation is intrinsic, but often the molecules can be oriented into fibers when isolated from the cells. The oriened fiber is placed in a collimated X-ray beam at right angles to the beam and the fiber diffraction pattern is recorded on a film placed a few centimeters away from the fiber. The direction parallel to the fiber axis and through the center of a fiber diffraction pattern is referred to as the meridian, and the direction perpendicular to this is called the equator (see Figure I.2.16). Fibers are usually composed of long, chain-like molecules, packed together with their axes parallel, or nearly parallel, to the fiber axis. The degree of order within fibers may vary considerably. A famous example is given by the fiber diffraction obtained from A and B DNA forms, shown in Figure I.2.16. In the cases like the A form of DNA (Figure I.2.16), the molecules are regularly arranged so that they form crystalline regions, but the different crystalline regions within a fiber are randomly oriented about the fiber axis. The diffraction patterns from such fibers are similar to single crystal rotation photographs, with all Bragg reflections registered at one time and with the appearance of arcs due to the disorder. In other cases, the degree of order is much lower. If molecules are randomly displaced relative to each other in the direction of fiber axis, discrete spots are only observed along the equator, and the higher layer lines have a continuous distribution of intensity along them, this is the case of B-DNA, shown in Figure I.2.16 (b). 39 Figure I.2.16. Fiber diffractions of A DNA (a) and B DNA (b). The meridian and the equator are shown. Discrete spots along the meridian indicate periodicity along the fiber axis. The diffraction along the equator gives information about the structure in projection down the fiber axis, thus, discrete spots along the equator are related with the lateral distance between molecules, an example of hexagonal arrangement is given in Figure I.2.17. Figure I.2.17. Projection down the fiber axis of a hexagonal arrangement of helical molecules, from the fiber diffraction pattern, the distance between helices can be measured. 40 Diffraction by helical molecules The theory of diffraction by helical molecules was first developed in 1952 by Cochran, Crick and Vand and by Stokes (unpublished). Crick showed that the diffraction from a helix occurs along a series of equidistant layer lines. The intensity along the layer lines is continuous and can be calculated via a “Fourier-Bessel Transform”. Bessel functions enter the equation because a cylindrical coordinate system is used. The variation of with n and x is shown in Figure I.2.18. Only the zero order Bessel function, , has a non-zero value at ; the value of for the first maximum decreases as n increases. Figure I.2.18. Bessel functions. Crick showed that for a continuous helix the order of Bessel function n, occurring on a certain layer line, is the same as the layer line number l. A continuous helix and its diffraction pattern are shown in Figure I.2.19. Because the order of Bessel function increases with layer line number, the position of the first strong peak moves further away from the meridional direction, generating the characteristic “helix cross”. 41 Figure I.2.19. A continuous helix and its diffraction pattern. The pitch P of the helix is shown. The position of the first strong peak is also inversely proportional to the radius of the helix. The spacing between the layer-lines is inversely proportional to the pitch (P) of the helix. A discontinuous helix can be considered as a set of scattering points, equally spaced along a helix, with p the vertical distance between such points and ω the turn angle between them. The pitch P can be directly derived from the first two values. In projection onto the helical axis (see Figure I.2.20), the structure has a regular repeat period p. This gives rise to meridional reflections on the diffraction patterns. The crosslike pattern seen in the diffraction of continuous helices is still visible and, in addition, it is repeated at each meridional reflection. This gives rise to characteristic diamondshaped regions above and below the centre of the pattern, as shown in Figure I.2.20. For a simple helix which repeats in one turn the spacing between layer lines is given by . The distance of the first meridional layer line along the meridian is given by . 42 Figure I.2.20. A discontinuous helix of ten residues in one turn and its diffraction pattern. The B-DNA is a simple helix which repeats in one turn. With its 10 base pairs per turn, the average angle ω is of about 36°. The spacing between the bases corresponds to 3.4 Å (p = 3.4 Å) and the pitch P is of about 34 Å. In frozen crystals like those studied here the value of p is usually smaller (3.2 Å-3.35 Å) II. CRYSTALLOGRAPHIC STUDY ON OLIGONUCLEOTIDE COILED-COILS II.1 INTRODUCTION: AT-rich DNA sequences Analysis of the human genome sequence has confirmed the presence of extensive noncoding regions (Lander et al., 2001). Such regions are also present in practically all eukaryotic genomes, but their biological role is unclear. Interestingly, in most cases, they are rich in AT base pairs. Centromeres and pericentromeric regions (Choo et al., 1997), introns (Lander et al., 2001), scaffold-associated regions or matrix attachment regions (Liebich et al., 2002), gene desert (Lander et al., 2001; Nobrega et al., 2003), complex genes (Nobrega et al., 2003), and some satellites (Sainz et al., 1989) are all very rich in AT base pairs. In contrast with the abundance of AT base pairs in extensive regions of the genome, there is no structure available for any protein interacting with a DNA fragment that only contains AT base pairs. There are also few structural studies of oligonucleotides with such sequences. An overview of the structure of thirty-three all-AT duplexes has been recently reported (Campos et al., 2006). Before this work, only a few alternating structures had been determined: d(ATAT) (Viswamitra et al., 1982), d(ATATAT) (Abrescia et al., 2002) and d(AT)6 (Campos et al., 2005). While d(ATAT) shows Watson-Crick base pairs, although in a nonstandard conformation, d(ATATAT) crystallizes as a duplex structure with Hoogsteen base pairs. On the other hand, d(AT)6 forms a coiled-coil probably in Hoogsteen conformation. 44 Hoogsteen base pairs have been known for more than 40 years. The Hoogsteen base pairing scheme was observed for the first time in the crystal structure of a complex of adenine and thymine bases methylated in the position linking to the sugars in the nucleosides (Hoogsteen, 1959). As shown in Figure II.1.1, in this mode of binding the purines are rotated 180° around the glycosidic bond and form two hydrogen bonds to the pyrimidines through N7 and N6, or O6 for C∙G base pairs. Protonation of cytosine at N3 is a prerequisite for C∙G Hoogsteen basepairs. The characteristic change to the syn conformation of the purine base adenine (whereas thymines are in the normal anti conformation) is also found in Z-form DNA, where the guanine undergoes a similar rotation. Hoogsteen base pairs were postulated for U(A∙U) triple helices (Felsenfeld et al., 1957). Such interactions have also been found in chemically modified nucleic acids (Hakoshima et al., 1981; Isaksson et al., 2001). Isolated base pairs have been reported in some protein/DNA complexes (Patikoglou et al., 1999; Nair et al., 2004) and occasionally in RNA (Leontis & Westhof, 1998). A∙T Hoogsteen and standard WatsonCrick hydrogen bondings are shown in Figure II.1.1. Figure II.1.1. Comparison of A∙T Watson-Crick (a) and Hoogsteen (b) base pairs in duplex conformation. The minor groove is facing downwards; hydrogen bond acceptors and donors are indicated by arrows; hydrogen bonds and C1’-C1’ distances are shown in dashed lines.The minor groove is narrower in the Hoogsteen case and has lost a hydrogen bond acceptor atom (adenine N3). The major groove has a similar appearance, but in the Hoogsteen conformation an additional external N3 atom is present. The conformation of the glycosidic angle is syn in the Hoogsteen adenine base and anti in all other cases. 45 Although the overall appearance of the Hoogsteen duplex is very similar to the standard B-form DNA, as shown in Figure II.1.2, there are important differences. The minor groove of Hoogsteen DNA is narrower due to the shorter C1’-C1’ distances (Figure II.1.1). The accessibility of the grooves toward interaction with solvent and proteins is different in the two forms. The N3 atom of adenine, which lies in the minor groove in B-form DNA, is now moved to the major groove. As a result, the minor groove becomes less electronegative with only one hydrogen bond acceptor represented by the O2 of thymine. This characteristic of the minor groove, together with its narrowness, makes it an appropriate target for interactions with hydrophobic groups. In the Hoogsteen duplex, the helical axis is found at the edge of the base pairs, approximately at the midpoint of the hydrogen bond between adenine N6 and thymine O4. The major groove is in the center of the helix, and the phosphates are externally located. This situation is also found in A-form DNA. Figure II.1.2. Comparison of ideal B DNA with Hoogsteen DNA (Abrescia et al., 2002). 52 II.3 Aim of the project Previous works show that the DNA sequences d(AT)6 (Campos et al., 2005) and d(AT)5 (unpublished results) generate coiled-coils with very different geometrical characteristics. The aim of this project is to study the properties of the coiled-coils. 1. With this purpose, fourteen oligonucleotides with sticky ended sequences have been crystallized. The presence of the sticky end determines the coiled-coil properties. The sequences that have been studied are (CG)n(AT)m and (AT)m(CG)n, plus some other very similar to those. The majority of them have and , so that the sticky end is usually represented by CG. 2. The geometrical characteristics of the coiled-coils have been studied. 3. It has been tried to determine whether the Hoogsteen base pairing influences or not the geometry of the super-coils. 4. As a complementary study, the melting temperatures of AT-rich oligonucleotides have been measured; the results are reported in Appendix III.1. The sequences studied in this thesis are summarized in Table II.1. Table II.1 List of the oligonucleotide sequences crystallized in this work. Dodecamers Decamers Octamers d(CGATATATATAT): CG(AT)5 d(CGCGATATATAT): (CG)2(AT)4 d(ATATATATATCG): (AT)5CG d(ATATATATATGC): (AT)5GC d(GCATATATATAT): GC(AT)5 d(CGATATGCATAT):CG(AT)2GC(AT)2 d(CGCGCGATATAT): (CG)3(AT)3* d(CGATATATAT): CG(AT)4 d(ATATATATCG): (AT)4CG d(ATATATATATT): (AT)5T d(CGTATATA): CG(TA)3 d(CGATATAT): CG(AT)3 d(ATATATCG): (AT)3CG Hexamers Tetramers d(CGATAT): CG(AT)2 d(CGAT)* *No crystals suitable for X-ray experiments have been obtained from the sequences (CG)3(AT)3 and d(CGAT). 53 II.4 Geometry of the coiled-coil As said is paragraph II.1, when the straight axis of a simple helix follows itself a helical path, then the structure is called a coiled-coil (Figure II.1.3). The supercoil conformation of DNA is well known from studies of circular molecules. The DNA in the nucleosome also forms a supercoiled structure. DNA oligonucleotides may also associate by following a helical path in some protein/DNA complexes (Bunting et al., 2003). However, prior to the study published in 2005 by Campos et al., no studies were available on isolated, regular, continuous coiled-coil DNA molecules. Ropes formed by supercoiled α-helices, though, are a classical biophysical model. In particular, the model has been used in the study of keratin (Fraser et al., 1964). In this work, studies on fourteen sticky ended oligonucleotides are presented. They form pseudo-continuous DNA helices. In about half of the cases, the individual duplexes axes are inclined a few degrees with respect to the major coil axis and thus generate a coiled-coil structure. The geometrical characteristics of the coiled-coils (also called superhelices) have been studied. In fact the coiled-coils generated by oligonucleotides with sticky ends may be considered as kinked coils, whose geometries are defined by the two angles θ and τ. The kink angle θ is formed by two consecutive duplexes, whereas τ is the torsion angle which relates three consecutive duplexes (see Figure II.4.1). This simplification assumes that the duplexes are perfectly straight. Small intrinsic bends in the duplexes will in fact be incorporated in the angle θ. If the structure is approximately continuous, τ is directly related with ω, the twist value of the individual base pairs. Figure II.4.1. (a) Lateral view of a DNA coiled-coil, the kink angle θ and the torsion angle τ are shown. The torsion angle τ represents the rotation of with respect to about the connective fragment . (b) Projections of the coil along . The τ angle is shown. 54 Although the geometry of a coiled-coil is determined by the kink angle θ and the torsion angle τ, from the experimental diffraction patterns the following parameters can be measured (see paragraph : the inclination β of the oligonucleotides, the length h of the repeating unit and the number N of duplexes per turn. A schematic view of two consecutive duplexes, part of a coiled coil, is shown in Figure II.4.2:  h is the length occupied by the repeating unit along the major axis of the coiled-coil;  β represents the inclination of the duplex axis with respect to the major coil axis;  l is the length of one duplex;  α is the turn angle, related to the number of duplex per turn, N;  R is the radius of the cylinder described by the coiled-coil;  a is the projection of two consecutive fragments along the major helix axis;  b is projection of two consecutive fragments onto the coiled-coil axis;  θ is the kink angle between two consecutive fragments;  θ’ is θ supplementary angle. A coiled-coil describes a cylinder with a radius R. The number N of duplexes per turn determines the turn angle α (with ). In the simple case of a coiled-coil with six duplexes per turn, α is equal to 60°. The projection of such a helix along the major coil axis is shown in Figure II.4.3. The turn angle α is always equal to or smaller than the torsion angle τ. These geometrical considerations are equally valid for more complex situations, when the duplexes form a coil with a non-integer value of N, for example twenty four duplexes in five turns. In the cases studied here, such a situation does not appear to be present. 55 Figure II.4.2. (a) Perspective view of two consecutive duplexes, part of a coiled coil. l is the length of one duplex; β is the inclination of one duplex axis with respect to the coiledcoil axis; R is the radius of the cylinder described by the coiled-coil; α is the turn angle; (b) the kink angle θ between two consecutive oligonucleotides is shown; θ’ is its supplementary angle. Figure II.4.3. Projection along the major axis of a coiled-coil with six oligonucleotides per turn. The turn angle α of 60° is shown. 56 II.4.1 Calculation of the geometrical parameters of the coiled-coil From the experimental diffraction patterns, the inclination β of each duplex, as well as the number N of duplexes per turn of the coiled-coil can be obtained; their values are given in the tables in which the experimental results are reported. The handedness of the coil cannot be calculated directly from these values. From β and N, the values of R, θ and τ can be calculated as follows.  Radius R. The turn angle α is given by the following equation (see Figu re II.4.3): (II.2) The relation between α and the β inclination is given by (see Figure II.4.2 a): (II.3) Rearranging the two equations, the radius R can be calculated as follows: (II.4)  Kink angle θ The sides of the triangle shown in Figure II.4.2 (a) are given by: (II.5) (II.6) Thus, from the Pythagorean Theorem, c is given by: (II.7) And, from the relation between c and θ shown in Figure II.4.2 (b), c is also given by: 57 (II.8) Thus the kink angle θ can be written as a function of the turn angle α and the β inclination angle: (II.9) (II.10) (II.11) Or, rearranging the equation: (II.12)  Torsion angle τ The value of τ can be calculated from α and β by the following formula (Van Meerssche & Feneau-Dupont, 1984): (II.13) The value of τ will be either positive or negative, depending on the handedness of the supercoil. It is interesting to consider two extreme cases. When we are faced with a flat polygon, with and . At the other extreme, when a straight coiled-coil is generated, that has been named a HASO structure, where HASO stays for Helical Arrangements of Stacked Oligonucleotides (Campos et al., 2006). In the latter case and τ and α have the same value. Examples The calculations given above allow a straightforward explanation of the large difference between the geometries of the coiled-coils observed for CG(AT)5 and CG(AT)4, which will be described later. Both should have a similar value of angle θ, since the CG base pairs which define the kink are the same. On the other hand the τ value should be quite different, since there is a difference of two base pairs in the length of oligonucleotide duplexes. 58 The Dodecamer d(CGATATATATAT), [CG(AT)5] As shown in paragraph II.5.1, the molecular structure of this dodecamer could be determined from the X-ray results which showed that it is a right handed coil. From the experimental pattern it could be derived that: , and , thus, from equation II.12, : And from equation II.13, τ is about 57°: In this case: and . The coiled-coil has a radius R of about 12.5 Å, given by equation II.4 ( ). The stereo view of the kink generated by the sticky end CG is shown in Figure II.4.4. Figure II.4.4. Stereo view of the coiled-coil generated by the sequence d(CGATATATATAT), the sticky ends (cyan) are shown. The discontinuity in the coil generates a kink of about 160° (see text for further details); the AT part of the molecule is shown in grey; the axes of the individual duplexes are in yellow (De Luchi et al., 2006). 59 The Decamer d(CGATATATAT), [CG(AT)4] The torsion angle τ is directly related to the twist value ω of the individual base pairs. With ten base pairs instead of twelve and being the average twist value per base pair of about 35°, τ is about ( ). A negative torsion angle is characteristic of left-handed coils. From the diffraction pattern, it is known that: , and , thus, from equation II.12: in this case the value of , of about 7.3°, is significantly different from the value; with a kink angle θ of about 180°, the coiled-coil is very smooth; the torsion angle τ is very similar to α . The radius R of the coiledcoil is of about 47 Å . The difference in the θ values found in the cases aforementioned should be attributed to changes in the intrinsic curvature of the duplexes. Packing interactions may also have an influence. It should be noted that, in the case of CG(AT)4, more complex coil structures, such as 24 residues in 5 turns, have been excluded. The diffraction from such coils would not coincide with the observed patterns. The two coils generated by CG(AT)4 nad CG(AT)5 are shown in Figure II.4.5. 60 Figure II.4.5. (a) Coiled-coil generated by the dodecamer CG(AT)5, with a pitch of about 220 Å and six oligonucleotides per turn. (b) Coiled-coil generated by the decamer CG(AT)4, with a pitch of about 660 Å and twenty-three oligonucleotides per turn. Values of N and have been plotted as a function of θ and τ and vice versa in Figure II.4.6 and Figure II.4.7. In Figure II.4.6 the values of θ and τ have been calculated for and . The geometrical characteristics of the coiled-coils generated by the sequences highlighted in Figure II.4.6 are summarized in Table II.2. Figure II.4.6. Values of N and as a function of θ and τ. Four examples are highlighted: d(CGTATATA), in red, d(ATATATCG), in blue, d(CGATATATATAT) in yellow and d(CGATATATAT) in green. 61 Figure II.4.7. Values of θ and τ as a function of N and . Table II.2. Geometrical characteristics of the coiled-coils generated by CG(TA)3, (AT)3CG, CG(AT)4 and CG(AT)5. Sequence β (°) N α (°)(1) (2) θ (°) θ' (°) (3) τ (°) R (4) CG(TA)3 5 4 90 0.062 172.93 7.06 0.0038 89.8 1.61 (AT)3CG 5 5 72 0.051 174.12 5.87 0.3126 71.8 1.93 CG(AT)4 28 23 15.65 0.064 172.7 7.3 0.971 -13.8 47.3 CG(AT)5 20 6 60 0.171 160.3 19.7 0.545 57 12.57 (1) Equation II.2. (2) Equation II.12. (3) Equation II.13. (4) Equation II.4. 68 Table II.4. Conformational parameters of Hoogsteen DNA*. Angles Atom α β γ δ ε ζ χ Twist C1’- C1’ Å Rise, Å A3 - - 23.4 134.9 -149.4 -117.3 63.8 27.50 8.2 3.61 T4 -70.0 143.1 72.4 105.5 -168.2 -92.3 -143.5 37.18 8.6 3.05 A5 -51.0 177.1 36.6 125.1 -167.0 -103.9 55.7 34.05 8.2 3.38 T6 -40.0 170.8 30.8 131.2 -69.4 -104.2 -113.6 42.16 8.1 3.33 A7 -62.4 175.1 59.3 124.9 -177.7 -90.0 46.5 27.11 8.0 3.68 T8 -57.2 -168.0 34.1 158.6 -141.2 -144.1 -83.8 42.16 8.0 3.33 A9 -78.9 145.7 65.4 118.4 149.1 -119.2 67.2 34.05 8.1 3.38 T10 -160.9 -88.7 104.4 107.4 100.7 -55.2 -104.4 37.18 8.2 3.05 A11 -146.1 -90.3 67.9 138.4 147.8 -96.7 87.2 27.50 8.6 3.61 T12 -87.1 -133.7 71.3 141.6 - - -78.9 - 8.2 - * The values have been calculated with the 3DNA program, based on C1’-C1’ vectors (Lu & Olson, 2003). The overall structure is shown in Figure II.5.4. It is a right handed coiled-coil, that shows kinks at the position of the CG base pairs, where the phosphodiester chain is interrupted (see Figure II.5.1). The kinks result in a strong compression on the major groove and opening of the minor groove, as it is apparent in Figure II.5.4, in agreement with classic studies (Dickerson et al., 1983). The absence of two phosphate residues facilitates this distortion. The central AT decamer forms a straight duplex, as it is evident in Figure II.5.3. The terminal A∙T base pairs form a large angle (equivalent to roll). The CG dimer sequence is compressed between both terminal A∙T pairs. Since the structure is practically isomorphous with d(AT)6, the origin of the kinks should be attributed to the discontinuity of the phosphodiester chain, rather than to the presence of a short CG stretch. 69 Figure II.5.4. View of two turns of the coiled-coil (a) and detail of two consecutive duplexes (b). The axis of the duplexes is shown in green, whereas the axis of the coiled-coil is shown in cyan. The molecules are projected onto the plane formed by the duplex axes (green), so that the widening of the minor groove in the kink region may be easily appreciated. It is interesting to note that decamer d(CGACGATCGT) also crystallizes as a continuous duplex with its sticky ends paired (Qiu et al., 1997), but as a standardWatson-Crick straight double helix. The results obtained with the dodecamer d[CG(AT)2GC(AT)2] (see paragraph II.5.6) also show straight double helices in the Bform. Since the starting sequence CGA is the same in the case of CG(AT)5, it is tempting to speculate that the coiled-coil conformation requires Hoogsteen base pairs. The coiled-coil is a stable, rigid structure which represents a new conformation of DNA, as part of the polymorphism found in AT sequences reviewed elsewhere (Abrescia et al., 2004). Such sequences are very abundant in non-coding regions of the genome (Abrescia et al., 2004), but their structure and function are not known. 70 II.5.2 d(CGCGATATATAT) d[(CG)2(AT)4] The sequence (CG)2(AT)4 has been crystallized at 13°C using the hanging-drop vapor diffusion technique and 2-methyl-2,4-pentanediol as a precipitant. Due to the high nucleation rate, instead of big single crystals, several small crystals, together with precipitate and thin needles, have been obtained. Only two crystals could be frozen and their diffraction pattern collected. From now on they will be referred to as “D18B2” and “D17A3”. The D18B2 crystal The D18B2 crystal was obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 1.0 mM Spermine Tetrahydrochloride, 2.5 mM MgCl2, 11.1 mM Thymidine and 1 µl of suspension of 4-aminophenylsulfon. Typical diffraction patterns of D18B2 crystal are shown in Figure II.5.5 and Figure II.5.6. The D18B2 crystal is isomorphous to crystals generated by the sequences d[CG(AT)5], d[(AT)5CG] and d(AT)6. The sequence (CG)2(AT)4 generates a coiledcoil, where the duplexes axes are inclined an angle of about 11° with respect to the major coil axis. Unlike d[CG(AT)5] and d[(AT)5CG] crystals, in this case, no streaks have been detected in the diffraction patterns, suggesting the absence of the screw disorder found in the crystals of the isomorphous sequences. Figure II.5.5. Oscillation patterns (2° and 3°) obtained from crystal D18B2, the long c* axis of the unit cell is approximately vertical. (a) The stacking reflections at 3.25 Å deviate about 11° ( ) from a meridional orientation. (b) Meridional 00l Bragg reflections (with ) are evident. 71 Figure II.5.6. Oscillation pattern (3°) of crystal D18B2; Bragg reflections in the equatorial region (with Miller indices 10l, 20l, 30l) are shown. No streaks are visible. The data set has a maximum resolution of about 5.5 Å. It could be indexed in P61 space group with the following unit cell: and . Dataset statistics are shown in Table II.5. Due to the low resolution of the data set, it has not yet been possible to solve the structure. Table II.5. Dataset statistics for crystal D18B2. Crystal D18B2 DNA sequence d(CGCGATATATAT) Space group P61 Unit cell (Å) Resolution (Å) 5.5 Wavelength (Å) 0.977 Unique reflections 782 Completeness (%) 96.7 Rint (%) 7.3 I/σ 11.7 Given the sequence of the oligonucleotide, the DNA duplexes are organized in a staggered fashion and thus give rise to a continuous DNA double helix with nicks in both strands, as shown in Figure II.5.7. Nicks in either strand can be located only at distances of four base pairs. 72 Figure II.5.7. Model of the organization of the oligonucleotides in the crystal. The unit cell parameters correspond to a cell that contains stacks of six dodecamers along the c direction (equivalent to 72 base pairs per unit cell) with their positions related by the helical P61 symmetry. The volume per base pair is of about 1900 Å3, which indicates a large amount of solvent in the crystal structure. The stacking reflections at about 3.25 Å are found at both sides of the meridian, indicating that the DNA duplexes are inclined an angle of about 11° from the meridional or c direction (Figure II.5.5 a). The absence of layer lines with continuous diffraction is probably due to the presence of four CG base pairs instead of only two, as in CG(AT)5 and (AT)5CG. The CG base pairs probably stabilize the structure. The height h occupied by one duplex along the c direction of the unit cell corresponds to about 37.9 Å ( ); the length l of one dodecamer, taking into account its inclination of about 11° is 38.6 Å ( ). The projection of one duplex onto the equatorial plane is thus of 7.4 Å (given by: ). The geometrical parameters of the coiled-coil generated by (CG)2(AT)4 are summarized in Table II.6. The kink angle θ between the two straight parts of the duplex (the (AT)4 and the (CG)2 part) is about 11°. The dimensions of the coiled coil are as follows: pitch (P) 227.5 Å; radius (R) 7.4 Å; inclination ( ) 11°; kink angle (θ) 11°; number N of dodecamers per turn, six. Table II.6. Geometrical parameters of the coiled-coil generated by (CG)2(AT)4. Crystal P (Å) N l (Å) Rise (Å) (°) R (Å) θ (°) α (°) τ (°) D18B2 227.5 6 38.6 3.22 11 7.4 169 60 59 P is the pitch of the coiled-coil. N is the number of oligonucleotides per turn. l is the length of one oligonucleotide (taking into account its inclination ). is the inclination of the axis of the minor helix with respect to the major coil. R is the radius of the cylinder described by the coiled-coil (eq. II.4). θ is the kink angle between two consecutive oligonucleotides (eq. II.12). α is the turn angle (eq. II.2). τ is the torsion angle (eq. II.13). 73 The D17A3 crystal The D17A3 crystal was obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 1.0 mM Spermine Tetrahydrochloride and 5 mM BaCl2. Typical diffraction patterns are shown in Figure II.5.8. The structure is different from the one described above for the D18B2 crystal. The DNA duplexes axes are almost parallel to the major coil axis. In fact, the stacking reflections at about 3.2 Å practically do not deviate from the meridional direction and appear like an arc rather than a sharp straight reflection. The first layer line of B form DNA is visible at about 32 Å resolution (Figure II.5.8 b). Therefore crystal D17A3 forms a standard continuous double helix with a repeat of about 10 base pairs per turn ( ) and an average twist per base pair, , of 36°. The absence of one phosphate in the phosphodiesteric chain does not alter the structure and the helix is practically continuous. Figure II.5.8. Oscillation patterns (5° and 15°) of the P17A3 crystal. (a) The stacking reflections appear like an arc, the inclination of the DNA duplexes is thus very small. (b) A few streaks are evident, but rather than a splitting of the layer line typical of coiled-coils, they appear to be the typical layer lines of the B DNA, only the 1st layer line is indicated. 74 II.5.3 d(ATATATATATCG) d[(AT)5CG] The sequence (AT)5CG was crystallized at 16 °C using the hanging-drop vapor diffusion technique and 2-methyl-2,4-pentanediol as a precipitant. Crystals (Figure II.5.9) have been obtained in the following conditions: 0.5 mM DNA duplex, 25 mM NaCacodylate pH 6.5, 1.5 mM Spermine Tetrahydrochloride, 50 mM KCl and 1 mM CoCl2. Only one crystal has been tested, from now on it will be referred to as “D34A5”. Figure II.5.9. (a, b) Microscopic photographs of (AT)5CG crystals. Typical diffraction patterns of crystal D34A5 are shown in Figure II.5.10. The diffraction pattern shows the coexistence of Bragg reflections and layer lines with continuous diffraction. The Bragg spots have a maximum resolution of about 6 Å. The data set could be indexed in a hexagonal space group with the following unit cell: a and b parameters about 26.3 Å and c parameter about 217 Å. The sequence (AT)5CG generates a coiled-coil with six duplexes per turn (i.e., 72 base pairs along the c direction of the unit cell). The volume per base pair is thus of about 1805 Å3 which indicates a large amount of solvent in the crystal ( ). The continuous layer lines appear at spacings that correspond to the 217 Å-repeat. The stacking reflections, at about 3.2 Å, are found at both sides of the meridian, indicating that the DNA duplexes are inclined an angle of about 15° from the meridional or c direction. Crystalline regions give rise to the Bragg spots, on the other hand partially disordered regions give rise to the continuous diffraction. The dimensions of the molecules appear to be the same in both regions, because the c spacings of the Bragg reflections and of the continuous layer lines are identical. 75 Figure II.5.10. Oscillation patterns (15°) of crystal D34A5. (a) The long c* axis of the unit cell is approximately vertical. Meridional Bragg reflections (006 and 0012) are evident. The prominent stacking reflections at about 3.2 Å resolution deviate about 15° ( ) from the meridional direction. (b) Oscillation pattern of the same crystal rotated 90° with respect to (a), it approximately corresponds to the equatorial region of the crystal. A hexagonal symmetry is evident. Due to the orientation of the crystal, in this region most of the reflections are overlapped. Given the sequence of the oligonucleotide, the DNA duplexes are organized in a staggered fashion and thus give rise to a continuous DNA double helix with nicks in both strands. As shown in Figure II.5.11, due to the terminal of guanine and cytosine, the pairing scheme is unique. Figure II.5.11. Model of organization of the oligonucleotides in the crystal. They form duplexes with sticky ends that generate infinitely long molecules with staggered nicks in both strands. 76 The continuous layer lines are organized in groups that emanate from the meridional region. A prominent group is centered on the seventh layer line, which corresponds to 31 Å ( ), equivalent to the first layer line of a DNA duplex, taking into account its β inclination of 15°. Scattering of the layer lines is centered around the 7th, 14th, 21st, etc., layer lines, which correspond to the first three layer lines of the original scattering of a continuous DNA duplex (see Figure II.5.10). The 21st layer line region is significantly weaker than the rest. The coiled-coil structure results in splitting of the layer lines of the original double helix. Both the Bragg reflections and the continuous layer lines can be explained by the same coiled-coiled model. The height h occupied by a dodecamer along the c direction of the cell corresponds to about 36.2 Å ( ). Its approximate length l is of about 36.8 Å ( ) and its projection onto the equatorial plane is of about 7 Å ( ), that is an estimation of the radius R of the coiled-coil. The geometrical parameters of the coiled-coil are summarized in Table II.7. Table II.7. Geometrical parameters of the coiled-coil generated by the sequence (AT)5CG. Crystal P (Å) N l (Å) Rise (Å) β (°) R (Å) θ (°) α (°) τ (°) D34A2 217 6 37.4 3.2 15 9.7 165 60 58.3 P is the pitch of the coiled-coil. N is the number of duplexes per turn. l is the length of one duplex (taking into account its inclination β). β is the inclination of the axis of the minor helix with respect to the major coil. R is the radius of the cylinder described by the coiled-coil (eq. II.4). θ is the kink angle between two consecutive duplexes (eq. II.12). α is the turn angle (eq. II.2). τ is the torsion angle (eq. II.13). 77 II.5.4 d(ATATATATATGC) d[(AT)5GC] The oligonucleotide d(ATATATATATGC) was crystallized at 13 °C and 17 °C using the hanging-drop vapor diffusion technique and 2-methyl-2,4-pentanediol as a precipitant. Several crystals were obtained in very similar conditions: 0.5 mM DNA duplex, 25-50 mM NaCacodylate pH 6.5, 1.0-2.0 mM Spermine Tetrahydrochloirde and 20 mM KCl, with or without the addition of 16.7 mM Thymidine. The presence of Thymidine did not improve the diffraction pattern of the crystals obtained. The best diffracting crystal (Figure II.5.12 b) was obtained at 13 °C in the following conditions: 0.5 mM DNA duplex, 25 mM NaCacodylate pH 6.5, 1.0 mM Spermine Tetrahydrochloirde, 50 mM KCl and MPD 30%. This crystal will from now on be referred to as “D31B1”. Figure II.5.12. (a) (AT)5GC crystals. (b) D31B1 crystal during data collection at the BM16 beamline at the ESRF, Grenoble. Typical diffraction patterns of crystal D31B1 are shown in Figure II.5.13, Figure II.5.14 and Figure II.5.15. The diffraction patterns show the coexistence of Bragg reflections and layer lines with continuous diffraction. The Bragg spots approximately lay on the layer lines. Due to several factors, as low resolution (5.5 Å), intrinsic symmetry of the structure, presence of a small pseudo-cell, only a limited number of spots is visible. The combination of all these factors prevented the automatic determination of the unit cell which has instead been manually determined. The stacking reflections, at about 3.30 Å resolution along the meridional axis (Figure II.5.17) indicate that this sequence forms a straight helix. Apparently, the diffraction patterns could be indexed as a pseudohexagonal unit cell with and . These parameters could be obtained indexing the diffraction patterns as shown in Figure II.5.13. 84 As shown in Figure II.5.19, the strong DNA stacking reflections are found at both sides of the meridian, indicating that the DNA duplexes are inclined about 24° from the meridional or c direction. The height h occupied by one oligonucleotide along the c direction of the unit cell is approximately 34.2 Å ( ). Thus, considering the β inclination of the duplexes of about 24°, the length l of one DNA duplex is of 37.5 Å ( ); the average rise of the DNA thus is of 3.12 Å ( ). Figure II.5.19. Oscillation pattern (15°) obtained from D34D3 crystal. The long c* axis of the unit cell is vertical. The prominent stacking reflections deviate about 24° (β) from the meridional orientation. Bragg spots with Miller indices 10l are also evident. 85 Figure II.5.20. Oscillation patterns (5°) of crystal D34D3. (a) An enlarged view of the meridional diffraction pattern is shown: layer lines spacing corresponds to half of Bragg spots spacings. The 8th layer line of the B-DNA and the Bragg spots with indices 0014 and 0028 are indicated. (b) The equatorial region of the diffraction pattern: a pseudo-hexagonal symmetry is visible. Scattering at the layer lines is centered on the 8th and 14th layer lines. They correspond to the first two layer lines of the original scattering of a continuous DNA duplex with about ten base pairs per turn. The spacing of the continuous diffraction is about 240 Å, i.e. half of the Bragg spacing along the c direction (479 Å). The first DNA layer line (or the 8th layer line of the dodecamer) is at about 29.9 Å resolution ( ), which is approximately the length of 10 base pairs of the dodecamer (taking into account the inclination β of the oligonucleotides: , where l is the length of a duplex of ten base pairs with an average rise of 3.2 Å). The diffraction pattern can be interpreted as due to a mosaic structure: crystalline regions that give rise to the Bragg spots and partially disordered regions in which the molecules are randomly displaced by vertical and rotational movement. The DNA duplexes are organized in a staggered fashion and thus give rise to a continuous double helix with nicks in both strands (Figure II.5.21). 86 Figure II.5.21. A model of organization of the oligonucleotides in the crystal. They form duplexes with sticky ends that generate infinitely long molecules, with staggered nicks in both strands. Data sets from other crystals have been collected (see Table II.8 for crystallization conditions): all of them show the coexistence of Bragg spots and continuous layer lines. Details relative to each of them are given in the paragraphs below. The D28C21 crystal Typical diffraction patterns of D28C21 crystal are shown in Figure II.5.22. The sequence generates a coiled-coil with 14 oligonucleotides per turn. Layer lines and Bragg spots have the same spacings of about 450-480 Å. The inclination β of the duplexes axis is of about 25° with respect to the major coil axis. Figure II.5.22. Oscillation patterns (3°) obtained from crystal D28C21. (a) The long c* axis of the unit cell is approximately vertical. Meridional Bragg reflections (0028 and 0042) are shown. The stacking reflections at 3.25 Å deviate about 25° (β) from a meridional orientation. (b and c) The spacing of the layer lines corresponds to spacing of the Bragg reflections; due to the strong fiber background, the spacing distances had to be manually measured with the funcction MEASURE CELL implemented in the program MOSFLM (Leslie, 1992). 87 The D28C1 crystal The diffraction pattern of D28C1 crystal appears slightly different with respect to the diffraction of crystals D28C21 and D34D3. Apparently, the diffraction pattern could be indexed as shown in Figure II.5.23 (b): Bragg spots with indices 007, 0014 and 1014 are shown. Reflection 0014 is not exactly at the same level of reflection 1014: it appears in between the hypothetical 1012 and 1014 reflections. Therefore its Miller index, instead of 0014, is 0027, and the coiled-coil has 6.75 oligonucleotides per turn (i.e. 27 duplexes in 4 turns). From the meridional diffraction pattern, the value of the c parameter of the unit cell can be determined: if the indexing shown in Figure II.5.23 (b) is correct, the c parameter corresponds to about 260 Å. If, as said before, the true index of reflection 0014 is 0027, the c dimension of the unit cell must be approximately doubled to 500 Å, the latter value would be in agreement with the c values found for the other crystals of this sequence. Figure II.5.23. Oscillation patterns (a, 5°; b, 15°) of crystal D28C1. (a) Approximately meriodional diffraction pattern. The stacking reflections at 3.28 Å deviate about 20° (β) from a meridional orientation. (b) Enlarged view of the meridional diffraction: Bragg reflections with hypothetical Miller indices 007 and 0014 are shown; also Bragg reflections in the equatorial region (Miller indices 10l) are visible. 88 The D34D6 crystal Typical diffraction patterns of PD34D6 crystal are shown in Figure II.5.24. In Figure II.5.24 (a), the long c* axis of the unit cell is approximately vertical. The stacking reflections at 3.25 Å deviate about 24° from the meridional orientation. Apparently the c parameter, manually measured, is of about 240 Å, but the presence of “double spots” suggests a bigger c parameter of about 900 Å (Figure II.5.24 c). Unfortunately, the presence of strong fiber diffraction does not allow an accurate determination of the unit cell parameters. Figure II.5.24. Oscillation patterns (a, 15°; b and c, 5°) obtained from crystal D34D6. (c) Enlarged view showing the presence of “double spots”. Table II.8. Summary of the characteristics of GC(AT)5 crystals, for all of them a pseudo-hexagonal symmetry is assumed. The maximum resolution is of about 11 Å. See text for further details. Crystal Unit cell (Å) β (°) l (Å) Rise (Å) Streaks N Crystallization conditions D34D3 24° 37.4 3.12 Yes 7 0.5 mM DNA, 50 mM NaCacodylate pH 6.5, 10 mM MgCl2, 1 mM CoCl2, 1.5 mM Spermine and 28% MPD. D28C21 a and b not determined 25° 37.8 3.15 Yes 14 0.5 mM DNA, 25 mM NaCacodylate pH 6.5, 50 mM KCl, 1 mM Spermine and 28% MPD. D28C1 22° - 3.25 Yes 7 or 6.75 0.5 mM DNA, 25 mM NaCacodylate pH 6.5, 20 mM NaCl, 1 mM Spermine and 30% MPD. D34D6 c apparently is about 240 Å, the presence of “double spots” clearly suggests a bigger c dimension of about 900 Å. a and b are about 50 Å 24° - 3.25 Yes - 0.5 mM DNA, 50 mM NaCacodylate pH 6.5, 50 mM KCl, 1 mM NiCl2, 1.5 mM Spermine and 30% MPD. 90 II.5.6 d(CGATATGCATAT) d[CG(AT)2GC(AT)2] The sequence d(CGATATGCATAT) was chosen in order to better understand the features of the structure of d[CG(AT)5] (De Luchi et al., 2006). Our results indicate that the DNA duplexes are organized in a staggered fashion and thus give rise to a continuous DNA double helix with nicks in both strands. The base pairing, shown in Figure II.5.25 appears to be unique. Figure II.5.25. Model of the organization of the oligonucleotides in the crystal. They form duplexes with sticky ends that generate infinitely long molecules, with staggered nicks in both strands. The sequence d(CGATATGCATAT) was crystallized at 17 °C by using the hanging-drop vapor diffusion technique and 2-methyl-2,4-pentanediol as a precipitant. Long needles were obtained from several solution conditions (Figure II.5.26): 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 0.5-1.5 mM Spermine Tetrahydrochloride, with or without additives (Thymidine, TMAO, caprolactam, MgCl2, BaCl2, CoCl2, CaCl2). The addition of divalent cations seems to be essential for a good diffraction, while the addition of the other additives (Thymidine, TMAO and Caprolactam) did not have any appreciable influence on the diffraction pattern of the crystals obtained. Seven crystals have been obtained (D50A5, D59D4, D50A6, D59D6, D59D6-1, D60A2 and D60A3); details relative to the each of them are reported in Table II.9. Figure II.5.26. Microscope photographs of crystals of d(CGATATGCATAT), all of them appeared as long needles. The following addivites have been used: (a) D59D6 crystal: 5 mM CaCl2, TMAO and Thymidine; (b) D59D4 crystal: 2.5 mM CaCl2; (c) D50A6 crystal: 5 mM CaCl2; (d) D60A3 crystal: BaCl2 and Caprolactam. 91 Typical diffraction patterns are shown in Figure II.5.27. The diffraction patterns show the coexistence of Bragg reflections and layer lines with continuous diffraction. The Bragg spots have a maximum resolution of 7.0 Å. The data set could be indexed on a hexagonal unit cell with the following parameters: and . The unit cell parameters correspond to a cell which contains three stacks of six dodecamers, equivalent to 216 base pairs per unit cell. The volume per base pair is thus 1998 Å3, which is in agreement with previous results. The volume per base pair indicates the presence of a large amount of solvent, which could explains the low resolution observed in the diffraction patterns. The continuous layer lines appear at spacings that correspond to the 230.5 Å-repeat. The same model can explain both the Bragg spots and the continuous layer lines. The strong stacking reflections are slightly off-meridional and not sharp, as shown in Figure II.5.27. This is evident in Figure II.5.27 (c), where the stacking appears like an arc more than like a sharp streak. In the same frame, the splitting of the layer lines, a typical feature of the coiled-coils, is also evident. The DNA duplexes are thus inclined from the meridional or c direction by an angle β of 0° to 5° degrees. Figure II.5.27. Oscillation patterns of three different crystals of d(CGATATGCATAT).(a) D60A2 crystal (3°); (b) D59D6 crystal (5°) and (c) D59D4 crystal(15°). 92 When the duplexes axes are parallel (i.e. ), the height h occupied by one dodecamer along the c direction corresponds to 38.4 Å ( ), with an average rise of 3.2 Å ( ). The dodecamers are organized end-to-end and build a continuous coil with nicks corresponding to the sticky ends, as shown in Figure II.5.25. Neighbor duplexes are practically coaxial. The overall rotation of one duplex with respect to its neighbor in a column is called Ω. The value of Ω can be exactly determined once the number N of duplexes stacked in one unit cell is known. must be an exact multiple of 360°: (II.14) with m a whole number. In the case of the sequence CG(AT)2GC(AT)2, , and . is also related to the individual base pair twist angle by the relation: (II.15) where n is the number of base pairs in a duplex and is the average twist of its base steps. In one unit cell there are stacks of 72 base pairs ( ). The average twist value for each base pair is thus of 35° ( ); consequently there are 10.3 base pairs per turn ( ). The average DNA twist in solution is 10.4 base pairs per turn, therefore it could be concluded that the absence of one phosphate does not change the organization of the double helix. The sequence d(CGATATGCATAT) generates a practically continuous double helix of standard B-form DNA. On the other hand, the average twist value of 35° is slightly smaller than the 35.9° value reported in the literature for mixed CG/AT sequences (Gorin et al., 1995). This observation indicates that although the overall structure of the DNA is not affected by the absence of a phosphate, the individual twist values suffer a small change. When (Figure II.5.27 (c)), all previous calculations are practically still correct. Due to the β inclination of helical axis with respect to the major coil axis, some parameters are slightly different. The crystal D59D4 has been indexed in the following hexagonal unit cell: and . The height h occupied by one dodecamer along the c direction is of about 38 Å ( ). Therefore, the length l of a dodecamer corresponds to 38.14 Å ( ). Its projection onto the equatorial plane is of 3.32 Å ( ). The radius of the coiled-coil can be estimated from the latter value. 93 In B-form DNA, the 9th layer line is generally found at 3.55 Å; on the other hand, in the meridional diffraction pattern shown in Figure II.5.28, the 9th layer line corresponds to 3.44 Å. This is probably due to the fact that the B-form-DNA 9th layer line coincides with the dodecamer 11th layer line, whose meridional spacing d corresponds to 3.44 Å (). Although the sequence d(CGATATGCATAT) was chosen just to improve the crystallization of d[CG(AT)5], a different structure, probably due to the presence of the central GC bases, was obtained. The four CG base pairs form Watson-Crick hydrogen bonds and therefore force the flanking ATs to also pair through Watson-Crick bonding instead of Hoogsteen, as seen in d[CG(AT)5] (De Luchi et al., 2006). Probably thanks to the coherence in the H-bonds throughout the whole molecule, the bends formed by the two terminal bases are less pronounced and the resulting structure is a standard B-DNA instead of a coiled-coil. The characteristics of the diffracting crystals are summarized in Table II.9 Figure II.5.28. Oscillation pattern (3°) obtained from crystal D60A2, which has . The 9th layer line of the B-form DNA, found at 3.44 Å instead of 3.55 Å, corresponds to the 11th layer line of the dodecamer d(CGATATGCATAT). In the table on the right, the meridional spacings d for the dodecamer have been calculated as follows: , where n is the layer line number. 100 The sequence CG(AT)4 gave rise to coiled-coiled structures, where the duplexes axes are inclined with respect to the major coil axis by 25°-32° (β), depending on the crystal; in the case of cystal P10B1, shown in Figure II.6.1, . Figure II.6.1. Oscillation pattern (15°) obtained from the crystal P10B1. (a) The long c* axis of the unit cell is approximately vertical. The stacking reflections at about 3.2 Å deviate about 28° (β) from a meridional orientation. (b) Enlarged view of the center of the diffraction. Meridional Bragg reflections close to 0044n reflections (with n a whole number) are shown. Given the sequence of the oligonucleotide, the DNA duplexes are organized in a staggered fashion that gives rise to a continuous DNA double helix with nicks in both strands, a model of the organization of the oligonucleotides is shown in Figure II.6.2. Figure II.6.2. A model of organization of the oligonucleotides in the crystal. They form duplexes with sticky ends that generate infinitely long molecules, with staggered nicks in both strands. Due to the presence of the sticky end d(CG), the pairing scheme appears to be unique. 0044 0088 00l, l close to 44n 101 Due to the extremely large c parameter of the unit cell, we encountered the “phioverlap” problem, which becomes extremely serious when the very long c axis gets close to being parallel to the beam. In order to avoid this problem, the detector had to be moved as far back as possible and the crystal had to be oriented with the long cell axis roughly (but not perfectly) aligned with the rotation axis of the goniometer. If the long axis and the beam are perfectly aligned, there will be an uncollected region along c*. It would also be useful to have a large detector and an X-ray beam with a diameter and a divergence as small as possible. The characteristics of the crystals of the sequence d(CGATATATAT) are summarized in Table II.12. Similar results were obtained in all cases, except in the P8C63 crystal which presents a larger unit cell. The P10A4 crystal Crystallization and Data Collection A hanging drop of 10 μl was prepared with the following composition: 0.33 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 50 mM KCl, 1 mM Spermine Tetrachloride and 10% of MPD. The drop was incubated at 18 °C over 800 μl of a reservoir consisting of an aqueous solution with 15% of MPD. Over more than two weeks, the MPD concentration of the reservoir was increased from the 15% to 33%. At this percentage of precipitant, crystals appeared suddenly and grew very rapidly. The crystal, with slightly curved faces and some irregularities, had a trapezoidal shape with the longest dimension of 100 μm, the other edges could not be measured. Due to the high sensitivity to the temperature, the crystal had to be handled with extreme care frozen at 4 °C. This crystal will from now on be referred to as “P10A4”. The P10A4 crystal was mounted in a nylon loop at 4 °C and flash frozen in liquid nitrogen. No extra cryoprotectant was used. Data collection was carried on at 100 K at beamline BM16 at the ESRF, Grenoble. Hundred eighty degrees (180°) of oscillation data with a rotation of 3° per image were collected, with a detector distance of 350 cm; one diffraction pattern was also collected at the detector distance of 150 cm, in order to record the stacking reflections of DNA (at about 3.25 Å resolution). 102 Typical diffraction patterns are shown in Figure II.6.3. The stacking reflections, found at both sides of the meridian, indicate that the DNA duplexes are inclined from the meridional or c direction of about 25° (β). Due to the orientation of the crystal, only 60° have been taken into account during the integration, the remained 120° correspond to a region of the crystal very close to the “equatorial” region of the diffraction, for this reason most of the reflections overlap, as shown in Figure II.6.3 (b) and their indexing and integration is not possible. A striking feature is the absence of h00 reflections. The Diffraction and the Unit Cell The diffraction patterns showed anisotropicity and very high mosaicity, as shown in Figure II.6.3 (a). The data have been indexed in the monoclinic system with the following unit cell: , , and . The ratio between the b and a parameters of the unit cell is equal to 1.67 ( ), which is very close to the square root of three ( ). In the case of the crystal P10A4, the fact that a multiplied by is only approximately equal to b, indicates a pseudohexagonal symmetry. The conversion of the monoclinic unit cell into a pseudohexagonal cell in shown in Figure II.6.4. Figure II.6.3. Oscillation patterns (3°) obtained from the P10A4 crystal. (a) The long c* axis is approximately vertical. The prominent stacking reflections at about 3.25 Å deviate about 25° (β) from a meridional orientation. (b) Oscillation pattern close to the equatorial region. The high inclination of the crystal respect to the rotation axes causes the overlapping of many reflections and does not allow their integration. 103 Figure II.6.4. The monoclinic cell parameters are , , and ; the monoclinic unit cell (red) can be easily converted in a pseudo-hexagonal unit cell (blue) with a and b approximately equal to 30 Å. In this case the ratio between a and b is not exactly equal to the square root of three ( ), in which case it would strongly suggest a hexagonal symmetry. The c parameter of 646.31 Å corresponds to a cell containing stacks of 22 decamers ( , with h equal to 29.4 Å); 0044n reflections, with n a whole number, are evident along the meridian, as shown in Figure II.6.5. The length of the repeating unit along the c direction is thus given by the Bragg distance d0044, which is equal to 14.7 Å and corresponds to half a decamer considering its β inclination of 25°. Figure II.6.5. Oscillation pattern (3°) of P10A4 crystal. The c* axis is approximately vertical; the meridional Bragg reflections (with Miller indices close to 0044) are s ho wn . 104 The P8C2, the P8D41, the P10B1 and the P9C3 crystals The P8C2, the P8D41, the P10B1 and the P9C3 crystals are practically isomorphous with the P10A4 crystal (see previous paragraph). Typical diffraction patterns are shown in Figure II.6.6, Figure II.6.7, Figure II.6.8 and Figure II.6.9. The P8C2 crystal The crystal P8C2 was obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 20 mM KCl and 1 mM Spermine Tetrahydrocloride. Typical diffraction patterns are shown in Figure II.6.6. Figure II.6.6. Oscillation patterns (5°) of the P8C2 crystal. (a) The long c* axis is approximately vertical; the stacking reflections at about 3.25 Å found at both sides of the meridian indicate a β inclination of about 25° of the duplexes with respect to the coiled-coil axis (b) Approximately equatorial region of the crystal, the b* and the a* axes are shown. 105 The P8D41 crystal The P8D41 crystal was obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 35 mM KCl and 1 mM Spermine Tetrahydrocloride. The data set has been indexed in the following orthorhombic unit cell: , , and , with 23 duplexes per turn. Typical diffraction pattern is shown in Figure II.6.7. Figure II.6.7. Meridional oscillation pattern (5°) of crystal P8D41 Bragg spots with Miller indices close to 0046 are evident, indicating that there are 23 duplexes per turn. 106 The P10B1 crystal The P10B1 crystal was obtained in the following conditions: 0.4 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 50 mM KCl and 1 mM Spermine Tetrahydrocloride. The meridional diffraction pattern of crystal P10B1 is shown in Figure II.6.1. Equatorial diffraction is shown in the figure below. Note the absence of h00 reflections, which are also absent in the other cases (see Figure II.6.3 b and Figure II.6.6). Figure II.6.8. Oscillation pattern (5°) of the equatorial region of crystal P10B1. 107 The P9C3 crystal The P9C3 crystal was obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 20 mM NaCl and 1 mM Spermine Tetrahydrocloride. The data set could not be automatically indexed. The β inclination of the duplexes axes is of about 30°. A typical diffraction pattern is shown in Figure II.6.9. Figure II.6.9. Meridional oscillation pattern (5°) of crystal PD9C3; the β inclination of the duplexes with respect to the major coil axis is of about 30°. 108 The P8C63 crystal Crystallization and Data Collection A hanging drop was set up in the following conditions: 0.5 mM DNA, 50 mM NaCacodylate pH 6.5, 20 mM KCl, 1 mM Spermine Tetrachloride and 5%MPD. The hanging drop was incubated at 18 °C over 800 μl of the reservoir consisting of an aqueous solution of MPD 25%. The drop developed precipitate seconds after being mixed only to become clear in some minutes. In a few days, the MPD concentration of the reservoir was increased from the 25% to 35%, at this percentage of precipitant, several needles and small crystals appeared, but no amorphous precipitate was present anymore. In order to eliminate the needles and the small crystals, the MPD concentration was reduced and the temperature was cyclically changed: 37 °C (30 minutes) 20 °C (1 hour) 16 °C (days)  14 °C (days). This cycle was repeated twice and eventually the temperature was maintained at 16 °C. Several small crystals were obtained which presented a “cubic” shape, with one edge of 50 µm, and the diagonal of 60 µm. The crystals have been frozen at 4 °C. This crystal will from now on be referred to as “P8C63”. 180° of oscillation data with a rotation of 1.5° per image were collected at the beamline BM16 of the ESRF in Grenoble. The Diffraction and the Unit Cell Typical diffraction patterns are shown in Figure II.6.10 and Figure II.6.11. The stacking reflections at about 3.25 Å, found at both sides of the meridian, indicate that the DNA duplexes are inclined about 32° (β) from the meridional or c direction (Figure II.6.10). The data set has been processed in three space groups: Orthorhombic, Monoclinic and Hexagonal (see Table II.11, page 116). The Orthorombic unit cell ( , and , C2221 space group) is approximately four times bigger than P10A4 cell, but shows some contraction (in fact and ). The volume per base pair of 1985 Å3 is therefore significantly smaller than in the P10A4 case (see Figure II.6.12). It appears that the various cycles of cooling and heating of the crystals described above have produced a more compact structure. The c parameter corresponds to a unit cell with stacks of 23 oligonucleotides. 109 As in P10A4 case, the relation between a and b parameters is described by the following equation: , suggesting a pseudo-hexagonal symmetry (see F igu r e II.6.4). Figure II.6.10. Oscillation patterns (1.5°) of crystal P8C63. The stacking reflections at about 3.25 Å deviate approximately 32° from a meridional orientation. Meridional reflections with Miller indices close to 0046 are indicated. Figure II.6.11. Oscillation pattern (1.5°) in a region close to the equator. Due to the inclination of the crystal, the view shown has a substantial deviation away from the true equator. A pseudo-hexagonal symmetry is clearly recognizable. 0046 0092 0092 116 Table II.11. Dataset statistics for crystal P8C63. In parenthesis are the values for the high resolution shell, its lower limits are shown in parenthesis in the resolution line. Crystal P8C63 Sequence d(CGATATATAT) Wavelength (Å) 0.9794 Detector distance (cm) 450 Oscillation range(°) 1.5 β (°) 32 Rise (Å) 3.25 N 23 Streaks No Space group C2221 C2 P63 Unit cell a = 58.921 b = 98.454 c = 645.37 a = 57.489 b ≈ 645 c = 56.217 β ≈ 120° a = b =57.38 c =645 Resolution 8 (8.10) 10 (10.11) 8 (8.10) Total Reflections 3374 8754 7035 Unique reflections 2125 4142 1131 Rint ( % ) 8.63 (15.74) 9.0 (29) 21 (38) I/σ (I) 5.25 (2.86) 2.68 (1.94) 3.64 (1.37) Completeness 91.2 (93.0) 59.2 (73.1) 62.4 (66.7) Eventually, a model in the P21 space group was used, with a unit cell similar to the P63 one: a and b equal to 57.4 Å, c equal to 658 Å, and γ to 120º. With respect to the processing, the c value is larger (658 Å instead of 645 Å reported in Table II.11); in fact the model presents twenty four duplexes per turn instead of twenty three. The asymmetric unit contained the two independent coils found in the smaller unit cell (see Figure II.6.13). The packing of such a unit cell is shown in Figure II.6.15. A section through the model is represented in Figure II.6.16, where the P21 unit cell is indicated, as well as a possible orthorhombic unit cell. The calculated fiber diffraction pattern for this model is practically identical to the diffraction given by the model in the smaller P1 cell and shown in Figure II.6.19. 117 Figure II.6.19. Comparison of the calculated fiber diffraction patterns of the models for the small P1 cell (a) and the larger P21 cell (b) assumed for the P8C63 crystal. In the effort of understand the structure generated by this oligonucleotide, the d(CGATATATAT) fragment has been studied by Dr C. Gonzales by NMR. This part of the work is still in a preliminary stage, but a different result was obtained in a solution with pH below 5. Crystallization trials have been set up at pH 4 and 5. Only amorphous precipitate and phase separation have been obtained (Figure II.6.20). Figure II.6.20. Phase separation and amorphous precipitate obtained at pH 4.5 (a) and 5.0 (b). 118 Summary of the CG(AT)4 structure  The sequence CG(AT)4 crystallizes with monovalent cations.  It generates coiled-coils with 22-24 oligonucleotide duplexes per turn.  It generates a coiled coil with an inclination β of about 30°.  The diffraction patterns are characterized by sharp stacking reflections and the absence of streaks: it has higher crystallinity than d(AT)5, which usually presents streaks in its diffraction patterns.  The maximum resolution of about 5 Å is not sufficient to determine the molecular structure of the crystals.  In analogy with the structure of CG(AT)5 (see paragraph II.5.1), the hydrogen bonds in the (AT)4 part of the structure might be in Hoogsteen conformation, however, there is no evidence of this hypothesis. The characteristics of CG(AT)4 crystals are summarized in Table II.12. Table II.12. Characteristics of CG(AT)4 crystals, not all the diffracting crystals are shown. Crystal Unit cell (Å) Space group Osc. range β N Streaks Crystallization conditions P8D41* P1 5° 25°-28° approx 23 No 0.5 mM DNA, 50 mM NaCac. pH 6.5, 35mM KCl, 1 mM Spermine and MPD 28% P10A4* P1 3° 25° 22 No 0.5 mM DNA, 50 mM NaCac. pH 6.5, 50mM KCl, 1mM Spermine and MPD 33% P8C63 C2221 P21 1.5° 32° 23 No 0.5 mM DNA, 50 mM NaCac. pH 6.5, 20mM KCl, 1mM Spermine and MPD 25% *Crystals P8D41 and P10A4 are practically isomorphous. 120 II.6.2 d(ATATATATCG) d[(AT)4CG] The sequence d(AT)4CG was chosen as a variation of d(CGATATATAT), to see if the change of the sticky end position could generate a different structure. The oligonucleotide was crystallized at 13 °C. The hanging-drop vapor diffusion technique was used with 2-methyl-2,4-pentanediol as a precipitant. The crystals (Figure II.6.21) were obtained in very similar conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 1-1.5 mM Spermine tetrahydrochloride, 1 mM CoCl2 and 10 mM MgCl2, with or without 16.7 mM Thymidine (the Thymidine had no appreciable influence on the diffraction of the crystals obtained). The crystals will be referred to as “D35A1-2”, “D35A1-3”, “D35A1-4”, “D35A4” and “D35A5”. All the crystals gave rise to fiber-like diffraction patterns with very few or no Bragg reflections. Typical diffraction patterns are shown in Figure II.6.22 and Figure II.6.23. The crystals used for the data collection grew in the same drop, but were frozen at different times: D35A1-2 crystal was frozen four weeks after the drop was set up (Figure II.6.22); D35A1-3 and D35A1-4 crystals were frozen four months after the drop was set up (Figure II.6.23 and Figure II.6.24). The d(AT)4CG sequence generates continuous coils, in the form of a standard B-DNA or of a coiled-coil. Figure II.6.21. Microscope photographs of d(ATATATATCG) crystals. They have been obtained in the following conditions: (a) D35A1 crystal: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 10 mM MgCl2, 1 mM CoCl2 and 1.0 mM Spermine tetrahydrochloride. (b) D35A4 crystal: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 10 mM MgCl2, 1 mM CoCl2 and 1.5 mM Spermine tetrahydrochloride. (c) D35A5 crystal: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 10 mM MgCl2, 1 mM CoCl2, 1.0 mM Spermine tetrahydrochloride and 16.7 mM Thymidine. 121 The diffraction patterns shown in Figure II.6.22 are typical of standard B-form DNA. The stacking reflections at approximately 3.25 Å resolution are found along the meridian, indicating that the continuous coil forms infinite straight columns of standard B-form DNA. In the diffraction patterns shown in Figure II.6.23 and Figure II.6.24, the stacking reflections at 3.2 Å are found at both sides of the meridian, indicating that the DNA duplexes are inclined a β angle of 12° to 25° (depending on the crystal) from the meridional or c direction. Figure II.6.22. Oscillation patterns (3°) of crystal D35A1-2 (frozen four weeks after the drop was set up). The two diffraction patterns are rotated 90° one respect to the other. This is a fiber-like diffraction of a B-form DNA, with a maximum intensity at about 3.2 Å resoloution, corresponding to the stacking reflection. The stacking reflections occur at the 10th DNA layer line and correspond to a pseudo-continuous DNA helix of ten stacked base pairs and a pitch P of about 32 Å (B-form DNA). 122 The data set could not be indexed, but a pseudo-hexagonal symmetry could be recognized from the diffraction patterns (see Figure II.6.23 (b) and Figure II.6.24). Bragg reflections of 10l order appeared only every 60°, suggesting a trigonal or pseudohexagonal symmetry. a and b unit cell parameters have been measured and correspond to about 27 Å ( ). The enlarged view of the meridional oscillation pattern shown in Figure II.6.25, reveals the existence of layer lines, whose spacing of about 920 Å corresponds to the unit cell c parameter. Figure II.6.23. Oscillation patterns (5°) of crystal D35A1-3 (frozen four months after the drop was set up). This is a fiber-like diffraction of a coiled-coil. The stacking reflections at 3.25 Å found at both sides of the meridian, indicate that the DNA duplexes are inclined about 17° (β) from the meridional or c direction. The two diffraction patterns are rotated of 90° one respect to the other: (a) approximate fiber meridional projection; (b) approximate fiber equatorial projection. The latter projection is visible due to the high inclination (of about 70°) of the rotation axis with respect to the meridional or c* axis. In (b) a pseudohexagonal symmetry can be recognized; it reflects the packing of the DNA columns projected onto the equatorial plane. 123 Figure II.6.24. Oscillation patterns (15°) of crystal D35A1-4; starting angles: (a) 0°, (b) 60° and (c) 120°; the c* axes is well oriented along the spindle axes, strong 10l reflections appear every 60°, suggesting a hexagonal symmetry. a and b unit cell parameters correspond to about 27 Å ( , corresponds to about 23.5 Å). The height h occupied by one duplex in the unit cell along the c direction can be measured from the first meridional spot in the diffraction pattern (see Figure II.6.25); it corresponds to about . The length l of the DNA duplex corresponds to ( ). Therefore there are 30 oligonucleotides per unit cell along the c direction (). 124 Figure II.6.25. Enlarged view of the oscillation pattern reported in Figure II.6.24 (c). The spacing between layer lines of about 920 Å corresponds to the unit cell c parameter. The height h occupied by one duplex along the c direction (measured from the first meridional spot) corresponds to about 31 Å. There are 30 duplexes per unit cell along the c direction ( ). Diffraction with 10l Miller indices is also evident. Given the sequence of the oligonucleotide, the DNA duplexes are organized in a staggered fashion and thus give rise to a continuous DNA double helix with nicks in both strands, as shown in Figure II.6.26. The duplexes axes can be parallel to the continuous coil axis or can deviate 12° to 25° from it. With these data it is not possible to formulate any hypothesis to explain the existence of the two different structures. In Table II.13 a summary of all diffracting crystals is reported. Figure II.6.26. Model of organization of the oligonucleotides in the crystal. They form duplexes with sticky ends that generate infinitely long molecules, with staggered nicks in both strands. Table II.13. Characteristics of the diffracting crystals of the sequence (AT)4CG. No Bragg reflections have been found in any of these diffraction patterns. Crystal Unit cell Osc. range β Rise (Å) Comments Crystallization conditions D34B3-2 D34B3-4 - +15° 0° 3.25 Pseudo-continuous coil, probably B-form. 0.5 mM DNA, 50 mM NaCac pH6.5, 10 mM MgCl2, 1 mM CoCl2 and 1.0 mM Spermine. D35A1-1 D35A1-2 - +15° 0° 3.2 Pseudo-continuous coil, probably B-form, crystal frozen four weeks after drop D35A1 was set up. 0.5 mM DNA, 50 mM NaCac pH6.5, 10 mM MgCl2, 1 mM CoCl2 and 1.0 mM Spermine. D35A1-3 Trigonal +5° 17° 3.25 Coiled-coil, crystal frozen four months after drop D35A1 was set up. Thirty duplexes per unit cell. 0.5 mM DNA, 50 mM NaCac pH6.5, 10 mM MgCl2, 1 mM CoCl2 and 1.0 mM Spermine. D35A1-4 - +15° 12° 3.25 Coiled-coil, crystal frozen four months after drop D35A1 was set up. 0.5 mM DNA, 50 mM NaCac pH6.5, 10 mM MgCl2, 1 mM CoCl2 and 1.0 mM Spermine. D35A5-1 D35A5-2 - +5° 25° 3.25 Coiled-coil, diffuse diffraction, only stacking visible. 0.5 mM DNA, 50 mM NaCac pH6.5, 10 mM MgCl2, 1 mM CoCl2, 1.0 mM Spermine and 16.7 mM Thymidine. 132 II.7.2 d(CGATATAT) d[CG(AT)3] and d(ATATATCG) d[(AT)3CG] d[CG(AT)3] The sequence CG(AT)3 has been crystallized using the hanging drop vapor diffusion technique and MPD as a precipitant; the crystals have been obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 1 mM Netropsine, 10 mM MgCl2 and 1 mM Spermine Tetrahydrochloride. Typical diffraction patterns are shown in Figure II.7.4. The crystal structure of the sequence CG(AT)3 is isomorphous with the structure of the sequence CG(TA)3. The unit cell parameters (manually determined) correspond to a trigonal unit cell with a and b parameters of about 46 Å and the c parameter of about 104 Å. The c value corresponds to a unit cell with stacks of four octamers along the c direction. The sequence CG(AT)3 generates crystals of straight pseudo-continuous coils (i.e. ). Figure II.7.4. Oscillation patterns (5°) of d(CGATATAT) crystals. The unit cell parameters, in a trigonal or pseudohexagonal space group, have been manually determined and are as follows: a and b about 46 Å and c about 104 Å. In (b) the Bragg spots with Miller indices 004 and 008 are shown. (c) Bragg spots with Miller indices 10l are evident. 133 d[(AT)3CG] The sequence (AT)3CG has been crystallized using the hanging drop vapor diffusion technique and MPD as a precipitant; the crystals have been obtained in the following conditions: 0.5 mM DNA duplex, 50 mM NaCacodylate pH 6.5, 50 mM KCl, 1 mM CoCl2 and 1 mM Spermine Tetrahydrochloride. Typical diffraction patterns obtained from crystals of d(ATATATCG) are shown in Figure II.7.5. Despite the poor diffraction, the unit cell has been manually determined; it corresponds to a trigonal or pseudohexagonal cell with a and b parameters of about 46 Å and c parameter apparently of about 130 Å. The c parameter corresponds to a unit cell with a stack of five duplexes along the c direction. The overall rotation Ω of one duplex with respect to its neighbor in a column is thus 288° ( , with and ). The average twist value per base pair is thus 36° ( ); the latter value is in agreement with those previously calculated for standard B-form DNA (Gorin et al., 1995). The typical DNA streaks are not clearly visible, in particular the second layer line is extremely weak, for this reason it can be speculated on a possible discontinuity that the change in hydrogen bonds type (from Watson-Crick for the d(CG) to Hoogsteen for the AT part) can create. The characteristics of the octamers studied in this work are summarized in Table II.14. Figure II.7.5. Oscillation patterns (15°) obtained from crystals of d(ATATATCG). The stacking reflections are found at about 3.25 Å resolution. Table II.14. Characteristics of the octamers studied in this work. Sequence Unit cell (Å) β Rise (Å) l N Resolution Comments Crystallization conditions CG(TA)3 ≈5° 3.25 26.13 4 6.5 Probably standard Bform. 0.5 mM DNA, 20 mM NaCacodylate pH 6.5, 20 mM KCl, 16.7 mM Thymidine, 1 μl 4-aminophenilsulfon, 1 mM Spermine tetrahydrochloride and MPD 20%. CG(AT)3 0° 3.24 26.9 4 16 Mainly streaks; isomorphous with CG(TA)3. 0.5 mM DNA, 50 mM NaCacodylate pH 6.5, 10 mM MgCl2, 1 mM Netropsine, 1 mM Spermine tetrachloride and MPD 32%. (AT)3CG ≈5° 3.23 26 5 16 Probably Hoogsteen. 0.5 mM DNA, 50 mM NaCacodylate pH 6.5, 1 mM CoCl2, 50 mM KCl , 1 mM Spermine tetrachloride and MPD 30%. 135 II.8 HEXAMER II.8.1 d(CGATAT), preliminary considerations The sequence CG(AT)2 has been crystallized using the hanging drop vapor diffusion technique and PEG 4000 as a precipitant. The hanging drops were prepared by mixing 1 μl of the DNA stock solution for a final concentration of 0.3 mM with 1 μl of 31 conditions of the Hampton Research crystallization screen for DNA, Natrix (HR2116). A birefringent precipitate was obtained in the following conditions: 80 mM Mg Acetate, 50 mM NaCacodylate pH 6.5, 30% PEG 4000. Drops with increasing Spermine concentration (0 mM to 6 mM) were set up. The hanging drops were incubated at 13 °C over 800 μl of a reservoir consisting of an aqueous solution of 30% PEG 4000. After a few weeks two-dimensional plates appeared in those drops with a Spermine concentration higher than 2 mM, while at lower concentration only precipitate or spherulites have been obtained. More than one month after the drops were set up, instead of the plates and the precipitate, needles started to appear: extremely thin needles at 2 mM Spermine concentration, clusters of big needles at higher Spermine concentrations (see Figure II.8.1). The needles had to be separated in order to obtain a single crystal suitable for the data collection. This crystal will from now on be referred to as “P52D3”. The Spermine/DNA ratio is much higher than for the other sequences studied in this thesis, for instance, in the CGAT5 case Spermine was three times more concentrated than DNA, while in the P52D3 case Spermine concentration is ten to seventeen times higher than DNA concentration. Besides, PEG has been used as a precipitant instead of MPD and the salt concentration (80 mM Mg Acetate) is relatively high. Figure II.8.1. Microscope photographs of crystals of CG(AT)2 obtained at Spermine different concentrations: (a, b) 0-1 mM Spermine; (c) 2-3 mM Spermine after one month the drop was set up; (d) 2 mM Spermine; (e) 4-6 mM Spermine. 136 The crystal was mounted in a nylon loop and flash frozen in liquid nitrogen. No extra cryoprotectant was used. Data collection was carried on at 100 K at beamline BM16, ESRF, Grenoble. 180° of oscillation data with a rotation of 2° per image were collected in a high resolution data set and another 180° of data in a low resolution pass. The images were integrated and scaled with the HKL package to a resolution of 2.6 Å. The crystal belongs to a hexagonal space group, with the following unit cell: and . Typical diffraction patterns are shown in Figure II.8.2, the stacking diffraction at about 3.2 Å is slightly off meridian indicating that the duplexes form a coiled-coil and their axes are slightly inclined with respect to the major axis. The curious shape of the stacking reflection might also indicate a peculiar arrangement of the junction. The data indicates that the crystal is formed by infinite parallel columns along the c dimension. Figure II.8.2. Oscillation patterns (2°) of crystal D52D3, the maximum resolution is about 2.6 Å. (b) Bragg reflections with Miller indices 009 and 0012 are shown. 137 II.9 SUMMARY AND CONCLUSIONS Sequences studied  All the sequences crystallized in this work are characterized by the presence of sticky ends, with the exception of d(AT)5T.  The overhanging sequences are either CG or GC.  The overhang is either at the 5’ or at the 3’ end of the sequence.  The central part of the sequences is represented by alternating fragments, (TA)n or (AT)n.  The following sequences have been studied: Overhanging sequence DNA sequences CG CG(TA)n n=3 CG(AT)n n=2, 3, 4, 5 (AT)nCG CGCG(AT)n n=3, 4, 5 n=4 GC GC(AT)n n=5 (AT)nGC n=5  The characteristics of the structures studied in this thesis are summarized in Table II.15. Conclusions 1. Practically all the sequences crystallized in this thesis are characterized by a high nucleation rate as well as a high crystal growth rate, with the exception of CG(AT)2 whose crystals grow within months. As a result the crystals have high mosaicity and often a fibrous structure. 2. The duplexes with sticky ends usually form infinite pseudo-continuous coiledcoils with staggered nicks. 3. Only the structure of the sequence CG(AT)5 could be determined at 3.1 Å resolution. It generates a right handed coiled coil with six duplexes per turn. The (AT)5 fragment is in the Hoogsteen conformation. The sticky end CG is assumed 138 to be in a standard Watson-Crick conformation, but that region appears disordered. The kink that gives rise to the coiled coil is attributed to the discontinuity of the phosphodiester chain. 4. All the sequences studied here pack with a hexagonal or pseudohexagonal symmetry. The geometry of the coiled-coils is determined by the angle θ between consecutive duplexes and the torsion angle τ. The latter is equivalent to the usual twist parameter ω. The majority of the sequences generate coiled-coils, with different β inclination and number N of residues per turn: N β Dodecamers 6; 6,75; 7; 14 11°; 20°; 25° Decamers 30-22-24 15°-25°-32° Octamers 4-5 0°-5° 5. Depending on the value of τ the coiled-coils may be either right handed, as in the case of CG(AT)5, or left handed, as in the case of CG(AT)4. 6. It is not clear whether the hydrogen bonding mode has an influence on the geometry of the coils or not, but it seems that a discontinuity in the hydrogen bonding is necessary for a coiled-coil to form. We suggest that the central alternating AT region forms Hoogsteen base pairs. However this conformation has only been firmly determined for d(ATATAT) and d(CGATATATATAT). 7. Special cases:  In the case of the sequence CGATATGCATAT the central CG base pairs force the flanking AT to form standard Watson-Crick hydrogen bonds, thus a standard B-form DNA is generated (see also Qiu et al., 1997).  In the case of (AT)5GC, the flipped out cytosine probably interacts with a neighbor AT base pair, the kink thus results less pronounced. The duplex appears to adopt a standard B-form.  The packing of the sequence CG(AT)4 varies depending on the precipitant concentration. Table II.15. Summary of the characteristics of the sequences crystallized in this thesis. The coiled-coils are probably Hoogsteen in most cases. In brackets are the numbers of oligonucleotides per turn. Sequence n=2 n=3 n=4 n=5 AT(AT)n Abrescia et al., 2002 Not studied Coiled-coil (work in progress) Campos et al., 2005 Coiled-coil CG(AT)n Continuous coil Coiled-coil Various types Coiled-coil CG(TA)n Not studied Coiled-coil Not studied Not studied GC(AT)n Not studied Not studied Not studied Coiled-coil (AT)nCG Not studied Coiled-coil I. B-form. Fiber diffraction II. Coiled-coil Coiled-coil (AT)nGC Not studied Not studied Not studied Flipped out cytosine. CGCG(AT)n Not studied Not studied Coiled-coil Not studied CG(AT)nGC(AT)n Not studied B-form Not studied Not studied 140 III. APPENDIX III.1 The influence of size on the thermal stability of oligonucleotides: the case of AT sequences. Notes & Tips The influence of size on the thermal stability of oligonucleotides: the case of AT sequences Daniela De Luchi, a Catherine Gouyette, b and Juan A. Subirana a,* a Department d’Enginyeria Quimica, ETSEIB, Universitat Polit eecnica de Catalunya, Av. Diagonal 647, Barcelona E-08028, Spain b Unit ee de Chimie Organique, Institut Pasteur, 28 rue du Dr. Roux, Paris 75724, France Received 23 May 2003 Since the early studies of Marmur and Doty [1], the melting temperature of DNA and oligonucleotides has been used for characterization. In our laboratory we are studying the structure of short AT-rich oligonucleotides by X-ray diffraction [2]. It is often necessary to know their thermal stability. Also we need to determine whether they show any sign of changes in conformation as a function of sequence and temperature. The theory of the melting transitions of oligonucleotides has been analyzed in detail by Marky and Breslauer [3]. However the theory is complex and it is not easy to use in a straightforward manner. In this paper we present a simple approach that allows a rapid determination of the influence of the various factors that determine the melting behavior of short oligonucleotides, in particular the influence of size. Materials and methods Oligonucleotides were synthesized on an automatic synthesizer by the phosphoramidite method and purified by gel filtration and reverse-phase HPLC. Samples were prepared for melting experiments by diluting appropriate aliquots from a concentrated stock solution of the oligonucleotides with 300 lL of the melting buffer (2 mM NaH2PO4, 6 mM Na2HPO4, pH 7.0, 1 M NaCl), for a final absorbance at 260 nm between 0.3 and 0.6 optical density in 1-mm-path-length cells at 20 °C. Thus the duplex concentration was about 50 lM for hexamers and 25 lM for dodecamers. Melting curves were obtained at 260-nm wavelength in a Varian Cary 100 spectrophotometer fitted with a thermoelectrically controlled sample holder. The heating rate was fixed at 0.5 °C min1and data were collected at 0.5 °C intervals up to 80 °C. Prior to the melting experiments, the samples were degasified and heated to 80 °C and then slowly annealed to the starting temperature. Water condensation on the cuvette exterior in the low-temperature region was avoided by flushing with a stream of dry nitrogen (4 L min1). However, due to the high humidity in Barcelona and the poor design of the commercial instrument chamber, it was often found imposible to work below 10 °C. The absorbance versus temperature curves were determined for sample and reference. Then the reference curve was subtracted from the sample curve. The resulting curves were smoothed and the maximum of the first derivative was taken to identify the melting temperature Tm. Results UV spectra The UV spectra were similar in all the samples that we have studied (results not shown), with absorption maxima in the range 260–263 nm depending on size and sequence. In alternating AT oligonucleotides we found upon denaturation a blueshift of the maximum in the range 0.6–1.5 nm. Nonalternating oligonucleotides did not show such a shift. No bimodal melting curves were observed. The values of Tmthat we have found are given in Table 1. Influence of concentration The process of oligonucleotide melting is an equilibrium process in which duplex structures and single * Corresponding author. Fax: +34-934010978. E-mail address: [email protected] (J.A. Subirana). 0003-2697/$ - see front matter Ó2003 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2003.08.008 Analytical Biochemistry 322 (2003) 279–282 ANALYTICAL BIOCHEMISTRY www.elsevier.com/locate/yabio