Simultaneous Topographic and Fluorescence Imaging of Single DNA

High-resolution fluorescence imaging of λ-phage DNA molecules, intercalated with the dye YOYO-1, has been performed by a SNOM/AFM based on a bent-typ...
1 downloads 0 Views 633KB Size
Anal. Chem. 2001, 73, 5984-5991

Simultaneous Topographic and Fluorescence Imaging of Single DNA Molecules for DNA Analysis with a Scanning Near-Field Optical/Atomic Force Microscope J. M. Kim,*,† T. Ohtani,† S. Sugiyama,† T. Hirose,‡ and H. Muramatsu§

Department of Food Engineering, National Food Research Institute, Kannondai, Tsukuba, Ibaraki 305-8642, Japan, Plant Resources Laboratory, Japan Atomic Energy Research Institute, Takasaki 370-1292, Japan, and R&D Center, Seiko Instruments Inc., 563 Takatsuka-shinden, Matsudo-shi, Chiba 270-2222, Japan

High-resolution fluorescence imaging of λ-phage DNA molecules, intercalated with the dye YOYO-1, has been performed by a SNOM/AFM based on a bent-type optical fiber probe. A modified design of the optical probe has been made, and successful near-field optical resolution has been obtained for the strongly stretched λ-phage DNA molecules. The best optical resolution was estimated at 45 nm for the dye-intercalated single λ-DNA molecules by a mean width evaluation. In our comparison between the far-field fluorescence and high-resolution near-field fluorescence images for the DNA, it has been found that the near-field images much better defined the intercalation state of the dye. Finally, the relation between the DNA shapes and the dye distribution states, and the discrimination between the double-stranded and single-stranded DNA molecules, are discussed by comparing the topography and fluorescence images of the SNOM/AFM. Because of its essential role in biology, DNA has been extensively studied by scanning probe microscopes (SPMs). In the atomic force microscope (AFM) studies, after a successful report for imaging DNA in 1992 by Bustamante et al.,1 numerous reports were published2 from single DNA imaging to DNAprotein interaction imaging. Recently, a few reports have also been published for DNA sequence studies done by direct physical mapping using the AFM.2,3 The scanning near-field optical microscope (SNOM)4 uses an aperture-formed probe to illuminate and scan a sample at a constant distance above the sample surface. * Current address: Seiko Instruments Inc., Chiba 270-2222, Japan. † National Food Research Institute. ‡ Japan Atomic Energy Research Institute. § Seiko Instruments Inc. (1) Bustamante, C.; Vesenka, J.; Tang, C. L.; Rees, W.; Gutfold, M.; Keller, R. Biochemistry 1992, 31, 22-26. (2) See some reviews for examples: (a) Lyubchenko, Y. L.; Jacobs, B. L.; Lindsay, S. M.; Stasiak, A. Scanning Microsc. 1995, 9 (3), 705-727. (b) Shao, Z.; Mou, J.; Czajkowsky, P. M.; Yang, J.; Yuan, J. J. Adv. Phys. 1996, 45 (1), 1-86. (3) Cherny, D. I.; Fourcade, A.; Svinarchuk, F.; Nielsen, P. E.; Malvy, C.; Delain, E. Biophys. J. 1998, 74, 1015-1023. (4) (a) Betzig, E.; Trautman, J. K. Science 1992, 262, 1422-1425. (b) Ambrose, W. P.; Goodwin, P. M.; Martin, J. C.; Keller, R. A. Phys. Rev. Lett. 1993, 72, 160-163. (c) Betzig, E.; Chichester, R. J. Science 1993, 262, 1422-1425. (d) Xie, X. S.; Dunn, R. C. Science 1994, 265, 361-364.

5984 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

The SNOM can simultaneously produce both force and optical images. This dual information means that the SNOM image can, in principle, better characterize the sample than the AFM alone. So far, however, the number of reports done by the SNOM for DNA imaging is small compared with that of the AFM. One of the main problems in the SNOM imaging is the resolution of the shear force image. The lateral resolution in the shear force controlled SNOM is inherently limited by the lateral dithering amplitude of the tip, and as the tip motion above the surface is driven by the shear force controller, it easier for the tip to crash into rougher samples. We have previously developed an imaging system called the scanning near-field optical/atomic force microscope (SNOM/ AFM).5 In this imaging system, a bent optical fiber is used as a cantilever and the distance control between the tip and a sample is based on a normal AFM method. Therefore, the SNOM/AFM can be operated by both contact mode (static force mode; AFM mode)6 and cyclic mode (dynamic force mode; DFM mode).7 As the DFM has the advantage of giving less damage to the tip and the sample surface than a normal contact-type AFM8 and giving better resolution than a shear force-type feedback, in principle, the resolution of the SNOM/AFM operated with the DFM mode is better than that of the shear force feedback microscope. Single DNA molecule imaging is interesting for examining the performance of the SNOM, and its application to genetic studies, because a DNA molecule has an absolute width of 2 nm. For the SNOM, the fluorescence imaging of single DNA molecules, intercalated with cyanine dyes such as YOYO-1 or YO-PRO-1, means the continuous optical imaging of the intercalated dye molecules. Some reports9 for the SNOM detection of single DNA molecules have been published with the optical resolution of 100 nm for natural or synthetic DNA. But none of these reports has (5) (a) Muramatsu, H.; Chiba, N.; Ataka, T.; Monobe, H.; Fujihira, M. Ultramicroscopy 1995, 57, 141-144. (b) Muramatsu, H.; Chiba, N.; Homma, K.; Nakajima, K.; Ohta, S.; Kusumi, A.; Fujihira, M. Appl. Phys. Lett. 1995, 66 (24), 3245-3247. (6) Muramatsu, H.; Chiba, N.; Fujihira, M. Appl. Phys. Lett. 1997, 71 (15), 2061-2063. (7) Tamiya, E.; Iwabuchi, S.; Nagatani, N.; Murakami, Y.; Sakaguchi, T.; Yokoyama, K.; Chiba, N.; Muramatsu, H. Anal. Chem. 1997, 69 (18), 36973701. (8) Martin, Y.; Williams, C. C.; Wickramasingge, J. Appl. Phys. 1987, 61, 47234726. 10.1021/ac010536i CCC: $20.00

© 2001 American Chemical Society Published on Web 11/10/2001

shown simultaneous topographic and optical images for natural DNA molecules. Because these reports have been focused either on small or oligonucleotide molecules, a more practical approach for a long DNA molecules has until now not been archived. Our previous report10 has shown simultaneous topographic and fluorescence images of aggregated λ-DNA done by SNOM/AFM though its resolution is not good. On the other hand, single DNA molecule imaging is a very fundamental requirement for the real application of SNOM/AFM on DNA researches. In this report, we visualize a YOYO-1-intercalated single λ-DNA molecule and demonstrate its intercalation state using highresolution near-field fluorescence imaging. The maximum fluorescence resolution for a λ-DNA molecule obtained has been about 45 nm, which exceeds the optical resolution of a previous report9a for the imaging of the oligonucleotide molecule. We also discuss the future possibilities of genetic studies by the discrimination between the single- and double-stranded DNA (dsDNA) molecules using the SNOM/AFM. EXPERIMENTAL SECTION DNA Solution. Double-stranded λ-phage DNA (48.5 kbp, 570 µg/mL) was obtained from Wako Pure Chemicals Inc., supplied in 10 mM Tris, 1 mM EDTA. The DNA solution was diluted to a final concentration of 5.7 × 10-3 ng/mL with N2 saturation and 1 mM DMSO included TE (pH 8) buffer solution. To visualize the DNA, we prestained it with dimeric cyanine dye YOYO-1 (Molecular Probes, absorption 491 nm, emission 509 nm), at a ratio of 1 dye/5 base pairs. The staining procedure is the one proposed by the manufacturer. Substrate and Spin Stretching. Cover glass was obtained from Matsunami Glass Ind., Ltd. (thickness 0.12-0.17 mm, width 24 × 50 mm), washed with acetone and double-distilled deionized water (DDW). Mica sheets of 0.01-mm thickness were purchased from HKK Inc., washed with acetone, and attached on the cover glass with an adhesive bond. After attaching and before the experiment, the mica was freshly cleaved and fixed on a steel ring. To obtain a topographic image and to stretch λ-DNA molecules, 3-(aminopropyl)triethoxysilane-coated mica (AP-mica) was used. The fabrication procedure of the AP-mica was followed from previous report11 except for the reaction time. We obtained a flatter surface by decreasing the reaction time from 2 h to 40 min. To stretch the DNA, we applied a spin stretching method reported by Yokota et al.12 The general procedure was similar to the original report, but a spin speed of 4500 rpm was applied to a 5-µL portion of DNA solution. This spin speed is slower than that of the original report. In our experiments, we have found that spin speeds greater than 5500 rpm produce partly broken DNA molecules, which are only confirmable in high-resolution surface images. In the far(9) (a) Deckert, V.; Zeisel, D.; Zenobi, R.; Vo-Dinh, T.; Anal. Chem. 1998, 70, 2646-2650. (b) Carcia-Parajo, M. F.; Veerman, J. A.; Ruiter, A. G. T.; van Hurst, N. F. Ultramicroscopy 1998, 71, 311-319. (10) (a) Muramatsu, H.; Homma, K.; Yamamoto, N.; Wang, J.; Sakata-Sogawa, K.; Shimamoto, N. Mater. Sci. Eng. 2000, C12, 29-32. (b) Muramatsu, H.; Homma, K.; Yamamoto, N.; Wang, J.; Sakata-Sogawa, K.; Shimamoto, N. In Near-Field Optics: Principles and Applications (The Second Asia-Pacific Workshop on Near Field Optics) Xing, Z., Motoichi, O., Eds.: World Scientific: Singapore, 1999; pp 67-72. (11) Bezanilla, M.; Manne, S.; Landy, D. E.; Lyubchenko, Y. L.; Hansma, H. G. Langmuir 1995, 11, 655-659. (12) Yokota, H.; SunWoo, J.; Sarikaya, M.; Engh, G.-v. d.; Aebersold, R. Anal. Chem. 1999, 71, 4418-4422.

Figure 1. Scanning electron microscope images of bent-type optical probe (a), cross section of the tip after treatment of focused ion beam (b), and optical transmission changes before (c) and after grinding (d) for 5 × 5 µm2. The grinding was performed for about 30 min using a transmission mode SNOM/AFM5a with a scan rate of 64 Hz, and with almost contact state. During the process, to obtain a good optical probe, the detection of direct signal changes is necessary and must stop if the optical transmission changes.

field fluorescence microscope, it was not clearly visible because of the far-field diffraction limit. Optical Probe Preparation. Panels a and b of Figure 1 show scanning electron microscope (SEM) images of bent-type optical fiber probes fabricated by a procedure of two-phase chemical etching, bending, aluminum layer deposition, and aperture formation for a single-mode optical fiber.6,13,14 To enhance the optical resolution, we modified the design of the optical fiber probe by two steps. First, the thickness of an aluminum layer was doubled, to an approximate thickness of 200 nm, to prevent the leakage of excitation light from the taper. In our observation, we found that the small leakage of laser light from the taper severely influences and decreases the optical resolution of DNA. Next, we ground the tip of the optical probe on chromium patterns to produce a flat tip end. The chromium part has dimensions of 2 µm × 2 µm × 20 nm on a quartz substrate and has a surface roughness of about a few nanometers. Detailed understanding of the chromium pattern can be obtained from previous reports.5b,13a This increasing of the aluminum layer thickness and grinding of the tip, however, may decrease topographical resolution due to the increase of the tip size. In the case of a DNA sample, the decrease of topographic resolution reaches about 3-4 nm. The optical probe sufficient to apply this process is dependent on the aperture size. Applying our method to the big aperture sized optical probes, in our experiment, possibly decreased both topographic and optical (13) (a) Chiba, N.; Muramatsu, H.; Ataka, T.; Fujihira, M. Jpn. J. Appl. Phys. 1995, 34, 321-324. (b) Muramatsu, H.; Homma, K.; Chiba, N.; Yamamoto, N.; Egawa, A. J. Microsc. 1999, 194 (Pt 2/3), 383-387. (14) Muramatsu, H.; Homma, K.; Chiba, N.; Yamamoto, N.; Egawa, A. J. Microsc. 1995, 61, 265-269.

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

5985

Figure 2. Schematic illustration of the SNOM/AFM imaging system for the fluorescence collection mode.

resolution of the probe. The reproducibility of this method has a strong dependence on the grinding process of the tip. Panels c and d of Figure 1 show a typical transmission change due to the grinding of the tip for the same area of Cr patterns. In Figure 1c, there is no bright area because the aperture is not opened, but in Figure 1d, the bright parts of the quartz substrate are revealed clearly, and that means the formation of an aperture during grinding. In our experiment, we have an 80% subhundred nanometer aperture-sized optical probes and a 30% success rate in producing less than 50-nm aperture-sized optical probes. SNOM/AFM System. The schematic illustration of the SNOM/AFM for the fluorescence collection mode is shown in Figure 2. The SNOM/AFM instrument is based on a conventional AFM unit (SPI 3800, Seiko Instruments Inc), which is capable of DFM function. A sharpened and bent optical fiber probe is mounted on a bimorph. The distance between tip and sample is controlled by a laser beam-deflecting AFM technique in which the cantilever vibrates only vertically at a resonant frequency. The amplitude of the cantilever vibration decreases with decrease of the tip-sample distance. This results in less influence of the vibration amplitude on the optical near-field and a stable source of excitation light. Light of 488 nm from a multiline Ar ion laser with the output power of ∼40 mW is coupled with the end of the optical probe and illuminated from the tip of the probe to a sample surface. After transmission through both the aperture of the optical fiber probe and the sample, the remaining light is detected by an avalanche photodiode (APD, EG&G, SPCM-AQR-16). With this configuration, the SNOM/AFM can provide topographic and optical images simultaneously with high resolution, though our design of the optical fiber probe may decrease the topographic resolution a little. RESULTS AND DISCUSSION Why We Stretch DNA. Figure 3 shows an example of a topography image (a) and a corresponding near-field fluorescence image (b) for the sample of not perfectly stretched λ-phage DNA on the surface of AP-mica. This sample was prepared by use of a relatively slow spin speed about 2000 rpm. As shown in Figure 3, 5986

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

the topographic image is not perfectly produced compared with a normal AFM image (data not shown) but DNA morphology is still discernible. In the case of the SNOM/AFM experiment, obtaining good topography and fluorescence images of DNA simultaneously is still a delicate technique. For an example, until now, there have been no reports for simultaneous topographic and near-field fluorescence images of a natural DNA. To obtain the single DNA topographic and fluorescence images simultaneously, a well-designed imaging system, a skilled operator, and a clean substrate are necessary. In the topographic image, the DNA is not perfectly stretched, and shows a few zigzags noted as (B) and (C). Also, in site D, the DNA shows an aggregation. In the case of the AFM study,2 the zigzag and the aggregation are not a problem because the sharp-pointed tip and the flexible cantilever of the AFM can successfully resolve the DNA. On the other hand, in the near-field fluorescence image, the aggregated site D, where each neighboring DNA is located at a smaller distance than the optical resolution, shows a big fluorescence intensity though its appearance is not directly understandable. One of the most interesting areas is the part around zigzags indexed as (B) and (C). In these parts, the local fluorescence intensities are strongly increasing, which is not detected in the strongly stretched area noted as (A). The radius of each zigzag is bigger than the optical resolution, i.e., the aperture size of the probe. And therefore, the higher optical intensity in these areas is not explainable by only the DNA structure-influenced fluorescence intensity increases. To understand these strong localized fluorescence signals, we have to think of a possible process. First, in our previous report,10 we have shown the heterogeneous intercalation of YOYO-1 into DNA when the ratio of dye to bp is 1:10. However, we used a dye/bp ratio of 1:5; this sample has also slightly random distribution. Also, we have to consider the external binding of YOYO-1, which occurs when the ratio of dye to bp is more than 1:8,15 combined with the internal intercalation. By the combination of intercalation and external binding, YOYO-1 may (15) Larsson, A.; Garlsson, C.; Jonsson, M.; Albinsson, B. J. Am. Chem. Soc. 1994, 116, 8459-8465.

Figure 3. Example of simultaneous topography (a) and near-field fluorescence images (b) for not perfectly stretched λ-DNA produced by the SNOM/AFM method. The sites B and C show big fluorescence intensity variation in the near-field fluorescence image. Scanned area is 5 × 5 µm2.

be more randomly distributed in the DNA than in our previous report. Second, the YOYO-1-intercalated DNA parts elongate more slowly than non-intercalated parts for the same applied force;16 i.e., the elongation of YOYO-1-intercalated DNA portions requires bigger force than non-intercalated DNA with the same time constant. The difference in the local stretching speed within a DNA molecule may influence the local stretching efficiency of the DNA. In this way, YOYO-1 seems to be localized more around zigzag sites, which results in increased fluorescence signals. To estimate the optical resolution of DNA, first, a line profile analysis was performed because it is a common method to evaluate optical resolution. In site A, the optical resolution reached about 40 nm but that of (B) and (C) reached about 100 nm at full width half-maximum (fwhm). For more easy understanding, an expanded view and a line profile analysis for the fluorescence image in Figure 3 is captured in Figure 4. The corresponding topography is indicated in Figure 3a as a recetangle. As shown in the line profile indicated as a-a′, the smaller fluorescence peak shows an optical resolution of about 40 nm, but the optical resolution of the bigger peak is almost double in height and width, which means the number of YOYO-1 around (a′) is bigger than that of around (a). As shown above, the direct line profile analysis in a fluorescence image of the DNA is not an adequate method to estimate the optical resolution of the DNA because the dye distribution is severely different with position. Therefore, mean width analyses have been performed in regions A, B, and C. The results showed 65, 55, and 100 nm for the regions A, B, and C, respectively. The difference in the measured values means that the difference by the selected area also exists. In principle, the mean width resolution may decrease the numerical value of the optical resolution and the difference due to the different concentration of fluorescent dye. However, it is a more credible value than that of the single line profile analysis for a single site. Also, the mean width resolution may decrease the numerical error of (16) Bennink, M. L.; Scha¨rer, O. D.; Lanaar, R.; Sakata-Sogawa, K.; Schins, J. M.; Johannes, S. K.; de Grooth, B. G.; Greve, J. Cytometry 1999, 36, 200208.

Figure 4. Expanded view and a line profile analysis for the nearfield fluorescence image of Figure 3b. Rectangular areas shown in Figure 3 correspondto this fluorescence image. Scanned area is 2.5 × 2.5 µm2.

optical resolution caused by the difference in scanning scales; i.e., the resolving power of one pixel is different in different scanning areas. In this paper, the mean width resolution of near-field Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

5987

Figure 5. Well-stretched single DNA for far-field fluorescence (a) and near-field fluorescence images (b, c) by the function of optical resolution. Optical resolution is 110 (b) and 70 nm (c) by mean area estimations. Image areas are approximately 15 × 15 (a) and 14 × 14 µm2 (b, c). The difference in optical resolution is mainly explainable by the difference om aperture size of the optical probes because every scan parameter is the same between (b) and (c) except the probe itself. (For a clearer understanding of (a), see the full color image available in Supporting Information.)

Figure 6. Expanded and contrast-controlled views of Figure 5 for optical resolution of 110 (a) and 70 nm (b). When the distance between two dye clusters is shorter than the size of the optical probe, the fluorescence signal produced is similar to a single cluster, but the width and shape of the signal indicate the difference in dye concentration as shown in the site C. Scanned areas are 5 × 5 µm2. Each capital in the line profile analyses is the pair to the capitals in the topographic and near-field fluorescence images.

fluroscence image is estimated by integrating the fluorescence profile of a 1 µm2 area, which located in an originally scanned 5 × 5 µm2 scale image. Also, we use real optical resolution of DNA molecules instead of the optical resolution of the optical probe itself. These two values are deeply related but show different numerical values because of the complexity of the sample. In the case of far-field optical mappings, improvements in resolution from ∼3 Mbp to 1-3 kbp have been made by the use of the stretching technique before analysis.17-21 Similarly, we can conclude that the real near-field optical resolution of the long DNA molecule can be estimated only after stretching the DNA. 5988 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

Fluorescence Images of Different Optical Resolution and Analyzing YOYO-1 Intercalation State. Comparisons between the far-field and near-field fluorescence images of two different (17) Wiegant, J.; Kalle, W.; Brookes, S.; Hoovers, J. M. N.; Dauwerse, J. G.; van Ommen, G. J. B.; Raap, A. K. Hum. Mol. Genet. 1992, 1, 587-591. (18) Parra, I.; Windle, B. Nat. Genet. 1993, 5, 17-21. (19) Heisanen, M.; Karhu, R.; Hellsten, E.; Peltonen, L.; Kallioniemi, O. P.; Palotie, A. BioTechniques 1994, 17, 928-933. (20) Posenberg, C.; Florijn, R. J.; van de Rijke, F. M.; Blonden, L. A. J.; Raap, T. K.; van Ommen, G. J. B.; den Dunnen, J. T. Nat. Genet. 1995, 10, 477479. (21) Weier, H.-U. G.; Wang, M.; Mullikin, J. C.; Zhu, Y.; Cheng, J.-F.; Greulich, K. M.; Bensimon, A.; Gray, J. W. Hum. Mol. Genet. 1995, 4, 1903-1910.

Figure 7. Discrimination between single- and double-stranded λ-DNA. In the topography image (a), three strands are imaged. In the nearfield fluorescence image (b), signals are only produced for the DNA A in the topography. Image areas are 5 × 5 µm2. Each capital in the line profile analyses is the pair to the capitals in the topographic and near-field fluorescence images. Scanning parameters are 0.12 Hz/line, 256 pixel × 256 line with gate time 10 ms. In the fluorescence image (b), the signal intensity 1 V corresponds to 256 counts.

resolutions for λ-DNA molecules are shown in Figure 5. (Note: Because the image areas are different between the far-field and near-field fluorescence images, a direct comparison between the far-field and near-field images is not possible.) In Figure 5, the highest resolution image (c), fluorescence resolution of about 70 nm, allows one to distinguish fluorescent clusters, which are not possible with the lower resolution near-field image (b) (resolution 110 nm) or the far-field image (a). In the case of the far-field image, the optical signal of a DNA molecule is vastly scattered in large areas, and the optical resolution of the image may exceed 500 nm at fwhm with a very poor S/N ratio. However, the image contrast seems to be decreased with increasing fluorescence resolution. For a detailed discussion, expanded views and line profile analyses through the DNA molecules are captured in Figure 6. As shown in the line profiles along the DNA molecules, the fluorescence intensities are approximately divided into three ranges, namely, a background signal, an average signal of the best abundant peak intensity, indicated with a solid line in the line profiles, and a bigger signal explainable by the DNA shape or the heterogeneous states of YOYO-1. When the optical resolution is increased as shown in Figure 6b, the fluorescence signals are resolved cluster by cluster in which the number of the dyes is not clear. The continuity of the near-field fluorescence signal for the DNA is proportional to the used dye/bp ratio (data

not shown). For example, when the dye/bp ratio was about 1:1, a relatively continuous signal was produced despite a high resolution of 60 nm. The shortest distances between each separated cluster in the fluorescence image are similar to the optical resolution of the probe itself, i.e., the shortest separated distances are about 110 nm for Figure 6a and 70 nm for Figure 6b. Therefore, the detection of the localized dye or dye cluster is limited by the optical resolution of the probe itself. For example, in Figure 6b, a distance between two dye clusters smaller than 70 nm produces a continuous fluorescence signal, noted as (C) in the line profile for (B). (Note: The width of (C) is more than 2-fold than that of the neighboring peaks.) In the case of signalto-noise ratios (the ratio of the average signal intensity compared to the background), Figure 6a shows a slightly bigger value (∼6.8) than that of Figure 6b (∼5.6) but not a severe difference to understand the optically resolved DNA shape or YOYO-1 intercalation state. On the other hand, the far-field image in Figure 5a can also resolve the schematic location of each dye cluster by the use of a high-performance image software though its resolution is limited by the far-field diffraction. In conclusion, though a higher resolution is required to get a real image, Figure 6b is believed to be the best image for understanding the real intercalation state of YOYO-1 to the DNA. It shows a striking difference compared to the far-field fluorescence image in Figure 5a. Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

5989

Information from the Comparison between the Topography and Fluorescence Image of DNA. One of the best advantages of the SNOM/AFM is to produce topographic and optical images simultaneously. It means already developed labeling techniques of the fluorescence microscope can be combined with the high-resolution imaging and manipulation properties of the AFM. Figure 7 shows an example of the merit of the SNOM/ AFM. In the topographic image of Figure 7a, three λ-DNA strands are resolved. Line profile analyses for the sites, noted as (A) and (B) in the topography image, show differences in height and width. In site A, the height and the width of the DNA are about 0.25 and 72 nm (SD 10%), respectively, but those of site B are about 0.14 and 45 nm (SD 20%) at half-heights and widths. The width of the DNA in the topographic image is very big compared with the real width of DNA and normal AFM imaging because of the influence of tip size on the resolved DNA size.22 In our surface analysis, which noted as (C) in both Figure 7a and line profile analyses, it is revealed that the smallest particle size for the resolved surface is about 37 nm. Therefore, it is assumable that the smallest resolvable DNA is than 37 nm in width. In the fluorescence image, on the other hand, there are fluorescence signals for the DNA A but no fluorescence signals for the DNA B as shown in Figure 7B. To understand these results, a few considerations are necessary. First, Johansen and Jacobsen23 showed the binding mode of YOYO-1 to DNA. In that report, they showed that each monomer unit of YOYO-1 intercalates between bases with the benzazolium ring system sandwiched between the pyrimidines and the quinolinium ring between the purine rings, causing the helix to unwind. Also, the distortion in the local DNA structure caused by YOYO-1 bis-intercalation has been observed by two-dimensional NMR spectroscopy. Another report by Akerman and Tuite24 shows the cleavage of ssDNA and dsDNA due to the bis-intercalation or external binding of the dye. Local unwinding and distortion in DNA structure can affect the whole structure, resulting in a separation of the dsDNA to the ssDNA over long storage times. In normal conditions, without the intercalation of YOYO-1, the clear separation of the double strand to the single strand requires the involvement of enzymes and functional proteins such as helicases and single-strand binding proteins (SSB). Another possibility is the existence of the ssDNA in the supplied solution from the manufacturer. Though YOYO-1 can intercalate in the ssDNA, in principle, in our investigation for the far-field fluorescence microscope images, we have found that observation of the YOYO-1-intercalated ssDNA is very difficult when it is stretched. Also, in the manufacture’s instruction for YOYO-1, it is described that the intercalation of YOYO-1 is optimized for dsDNA, but it is possible for a small quantity of YOYO-1 to be intercalated or bound externally to the ssDNA. Therefore, in the mixes of both ssDNA and dsDNA, staining the ssDNA globally is practically impossible using this dye/bp ratio. For a quantitative analysis between ssDNA and dsDNA, we have made a topographic profile model as shown in Figure 8. The two models are selected for the comparison of calculated crosssectional areas; i.e., the cross-sectional area means the area occupied by the DNA in the line profile analysis of a topographic (22) Allen, M. J.; Hud, N. V.; Balooch, M.; Tench, R. J.; Siekhaus, W. J.; Balhorn, R. Ultramicroscopy 1992, 42-44, 1095-1100. (23) Johansen, F.; Jacobsen, J. P. J. Biomol. Struct. Dyn. 1998, 16, 205-222. (24) Akerman, B.; Tuite, E. Nucleic Acids Res. 1996, 24 (6), 1080-1090.

5990 Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

Figure 8. Model of a topographic profile for ss- or dsDNA, which is treated as a bundled single cylindrical shape (a), with a large tipsized optical fiber probe. A model of a topographic profile for three bundled DNA (b) in which each dsDNA is included in the calculation and used for comparison with model a. Comparison between the calculated areas of these two models and the measured crosssectional areas from Figure 7a by the number of strands (c). Each symbol (2), (9), and (b) indicates the results done by model (a), (b), and measured data in the experiment, respectively.

image.10b In Figure 8, model a considers all bundles of DNA as cylindrically shaped material in which the diameter is proportional to the width of DNA strands. Model b considers each strand as separated cylinders. In Figure 8c, two circular points correspond with the measured areas for the DNA A and the DNA B from Figure 7a. As shown in Figure 8c, the comparison between these models and measured areas with an assumed tip size of 40 nm, which is the smallest particle size in the background analysis of Figure 7a at fwhm, shows fundamentally corresponding results, though small errors still exist. The difference between the two models is not observed at strand 6 in Figure 8c. Previously, we reported the measured areas for multiple-stranded DNA molecules.10b In the cases of the DNA A and the DNA B, the calculated and measured data show a good correspondence, and the calculated data by model a indicate the strand compositions of Figure 7a.

In this comparison between the topographic image and the fluorescence image, though more consideration may be required by other methods, we can conclude that (A) is a dsDNA molecule and (B) is a denatured ssDNA molecule in Figure 7a. In principle, the discrimination between ssDNA and dsDNA from the other SPM25 or fluorescence microscope images was difficult because the other SPM methods could not produce a fluorescence imags and because a normal fluorescence microscope could not produce a topographic image. But in SNOM/AFM, if topographic and optical images are simultaneously produced, it is possible to discriminate these by comparing the topographic and fluorescence images. It has an important implication for the possibility of studying ssDNA using the SNOM/AFM method. For example, we can investigate a local DNA mutation by the hybridization of DNA probes by only using dye YOYO-1 as an indicator for the location. The optical resolution in Figure 7b is observed as about 45 nm from a mean width profile. The selected area is indicated as a rectangle in Figure 7b, and its mean width profile is presented as (D) in the line profile data. Recently, a group26 has succeeded in obtaining a near-field fluorescence image for a dye molecule with an optical resolution of about 20 nm, and therefore, more enhancement of the optical resolution in the SNOM/AFM is expected for single DNA imaging. (25) (a) Adam, T.; Woolley; Kelly, R. T. Nano Lett. 2001, 1 (7), 345-348. (b) Michalet, X. Nano Lett. 2001, 1 (7), 341-343. (26) (a) Hosaka, N.; Saiki, T. 6th international conference on near field optics and related technique; Twente, The Netherlands, 2000; p 173. (b) Sakai, T.; Matsuda, A. Appl. Phys. Lett. 1999, 74, 2773-2775.

CONCLUSIONS In this paper, we present the high-resolution near field fluorescence images of a λ-DNA molecule for describing the intercalation of dye YOYO-1 into the DNA done by the SNOM/ AFM method. In the fluorescence images, by achieving higher optical resolution, the intercalation sites of the dye are resolved with separated fluorescence signals, which are difficult to obtain from a direct far-field fluorescence image or a poor-resolution nearfield optical image. Particularly, we have described the possibility of discrimination between dsDNA and ssDNA that will enlarge the application of the SNOM/AFM method in the field of genetic studies. Currently, the best optical resolution obtained for single DNA is about 45 nm, which corresponds to a resolution of more than λ/10 of excitation light with a mean width evaluation. ACKNOWLEDGMENT Authors thank the Bio-oriented Technology Research Advancement Institute for support of the research and Prof. T. Ushiki of Niigata University, Prof. K. Fukui of Osaka University, and Dr. P. Degenaar of JAIST for their kind discussions. SUPPORTING INFORMATION AVAILABLE Full color image of Figure 5. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review May 9, 2001. Accepted September 23, 2001. AC010536I

Analytical Chemistry, Vol. 73, No. 24, December 15, 2001

5991