Guanine, one of the four DNA bases, has been observed by tunneling microscopy to form a two-dimensional ordered structure on two crystalline substrates, ...
Proc. Nati. Acad. Sci. USA
Vol. 88, pp. 8003-8005, September 1991
Biophysics
Two-dimensional ordering of the DNA base guanine observed by scanning tunneling microscopy W. M. HECKL*t, D. P. E. SMITHf, G. BINNIG*, H. KLAGGESt, T. W. HANSCH*, AND J. MADDOCKS§ *Universitat MOnchen, Sektion Physik, and *IBM Physics Group, Schellingstrasse 4, 8000 Mtnchen 40, Federal Republic of Germany; and §University of
Sheffield, Department of Therapeutics, Sheffield S10 2JF, United Kingdom
Contributed by G. Binnig, June 10, 1991
ABSTRACT Guanine, one ofthe four DNA bases, has been observed by tunneling microscopy to form a two-dimensional ordered structure on two crystalline substrates, graphite and MoS2. The two-dimensional lattice formed by guanine is nearly identical on the two surfaces, and heteroepitaxy appears to be the growth mechanism in both cases. Although the resolution of molecular details is superior for the graphite substrate, the simpler results on MoS2 are not only easier to interpret but also facilitate the understanding of the more complex images on graphite. We propose that the interfacial structure is composed of linear chains of hydrogen-bonded molecules aligned into a closely packed two-dimensional array.
Several recent studies have examined the possibility of imaging DNA by scanning tunneling microscopy (STM) (1-3). However, the ultimate goal-to read the code contained in the strands-has not yet been achieved. To do so it must be possible to recognize clearly and distinguish between the four bases of the genetic code. One essential requirement for imaging small organic molecules by STM is to find experimental preparation conditions whereby the molecules stick firmly to the substrate and form a highly stable layer. This is necessary to withstand the forces of the STM tip during imaging. Compared to the binding of complete DNA strands to the basal planes of MoS2 or graphite, the adsorption of "naked" nucleotide bases is favored by their greater hydrophobicity and, as is shown in this report, by their ability to register with the substrate, thus forming a stable twodimensional ordered array. No signs of mobility were observed during investigations, which lasted up to 1 h in a single region. That such an array can form is probably due to the intermolecular hydrogen bonding capability of the DNA bases (4). Ordered molecular layers of organic molecules, such as benzene (5), alkanes (6), or liquid crystals (7, 8), are the most prominent examples to have been imaged recently by STM at atomic resolution. STM results of adenine, one of the four DNA bases, have recently been obtained on graphite by Allen et al. (9). Using a sample preparation technique similar to theirs (9), we prepared samples of guanine on the surfaces of natural MoS2 crystals and highly oriented pyrolytic graphite. We often observed steps from the bare substrate to monolayers of guanine. The results presented in this paper were obtained on such monolayer islands. The character of the steplines provides additional information since the orientation of the molecular lattice with respect to the substrate can be measured there. In the STM images the bases look quite different depending on whether they are deposited on MoS2 (Fig. 1) or on graphite (see Fig. 3). The main differences are that on MoS2 the bases appear as distinct well-isolated nearly structureless blobs; on graphite more details become visible, although it is still not possible to determine the The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
FIG. 1. (A) STM image of the MoS2 substrate showing the uppermost sulfur atoms, 3.16 A apart. The linear dimensions in B are twice those in A.
(B) STM image of guanine ad-
C
-
sorbed to the surface of MoS2 showing an ordered array of guanine molecules forming ,,parallel rows with alternating molecular directions. The mol-I ecules in a row form a zig-zag with an angle of 95°and with an average center-to-center molecular distance of 7.7 A. (C) Magnified region of B with superimposed model and the unit cell with dimensions a = 10.9 Aand b = 19 A. The STM images were taken in the constant current mode with a tip voltage Ut = -0.7 V and a tunnel current It = 800 pA for A and Ut = -1.3 V and It = 50 pA for B and C. The sample was prepared by applying a microliter of guanine (2-amino-6-hydroxypurine) in aqueous solution (0.01 mg/ml) to the surface of MoS2 heated to =120TC. This temperature provided enough thermal energy to allow the spreading and diffusion of the guanine molecules. We believe that this "sizzling" technique is also responsible for providing enough thermal energy [> 10 kcal/mol (10)] to get rid of most of the water molecules.
location of individual atoms. The results on MoS2 are of crucial importance as they allow an unambiguous interpretation of the STM images and enhance our understanding of the results on graphite. Fig. 1B shows an ordered array of guanine molecules adsorbed to the MoS2 surface. The molecules form parallel rows with alternating directions of the long axis of the Abbreviation: STM, scanning tunneling microscopy. tTo whom reprint requests should be addressed.
8003
8004
Biophysics: HeckI et al.
Proc. Natl. Acad. Sci. USA 88 (1991)
FIG. 3. (A) STM image of rC the graphite substrate showing the graphite atoms, 2.46 A apart. The linear dimensions in B and C are half those in A. (B) to STM image of guanine adsorbed on graphite. (C) Same image with superimposed model where the loops approximate the outline of the molecule. Black lines indicate H-O bridge bonds and white dots_ indicate H-N bridge bonds.NW The STM images were taken in the constant-current mode with Ut = -0.8 V and It = 700 pA for A and -1.4 V and 50 pA for B and C. As for MoS2, the molecular lattice is such that every fourth molecule within a chain is located at an equivalent position with respect to the graphite. The placement of the model on top ofthe data is constrained by the assumption that mirror symmetry exists between neighboring rows. Using the substrate vectors g, = [100] and g2 = [110], which are 2.46 A long, we find the molecular unit cell vectors a = 8g, and b = -5g, + 10g2, giving the unit cell an area of 80 times the substrate unit cell area. A
FIG. 2. Model of guanine on MoS2 showing the arrangement deduced from STM images. The number of molecules per chain length is in excellent agreement with the measurements. The experimentally measured periodicity within a chain is 10.9 A compared to 10.8 A in the model. This periodicity explains the orientation of the molecular lattice with respect to the substrate. Seven sulfur lattice spacings of 22.1 A correspond within 2% to two zig-zags of the chain. Hence every fourth molecule of a chain is again in an equivalent position with respect to the MoS2 lattice. From the experimental results, we derive that the distance to the next neighboring row is 19 A (corresponding to unit cell vector b) and that the next neighboring rows are phase-shifted in the direction of the main axis of the chains by half a lattice constant with respect to each other. The unit vectors ofthe guanine lattice (a and b) can be written in terms ofthe substrate unit vectors s1 = [100] and s2 = [010], where s1 and s2 are 3.16 A long. We find a = 3.5s1 and similarly b = -3s, + 7s2. Every even multiple of b lies on a commensurate position. The smallest commensurate vector a would then be 7s, giving the unit cell an area of 49 times the substrate unit cell area.
molecule. The row-like character of the array is most clearly visible at the edges of the monolayer islands as some of the molecular rows extend somewhat further across the stepline than others. The molecules within one row form a zig-zag with an angle of 95 ± 20 and with an average center-to-center molecular distance of 7.7 ± 0.2 A. By reducing the gap resistance below about 109 fQ, we let the tip pierce the guanine layer such that the atomic structure of the underlying MoS2 substrate became visible (Fig. LA). By this method (7) the orientation of the monolayer with respect to the substrate lattice was measured, yielding the same results as the observations around the edges ofthe islands. The main axis (nearly vertical in Fig. 1B) of the molecular rows is parallel to the [100] direction of the MoS2 (Fig. 1A). The unit cell of four molecules is indicated in Fig. 1C with unit cell vectors a = 10.9 A, a being parallel to the [100] direction of MoS2, and b =
19 A. By connecting the molecules in the direction perpendicular to the rows, another zig-zag structure is visible, as indicated in Fig. 1C. We have assumed that, independent of its registry with the substrate, the location of the molecule on MoS2 is represented by the location of the blobs in the measurement. We will show later that this assumption cannot be made for graphite, where the coupling of the molecules to the substrate more strongly modulates the electronic structure of the molecules. With the above-mentioned assumption, we represent (in Fig. 1C) the molecular chains by zig-zags where the dots mark equivalent sites of the molecules. Fig. 2 shows a model, deduced from STM images, of guanine molecules on the uppermost sulfur layer of MoS2. The model we propose is the most closely packed array possible with a unit cell of four molecules. The unit cell dimensions that we measure by STM correspond, within experimental error, to the commensurate lattice shown in the model. Besides registry of the molecules with the substrate lattice, intermolecular hydrogen bridge bonds are also likely responsible for the observed stability of the array. We propose that hydrogen bonding determines the size and shape of the unit cell, and that registry with the substrate determines its orientation. Double hydrogen bridge bonds to oxygen alternate with those to nitrogen, forming a zig-zag row of equidistant molecules. The molecules point in alternating directions and can in principle form an infinitely long chain (Fig. 2). The hydrogen bonds have a calculated binding enthalpy of 8 kcal/mol for oxygen and about 3 kcal/mol for nitrogen at the 6' ring (1 cal = 4.184 J). The binding enthalpy values were calculated by the semiempirical molecular dynamics program DISCOVER (Biosym Technologies, San Diego) for two molecules brought together from infinity to the energy-minimized hydrogen bonding geometry. The experimentally observed bond lengths are 2 A, which is within the range of known hydrogen bond lengths for base pairs (4).
Biophysics: HeckI et al. The interpretation of the results on MoS2 is relatively easy. Although the molecules are closely packed, they are distinct from their neighbors. The hydrogen bridge bonds and the atomic structure of the molecules do not contribute significantly to the tunneling current. If the substrate is changed from MoS2 to graphite the situation is quite different (Fig. 3). STM images of guanine on graphite are much more detailed and complicated than those on MoS2. Moreover, the rows of guanine molecules are parallel to the [110] direction of the graphite substrate. We find, however, that the overall molecular structure is the same for both substrates. The unit cell dimensions are slightly different due to the different substrate lattice constants. In Fig. 3C we present a model for the location of the molecular lattice with respect to the measurements on graphite. The model is constructed based on the findings on MoS2. The outline of a guanine molecule is represented by a ring and the most prominent features of the measurements, the brightest spots in the image, are assigned to the oxygen-hydrogen bridge bonds. The images on graphite still remain quite difficult to interpret. It seems that the registry with the graphite substrate plays an essential role in how the molecules appear in the images. Molecules with a different registry to the graphite lattice appear differently in the STM image. This suggests that on graphite the coupling of molecular states to the substrate plays an important role and one cannot simply consider isolated molecular states. Furthermore, unlike on MoS2, it appears that the bonds between the molecules also contribute to the tunneling current. This shows that there is
Proc. Natl. Acad. Sci. USA 88 (1991)
8005
a need for improved numerical simulations to explain the STM contrast for these complex adsorbed systems. In this report we have demonstrated the potential of STM to visualize the detailed structure of isolated DNA bases on
different substrates, thus raising interesting questions about the possibility of sequencing DNA. To date we have also been able to distinguish clearly between isolated purine and pyrimidine bases in our STM images. These results and our attempts to decode entire DNA strands will be discussed elsewhere. 1. Arscott, P. G., Lee, G., Bloomfield, V. A. & Evans, D. F. (1989) Nature (London) 339, 484-486. 2. Dunlap, D. D. & Bustamante, C. (1989) Nature (London) 342, 204-206. 3. Driscoll, R. J., Youngquist, M. G. & Baldeschwieler, J. D. (1990) Nature (London) 346, 294-296. 4. Saenger, W. (1984) Principles of Nucleic Acid Structure (Springer, New York). 5. Ohtani, H., Wilson, R. J., Chiang, S. & Mate, C. M. (1988) Phys. Rev. Lett. 60, 2398-2401. 6. McGonigal, G. C., Bernhardt, R. H. & Thompson, D. J. (1990) Appl. Phys. Lett. 57, 28-30. 7. Smith, D. P. E., Horber, J. K. H., Binnig, G. & Nejoh, H. (1990) Nature (London) 344, 641-644. 8. Hara, M., Iwakabe, Y., Tochigi, K., Sasabe, H., Garito, A. F. & Yamada, A. (1990) Nature (London) 344, 228-230. 9. Allen, M. J., Balooch, M., Subbiah, S., Tench, R. J., Siekhaus, W. & Balhorn, R. (1991) Scanning Micros. 5, in press. 10. Marx, K. A. & Ruben, G. C. (1984) J. Biomol. Struct. Dyn. 1, 1109-1132.
