Anal. Chem. 2007, 79, 2199-2206
Articles
TOF-SIMS 3D Biomolecular Imaging of Xenopus laevis Oocytes Using Buckminsterfullerene (C60) Primary Ions John S. Fletcher,* Nicholas P. Lockyer, Seetharaman Vaidyanathan, and John C. Vickerman
Manchester Interdisciplinary Biocentre, Centre for Instrumentation and Analytical Science, School of Chemical Engineering and Analytical Science, The University of Manchester, Manchester, M1 7ND, UK
Time-of-flight secondary ion mass spectrometry (TOFSIMS) using buckminsterfullerene (C60) as the primary ion source has the ability to generate chemical images of surfaces with high sensitivities and minimal chemical damage. We studied the application of C60+ to depth profile a biological cell surface in a controlled manner and to subsequently image the revealed subsurfaces, in order to generate three-dimensional molecular images of the biological system. Such an analytical tool not only enables the surface localization of molecular species to be mapped but also enables the biomolecular distribution as a function of depth to be investigated with minimal sample preparation/intervention. Here we demonstrate the technique with a freeze-dried Xenopus laevis oocyte, which is a single cell. A C60+ ion beam was used with computercontrolled analyses and etch cycles. Mass spectra derived from the surface revealed peaks corresponding to cholesterol (m/z 369) and other lipids at m/z 540-570 and 800-1000, in the positive ion mode, and lipid fatty acid side chains (e.g., m/z 255) in the negative ion mode. To our knowledge, this is the first demonstration of the 3D biomolecular imaging within an actual biological system using TOF-SIMS. The ability to map the distribution and localization of biomolecular species in biological systems is of considerable interest in biology and medicine. Time-of-flight secondary ion mass spectrometry (TOF-SIMS) has been successfully employed for mass spectrometric imaging of biological samples with relatively high lateral resolution.1-7 Liquid metal ion beams can be routinely * To whom correspondence should be addressed. E-mail: John.Fletcher@ manchester.ac.uk. (1) Gazi, E.; Dwyer, J.; Lockyer, N.; Gardner, P.; Vickerman, J. C.; Miyan, J.; Hart, C. A.; Brown, M.; Shanks, J. H.; Clarke, N. Faraday Discuss. 2004, 126, 41-59. (2) Lockyer, N. P.; Vickerman, J. C. Appl. Surf. Sci. 2004, 231-2, 377-384. (3) Ostrowski, S. G.; Van, Bell, C. T.; Winograd, N.; Ewing, A. G. Science 2004, 305, 71-73. (4) Sjovall, P.; Lausmaa, J.; Johansson, B. Anal. Chem. 2004, 76, 4271-4278. (5) Chandra, S. Eur. J. Cell Biol. 2005, 84, 783-797. (6) Nygren, H.; Borner, K.; Hagenhoff, B.; Malmberg, P.; Mansson, J. E. Biochim. Biophys. Acta: Mol. Cell Biol. Lipids 2005, 1737, 102-110. (7) Winograd, N. Anal. Chem. 2005, 77, 142a-149a. 10.1021/ac061370u CCC: $37.00 Published on Web 02/16/2007
© 2007 American Chemical Society
focused to spot sizes below 500 nm, and the introduction of cluster projectiles such as Aun+ and Bin+ has increased the ability to eject and detect intact biomolecules. Popular test samples for biological SIMS imaging are sections of mouse or rat brain where analysis with Au3+ (and more recently Bin+) can provide maps of the distribution of various lipids and other biomolecules such as vitamin E with m/z values up to and in some cases beyond 1000.4,8 MALDI imaging of tissue sections following the SIMS analysis has also been reported with higher lateral resolution SIMS images of smaller molecules being combined with MALDI images mapping distribution of protein in the sample and demonstrating the complimentary nature of the two techniques.9 SIMS imaging of small individual cells is much more challenging although it has been successfully demonstrated in some instances. For example, Ostrowski et al. have demonstrated changes in lipid distribution at the point of contact between mating tetrahymena cells.3 In order to aid the detection of molecular species, methods such as cationization by the addition of a metal such as silver have been used and such analysis has yielded images of the distribution of phospholipid and cholesterol in single cells.10 Enhancement of the SIMS signal by the addition of a matrix has also been shown to generate increased secondary ion signals from higher mass species, particularly when combined with bombardment using polyatomic ion beams. However, such methods add an extra level of complexity to the sample preparation, removing the sample from its natural state, and may limit the ultimate spatial resolution achievable in the image.11 Although successful in terms of lateral resolution, the use of liquid metal ion sources limits the detection of intact biomolecular species to the upper most layer(s) of the analyte. However, there is a strong desire also to investigate molecular distribution as a function of depth in biological cell and tissue specimens. It is the realization of such ambitions that have provided some of the (8) Touboul, D.; Kollmer, F.; Niehuis, E.; Brunelle, A.; Laprevote, O. J. Am. Soc. Mass Spectrom. 2005, 16, 1608-1618. (9) Brunelle, A.; Touboul, D.; Laprevote, O. J. Mass Spectrom. 2005, 40, 985999. (10) Sjovall, P.; Lausmaa, J.; Nygren, H.; Carlsson, L.; Malmberg, P. Anal. Chem. 2003, 75, 3429-3434. (11) McDonnell, L. A.; Heeren, R. M.; de Lange, R. P.; Fletcher, I. W. J. Am. Soc. Mass Spectrom. 2006, 17, 1195-1202.
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007 2199
impetus in the development of a range of polyatomic primary ion sources. Traditionally, analysis with conventional monatomic ion beams has only been capable of yielding molecular information from the upper surface of the analyte. This is due to an inherent destructive feature of the technique. The sample is sputtered by bombardment with a high (typically keV) primary ion beam. This generates a mixture of predominantly neutral species accompanied by a small portion (∼1%) of secondary ions. It is these ions that are then extracted and analyzed using TOFMS. As the primary ion impinges on a surface, it will deposit its energy not only on the surface but, to varying degrees, in the subsurface as well. This transfer of energy to the subsurface region of the sample can result in the breaking of chemical bonds. Thus, upon continued bombardment of the same area of the sample, the secondary ion yield of the more chemically and biologically characteristic molecular ions is severely reduced (in many cases to zero) and is accompanied by a corresponding tendency to generate more low-mass hydrocarbon and elemental signal. The rate of disappearance of a molecular signal due to damage accumulation (with respect to primary ion dose) is described by the damage cross section.12,13 Analysis in the regime of low primary ion dose where the data are considered to be indicative of the pristine surface is termed static SIMS. Analyses using high primary ion doses to map elemental species as a function of depth as the sample is etched away during the sputtering process are common place in the semiconductor field and such experiments are generally referred to as dynamic SIMS. Initial experiments using SF5+ primary ions to analyze polymers demonstrated that the generation of significant increases in secondary ion yield was accompanied by a reduction in the rate of damage accumulation compared to conventional monatomic primary ion bombardment.14 The ratio of the secondary ion yield to the damage cross section can be used to describe the efficiency of the primary particle. In some cases, particularly with materials that have high sputter yields, molecular depth profiling with SF5+ has been demonstrated with little if any chemical damage evidenced in the SIMS spectrum.15-17 In particular, impressive results have been demonstrated of the imaging of pharmaceuticals in a poly(lactic acid) film.18 The most efficient ion beam currently available for routine analysis is C60+. Initial experiments using such ion beam systems demonstrated substantial increases in secondary ion yield and efficiency.19-22 A number of recent publications have detailed analyses of a range of test systems alluding to the possibility of (12) Nguyen, T. C.; Ward, D. W.; Townes, J. A.; White, A. K.; Krantzman, K. D.; Garrison, B. J. J. Phys. Chem. B 2000, 104, 8221-8228. (13) Benninghoven, A. Angew. Chem., Int. Ed. Engl. 1994, 33, 1023-1043. (14) Kotter, F.; Benninghoven, A. Appl. Surf. Sci. 1998, 133, 47-57. (15) Wagner, M. S.; Castner, D. G. Appl. Surf. Sci. 2004, 231-2, 366-376. (16) Mahoney, C. M.; Roberson, S.; Gillen, G. Appl. Surf. Sci. 2004, 231-2, 174178. (17) Mahoney, C. M.; Roberson, S. V.; Gillen, G. Anal. Chem. 2004, 76, 31993207. (18) Gillen, G.; Fahey, A.; Wagner, M. S.; Mahoney, C. M. Appl. Surf. Sci. 2006, 252 (19), 6537-6541. (19) Weibel, D.; Wong, S.; Lockyer, N.; Blenkinsopp, P.; Hill, R.; Vickerman, J. C. Anal. Chem. 2003, 75, 1754-1764. (20) Wong, S. C. C.; Hill, R.; Blenkinsopp, P.; Lockyer, N. P.; Weibel, D. E.; Vickerman, J. C. Appl. Surf. Sci. 2003, 203, 219-222. (21) Weibel, D. E.; Lockyer, N.; Vickerman, J. C. Appl. Surf. Sci. 2004, 231-2, 146-152.
2200
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
3D molecular analysis of biocellular systems. In particular, low damage depth profiling has been demonstrated on LangmuirBlodgett multilayer films,23 biopolymer films,24 amino acids in ice,25,26 and small peptides in trehalose.27 Biological species such as phospholipids and cholesterol have also been shown to be stable to depth profiling. Following the formation of an altered layer, a steady state is reached where, in a pure sample, the signal intensity is stable. Thus, in more complex multicomponent systems such as cells and tissues, variation of intensity with depth is expected to represent changes in chemistry/concentration.28 In this report, we present a first account of 3D molecular imaging of a biological cell, using TOF-SIMS. To date, this important breakthrough in mass spectral capability has not been described in the literature. We discuss the initial application of TOF-SIMS using C60+ ions to sputter etch and image Xenopus laevis oocyte, thus characterizing chemical changes in successive layers of a biological cell with minimal sample preparation/ intervention. X. laevis oocyte is a well-established cell model that has been extensively used in many branches of experimental biology and pharmacological research. It is a large single cell about 0.8-1.3 mm in diameter, whose cellular compartments are so large that they are accessible to cell biologists for manipulation and visualization. The oocyte nucleus, called the germinal vesicle, is large enough to be prised out intact and experimentally examined, relatively easily. In addition, X. laevis oocyte is remarkably resistant to osmotic challenges.29 This facilitates preparation cells for mass spectral imaging as they are reasonably resistant to washing of exogenous substances using deionized water. Such considerations motivated our choice of experimental system for this investigation. EXPERIMENTAL SECTION Biological Sample Preparation. Xenopus oocytes were kindly donated by Prof. Mark Boyett (The University of Manchester). The oocyte pool, maintained in Barth’s solution, contained oocytes from different stages. Stage IV-VI oocytes were selected, rapidly washed (less than 1 min) in MilliQ water using Pasteur pipettes, gently placed on silicon wafers (5 × 5 mm), and allowed to dry slightly in air. The oocyte was then plunge frozen in liquid nitrogen-cooled liquid propane and stored under liquid nitrogen until freeze-drying. The cryofixed oocyte was freeze-dried (10-3 mbar) overnight and transferred onto conducting copper tape prior to TOF-SIMS analysis at room temperature. TOF-SIMS. TOF-SIMS analysis was performed using a BioToFSIMS instrument, the design of which has been described (22) Wong, S. C. C.; Lockyer, N. P.; Vickerman, J. C. Surf. Interface Anal. 2005, 37, 721-730. (23) Sostarecz, A. G.; McQuaw, C. M.; Wucher, A.; Winograd, N. Anal. Chem. 2004, 76, 6651-6658. (24) Fletcher, J. S.; Conlan, X. A.; Jones, E. A.; Biddulph, G.; Lockyer, N. P.; Vickerman, J. C. Anal. Chem. 2006, 78, 1827-1831. (25) Wucher, A.; Sun, S. X.; Szakal, C.; Winograd, N. Anal. Chem. 2004, 76, 7234-7242. (26) Conlan, X. A.; Lockyer, N. P.; Vickerman, J. C. Rapid Commun. Mass Spectrom. 2006, 20, 1327-1334. (27) Cheng, J.; Winograd, N. Anal. Chem. 2005, 77, 3651-3659. (28) Jones, E. A.; Lockyer, N.; Vickerman, J. C. Int. J. Mass Spectrom. 2007, 260, 146-157. (29) Kelly, S. M.; Butler, J. P.; Macklem, P. T. Comp. Biochem. Physiol. A 1995, 111, 681-691.
Figure 1. Schematic of the formation of the 3D data set by the stacking of sequential 2D images from the depth profile and an illustration of the type of figure output from Matlab with slices removed for visualization of the interior of the sample.
elsewhere.30 The instrument is currently equipped with a wien filtered 40 kV C60 ion gun (Ionoptika Ltd.) and a wien filtered 25 kV LMIG fitted with a Au:Ge eutectic source (Ionoptika Ltd.) providing Au+ and Au3+. The sample stage can be cooled to a base temperature of ∼100 K for the analysis of cryogenic samples, and the preparation chamber is fitted with an in vacuo freezefracture system for frozen-hydrated samples (Kore Technology Ltd., Cambridge, UK). The sample was held at ground potential during ion impact. Secondary ions were directed into a two-stage reflectron TOF-MS by applying a delayed extraction pulse of 2.5 kV. The field of view of the mass analyzer is particularly large for this type of instrument (∼2 mm diameter) permitting the analysis of relatively large samples without moving the sample stage or scanning the secondary ion optics. Ions were postaccelerated to 20 keV and detected using a dual microchannel plate assembly. Flight times are recorded on a 1-ns time-to-digital converter (Fast Comtec GmbH). Charge compensation during analysis is performed by pulsing low-energy (25 eV) electrons onto the sample between successive primary ion pulses. Computercontrolled analysis and etch cycles are facilitated through the acquisition software allowing unattended operation of the instrument. The field of view for etching and analysis can be independently set to avoid crater-edge effects. During etching of insulating samples using a dc primary ion beam, a dc low-energy (25 eV) electron beam is used to offset sample charging. Although it has been shown that even low-energy electrons can result in chemical damage at high flux density,31 under the methodology described here, these affects are minimized by high sputter rates under dc ion bombardment. Three-Dimensional Imaging. A single freeze-dried oocyte cell was subjected to controlled C60+ ion beam etching and TOF-SIMS imaging to visualize biomolecular distribution on and within the cell. The entire cell (∼1 mm in size) was contained within a single raster field of the 40 keV C60+ beam. The C60+ beam was loosely focused to match the image pixel size of 5 µm, providing a dc beam current of 0.5 nA. (30) Braun, R. M.; Blenkinsopp, P.; Mullock, S. J.; Corlett, C.; Willey, K. F.; Vickerman, J. C.; Winograd, N. Rapid Commun. Mass Spectrom. 1998, 12, 1246-+. (31) Gilmore, I. S.; Seah, M. P. Appl. Surf. Sci. 2002, 187, 89-100.
2D SIMS images were collected by scanning the ion beam over the sample and acquiring a mass spectrum at each pixel. By summing a number of these frames, the amount of signal is increased and also compensates for any small fluctuations in the sampling such as ion beam current. All the 2D images were acquired using the same number of frames and hence the same primary ion beam dose. For improved stability, and ease of operation, the image/etch cycles were computer controlled and automated to allow the analysis to continue over night. 3D images were generated by stacking the 2D data generated at each layer of the depth profile as is illustrated in Figure 1. The data were stretched in the z (depth) direction to create a data cube. Individual peaks, groups of peaks, or m/z ranges were selected and integrated to generate a matrix of intensities at each pixel in the 3D array. This matrix was then plotted using Matlab (The MathWorks Inc.). Prior to plotting, the data were smoothed (three dimensionally), this reduces the spatial resolution slightly and reduces the maximum number of counts associated with any given point. However, the average counts/pixel is increased, allowing easier visualization. These count values were scaled to the maximum per pixel and mapped onto a “Jet” color scale. The number of counts to which each color corresponds in any given figure is indicated alongside the intensity scale bar. In order to visualize the change in signal intensity within the volume, slices were selected through regions of interest, also in Figure 1. Although it was impossible to ascertain the exact amount of material eroded during the experiment, an approximation can be made by optical microscopy. The oocyte was ∼1 mm in diameter, and a comparison of the micrograph of the oocyte prior to and following the experiment indicates an erosion depth of ∼100 µm. This estimation was used to generate a depth scale for the 3D maps based on etch dose. Thus, the z axes are labeled in nanometers starting at the bottom of the etch crater. It is possible however to calculate the sputter rate of our ion beam through a flat sample of cholesterol coated onto silicon from solution in chloroform. A number of etch craters were created and the dimensions of the craters measured using an AFM (Nanopics 2100, Seiko Instruments). We calculate that the number of cholesterol molecules removed per impacting 40 keV C60+ is in Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
2201
Figure 2. Optical micrograph of oocyte cell mounted on copper tape for TOF-SIMS analysis (a) before ion beam etching and (b) following 40 keV C60+ etching. TOF-SIMS spectra of X. laevis oocyte (integrated over the scanned area), showing peaks above m/z 200 in the (c) positive and (d) negative ion modes, after a primary ion dose of 1 × 1015 ions/cm2. (e) and (f) are representative secondary ion images associated with m/z 369 (cholesterol [M - H2O + H]+) and 281 (oleic acid [M - H]-), respectively.
the region of 180 and extrapolating to the etch dose used in the oocyte depth profile gives a predicted erosion depth of 175 µm. This is greater than that estimated by microscopy, but the difference may be explained by the change in incidence angle. In the oocyte analysis, the ion beam impact was closer to the normal of the sample face compared to an incline of 50° on the flat sample (see Figure 2a and b). The sputter rate would be expected to be much lower at normal incidence. For less topographically demanding samples, this should provide a useful first approximation for sputter rate. A further consideration regarding depth estimation of signals in the sample concerns possible preferential sputtering of different regions of the sample based on the physical properties of that area and also the angle of the surface relative to the incident ion beam. Hence, although all areas of the sample experienced the same primary ion dose, the extent of erosion may vary. This phenomenon is well established in dynamic SIMS of inorganic samples, but has yet to be fully explored in organic systems. RESULTS AND DISCUSSION Figure 2 contains optical (microscopic) images of the freezedried oocyte in the vacuum chamber as introduced (a) and following the C60 experiment (b). It is clear that the freeze-drying process has not maintained the morphology of the sample (originally spherical) and has resulted in the collapse of the cell wall on the upper left of the sample. The ion beam impacted the sample from the left at an incline angle of 50°. TOF-SIMS analysis of the oocyte provides secondary ions signals of significant intensity from species with m/z values up to ∼1000. Further, under 40 keV C60+ bombardment, these species persist in the mass spectrum following primary ion dose densities above 1 × 1015 ions/cm2. Distinct m/z regions of interest are 2202 Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
evident in the mass spectrum with groups of peaks occurring at m/z 540-700 and a similar set of peaks between m/z 800 and 1000. The intense feature visible in the positive ion spectrum is attributed to the cholesterol [M + H - H2O]+ ion at m/z 369. The m/z 540-700 envelop contains peaks at m/z 548, 574, and 576 that can be attributed to phosphatidylcholines with fatty acid side chains of composition C16:0, C18:2, and C18:1, respectively, without the phosphocholine head group. These assignments are corroborated by the detection of the corresponding fatty acid peaks in the negative ion spectra, i.e., peaks at m/z 255, 279, and 281 for fatty acid side chains of composition C16:0, C18:2, and C18:1, respectively. The lipid composition of X. laevis oocyte membrane has a significant proportion of phosphatidylcholines, which in combination with sphingomyelins constitute the major membrane lipid components of the light membrane fraction.32 Phosphatidylcholines are also the major phospholipids in whole oocytes.33 The peaks between m/z 800 and 1000 are present in both the positive and negative ion spectra. The difference between the peaks (m/z 26) suggests that they are also of lipid origin of different fatty acid compositions. It is possible that these originate from glycosphingolipids that are known membrane constituents, especially in microdomains termed caveolae.34 The 3D images for phosphocholine peaks (combined information from m/z 58, 86, 166, and 184), cholesterol (m/z 369), and the envelopes at m/z 540-650 and 815-960 are shown in Figure 3. The images are cut along two dimensions to display the changes in the third dimension (depth) so that changes in the localization (32) Hill, W. G.; Southern, N. M.; MacIver, B.; Potter, E.; Apodaca, G.; Smith, C. P.; Zeidel, M. L. Am. J. Physiol. Renal Physiol. 2005, 289, F217-224. (33) Stith, B. J.; Hall, J.; Ayres, P.; Waggoner, L.; Moore, J. D.; Shaw, W. A. J. Lipid Res. 2000, 41, 1448-1454. (34) Sadler, S. E. J. Cell. Biochem. 2001, 83, 21-32.
Figure 3. 3D biochemical images of freeze-dried oocyte, showing changes in (a) phosphocholine peaks m/z 58, 86, 166, and 184, (b) signal summed over the m/z range 540-650, (c) signal summed over the m/z range 815-960, and (d) cholesterol peak at m/z 369. Color scale normalized for total counts per pixel for each variable (m/z range).
of the imaged species can be seen at different depths (etched surfaces). It must be noted that the depth dimension (in nm) is stretched relative to the other two dimensions (in µm) for ease of visualization. Given our dimensions of depth, it is likely that we are predominantly imaging the membrane component of the oocyte. As can be seen, the imaged species are localized differently, not only along the lateral dimensions but also along the depth. In some cases, as illustrated by the image of the m/z 815960, there is an apparent reappearance of signal after its absence at some depths, which may be due to the intermittent presence of other biochemical species, such as proteins or carbohydrates that are known to be interspersed in the membrane. This illustrates the power of analyzing subcellular chemistry in three dimensions. Of note is the fact that many of the species actually have a maximum intensity below the surface of the sample. This is attributed to a removal of overlying material and hence uncovering of the imaged species. A closer look at the peaks in the mass range m/z 540-650 (Figure 3) shows that the 3D distribution of intensity of individual fragments is different. As detailed above, these species are assigned to phospholipids; the fragments in this mass range representative of the molecular species resulting from loss of the “headgroup” part of the molecule. Hill et al.32 performed mass spectrometric analysis of lipids extracted from X. laevis oocytes and reported that the majority of the lipids associated with the oocyte have phosphatidylcholine termini and the most abundant of these are those with fatty acid chains C16:0, C18:1, and C18:2. The variation in distribution of these fragments is demonstrated
in Figure 4a-c, respectively. Figure 4d shows a further variation in distribution of an as yet unassigned peak at m/z 600. Such illustrations clearly demonstrate the ability of TOF-SIMS, coupled to C60+ ion beam systems, to map biochemical species both laterally and as a function of analysis depth. The sputtering process used in SIMS makes it extremely difficult to remove intact protein molecules.35 However, a number of studies have indicated a range of amino acid fragments including immonium ions15 that can be readily observed in TOFSIMS spectra, and in some cases, with the use of multivariate analysis, discrimination of individual proteins and even conformation can be deduced.36 When selecting amino acid-related peaks to image, care must be taken to avoid those where isobaric interference may be a problem. Figure 5 shows the corresponding 3D image of the combined signal from amino acid fragment peaks at m/z 30, 44, 60, 120, 130, 159, 170, and 171. Unlike the lipid fragments, these show maximum intensity on the upper surface of the oocyte. It is postulated that these originate from follicle cells that form a lattice over the oocytes. Our sample preparation procedure is not expected to have removed the entirety of these cells. It is noteworthy that the biomolecular species discussed above persist in the TOF-SIMS analysis even after primary ion dose (35) Garrison, B. J.; Delcorte, A.; Zhigilei, L.; Itina, T. E.; Krantzman, K. D.; Yingling, Y. G.; McQuaw, C. M.; Smiley, E. J.; Winograd, N. Appl. Surf. Sci. 2003, 203, 69-71. (36) Michel, R.; Pasche, S.; Textor, M.; Castner, D. G. Langmuir 2005, 21, 12327-12332.
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
2203
Figure 4. 3D biochemical images of freeze-dried oocyte, showing changes in the individual peaks of the envelop m/z 540-650: m/z (a) 548, (b) 574, (c) 576, and (d) 600. Color scale is normalized to maximum counts per pixel for each individual fragment ion.
Figure 5. 3D biochemical image of freeze-dried oocyte, showing the distribution of the combined signal from amino acid fragments (m/z 30, 44, 70, 120, 130, 159, 170 and 171), attributable to proteins.
densities of 1016 ions/cm2. This illustrates the ability to acquire relatively high-mass signals diagnostic of particular cellular chemistry well beyond the traditional static limit. To our knowledge, this is the first time this has been demonstrated on a real cell system and may be a particular consequence of the low damage cross section for polyatomic beams such as C60+. The persistence of significant intensity with dose (depth) is more clearly illustrated in Figure 6, where the variation of a number of fragments with increasing ion dose is plotted. Such data further establish the ability of the C60 beam to generate stable high-mass molecular signals following the removal of several micrometers of overlying material. 2204 Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
Figure 6. Variation in relative intensity of m/z 184, the mass range m/z 300-320, 369, and 576 with increasing primary ion dose. Signals are normalized to their initial intensity for clarity.
The negative TOF-SIMS images (Figure 7) show a similar persistence of molecular biochemistry at high doses. As noted earlier, a number of fatty acid species including myristate (m/z 227), palmitate (m/z 255), and oleate (m/z 281) are observed even at PIDD 1016 ions/cm2. The images demonstrate the same persistence in high-mass signal as was observed in the positive ion spectrum and also a change in the distribution of the signal from each ion following etching. Generally, there is a shift in the maximum intensity of these phospholipid-associated peaks from an approximately uniform distribution to one of localization predominantly in the outer regions of the oocyte. This cor-
Figure 7. Example negative ion TOF-SIMS images of freeze-dried oocyte exposed to 40 keV C60+. PO3- (m/z 79), palmitate (m/z 255), and oleate (m/z 281). Primary ion dose densities and secondary ion m/z values are given in the figure. Images are 256 × 256 pixels, analysis area ∼1200 × 1200 µm2.
Figure 8. Effect of normalization on the apparent distribution of lipid peaks in the m/z 540-650 range. Raw data (a), signal in each layer scaled up to that in the initial image (b), and cholesterol signal in each pixel normalized to total ion intensity in each pixel (c).
roborates our assumption that these are membrane-associated species that are removed during etching but remain where fresh membrane remains to be analyzed on the edges of the sample. Although it is difficult to ascertain if the distributions observed have been disturbed by the sample preparation or if any matrix effects influence the analysis, the following distributions can be observed laterally and with depth. We see clear differences in distributions of a range of mass fragments. Of particular significance is the persistence of significant secondary ion signal from more chemically/biologically characteristic higher mass fragments. Simply freeze drying a sample does not always preserve morphology as is clear from the microscope image of the oocyte. We envisage that future analyses will require preserving the sample in a hydrated state. Even then the freezing process is critical to maintaining lifelike distributions of biomolecules within the sample. Very rapid cooling is required, and the sample must not be allowed to warm during transfer to the instrument as the sample should ideally remain in an amorphous state. The formation of crystalline ice in the sample will undoubtedly disrupt chemical distribution. For specific applications, chemical fixation may be applicable although analyses of samples prepared by such means would not take full advantage of the SIMS technique as much information would already be lost.
A further complication in the interpretation of the SIMS data arises from the influence of the surrounding chemical environment, the so-called matrix effect. Signals from many molecules may be suppressed or enhanced depending on the local environment leading to misinterpretation of chemical distribution.37 In the case of biological samples, where proton transfer plays a major role in secondary ion formation, the relative gas-phase basicities provide a qualitative assessment of this effect. Our group is actively researching the area, and a more detailed discussion of these effects will be included in forthcoming publications. Changes in topography will also need to be addressed if truly lifelike images are to be generated from the SIMS data. In certain cases, scanning probe microscopy may be able to provide topographical maps of the sample prior to and possibly following SIMS analysis. Given the development of suitable software, these data could be used to scale the data in the z direction although topographical variation of the extent seen on samples such as the oocyte would be too great to be overcome by these means. Further complications may arise from changes in the sputter rate though the sample during analysis; for example, in a frozen hydrated sample, the cytoplasm may be eroded more rapidly than compartments such as the nucleus; we plan to assess the influence of these effects by labeling (37) Jones, E. A.; Lockyer, N.; Vickerman, J. C. Appl. Surf. Sci. 2006, 252 (19), 6727-6730.
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
2205
of specific sample regions and comparing the SIMS results with those from other optical-based microscopy methods. This sample was particularly topographically challenging, both due to the original shape of the oocyte and also from changes in topography during analysis. It can be seen in Figure 3 that toward the end of the analysis there is a gradual decrease in the signal from all the species. This is probably an artifact specific to the analysis of this oocyte. The etching of the sample eroded it in such a way that the incline on the left of the oocyte became gradually steeper; this resulted in an accompanying loss in secondary ion extraction efficiency. To compensate for this loss, one approach would be to normalize the secondary ion intensity in each layer or pixel by either a common fragment or total ion intensity. Figure 8 shows that following normalization of each layer of the stack to the total ion signal in the first image (thus accounting for the loss of signal due to the analysis induced topography change) a much more even secondary ion distribution with depth is observed. Alternatively normalizing the signal in each pixel to the total ion signal in the same pixel could potentially correct for the topography changes due to the initial shape of the sample and as a consequence of the analysis. Again the distribution is more uniform with depth, but in this case, there is a change in the lateral distribution of the signal in each layer. Normalization methods such as these add an extra level of complexity to the interpretation and require careful investigation. We are inclined to agree with Rangarajan and Tyler38 following their thorough study of such effects on model samples that “a greater degree of caution is recommended when interpreting TOF-SIMS images from topographically complex samples.” Consequently, at this stage in the development of 3D imaging, the images presented here are based on the raw intensity variations in images acquired using equivalent primary ion dose. CONCLUSIONS The utilization of very high efficiency primary ion beams such as C60+ adds a new dimension (depth) to the molecular imaging (38) Rangarajan, S.; Tyler, B. J. J. Vac. Sci. Technol., A 2006, 24, 1730-1736. (39) Fletcher, J. S.; Lockyer, N. P.; Vickerman, J. C. Surf. Interface Anal. 2006, 38, 1393-1400.
2206
Analytical Chemistry, Vol. 79, No. 6, March 15, 2007
technique. In this paper, we have demonstrated that the ion beam technology exists that is capable of generating 3D biological images. Future experiments will need to more fully address complications in sample preparation that arise due to the ultrahigh vacuum requirements of the instrument. In this report, the lateral resolution was not a major concern due to the size of the sample. The imaging of smaller systems such as cells with dimensions on the order of tens of micrometers requires much more highly focused ion beams. Currently liquid metal ion sources producing ions such as Ga+, Aun+, and Bin+ offer the best routine lateral resolution; however, 40 keV C60 ion beam systems have recently been shown to be focusable to spot sizes below 200 nm.39 As smaller pixel sizes are achieved, the useful ultimate resolution becomes dependent on the efficiency of the ion beam. We have demonstrated with the oocyte analysis that images may be acquired following primary ion bombardment far in excess of the traditional static limit, indicating a dramatic improvement to the theoretical limit of useful spatial resolution. The ability to perform three-dimensional molecular imaging experiments obviously results in much longer analysis times that we believe will produce a demand for a new generation of high-throughput TOF-SIMS instrumentation. This will require a radical change in the way TOFSIMS data are acquired, utilizing methods for improving the duty cycle of the instruments. To fully capitalize on the vast amount of data generated in such an experiment, multivariate analysis may play an important role in data analysis and interpretation. ACKNOWLEDGMENT This work was funded by The Engineering and Physical Sciences Research Council (EPSRC) and The Biotechnology and Biological Sciences Research Council (BBSRC) UK. The authors gratefully acknowledge Dr. Alex Henderson, Prof. Andreas Wucher, and Dr Robert Mart for assistance and advice regarding data visualization and Matthew Baker for sourcing the oocytes.
Received for review July 26, 2006. Accepted December 15, 2006. AC061370U
