Differentiation between Single Bladder Cancer Cells Using Principal

Article pubs.acs.org/ac

Differentiation between Single Bladder Cancer Cells Using Principal Component Analysis of Time-of-Flight Secondary Ion Mass Spectrometry Justyna Gostek,† Kamil Awsiuk,‡ Joanna Pabijan,† Jakub Rysz,‡ Andrzej Budkowski,‡ and Malgorzata Lekka*,† †

The Henryk Niewodniczanski Institute of Nuclear Physcis, Polish Academy of Sciences, Radzikowskiego 152, 31-342 Cracow, Poland ‡ The Marian Smoluchowski Institute of Physics, Jagiellonian University, Reymonta 4, 30-059 Cracow, Poland S Supporting Information *

ABSTRACT: Time-of-flight-secondary ion mass spectrometry (TOF-SIMS) mass spectra measurements combined with an appropriate sample preparation protocol are the powerful tools to obtain unique information about the chemical composition of biological materials. In our studies, two questions were addressed, i.e., whether it is possible to develop a fixative-based sample preparation protocol and whether it allows one to distinguish between cells originating from various stages of cancer progression. Therefore, four human bladder cancer cell lines (with distinct malignancy degree) have been investigated. A chemical fixation protocol has been used for TOF-SIMS measurements, and mass spectra were obtained using a Bi3+ primary ion beam. The principal component analysis (PCA) has been applied to analyze the whole range of mass spectra (without preselection of any particular masses) using two approaches of data preprocessing, namely, mean centering and autoscaling. The PC3 versus PC2 plot has showed significant differences between nonmalignant cancer cells and the cancerous ones for both of preprocessing approaches. The analysis of mass spectra of human bladder cells allows one to find a list of mass peaks with intensities significantly larger in cancerous bladder cells compared to nonmalignant cell cancer of the ureter (HCV29 cells). These findings show that TOF-SIMS in combination with PCA can be used to identify reference, human bladder cells from cancerous ones.

D

clusters, and whole molecules from its surface. Most of the emerging particles are neutral, but some of them are ionized and can be analyzed with the mass spectrometer. Therefore, it is possible to obtain the information on chemical composition from very superficial layers of a sample surface (i.e., less than several dozen of nanometers). Thickness of a probing layer seemed to be a potential advantage considering the fact that the majority of cancer-related alterations are associated with changes in the surface receptor expression and composition.7 The TOF-SIMS providing the information on chemical composition seems to be a technique capable of detecting changes related to oncogenic transformation. One of the first studies on chemical composition of cancerous cells have been presented by Gazi et al.8 In this work, the TOF-SIMS technique has been used in imaging mode to localize various species that contain K, Ca, and Mg ions. The obtained results complemented those obtained from Fourier transform-infrared (FT-IR) spectroscopy. The findings

espite much research in bladder cancer, it is still difficult to predict tumor progression, optimal therapy, and clinical outcome for this type of cancer.1 Usually, bladder cancer develops by a multistage process with the symptoms, which, at the initial stage, are very similar to the common infections leading to misleading diagnosis. As a result, most of the bladder cancer patients die from the invasive, metastatic transitional cell carcinoma.2 Therefore, early diagnosis of bladder cancer is a prerequisite for a good prognosis. Cancer progression is a multistep process leading to alterations in cellular cytoarchitecture and biophysical properties that are associated with biochemical changes.3,4 Various studies on cancer progression show a wide range of changes in chemical composition leading to a conclusion that a general, single change of a specific molecular compound does not exist.5 The characterization of cancer-related changes in a quantitative manner with respect to normal, benign, and cancer cells defines a new target area for novel diagnostic approaches. Recent works have shown that time-of-flight secondary ion mass spectrometry (TOF-SIMS) is a widely applied technique in materials science.6 In this method, a sample is bombarded by the primary ion beam that induces desorption of atoms, © 2015 American Chemical Society

Received: August 6, 2014 Accepted: February 17, 2015 Published: February 17, 2015 3195

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201

Article

Analytical Chemistry showed that any molecular compounds, present at millimolar concentrations within the cytoplasm, can be detected by the TOF-SIMS technique. Two years later, Kulp et al.9 focused on the differentiation between three human breast cancer cell lines (MCF-7, T47D, and MDA-MB-231). They characterized rare chemical changes in breast cancer cell lines that enabled their differentiation based on TOF-SIMS spectra recorded for cell homogenates. The differentiation between different fractions of cell homogenates was obtained using statistical means for data reduction. These results lead to other studies on the applications of the TOF-SIMS technique to discriminate cancerous cells from reference ones. The comparison between human prostate cancer types originating from metastasis to bone (PC-3), to lymph node (LNCaP) and nonmalignant normal adult prostate epithelial cells showed nicely separated group of results attributed to each studied cell line.10 Tissue samples were examined in an analogous way. The TOF-SIMS mass spectra combined with multivariate analysis were used to differentiate between colon cancer and normal colon mucosa,11 followed by other studies on disease-related lipid fingerprints in the mouse model of cystic fibrosis.12 In all presented papers, independent of the sample type (single cells or tissues), the TOF-SIMS measurements employed a relatively complicated preparation method that involves freezing in liquid nitrogen in various ways. Mainly, the sample preparation for TOF-SIMS involves either freezefraction of single cells or frozen tissue slices. In our studies, we addressed two questions. First, whether it is possible to develop a fixative-based sample preparation protocol and, second, whether it allows one to distinguish between cells originating from various stages of cancer progression. In our research, we would like to demonstrate the functionality of the TOF-SIMS technique to study biological samples without using freezingbased methodology for sample preparation. In our data analysis, we apply principal component analysis (PCA) to full mass spectra, without predefining any specific masses. Thus, the nonmalignant epithelial cells of the ureter were compared to three cancerous cell lines, characterized by distinct histological malignancy degree. The recorded mass spectra (up to 800 Da) were analyzed using the PCA approach without focusing on any particular mass. This allowed detection of differences in the chemical composition of the cell membrane surface related to cancer cells characterized by distinct histological grades. The PCA weights all of the available variables to provide the maximum discrimination between data.

atmosphere. The HCV29 and T24 cells were grown in RPMI1640 medium (Sigma) supplemented with 10% fetal bovine serum (FBS, Sigma). The HTB-9 cells were grown in RPMI supplemented with 10% FBS, 1% HEPES (4-(2-hydroxyethyl)1-piperazineethanesulfonic acid, Sigma), and 1% sodium pyruvate (Sigma). Lastly, the HT1376 cells were grown in Eagle’s medium (EMEM, LGC Standards) supplemented with 10% FBS (LGC Standards). After a few passages, cells were seeded on previously prepared silicon substrates placed in Petri dishes (Sarstedt) and further cultured for 48 h in the corresponding media and culture conditions. Silicon Substrates for Cell Growth. The substrate used for cell growth for TOF-SIMS was a silicon. From commercially available silicon wafers (Si-Mat, Germany), squares with the size of 1 cm × 1 cm were cut. Such prepared substrates were cleared with pressurized nitrogen and sterilized with a UV lamp (1 h on both sides). Preparation of Cells for TOF-SIMS. The TOF-SIMS experiment requires a special treatment of samples, due to vacuum conditions, needed during the measurements.13 Cells, previously cultured on silicon surfaces, were chemically fixed using paraformaldehyde followed by drying with alcohol.14 Briefly, after 48 h of culture, cells attached to the silicon surface were fixed with 3.7% paraformaldehyde dissolved in phosphate buffered saline (PBS, Sigma), for 15 min in the CO2 incubator (NuAire). Then, such prepared samples were washed with various PBS buffers for 1 min. The applied PBS buffers were prepared by diluting PBS in deionized water (Cobrabid purification system, 18 MΩ cm) at ratios 1:2 and 1:4. Next, the samples were washed in deionized water alone for 1 min and they were consecutively moved to alcohol substitutions (anhydrous ethyl alcohol, 99.8%, Avantor Performance Materials Poland S.A.). These steps aimed at the sample dehydration. For this purpose, six solutions of ethyl alcohol in deionized water were prepared, starting from 40% up to 90%, every 10%. Silicon wafers with cells were immersed in them in sequence for 30 s. The final rinsing was performed in pure anhydrous ethyl alcohol. Next, samples were transferred into the lock-in chamber of the SIMS apparatus in a vacuum sealed vessel to maintain a sterile environment. All protocol steps were completed under the laminar flow chamber (NuAire) to provide sterile conditions. Also, for a single experiment, all samples were prepared in one session to avoid any differences stemming from the preparation process. As reference, bare silicon surfaces, treated in an analogous way as cells (i.e., sterilized, cultured without cells, and dehydrated) were used. TOF-SIMS Measurements. In all experiments, TOF-SIMS 5 apparatus (ION-TOF GmbH, Munster) equipped with a 30 keV bismuth liquid metal ion gun was used. High-resolution m/ z mass spectra of secondary ions emerging from samples bombarded with the Bi3+ ion beam were acquired with a timeof-flight mass spectrometer. In the TOF-SIMS measurements, two types of samples were measured, namely, (1) silicon surface without cells (blank samples) and (2) silicon surface with cells. To probe single cells, the area of 150 μm × 150 μm was chosen. The primary beam current and time of acquisition were chosen to maintain static conditions (primary ion fluency 6500. In a single

■

EXPERIMENTAL SECTION Cell Culture. Four human bladder cancer cell lines were studied: (i) HCV29, nonmalignant epithelial cells of the ureter (established in Fibiger Institute, Copenhagen, Denmark); (ii) HTB-9, urinary bladder carcinoma (ATCC, LGC Standards); HT1376, urinary bladder carcinoma (ATCC, LGC Standards); T24, transitional cell carcinoma of human bladder (ATCC, LGC Standards). Only HCV29 cells represent a nonmalignant cancer whereas three other cell lines are metastatic cells: HTB-9 (histological grade II); HT1376 (histological grade III); and T24, transitional cell carcinoma. Histological grade describes the degree of abnormality in cancerous cells. It can be treated as a measure of differentiation, describing to what extent cancer cells are similar in appearance and function to healthy cells of the same tissue type. The cells were grown in 25 cm2 culture flasks (Sarstedt) in the incubator (NuAire) providing 37 °C and a 95% air/5% CO2 3196

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201

Article

Analytical Chemistry experiment, eight samples with cultured cells (two samples per each cell line) were measured to avoid any instrumental drift. Thus, from a single silicon substrate, mass spectra of five cells were acquired. In total, 10 mass spectra for each cell line were recorded within one experiment. Also, each time a reference surface (i.e., a bare silicon surface undergoing the same protocols as silicon surfaces with cells) was measured. For reference samples, five mass spectra were collected. For the chosen conditions, measurement of one spectrum took no more than 10 min. Principal Component Analysis (PCA). Principal component analysis (PCA) is a well-known method to analyze large data sets.15,16 The basic idea of PCA is to reduce the dimensionality of a data set, while retaining as much as possible the variation present in the original predictor variables. In mathematical terms, PCA sequentially maximizes the variance of a linear combination of the original predictor variables.17 In our case, the PCA was applied to a whole mass spectra acquired from the entire scanned region (150 μm × 150 μm), without specifying any particular mass present within the range between 1 and 500 Da. The upper limit of the mass window has been arbitrary chosen, focusing at the area under each single peak, i.e., mass spectra were analyzed in order to find every fragments visible as a peak in at least one mass spectrum. In such a way, the list of all peaks present in the TOF-SIMS spectra was generated for each measurement. This list was later used for the PCA carried out within the mass window between 1 and 500 Da (832 masses have been identified). The data were normalized to the sum of all peaks in the spectra, and peak areas were used for the PCA. The peak areas were calculated using the SurfaceLab 6 Software provided with the apparatus and organized in a matrix composed of 45 columns (corresponding to each mass spectrum) and 832 rows (corresponding to identified masses). Afterward, the PCA was carried out with the aid of the PLS Toolbox (eigenvector Research, Manson, WA) for MATLAB (MathWorks, Inc., Natick, MA) software. AFM Topography Measurements. To verify whether the preparation protocol applied to the samples influenced the morphology of cells, the surface of dried cells was imaged using AFM working in contact mode (Park Systems XE120 microscope). Images (scan size 90 μm × 90 μm) were recorded using cantilevers (MLCT-D, Bruker) characterized by a nominal spring constant of 0.03 N/m and an open angle of 20°. Cells were imaged in ambient atmosphere with a scan rate of 0.5 Hz and set point of 0.3−0.5 nN.

Figure 1. Error mode images of individual cells recorded for (a) HCV29, (b) HTB-9, (c) HT1376, and (d) T24 human bladder cells.

sections were drawn (Figure 2). The thickness of cells prepared based on the proposed protocol varies within tens (20−25 nm) to hundreds of nanometers (up to 400 nm) for each cell line.

Figure 2. Exemplary image of cross sections guided through different locations on a dried cell.

Taking into account the size of the Bi3+ primary ion beam, the thickness of a layer, from which ions, molecules, or their fragments are removed, and a cell thickness, we can assume that the influence of the underlying silicon substrate is negligible. Effects of the Various Composition of Culture Media. The studied cell lines were cultured in media composed of various constituents required for their normal growth. To check the effect of culture media, the silicon surfaces were exposed to three different buffers used for human bladder cell cultures, i.e., EMEM (used in HT1376 culture), RPMI 1640 (used for HCV29 and T24 cells), and RPMI 1640 supplemented with HEPES and sodium pyruvate (used for HTB-9 culture), for 24 h at 37 °C in an CO2 incubator. After the exposure, Si surfaces underwent the same chemical protocol for sample preparation as cells. As a next step, samples were measured with the TOFSIMS technique. The collected mass spectra were analyzed in the same way as mass spectra recorded for cells (described in the Experimental Section). The exemplary mass spectra of the silicon surface exposed to RPMI 1640 culture medium is shown in Figure 3a. The TOF-SIMS mass spectrum of bare silicon surface, treated in a similar way as silicon with cells, possesses the organic compounds on its surface (Figure 3a), being the result of medium components deposition. Simultaneously, this

■

RESULTS AND DISCUSSION Effects of the Preparation Protocol on Cell Structure. Figure 1 presents the images of cells that underwent the applied drying protocol using multistep washing in diluted solutions of anhydrous alcohol. The so-called “error mode” images are presented since they nicely show tiny features on a cellular surface, measured by the AFM technique. The AFM images of cellular topography show nicely spread out cells with a clearly visible cell nucleus. In particular, the morphology of HCV29 and T24 cells after drying is similar to that observed for living cells, as reported in Lekka et al.18 We conclude that the applied protocol for sample preparation used for TOF-SIMS measurements prevents cell shape and morphology. The dried cells show no visible destruction neither in cell nucleus nor membranous part of the cell. To estimate the thickness of dried cells, further studied by TOF-SIMS, the cross 3197

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201

Article

Analytical Chemistry

is a statistical method that highlights the similarities and differences among samples.17 A principal component is a linear combination of the original variables between samples. Combinations are chosen to show the maximum variation in the samples. The PCA results in mathematical transformation of data to give a scores matrix of column vectors and a loadings matrix of row vectors, with as many columns as the original data matrix. Initial variables are transformed into a set of variables called principal components (PC). It is well-known that the final result obtained from PCA might be influenced by the way of data preprocessing.17 In our studies, two methods of the data preprocessing have been applied, namely, mean centering and autoscaling in order to study their influence on the final results of analysis. It will be shown that when mean centering is applied, the PCA-based separation between samples is influenced by small masses, while autoscaling consider larges masses. Initial analysis of the PCA loadings, corresponding to the scores in both approaches of data preprocessing for the all human bladder cell groups, revealed significant complexity of the recorded mass spectra (Figure 5).

Figure 3. (a) Exemplary TOF-SIMS mass spectra recorded for silicon surface exposed to RPMI 1640 media used for HCV29 and T24 cells growth. (b) Principal components PC2 versus PC1 for PCA of spectral data for human bladder cells.

underlies the importance to use these samples as reference ones to subtract such an effect. The results of PCA (Figure 3b and Figure S1 in the Supporting Information) showed separate groups of silicon surfaces resulting from the culture media composition. PCA of a Whole Mass Spectra Recorded for Human Bladder Cells. Typical TOF-SIMS spectra of measured cell lines are presented in Figure 4. The recorded mass spectra

Figure 4. TOF-SIMS mass spectra recorded for all studied human bladder cell lines. Figure 5. Loadings plot for PC1, PC2, and PC3 components resulted from TOF-SIMS spectra obtained using (a) mean centering and (b) autoscaling approaches of data preprocessing, determined for the studied bladder cancer cells.

showed large numbers of peaks with distinct intensities, independent of the cell type. Numerous peaks point out, however, the inability to define neither a single peak nor a group of peaks that characterized a given human bladder cell line. The complexity of the spectral signatures associated with biological materials is widely known.19,20 Therefore, their interpretation can be facilitated by chemometric analysis based on finding the greatest mass spectral differences. Thus, differences between cells can be rapidly identified. The PCA

From the analysis of loadings, one can see that there is no single mass that is significantly different for each principal component (i.e., PC1, PC2, or PC3). When PC components are plotted against each other, the greatest differentiation between human bladder cancer cells is obtained for the third principal component (PC3) plotted as a 3198

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201

Article

Analytical Chemistry

actin filaments.19,21 Carcinoma cells (HTB-9, HT1376, and T24) have a rounded shape and their actin cytoskeleton is poorly differentiated.21,22 Their smaller distribution of data points suggest very similar chemical compositions in each of the studied cell lines. Principal components define an average model spectrum describing all studied cells. Figure 7 presents a measure of how well each cell conforms with the model Q-residual.

function of the second principal component (PC2). All data points cluster in five different groups corresponding to different cell lines and silicon substrate (Figure 6 and Figures S3−S5 in

Figure 7. Q residual versus sample type determined for the studied human bladder cell lines. Data are presented as a mean ± standard deviation.

The spread of Q-residuals corresponding to the nonmalignant line reflects a large variety of HCV29 cells. Similarity between cells derived from metastatic lines is manifested by smaller values of Q-residuals which are also less scattered. This conclusion is valid for both preprocessing approaches used priori to principal component computation. Abnormality of Human Bladder Cancer Cells. Histological grade describes the degree of abnormality in cancer cells. It is a parameter derived from microscopic appearance of cancerous cells. It can be treated as a measure of differentiation describing to what extent cancer cells are similar in appearance and funtionality to healthy cells of the same tissue type. The severity of changes are frequently classified into four degrees. The cells of grade I are often well-differentiated and they are generally considered to manifest the least aggressive behavior. Conversely, the cells of grade III or grade IV denote usually poorly differentiated or undifferentiated high-grade tumors and are generally the most aggressive in behavior. In our studies, the histological grade of HT1376 and T24 cells is very similar, i.e., grade III and grade III/IV, correspondingly. Therefore, overlapping data points distributions are expected. HTB-9 cells are carcinoma with histological grade II; therefore, it is expected that the distribution of these data points will be separated both from HCV29 and HT1376/T24 cells. The PCA, almost independent of the calculation approach (with or without considering the silicon surfaces exposed to culture medium; applied preprocessing methods), shows clear separation of HCV29 (nonmalignant) cells. The results obtained for all three cell lines of cancerous cells overlap to a various extent, however, indicating lack of correlation with histological degree. On the other hand, these results are supported by the mechanical properties of these cells that are indirectly determined by chemical composition of the actin cytoskeleton and cell membrane, as recently reported by Ramos et al.21 Similarly, as in the PCA, the largest variations of mechanical properties have been observed for HCV29 cells while distribution of elastic properties for cancerous cells were narrower.

Figure 6. Principal components plots: (PC3 against PC2) showing grouped data points that corresponds to each sample type (a cell line or a silicon surface). The analysis based on the PCA reduction of TOF-SIMS mass spectra, performed using two different approaches of data preprocessing (i.e., mean centering (a) and autoscaling (b)). Each point on a plot denotes a whole mass spectrum acquired from the entire scanned region (150 μm × 150 μm) (n = 8−10 mass spectra).

the Supporting Information), regardless of the used preprocessing approach. The observed separation between all studied cell lines strongly indicates the difference in their chemical compositions, whereas the distribution of points brings the information on how homogeneity of the population of a specific cell type. Analogous grouping is observed for the PCA that includes the data recorded only for the cell, without considering the mass spectra collected for the silicon surface exposed to culture medium (Figures S7−S9 in the Supporting Information). The separation of the studied cell lines is clearly visible; however, depending on the preprocessing approach applied, a distinct character of point distributions is observed. The mean centering (dominated by ions and molecule fragments with smaller masses) is less than 150 Da. Figure 6a shows all data separated according to sample type. The autoscalling, however, seems to reflect better the properties of the studied cell lines (Figure 6b). The data points representing the mass spectra obtained for the HCV29 nonmetastatic cell line are apparently separated from the remaining three cell lines of metastatic cells. The distribution of points is the widest out of all distributions. This might indicate that chemical composition of HCV29 bladder cells in more variable as compared to cancerous bladder cells. Alternately, morphology of these cells differs and all mass spectra are influenced by the underlying silicon substrate. Both explanations seemed to be applicable. The HCV29 cells vary significantly in shape denoting both a distinct organization of 3199

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201

Article

Analytical Chemistry Identification of Mass Peaks. Analysis through direct observation of the TOF-SIMS spectra is a difficult task since, independent of the studied cells, a wide range of masses can be excited from the cellular surface (the mass accuracy is 0.01 Da, Figure 4). We classified the importance of different masses in the discrimination between the studied cell lines based on the values of PC2. It has been noticed that nonmalignant HCV29 cells have a positive values of PC2 while malignant HTB9, HT1376, and T24 cell lines have negative values (independent of preprocessing approaches, better visible for autoscaling). To simplify the analysis, the mean value and the standard deviation (SD) of the PC2 were calculated, resulting in values of 1.33 × 10−9 ± 0.0371 and 0.0015 ± 0.0347, for mean centering and autoscalling approaches, respectively. Then, all masses with the values of PC2 below −2 SD were chosen, assuming that they caused the largest differentiations between nonmalignant and malignant cells. The results of such analysis are presented in Tables 1 and 2 for autoscaling and mean centering approaches of preprocessing, correspondingly.

The performed analysis shows that depending on the applied preprocessing approach, distinct sets of signals strongly contribute to the principal components. Alkyl chain fragments (masses up to 95 Da) originating from the lipid tails of fatty acids give the strongest contribution to PC when meancentering is used. When autoscaling is used, larger fragments corresponding to both fatty acids and proteins became more relevant. Masses that differentiate values of PC2 can be assigned mostly as lipids; however, there is also a mass peak characteristic for arginine.23−27

■

Cancer progression is characterized by the aberrant growth of cells with genetic mutations of genes controlling cell proliferations and survival. It is a multistep process proceeding through many different stages, governed by successive changes in gene expression and the encoded products.28 Cancer, even if diagnosed for a specific organ or tissue, is characterized by hundreds of variations, each with a specific pattern of behavior and genetic makeup. Frequently, this leads to difficulties in the determination of a biomarker specific for a given cancer type. Bladder cancer is difficult to be diagnosed due to its identification in the very advanced stages of the disease. Moreover, there are no biomarkers that could be applied in the bladder cancer diagnosis. Therefore, high-resolution techniques that enable the identification of a group of nonspecific changes are crucial to help the diagnosis. In our studies, we have demonstrated that the TOF-SIMS technique together with the PCA of mass spectra can be used to distinguish between various types of cancer cells. Our findings show that the invasive phenotype of human bladder cancer cells can be correlated with the alterations in chemical compositions of the studied cells. There are many TOF-SIMS instruments that are probably accessible to research focusing on biology-related samples. Considering this, the amount of scientific papers investigating biological materials is surprisingly low. One of the reasons is lack of data showing that other sample preparation methods (without any freezing based methodology) can be applied for TOF-SIMS measurements. Another reason is the fact that mass spectra are far more complicated as compared to simpler organic structures. In our work, we have demonstrated that (1) the methodology that involves fixing and dehydrating steps leads to results showing separation between normal and cancerous cells; and (2) there is no need to know the masses that discriminate cells by their chemical properties. This is a huge advantage in terms of potential application of the TOFSIMS technique to identify the pathological state of the sample. The score plot PC3 versus PC2 clearly distinguishes the nonmalignant cell line HCV29 from the malignant cell lines. The knowledge which masses are mostly responsible for discrimination is very important since it may bring to our improvements on mechanics leading to chemical changes on a cellular surface. Further analysis of mass spectra allows one to find a list of mass peaks with intensities that are significantly higher in cancer cell lines (in particular, for transitional cell carcinoma, T24 cells) than in nonmalignant cell cancer of the ureter (i.e., HCV29 cells).

Table 1. Masses with PC2 Values below −2SD Determined Based on Autoscaling Type of Pre-Processing mass [u]

formula

assignment

60.09 88.08 97.07 102.10 114.14 127.11 132.09 139.11 141.12 144.08 150.10 151.10 159.09 170.19 200.14 226.16

C3H10N+ C3H10N3+ C6H9O+ C4H12N3+ C8H18+ C5H11N4+ C10H12+ C9H15O+ C8H15NO+ C10H10N+ C10H14O+ C10H15O+ C10H11N2+ C11H8NO+ C11H20O3+ C13H22O3+

fatty acid−lipid heada

a

Assignment based on literature data.23−27 based on literature data.23−27

fatty acid−lipid heada ARG, argininea fatty acid−lipid tailb ARG, argininea fatty acid−lipid tailb fatty acid−lipid headb

α-tocopherolb α-tocopherola TRP, tryptophana TRP, tryptophana

b

Proposed assignment

Table 2. Masses with PC2 Values below −2SD Obtained after Mean Centering Type of Pre-Processing

a

mass [u]

determined formula

assignment

58.07 67.06 73.07 81.08 83.10 84.09 86.11 91.06 95.10

C3H8N+ C5H7+ C2H7N3+ C6H9+ C6H11+ C6H12+ C6H14+ C7H7+ C7H11+

fatty acid−lipid heada fatty acid−lipid taila ARG, argininea fatty acid−lipid taila fatty acid−lipid taila fatty acid−lipid tailb fatty acid−lipid tailb fatty acid−lipid taila CHOL, cholesterola

Assignment based on literature data.23−27 based on literature data.23−27

b

CONCLUSIONS

Proposed assignment

The majority of masses can be assigned as lipids, but there are also signals typical for amino acids, namely, arginine and tryptophan.23−27 3200

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201

Article

Analytical Chemistry

■

(26) Nygren, K. A. N.; Richter, K.; Chen, Y. U. N.; Dangardt, F.; Friberg, P.; Magnusson, Y. Microsc. Res. Tech. 2007, 835, 828−835. (27) Passarelli, M. K.; Winograd, N. Biochim. Biophys. Acta 2011, 1811, 976−990. (28) Hart, I. M.; Saini, A. Lancet 1992, 339, 1453−1457.

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by the National Science Centre (NCN) Project Number DEC-2013/11/N/ST4/01860. REFERENCES

(1) Burchardt, M.; Burchardt, T.; Shabsigh, A.; De La Taille, A.; Benson, M. C.; Sawczuk, I. Clin. Chem. 2000, 46, 595−605. (2) Kim, J. C.; Steinberg, G. D. J. Urol. 2001, 165, 745−756. (3) Suresh, S. Acta Biomater. 2007, 3, 413−438. (4) Kumar, S.; Weaver, V. M. Cancer Metastasis Rev. 2009, 28, 113− 127. (5) Hanahan, D.; Weinberg, R. A. Cell 2011, 144, 646−674. (6) Gianuzzi, L. A.; Stevie, F. Introduction to Focused Ion Beams: Instrumentation, Theory, Techniques, and Practice; Springer: New York, 2005. (7) Varki, A.; Kannagi, R.; Toole, B. P. Glycosylation Changes in Cancer. In Essentials of Glycobiology , 2nd ed.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 2009; Chapter 44, pp 1− 11. (8) 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 discussion 77−92.. (9) Kulp, K. S.; Berman, E. 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. (10) Baker, M. J.; Brown, M. D.; Gazi, E.; Clarke, N. W.; Vickerman, J. C.; Lockyer, N. P. Analyst 2008, 133, 175−179. (11) Park, J.-W.; Shon, H. K.; Yoo, B. C.; Kim, I. H.; Moon, D. W.; Lee, T. G. Appl. Surf. Sci. 2008, 255, 1119−1122. (12) Brulet, M.; Seyer, A.; Edelman, A.; Brunelle, A.; Fritsch, J.; Ollero, A.; Laprevote, A. J. Lipid Res. 2010, 51, 3034−3045. (13) Chandra, S.; Morrison, G. H. Biol. Cell 1992, 74, 31−42. (14) Prauzner-Bechcicki, S.; Wiltowska-Zuber, J.; Budkowski, A.; Lekka, M.; Rysz, J. Acta Phys. Polym., A 2012, 121, 533−539. (15) Tipping, M. E.; Bishop, C. M. J. R. Stat. Soc. Ser. B 1999, 61, 611−622. (16) Bro, R.; Smilde, A. K. Anal. Methods 2014, 6, 2812. (17) Ringnér, M. Nat. Biotechnol. 2008, 26, 303−304. (18) Lekka, M.; Gil, D.; Dabros, W.; Jaczewska, J.; Kulik, A. J.; Lekki, J.; Stachura, Z.; Stachura, J.; Laidler, P. J. Mol. Recognit. 2011, 24, 833− 842. (19) Berman, E. S. F.; Wu, L.; Fortson, S. L.; Kulp, K. S.; Nelson, D. O.; Wu, K. J. Surf. Interface Anal. 2009, 41, 97−104. (20) Belu, A. M.; Graham, D. J.; Castner, D. G. Biomaterials 2003, 24, 3635−3653. (21) Ramos, J. R.; Pabijan, J.; Garcia, R.; Lekka, M.; Beilstein, J. Nanotechnology 2014, 5, 447−457. (22) Lityńska, A.; Przybyło, M.; Pocheć, E.; Laidler, P. Acta Biochim. Polym. 2002, 49, 643−650. (23) Debois, D.; Bralet, M. P.; Le Naour, F.; Brunelle, A.; Lapravote, O. Anal. Chem. 2009, 81, 2823−2831. (24) Anderton, C. R.; Vaezian, B.; Lou, K.; Frisz, J. F.; Kraft, M. L. Surf. Interface Anal. 2012, 44, 322−333. (25) Leefmann, T.; Heim, C.; Siljeström, S.; Blumenberg, M.; Sjövall, P.; Thiel, V. Rapid Commun. Mass Spectrom. 2013, 27, 565−581. 3201

DOI: 10.1021/ac504684n Anal. Chem. 2015, 87, 3195−3201