J. Phys. Chem. 1993,97, 3022-3027
3022
Photoelectron Imaging of DNA: A Study of Substrates and Contrast Douglas L. Habliston, C. Bruce Birrell, a d 0. H8ym Crifnth’ Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403
Gertrude F. Rempfer Department of Physics, Portland State University, Portland, Oregon 97207 Received: September 29, 1992; In Final Form: Nwember 10. 1992
Photoelectron imaging is a surface technique in which low-energy electrons are ejected from the specimen, usually by UV light. In biological applications, the choice of substrates is an important factor. Substrates must be conductive, have low photoelectron emission, be optically flat and stable, and promote spreading of biological specimens. The results of a survey of substrate materials for photoelectron imaging of DNA and other biological macromolecules are reported. On the basis of material contrast observed in the photoelectric measurements, it is predicted that DNA complexes with small proteins should be observable by photoelectron microscopy. Calculations show that topographical contrast, in addition to material contrast, is important in imaging nucleic acidsand proteins. Improved photoelectron images of duplex DNA were recorded using the k t of the substrates examined: a thin layer of magnesium fluoride on chromium-coated glass. These photoelectron images were obtained without staining, coating, or metal-shadowing the DNA.
e-
Iatraduction
Recent progress in photoelectronimaging is reaching the point where studies of DNA and DNA-protein complexes, as well as other cellular components, are becoming pra~tical.l-~Photoelectron imaging can involve excitation by a wide range of photon sources (e.g., UV lamps or lasers, synchrotron radiation, or even a conventional X-ray source). In biological applications the excitation is generally a mercury arc lamp. Electrons are photoejected from the surface as illustrated in Figure 1. The low-energy electrons are accelerated and then imaged by means of a series of electron lenses in an instrumentcalled a photoelectron microscope (PEM), diagramed in Figure 2.& In contrast to techniques in which information is obtained sequentially by scanning a tip or finely focused beam, PEM utilizes parallel imaging. PEM provides a unique material contrast mechanism based on differences in valence-state ionization potentials of molecules. This method also has high topographical contrast and depth resolution in biological specimens due to the low energy of the electrons (e.g., 0.1-1 eV) released by UV excitation. The IateralresolutionofPEM iscurrently7-10nm,with thetheoretical possibility of about 1-nm resolution when an aberration corrector is developed for this As with all other surface imaging techniques, the results are dependent on a careful choice of substrates. To date, there have been no published studies of substrates for photoelectron imaging of biological specimens. The purpose of this paper is to present data on some of the substrates we have been testing, to discuss contrast mechanisms, and to present improved photoelectron images of duplex DNA spread on one of these substrates.
Experimental Methods Substrate Reparaha All substrates were deposited on 5-mmdiameter No. 1 round glass microscope cover slips (Bellco Giass, Inc.). The cover slip discs were cleaned by first sonicating for 1h in a dilute solutionof MICROdetergent in water (International Products Corp.) followed by extensive deionized water washes and overnight soaking in 3 M HCI-3% H202 in deionized water. Before air-drying from distilled acetone in a laminar-flow clean air cabinet, the discs were rinsed extensively with deionized water. Chromium,gold, and magnesium fluoride were deposited over a period of 1-2 min by resistively heated evaporation in an oil0022-3654/93/2097-3022$04.00/0
J
hv
Figwe 1. Diagram illustrating photoemission of DNA spread on a solid substrate. The surface is broadly illuminated with UV light (hv),which photoejects the information-carrying electrons (e-). The initially slowmoving electrons arc accelerated and focused by electron lensea in a photoelectron micrmpc.
free Varian FC- 12 vacuum evaporator at a pressure of approximately 1 X le7Torr. Thicknesses were monitored by means of an Inficon XTM thin film thickness monitor utilizing a quartz crystal oscillator. Evaporated carbon films were prepared by resistively heating carbon rods in a spring-loaded carbon rod evaporation unit mounted in an Edwards E306A vacuum evaporator. Carbon film thickness was measured both with a quartz crystal oscillator thin film monitor and by optical absorptiongon rectangular glass cover slips coated simultaneously with the cover slips used for substrate emission measurements. All of the substrates prepared by evaporation were placed approximately 10 cm from the evaporative source and were at room temperature. Tin oxide-coated glass cover slips were prepared as described previously.1o Spermine was obtained from Sigma Chemical Co. Spermine-derivatized dextran was prepared by the procedure of Hicks and Moldayll except that spermine was used instead of ethylenediamine. Chromium-coated discs were overcoated with spermine-derivatizeddextran by floating the discs on aqueous solutions (1 mg/mL) of this material on Parafilm for 10 min. Excess spermine-derivatizeddextran was removed by edge-blottingfollowed by floating the discs on several drops of distilled water (edge-blottingthe discs before moving on to the next drop) and air-drying. This treatment caused the normally hydrophobic chromium oxide surfaceto become strongly hydrophilic. Bovine serum albumin (BSA) (Cohn fraction V, US. BioQ 1993 American Chemical Society
The Journal of Physical Chemistry, Vol. 97, No. 12, 1993 3023
Photoelectron Imaging of DNA
Relative Electron Emission 0
Chromium Tin Oxide Carbon
Gold MsF2 Deriv. Dextran DNA
U
--
'1\I
I
INTERMEOIATE LENS
BSA Lysozyme RecA
I
PROJECT ION LENS
PHOTOGRAPHIC
FILM
FLUORESCENT SCR€EN
Figure 2. Simplified diagram of a photoelectron microscope. The specimen is at high negative potential (e.g.,-30 kV). The light is reflected from the polished anodemirror onto the specimen. Electronsphotoemitted from the specimen surface are accelerated in the cathodeanode gap and then imaged by a set of electron lenses resembling those of a transmission electron microscope. The PEM is of ultrahigh-vacuum design in order to avoid contamination of the specimen.
chemical Corp.), chicken egg white lysozyme (Sigma Chemical Co.), and RecA protein (U.S. Biochemical Corp.) were used without further purification. Tobacco mosaic virus (TMV) was a gift of T. Shalla. DNA used for photoelectric measurements was plasmid pBR322 from New England Biolabs. Approximately 80% of the preparation consisted of supercoiled plasmids with the remainder being a mixture of relaxed circles and linear DNA. Plasmid and protein samples were deposited from aqueous solutions on glass cover slips coated first with 5 nm of chromium immediatelyfollowed by 5 nm of MgF2 without breaking vacuum between the evaporations. To prepare the thin films, 50-pL drops of DNA plasmids (1 pg/mL) or protein (1 mg/mL) were placed on Parafilm, and the discs were floated on the drops for 10 min, followed by extensive rinsing with water for 1-2 min to remove unbound DNA or protein, blotted, and air-dried. Preparation of DNA Samples for Photoelectron Imaging. Plasmid +XI74 R F I1 (relaxed form) was obtained from New England Biolabs as a 1 mg/mL stock in 10 mM Tris-HC1,l mM EDTA, pH 8.0. The plasmid solution was diluted to 10 ng/mL with distilled H20, and a 40-pL drop was placed on a piece of Parafilm. A 5-mm chromium-coated glass cover slip overcoated with approximately 5 nm of MgF2 was floated on the DNAcontaining drop for 1 min and transferred to a 40-pL drop of distilled H20 containing approximately 1OloTMV per milliliter. After 1 min, the specimen disc was rinsed by a continuous flow procedure in which water was rapidly added to the drop and aspirated off over the course of another minute. The cover slip was then removed from the drop, and the very small amount of H20 remaining on the cover slip rapidly evaporated. The TMV was added to provide a larger and easily distinguished object to aid in focusing.
Figure 3. Relative electron emission of different substrate materials in comparison with thin films of DNA and protein (BSA, lysozyme, and RecA). All sampleswere deposited on glass cover slips coated with 5 nm of chromium. The other evaporated specimens (carbon, gold, and magnesium fluoride) are also 5 nm thick. These data were collected in the photoelectron microscope equipped with an Osram HBO 100 W/2 mercury short-arc lamp. The emission currents are plotted on a log scale.
PhotoelectronMicroscopy. The photoelectronmicroscope used in this study was an instrument of oil-free vacuum design to avoid specimencontamination. The instrument design,computer-aided control, and image processing system are described in detail elsewhere.12J3Excitation was provided by two Osram HBO 100 W/2 short-arc mercury lamps. Intensity plots were obtained by digitizing original electron image film negatives by means of a Panasonic WV-CD5O CCD video camera equipped with a 50mm Nikon Nikor macro lens coupled to Imaging Technology Inc. Series 150 image processing modules interfaced to a Silicon Graphics IRIS 4D20 workstation using image analysis software by Wyndham Hannaway and Associates.13 Beam Current Measurements. Electrons were collected on the output fiber optic centered on the electron optical axis on the bottom flange of the photoelectron microscope. The electron optical system was focused so that the maximum number of electrons emitted from the sample surface was projected onto the aluminized phosphor-coated fiber optic. The ground-insulated aluminized coating was connected to a Keithley Model 2632 picoammeter through a BNC feed-through and cable. Beam currents were recorded in 1-min increments over a 15-min time period. Output from the UV sources was adjusted such that, at this low magnification, the beam currents of the most photoemissive and least photoemissive samples were in the sampling range of the electrometer. The relative photoelectron currents are reproducibleto within k20% except magnesium fluoride over chromium, which had somewhat larger variability. Values reported here are the averages of measurements on at least three specimens.
Results and Discussion
Properties of Substrates. The ideal substrate for biological applications of PEM would have the following properties: (1) sufficient conductivity so that electrons emitted from the object can be replenished from the cathode via the substrate, (2) low photoelectronemission yield over the wavelength region of interest, (3) a flat surface, (4) stability in air and water, and ( 5 ) surface properties that promote the binding and spreading of biological specimens. The relative photoelectron emission of several substrate materials is given in Figure 3. All of these substrates
3024 The Journal of Physical Chemistry, Vol. 97, No. 12, 1993
pass the conductivity criterion. Because the specimen surface is illuminated during photoelectron current measurements, photoconductivity may complement the intrinsic low conductivity of the thin films of magnesium fluoride, spermine-derivatized dextran, DNA, and the proteins BSA, lysozyme, and RecA. The light source is a broad spectrum mercury arc lamp with a shortwavelength cutoff at 200 nm. Although not monochromatic, this lamp has the virtue of being the standard excitation source in photoelectron imaging of biological specimens. Thus, the data of Figure 3 aredirectly applicablein comparing substrate materials for actual biological experiments. Screening by the second criterion, some substrates popular in other surface techniques, such as carbon and gold, are not suitable becauseof their relatively high photoemission. The tin oxide-coated glass is less photoemissive than gold and carbon. However, tin oxide fairs poorly on criterion 3, because we have found it difficult to prepare flat and uniform surfaces as judged by photoelectron images (not shown). Stainless steel rods exhibit about the same level of electron emission as the chromium-coated glass discs. Both are probably covered with a thin layer of metal oxide, since they have been exposed to air. In any case, the chromium-coated glass discs have many advantages over stainless steel rods. The chromiumcoated glass cover slips can be prepared in relatively large batches and are disposable whereas the stainless steel rods are more expensive to prepare, inherently have scratch marks, and must be polished after every use. In addition, the thin chromium layer issemitransparent. The thickness of 5-nmchromiumwasselected because there is sufficient conductivity for PEM while retaining enough optical transmittance (Le., 30% transmittance to visible light) to permit viewing of the samespecimen by light microscopy. Chromium adheres well to glass and forms a reproducible and optically flat substrate for the deposition of other materials, for example, the thin films of carbon, magnesium fluoride, and spermine-derivatized dextran used to obtain the data of Figure 3. We concludethat chromium is the most useful of the conductive coatings examined. However, a chromium metal or metal oxide surface is not optimal for the binding and spreading of nucleic acids and proteins. For this reason the chromium is coated with a thin layer of magnesium fluoride. Not only does the magnesium fluoride facilitate the binding and spreading of DNA, but it suppresses the background photoelectron emission of the chromium (see Figure 3). Magnesium fluoride is commonly used in fabricating lenses and windows and as an optical coating. It does not absorb light in this wavelength region and therefore does not emit a significant number of electrons. Lithium fluoride is also widely used in optics, but it is too water soluble for substrates that must be exposed to aqueous samples. Derivatized dextran, although not as uniform as magnesium fluoride-coated chromium, also promotes binding and spreading of DNA. Material Contrast in Photoelectron Imaging. Surface imaging techniques ideally have two sources of contrast: one to provide information regarding the nature of the specimen (material contrast) and the second to identify its location on the uneven specimen surface (topographic contrast). In photoelectron imaging, the material contrast is provided by the valence state electrons released by UV light. For organic and biological specimens,material contrast is best considered to be an empirical quantity, since many factors are involved. It is known, for example, that there is a "brightening effect" whereby the relatively low initial electron emission rises markedly with exposure to UV light and typically levels off after a few minutes in the PEM.14 The brightening effect is large for amino acids and proteins and much less for nucleic acids. This effect is presumably caused by photochemistry. (It is the opposite from the 'bleaching effect" observed in fluorescence microscopy.) The data of Figure 3 provide useful information on material contrast because they are
Habliston et al.
Figure 4. Diagram illustrating the emission of electrons from DNA or other small obicct in the DhotoelwtrOII microscope. The electrons are emitted in a wide range of directions. The eleckc field between the cathode (specimen) and anode accelerates these electrons and c a w them to follow parabolic trajectories.
recorded after the brightening effect has occurred in the a c t a 1 imaging environment. From Figure 3, the thin layer of DNA is clearly more photoemissive than a magnesium fluoride-coated chromium on glass substrate. To decide whether DNA would be visible from material contrast alone, it is instructive to estimate the brightnessor intensity ratio (Z)between the resolution element containing DNA and the substrate background according to the relation
wherefi is the fractional area occupied by DNA in the resolution element, 1 -fi is the fractional area of the exposed substrate in the resolution element, Y2 is the photoelectron quantum yield of DNA, and YIis the photoelectronquantum yield of the substrate. The current resolution of PEM is on the order of 7-10 nm7 and the diameter of duplex DNA is 2 nm; thusf2 = 0.2. From Figure 3, the ratio Y ~ / Y=I6,and for a resolution of 10 nm, eq 1 gives I = 2. This indicates that from material contrast alone the duplex DNA should be visible on the magnesium fluoride-coated chromium substrate. In order to assess the feasibility of imaging protein-DNA complexes, it is useful to compare the relative photoemission of these two types of biopolymers. Bovine serum albumin (BSA), egg white lysozyme, and RecA protein were selected as representative proteins for this study. All three gave similar results, as shown in the last three bars in Figure 3. The photoemission of proteins under irradiation of UV light from a mercury shortarc lamp is clearly greater than that from DNA, suggesting that proteins will show up as bright spots on duplex DNA. Shrinking the size of the resolution element will enhance this material contrast. If the instrumental resolution can be improved from 10 to 2 nm by means of aberration correction, as has been recently suggested: thenf2 = 1, and from eq 1, I = 6 for DNA on the substrate and I = 12 for a 2-nm diameter globular protein bound to DNA. Topographic Contrast in Pbotoekbon Imaging. Figure 4 illustrates the image formation process in a photoelectron microscope. The low-energy electrons are emitted in a wide range of directions. The high electric field between the cathode (specimen) and the grounded anode causes the electrons to follow approximately parabolic trajectories. The tangents to these parabolas intersect at a point that is approximately the same distance behind the cathode as the anode is in front of it. The point of intersection of the tangents defines thevirtual specimen. Local topography has an effect on the direction of the emission velocity and on the local accelerating field experienced by the slow-moving electrons as they leave the surface. The net effect is a contrast enhancement (dark or bright, depending on the placement of the aperture stop) at the edges of protrusions, steps, or depressions.lsJ6 Step heights as small as 3 nm on surfaces
Photoelectron Imaging of DNA TO ANODE ATz=]I
7 ELECTRON TRAJECTORY
TOPOGRAPHIC RIDGE\
The Journal of Physical Chemistry, Vol. 97, No. 12, 1993 3025 approximates a hyperbolic field, with the substrate and the side of the ridge serving as asymptotes. By making use of the hyperbolic field approximation, one can greatly simplify the calculation of the electron trajectories. The equations of motion are separable in the YZ’coordinate system (dashed axes at 4S0 to the I 2 axes), and the solutions are readily obtained. The expression for the hyperbolic field in terms of y’and z’takes the form
v = k(z’2-yf2)
Figure5 Trajectory of an electronemitted from the side of a topographic ridge. The electron has a transverse.velocity which is the resultant of the initial transverse.velocity and the velocity acquired from acceleration by the transverse component of the microfield at the ridge.
with little or no material contrast have been detected experimentally.” The effect of topography on the electron beam has been solved analytically for several simple geometries.l* Electrons emitted from the side of a topographical feature playa role in contrast formation. Theseelectrons have transverse componentsofvelocity(uL). Thereare twoeffacts thatcontribute to uI . One is the component of the emission velocity perpendicular to the optical axis, and the other is the result of a sideward acceleration of the electron by the transverse component of the microfield at the tilted surface of the topographic feature. The directions of electrons emerging from a point on a surface aredistributed around an axis normal to thesurface. If the surface normal is tilted away from the optical axis, the components of the emission velocity at right angles to the optical axis are shifted in the direction of tilt. For the center ray, the initial transverse component is
ueL = u, sin 8,
(2) where u, is the emission velocity and eT is the tilt angle of the emitting surface. The additional effect of the electric field at the tilted surface contributes to the resultant transverse velocity uI. After accelerationto the final velocity D, by the potential difference between the cathode and the anode, the angle which the ray makes with the axis is
e, = sin-’ u I / u ,
(3) The electron bundle ends up off-center in the aperture plane, and a smaller percentage of the bundle is transmitted by the aperture stop (assumed centered) than in the case of a centered beam. This discrimination results in contrast because it reduces the illumination in the image at the sides of a topographic feature relative to that at the top. The transverse velocity acquired as a result of the sideward acceleration of the electron beam (caused by the electric field at the specimen surface) depends on the size of the topographic feature, whereas the transverse component of the initial velocity depends not on the size but on the slope. Thus, the relative importance of these two effects in producing contrast depends on the scale of the topography. We calculate here the effect of topography on the transversevelocityfor the case of a ridge having a square cross section of height and width h 1, z’is described by a sinh rather than a cosh function.) The velocities in thez’andy’directions aregiven by the timederivatives of eqs 6a and 6b. The electron travels in the hyperbolic field out to the boundary, where it makes a transition to a parabolic path. The velocity at the end of the hyperbolic path is transferred to the parabolic path. For the present calculation we are interested in the velocity transverse to the optic axis. The electron’s emission velocity u, is assumed to be normal to the side of the ridge. The initial conditions are z’,, = ( 1 / 2 ) h / f i = 0 1 2 , y’,, = - ( 1 / 2 ) h / &
z’,, = y; = U
J f i
= -012 (74 (7b)
At the hyperbolic boundary, z‘= D and, since 2’0 = 012,cosh(wt +) = 2 cosh $. From this condition, wt can be found and used to obtain y’ and the velocity components at the end of the hyperbolic region. The resulting velocity transverse to the optic axis can then be used in eq 3 to find the slope of the electron trajectory after acceleration to the final velocity us. For an electron leaving the surface with zero initial velocity, eqs 6 simplify to
+
z’ = z’,, cosh wt
(84
y ’ = y’,, cos wt
(8b)
In this case, the transverse velocity is acquired solely in the hyperbolic field. The velocity components can be obtained from the derivatives of eqs 8. The resultant velocity at the boundary can be calculated by combining the components quadratically or
3026
Habliston et al.
The Journal of Physical Chemistry, Vol. 97, No. 12, 1993
Figure 6. Photoelectron image of naked 4x174 DNA plasmids. The substrate is a chromium-coated glass cover slip overcoated with a thin layer of MgF2. The exposure time was 70 s. White arrow: one of the many plasmids showing crossover of two DNA strands. Dark arrow: a tobacco mosaic virus (TMV) marker, used in focusing the image.
from the energy relation ( 1 /2)mu2 = evboundary = (1 / 4 ) ( h / l )Va, where Va is the accelerating voltage. Thedirectionof the resultant velocity, determined by the components, makes an angle of about 74O with the optic axis. The angle 0, ( = U J U a ) of the trajectory after acceleration can be written
e, = ( v ~ u n d , r ,sin / ~740 a ) =1 0.5(h/Z)'/2(0.96) ~2 = 0.48(/~/Z)'/~ (9) In general, both the emissionvelocity and the transverse electric field contribute appreciably to the transverse velocity. In UV photoelectron microscopy the emission energies range from 0 to about 1 .O eV, and representative values of the acceleratingvoltage and cathode-anode spacing are Va = 30 kV and 1 = 3 mm. For a step of height 300 nm and transverse emission energy of 0.3 eV (Le., with uc normal to the side of the step), 0, = 5.9 mrad. After divergence by the anode aperture lens, the angle entering the objective lens becomes 81 = (3/2)0, = 8.9 mrad. This angle is well outside theangle of rays transmitted by theobjectiveaperture stop. The limiting angle for a typical aperture diameter of 50 pm, and focal length of 7.5 mm is a1 = 3.3 mrad. As a consequence, the image in the vicinity of the face of the step appears dark. In the foregoing example the calculated contribution due to the sideways acceleration in the electric field turned out to be somewhat larger than that due to theemission velocity. However, for very small topographical features the effect of the transverse field becomes much less important. For example, if we simulate the topography of a DNA strand by choosing h = 20 A, the result given by eq 9 for the effect of the field alone is
8, = 0.48(20/3
X
= 0.4 mrad
After divergence by the aperture lens at the anode, the angle entering the objective lens is increased to
8 , = (3/2)8, = 0.6 mrad With an aperture limiting the angle to 3.3 mrad, the contrast effect is small; the electron is off-center by only about 20% of the aperture radius as it passes through the aperture plane. On the other hand, with even 0.1 eV of (transverse) emission energy, the
above deflection angles are increased to
8, = 1.8 mrad and 8, = 2.7 mrad The electron is now displaced by about 80%of the aperture radius at the aperture plane. Thus, for an average transverse emission energy of 0.1 eV, a contrast ratio approaching2 can be expected. With larger emission energies and/or smaller aperture sizes, larger contrast ratios are possible. Other contrast mechanisms may contribute to making DNA and DNA-protein complexes visible in photoelectron imaging. For example, the de Broglie wavelength of a 0.5 eV electron is 17 A. Since this is of the order of the diameter of DNA, electron interferencecan beexpected and may further enhance thecontrast. Photoelectron Imaging of DNA. Taking advantage of the information on substrate materials presented in Figure 3, we have obtained a substantial improvement in images. As an example, Figure 6 is of a naked DNA plasmid preparation spread on the magnesium fluoride over chromium substrate. This photoelectron image was obtained without metal shadowing, coating, or staining, as is normally required in TEM in order to enhance contrast. These photoelectron imagesof DNA are stable over a period of hours. Figure 6 shows only a fraction of the field of view recorded in a typical photoelectron micrograph. Information can be obtained simultaneously from a large number of plasmids (or an equivalent amount of well-spread chromosomal DNA) because PEM, like TEM, is a parallel imaging method. An enlargement of a naked DNA plasmid is shown in Figure 7A. The point of crossover of one duplex over another is clearly visible (see arrow). The relative brightness of the DNA at the crossover point is increased by at least a factor of 2, as shown in the intensity plot of Figure 7B. This is a consequence of two contributing factors: topographic contrast and the larger fractional area cf2) of the image element occupied by DNA. Since the DNA is only 2 nm in diameter, the ability to detect the overlapping DNA duplexes is a strong indication that it will be possible to detect proteins of 2-nm diameter or less bound to DNA. Proteins would be more readily detected because of the added material contrast. In previous work, we reported photoelectron images of bacteriophage Tq DNA and of DNA plasmids spread with cytochrome c (Kleinschmidttechnique).' The cytochrome ccoats the DNA, increasing the effective diameter to 7-10 nm. Naked
Photoelectron Imaging of DNA
The Journal of Physical Chemistry, Vol. 97, No. 12, 1993 3027 from that of other imaging methods and could be useful, for example, in studies of gene transcriptiorl complexes. Topographic contrastis also significantsincephotoelectron imaging is unusually sensitive to local topography. Calculations show that, for macromolecules with dimensions on the order of DNA, the topographical contrast alone should be sufficient to permit detectionof these macromolecules. For these very small objects, the magnitude of the initial transverse velocity of the emitted electrons is more important than the sideward acceleration of the emitted electrons produced by the microfield at the object, although the latter is often considered to be the main origin of topographic contrast in photoelectron microscopy. The combination of material contrast and high topographic contrast makes photoelectron imaging a promising approach to studies of nucleic acids and nucleic acid-protein complexes. As a practical example, significantly improved images of DNA plasmids have been obtained using the new substrate, without any staining, coating, or shadowing of the DNA. The images are stable over a period of hours, indicatingthat any photochemistry taking place does not degrade the image quality. The points at which one DNA duplex crosses over another are observed, confirming the sensitivity of photoelectron imaging to very fine topographical detail.
Figure 7. (A) Higher magnification photoelectron image of a naked
4x174 D N A plasmid. The points where the duplex D N A strands crossover show up clearly (see arrow). (B) A perspective type view of the image in (A) with the intensity ratio (I)presented as the vertical height at each point along the D N A .
plasmid DNA on a chromium-coated glass substrate was barely detectable, but large E. coli RNA polymerase protein complexes bound to DNA were readily imaged. These earlier observations are consistent with the beam current data presented here. Improvement in substrates to achieve lower emission and greater uniformity contributes to the achievement of higher quality photoelectron images of DNA. Conclusions The significance of photoelectron imaging of DNA lies in the information content of the photoelectric effect and the nature of the imaging process. In order to make use of this information, substrates must be identified which provide a uniform background. A series of possible substrates have been screened, and one substrate has been identified (magnesium fluoride over chromium on glass) which provides a significant improvement. During the course of this work a better understanding of material contrast and topographical contrast of biological macromolecules has been achieved. The material contrast, although empirical, is reproducible. Photoelectron beam current data indicate that there is substantial contrast between proteins and DNA. Since the photoemissivity of proteins is greater, it should be possible to image proteins bound to DNA. This source of contrast differs
Acknowledgment. We thank our colleagues Dr. Karen K. Hedberg, Walter P. Skoczylas, and Denis Desloge for useful discussions. O.H.G. acknowledges the early training of his mentor, Dr. Harden M. McConnell, who instilled in his students enthusiasm for tackling challenging problems. This work was supported by PHS Grant CA 11695 from the National Cancer Institute. References and Notes (1) Griffith, 0. H.; Habliston, D. L.; Birrell, G. B.; Schabtach, E. Biopolymers 1990, 29, 1491-1493. (2) Griffith, 0. H.; Habliston, D. L.; Birrell, G. B.; Skoczylas, W. P. Biophys. J . 1990,57,935-941. (3) Birrell, G. B.; Hedberg, K. K.; Habliston, D. L.; Griffith, 0. H. Ultramicroscopy 1991, 36, 235-251. (4) Griffith, 0. H.; Engel, W. Ultramicroscopy 1991, 36, 1-28. (5) Mundschau, M. Ultramicroscopy 1991, 36, 29-51. (6) Griffith, 0. H.; Rempfer, G. F. In Aduances in Opticaland Electron Microscopy; Barer, R., Cosslett, V. E., Eds.; Academic Press: London, 1987; Vol. 10, pp 269-337. (7) Rempfer, G. F.; Griffith, 0.H. Ultramicroscopy 1989,27,273-300. Rempfer, G. F.; Griffith, 0. H. Ultramicroscopy 1992, 47, 35-54. (8) Skoczylas, W. P.; Rempfer, G. F.; Griffith, 0. H. Ultramicroscopy 1991,36, 252-261. (9) Moretz, R. C.; Johnson, H. M.; Parsons, D. F. J . Appl. Phys. 1968, 39,5421-5424. (10) Nadakavukaren, K. K.; Chen, L. B.; Habliston, D. L.; Griffith, 0. H. Proc. Natl. Acad. Sci. U.S.A. 1983,80,4012-4016. (1 1) Hicks, D.; Molday, R. S. Inuest. Ophthalmol. Visual Sci. 1985.26, 1002-1013. (12) Rempfer, G. F.; Skoczylas, W. P.; Griffith, 0. H. Ultramicroscopy 1991.36, 196-221. (13) Habliston, D. L.; Baker, B.; Griffith, 0. H. Ultrumicroscopy 1991, 36,222-234. (14) Griffith, 0. H.; Holmbo, D. L.; Habliston, D. L.; Nadakavukaren, K. K. Ultramicroscopy 1981,6, 149-156. (15) Schwarzer, R. A. Microsc. Acta 1981, 84, 51-86. (16) Soa, E.-A. Exp. Tech. Phys. 1965, 13, 393-408. (17) Reimer, L.; Schulte, E.; Schulte. Chr.; Schur, K. Beitr. Elektronenmikroskop. Direktabb. Oberjl. 1972,5, 987-1006. (18) Rempfer, G . F.; Nadakavukaren, K. K.; Griffith, 0. H. Ultramicroscopy 1980, 5, 437-448.