MICROSCOPY RESEARCH AND TECHNIQUE 36:301–312 (1997) Imaging of BrdU-Labeled Human Metaphase Chromosomes With a High Resolution Scanning Ion Microprobe RICCARDO LEVI-SETTI,1 * JAN M. CHABALA,1 KONSTANTIN GAVRILOV,1 RAFAEL ESPINOSA III, 2 AND MICHELLE M. LE BEAU2 1The Enrico Fermi Institute and Department of Physics, The University of Chicago, Chicago, IL 60637 of Hematology/Oncology, Department of Medicine, The University of Chicago, Chicago, IL 60637 2Section KEY WORDS chromosome banding-nucleoside SIMS mapping-sister chromatid exchangefragile site-Giemsa ABSTRACT Detailed maps of the A-T distribution within human mitotic chromosomes labeled with BrdU are obtained with a high resolution scanning ion microprobe through the detection of bromine by imaging secondary ion mass spectrometry (SIMS). Corresponding maps of the emission loci of the molecular ion CN2 describe the overall DNA, RNA and protein distribution in the chromosomes. Several chromosome preparations exhibit base-specific banding patterns (SIMSbands) which mimic the well known G- or Q-bands resulting from conventional staining methods for optical microscopy. SIMS-bands are more noticeable in mitotic cells at the first cell cycle and after in situ denaturation or Giemsa staining. Sister chromatid exchanges (SCE) at the second cell cycle and beyond, occurring both spontaneously and promoted following cell culture exposure to the chemical aphidicolin (an inhibitor of DNA replication), can be visualized readily from the relative label signal intensities between sister chromatids. The comparison of base-specific label maps with CN2 maps, in conjunction with the appearance of base-specific banding patterns, is informative about protein survival and/or removal following different chromosome preparation protocols. In addition, the resulting condensation state of the chromosomes can be appraised during SIMS analysis from the sample topography (imaged via the collection of mass-unresolved secondary ions). We demonstrate that imaging SIMS is a powerful complement to existing methods for the study of banding mechanisms and for the elucidation of chromosome structure. The advantages of this novel approach to the systematic and quantitative study of cytogenetic phenomena and methodologies are still largely untapped. Microsc. Res. Tech. 36:301–312, 1997. r 1997 Wiley-Liss, Inc. INTRODUCTION The University of Chicago Scanning Ion Microprobe (UC SIM) utilizes Secondary Ion Mass Spectrometry (SIMS) imaging to map the distribution of elements and several chemical compounds in soft and mineralized biological material, at a typical image resolution of 30–40 nm (Levi-Setti, 1988). The instrument uses a 40–60 keV energy, focused heavy ion probe to scan the surface of a material, leading to the emission of secondary ions which are constituents of the material itself. These secondary ions are then collected and analyzed by a mass spectrometer, and recorded to generate mass-resolved images. Because elemental isotopes can be detected by this approach, the method lends itself to the use of isotope tracers that need not be radioactive. High-resolution SIMS imaging microanalysis with the UC SIM has already proven to be of unique value in a number of biological applications (Chabala & LeviSetti, 1987; Levi-Setti, 1988). More recently, we applied the SIMS methodology to the mapping of labeled nucleosides in human metaphase chromosomes (for a review, see Levi-Setti and Le Beau, 1992), and in Drosophila polytene chromosomes (Levi-Setti et al., 1994a). Here we discuss recent advances in the study of human metaphase chromosomes. r 1997 WILEY-LISS, INC. The technique was first tested using bromodeoxyuridine (BrdU, a thymidine analog) and 14C-thymidine as labels, to explore the feasibility of studying sister chromatid exchange (SCE) and fragile sites by this method (Hallégot et al., 1989). SCEs result from a spontaneous recombination, by which homologous DNA sequences cross over between sister chromatids, and are thought to be associated with various pathologies. Their rate of occurrence may be promoted by chemicals and may serve as an index of the mutagenic properties of these chemicals. Fragile sites are loci where chromosomes are prone to break and undergo rearrangements, including homologous and non-homologous recombination. In a few cases, these fragile sites are known to be related to genetic abnormalities. However, their pathological role, genetic and molecular basis, and possible connection with SCE, are generally not known. The label signatures in these experiments were the secondary ion 81Br (free of the PO3 interference that affects the other Br isotope at mass 79) and 28CN2 ( 14C14N), also Received 5 January 1996; accepted in revised form 5 March 1996. *Correspondence to: Prof. Riccardo Levi-Setti, The Enrico Fermi Institute, The University of Chicago, 5640 S. Ellis Avenue, Chicago, IL 60637, U.S.A. Grant from: National Science Foundation, Grant number: BIR 9317959. 302 R. LEVI-SETTI ET AL. Fig. 1. Schematized representation of the components of the University of Chicago Scanning Ion Microprobe (UC SIM). By means of a ‘‘switchyard’’ quadrupole, the secondary ion beam can be directed either into an Extranuclear 300 RF quadrupole mass filter, or into a modified Finnigan MAT 90 magnetic sector mass spectrometer. (From Levi-Setti et al., 1994, reproduced by permission of The Royal Microscopical Society). devoid of significant interferences and copiously emitted from DNA. Additional information concerning the chemical structure of the chromosomes can be obtained through mapping with the latter molecular ion in its normal 26CN2 form, which accurately describes the cumulative distribution of proteins, as well as of DNA and RNA, providing detailed histological SIMS maps. Subsequent studies using the above labels did succeed in detecting the differential label staining expected for sister chromatids, as well as possible examples of SCE (Levi-Setti and Le Beau, 1992). It also became apparent that the nucleoside distribution along the chromatids was not uniform, and that banding patterns could be recognized. In a few examples, these ‘‘SIMS bands’’ appeared similar to the well known patterns resulting from the trypsin-Giemsa stain (G-bands) and the fluorescent quinacrine mustard stain (Q-bands). The SIMS bands acquire particular significance, as will be discussed further below, because by this approach we hope to shed light on the true biochemical nature of the banding patterns observed with the more conventional staining methods. This is an issue that has not yet been resolved (Sumner, 1990) and that has been thought, in rather unspecified terms, to involve DNA-protein interactions. The ability to differentiate isotope-tagged DNA from proteins using the SIMS technique is particularly relevant in regard to this hypothesis. Although encouraging, our previous attempts to map base-specific label distributions in human metaphase chromosomes were frustrated by the relatively poor transmission (,0.2%) of the RF quadrupole mass filter incorporated in the UC SIM for SIMS analysis at that time. The resulting detected counts were often insufficient to construct images with statistically useful de- ION MICROPROBE IMAGING OF HUMAN CHROMOSOMES 303 Fig. 2. (a): 81Br2 SIMS map of a BrdU-labelled mitotic cell, obtained by adding 7 single-pass maps, 35 minutes overall scan time. Three chromosomes overlap an unrelated cell nucleus. The image displays 9.1 3 105 counts. Sister chromatids are equilabeled, indicating that the cell is undergoing the first mitosis following exposure to BrdU. 45 µm full scale. (b): Same as the image in (a), shown in inverted contrast to facilitate comparison with optical staining. (c): Magnified detail of (b), exhibiting banding patterns (SIMSbands). 2.0 3 105 counts displayed. 23 µm full scale. (d): 26CN2 map of area in (c), from a single-pass scan of 2.2 minutes duration. 1.2 3 106 counts displayed. The structure of the basespecific map in (c) is not visible in this map, which includes the overall contribution from DNA, RNA and proteins. tail. In addition, difficulties were encountered in obtaining satisfactory spreading of mitotic cells on conducting substrates (e.g., Au foils), which exhibit hydrophobic behavior. The limitations resulting from low count statistics have been largely overcome in the present investigation, which takes advantage of a radical upgrade of the UC SIM. The instrument is now coupled to a high performance magnetic sector mass spectrometer and exhibits a better than fifty-fold increase in SIMS sensitivity (Chabala et al., 1995; Levi-Setti et al., 1994b). At the same time, a satisfactory protocol has been developed to obtain consistently well-distributed chromosome spreads on conducting substrates (LeviSetti et al., 1995). It is now feasible to pursue several of the original goals of our program, concentrating here on the SIMS analysis of alcohol-acid-fixed, BrdU-labelled chromosomes. Here we present details of the appearance and occurrence of base-specific SIMS-banding in relation to cell division and to in situ procedures such as DNA denaturation and Giemsa staining. These procedures will be shown to deplete the protein content of the chromosomes to some extent. The differential BrdU labelling of sister chromatids as a function of the number of cell divisions will be shown to enable the routine detection of SCEs occurring in normal mitotic cell cultures, and at an increased rate following exposure of the cultures to the chemical aphidicolin. The UC SIM can also yield images of the surface topography of a sample by collecting the overall ion-induced secondary ion (ISI) emission from the sample; this feature will be used to illustrate the morphological changes associated with the different chromosome preparation procedures mentioned above. MATERIALS AND METHODS Sample Preparation Because of the irreversible structural incorporation of the BrdU label into DNA, the standard culture and specimen preparation techniques employed for optical microscope observation are adequate for SIMS analysis. However, the requirements that the substrates be conducting to prevent electrical charging, and at the same time wettable by the mitotic cell fixative, necessitates the development of a special methodology. The chromosome samples were obtained from human peripheral blood lymphocytes of healthy individuals, cultured in BrdU-containing medium following the protocol previously described (Levi-Setti and Le Beau, 1992). Relevant to the present context is the incubation time (72 hours), which allows, on the average, three cell divisions to occur. After arresting the cells in mitosis with Colcemid (0.05 µg/ml for 45 minutes, leading to a wide range of chromosome condensation levels), the cultures were treated with hypotonic KCl (0.075 M, 8 minutes) followed by 4–6 changes of fixative (absolute methanol: glacial acetic acid, 3:1). This method of fixation is known to remove some of the histone (primarily H1) and non-histone proteins from the chromosomes (Sumner, 1990). However, estimates of this loss vary over a wide range. As will be shown below, our observations suggest that a substantial amount of these proteins is 304 R. LEVI-SETTI ET AL. Fig. 3. (a): Sum of 8 81Br2 SIMS scans totaling 36 minutes, 9.8 3 106 counts, 38 µm vertical full scale. The label intensity ratio between sister chromatids of this BrdU-labeled cell averages close to the expected 1:1/2 ratio for second division. (b): Magnified detail of (a), showing structure in the packing of chromatin, but no recognizeable banding. 3.6 3 105 counts, 19 µm full scale. (c): Same as in (a), shown in inverted contrast. Three SCEs are marked by pointers. (d): Another detail of the same chromosome spread, exhibiting 2 additional SCEs. 3.5 3 105 counts, 19 µm full scale. still present after the above exposure to acid fixative. Instead of using glass slides as substrate, as is routine for optical microscope observation, the cell suspensions were dropped onto glass coverslips that have been lightly Au-coated and treated with a non-ionic surfactant (Triton X-100, 0.1%), and air dried. This procedure ensured uniform wetting of the substrate by the cell suspension until dry, a feature that is essential to obtaining well distributed chromosomes which are free of deformity. For SIMS analysis, the samples were further coated with a thin, sputter-deposited layer of gold. Several variants to the above standard preparation protocol were introduced, both at the culture level and in situ, to explore the chemical basis of a number of cytogenetic phenomena and procedures: 1) To promote the occurrence of SCE and the expression of fragile sites, some of the BrdU-containing cultures were exposed to the chemical aphidicolin (0.4 µM) for 24 hours at 37°C (Rassool et al., 1992). 2) Toward elucidating aspects of the staining process, some of the samples were Giemsa and trypsin-Giemsa stained, following standard procedures. 3) To explore the chemical changes associated with DNA denaturation, some of the samples were subjected to the procedures for fluorescent in situ hybridization (FISH), except exposure to the DNA probe, according to the formamide/SSC protocol of Rowley et al. (1990). ION MICROPROBE IMAGING OF HUMAN CHROMOSOMES 305 Fig. 4. (a): 81Br2 SIMS map of BrdU-labelled mitotic cell cultured in the presence of aphidicolin, shown in inverted contrast. The average number of SCEs/cell in this sample is 2.5 greater than for a normal culture. Sum of 9 single-pass maps, totaling 39.3 minutes. 3.7 3 105 counts, 47 µm full scale. (b): The same cell as in (a), imaged with 26CN2. Some structure, due to the condensation pattern of the chromatin, is detectable here. 3.0 3 106 counts from a single scan in 2.2 minutes. (c): Detail of another aphidicolin-treated cell, exhibiting 5 SCEs. 1.8 3 105 81Br2 counts acquired in 4 single-pass scans totaling 17 minutes. 17 µm full scale. (d): 26CN2 map corresponding to the area in (c), revealing a break (possibly due to a fragile site) in the q arm of the left chromatid of the uppermost left chromosome. This occurs at the location of an SCE, as shown in (c). 5.1 3 106 counts from a single scan in 2.2 minutes. Scanning Ion Microprobe An exploded schematic of the recently upgraded UC SIM, in the configuration employed in this investigation (magSIMS), is shown in Figure 1. It differs from that originally described (Levi-Setti et al., 1986) by its coupling with a high performance magnetic sector mass spectrometer (modified Finnigan MAT 90) for SIMS analysis. The primary column is unchanged, relative to previous descriptions, and for the present work was set to focus a 6 pA Ga1 ion beam, extracted from a liquid metal ion source, to a spot 35 nm wide. The ejected secondary ions are collected and energy-analyzed by a spherical 90° electrostatic sector. The secondary ion transport system has been entirely redesigned, and the overall collection and transmission efficiency in the magSIMS mode is now estimated at ,20%, at a mass resolution m/Dm , 550, independent of mass. Compared with the former RF quadrupole mass filter performance, where the transmission efficiency dropped by a factor of 5 with increasing mass, at a mass resolution m/Dm , 200, this represents an improvement in overall detection efficiency by factors of 20–100 over the periodic table of the elements (Chabala et al., 1995). To ensure even transmission of the SIMS transport system while the point of primary ion impact is scanned over the field of view, the secondary ion beam is synchronously descanned (dynamic emittance matching) so as to be constantly redirected toward the center of the entrance slit of the magnetic spectrometer. Image Acquisition and Resolution The SIMS signals are detected by an active film electron multiplier (ETP AF820, pulse counting) capable of accepting count rates up to 50 MHz. Images, containing 512 3 512 picture elements in a square raster from single scans, are stored in digital memory for subsequent analysis with a KONTRON IMCO image processing system. Mitotic cells are located by scanning the samples using the 81Br2 SIMS signal displayed on a CRT, at fields of view 160 µm wide. Analytical maps are then recorded from single pass scans at fields of view of width in the range of 20–50 µm, to encompass the entire mitotic cell. The 6 pA Ga1 probe erodes the chromosomes very slowly: We estimate that about 300 scans, 40 3 40 µm2 in size, at a dwell time of 1 µs/pixel, are needed to completely erode a set of BrdU-labelled chromosomes prepared by the standard preparation procedure above. Counting rates cannot exceed 256 counts/pixel, the limit imposed by the 8 bit image processor memory, to avoid image saturation. This limit is generally reached for the abundantly emitted CN2 in a single pass scan at a dwell time of 1 µs/pixel. However, for the 81Br2 label, the sum of 5–10 such single pass maps for a particular mitotic cell is needed (added together after registration of the images to compensate for occasional minor specimen drift) before the computer saturation limit is reached. Such stacks of sequential images effectively provide an in-depth view of elemental distributions (SIMS tomography) and may be used for 3D image 306 R. LEVI-SETTI ET AL. Fig. 5. (a) 81Br2 map of a BrdU-labelled mitotic cell stained with Giemsa. Here the Br content of eosin overshadows the label, and originates recognizeable Giemsa banding patterns of equal intensity for both sister chromatids. 9.0 3 106 counts from a single scan in 2.2 minutes. 40 µm full scale. (b) 26CN2 map of the area in (a). The bright halo surrounding the chromosomes is attributed to the thiazines residue from Giemsa. The apparent lack of emission from the chromosome interior is a photographic reproduction artifact; the signal intensity within chromatids in fact equals that of unstained chromosomes. 2 3 107 counts from a single scan in 2.2 minutes. (c) Another example of a BrdU-labelled mitotic cell stained with Giemsa. 7.7 3 106 81Br2 counts from 5 single-pass scans in 44 minutes. 47 µm full scale. (d) Magnified detail of (c) showing well defined SIMS bands. 2.1 3 106 counts, 25 µm full scale. reconstruction. To emphasize relevant features and to smooth the graininess of low signal regions of the images, gray levels may be minimally averaged using lowpass routines as needed, without affecting the image resolution. The relationship between probe size and image resolution as a function of signal count statistics and pixel spacing (related to the magnification and raster size) has been extensively studied (Levi-Setti et al., 1987). The image resolution is defined in terms of an edge spread function, measuring the sharpness of the boundary separating two regions of different signal density. Experimentally, this quantity is obtained from line scans across images of sharp boundaries. Only for signal density differences across an edge in excess of ,10 counts/pixel does the edge spread function approach the probe size. Here, this condition is usually satisfied for the CN2 images and for the sum of several 81Br2 maps, provided that the probe spacing in the raster be less than the probe diameter (for a raster of 512 3 512 and a probe diameter of 35 nm, the field of view must be 18 µm wide or less). At lower magnifications, the edge spread function approaches, instead, the spacing between pixels, which can be greater than the probe size. The images presented here were taken from areas 20 to 50 µm wide, leading to an estimate of the edge spread function, for high counts/pixel regions, in the range of ,40–100 nm. BrdU Labeling and Its Detection BrdU, incorporated in the chromosomes as a thymidine analog, is the carrier of the Br label that is detected by SIMS. Br is present in its two isotopic components 79Br and 81Br in their natural abundances, 50.54% and 49.46% respectively, and is predominantly emitted as a negative ion. As previously shown (Hallégot et al., 1989), PO32 represents a major interfering molecule at mass 79 (,39%), that could not be separated from 79Br2 at the limited mass-resolving power (Dm/m , 200) of the RF quadrupole mass filter used in previous studies. We have shown (Levi-Setti et al., 1994b) that the magSIMS upgraded version of the UC SIM can indeed separate the two species at mass 79 (at Dm/m , 3,400), at the expense however of a reduced SIMS system transmission. In view of the much improved detection sensitivity of magSIMS for 81Br2 (,1%) compared with that reported previously (0.01% from Levi-Setti and Le Beau, 1992), we opted as an unwarranted effort the use of the potential additional contribution at mass 79. As is well known, and reviewed in detail by Levi-Setti and Le Beau (1992), the relative amounts of label that are incorporated within sister chromatids from the BrdU-containing culture medium varies with cell cycle, because the label is only introduced in newly-synthesized DNA during chromosome replication, starting from unlabeled and one parental DNA. Thus at the first cell cycle, each chromatid contains one labeled and one parental (unlabeled DNA) strand: the sister chromatids carry the same quantity of label, so that the relative label amounts stand in the ratio 1/2:1/2 (taking as 1 the content of fully labelled DNA). It follows from combinatorial arithmetics that after two S phases, one chromatid will carry the label in both DNA strands, the other only in one, and the label ratio between sister chroma- ION MICROPROBE IMAGING OF HUMAN CHROMOSOMES 307 Fig. 6. (a) 81Br2 map of a BrdU-labelled mitotic cell after DNA denaturation with FISH protocol. 6.7 3 105 counts from 7 single-pass scans, 31 minutes, 36 µm full scale. (b) 26CN2 map of the area in (a), showing fine condensation structure not visible in undenatured samples. 2.4 3 106 counts from single scan in 2.2 minutes. (c) Detail of another denatured mitotic cell, with markers pointing at 3 SCEs. 1.9 3 105 counts, 7 single-pass scans in 31 minutes, 24 µm full scale. (d) 26CN2 map of the area in (c), showing coiling of the chromatid fibers. 4.0 3 106 counts from single scan, 4.4 minutes. tids will be 1:1/2. At the third cell division, ratios 1:1 and 1:1/2 should occur with equal probability. Because the SIMS signal is proportional to the concentration of the element being detected, it is possible to identify the number of divisions of a particular mitotic cell in the presence of BrdU from the measured ratio of Br label intensities between sister chromatids. Importantly, at the second division and beyond, where one chromatid may carry twice the amount of label as the other, it is possible to detect SCEs by simple inspection of the SIMS images. In the present experiment the cells were cultured for the average time (72 hours) necessary for 2 3-cell cycles to occur; therefore, examples of all 3 above occurrences are represented in the results to follow. It should be emphasized that the above predictions are valid, provided the entire label content of the chromosomes is recorded. This would require the total sputter erosion of the sample, a goal that is seldom attainable. Because one of our 40 3 40 µm2 SIMS scans probes only the outermost layers of the samples (,6 monolayers), the condensation form of the chromosomes and the spatial distribution of the proteins may affect the actual surface density of DNA detected by the probe. As a consequence, the ratios of label intensities between chromatids that we measure may exhibit substantial variation in individual scans, approaching the predicted values only after averaging over a large number of chromosome maps. RESULTS Several mitotic cells were located and analyzed for each of the sample categories. Representative examples of 81Br2 and 26CN2 SIMS maps are presented here for several of these categories. A few color-coded correlative maps are shown in the attached color plate. BrdU-Labeled Chromosomes A partial mitotic cell resulting from the first division is shown in Figure 2. Forty-three chromosomes are well separated. Figure 2a is a 81Br2 map where each detected count is displayed to form a bright image, reminiscent of fluorescent staining microscopy. Figure 2b is an inverted contrast version of the same, to facilitate comparison with Giemsa-stained samples seen by transmission in the optical microscope. We note first that the well-separated chromatids of each chromosome have similar average image density, as expected from equal incorporation of BrdU in the first cell division. This is quantitatively confirmed by measuring the average signal intensity in each chromatid, relative to that within the cell nucleus, taken as a calibration standard. This ratio is close to 1/2 of a similar ratio measured for the brighter chromatid of chromosomes from the second cell division (where both DNA strands are labeled), indicating that we observe here only one labeled strand. We also note that, although the label density is essentially continuous for most chromatids, SIMS bands are discernable, suggestive of Q-bands in Figure 2a, or G-bands in 2b, although perhaps not as sharply defined. These banding patterns are more apparent in the magnified detail of Figure 2c. Also to be noted is the homologous appearance of the SIMS bands for sister chromatids, as expected for homologous label distributions (identified here as A–T sequences). The 308 R. LEVI-SETTI ET AL. Fig. 7. Topographic images of mitotic cells, obtained with mass-unresolved secondary ions. (a) and (b) show 2 different condensation states of BrdU-labelled chromosomes, 37 and 39 µm full scale respectively. (c) Collapsed appearance of chromosomes after denaturation in 70% formamide 4 3 SSC, 70°C, following FISH protocol, 36 µm full scale. (d) Highly condensed and resolved chromatid appearance after Giemsa staining, 47 µm full scale. overall chromosome morphology revealed by these maps, together with the SIMS banding patterns, identifies most of the chromosomes at first sight. Finally, Figure 2d shows a 26CN2 map, corresponding to the detail of Figure 2(c). Here the structure present in the 81Br2 maps is generally absent, the images of sister chromatids often merge together. This masking must be attributed to the remainder of base pairs and proteins convoluted with the A–T pairs. The brighter border of the chromosome images is attributable to edge effects (artifacts of the SIMS technique), not present in the label images, also suggesting a protein coating of the DNA. Figure 3 shows 81Br2 SIMS maps of a partial metaphase cell from the second cell division, in both normal and inverted contrast. Here the differential label content of sister chromatids is clearly visible, and measured, on the average, to be in the expected ratio 1:1/2. This leads to the detectability, by inspection, of several examples of SCEs, pointed out by markers. The image intensity for the chromatids with only one labeled DNA strand, comparable to that of the chromatids in Figure 2, appears smaller than the latter due to photographic reproduction constraints. The SIMS banding patterns evident in the singly-labeled chromatids of Figure 2 are replaced in the doubly labeled chromatids of Figure 3 by an essentially continuous knotted structure, suggestive in some details of the coiled structure of the chromatid fiber of metaphase chromosomes discussed by, e.g., Gasser et al. (1986). The average rate of occurrence of SCEs, measured from several mitotic cells belonging to the same cell culture from an indi- vidual healthy donor (control), is approximately 4.3/ cell, or about 9.4% of the chromosomes. This figure does not include apparent SCEs occurring at the centromere (often included in the count), which cannot be distinguished from a simple twist in the chromosome arms. BrdU Labeling Plus Aphidicolin Treatment Exposure to aphidicolin (Rassool et al., 1992) of cultures from the same peripheral blood sample as the control results in a marked increase in the average rate of occurrence of SCEs: about 10.3/cell, or 23% of the chromosomes, ,2.5 times that for the control. Examples of these mitotic cells are shown in Figure 4. The 81Br2 map of the metaphase cell of Figure 4a exhibits 5 SCEs and 7 centromeric twists. The identification of these features is facilitated by comparison with the CN2 map of Figure 4b, which characterizes the chromosome appearances and allows a precise localization of the centromeres. A higher magnification detail of another mitotic cell is shown in Figure 4c and d. Here 5 SCEs and 1 twist can be identified, the uppermost left SCE occurring at the site of a break in the left chromatid (a possible fragile site induced by aphidicolin) visible in the CN2 map. BrdU Labeling Plus Giemsa The Br content of eosin (C20H6Br4Na2OS) in Giemsa (a mixture of eosin with the thiazines methylene blue and azure B) was found to overshadow (by at least a factor of 10) the BrdU label, masking the differentiation in label content of sister chromatids. Nonetheless it is ION MICROPROBE IMAGING OF HUMAN CHROMOSOMES 309 Fig. 8. Comparison of optical microscope images of trypsin-Giemsa stained mitotic cells, with base-specific SIMS maps from BrdU-labelled cells. (a) and (b) optical microscope, 50 and 33 µm full scale respectively. (c) 81Br2 SIMS map, second division, sum of 16 single pass maps recorded for SIMS tomography, 5.6 3 105 counts, 70 minutes, 39.1 µm full scale. (d): 81Br2 SIMS map, third division, sum of 9 single-pass scans, 6.7 3 105 counts, 39.3 minutes, 32 µm full scale. still useful to investigate the chemical changes associated with Giemsa staining using SIMS. The 81Br2 maps of 2 Giemsa-stained chromosome spreads are shown in Figure 5a and c. Figure 5b is a CN2 map corresponding to (a), and Figure 5d a magnified detail of 5c. Several features stand out in these SIMS maps. First, the apparent absence of CN2 emission from the chromosomes in 5b is a visual artefact of the photoreproduction. In fact, the measured CN2 signal intensity within the chromosomes corresponds closely to that measured for e.g., Figures 2d or 4d in the absence of Giemsa staining, although the chromatids appear better separated and closely mimic the 81Br2 images in the present case. This comparison to non-stained samples suggests that some of the proteins not closely packed with DNA may have been removed by staining, and that the original histone proteins-DNA complex may either have been left untouched, or partially replaced by the nitrogen-carrying components of Giemsa (thiazines). Second, CN2 originates from a bright halo surrounding the chromosomes, with signal intensity about double that of the chromosome interior. This halo can be attributed to the residues of thiazines (eosin contains no nitrogen) left by capillarity after in situ Giemsa staining. More importantly, the Br2 distribution, in this case representing the eosin distribution, shows transverse bands, not surprisingly similar to the Giemsa patterns. We shall return to these observations, related to the banding mechanism of optical stains, in the discussion section. BrdU Labeling Plus DNA Denaturation Two examples of mitotic cells subjected to the FISH procedure (without DNA probe) are shown in Figure 6. Both originate from the third cell division, because of the presence of chromosomes exhibiting 1:1 and 1:1/2 label ratios between sister chromatids in the 81Br2 SIMS maps (Figs. 6a and c). Two SCEs are marked by pointers in Figure 6c. The label intensities within the chromatids are comparable to those measured for nondenaturated samples. Of particular interest are the corresponding CN2 maps of Figure 6b, and especially that of Figure 6d. In contrast to the rather uniform maps for CN2 previously shown for non-denaturated samples (see Fig. 2d and 4d), these maps exhibit remarkable structure and evidence of a coiling arrangement of the chromatid fiber. Perhaps unexpectedly, the CN2 signal within the chromatids is about twice as intense as that for non-denaturated chromosomes, seemingly in contradiction with the evidence that much of the protein coating, normally widening the slender structure of the coiled chromatin, has been removed by the denaturation process. We propose an explanation for this apparent discrepancy by referring to the condensation state of the chromosomes, revealed by their topography. This is shown in Figure 7, where Figures 7a and b show the appearance of two different presentations of mitotic cells among our samples, prior to any further in situ procedure; Figure 7c shows the result of denaturation, and Figure 7d that of Giemsa staining. The denatured chromosomes in Figure 7c appear flattened with a thin raised rim, suggestive of the collapsed chromosome state reported (Gormley and Ross, 1972; Sumner, 1990) following treatment with hot 23 SSC, part of the ASG (acetic/saline/Giemsa) staining method of Sumner et al. (1971). We can surmise that the intensity of the SIMS signal is a function of the aspect Figure 9. ION MICROPROBE IMAGING OF HUMAN CHROMOSOMES ratio of the chromatin exposed to the primary probe. A collapsed state would then favor a better secondary ion yield than a more compacted normal state, or a contracted state such as that indicated as a result of Giemsa staining in Figure 7d. DISCUSSION The images presented in the previous section are illustrative of the direct information that high resolution SIMS imaging can provide concerning chromosomal structure and composition. In conjunction with BrdU labeling, the method enables the detection of SCEs at the base-specific level in images of comparable contrast and resolution to those obtained with fluorescent dye techniques. Depth compositional information from sequential SIMS mapping (although not exploited in this presentation) is also available to explore the base-specific 3D structure of the chromosomes. In conjunction with the label maps, the corresponding images taken with the CN2 secondary ions provide complementary information about the overall chromosome morphology that may occasionally be valuable. Thus, in one example shown (Figures 4c and d), this comparison has made it possible to detect a break in one chromatid (possibly due to a fragile site), coincident with the occurrence of an SCE. Furthermore, the difference between these two sets of images, representing respectively the DNA core and the overall DNA-protein complex, reveals the level of protein digestion due to the fixation process and subsequent in situ procedures. At the same time, images of the surface topography indicate the condensation state of the chromosomes under investigation. Before discussing our observations that may be relevant to the mechanism of chromosome banding, it is Fig. 9. (a) 81Br2 SIMS map of portion of a BrdU-labelled mitotic cell. Sister chromatids are equilabeled, indicating first cell division where only one DNA strand of each chromatid is labeled. The image is shown on a pseudocolor scale which emphasizes a banded structure due to base-specific (A-T) condensation (blue:zero signal; yellow–orange: maximal signal). A coiled structure of the chromatin fibers is also apparent in several chromosomes. (b) 81Br2 SIMS map of a BrdUlabelled mitotic cell, where the label intensity ratio between sister chromatids averages close to the expected 1:1/2 ratio for second cell division. Several occurrences of sister chromatid exchange (SCE) can be detected in this micrograph. (c) False color superposition of a 81Br2 SIMS map (A-T pair distribution) of a BrdU-labelled mitotic cell (green), with the corresponding 26CN2 map (blue). The latter includes the overall contribution from DNA, RNA and proteins. This correlative presentation emphasizes the occurrence of SCEs. (d) Portion of BrdU-labelled mitotic cell cultured in the presence of the chemical aphidicolin. The average rate of occurrence of SCEs in this sample is 2.5 times that of a normal culture. The image is a color-coded superposition of 81Br2 (green) and 26CN2 (blue) SIMS maps. Four SCEs are pointed out by arrows. In this example, the 26CN2 distribution extends beyond the chromatid outline, possibly due to residues of nuclear material. (e) Portion of another BrdU-labelled mitotic cell cultured in the presence of the chemical aphidicolin. 81Br2 is color coded in white, 26CN2 in blue. Here the latter map helps identify a break in the p-arm of the left chromatid in the chromosome located in the NW corner. This break occurs in conjunction with an SCE and is probably due to a fragile site. Several additional examples of SCEs are clearly identifiable in this correlative image. (f ) Correlative superposition of 81Br2 (white) and 26CN2 (blue) SIMS maps for a BrdU-labeled mitotic cell stained with Giemsa. Here the Br content of eosin overshadows the label, and originates recognizeable Giemsa banding patterns of equal intensity for both sister chromatids. The bright halo surrounding the chromosomes is attributed to the thiazine residue from Giemsa. 311 appropriate to compare in Figure 8 the optical micrographs of trypsin-Giemsa stained chromosomes (Figure 8a and b), with base-specific SIMS maps of BrdUlabeled chromosomes (Figure 8c and d). As previously noted in regard to Figure 3, fully-labeled chromatids show an essentially continuous label (or A-T pair) distribution, with only occasional suggestion of banding, never differentiated at a level comparable to the Giemsa bands. Even after denaturation, the correlation between SIMS and Giemsa banding (see Figure 8) for fully-labeled chromatids is at best rather weak. Only in monofilarly labeled chromosomes, such as those in Figure 2, banding suggestive of the Giemsa patterns, that can be associated with possible regions of A-T enrichment are discernable. These observations militate against attributing Giemsa banding exclusively to patterns of base-specific condensation, at least in compacted metaphase chromosomes such as those examined in the present survey. The pattern of overall chromosomal condensation becomes visible instead in the CN2 maps of denatured chromosomes. Again however, this pattern, although structured, appears not selectively differentiated. In view of the Br content of eosin, the detection of Giemsa-like bands in the Br2 maps of Giemsa-stained chromosomes, at an intensity level that is higher than that provided by BrdU labeling, simply reinforces the notion that eosin is associated with G-positive bands. By and large, these observations are consistent with the discussion of G-banding summarized by Sumner (1990). Hints of the possible role of divalent cations are found in the literature and are also referred to by this author. We are presently investigating the distribution of Ca and Mg in our samples. It should be noted that our observations on denaturated chromosomes represent a successful preliminary to the localization of specific labelled probes, to develop, in analogy with FISH, a methodology that should be called SISH (for SIMS in situ hybridization). This goal seems attainable using, e.g., BrdU-labelled probes (with chromosomes not carrying the same label), or probes labelled with other thymidine analogs such as iododeoxyuridine or fluorodeoxyuridine. By a similar approach, using stable isotope labels such as, e.g., deuterium it seems feasible to aim at the localization of regulatory proteins, in conjunction with SISH, for specific gene sequences. Further investigations along these lines, with the UC SIM, are being formulated at this time. ACKNOWLEDGMENTS This report is based on work supported by the National Science Foundation under grant BIR 9317959. We wish to thank Dr. Pamela L. Strissel for informative discussions. REFERENCES Chabala, J.M., and Levi-Setti, R. (1987) High-resolution chemical mapping of biological material by use of a heavy ion probe. In: Analytical Electron Microscopy—1987. D.C. Joy, ed. San Francisco Press, San Francisco, CA, USA, pp. 203–207. Chabala, J.M., Soni, K.K., Li, J., Gavrilov, K.L., and Levi-Setti, R. (1995) High-resolution chemical imaging with scanning ion probe SIMS. Int. J. Mass Spectr. Ion Processes, 143:191–212. Gasser, S.M., Laroche, T., Falquet, J., Boy de la Tour, E., and Laemmli, U.K. (1986) Metaphase chromosome structure. Involvement of topoisomerase II. J. Mol. Biol., 188:613–629. 312 R. LEVI-SETTI ET AL. Gormley, I.P., and Ross, A. (1972) Surface topography of human chromosomes examined at each stage during ASG banding procedure. Exp. Cell Res., 74:585–587. Hallégot, P., Girod, C., Le Beau, M.M., and Levi-Setti, R. (1989) High spatial resolution SIMS imaging of labelled human chromosomes. In: Secondary Ion Mass Spectrometry-SIMS VII. A. Benninghoven, C.A. Evans, K.D. McKeegan, and H.A. Storms, eds. John Wiley & Sons, New York, NY, USA, pp. 327–330. Levi-Setti, R., Chabala, J., and Wang, Y.L. (1987) Micro-secondary ion mass spectrometry: Physical and instrumental factors affecting image resolution. Scanning Microsc. Suppl., 1:13–22. Levi-Setti, R. (1988) Structural and microanalytical imaging of biological materials with heavy-ion probes. Ann. Rev. Biophys. Biophys. Chem., 17:325–347. Levi-Setti, R., and Le Beau, M.M. (1992) Cytogenetic applications of high resolution secondary ion imaging microanalysis: detection and mapping of tracer isotopes in human chromosomes. Biol. Cell, 74:51–58. Levi-Setti, R., Chabala, J.M., and Smolik, S. (1994a) Nucleotide and protein distribution in BrdU-labelled polytene chromosomes revealed by ion probe mass spectrometry. J. Microsc., 175:44–53. Levi-Setti, R., Chabala, J.M., Gavrilov, K., Li, J., Soni, K.K., and Mogilevsky, R. (1994b) Imaging with a high-resolution SIM. In: Secondary Ion Mass Spectrometry-SIMS IX. A. Benninghoven, Y. Nihei, R. Shimizu, and H.W. Werner, eds. John Wiley & Sons, Chichester, England, pp. 233–237. Levi-Setti, R., Chabala, J.M., Gavrilov, K., Espinosa III, R., and Le Beau, M.M. (1995) Advances in imaging SIMS studies of BrdUlabelled human metaphase chromosomes. In: Secondary Ion Mass Spectrometry-SIMS X. A. Benninghoven, ed. John Wiley & Sons, Chichester, England, in press. Rassool, F.V., Le Beau, M.M., Neilly, M.E., van Melle, E., Espinosa III, R., and McKeithan, T.W. (1992) Increased genetic instability of the common fragile site at 3p14 after integration of exogenous DNA. Am. J. Hum. Genet., 50:1243–1251. Rowley, J.D., Diaz, M.O., Espinosa III, R., Patel, Y.D., van Melle, E., Ziemin, S., Taillon-Miller, P., Lichter, P. Evans, G.A., Kersey, J.H., Ward, D.C., Domer, P.H., and Le Beau, M.M. (1990) Mapping chromosome band 11q23 in human acute leukemia with biotinylated probes: Identification of 11q23 translocation breakpoints with a yeast artificial chromosome. Proc. Nat. Acad. Sci. USA, 87:9358– 9362. Sumner, A.T. (1990) Chromosome banding. Unwin Hyman, London, England, p. 92. Sumner, A.T., Evans, H.J., and Buckland, R.A. (1971) New technique for distinguishing between human chromosomes. Nature New Biol., 232:31–32.