Mass Spectral Imaging of Glycophospholipids, Cholesterol, and

Manchester Interdisciplinary Biocentre, School of Chemical Engineering and ... Penn State University, 104 Chemistry Building, University Park, Pennsyl...
0 downloads 0 Views 2MB Size
Langmuir 2008, 24, 11803-11810

11803

Mass Spectral Imaging of Glycophospholipids, Cholesterol, and Glycophorin A in Model Cell Membranes Matthew J. Baker,*,†,‡ Leiliang Zheng,‡ Nicholas Winograd,‡ Nicholas P. Lockyer,† and John C. Vickerman† Manchester Interdisciplinary Biocentre, School of Chemical Engineering and Analytical Science, The UniVersity of Manchester, Manchester, M1 7DN, U.K., and Department of Chemistry, Penn State UniVersity, 104 Chemistry Building, UniVersity Park, PennsylVania 16802 ReceiVed April 9, 2008 Time of flight secondary ion mass spectrometry (ToF-SIMS) and the Langmuir-Blodgett (LB) technique have been used to create and analyze reproducible membrane mimics of the inner and outer leaflets of a cellular membrane to investigate lipid-protein and lipid-lipid interactions. Films composed of phospholipids, cholesterol and an integral membrane protein were utilized. The results show the outer membrane leaflet mimic (DPPC/cholesterol/glycophorin A LB film) consisting of a single homogeneous phase whereas the inner membrane leaflet mimic (DPPE/cholesterol/ glycophorin A LB film) displays heterogeneity in the form of two separate phases. A DPPE/cholesterol phase and a glycophorin A phase. This points to differences in membrane domain formation based upon the different chemical composition of the leaflets of a cell membrane. The reliability of the measurements was enhanced by establishing the influence of the matrix effect upon the measurement and by utlilizing PCA to enhance the contrast of the images.

Introduction The cell membrane comprises the surface of all living cells. It is formed of a fluid lipid bilayer in which embedded proteins carry out a myriad of functions including acting as enzymes, ion pumps, transport proteins and receptors for hormones. The bilayer assembly regulates the entry and exit of most solutes and ions, with few substances being able to diffuse through unaided.1 The cell membrane is a very dynamic structure whose behavior is often described by a fluid mosaic model2 whereby all lipid or protein molecules in the biological membrane diffuse more or less freely as a two-dimensional liquid. As a direct consequence, both types of molecules would be expected to be randomly distributed within the membrane. More recent experiments suggest the situation is more complex due to the occurrence of both a transverse and lateral regionalization within the bilayers. The observation of micro- and macrodomains is widespread.3 Recent studies have focused upon lipids in the cellular membrane, with special emphasis placed upon elucidating their role in transport and signaling within the cell via organized lipid domains. There are many diseases that are thought to utilize cellular domains. The exit of HIV from a cell depends upon membrane rafts which contain HIV spike proteins4 and the study of Alzheimer’s disease has shown that lipid rafts are involved in protein regulation and trafficking.5 The complexity of membranes associated with live cells makes it difficult to acquire meaningful information about these domains. Model systems however, such as those fabricated using the Langmuir-Blodgett * To whom correspondence should be addressed. E-mail: M.J.Baker@ manchester.ac.uk. Phone: +44(0)161 306 4440. † The University of Manchester. ‡ Penn State University. (1) Lawrence E., Henderson’s Dictionary of Biological Terms, 11th ed.; Longman: Harlow, England, 445. (2) Singer, S. J.; Nicholson, G. L. Science. 1972, 175, 720–731. (3) Tocanne, J. F.; Cezanne, L.; Lopez, A.; Piknova, B.; Schram, V.; Tournier, J. F.; Welby, M. Chem. Phys. Lipids 1994, 73, 139–158. (4) Simons, K.; Ehehalt, R. J. Clin. InVest. 2002, 110(5), 597–603. (5) Simons, M.; Keller, P.; Dichgans, J.; Schulz, J. B. Neurology 2001, 57, 1089–1093. (6) McQuaw, C. M.; Sostarecz, A. S.; Zheng, L.; Ewing, A. G.; Winograd, N. Langmuir. 2005, 21, 807–813.

(LB) technique, are useful in decreasing the complexity by creating well defined, reproducible membrane mimics.6 Recent experiments have utilized time-of-flight secondary ion mass spectrometry (ToF-SIMS) to examine LB films, lipid interactions and the process of domain formation in cellular membranes.7-11 These studies utilize the chemical imaging ability and the high surface sensitivity of ToF-SIMS to show that it is an excellent tool to examine model membrane systems. So far, the work conducted with ToF-SIMS and LB model membrane systems has been largely limited to films containing phospholipids (such as dipalmitoylphosphatidylethanolamine (DPPE), dipalmitoylphosphatidylcholine (DPPC) and sphingomyelin (SM) and cholesterol). An important step for the building of simple but more representative biological membranes is to insert purified integral membrane proteins into well-defined lipid model membranes.12 In one recent experiment, a ternary LB monolayer film consisting of DPPC, dipalmitoylphosphatidylglycerol (DPPG) and surfactant protein B (SP-B) has been examined by ToFSIMS to determine the lipidic interaction partner of SP-B.13 This mammalian pulmonary surfactant is a complex lipid/protein mixture secreted by alveolar epithelial cells. The results show the lipid partner of SP-B to be DPPC rather than the generally accepted DPPG. The mammalian pulmonary surfactant study reveals an unexpected result that is in conflict with other experimental data and is yet to be resolved. To establish that lipid/protein interactions can be highly specific, here we utilize the integral membrane protein, glyco(7) McQuaw, C. M.; Sostarecz, A. S.; Zheng, L.; Ewing, A. G.; Winograd, N. Appl. Surf. Sci. 2006, 252, 6716–6718. (8) Biesinger, M. C.; Paepegaey, P. Y.; McIntyre, N. S.; Harbottle, R; Petersen, N. O. Anal. Chem. 2002, 74, 5711–5716. (9) Bourdos, N.; Kollner, F.; Benninghoven, A.; Ross, M.; Sieber, M; Gall, H. J. Biophys. J. 2000, 79, 357–369. (10) Harbottle, R. R.; Nag, K.; McIntyre, N. S.; Possmayer, F; Petersen, N. O. Langmuir 2003, 19, 3698–3704. (11) Ross, M.; Steinem, C.; Galla, H. J.; Janshoff, A. Langmuir 2001, 17, 2437–2445. (12) Vance, D. E., Vance J. E. New ComprehensiVe Biochemistry, Volume 31, Biochemistry of Lipids, Lipoproteins and Membranes; Elsevier: Amsterdam, 1996, 11. (13) Breitensten, D.; Batneburg, J. J.; Hagenhoff, B.; Galla, J. Biophys. J. 2006, 91, 1347–1356.

10.1021/la802582f CCC: $40.75  2008 American Chemical Society Published on Web 09/13/2008

11804 Langmuir, Vol. 24, No. 20, 2008

phorin A, to determine whether the probability of domain formation depends upon the nature of the surrounding lipid. Model membrane systems consisting of phospholipids, cholesterol and glycophorin A are constructed to imitate the inner and outer leaflets of a cellular membrane. A well-studied example is the cell membrane of the human erythrocyte (red blood cell), which consists of 43.6% total lipids (32.5% Phospholipid, 11.1% Cholesterol), 49.2% proteins and 7.2% carbohydrates in weight percents.14 The inner leaflet of the human erythrocyte membrane contains mainly phosphatidylethanolamine (PE) and the outer leaflet contains mainly phosphatidylcholine (PC) and SM.15 The DPPC lipid combined with cholesterol and glycophorin A are used to represent the outer leaflet and DPPE combined with cholesterol and glycophorin A are used to represent the inner leaflet. The components of the films are combined in the same ratios as in the erythrocyte membrane. The results show that the DPPC/cholesterol/glycophorin A LB films exhibit a single homogeneous phase, in accord with the fluid mosaic model, whereas the DPPE/cholesterol/glycophorin A LB film exhibits heterogeneity and domains within the LB film. The reliability of the measurements has been considerably enhanced by establishing the influences of lipid mixtures on the SIMS secondary ion yield and by utilizing principal component analysis (PCA) to enhance the contrast of the images. In general, our results suggest that the presence of integral membrane proteins can exert both lateral and longitudinal regionalizations within the construct of the model membrane mimics.

Materials and Methods Materials. DPPC, DPPE, cholesterol (Avanti Polar Lipids, AL, USA) glycophorin A, 16-mercaptohexanoic acid, methanol and chloroform (Sigma-Aldrich, MO, USA) were obtained and used without further purification. LB Film Construction. Single crystal (1 0 0) 3 in. silicon wafers were cut and piranha etched (3:1 H2SO4/H2O2) before further treatment. A layer of 100 Å of Cr followed by 2000 Å of Au was then deposited onto clean silicon as described previously.16 A solution of 1 mM 16-mercaptohexanoic acid in propan-2-ol was used to form self-assembled monolayers on gold. The LB films were prepared using a Kibron µ Trough S-LB (Helsinki, Finland) with an aqueous subphase of 70 mL Milli-Q purified room temperature water. The final resistivity of the water was 18.2 MΩ with a total organic content of less than 5 ppb. All lipid and protein mixtures were dissolved in a 9:1 chloroform/ methanol solution. After application the film was exposed to air for 20 min to ensure complete solvent evaporation. Trough barriers were computer controlled to allow uniform compression and constant feedback when depositing monolayers. Surface pressure-area isotherms were obtained. Films were deposited vertically onto a SAM substrate at 5 mN m-1 and the pulling rate was 3 mm min-1. In this model system, the phospholipid content of the LB films is represented by one phospholipid (DPPC for the outer leaflet and DPPE for the inner leaflet) and the protein content is represented by the single integral membrane protein glycophorin A. The LB films were prepared not only as reference materials for future study but also to imitate the inner and outer leaflets of the cellular membrane. Five films were prepared; DPPC/cholesterol; DPPE/cholesterol; glycophorin A; DPPC/cholesterol/glycophorin A and DPPE/ cholesterol/glycophorin A. The components of DPPE/cholesterol/ glycophorin A and DPPC/cholesterol/glycophorin A were combined in the correct ratio to imitate the erythrocyte membrane. This procedure results in the synthesis of a film imitating the outer leaflet with a molar ratio of DPPC 60%:Cholesterol 38%:Glycophorin A (14) Yawata, Y. Cell membrane: the red blood cell as a model; Wiley: New York, 2003. (15) Rothman, J. E.; Lenard, J. Science 1977, 195, 743–753. (16) Fisher, G. L.; Hooper, A. E.; Opila, R. L.; Allara, D. L.; Winograd, N. J. Phys. Chem. B 2000, 104(14), 3267–3273.

Baker et al.

Figure 1. Surface pressure-area isotherms for inner and outer membrane leaflet LB mimics. The outer leaflet plot is displaced upward by ∼0.5 nN m-1 for purposes of visual clarity.

2% and a film imitating the inner leaflet with a molar ratio DPPE 65%:Cholesterol 33%:Glycophorin A 2%. The surface pressurearea isotherms for these two films are shown in Figure 1. The surface pressure area isotherms for the single components (not shown) match those reported earlier.6,7 These plots confirm the purity and concentration of the solutions used to make the films. The phase changing point for the DPPE ternary film is ∼52 Å2 and for the DPPC ternary film is ∼57 Å2, and is found to be reproducible to less than 1 Å. Monolayer deposition was confirmed by ellipsometry. The structure of the supported LB lipid monolayer is expected to be the same as at the air-water interface.9,10 Sample Preparation for Matrix Effect Study. Cholesterol, DPPE, DPPC and glycophorin A were obtained from Sigma-Aldrich and used without further purification. Solutions in chloroform were prepared in the 3 component LB films (DPPE 65%:Cholesterol 33%: Glycophorin A 2% and DPPC 60%:Cholesterol 38%:Glycophorin A 2%). The SIMS information was obtained from the single components as well as the mixed component films. Sample preparation consists of pipetting 5 µL of each solution was onto a silicon wafer by spin casting for 30 s to ensure uniform solvent evaporation. ToF-SIMS Analysis. The mass spectral imaging was carried out on a specially constructed SIMS instrument described elsewhere.17 Analysis was performed with a 20 keV Ga+ primary ion beam system (Ionoptika Ltd., UK). Secondary ions were analyzed in a two-stage reflectron mass spectrometer (Kore Technology Ltd., U.K.). The primary ion dose density did not exceed 1 × 1012 ions cm-2. The ternary mixtures for the matrix effect experiments were analyzed with a 20 keV Au+ primary ion beam system. No evidence for sample charging was observed in any of these experiments. Data Analysis. Initial image processing was performed using specially constructed software. Principal component analysis was performed using software written using Matlab. Principal component analysis is an unsupervised technique which is used to reduce the number of variables used to represent a complex data set with minimal loss of information, to identify relationships between variables and to identify relationships between samples. In the case of SIMS image analysis, the ion peaks are considered as variables and the image pixels as samples.18 A comprehensive review and explanation of the imaging PCA technique has been published recently.19 (17) 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– 1252. (18) Tyler, B. J. Appl. Surf. Sci. 2006, 252, 6875–6882. (19) Tyler, B. J.; Rayal, G.; Castner, D. G. Biomaterials 2007, 15, 2412–2423.

Mass Spectral Imaging in Model Cell Membranes

Langmuir, Vol. 24, No. 20, 2008 11805

Table 1. Diagnostic Fragment Ions from Cholesterol, DPPC and DPPE That Are Present in DPPC/Cholesterol and DPPE/ Cholesterol Films DPPE and DPPC DPPE and DPPC cholesterol cholesterol DPPC DPPC DPPC DPPC DPPC DPPC DPPE DPPE

Table 2. Common Mass Spectral Ions of Amino Acids Found in Glycophorin A Compiled from References.22-26 Formulas Are Shown Where Given

dipalmitoyl [C35H67O4]+

551.50

alanine (Ala)

palmitoyl [C19H37O3]+

313.27

arginine (Arg)

cholesterol [M - OH]+ cholesterol [M - H]+ phosphocholine headgroup [C8H19NPO4]+ phosphocholine headgroup [C5H15NPO4]+ phosphocholine headgroup [C5H13NPO3]+ phosphocholine headgroup [C5H14NO]+ phosphocholine headgroup [C5H12N]+ DPPC [M + H]+ phosphoethanolamine headgroup [C2H9NPO4]+ phosphoethanolamine headgroup [C2H7NPO3]+

369.35 385.35 224.11 184.07 166.06 104.11 86.10 734.56 142.03 124.02

Results and Discussion The goal of this study was to investigate lateral and transverse regionalizations of an integral membrane protein in the cell membrane. To study this we utilize inner and outer membrane mimics based upon the asymmetry of phospholipids in the cell membrane. Reference LB films of each component were analyzed to identify diagnostic fragment ions and to investigate the degree to which the matrix effect inherent to the system under analysis might influence the observed data. Once this has been achieved the measurements on the inner and outer membrane leaflets were analyzed and any resulting domains verified in light of the matrix effect results. PCA was then performed on the mass spectral images to increase the contrast of the images and refine the peak assignments associated with the domains. Reference LB Films. To establish expected secondary ion intensities and to determine any interaction between the components that might lead to artifacts, a series of control films consisting of lipid and cholesterol were examined in detail as summarized in Table 1. The representative peaks chosen are consistent with those used in earlier studies.20,21 In this study the fatty acid tail group at m/z 551.50 and PE headgroup at m/z 142.03 is utilized to identify the presence of DPPE in the inner leaflet mimic. The PC headgroup at m/z 184.07 identifies DPPC in the outer leaflet mimic and m/z 369.35 is utilized to monitor the distribution of cholesterol. Protein molecular ions or even large peptide fragments are not usually observable in SIMS. A number of papers have shown that particular ions and their behavior can be followed via the immonium ions and other fragments from their constituent amino acids.27,28 Glycophorin A is composed of glutamic acid (Glu), isoleucine (Ile), proline (Pro), serine (Ser), threonine (Thr), valine (Val), alanine (Ala), leucine (Leu), arginine (Arg), asparagine (Asn), aspartic Acid (Asp), glutamine (Gln), glycine (Gly), histidine (His), lysine (Lys), methionine (Met), phenylalanine (Phe) and tyrosine (Tyr). Glycophorin A contains no cysteine (Cys) or tryptophan (Trp). Amino acid common mass spectral ions compiled from refs 22-26 are shown in Table 2. The ToF(20) McQuaw, C. M.; Zheng, L.; Ewing, A. G.; Winograd, N. Langmuir 2007, 23, 5645–5650. (21) Ostrowski, S. G.; Szakal, C.; Kozole, J.; Roddy, T. P.; Xu, J.; Ewing, A. G.; Winograd, N. Anal. Chem. 2005, 77(19), 6190–6196. (22) Kulp, K. S.; Berman, S. F.; Knize, M. G.; Shattuck, D. L.; Nelson, E. J.; Wu, L.; Montgomery, J. L.; Felton, J. S.; Wu, K. J. Anal. Chem. 2006, (78), 3651–3658. (23) Sanni, O. D.; Wagner, M. S.; Briggs, D.; Castner, D. G.; Vickerman, J. C. Surf. Interface Anal. 2002, 33, 715–728. (24) Wagner, M. S.; Castner, D. G. Langmuir 2000, 17, 4649–4660. (25) Jochims, H. W.; Schwell, M.; Chotin, J. L.; Clemino, M.; Dulieu, F.; Baumgartel, H.; Leach, S. Chem. Phys. 2004, 298, 279–297. (26) Static SIMS Library Version 4.0.1.35, Surface Spectra Ltd 19992006.

asparagine (Asn) aspartic acid (Asp) glutamic acid (Glu) glutamine (Gln) glycine (Gly) histidine (His) isoleucine (Ile) leucine (Leu) lysine (Lys) methionine (Met) phenylalanine (Phe) proline (Pro) serine (Ser) threonine (Thr) tyrosine (Tyr) valine (Val)

44 [C2H6N]+,57, 89 [NH2CH3 CHCOOH]+, 81, 90, 112, 134, 179 43 [CH3N2]+, 55, 57, 59, 70, 73 [C2H7N3]+, 100 [C4H10N3]+, 101 [C4H11N3]+, 112, [C5H10N3]+, 127 [C5H11N4]+, 175 56, 70 [C3H4NO]+, 87 [C3H7N2O]+, 88 [C3H6NO2]+, 98 [C4H4NO2]+, 133, 155, 177 57, 88 [C3H6NO2]+, 134, 155, 219, 279 70, 84 [C4H6NO]+, 102 [C4H8NO2]+, 130, 148, 175, 56, 84 [C4H6NO]+, 101, 130, 147, 279 30 [CH4N]+, 57, 76, 98, 120, 178, 200 55, 69, 81 [C4H5N2]+, 82 [C2H6N2]+, 95, 110 [C5H8N3]+, 156 56, 58, 69, 86 [C5H12N]+, 132, 263 55, 57, 70, 86 [C5H12N]+, 132, 263 55, 56, 69, 84 [C5H10N]+, 104, 147 56, 61 [C2H5S]+, 91, 120, 150, 166 51, 77, 91, 103, 120 [C8H10N]+, 131 [C9H7O]+, 166 68 [C4H6N]+, 70 [C4H8N]+, 116, 117, 138, 231 57, 60 [C2H6NO]+, 71 [C3H3O2]+, 91, 116, 106, 128 56, 57, 69 [C4H5O]+, 74 [C3H8NO]+, 116, 120, 239 107 [C7H7O]+, 116, 123, 136 [C8H10NO]+, 165, 182 55, 57, 59, 72 [C4H10N]+, 83 [C5H7O]+, 118, 235

SIMS spectra of the positive ions of glycophorin A, DPPE/ cholesterol and DPPC/cholesterol LB films with SIMS peaks attributable to their components (Tables 1 and 2) are shown in Figure 2. Inspection of these tables and figures show that to monitor the presence of GpA with a unique set of ions is not straightforward. A number of the amino acid immonium ions have isobaric interferences with ions from the lipids also present in the systems under study. Initially it was thought that m/z 59, 70, 71, 72, 84 and 102 reflecting immonium and other ions from valine, proline, serine and glutamic acid would provide a unique set as these 6 amino acids make up 57.9% of the composition of glycophorin A. However as will be seen, the subsequent imaging study of the DPPE/cholesterol/glycophorin A film suggested the formation of two domains, one seemed to be exclusively DPPE and cholesterol and the other was characterized by m/z 59 and 72 while the other peaks were uniform across the whole film. This suggested that the signals at m/z 70, 71, 84 and 102 were not uniquely characteristic of a single component. A PCA study (see later for details) showed that one domain was characterized by the peaks characteristic of DPPE and cholesterol while peaks at m/z 45, 59, 72, 89 and 144 distinguished the other domain. Utilizing PCA and interrogation of the raw data we find that m/z 59, 72 and 89 are attributable to valine, alanine and arginine. Our hypothesis would be that these are the peaks in this system that are unique to glycophorin A. The m/z 89 peak [NH2CH3CHCOOH]+ is a fragment ion attributable to alanine.25,26 Of course we cannot be absolutely sure that there are no contributions from the other components, but on the basis of this suggestion the system appears self-consistent. We have (27) Wagner, M. S.; Horbett, T. A.; Castner, D. G. Langmuir 2003, (5), 1708– 1715. (28) Wagner, M. S.; Tyler, B. J.; Castner, D. G. Langmuir 2002, (8), 1824– 1835.

11806 Langmuir, Vol. 24, No. 20, 2008

Baker et al.

Figure 2. ToF-SIMS spectra of (A) glycophorin A LB film, (B) DPPE/cholesterol LB film and (C) DPPC/cholesterol LB film with SIMS peaks attributable to their components labelled. The spectra are all collected under the same operating conditions, described in the text, and as such, the intensity scales are comparable.

therefore used these peaks as characteristic of glycophorin A. Assuming there are no conformation effects that might hide particular amino acid components at the film surface from interrogation by the SIMS primary ion beam, the intensity of any one or the sum of these peaks can be used to monitor the concentration of the protein in the mixed system relative to a pure protein film. Evaluation of Possible Matrix Effects. Previous studies have shown that in multi - component systems involving lipids the secondary ion yields from each component strongly depend upon

the chemical composition of the film. For example it has been reported that the presence of PC suppresses the ion intensity of PE.21 To test for these effects ToF-SIMS spectra were obtained from the pure samples of DPPC, DPPE, cholesterol and glycophorin A and compared to the SIMS spectra from ternary mixtures. Secondary ion intensities of diagnostic fragment ions were determined and converted into an effective secondary ion yield by normalizing to the incident beam current. The headgroup fragment m/z 184 was used to identify DPPC, m/z 369 to identify cholesterol and headgroup fragment m/z 142 to identify DPPE.

Mass Spectral Imaging in Model Cell Membranes

Langmuir, Vol. 24, No. 20, 2008 11807

Table 3. Secondary Ion Yields from Pure Films, Expected Secondary Ion Yield for the DPPC and DPPE Ternary Mixtures and Measured Secondary Ion Yields for the DPPC and DPPE Ternary Mixtures yield of pure compounda

expected yield DPPC ternary mixtureb

measured yield DPPC ternary mixture

184 (DPPC)

1.41 × 10 -4

8.46 × 10-5 (60%)b

1.30 × 10-4

142 (DPPE)

3.84 × 10-6

369 (cholesterol) amino acid sum (glycophorin A)

4.91 × 10-6 3.51 × 10-5

ion m/z (origin)

1.87 × 10-6 (38%) 7.02 × 10-7 (2%)

3.98 × 10-6 2.91 × 10-5

expected yield DPPE ternary mixtures

measured yield DPPE ternary mixture

2.50 × 10-6 (65%)

3.84 × 10-6

1.62 × 10-6 (33%) 7.02 × 10-7 (2%)

3.22 × 10-6 4.92 × 10-5

a Yield is determined as the number of secondary ions per incident primary ion. b Percentages represent the molar percent of each component in the ternary film.

Figure 3. (A) Expected secondary ion yield for DPPC, cholesterol and glycophorin A based upon pure yield and molar ratio (blue) and the measured yield observed from the DPPC ternary mixture (red). (B) The expected secondary ion yield for DPPE, cholesterol and glycophorin A based upon pure yield and molar ratio (blue) and the measured yield observed from the DPPE ternary mixture (red).

Figure 4. Mass spectral images of glycophorin A LB film (total ion ) 0-369 counts, m/z 59 ) 0-8 counts, m/z 72 ) 0-5 counts, m/z 89 ) 0-3 counts).

To identify glycophorin A the sum of m/z 59, 72 and 89 was utilized. Secondary ion yields obtained from the pure compounds and the secondary ion yields expected for the ternary mixtures based upon molar ratio percentage are shown in Table 3. These numbers are compared with the measured values, also shown in Table 3. The expected secondary ion yields (blue) and the measured secondary ion yields (red) for the DPPC and DPPE ternary mixtures are shown in Figure 3. For the DPPC and DPPE ternary mixtures none of the components are being suppressed. There is a small yield enhancement of DPPE, DPPC and cholesterol in the two ternary mixtures and larger yield enhancement of GpA in both ternary mixtures. These results show that the

Figure 5. Mass spectral images of DPPC/cholesterol/glycophorin A LB film (total ion ) 0-247 counts, m/z 184 ) 0-12 counts, m/z 369 ) 0-5 counts, m/z 59 ) 0-4 counts, m/z 72 ) 0-4 counts).

detection of all three components of each ternary mixture is possible when they are all present in a chemical environment that contains them. Glycophorin A Reference LB Film. ToF-SIMS chemical images of the glycophorin A film are shown in Figure 4. The images depict a single homogeneous phase for glycophorin A based upon diagnostic fragment ions attributable to valine,

11808 Langmuir, Vol. 24, No. 20, 2008

Figure 6. Mass spectral images of the DPPE/cholesterol/glycophorin A LB film (total ion ) 0-382 counts, m/z 551.5 ) 0-3 counts, m/z 369 ) 0-9 counts).

Figure 7. Mass spectral images of the DPPE/cholesterol/glycophorin A LB film showing peaks attributable to amino acids. (m/z 59 ) 0-5 counts, m/z 70 ) 0-5 counts, m/z 71 ) 0-10 counts, m/z 72 ) 0-4 counts, m/z 84 ) 0-4 counts, m/z 89 ) 0-16 counts m/z 102 ) 0-3 counts).

Figure 8. Principal component 1 scores plot for the DPPE/cholesterol/ glycophorin A. The total ion image associated with this data is shown in Figure 6.

arginine and glutamic acid. Images from this reference film illustrates that it is possible to construct an LB film from glycophorin A and detect amino acid fragments from the film. Defects in the silicon, deposition of the gold layer or addition of the SAM layer can lead to defects observable on the micron scale. This will affect the production of SIMS signal in the particular area of the defect. However as can be seen from Figure 4 total ion image the defects are small and do not affect the monolayer formation on surrounding areas. Outer Membrane Leaflet LB Mimic (DPPC/Cholesterol/ Glycophorin A). With mass spectra and matrix effects of each component known, it is feasible to acquire ToF-SIMS images

Baker et al.

of the synthesized membrane mimics. Images from the DPPC/ cholesterol/glycophorin A LB film for each characteristic mass are shown in Figure 5. No domain formation is observable from the total ion image. The peaks from the phospholipid component at m/z 184.07, the cholesterol component at m/z 369.35 and valine/ arginine related ions from Glycophorin A components at m/z 59.07 and 72.08 are also uniform in coverage. The overlay image represents peaks from all three components in their respective colors summed to represent lateral distribution. From all of these representations the DPPC/cholesterol/glycophorin A LB film appears to form a single homogeneous phase on the scale observable by ToF-SIMS. The homogeneity of the films has been confirmed with repeat measurements on two outer membrane leaflet LB mimic films. Inner Membrane Leaflet LB Mimic (DPPE/Cholesterol/ Glycophorin A). The ToF-SIMS images of the inner leaflet mimic are shown in Figure 6. The distribution of the dipalmitoyl tailgroup peak at m/z 551.5 exhibits clear heterogeneity. Moreover, the PE fragments at m/z 142.03 (not shown) and 124.02 (not shown) and the cholesterol peaks at m/z 369.35 and 385.35 (not shown) exhibit an identical pattern. This information suggests there is a co-localization of DPPE and cholesterol, with voids of signal throughout the image that are not occupied by these species. The mass spectral images of the DPPE, cholesterol and glycophorin A LB film of SIMS peaks m/z 59, 70, 71, 72, 84 and 102 are shown in Figure 7. The images show that only some of these expected amino acid peaks show heterogeneity throughout the film. This illustrates the issue referred to earlier, that because of the presence of mass interference from the three components of the film only m/z 59, 72 and 89 which can be attributed to valine and alanine show heterogeneity throughout the film and appear in the opposite spaces to the DPPE and cholesterol signal whereas m/z 70, 71, 84 and 102 show signal from across the images, these peaks are observable in the DPPE and cholesterol LB film. The three components of the DPPE, cholesterol and glycophorin A film are clearly visible in Figures 6 and 7. The mass spectra images show that DPPE and cholesterol are colocated and anticorrelated to areas which we have argued are characteristic of protein. If our analysis is correct the heterogeneity reveals that the inner membrane film contains two separate phases; a glycophorin A phase and a DPPE and cholesterol phase on the scale observable by ToF-SIMS. The heterogeneity of the film has been confirmed by repeat measurements on the two inner membrane leaflet LB film mimics. Principal Component Analysis. To refine these assignments PCA is employed to determine which peaks distinguish these domains and to increase the contrast of the images. This analysis was performed on the two total ion images previously shown in Figures 5 and 6. The mass range 1-1000 Da was utilized to include the major peaks from all three components of the LB films. The scores plot for principal component 1 of the SIMS total ion image for the DPPE/cholesterol/glycophorin A LB film in Figure 6 is shown in Figure 8. Each principal component score has an associated loading which shows the SIMS peaks responsible for the discrimination observed in the image. The loadings, as shown in Figure 9, have a positive and negative direction on the y axis. The peaks observable in these directions correspond to the color scale at the side of the image. If there is a high concentration of peaks in the positive direction of the y axis this is depicted as yellow through to red. If there is a high concentration of peaks in the negative direction of the y axis in the image this area is depicted

Mass Spectral Imaging in Model Cell Membranes

Langmuir, Vol. 24, No. 20, 2008 11809

Figure 9. Loading plots for principal component 1 (A) m/z 1-1000, (B) m/z 0-200, (C) m/z 360-400 and (D) m/z 545-560.

Figure 10. Principal component 1 scores plot for the DPPC/cholesterol/ glycophorin A total ion image shown in Figure 4B.

as shades of blue. This method allows visualization of the areas of the film with different surface chemistry and provides mass spectral information about the discrimination. During PCA, the spectra of the images are binned to nominal mass (1 amu) to ease computational demands. This means that the dipalmitoyl tailgroup peak at m/z 551.5 appears at m/z 552 in the loadings, as seen in Figure 9D. This identification has been verified as the dipalmitoyl tailgroup peak at m/z 551.5 from the original data. The peaks in the negative direction are responsible for the blue areas on the principal component 1 scores image, the peaks responsible originate from DPPE (m/z 552), cholesterol (m/z 369 and 385) and contributions from low molecular mass fragments that can be attributed to hydrocarbons. None of the loadings in the positive direction are characteristic of DPPE or cholesterol and as we have argued earlier they can be attributed to glycophorin A. m/z 59, 72 and 89 are attributed to valine and alanine and the identified peaks localize to the same area of the image. These mass spectral ions appear in the original data for the glycophorin A LB film and localize to the same area of the

image as the identified amino acid peaks in the DPPE, cholesterol and glycophorin A LB film (Figure 7). Hence, the mass spectral ions identified by PCA can be traced back to ions identified as originating from glycophorin A for one domain and from cholesterol and DPPE for the other domain. It is interesting to compare the PCA scores of the heterogeneous DPPE ternary film to the more uniform DPPC ternary film. The scores plot for principal component 1 of the SIMS total ion image for the DPPC/cholesterol/glycophorin A LB film in Figure 5 is shown in Figure 10 and the loadings plots for principal component 1 is shown in Figure 11. The scores image shows no evidence for any distinct protein domains. The PCA analysis of the DPPC/cholesterol/glycophorin A LB film supports the conclusion drawn from the original SIMS data that the DPPC/ cholesterol/glycophorin A LB film exists as a single phase. The negative loadings m/z 73 and 147 in Figure 11 relate to the contaminant polymethylsiloxane. Figure 2 showed that this is a contaminant of the protein samples. SIMS is very sensitive to this contaminant and the signal intensity does not indicate that the contamination is high, however it might be argued that it could stimulate or influence domain formation among the three main film components. It clearly does not in the DPPC/cholesterol/ glycophorin A film because the only heterogeneity is due to the presence of areas of PDMS. In the case of the DPPE/cholesterol/ glycophorin A film the PDMS ions do not show up as distinguishing peaks for either domain.

Conclusions The LB film outer leaflet membrane mimic shows the existence of a single homogeneous phase across the DPPC/cholesterol/ glycophorin A LB film and the inner membrane leaflet displays two phases, a DPPE/cholesterol phase and a glycophorin A phase. These results suggest there is a transverse regionalization of glycophorin A between the inner and outer leaflet of a cell membrane. This behavior may have implications for the fluid mosaic model whereby all lipid and protein molecules are suggested to diffuse more or less freely in the cell membrane. Quantitative analysis of the precise chemical composition of the observed domains is influenced by the presence of matrix

11810 Langmuir, Vol. 24, No. 20, 2008

Baker et al.

Figure 11. Loading plots for principal component 1 (A) m/z 1-1000, (B) m/z 50-150.

effects. Our investigations show ToF-SIMS detection of lipid and protein signals is possible when they are combined together. Thus our interpretation of the ToF-SIMS images in terms of homogeneous DPPC/cholesterol/glycophorin A LB films and a two domain system in the DPPE/cholesterol/glycophorin A LB films should not be compromised by matrix effects. These two monolayer films differ only by the chemical nature of the phospholipid headgroup, yet the distribution of membrane protein is dramatically different. The choline headgroup consist of an -N(CH3)3, while the ethanolamine terminates with a -NH3 group. We speculate that this different chemistry allows strong DPPE-DPPE hydrogen bonding interactions that encourage tight structural integrity, opening the possibility of protein exclusion.5 The phosphocholine molecules do not exhibit this attraction and are free to mix with the various components in the film. At this point, it is unclear whether these same factors drive asymmetry in a real biological membrane since much more complex chemistry is involved. For example, recent studies of a ternary mixture of cholesterol, SM and palmitoyloleoylphosphatidylcholine (POPC) show that the degree of tailgroup saturation is important in determining the composition and degree of domain

formation.20 ToF-SIMS has been shown as a sensitive technique for the analysis of lateral and longitudinal regionalizations in cell membrane mimics. In this investigation we have demonstrated that such sensitivity can be utilized to identify a transverse regionalization of an integral membrane protein that is dependent upon phospholipid composition of the membrane mimic. Acknowledgment. The authors acknowledge the World University Network, Engineering and Physical Sciences Research Council (EPSRC) under Grant No. GR/T05646/01 and the EPSRC Life Sciences Interface under Grant No. EP/C008251/1 for funding to carry out this research and Dr. Alex Henderson for his help and expertise on the multivariate image analysis. Financial support from the National Institute of Health under Grant No. EB002016-13 and the National Science Foundation under Grant No. CHE-555314 are also acknowledged. Supporting Information Available: Experimental details. This material is available free of charge via the Internet at http://pubs.acs.org. LA802582F