下载掌阅APP，畅读海量书库
立即打开
Double Helix at Atomic Resolution
J. M. Rosenberg et al .
Editor’s Note
By this time, the double-helix structure of DNA proposed by Francis Crick and James Watson was not in doubt. All the same, no one had carried out X-ray crystallography at sufficiently high resolution to see the positions of the atoms directly. That is what Alexander Rich, a postdoctoral student of Linus Pauling, and his colleagues report here—not for DNA itself, but for a double-helical form of a small fragment of RNA. This offered the first direct view of a Watson–Crick base pair, and confirmed its structure beautifully. Nadrian Seeman, one of Rich’s coauthors, went on to pioneer the use of DNA as a construction material for molecular nanotechnology. 中文
The sodium salt of the dinucleoside phosphate adenosyl–3′, 5′–uridine phosphate crystallizes in the form of a right-handed antiparallel double helix with Watson–Crick hydrogen bonding between uracil and adenine. A sodium ion is located in the minor groove of the helix complexed to both uracil rings. 中文
MANY important functions in molecular biology are determined by antiparallel double helical nucleic acids with complementary base pairing between the two strands, as first described by Watson and Crick 1 . Although considerable work has been carried out on the structures of double helical nucleic acids 2-8 , the fine details of their molecular architecture have never been available at atomic resolution. The X-ray diffraction patterns shown by fibres of DNA and RNA characteristically die out at resolution greater than about 3 Å. Because of this, there has been considerable discussion 9-13 about the actual nature of the detailed stereochemistry and hydrogen bonding in double helical DNA. One way to obtain additional information about these atomic details is to crystallize oligonucleotide fragments of known sequence which may form double helices in the crystal lattice. Here we report the single crystal X-ray analysis of the sodium salt of adenosyl–3′, 5′–uridine phosphate (ApU) which forms a right-handed antiparallel double helix in which the ribose phosphate backbones are held together by Watson–Crick hydrogen bonding between the adenine and uracil residues. This is the first crystal structure in which the atomic details of double helical nucleic acids can be visualized. In addition, this is the first single crystal structure showing Watson–Crick base pairing between adenine and uracil. 中文
The sodium salt of ApU (Miles) was mixed stoichiometrically with bromotertiary butyl amine in a 40% solution of 2–methyl–2,4–pentanediol (MPD). A small volume of the solution was placed in equilibrium with a large reservoir of 60% MPD in a closed container which was stored at 4℃. After standing for 2 weeks, small prismatic crystals with well-defined faces began to appear. Crystal growth continued slowly for several months and yielded crystals suitable for X-ray analysis. The amine was put into the solution in the hope that it might become the cation of the structure; however, subsequent analysis revealed that the sodium salt had crystallized rather than the bromoamine. A crystal measuring 0.2 mm×0.15 mm×0.05 mm was mounted on the tip of a glass fibre for X-ray analysis. The crystal was found to be monoclinic, space group P2 1 , with cell dimensions a =18.025 Å, b =17.501 Å, c =9.677 Å, β =99.45°. The crystal density measured in a density gradient was 1.53. In order to obtain a calculated density near the observed density, it was necessary to assume four molecules of Na + ApU – and twenty-two water molecules in the unit cell. This surprisingly high degree of hydration was subsequently shown to be a low estimate as solution of the structure revealed twenty-four water molecules in the unit cell. Three-dimensional X-ray diffraction intensity data were collected out to a resolution of 0.8 Å, with a “Picker” FACS-1 diffractometer in an ‘Omega’ step scan mode, using Nickel filtered CuKα radiation. The data were collected at 8℃ and were corrected for Lorentz and polarization effects; no absorption correction has been applied because of the low mass absorption coefficient (18.0). 中文
Although there was no crystallographic two-fold axis in the lattice, the presence of a peak nearly 40% the height of the origin on the Harker section of the Patterson function suggested a non-crystallographic two-fold axis in the structure. From this we inferred the double helical nature of ApU. This peak was assumed to contain all vectors from each atom of one ApU molecule to the corresponding atom of the other independent molecule, symmetry related. From previous work with the protonated dinucleoside phosphate uridyl–3′,5′–adenosine phosphate 14 (UpA), it was known that resolution difference Patterson techniques were effective in discriminating vectors arising from two second-row atoms in a large structure. (Resolution difference Patterson techniques are those in which two Patterson or superposition functions, for example, multiple minimum functions or N -atom symmetry minimum functions, are compared. These two Pattersons are (1) the standard F 2 Patterson, calculated using all the diffraction data, and (2) a Patterson calculated only from the higher-order reflexions whose F s should contain a larger contribution from the heavier atoms sought as they are relatively denser near the atomic centres. The origins of both Pattersons are normalized to the same value. Peaks arising from heavy atom–heavy atom vectors ought to be relatively more prominent than overlapping light atom–light atom vectors in the second map. In this work, the first Patterson contained all the 1 Å data, and the second function contained the data in the shell between 1.5 Å and 1 Å. 中文
UpA contained seventy-seven first-row atoms. Thus, we hoped to locate the phosphorus atoms by using this vector as the basis vector of a Resolution Difference 2-Atom Symmetry Minimum Function 15,16 (RDSMF(2)). The RDSMF(2) initially indicated the wrong location for the phosphorus atoms, however, which was obvious when Fourier refinement procedures failed to reveal the structure. An ( E 2 –1) Patterson ( E is the quasinormalized structure factor) was therefore calculated and the map and its Fourier coefficients were corrected for non-negativity and minimal bond lengths according to the procedure proposed by Karle and Hauptman 17 : (1) all negative points in the Patterson map and all points within a 0.9 Å radius of the origin were zeroed; (2) the map was Fourier transformed and all resultant ( E 2 –1) coefficients less than –1 were raised to –1; (3) a new Patterson map was calculated from these coefficients, and the process was iterated until convergence was obtained at forty cycles. From the final amplitudes, a set of thermally sharpened F s were generated, and used to calculate Patterson functions. The RDSMF(2) calculated from these Pattersons revealed the proper phosphorus locations. Approximate phosphate orientations were derived from Patterson superpositions. Fourteen cycles of Fourier refinement using the corrected E s as Fourier amplitudes revealed the two molecules of ApU, the sodium ions and four water molecules. The other water molecules were located in a series of difference syntheses. The structure has been refined using isotropic thermal parameters and only the 1 Å observed data by full matrix least squares. The current R factor is 0.091. (The discrepancy factor, R , is defined as R =∑│| F o |–| F c |│/∑| F o |, where | F o | and | F c | are the amplitudes of the observed and calculated structure factors, respectively.) Because this is one of the larger non-centrosymmetric biological crystal structures solved without isomorphous replacement, we will expand on the details of the solution in a later publication. 中文
In the analysis of this structure, there are two striking features. The non-crystallographic pseudo two-fold axis mentioned above rotates one ApU molecule into the other, so that the structure forms a segment of a right-handed antiparallel double helix in which the bases are hydrogen bonded to each other in the Watson–Crick manner, as shown in Figs. 1 and 2. This is the familiar hydrogen bonding between adenine and uracil which is believed to occur quite generally in double helical nucleic acids. Nonetheless, this type of hydrogen bonding had never been seen previously in single crystal X-ray analysis. It should be noted that the two-fold axis to which we are referring lies half-way between the base planes, rather than in those planes. Axes in the planes are the ones usually discussed in the literature, but the periodic nature of double helices generates symmetry elements at both locations. 中文
Fig. 1. View of the crystal structure perpendicular to the base planes showing hydrogen bonding (dashed lines) as well as the base stacking interactions. The darkest portions of the figure are those nearest the viewer. The rotational relationship is easily seen by comparing the front (black) and rear (white) glycosidic bonds.
The other prominent feature was that the crystal was heavily hydrated, so that the ApU molecules are surrounded by large numbers of water molecules. This observation, coupled with the non-crystallographic nature of the two-fold axis, leads us to believe that the double helical nature of ApU is a function of the molecules themselves, rather than crystal packing forces. 中文
The separation between the base pairs is 3.4 Å and, as can be seen in Fig. 1, there is a considerable degree of stacking in the structure. The NH...O hydrogen bonds between the N6 amino group of adenine and the O4 carbonyl oxygen of uracil (see Fig. 3 for the numbering scheme) have bond lengths of 2.95 Å and 2.91 Å. The NH...N hydrogen bonds between uracil N3 and adenine N1 have lengths of 2.82 Å and 2.86 Å. The important distances across the base pairs between the two ribose carbon atoms C1′ of the glycosidic linkage, are 10.50 Å and 10.53 Å. Similar distances to these have been obtained by analysis of single crystals of intermolecular complexes of purine and pyrimidine derivatives 18 but this is the first time that these distances have been observed in a molecule which forms a fragment of a double helix. 中文
Fig. 2. Comparison of the structures of ApU (upper) and RNA 11 (lower). This view is approximately perpendicular to the helix axis which is indicated by the vertical dashed line.
Fig. 3. Perspective view of the structure as seen from the minor groove showing the numbering scheme and the coordination of the sodium ions. Both sodium ions (●) have distorted octahedral coordination involving water molecules and other oxygens. The central sodium ion is 2.36 Å from the uracil O2 atoms.
The helical form of the molecules can be seen in Fig. 1 in which one is looking in a direction perpendicular to the base pair plane. There is clearly a rotational relationship between the glycosidic bonds (C1′ to adenine N9 or C1′ to uracil N1) of the front and rear base pairs. Another view of the structure parallel to the stacked hydrogen bonded bases is shown in Fig. 2. The right-handed helical rotation can readily be seen by observing the orientations of adjoining ribose residues on the ribose phosphate chain. This structure is similar to that which has been deduced from studying double-stranded viral RNA. The naturally occurring material is called RNA 11. Fig. 2 shows similar views of ApU and RNA 11, where the dashed vertical line represents the approximate helix axis. The continued extension of the ApU nucleotide pairs would generate a right-handed double-stranded helix similar to RNA 11. It is important to note that although the helical parameters of RNA 11 can be derived directly from an X-ray analysis of the fibre, the positions of the atomic centres are of necessity inferred from data derived from single crystal studies of small molecules rather than from direct observation. 中文
Although the two molecules of ApU in the asymmetric unit are very similar, they are not quite identical. The bond lengths and angles are all within expected range. The conformations of the ribose residues are all 3′-endo 19 , and the nucleosides are orientated in the anticonformation 20 , with torsion angles about the glycosidic bonds 19 of 2° and 7° for the adenosine residues, and 29° and 30° for the uridines. Comparison with the protonated UpA structure 21,22 suggests that discrepancies between adenine and uracil torsion angles are caused by the differences between the 3′ and 5′ ends of the molecule as the same pattern was obtained with respect to the 3′ and 5′ ends in that molecule which had the opposite base sequence. 中文
In the protonated structure UpA, it was noted that there were short distances between adenosine C8 and O5′, as well as uridine C6 and O5′ (ref. 14). At that time, it was suggested that this might be a result of an intramolecular attraction which, if of a general nature, would help stabilize the anticonformation of the nucleotide. Accordingly, the related distances in this structure were examined with great interest. Three of the four independent distances were greater than or equal to 3.3 Å; thus, there was no evidence of any interaction. But the distance between one adenine C8 and its ribose O5′ was 3.07 Å. As this O5′ is not covalently linked to a phosphorus atom, the phenomenon may not be relevant to the nucleotides in a helical polynucleotide chain. Nevertheless, for non-helical and 3′ terminal nucleic acid structures, this interaction may yet be seen to play an important role. 中文
The individual double helical fragments are separated in the crystal lattice by a spacing of 3.4 Å, with considerable overlapping of the adenine residues. Besides this intermolecular base stacking, the other important interaction in the direction perpendicular to the base pairs involves the proximity of O1′ of both adenosine riboses with uracil rings. A similar type of interaction was found to be important in the crystal structure of UpA 21,22 . 中文
The crystal structure can be visualized as a rod-like entity. The central portions of these rods contain the stacked base pairs. These are flanked by the ribose phosphate backbones, both of which in turn are surrounded by the solvent structure. The large amount of water allows the crystal structure to assume a conformation which is minimally perturbed by lattice interactions. Thus, the double helical structure which we observe here bears a marked resemblance to those structures proposed for solvated polynucleotides. It should be noted in passing, however, that the water is held rather firmly, as the intensity data were collected from the crystal while it was mounted in air. 中文
The ribose phosphate chains are packed together with the phosphate groups facing each other. In the layer between the ribose phosphate moieties are sodium ions which are found in two distinct sites: one, complexed between phosphate groups, and the other, surprisingly, bound to the uracil residues. The sodium ions and their ligands are shown in Fig. 3. The sodium ion in the centre is located on the pseudo two-fold axis in the minor groove of the helix with octahedral coordination which includes the free O2 atoms of the uracils; the remaining ligands are water molecules. As shown in Fig. 3, the other sodium ion rests on an intermolecular pseudo-dyad axis, also exhibiting octahedral coordination, which includes two adenosyl O3′ atoms and two phosphate oxygen atoms. 中文
The stability of the sodium ion position between two phosphate groups is readily understood on electrostatic grounds. What is not so obvious is why the second sodium ion is complexed to the two uracils in the minor groove of the helix. Five angstroms away from this site there is another position between two phosphate groups which could comfortably accommodate the sodium ion and is occupied only by solvent. It is possible that the structure is stabilized considerably by the sodium ion in the minor groove site. Should this prove to be of more general occurrence in other crystals containing A–U sequences, it is possible that this type of coordination may be of importance in polynucleotide structure and function. 中文
As noted above, one of the most significant features of the organization of the crystal structure of ApU is that the two independent molecules in the asymmetric unit have very similar but not quite identical conformations. Thus, they are related by a pseudo two-fold rotation axis while in an antiparallel double helical polynucleotide there is presumed to be a true two-fold axis which relates the backbones of the two antiparallel chains. We do not completely understand why the crystal structure of ApU did not adopt a true two-fold axis. In this regard, however, we recently discovered that the closely related dinucleoside phosphate, guanosyl-3′, 5′-cytidyl phosphate does indeed form a right-handed antiparallel double helix in the crystalline state in which this two-fold axis is crystallographic (unpublished results of R. O. Day, N. C. Seeman, J. M. R., and A. R.). This recent observation lends support to our earlier emphasis on the importance of the pseudo two-fold axis in the present crystal structure. 中文
Among the significant facts which we learn by analysing double helical fragments are the important detailed parameters determining the conformation of the polynucleotide chain, especially those parts dealing with the geometry of the phosphate group. The structure of the ribose phosphate chain is of central importance in understanding the physical properties and behavior of polynucleotide chains 23 . This information should prove valuable in interpreting the details of the molecular structure in polynucleotide double helices, as well as in the more complex forms of RNA such as those observed in tRNA 24 . 中文
This research was supported by grants from the National Institutes of Health, the National Science Foundation and the American Cancer Society. N. C. S. is a postdoctorate fellow of the Damon Runyon Foundation, J. M. R. is a predoctoral trainee of the National Institutes of Health, F. L. S. is a postdoctoral fellow of the American Cancer Society, H. B. N. was an NIH postdoctoral fellow. 中文
We thank Bob Rosenstein, Roberta Ogilvie Day, Don Hatfield, Sung-Hou Kim and Gary Quigley for useful discussions and encouragement, and John Genova and Tim O’Meara for technical assistance. 中文
( 243 , 150-154; 1973)
John M. Rosenberg, Nadrian C. Seeman, Jung Ja Park Kim, F. L. Suddath, Hugh B. Nicholas and Alexander Rich
Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
Received January 3, 1973.
References: