Spatial-Resolution Limits in Mass Spectrometry Imaging of Supported

Anal. Chem. 2010, 82, 2426–2433

Spatial-Resolution Limits in Mass Spectrometry Imaging of Supported Lipid Bilayers and Individual Lipid Vesicles Anders Gunnarsson,†,| Felix Kollmer,‡ Sascha Sohn,§ Fredrik Ho¨o¨k,*,† and Peter Sjo¨vall*,†,| Department of Applied Physics, Division of Biological Physics, Chalmers University of Technology, SE-412 96 Go¨teborg, Sweden, ION-TOF GmbH, Heisenbergstrasse 15, D-48149, Mu¨nster, Germany, Physikalisches Institut der Universita¨t Mu¨nster, Wilhelm-Klemm-Strasse 10, D-48149 Mu¨nster, Germany, and Department of Chemistry and Materials Technology, SP Technical Research Institute of Sweden, P.O. Box 857, SE-501 15 Borås, Sweden The capabilities of time-of-flight secondary ion mass spectrometry (TOF-SIMS) with regards to limits in lateral resolution for biological samples are examined using supported lipid bilayers and individual lipid vesicles, both being among the most commonly used cell membrane mimics. Using supported 1-palmitoyl-2-oleoyl-sn-glycero3-phosphocholine (POPC) bilayers confined to a SiO2 substrate by a chemically modified gold surface, the edge of the lipid bilayer was analyzed by imaging TOFSIMS to assess the lateral resolution. The results using 80 keV Bi32+ primary ions show that, under optimized conditions, mass spectrometry imaging of specific unlabeled lipid fragments is possible with sub-100 nm lateral resolution. Comparison of the secondary ion yields for the phosphocholine ion (m/z 184) from a POPC bilayer using C60+ or Bi3+ primary ions showed similar results, indicating an advantage of Bi3+ primary ions for high-resolution imaging of lipid membranes, due to their better demonstrated focusing capability. Moreover, using 300 nm vesicles of different lipid composition, the capability to detect and chemically identify individual submicrometer lipid vesicles at separations down to ∼1 µm is demonstrated.

and lipid model systems.9-12 This development has been stimulated by the increased use of cluster primary ion sources, such as Aun+, Bin+, and C60+, which by increasing the secondary ion yield of high-mass fragment and molecular ions13 has greatly improved the capability to analyze biomolecules, in particular lipids, by TOF-SIMS in the mass range up to a few thousand daltons. In comparison to other imaging mass spectrometry techniques, the advantage of TOF-SIMS is the capability to provide detailed chemical information at high lateral resolution. The analysis of specific organic molecules by TOF-SIMS is primarily made possible by carrying out the analysis in the so-called static SIMS mode, i.e., analysis of the original sample surface at low doses of incident (primary) ions, at which the surface has not yet been molecularly damaged.14 In contrast, analysis in the dynamic SIMS mode is carried out at high primary ion dose densities on a molecularly damaged and continuously eroding surface which normally means that the chemical information is limited to what can be provided by atomic or diatomic fragment ions. With this limitation in chemical information, however, dynamic SIMS has the capability to image the lateral distribution of emitted secondary ions in biological structures with high sensitivity at lateral resolutions down to 50-100 nm.15-17 In addition, using isotopic labeling, specific lipids were recently imaged using dynamic SIMS

Time-of-flight secondary ion mass spectrometry (TOF-SIMS) has emerged as an important technique for analysis of biointerfaces, providing chemical information from biological samples at submicrometer resolution without the need of labeling.1 A growing number of studies using TOF-SIMS have demonstrated the possibility of mapping the spatial distribution of a variety of phospholipids and cholesterol in biological tissues,2-5 cells,6-8

(4) Altelaar, A. F. M.; van Minnen, J.; Jimenez, C. R.; Heeren, R. M. A.; Piersma, S. R. Anal. Chem. 2005, 77, 735. (5) Sjovall, P.; Lausmaa, J.; Johansson, B. Anal. Chem. 2004, 76, 4271. (6) Sjovall, P.; Lausmaa, J.; Nygren, H.; Carlsson, L.; Malmberg, P. Anal. Chem. 2003, 75, 3429. (7) Parry, S.; Winograd, N. Anal. Chem. 2005, 77, 7950. (8) Ostrowski, S. G.; Van Bell, C. T.; Winograd, N.; Ewing, A. G. Science 2004, 305, 71. (9) Prinz, C.; Hook, F.; Malm, J.; Sjovall, P. Langmuir 2007, 23, 8035. (10) Ross, M.; Steinem, C.; Galla, H. J.; Janshoff, A. Langmuir 2001, 17, 2437. (11) Michel, R.; Subramaniam, V.; McArthur, S. L.; Bondurant, B.; D’Ambruoso, G. D.; Hall, H. K.; Brown, M. F.; Ross, E. E.; Saavedra, S. S.; Castner, D. G. Langmuir 2008, 24, 4901. (12) McQuaw, C. M.; Zheng, L. L.; Ewing, A. G.; Winograd, N. Langmuir 2007, 23, 5645. (13) Kollmer, F. Appl. Surf. Sci. 2004, 231-232, 153. (14) Vickerman, J. C.; Briggs, D. Eds., TOF-SIMS: Surface Analysis by Mass Spectrometry; IM Publications and SurfaceSpectra Limited: Charlton, Chichester, West Sussex, UK, 2001. (15) Quintana, C.; Wu, T. D.; Delatour, B.; Dhenain, M.; Guerquin-Kern, J. L.; Croisy, A. Microsc. Res. Tech. 2007, 70, 281. (16) Chandra, S.; Smith, D. R.; Morrison, G. H. Anal. Chem. 2000, 72, 104a.

* To whom correspondence should be addressed. E-mail: fredrik.hook@ chalmers.se (F.H.); peter.sjo ¨[email protected] (P.S.). Phone: +46 (0) 31-772 6130 (F.H.); +46 10 516 5299 (P.S.). Fax: +46 (0) 31-772 3134 (F.H.); +46 33 10 3388 (P.S.). † Chalmers University of Technology. ‡ ION-TOF GmbH. § Physikalisches Institut der Universita¨t Mu ¨ nster. | SP Technical Research Institute of Sweden. (1) Brunelle, A.; Touboul, D.; Laprevote, O. J. Mass Spectrom. 2005, 40, 985. (2) Sjovall, P.; Johansson, B.; Lausmaa, J. Appl. Surf. Sci. 2006, 252, 6966. (3) Nygren, H.; Borner, K.; Hagenhoff, B.; Malmberg, P.; Mansson, J. E. Biochim. Biophys. Acta 2005, 1737, 102.

2426

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

10.1021/ac902744u  2010 American Chemical Society Published on Web 02/17/2010

at a lateral resolution of roughly 100 nm.17 In another, newly developed instrument, analysis is carried out in the dynamic SIMS mode using C60+ primary ions.18 Due to the significantly reduced molecular damage caused by the C60+ ions, as compared to the damage caused by the atomic primary ions typically used in dynamic SIMS, the new instrument is expected to enable three-dimensional (3D) mapping of organic molecules in biological samples. Although promising, the technique relies on the amount of molecular information retained on the sample surface after C60+ sputtering, which has proven to be highly variable and still poorly understood.19-21 Other imaging mass spectrometry techniques such as imaging matrix-assisted laser desorption ionization (MALDI)22 and desorption by electrospray ionization (DESI)23 have the important advantage to cover a wider mass range than the SIMS-based methods, allowing for mapping of proteins and nucleotides. However, the spatial resolution is currently limited to about 25-50 µm for imaging MALDI24 and around 100 µm for DESI.25,26 The lateral resolution that can be obtained by TOF-SIMS is mainly limited by two factors, which both are connected to the primary ion source used in the analysis, namely, (i) the focus of the primary ion beam and (ii) the secondary ion yield of the specific substance and the specific secondary ion probed. When the two major types of primary ion sources currently used in TOFSIMS analysis of biological samples, i.e., Bin+ and C60+, are compared, the bismuth cluster ion source has the advantage of having both a narrow beam focus and a high secondary ion yield of large characteristic ions from organic materials. This source, which is based on the liquid metal ion gun principle, routinely provides a beam focus diameter of 100-200 nm, whereas the C60+ source cannot be focused down below ∼1 µm without serious sacrifice in the primary ion current (pulsed mode).27 Furthermore, experimental13,28,29 and theoretical30 comparisons indicate that Bin+ and C60+ primary ions provide similar secondary ion yields from organic materials. The possibility to accurately map the chemical distribution of specific substances also depends on the properties of the analyzed sample, including factors such as the surface concentration, the chemical environment (matrix effects), surface topography, and possible relocation of substances during sample preparation. In (17) Kraft, M. L.; Weber, P. K.; Longo, M. L.; Hutcheon, I. D.; Boxer, S. G. Science 2006, 313, 1948. (18) Fletcher, J. S.; Rabbani, S.; Henderson, A.; Blenkinsopp, P.; Thompson, S. P.; Lockyer, N. P.; Vickerman, J. C. Anal. Chem. 2008, 80, 9058. (19) Jones, E. A.; Lockyer, N. P.; Vickerman, J. C. Anal. Chem. 2008, 80, 2125. (20) Debois, D.; Brunelle, A.; Laprevote, O. Int. J. Mass Spectrom. 2007, 260, 115. (21) Wucher, A.; Cheng, J.; Zheng, L.; Winograd, N. Anal. Bioanal. Chem. 2009, 393, 1835. (22) Heeren, R. M. A.; Sweedler, J. V. Int. J. Mass Spectrom. 2007, 260, 89. (23) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004, 306, 471. (24) Touboul, D.; Roy, S.; Germain, D. P.; Chaminade, P.; Brunelle, A.; Laprevote, O. Int. J. Mass Spectrom. 2007, 260, 158. (25) Wiseman, J. M.; Ifa, D. R.; Venter, A.; Cooks, R. G. Nat. Protoc. 2008, 3, 517. (26) Ifa, D. R.; Wiseman, J. M.; Song, Q. Y.; Cooks, R. G. Int. J. Mass Spectrom. 2007, 259, 8. (27) Hill, R.; Blenkinsopp, P. W. M. Appl. Surf. Sci. 2004, 231-232, 936. (28) Kersting, R.; Hagenhoff, B.; Kollmer, F.; Mollers, R.; Niehuis, E. Appl. Surf. Sci. 2004, 231-232, 261. (29) Touboul, D.; Kollmer, F.; Niehuis, E.; Brunelle, A.; Laprevote, O. J. Am. Chem. Soc. 2005, 16, 1608. (30) Seah, M. P. Surf. Interface Anal. 2007, 39, 890.

real biological samples, these properties are normally highly uncertain, and it is therefore difficult to obtain well-defined assessments of the imaging capabilities of TOF-SIMS using such samples. In this work, the limits in lateral resolution for static TOF-SIMS analysis of biological samples are therefore investigated using two well-defined and highly relevant biological model systems, the supported lipid bilayer and the lipid vesicle. As mimics of the cell membrane, both model systems represent important biological structures that have proven to be particularly suitable for TOF-SIMS analysis and have been the focus of many recent TOF-SIMS studies of biological cells and tissues.1-8 In this context, phosphatidylcholine (PC) is one of the most frequently studied lipids, partly due to its biological relevance and high abundance in cell membranes but also due to the high secondary ion yield of its characteristic peak at m/z 184. Furthermore, cell membrane mimics are presently a very active area of research due to their high importance in various fields ranging from pharmaceutical industry to fundamental cell biology research.31,32 Using supported lipid bilayers on SiO2 confined by a chemically modified gold surface, we here assess the lateral resolution by analyzing the sharpness of the imaged borderline between the lipid bilayer and the chemically modified gold surface. This sample is a biomolecular equivalent to the inorganic strip-patterned samples that have been used previously to evaluate the lateral resolution of TOF-SIMS instruments.33,34 In addition, Bi3+ and C60+ primary ions are compared with respect to the secondary ion yield and useful lateral resolution obtained in the analysis of supported POPC bilayers. For the second type of sample, we address issues related to analysis of individual submicrometer-sized biological objects by analyzing lipid vesicles (Ø ∼ 300 ± 100 nm) of different lipid compositions adsorbed onto a SiO2 surface at low surface coverage. EXPERIMENTAL SECTION Substrate Preparation. The lipid bilayer structures were prepared on gold-patterned SiO2 substrates, as shown in Figure 1A. The gold pattern was made using conventional optical lithography and consisted of ∼1 × 2 mm2 gold patches, each containing three strip-shaped windows (4, 7, and 100 µm wide, respectively) exposing the underlying SiO2 surface. Oxidized silicon wafers were spin-coated with two layers of resist, LOR3A, and baked in 180 °C for 20 min followed by S1813 and baked on hot plate for 90 s (MicroChem Corporation, Massachusetts). The resist was exposed for 6 s (Karl Su ¨ss MJB 3 mask aligner, Germany) and developed in MF319 for 60 s followed by rinsing in Milli-Q and drying with N2. Metal evaporation consisted of 5 nm titanium for adhesion followed by 50 nm gold (Pfeiffer Classic 500, U.S.A.) before subsequent lift off by gentle sonication in acetone followed by remover 1165 and rinsing in Milli-Q water. Scanning electron microscopy (SEM) was used to examine the lithographic gold pattern and, in particular, the sharpness of the Au/SiO2 edges (Figure 1A-C). From the high-magnification (31) Chan, Y. H. M.; Boxer, S. G. Curr. Opin. Chem. Biol. 2007, 11, 581. (32) Cooper, M. A. J. Mol. Recognit. 2004, 17, 286. (33) Senoner, M.; Wirth, T.; Unger, W.; Osterle, W.; Kaiander, I.; Sellin, R. L.; Bimberg, D. Surf. Interface Anal. 2004, 36, 1423. (34) Senoner, M.; Unger, W. E. S. Surf. Interface Anal. 2007, 39, 16.


2427

Figure 1. (A) Micrograph of the gold pattern on the SiO2 surface showing three SiO2 stripes, 4, 7, and 100 µm wide, respectively. (B and C) SEM images of a 7 µm stripe. (C) The sample is tilted 30° where the top of the gold film is emphasized with the vertical white line. The inset in panel C shows a typical line profile across the Au-SiO2 interface (marked with the purple line), where the width of the interface can be estimated to 20 times through a 100 nm polycarbonate membrane (Whatman, U.S.A.). For the single-vesicles experiments, a 400 nm polycarbonate membrane was used instead. Since the phase transition temperature of the D13-DPPC lipids is around 40 °C, these lipids where both hydrated and extruded at elevated temperature (∼50 °C). The final vesicle suspension was stored in a refrigerator at 4 °C. Size distribution measurements using 2428


dynamic light scattering was performed on a Zetasizer Nano (Malvern, U.K.). Supported lipid bilayers were formed by incubating the goldpatterned SiO2 surface with lipid vesicles (0.2 mg/mL lipids) for 30 min, following an established protocol.35 Bilayer formation was verified by measuring the lipid mobility using fluorescence recovery after photobleaching (FRAP)36 (Figure 1D). Recovery was observed in all three SiO2 stripes, confirming full mobility and bilayer formation. For the FRAP measurements, 1 wt % of the lipids was replaced with lissamine rhodamine B 1,2-dihexadecanoyl-sn-glycero-3-phosphatidylethanolamine (rhodamine-DHPE, Invitrogen, U.K.). The inertness of the chemically modified gold surface is verified by the dark appearance of the gold area in Figure 1D, where only a few adsorbed intact lipid vesicles are visible. Adsorption of intact individual lipid vesicles was performed at a lipid concentration of 2 µg/mL. HEPES buffer was used throughout all experiments. TOF-SIMS Analysis. Prior to TOF-SIMS analysis, the supported lipid bilayer and lipid vesicle samples were first rinsed in Milli-Q water to remove ions from the buffer and then plungefrozen in liquid propane at -185 °C. In order to minimize lipid migration prior to analysis of the bilayer edge, the supported bilayer samples were freeze-dried inside the TOF-SIMS vacuum chamber and immediately analyzed at T ) -80 °C.9 For the vesicle samples, lipid migration is less critical, and these samples were therefore freeze-dried on a liquid-nitrogen-cooled metal block (35) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397. (36) Jonsson, P.; Jonsson, M. P.; Tegenfeldt, J. O.; Hook, F. Biophys. J. 2008, 95, 5334.

Figure 2. Schematic illustration of the two different lipid membrane model systems: (A) the supported lipid bilayer and (B) intact lipid vesicles formed on SiO2 surrounded by PEG-modified Au. Illustration is not to scale.

under vacuum conditions overnight and then analyzed by TOFSIMS at room temperature. The imaging TOF-SIMS analyses were carried out in two different instruments at settings optimized for high image resolution. The high lateral resolution analysis of the supported lipid bilayer samples was done in a prototype TOF-SIMS V instrument (ION-TOF GmbH, Germany) equipped with a Bin+ cluster liquid metal ion source running at an acceleration voltage of 40 kV (as compared to the “standard” 25 kV). With the use of the higher acceleration voltage, a significantly improved beam focus was obtained (focus diameter ∼50 nm as compared to ∼150 nm for the “standard” system used for the lipid vesicle samples, see below). The smaller beam focus was particularly important in the analysis of the lipid bilayer sample, where the edge of the prepared bilayer is expected to be very sharp, allowing for a precise determination of the ultimate spatial resolution. The lipid bilayer analyses were carried out using the following parameters: 80 keV Bi32+ primary ions, pulsed current 13 fA, pulse width 170 ns, analysis area 21 × 21 µm2, number of pixels 512 × 512, pulses per pixel 8, dose density 4.3 × 1012/cm2, single scan. The lipid vesicle samples were analyzed in a TOF-SIMS IV instrument (ION-TOF GmbH, Germany) using 50 keV Bi32+ primary ions (pulsed current 40 fA, pulse width 100 ns, 512 × 512 pixels (Figure 5) or 256 × 256 pixels (Figure 6), dose densities 3.5 × 1012 cm-2 (Figure 5) and 3.1 × 1011 cm-2 (Figure 6), multiple scans. The secondary ion yield measurements were carried out in the TOF-SIMS IV instrument (same as in the vesicle analysis) optimized for high mass resolution (bunched mode, mass resolution m/∆m ∼ 5000), using 25 keV Bi3+ primary ions at a pulsed current of 0.1 pA and 10 keV C60+ primary ions at a pulsed current of 0.11 pA. The analysis area was 65 × 65 µm2, but only data from the center 31 × 31 µm2 was used in the analysis in order to minimize uneven dose densities due to the large beam diameter of the C60+ ion beam (40 µm). The diameter of the Bi3+ beam was 3.3 µm in these measurements. RESULTS AND DISCUSSION The two types of lipid bilayer systems analyzed in this work are schematically shown in Figure 2, i.e., (a) a supported POPC bilayer displaying a sharp edge against the surrounding functionalized gold surface and (b) separated individual lipid vesicles adsorbed on a SiO2 surface. In both cases, the lipid bilayer structures were prepared on strip-shaped SiO2 areas sur-

rounded by Au regions functionalized with a self-assembled monolayer of thiol-PEG37 to prevent lipid vesicle adsorption. The main purpose of the Au pattern is to provide a sharp edge of the POPC bilayer for the lateral resolution measurements. However, the hydrophilic PEG-modified Au regions also has an advantageous water-stabilizing effect, resulting in the formation of a thin water film that covers the bilayer structures on the strip-shaped SiO2 areas during the rinsing and plungefreezing steps prior to the freeze-drying of the sample. Imaging Analysis of a Supported Lipid Bilayer Edge. In order to optimize the focus of the primary ion beam and thus the lateral resolution, these measurements were carried out using a Bi cluster ion source capable of producing Bi32+ primary ions at 80 keV energy (acceleration voltage 40 kV). All samples were aligned such that the edges of the SiO2 strips were parallel to the primary ion beam in order to minimize possible screening of primary ions due to differences in topography (the gold pattern is, on average, 55 nm thick, whereas the bilayer on the SiO2 surface is only 5 nm), which could possibly affect the yield of secondary ions close to the edge. Figure 3 shows an example of ion images and intensity profiles obtained from TOF-SIMS analysis of a supported POPC bilayer stripe. The ion image in Figure 3A shows the signal intensity distribution of the characteristic fragment of phosphatidylcholine at m/z 184 over an 18 × 18 µm2 area. The signal intensity is only originating from the SiO2 area, demonstrating that the bilayer is well-confined within the stripe. Parts B and C of Figure 3 show signal intensity profiles across the POPC/Au border, providing a way to quantify the sharpness by which the bilayer edge is reproduced in the m/z 184 ion image. The intensity profile in Figure 3B was obtained from the area indicated in Figure 3A by addition of multiple horizontal line profiles along the direction parallel to the bilayer edge (3.6 µm). From this profile, a lateral resolution of 60 nm is obtained (defined as the distance between 16% and 84% of the maximum signal intensity).34 Parts D and E of Figure 3 show the corresponding ion image and signal intensity profile for a characteristic fragment of thiol-PEG at m/z 45 (C2H5O+). This fragment displayed the highest contrast in the ion images, due to the high secondary ion yield from thiol-PEG and low yield from POPC. The m/z 45 ion image (Figure 3D) shows strong signal intensity localized to the gold area, indicating that the surface modification was well-preserved after freeze-drying. Using the same analysis of a corresponding intensity profile across the Au/SiO2 edge in the m/z 45 ion image, the resolution was also in this case determined to 60 nm (Figure 3E). In a static SIMS measurement, the lateral resolution that can be obtained is limited both by the focus of the primary ion beam and on the detection efficiency of the specific secondary ion that is used to monitor the analyte substance.38 The latter reflects the fact that the number of secondary ions that can be emitted per unit area is limited, due to the molecular damage caused by the primary ion bombardment. Limitations in lateral resolution due to detection efficiency become clear when intensity profiles from areas of different sizes are compared. By adding line profiles over (37) Dahlin, A. B.; Jonsson, M. P.; Hook, F. Adv. Mater. 2008, 20, 1436. (38) Kotter, F.; Benninghoven, A. Appl. Surf. Sci. 1998, 133, 47.


2429

Figure 3. Positive ion images of the C5H15NPO4+ ion at m/z 184 originating from the POPC lipid headgroup (A) and the C2H5O+ ion at m/z 45 originating from the thiol-PEG (D) from a freeze-dried POPC bilayer in the 7 µm SiO2 stripe. The field of view is 18 × 18 µm. Horizontal intensity profiles (counts per horizontal pixel) across the interface for m/z 184 (B-C) and m/z 45 (E-F) obtained from the ion images in A and D, respectively, including estimates of the spatial resolution. For each horizontal pixel (41 nm), the signal intensity was added over a vertical distance of 3.6 µm (B and E), as indicated by the red areas in A and D, and of 0.25 µm (C and F), approximately corresponding to the width of one of the horizontal red lines in A and D.

Figure 4. Measured secondary ion yield for the m/z 184 ion as a function of accumulated primary ion dose density during TOF-SIMS analysis of a supported POPC bilayer using Bi3+ and C60+ primary ions. Fitted curves are superimposed on the measured data, representing a single-exponential decay for Bi3+ and a double-exponential decay for C60+.

an extended area, the signal intensity profiles in Figure 3, parts B and E, show relatively high signal-to-noise ratios which, in turn, allows for a sharp definition of the interface position (60 nm). However, if the area over which the intensity line profiles are added is reduced, the signal intensities are reduced, possibly, making the measured edge position less defined. Parts C and F of Figure 3 show signal intensity profiles obtained by addition of six horizontal line profiles, representing a vertical distance of 246 nm (indicated by one of the horizontal red lines in Figure 3, parts A and D). In the case of the phosphocholine profile (Figure 3C), the POPC bilayer stripe is still visible, but the signal intensity is very low, approximately 2.2 counts per horizontal pixel of 41 nm. The weak signal intensity makes the position of the interface considerably more uncertain, demonstrating that the image resolution is now limited by the signal intensity rather than the focus of the primary ion beam. 2430


A related property is the so-called useful lateral resolution, L, which is defined as the size of the smallest square (L2), from which a significant signal can be obtained.29 The useful lateral resolution was recently determined to be around 100 nm for the phosphocholine ion at m/z 184 in a supported POPC bilayer, indicating that a surface area of 100 × 100 nm2 is required in order to obtain a significant signal (arbitrarily defined to 4 counts).9 The ultimate lateral resolution in two dimensions that presently can be obtained is thus limited to this area even if a substantially smaller primary ion focus is used. In fact, the measured signal intensity (2.2 counts) in the narrow line profile (Figure 3C) is somewhat smaller but still in good agreement with the reported useful lateral resolution (100 nm), considering that the probed area in each horizontal pixel is approximately 100 × 100 nm2 (41 × 246 nm2). In the case of thiol-PEG, monitored by the m/z 45 signal, the intensity profile shows a significantly higher signal-to-noise ratio (Figure 3F), demonstrating a considerably higher useful lateral resolution and a more well-defined interface position than what was obtained for the m/z 184 signal representing the POPC bilayer. Comparison between Bi3 and C60 Primary Ions in the Analysis of Lipid Bilayers. The narrow focus of the Bi cluster ion source as compared to the C60+ source makes Bin+ highly suitable as primary ions in high-resolution TOF-SIMS imaging of biological samples. However, in order to investigate the possible advantage of C60+ primary ions with regard to detection sensitivity, e.g., if a narrow beam focus is not needed for the analysis, the secondary ion yields and useful lateral resolutions were determined using Bi3+ and C60+ primary ions in the analysis of a supported lipid bilayer. Figure 4 shows the measured secondary ion yields of the m/z 184 fragment ion as a function of accumulated primary ion dose density (PIDD) in the analysis of a supported POPC bilayer, using Bi3+ and C60+ primary ions. In both cases, the secondary ion

Figure 5. Chemical structure and corresponding positive ion spectra for POEPC, POPC, and D13-DPPC, respectively. The characteristic ions used to identify each lipid are marked by their corresponding mass-to-charge ratios.

yield starts at a similar value of around 0.017 and then decays to zero at high primary ion doses, reflecting the fact that the sample molecules are consumed during the analysis. The results show that the initial secondary ion yield is slightly higher using Bi3+ primary ions (0.0186 as compared to 0.0164 for C60+), but the decay of the yield as a function of PIDD is significantly slower using C60+ primary ions, indicating less molecular damage caused by the C60+ primary ions. In the case of Bi3+ primary ions, the measured secondary ion yield was fitted to a single-exponential decay (see Figure 4), from which a useful lateral resolution of 93 nm could be calculated, in excellent agreement with previous results.9 In contrast, the secondary ion yield using C60+ primary ions could not be fitted satisfactory using a single-exponential decay. Instead, the C60+ curve was fitted to a double-exponential decay (Figure 4), consistent with two parallel mechanisms causing molecular damage with two different disappearance cross sections. Using the fact that the maximum number of detectable secondary ions is obtained by integration of the SI yield versus PIDD curve, a useful lateral resolution of 67 nm was calculated from the measured data. These results indicate that the signal intensities that can be obtained from a lipid bilayer (m/z 184 fragment) are very similar when using Bi3+ or C60+ primary ions. Furthermore, the signal decay down to zero observed in Figure 4 clearly shows that these 2D lipid membrane structures are best analyzed in the static regime, also when using C60+ primary ions. However, for depth profiling of 3D membrane structures, the results indicate that sputtering is better performed using C60+ ions due to the smaller molecular damage caused by this ion. Thus, Bi3+ and C60+ are an optimum combination for a dual-beam 3D depth profiling

Figure 6. (A and B) Positive ion images of freeze-dried single POPC and POEPC vesicles adsorbed on the 100 µm SiO2 stripe showing specific fragment peaks for both lipids (m/z 184) and for POEPC only (m/z 212). Spots, each consisting of a small cluster of pixels with intensity above the background, are identified as individual vesicles. The field of view 50 × 50 µm2. (C) Overlay of the positive ion images (A and B) with the characteristic signal for both lipids (m/z 184) in green and the characteristic signals for POEPC only (m/z 212) in red. Red and green arrows indicate POEPC and POPC vesicles, respectively. (D) Positive ion spectra from ROIs of a single POPC and POEPC lipid vesicle, respectively. The characteristic peaks at m/z 184 and m/z 212 are clearly visible from the POEPC vesicle. In contrast, the POPC vesicle shows only the characteristic peaks at m/z 184, whereas no peak is observed at m/z 212. The unusually high noise level at m/z 191 and 207 originates from trace amounts of contaminants.

mode. Regarding the charge of the primary ions, Bi32+ produces a slightly more narrow beam focus and is therefore more suitable for high-resolution imaging, although only minor differences are expected in the secondary ion yields between Bi3+ and Bi32+.39 Imaging Analysis of Individual Submicrometer Phospholipid Vesicles. Although the capability to image a sharp edge (shown above) represents a common method to measure the lateral resolution,34 the second sample system, consisting of single lipid vesicles adsorbed on a SiO2 surface (Figure 2B), was studied in order to evaluate the capability of TOF-SIMS to image small point-shaped biological structures. Hence, analysis of the latter system more directly demonstrates the lateral (39) Nagy, G.; Lu, P.; Walker, A. V. J. Am. Chem. Soc. 2008, 19, 33.


2431

resolution as limited both by the detection efficiency and the primary ion beam focus. Intact lipid vesicles (Ø ) 300 ± 100 nm, measured by dynamic light scattering [DLS]) were adsorbed at low concentrations on the SiO2 area of the Au-patterned substrate. In order to investigate the capability to chemically identify single lipid vesicles and to verify that the detected lipid structures represent single vesicles and not clusters, vesicles composed of different lipids were mixed and analyzed simultaneously. Figure 5 shows the molecular structures of the lipids used in the vesicles (POEPC, POPC, and D13-DPPC) and reference spectra obtained from TOFSIMS analysis of the pure lipids. The rationale for using POEPC and POPC is to demonstrate the capability to detect and chemically distinguish unlabeled vesicles made of different lipids. The spectra from POEPC and POPC display common peaks at m/z 58, 86 and 184, reflecting the similar molecular structures of the two lipids. However, the extra ethyl group bound to the phosphate group in POEPC gives rise to a strong peak at m/z 212 not present in the spectrum from POPC. The signal from m/z 212 can therefore be used to distinguish between POPC and POEPC vesicles. Figure 6 shows an example of results obtained from a sample containing adsorbed POEPC and POPC vesicles. The ion images of the characteristic ions at m/z 184 (Figure 6A) and m/z 212 (Figure 6B) both show relatively large numbers of spots, consistent with detection of individual lipid vesicles. A detailed inspection of the images shows that all spots in the m/z 212 ion image are also present in the m/z 184 ion image. This is consistent with the reference spectra for POEPC (Figure 5), indicating that the spot structures showing signal at both m/z 184 and m/z 212 correspond to POEPC vesicles. In contrast, a number of spots can be observed in the m/z 184 ion image that are not present in the m/z 212 image, consistent with the detection of POPC vesicles (Figure 5). This observation can be clearly seen in Figure 6C, which shows an overlay of the two ion images with the signal intensity from m/z 184

(POPC and POEPC) in green and the signal intensity from m/z 212 (POEPC) in red. Due to the presence or absence of the m/z 212 fragment, POEPC and POPC vesicles (some of which are marked by arrows) appear as colocalized red and green spots or green spots, respectively. The observation of individual vesicles is further confirmed from single lipid vesicle spectra (Figure 6D), obtained by selecting a region of interest (ROI) corresponding to one vesicle of each type. The single-vesicle spectra clearly demonstrate the capability to distinguish between POPC and POEPC vesicles, based on the absence or presence of signal at m/z 212 in combination with the observation of a peak at m/z 184. Note in particular that this identification of different lipids was made without labeling. The recorded signal intensities from single vesicles were determined for five POPC vesicles and five POEPC vesicles in Figure 6, using single-vesicle ROIs. The results showed an average signal intensity of 57 counts for each POPC vesicle at m/z 184 and 125 counts for each POEPC vesicle at m/z 212, indicating significantly higher detection efficiency for POEPC. The surface bilayer area of a 300 nm vesicle is about 28 times larger than a 100 × 100 nm2 area from which a maximum signal of 4 counts is expected for POPC (based on the 100 nm useful lateral resolution9). Therefore, the maximum signal intensity that can be expected from a 300 nm POPC vesicle is approximately 110 counts, i.e., roughly a factor of 2 larger than what is observed (57 counts). Taking into account that the somewhat lower value of the observed signal intensity can be partially explained by incomplete consumption of the sample molecules (dose density 3.5 × 1012 ions/ cm2), the agreement with the expected value is very good. Although the analysis above demonstrates that vesicles of different lipid composition can be imaged and chemically identified, the presence of clusters consisting of several vesicles cannot be entirely ruled out. To further establish whether the imaged lipid spots correspond to single vesicles or vesicle clusters, a similar experiment was carried out in which the POEPC vesicles

Figure 7. (A-F) Positive ion images from freeze-dried single POPC and D13-DPPC vesicles adsorbed on the 100 µm SiO2 stripe showing three specific fragment peaks for each lipid (m/z 58, 86, and 184 u for POPC and m/z 66, 98, and 197 for D13-DPPC). Some of the spots present in all three images from the same lipid are highlighted with red and green squares. The signal in the top of the m/z 58 and m/z 86 images is originating from mass interferences with fragment ions from thiol-PEG. The field of view 100 × 100 µm2. (G) Overlay of the positive ion images (A-F) with the added signal from the three characteristic ions for POPC (m/z 58 + 86 + 184) in red and the three characteristic signals for D13-DPPC (m/z 66 + 98 + 197) in green. Inset: Positive ion spectra from ROI for two individual lipid vesicles (marked by the red and green arrows) indicating a POPC and D13-DPPC vesicles, respectively. 2432


were replaced by vesicles made of deuterated phosphatidylcholine (D13-DPPC). The rationale for using this lipid is to use two different lipids which produce characteristic peaks of similar high yield but at different positions in the mass spectrum, thus avoiding common or overlapping peaks (see Figure 5). In the case of vesicle clusters present on the surface, one would expect that some of these clusters contain vesicles of both types and, consequently, signal from both lipids would be detected in the same spot. If spots containing both lipids are not detected on the surface, it can be concluded that the lipid spots correspond to single lipid vesicles and not to vesicle clusters. In Figure 7A-F, ion images of the three characteristic peaks for each lipid are shown separately. Each image shows a relatively large number of spots consistent with detection of individual lipid vesicles. Furthermore, detailed inspection of the images shows that many of the spots are common for the m/z 58, 86, and 184 images, confirming the assignment of these spots to POPC vesicles (some of which are highlighted by red squares), and other spots are common for the m/z 66, 98, and 197 images, which thus can be assigned to D13-DPPC vesicles (green squares). An overlay of all six ion images is shown in Figure 7G, in which the added signal intensity from m/z 58, 86, and 184 (POPC) is shown in red and the added intensity from m/z 66, 98, and 197 (D13-DPPC) is shown in green. The image clearly shows individual spots representing the two different types of lipids. Furthermore, there are no overlapping signals from the two different lipids (which would result in yellow spots), demonstrating that each spot represents a single lipid vesicle and not a cluster of several lipid vesicles. This is supported by single lipid vesicle spectra (inset), obtained by selecting a small ROI for one lipid vesicle of each lipid type (red and green arrows in Figure 7G). Similar to the analysis of the POEPC and POPC vesicle mixture above (Figure 6), the single-vesicle spectra in Figure 7 show that each lipid vesicle can be chemically identified using the characteristic peaks at m/z 184 and m/z 197 for POPC and D13-DPPC, respectively. Figures 6 and 7 thus show that individual submicrometer-sized lipid vesicles can be detected and chemically identified on a substrate surface by TOF-SIMS. In addition, image analysis of the

spots in Figures 6 and 7 indicate a full width at half-maximum (fwhm) diameter of ∼800 nm, suggesting that the lipid molecules of the collapsed vesicle are reasonably well-retained at the original location, although no further information about the structure of the collapsed vesicle after freeze-drying can be obtained. With regard to lateral resolution, these results indicate that TOF-SIMS analysis using Bi3 primary ions is capable of detecting and spatially (and chemically) distinguishing individual lipid vesicles ∼300 nm in diameter at a spatial separation on the surface down to ∼1 µm. For this type of measurement, our results clearly demonstrate the importance of using a primary ion source that provides both a narrow focus and a high detection efficiency of the probed molecule.

(40) Gunnarsson, A.; Sjovall, P.; Hook, F. Nano Lett. 2010, 10, 732.

AC902744U

CONCLUSIONS We have investigated the limits in lateral resolution for TOFSIMS analysis of biological samples using two different cell membrane mimics. The results demonstrate that a sharp lipid bilayer edge can be imaged with 60 nm lateral resolution, by detection of a large secondary ion fragment specific to phosphatidylcholine. Furthermore, our results demonstrate the importance of using a primary ion source which provides both a narrow beam focus and a high secondary ion yield in order to produce highresolution images of biological structures using TOF-SIMS. The results also demonstrate the possibility to detect and chemically identify individual submicrometer lipid vesicles at separations down to ∼1 µm, which may potentially allow investigations of lipid compositions of small native vesicles derived or naturally secreted from living cells. Currently, we also explore the possibility to utilize lipid vesicles of different lipid compositions as chemical barcodes in diagnostic contexts.40 ACKNOWLEDGMENT This work was supported by EC FP6 funding (contract no. 005045 NANOBIOMAPS), the Swedish Governmental Agency for Innovation Systems (VINNOVA), the SSF funded Ingvar program, and the Swedish Research Council (Grant 2005-3140). Received for review December 2, 2009. Accepted February 3, 2010.


2433

Spatial-Resolution Limits in Mass Spectrometry Imaging of Supported

Recommend Documents