Surface-Enhanced Raman Scattering-Based Detection of Cancerous Renal Cells Sevda Mert, Mustafa C ¸ ulha* Department of Genetics and Bioengineering, Faculty of Engineering, Yeditepe University, Atasehir, Istanbul 34755 Turkey
Surface-enhanced Raman scattering (SERS) is used for the differentiation of human kidney adenocarcinoma, human kidney carcinoma, and non-cancerous human kidney embryonic cells. Silver nanoparticles (AgNPs) are used as substrate in the experiments. A volume of colloidal suspension containing AgNPs is added onto the cultured cells on a CaF2 slide, and the slide is dried at the overturned position. A number of SERS spectra acquired from the three different cell lines are statistically analyzed to differentiate the cells. Principal component analysis (PCA) combined with linear discriminate analysis (LDA) was performed to differentiate the three kidney cell types. The LDA, based on PCA, provided for classification among the three cell lines with 88% sensitivity and 84% specificity. This study demonstrates that SERS can be used to identify renal cancers by combining this new sampling method and LDA algorithms. Index Headings: Surface-enhanced Raman scattering; SERS; Renal carcinoma; Single cell; Differentiation; Principal component analysis-linear discriminate analysis; PCA-LDA.
INTRODUCTION Cancer continues to be one of the deadliest diseases worldwide. In the last two decades, a significant effort has been devoted to diagnosis and treatment. However, the diverse causes of malignance hinders the development of a solid approach for treatment. Therefore, early diagnosis gains importance in the successful treatment of the disease. Conventional cancer diagnostic approaches such as positron emission tomography (PET), computed tomography, magnetic resonance, and ultrasound are routinely used for cancer diagnosis.1 Magnetic resonance imaging identifies changes in the tissue morphology due to abnormal cell growth, while PET gives information about malignant tumors where cellular activity is increased.2 These techniques provide information about the disease mostly after the earliest stage of onset. Therefore, there is a need for the development of novel approaches for early detection. The use of spectroscopic techniques such as fluorescence,3 infrared (IR),4 and Raman5 are gaining importance in clinical research in recent years since they can provide fast molecular-level information with minimal surgical intervention or biopsy. The utility of vibrational spectroscopic techniques, IR and Raman, has long been investigated for tissue differentiation.6–8 A Raman spectrum can give direct information about molecular composition of a sample. This non-destructive and nonReceived 26 August 2013; accepted 11 February 2014. * Author to whom correspondence should be sent. E-mail: mculha@ yeditepe.edu.tr. DOI: 10.1366/13-07263
Volume 68, Number 6, 2014
invasive technique has been employed for population investigation for early diagnosis and the following phases of treatments.9 Raman spectroscopy especially has been used to examine samples obtained from breast,10 brain,6 cervical,11 nasopharyngeal,12 lung,13 and skin14 tissue in the literature. Surface-enhanced Raman spectroscopy (SERS) can be considered as a mode of Raman spectroscopy, which allows for the enhancement of Raman scattering of up to 108 when a molecule or molecular structure is located in the close vicinity of nanostructured noble metal surfaces.15 The sensitivity increase in SERS resulted in an enormous interest from the scientific communities from various disciplines to employ the technique for the detection and identification of DNA,16 proteins,17 microorganisms18 and living cells,19 and tissue.6 The first use of SERS at the living cancer cell level for the detection of antitumor drug concentrations was reported in 1991.20 Shortly after, a pioneering study demonstrated that SERS could be used to collect biomolecular information from living cells at the singlecell level.19 In that study, gold nanoparticles (AuNPs) internalized by the cells were used as substrates. In the following years, several reports concerning the use of SERS on living cells were appeared in the literature.13,21–31 All these studies demonstrate that SERS is slowly evolving as a novel technique to obtain molecular-level information at the single-cell level. Kidney cancer is almost 2% of all cancer cases around the world, although the most common form of kidney cancer, renal cell carcinoma (RCC), has increased in the West compared to Asia.32 Renal cell carcinoma has similar symptoms to many diseases. Only about 5% to 10% of people with RCC show classic symptoms such as abdominal mass, flank pain, and hematuria, and only 9% of RCC cases appear with these symptoms.33 Therefore, the symptoms of RCC can be easily confused with other diseases, and it makes diagnosis very difficult. With early diagnosis, RCC patients have a high survival rate. Research over the past 30 years on early detection, diagnosis, and treatment for patients with RCC has been increased and provided new imaging techniques for tumorous formations.34 In our study, we demonstrate the power of SERS using citrate-reduced silver nanoparticles (AgNPs) as substrates for the differentiation of cancer and healthy kidney cells. Human kidney adenocarcinoma (ACHN) cells with higher tumorigenic and metastatic potentiality, human kidney carcinoma (A-498) cells with lower tumorigenic potentiality, and human kidney embryonic (HEK 293; non-cancerous) cells were chosen as model cells. The sample preparation method that we employed
0003-7028/14/6806-0617/0 Q 2014 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
617
in our previous study for protein detection and identification with SERS was used in the study.35 The method uses the drying of samples containing colloidal AgNPs and analytes at the overturned position. This allows for a more uniform distribution of AgNPs on the droplet area instead of getting them jammed at the liquid–air–solid contact interface. The SERS spectra obtained from three cell lines were processed with linear discriminate analysis (LDA)36 based on the principal component analysis (PCA)37 method to differentiate the healthy cells from cancer cells. The results of this study with sensitivity and specificity of 87.8% and 84.4%, respectively, indicate that the SERS spectra obtained from the cells can be used for the diagnosis of the cancer.
MATERIALS AND METHODS Cell Cultures and Sample Preparation. Renal cancer cells (ACHN and A-498) and normal cells (HEK 293) were obtained from the American Type Culture Collection (ATCC). The human kidney adenocarcinoma cell line has high tumorgenic properties, while the A-498 cell line has low tumorigenic properties. Thus, HEK 293 is a noncancerous and immortalized cell line. The cells were taken out from a 80 8C freezer, thawed, and washed. Then, ACHN, A-498, and HEK 293 cells were incubated in T-75 flasks using glucose Dulbecco’s modified Eagle’s medium (DMEM; Sigma-Aldrich, Germany) with 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 1% penicillin/streptomycin in an incubator with a humidified 5% CO2 atmosphere at 37 8C prior to subculturing. For the subculturing of the HEK 293, A-498, and ACHN cells, the medium was removed and rinsed with phosphate buffered saline (PBS, 10X) solution, and 2 ml of 0.025% trypsin solution was added into the culture plate and placed at 37 8C until the cells detached. Then, the fresh medium was added to deactivate trypsin, and the cultures were transferred into the centrifuge tubes and centrifuged at 1500 rpm for 5 min. After the supernatant was removed, the fresh medium was added and dispensed into a new flask containing the fresh medium. After twice subculturing, the cells were moved to six-well cultivation plates with CaF2 slides and incubated at 37 8C. The cells were directly grown on CaF2 slides using standard procedure. Before the SERS measurements, the culture medium was removed, and the cells were washed three times with PBS. Then, CaF2 slides were rinsed with water to remove remnant salts. After the CaF2slides were dried, 2 lL of colloidal AgNPs was added onto the slides, and the slides were turned upside down to dry in a cell culture hood in sterile conditions. Synthesis of Colloidal Silver Nanoparticles (AgNPs). The colloidal AgNPs were synthesized by using sodium citrate as a reducing agent.38 Briefly, 90 mg of AgNO3 was dissolved in 500 mL of water, and the solution was kept on the heater until boiling. Then, 10 mL of 1% sodium citrate was added into the solution, and the solution was kept for 1.5 h until the volume reached the half volume. The concentration of this suspension was called 1X. The final concentration of AgNPs colloidal suspension was adjusted to 4X by centrifuging the 1X suspension and removing 3/4 of the supernatant for the use in SERS experiments. The characterization was
618
Volume 68, Number 6, 2014
performed with a Zetasizer (NanoSizer, Malvern, UK) and a ultraviolet/visible (UV/Vis) spectrometer (Perkin Elmer Lambda 25-UV). The average size distribution of the solution was around 60 nm, and the maximum absorption was recorded at 420 nm. Surface-Enhanced Raman Scattering (SERS) Measurements. A Renishaw InVia Reflex Raman microscopy system (Renishaw Plc., New Mills, Wotton-under-Edge, UK) equipped with an 830 nm diode was calibrated by using the silicon phonon mode at 520 cm 1. The incident laser power was 7.5 mW on the sample, and the spectral data acquisition time was 10 s. A 503 microscope objective (NA: 0.50) was used in the experiments. The SERS spectra were acquired over a spectral range of 500–1800 cm 1 because the specific Raman bands were observed in this range. The SERS spectra were collected over an area of 10 3 10 lm2 on the cell using the map image acquisition method function in WIRE 2.0 software with 1 lm steps collected. The experiment was repeated at least three times, and the spectral analyses were carried out by the WIRE 2.0 software. Data Analysis. The total number of spectra collected on a cell over 10 3 10 lm2 area was 100. A minimum of 30 cells for each cell line was used for the SERS acquisition for each experiment. Each of 100 SERS spectra obtained from each single cell was normalized to reduce the variations in intensity, and the total SERS spectra obtained from each cell were averaged to obtain a mean SERS spectrum. The high-dimensional SERS dataset was processed by the SPSS software package (SPSS Inc., Chicago), which is used for statistical analysis in the social sciences. The data mining, discrimination, and graphing techniques of SPSS are performed to differentiate the cell types. The principal component analysis (PCA) method was used to reduce the dimension of the SERS spectral dataset space by generating standardized matrices and principal component (PC) scores, the most significant variables produced from original spectra. Then, the linear discriminant analyses (LDA) method was used as a classification method. The most significant PC scores obtained by using independent sample ttests were used as variables in LDA. The LDA method provides discriminant functions, which increase the variances in SERS data between the different types of cells while decreasing the variances between the cells of the same type. The leave-one-out cross-validation method was used to indicate the classification results of PCA and LDA models accurately.
RESULTS AND DISCUSSIONS The CaF2 slides were used to culture the cells due to the background interference from the glass slide. After the cells were grown on the CaF2 slides, the medium was removed by washing with PBS, and remnants of PBS were washed with deionized and autoclaved water to avoid contamination. The slide with cells was placed in a safety cabinet until it was dried. Then, a 2 ll droplet of colloidal suspension containing AgNPs was added onto the cells while viewing under a light microscope and left to dry by turning the slide upside down. Multiple 2 ll droplets were spotted onto the cells grown on the CaF2 slide in order to place the AgNPs on more cells when
FIG. 1. (a) Image of ACHN cells under a 403 objective (NA:0.5), (b) image of setup used for droplet, (c) images of droplet area dried at suspended and (d) regular positions under 43 objective (NA:0.13).
necessary. A 403 microscope objective was used to observe the location of the cells and the aggregates of AgNPs in dried droplet area for each position. Figure 1 shows the sample preparation process (Figs. 1a and 1b) and the images (Figs. 1c and 1d). When a small volume of colloidal suspension containing AgNPs is placed on a surface, a phenomenon call ‘‘coffee ring’’ takes place and jams almost all of the particles at the liquid–solid–air interface. This causes the formation of very tightly packed particles at the droplet edges.39,40 For optimal SERS activity, a loose packing of noble metal particles is necessary.41–43 In order to avoid the unequal distribution AgNPs on the droplet area and jamming them at the droplet peripherals, the droplet placed onto the cells grown on CaF2 was overturned and dried at the suspended position (Fig. 1b). Figures 1c and 1d show the light microscopy images of the droplet areas at the suspended and regular position, respectively. As seen in the figure, the AgNPs are accumulated at the periphery of the droplet area when the droplet is dried at the regular position, while they are fairly distributed in the droplet area dried at the suspended position. The cells grown on CaF2 are washed with PBS and then water before letting them dry at room temperature. During drying, it is highly possible that most of the cells retained their morphology while a few of them crystallized. The washing procedure of the cells grown on CaF2 does not affect the cell morphology (Fig. 1a), and they were visible to select under microscope during the additional AgNP colloidal suspension and their selection for SERS measurements. The scanning electron microscopy (SEM) images of ACHN cells before and after addition of the AgNP colloidal suspension are given in Figs. 2a and 2b and Figs. 2c and 2d, respectively. As seen, the addition of colloidal suspension damages the cell integrity. It is clear that cell necrosis, or unprogrammed death, is an unfortunate outcome of the unexpected condition, drying, and adding colloidal suspension during the sample preparation. However, the AgNPs and their aggregates remained in the ruptured cell area. Therefore, the spectral features on the SERS spectra originate from biomacromolecules and ionic species composing the cell. The SERS spectra from an area of 10 3 10 lm2 on a cell were acquired by using the mapping method of WIRE 2.0 software with 1 lm steps. A total of 100 SERS spectra from one cell and a minimum of 30 cells for each cell line
FIG. 2. SEM images of ACHN cells (a) and (b) before, and (c) and (d) after treatment with AgNPs containing colloidal suspension.
were acquired. A minimum of 3000 SERS spectra from each cell line were collected, and the total SERS spectra acquired from healthy and cancer cell lines were averaged to one mean SERS spectrum, presented in Fig. 3a. The mean SERS spectra for each cell line was normalized at 529 cm 1 (Fig. 3b) to visualize the differences in spectral bands of the three cell types. As seen, there are several differences in the intensity and the shape of the bands appearing on each spectrum. The tentative band assignments based on the SERS spectra of 28 reference compounds with biological origin (provided in the Supplemental Material) are listed in Table I. A strong band at around 667 cm 1 is observed for each cell line, and this band is probably associated with m(C–S) stretching, which is usually observed in the range of 600–700 cm 1 as an intense band (Fig. S4b). Although there could be several molecular species such as amino acids, peptides, peptide derivatives, and proteins possessing a C–S bond and they may selectively interact with the AgNPs, the band at 660 cm 1 in cancer cell lines is claimed to be related to glutathione (GSH)43,44, a tripeptide containing cysteine. Glutathione is found in all human tissues and protects the fatty tissues from hazardous effects of free radicals.45 The band at around 660 cm 1 can also be attributed to uracil and lipoproteins (see Supplemental Material). The band at 732 cm 1 may originate from adenine, phospholipids, proteins, and nicotinamide adenine dinucleotide hydrogen (NADH; see Supplemental Material). For cellular growth and metabolism, NADH is required, and the increased concentration in cancer cells of this molecule may indicate this consequence. The band at around 800 cm 1 on the spectra of the cancer cells may originate from cytosine, tyrosine, and/or lactic acid. The increase in the concentration of these molecules can be related to the increased metabolism in cancer cells due to abnormal cellular growth. The band at around 906 cm 1, which is more intense in the spectra of cancer cells than of healthy cells, is assigned to proteins and
APPLIED SPECTROSCOPY
619
FIG. 3. (a) Mean SERS spectra from cultured cells: ACHN, A-498, and HEK 293. (b) Mean SERS spectra normalized to the intensity 529 cm
1
peak.
TABLE I. Tentative band assignments of SERS spectra of the cell lines Raman shift/ cm
1
Major assignments
HEK
ACHN
A-498
Saccharides/proteins Lipoprotein/RNA/glutathione Adenine (DNA/RNA)/phospholipids/proteins/NADH Cytosine (RNA/DNA)/proteins/lactic acid/glutathione Saccharides/proteins Phospholipids/proteins Proteins Proteins/lipids Proteins Cytosine and guanine ( DNA/RNA)/proteins Proteins/adenine (DNA/RNA) Proteins/DNA/NADH DNA/proteins/lactic acid Phospholipids/proteins DNA/proteins Phospholipids
530 668 732 797 911 965 993 1050 1102 1180 1278 1330 1412 1446 1577 –
530 670 732 800 914 965 993 1050 1102 – 1278 1333 – 1447 1576 1654
– 660 725 800 906 964 998 1040 1102 1180 1278 1330 1414 1448 1577 –
620
Volume 68, Number 6, 2014
FIG. 4. The PC scatter plots show the principal components of 30 cells from each cell line (ACHN, A-498, and HEK 293). (a) PC 1 and PC 2, (b) PC 1 and PC 3, (c) PC 2 and PC 3, (d) PC 1 and PC 4, (e) PC 2 and PC 4, and (f) PC 3 and PC 4.
APPLIED SPECTROSCOPY
621
mainly glucose or other possible monosaccharides from the SERS spectra of the biomolecules listed in the Supplemental Material. The band at around 964 cm 1 is attributed to phospholipids and proteins, and the band at 993 cm 1 is assigned to phenylalanine and aspartic acid from proteins. The anticipated increase in the intensity of this band was observed on the spectra of cancer cells compared to the spectra of the HEK 293 cells. This increase in phospholipids concentration is obtained in cancer cells due to the possible increase in fatty acid production. As the cell division is accelerated in cancer cells, they need to produce new phospholipids to maintain cell membrane integrity. In addition, as mentioned above, the energy need in cancer cells increases due to the increased metabolism. The bands in the region of 1000–1200 cm 1 may originate from lipids, proteins, and nucleic acids. The bands in the region of 1200–1300 cm 1 may originate from tyrosine and adenine. The band examined at around 1300 cm 1 is broader for cancer cells than the band at this wavenumber for the healthy cells. The difference in the intensity of the bands at 529, 1414, and 1446 cm 1 can be explained with the concentration difference of saccharides, proteins, and lipids in the cancer cells. The bands in the region of 1500–1600 cm 1 are attributed to the pyrimidine ring of nucleotides and/or proteins.46 In addition, the band at around 1576 cm 1, which was assigned to Amide II46 and/or DNA, is broader for the A-498 cell line than the band on the spectra of the other cell lines, but the intensity of this band is higher for the one for the cancer cells than the one for healthy cells. Linear discriminate analysis (LDA), a classical discriminating method, was used in our study to differentiate the cells in high reliability. The high-dimensional SERS dataset was reduced, and new relevant variables, principal component (PC) scores, were obtained by using PCA, which is an eigenvalue–eigenvector method. Four PC component eigenvectors were extracted, and PC 1 (44, 25% of the total variance), PC 2 (34, 51%), PC 3 (13, 86%), and PC 4 (2, 31%) were plotted by using a scatter plot to show the values of the three variables, ACHN, HEK, and A-498 cell lines (Fig. 4). Each dot represents one cell. Here, PC 1 and PC 2 vectors could identify the cells with a sensitivity and specificity of 85% and 84%, respectively, while others are grouped in a low correlation. Also, PC 1 and PC 3 could classify the cell lines with a sensitivity and specificity of 70% and 68%, respectively, while PC 1 and PC 4 could distinguish the cells with sensitivity and specificity of 70% and 69%, respectively. As well, PC 2 and PC 3 could diagnose the three types of cells with a sensitivity and specificity of 71% and 69%, respectively. In addition, PC 2 and PC 4 could differentiate the cell lines with a sensitivity and specificity of 76% and 73%, respectively, while PC 3 and PC 4 could separate the cell lines with a sensitivity and specificity of 46% and 42%, respectively. Here, PC 1 and PC 2 vectors have large means and significant differences between the groups compared to PC 3 and PC 4 vectors. Principal component analysis is not the best classification method for our SERS dataset, although it is a good way to summarize high datasets into subsets of variables. Linear discriminant analysis
622
Volume 68, Number 6, 2014
FIG. 5. LD 1 and LD 2 scatter plot of three types of cells.
arranges discriminant functions by maximizing the variances in the SERS data between different cell groups while minimizing the variances between the cells of the same group. The most significant PCs obtained by using independent-sample t-tests were used in LDA. The scores on LD 1 and LD 2 were also plotted by using a scatter plot (Fig. 5). The ACHN, HEK, and A-498 cells are differentiated with a sensitivity and specificity of 88% and 84%, respectively, using LDA. The differentiating results are obtained using the leave-one-out cross-validation method of PCA and LDA models. As a result, the clustering of three cells with sensitivity and specificity of 88% and 84%, respectively, is a high rate to detect the cancer types of RCC.
CONCLUSION In this study, the use of SERS to identify renal cancers by combining this new sampling method and LDA algorithms was demonstrated. The citrate-reduced AgNPs were used as SERS substrates, and the AgNP colloidal suspension placed onto the cells grown on the slide was dried at the suspended position. Since the AgNPs and their aggregates are localized on the cell surface area with this sample preparation method, the observed bands on the SERS spectra of the cells can be attributed to the released metabolites or biomolecules originating from the cell surface and intracellular molecules and molecular structures due to the cell rupture. The observed changes on the SERS spectra can be related to altered biological function in healthy versus cancer cells, which could be used for early diagnosis or cell typing. This study was the first to differentiate the renal cancer cells from healthy cells with high correlation by using algorithms-based PCA and LDA on the SERS spectra set. This diagnostic approach offers the possibility and simplicity for the future studies of SERS in categorizing and typing cell lines.
ACKNOWLEDGMENTS The authors thank Associate Professor Dilek Telci for providing cell lines. The financial support of Yeditepe University and TUBITAK is also greatly acknowledged. SUPPLEMENTAL MATERIAL All supplemental materials mentioned in the text, including six figures, one table, and a description of the procedure, can be found in the online version of the journal at http://www.s-a-s.org. 1. L. Fass. ‘‘Imaging and Cancer: A Review’’. Mol. Oncol. 2008. 2(2): 115-152. 2. X. Gao, Y. Cui, R.M. Levenson, L.W. Chung, S. Nie. ‘‘In Vivo Cancer Targeting and Imaging with Semiconductor Quantum Dots’’. Nat. Biotechnol. 2004. 22: 969-976. 3. J.R. Lakowicz. ‘‘Principles of Fluorescence Spectroscopy’’. New York: Springer, 2006. 3rd ed. 4. D. Lin-Vien, N.B. Colthup, W.G. Fateley, J.G. Graselli. ‘‘The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules’’. San Diego: Academic Press, 1991. 5. N.B. Colthup, L.H. Daly, S.E. Wiberley. ‘‘Introduction to Infrared and Raman Spectroscopy POD’’. San Diego: Academic Press, 1990. 3rd ed. 6. O. Aydin, M. Altas, M. Kahraman, O.F. Bayrak, M. Culha. ‘‘Differentiation of Healthy Brain Tissue and Tumors Using Surface-Enhanced Raman Scattering’’. Appl. Spectrosc. 2009. 63(10): 1095-1100. 7. O. Aydin, M. Kahraman, E. Kilic, M. Culha. ‘‘Surface-Enhanced Raman Scattering of Rat Tissues’’. Appl. Spectrosc. 2009. 63(6): 662-668. 8. R. Petry, M. Schmitt, J. Popp. ‘‘Raman Spectroscopy—A Prospective Tool in the Life Sciences’’. ChemPhysChem. 2003. 4(1): 14-30. 9. C. Kendall, N. Stone, N. Shepherd, K. Geboes, B. Warren, R. Bennett, H. Barr. ‘‘Raman Spectroscopy, A Potential Tool for the Objective Identification and Classification of Neoplasia In Barrett’s Oesophagus’’. J. Pathol. 2003. 200(5): 602-609. 10. C. Yu, E. Gestl, K. Eckert, D. Allara, J. Irudayaraj. ‘‘Characterization of Human Breast Epithelial Cells by Confocal Raman Microspectroscopy’’. Cancer Detect. Prev. 2006. 30(6): 515-522. 11. F.M. Lyng, E.O´. Faola´in, J. Conroy, A.D. Meade, P. Knief, B. Duffy, M.B. Hunter, J.M. Byrne, P. Kelehan, H.J. Byrne. ‘‘Vibrational Spectroscopy for Cervical Cancer Pathology, from Biochemical Analysis to Diagnostic Tool’’. Exp. Mol. Pathol. 2007. 82(2): 121-129. 12. S. Feng, R. Chen, J. Lin, J. Pan, G. Chen, Y. Li, M. Cheng, Z. Huang, J. Chen, H. Zeng. ‘‘Nasopharyngeal Cancer Detection Based on Blood Plasma Surface-Enhanced Raman Spectroscopy and Multivariate Analysis’’. Biosens. Bioelectron. 2010. 25(11): 2414-2419. 13. X. Qian, X.-H. Peng, D.O. Ansari, Q. Yin-Goen, G.Z. Chen, D.M. Shin, L. Yang, A.N. Young, M.D. Wang, S. Nie. ‘‘In Vivo Tumor Targeting and Spectroscopic Detection with Surface-Enhanced Raman Nanoparticle Tags’’. Nat. Biotechnol. 2007. 26: 83-90. 14. S. Sigurdsson, P.A. Philipsen, L.K. Hansen, J. Larsen, M. Gniadecka, H.C. Wulf. ‘‘Detection of Skin Cancer by Classification of Raman Spectra’’. IEEE Trans. Biomed. Eng. 2004. 51(10): 17841793. 15. K. Kneipp, M. Moskovits, H. Kneipp. ‘‘Surface-Enhanced Raman Scattering: Physics and Applications’’. Berlin: Springer-Verlag, 2006, vol. 103. 16. K. Kneipp, H. Kneipp, V.B. Kartha, R. Manoharan, G. Deinum, I. Itzkan, R.R. Dasari, M.S. Feld. ‘‘Detection and Identification of A Single DNA Base Molecule Using Surface-Enhanced Raman Scattering (SERS)’’. Phys. Rev. E. 1998. 57: 6281(R)-R6284(R). 17. Y. Wang, H. Wei, B. Li, W. Ren, S. Guo, S. Dong, E. Wang. ‘‘SERS Opens A New Way in Aptasensor for Protein Recognition with High Sensitivity and Selectivity’’. Chem. Commun. 2007. 48: 5220-5222. 18. W.R. Premasiri, D.T. Moir, M.S. Klempner, N. Krieger, G. Jones II, L.D. Ziegler. ‘‘Characterization of the Surface Enhanced Raman Scattering (SERS) of Bacteria’’. J. Phys. Chem. B. 2005. 109(1): 312320. 19. K. Kneipp, A.S. Haka, H. Kneipp, K. Badizadegan, N. Yoshizawa, C. Boone, K.E. Shafer-Peltier, J.T. Motz, R.R. Dasari, M.S. Feld. ‘‘Surface-Enhanced Raman Spectroscopy in Single Living Cells Using Gold Nanoparticles’’. Appl. Spectrosc. 2002. 56(2): 150-154.
20. I.R. Nabiev, H. Morjani, M. Manfait. ‘‘Selective Analysis of Antitumor Drug Interaction with Living Cancer Cells As Probed by Surface-Enhanced Raman Spectroscopy’’. Eur. Biophys. J. 1991. 19(6): 311-316. 21. G. Breuzard, J.F. Angiboust, P. Jeannesson, M. Manfait, J.M. Millot. ‘‘Surface-Enhanced Raman Scattering Reveals Adsorption of Mitoxantrone on Plasma Membrane of Living Cells’’. Biochem. Biophys. Res. Commun. 2004. 320(2): 615-621. 22. C. Eliasson, A. Lore´n, J. Engelbrektsson, M. Josefson, J. Abrahamsson, K. Abrahamsson. ‘‘Surface-Enhanced Raman Scattering Imaging of Single Living Lymphocytes with Multivariate Evaluation’’. Spectrochim. Acta, Part A. 2005. 61(4): 755-760. 23. X. Huang, I.H. El-Sayed, W. Qian, M.A. El-Sayed. ‘‘Cancer Cell Imaging and Photothermal Therapy in the Near-Infrared Region by Using Gold Nanorods’’. J. Am. Chem. Soc. 2006. 128(6): 2115-2120. 24. X. Huang, I.H. El-Sayed, W. Qian, M.A. El-Sayed. ‘‘Cancer Cells Assemble and Align Gold Nanorods Conjugated to Antibodies to Produce Highly Enhanced, Sharp, and Polarized Surface Raman Spectra: A Potential Cancer Diagnostic Marker’’. Nano Lett. 2007. 7(6): 1591-1597. 25. J.-H. Kim, J.-S. Kim, H. Choi, S.-M. Lee, B.-H. Jun, K.-N. Yu, E. Kuk, Y.-K. Kim, D.H. Jeong, M.-H. Cho, Y.-S. Lee. ‘‘Nanoparticle Probes with Surface Enhanced Raman Spectroscopic Tags for Cellular Cancer Targeting’’. Anal. Chem. 2006. 78(19): 6967-6973. 26. K. Kneipp, H. Kneipp, J. Kneipp. ‘‘Surface-Enhanced Raman Scattering in Local Optical Fields of Silver and Gold Nanoaggregates from Single-Molecule Raman Spectroscopy to Ultrasensitive Probing in Live Cells’’. Acc. Chem. Res. 2006. 39(7): 443450. 27. S. Shanmukh, L. Jones, J. Driskell, Y. Zhao, R. Dluhy, R.A. Tripp. ‘‘Rapid and Sensitive Detection of Respiratory Virus Molecular Signatures Using a Silver Nanorod Array SERS Substrate’’. Nano Lett. 2006. 6(11): 2630-2636. 28. T. Vo-Dinh, L.R. Allain, D.L. Stokes. ‘‘Cancer Gene Detection Using Surface-Enhanced Raman Scattering (SERS)’’. J. Raman Spectrosc. 2002. 33(7): 511-516. 29. Z. Wang, A. Bonoiu, M. Samoc, Y. Cui, P.N. Prasad. ‘‘Biological pH Sensing Based on Surface Enhanced Raman Scattering Through a 2-aminothiophenol-Silver Probe’’. Biosens. Bioelectron. 2008. 23(6): 886-891. 30. W. Xie, L. Wang, Y. Zhang, L. Su, A. Shen, J. Tan, J. Hu. ‘‘Nuclear Targeted Nanoprobe for Single Living Cell Detection by SurfaceEnhanced Raman Scattering’’. Bioconjugate Chem. 2009. 20(4): 768773. 31. S. Feng, J. Lin, Z. Huang, G. Chen, W. Chen, Y. Wang, R. Chen, H. Zeng. ‘‘Esophageal Cancer Detection Based on Tissue SurfaceEnhanced Raman Spectroscopy and Multivariate Analysis’’. Appl. Phys. Lett. 2013. 102(4): 043702-043704. 32. W.-H. Chow, L.M. Dong, S.S. Devesa. ‘‘Epidemiology and Risk Factors for Kidney Cancer’’. Nat. Rev. Urol. 2010. 7(5): 245-257. 33. B. Ljungberg, D.C. Hanbury, M.A. Kuczyk, A.S. Merseburger, P.F. Mulders, J.-J. Patard, I.C. Sinescu. ‘‘Renal Cell Carcinoma Guideline’’. Eur. Urol. 2007. 51(6): 1502-1510. 34. W.C. Black, H.G. Welch. ‘‘Advances in Diagnostic Imaging and Overestimations of Disease Prevalence and the Benefits of Therapy’’. N. Engl. J. Med. 1993. 328(17): 1237-1243. 35. S. Keskin, M. Culha. ‘‘Label-Free Detection of Proteins From DriedSuspended Droplets Using Surface Enhanced Raman Scattering’’. Analyst. 2012. 137(11): 2651. 36. M. Li, B. Yuan. ‘‘2D-LDA: A Statistical Linear Discriminant Analysis for Image Matrix’’. Pattern Recognit. Lett. 2005. 26(5): 52-532. 37. I.T. Jolliffe. ‘‘Principal Component Analysis’’. New York: John Wiley & Sons Inc. 2005. 38. P.C. Lee, D. Meisel. ‘‘Adsorption and Surface-Enhanced Raman of Dyes on Silver and Gold Sols’’. J. Phys. Chem. 1982. 86. (17): 33913395. 39. X. Shen, C.-M. Ho, T.-S. Wong. ‘‘Minimal Size of Coffee Ring Structure’’. J. Phys. Chem. B. 2010. 114(16): 5269-5274. 40. H. Hu, R.G. Larson. ‘‘Marangoni Effect Reverses Coffee-Ring Depositions’’. J. Phys. Chem. B. 2006. 110(14): 7090-7094. 41. J. Jiang, K. Bosnick, M. Maillard, L. Brus. ‘‘Single Molecule Raman Spectroscopy at the Junctions of Large Ag Nanocrystals’’. J. Phys. Chem. B. 2003. 107(37): 9964-9972. 42. V. Safonov, V. Shalaev, V. Markel, Y.E. Danilova, N. Lepeshkin, W. Kim, S. Rautian, R. Armstrong. ‘‘Spectral Dependence of Selective
APPLIED SPECTROSCOPY
623
Photomodification in Fractal Aggregates of Colloidal Particles’’. Phys. Rev. Lett. 1998. 80(5): 1102. 43. K. Kneipp, H. Kneipp, P. Corio, S. Brown, K. Shafer, J. Motz, L. Perelman, E. Hanlon, A. Marucci, G. Dresselhaus. ‘‘SurfaceEnhanced and Normal Stokes and Anti-Stokes Raman Spectroscopy of Single-Walled Carbon Nanotubes’’. Phys. Rev. Lett. 2000. 84(15): 3470.
624
Volume 68, Number 6, 2014
44. J. De Gelder, K. De Gussem, P. Vandenabeele, L. Moens. ‘‘Reference Database of Raman Spectra of Biological Molecules’’. J. Raman Spectrosc. 2007. 38(9): 1133-1147. 45. R. Brigelius-Flohe. ‘‘Tissue-Specific Functions of Individual Glutathione Peroxidases’’. Free Radical Biol. Med. 1999. 27(9-10): 951965. 46. Z. Movasaghi, S. Rehman, I.U. Rehman. ‘‘Raman Spectroscopy of Biological Tissues’’. Appl. Spectrosc. Rev. 2007. 42(5): 493-541.
Surface-enhanced Raman scattering (SERS) is used for the differentiation of human kidney adenocarcinoma, human kidney carcinoma, and non-cancerous human kidney embryonic cells. Silver nanoparticles (AgNPs) are used as substrate in the experiments. A volume of colloidal suspension containing AgNPs is added onto the cultured cells on a CaF2 slide, and the slide is dried at the overturned position. A number of SERS spectra acquired from the three different cell lines are statistically analyzed to differentiate the cells. Principal component analysis (PCA) combined with linear discriminate analysis (LDA) was performed to differentiate the three kidney cell types. The LDA, based on PCA, provided for classification among the three cell lines with 88% sensitivity and 84% specificity. This study demonstrates that SERS can be used to identify renal cancers by combining this new sampling method and LDA algorithms. Index Headings: Surface-enhanced Raman scattering; SERS; Renal carcinoma; Single cell; Differentiation; Principal component analysis-linear discriminate analysis; PCA-LDA.
INTRODUCTION Cancer continues to be one of the deadliest diseases worldwide. In the last two decades, a significant effort has been devoted to diagnosis and treatment. However, the diverse causes of malignance hinders the development of a solid approach for treatment. Therefore, early diagnosis gains importance in the successful treatment of the disease. Conventional cancer diagnostic approaches such as positron emission tomography (PET), computed tomography, magnetic resonance, and ultrasound are routinely used for cancer diagnosis.1 Magnetic resonance imaging identifies changes in the tissue morphology due to abnormal cell growth, while PET gives information about malignant tumors where cellular activity is increased.2 These techniques provide information about the disease mostly after the earliest stage of onset. Therefore, there is a need for the development of novel approaches for early detection. The use of spectroscopic techniques such as fluorescence,3 infrared (IR),4 and Raman5 are gaining importance in clinical research in recent years since they can provide fast molecular-level information with minimal surgical intervention or biopsy. The utility of vibrational spectroscopic techniques, IR and Raman, has long been investigated for tissue differentiation.6–8 A Raman spectrum can give direct information about molecular composition of a sample. This non-destructive and nonReceived 26 August 2013; accepted 11 February 2014. * Author to whom correspondence should be sent. E-mail: mculha@ yeditepe.edu.tr. DOI: 10.1366/13-07263
Volume 68, Number 6, 2014
invasive technique has been employed for population investigation for early diagnosis and the following phases of treatments.9 Raman spectroscopy especially has been used to examine samples obtained from breast,10 brain,6 cervical,11 nasopharyngeal,12 lung,13 and skin14 tissue in the literature. Surface-enhanced Raman spectroscopy (SERS) can be considered as a mode of Raman spectroscopy, which allows for the enhancement of Raman scattering of up to 108 when a molecule or molecular structure is located in the close vicinity of nanostructured noble metal surfaces.15 The sensitivity increase in SERS resulted in an enormous interest from the scientific communities from various disciplines to employ the technique for the detection and identification of DNA,16 proteins,17 microorganisms18 and living cells,19 and tissue.6 The first use of SERS at the living cancer cell level for the detection of antitumor drug concentrations was reported in 1991.20 Shortly after, a pioneering study demonstrated that SERS could be used to collect biomolecular information from living cells at the singlecell level.19 In that study, gold nanoparticles (AuNPs) internalized by the cells were used as substrates. In the following years, several reports concerning the use of SERS on living cells were appeared in the literature.13,21–31 All these studies demonstrate that SERS is slowly evolving as a novel technique to obtain molecular-level information at the single-cell level. Kidney cancer is almost 2% of all cancer cases around the world, although the most common form of kidney cancer, renal cell carcinoma (RCC), has increased in the West compared to Asia.32 Renal cell carcinoma has similar symptoms to many diseases. Only about 5% to 10% of people with RCC show classic symptoms such as abdominal mass, flank pain, and hematuria, and only 9% of RCC cases appear with these symptoms.33 Therefore, the symptoms of RCC can be easily confused with other diseases, and it makes diagnosis very difficult. With early diagnosis, RCC patients have a high survival rate. Research over the past 30 years on early detection, diagnosis, and treatment for patients with RCC has been increased and provided new imaging techniques for tumorous formations.34 In our study, we demonstrate the power of SERS using citrate-reduced silver nanoparticles (AgNPs) as substrates for the differentiation of cancer and healthy kidney cells. Human kidney adenocarcinoma (ACHN) cells with higher tumorigenic and metastatic potentiality, human kidney carcinoma (A-498) cells with lower tumorigenic potentiality, and human kidney embryonic (HEK 293; non-cancerous) cells were chosen as model cells. The sample preparation method that we employed
0003-7028/14/6806-0617/0 Q 2014 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
617
in our previous study for protein detection and identification with SERS was used in the study.35 The method uses the drying of samples containing colloidal AgNPs and analytes at the overturned position. This allows for a more uniform distribution of AgNPs on the droplet area instead of getting them jammed at the liquid–air–solid contact interface. The SERS spectra obtained from three cell lines were processed with linear discriminate analysis (LDA)36 based on the principal component analysis (PCA)37 method to differentiate the healthy cells from cancer cells. The results of this study with sensitivity and specificity of 87.8% and 84.4%, respectively, indicate that the SERS spectra obtained from the cells can be used for the diagnosis of the cancer.
MATERIALS AND METHODS Cell Cultures and Sample Preparation. Renal cancer cells (ACHN and A-498) and normal cells (HEK 293) were obtained from the American Type Culture Collection (ATCC). The human kidney adenocarcinoma cell line has high tumorgenic properties, while the A-498 cell line has low tumorigenic properties. Thus, HEK 293 is a noncancerous and immortalized cell line. The cells were taken out from a 80 8C freezer, thawed, and washed. Then, ACHN, A-498, and HEK 293 cells were incubated in T-75 flasks using glucose Dulbecco’s modified Eagle’s medium (DMEM; Sigma-Aldrich, Germany) with 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 1% penicillin/streptomycin in an incubator with a humidified 5% CO2 atmosphere at 37 8C prior to subculturing. For the subculturing of the HEK 293, A-498, and ACHN cells, the medium was removed and rinsed with phosphate buffered saline (PBS, 10X) solution, and 2 ml of 0.025% trypsin solution was added into the culture plate and placed at 37 8C until the cells detached. Then, the fresh medium was added to deactivate trypsin, and the cultures were transferred into the centrifuge tubes and centrifuged at 1500 rpm for 5 min. After the supernatant was removed, the fresh medium was added and dispensed into a new flask containing the fresh medium. After twice subculturing, the cells were moved to six-well cultivation plates with CaF2 slides and incubated at 37 8C. The cells were directly grown on CaF2 slides using standard procedure. Before the SERS measurements, the culture medium was removed, and the cells were washed three times with PBS. Then, CaF2 slides were rinsed with water to remove remnant salts. After the CaF2slides were dried, 2 lL of colloidal AgNPs was added onto the slides, and the slides were turned upside down to dry in a cell culture hood in sterile conditions. Synthesis of Colloidal Silver Nanoparticles (AgNPs). The colloidal AgNPs were synthesized by using sodium citrate as a reducing agent.38 Briefly, 90 mg of AgNO3 was dissolved in 500 mL of water, and the solution was kept on the heater until boiling. Then, 10 mL of 1% sodium citrate was added into the solution, and the solution was kept for 1.5 h until the volume reached the half volume. The concentration of this suspension was called 1X. The final concentration of AgNPs colloidal suspension was adjusted to 4X by centrifuging the 1X suspension and removing 3/4 of the supernatant for the use in SERS experiments. The characterization was
618
Volume 68, Number 6, 2014
performed with a Zetasizer (NanoSizer, Malvern, UK) and a ultraviolet/visible (UV/Vis) spectrometer (Perkin Elmer Lambda 25-UV). The average size distribution of the solution was around 60 nm, and the maximum absorption was recorded at 420 nm. Surface-Enhanced Raman Scattering (SERS) Measurements. A Renishaw InVia Reflex Raman microscopy system (Renishaw Plc., New Mills, Wotton-under-Edge, UK) equipped with an 830 nm diode was calibrated by using the silicon phonon mode at 520 cm 1. The incident laser power was 7.5 mW on the sample, and the spectral data acquisition time was 10 s. A 503 microscope objective (NA: 0.50) was used in the experiments. The SERS spectra were acquired over a spectral range of 500–1800 cm 1 because the specific Raman bands were observed in this range. The SERS spectra were collected over an area of 10 3 10 lm2 on the cell using the map image acquisition method function in WIRE 2.0 software with 1 lm steps collected. The experiment was repeated at least three times, and the spectral analyses were carried out by the WIRE 2.0 software. Data Analysis. The total number of spectra collected on a cell over 10 3 10 lm2 area was 100. A minimum of 30 cells for each cell line was used for the SERS acquisition for each experiment. Each of 100 SERS spectra obtained from each single cell was normalized to reduce the variations in intensity, and the total SERS spectra obtained from each cell were averaged to obtain a mean SERS spectrum. The high-dimensional SERS dataset was processed by the SPSS software package (SPSS Inc., Chicago), which is used for statistical analysis in the social sciences. The data mining, discrimination, and graphing techniques of SPSS are performed to differentiate the cell types. The principal component analysis (PCA) method was used to reduce the dimension of the SERS spectral dataset space by generating standardized matrices and principal component (PC) scores, the most significant variables produced from original spectra. Then, the linear discriminant analyses (LDA) method was used as a classification method. The most significant PC scores obtained by using independent sample ttests were used as variables in LDA. The LDA method provides discriminant functions, which increase the variances in SERS data between the different types of cells while decreasing the variances between the cells of the same type. The leave-one-out cross-validation method was used to indicate the classification results of PCA and LDA models accurately.
RESULTS AND DISCUSSIONS The CaF2 slides were used to culture the cells due to the background interference from the glass slide. After the cells were grown on the CaF2 slides, the medium was removed by washing with PBS, and remnants of PBS were washed with deionized and autoclaved water to avoid contamination. The slide with cells was placed in a safety cabinet until it was dried. Then, a 2 ll droplet of colloidal suspension containing AgNPs was added onto the cells while viewing under a light microscope and left to dry by turning the slide upside down. Multiple 2 ll droplets were spotted onto the cells grown on the CaF2 slide in order to place the AgNPs on more cells when
FIG. 1. (a) Image of ACHN cells under a 403 objective (NA:0.5), (b) image of setup used for droplet, (c) images of droplet area dried at suspended and (d) regular positions under 43 objective (NA:0.13).
necessary. A 403 microscope objective was used to observe the location of the cells and the aggregates of AgNPs in dried droplet area for each position. Figure 1 shows the sample preparation process (Figs. 1a and 1b) and the images (Figs. 1c and 1d). When a small volume of colloidal suspension containing AgNPs is placed on a surface, a phenomenon call ‘‘coffee ring’’ takes place and jams almost all of the particles at the liquid–solid–air interface. This causes the formation of very tightly packed particles at the droplet edges.39,40 For optimal SERS activity, a loose packing of noble metal particles is necessary.41–43 In order to avoid the unequal distribution AgNPs on the droplet area and jamming them at the droplet peripherals, the droplet placed onto the cells grown on CaF2 was overturned and dried at the suspended position (Fig. 1b). Figures 1c and 1d show the light microscopy images of the droplet areas at the suspended and regular position, respectively. As seen in the figure, the AgNPs are accumulated at the periphery of the droplet area when the droplet is dried at the regular position, while they are fairly distributed in the droplet area dried at the suspended position. The cells grown on CaF2 are washed with PBS and then water before letting them dry at room temperature. During drying, it is highly possible that most of the cells retained their morphology while a few of them crystallized. The washing procedure of the cells grown on CaF2 does not affect the cell morphology (Fig. 1a), and they were visible to select under microscope during the additional AgNP colloidal suspension and their selection for SERS measurements. The scanning electron microscopy (SEM) images of ACHN cells before and after addition of the AgNP colloidal suspension are given in Figs. 2a and 2b and Figs. 2c and 2d, respectively. As seen, the addition of colloidal suspension damages the cell integrity. It is clear that cell necrosis, or unprogrammed death, is an unfortunate outcome of the unexpected condition, drying, and adding colloidal suspension during the sample preparation. However, the AgNPs and their aggregates remained in the ruptured cell area. Therefore, the spectral features on the SERS spectra originate from biomacromolecules and ionic species composing the cell. The SERS spectra from an area of 10 3 10 lm2 on a cell were acquired by using the mapping method of WIRE 2.0 software with 1 lm steps. A total of 100 SERS spectra from one cell and a minimum of 30 cells for each cell line
FIG. 2. SEM images of ACHN cells (a) and (b) before, and (c) and (d) after treatment with AgNPs containing colloidal suspension.
were acquired. A minimum of 3000 SERS spectra from each cell line were collected, and the total SERS spectra acquired from healthy and cancer cell lines were averaged to one mean SERS spectrum, presented in Fig. 3a. The mean SERS spectra for each cell line was normalized at 529 cm 1 (Fig. 3b) to visualize the differences in spectral bands of the three cell types. As seen, there are several differences in the intensity and the shape of the bands appearing on each spectrum. The tentative band assignments based on the SERS spectra of 28 reference compounds with biological origin (provided in the Supplemental Material) are listed in Table I. A strong band at around 667 cm 1 is observed for each cell line, and this band is probably associated with m(C–S) stretching, which is usually observed in the range of 600–700 cm 1 as an intense band (Fig. S4b). Although there could be several molecular species such as amino acids, peptides, peptide derivatives, and proteins possessing a C–S bond and they may selectively interact with the AgNPs, the band at 660 cm 1 in cancer cell lines is claimed to be related to glutathione (GSH)43,44, a tripeptide containing cysteine. Glutathione is found in all human tissues and protects the fatty tissues from hazardous effects of free radicals.45 The band at around 660 cm 1 can also be attributed to uracil and lipoproteins (see Supplemental Material). The band at 732 cm 1 may originate from adenine, phospholipids, proteins, and nicotinamide adenine dinucleotide hydrogen (NADH; see Supplemental Material). For cellular growth and metabolism, NADH is required, and the increased concentration in cancer cells of this molecule may indicate this consequence. The band at around 800 cm 1 on the spectra of the cancer cells may originate from cytosine, tyrosine, and/or lactic acid. The increase in the concentration of these molecules can be related to the increased metabolism in cancer cells due to abnormal cellular growth. The band at around 906 cm 1, which is more intense in the spectra of cancer cells than of healthy cells, is assigned to proteins and
APPLIED SPECTROSCOPY
619
FIG. 3. (a) Mean SERS spectra from cultured cells: ACHN, A-498, and HEK 293. (b) Mean SERS spectra normalized to the intensity 529 cm
1
peak.
TABLE I. Tentative band assignments of SERS spectra of the cell lines Raman shift/ cm
1
Major assignments
HEK
ACHN
A-498
Saccharides/proteins Lipoprotein/RNA/glutathione Adenine (DNA/RNA)/phospholipids/proteins/NADH Cytosine (RNA/DNA)/proteins/lactic acid/glutathione Saccharides/proteins Phospholipids/proteins Proteins Proteins/lipids Proteins Cytosine and guanine ( DNA/RNA)/proteins Proteins/adenine (DNA/RNA) Proteins/DNA/NADH DNA/proteins/lactic acid Phospholipids/proteins DNA/proteins Phospholipids
530 668 732 797 911 965 993 1050 1102 1180 1278 1330 1412 1446 1577 –
530 670 732 800 914 965 993 1050 1102 – 1278 1333 – 1447 1576 1654
– 660 725 800 906 964 998 1040 1102 1180 1278 1330 1414 1448 1577 –
620
Volume 68, Number 6, 2014
FIG. 4. The PC scatter plots show the principal components of 30 cells from each cell line (ACHN, A-498, and HEK 293). (a) PC 1 and PC 2, (b) PC 1 and PC 3, (c) PC 2 and PC 3, (d) PC 1 and PC 4, (e) PC 2 and PC 4, and (f) PC 3 and PC 4.
APPLIED SPECTROSCOPY
621
mainly glucose or other possible monosaccharides from the SERS spectra of the biomolecules listed in the Supplemental Material. The band at around 964 cm 1 is attributed to phospholipids and proteins, and the band at 993 cm 1 is assigned to phenylalanine and aspartic acid from proteins. The anticipated increase in the intensity of this band was observed on the spectra of cancer cells compared to the spectra of the HEK 293 cells. This increase in phospholipids concentration is obtained in cancer cells due to the possible increase in fatty acid production. As the cell division is accelerated in cancer cells, they need to produce new phospholipids to maintain cell membrane integrity. In addition, as mentioned above, the energy need in cancer cells increases due to the increased metabolism. The bands in the region of 1000–1200 cm 1 may originate from lipids, proteins, and nucleic acids. The bands in the region of 1200–1300 cm 1 may originate from tyrosine and adenine. The band examined at around 1300 cm 1 is broader for cancer cells than the band at this wavenumber for the healthy cells. The difference in the intensity of the bands at 529, 1414, and 1446 cm 1 can be explained with the concentration difference of saccharides, proteins, and lipids in the cancer cells. The bands in the region of 1500–1600 cm 1 are attributed to the pyrimidine ring of nucleotides and/or proteins.46 In addition, the band at around 1576 cm 1, which was assigned to Amide II46 and/or DNA, is broader for the A-498 cell line than the band on the spectra of the other cell lines, but the intensity of this band is higher for the one for the cancer cells than the one for healthy cells. Linear discriminate analysis (LDA), a classical discriminating method, was used in our study to differentiate the cells in high reliability. The high-dimensional SERS dataset was reduced, and new relevant variables, principal component (PC) scores, were obtained by using PCA, which is an eigenvalue–eigenvector method. Four PC component eigenvectors were extracted, and PC 1 (44, 25% of the total variance), PC 2 (34, 51%), PC 3 (13, 86%), and PC 4 (2, 31%) were plotted by using a scatter plot to show the values of the three variables, ACHN, HEK, and A-498 cell lines (Fig. 4). Each dot represents one cell. Here, PC 1 and PC 2 vectors could identify the cells with a sensitivity and specificity of 85% and 84%, respectively, while others are grouped in a low correlation. Also, PC 1 and PC 3 could classify the cell lines with a sensitivity and specificity of 70% and 68%, respectively, while PC 1 and PC 4 could distinguish the cells with sensitivity and specificity of 70% and 69%, respectively. As well, PC 2 and PC 3 could diagnose the three types of cells with a sensitivity and specificity of 71% and 69%, respectively. In addition, PC 2 and PC 4 could differentiate the cell lines with a sensitivity and specificity of 76% and 73%, respectively, while PC 3 and PC 4 could separate the cell lines with a sensitivity and specificity of 46% and 42%, respectively. Here, PC 1 and PC 2 vectors have large means and significant differences between the groups compared to PC 3 and PC 4 vectors. Principal component analysis is not the best classification method for our SERS dataset, although it is a good way to summarize high datasets into subsets of variables. Linear discriminant analysis
622
Volume 68, Number 6, 2014
FIG. 5. LD 1 and LD 2 scatter plot of three types of cells.
arranges discriminant functions by maximizing the variances in the SERS data between different cell groups while minimizing the variances between the cells of the same group. The most significant PCs obtained by using independent-sample t-tests were used in LDA. The scores on LD 1 and LD 2 were also plotted by using a scatter plot (Fig. 5). The ACHN, HEK, and A-498 cells are differentiated with a sensitivity and specificity of 88% and 84%, respectively, using LDA. The differentiating results are obtained using the leave-one-out cross-validation method of PCA and LDA models. As a result, the clustering of three cells with sensitivity and specificity of 88% and 84%, respectively, is a high rate to detect the cancer types of RCC.
CONCLUSION In this study, the use of SERS to identify renal cancers by combining this new sampling method and LDA algorithms was demonstrated. The citrate-reduced AgNPs were used as SERS substrates, and the AgNP colloidal suspension placed onto the cells grown on the slide was dried at the suspended position. Since the AgNPs and their aggregates are localized on the cell surface area with this sample preparation method, the observed bands on the SERS spectra of the cells can be attributed to the released metabolites or biomolecules originating from the cell surface and intracellular molecules and molecular structures due to the cell rupture. The observed changes on the SERS spectra can be related to altered biological function in healthy versus cancer cells, which could be used for early diagnosis or cell typing. This study was the first to differentiate the renal cancer cells from healthy cells with high correlation by using algorithms-based PCA and LDA on the SERS spectra set. This diagnostic approach offers the possibility and simplicity for the future studies of SERS in categorizing and typing cell lines.
ACKNOWLEDGMENTS The authors thank Associate Professor Dilek Telci for providing cell lines. The financial support of Yeditepe University and TUBITAK is also greatly acknowledged. SUPPLEMENTAL MATERIAL All supplemental materials mentioned in the text, including six figures, one table, and a description of the procedure, can be found in the online version of the journal at http://www.s-a-s.org. 1. L. Fass. ‘‘Imaging and Cancer: A Review’’. Mol. Oncol. 2008. 2(2): 115-152. 2. X. Gao, Y. Cui, R.M. Levenson, L.W. Chung, S. Nie. ‘‘In Vivo Cancer Targeting and Imaging with Semiconductor Quantum Dots’’. Nat. Biotechnol. 2004. 22: 969-976. 3. J.R. Lakowicz. ‘‘Principles of Fluorescence Spectroscopy’’. New York: Springer, 2006. 3rd ed. 4. D. Lin-Vien, N.B. Colthup, W.G. Fateley, J.G. Graselli. ‘‘The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules’’. San Diego: Academic Press, 1991. 5. N.B. Colthup, L.H. Daly, S.E. Wiberley. ‘‘Introduction to Infrared and Raman Spectroscopy POD’’. San Diego: Academic Press, 1990. 3rd ed. 6. O. Aydin, M. Altas, M. Kahraman, O.F. Bayrak, M. Culha. ‘‘Differentiation of Healthy Brain Tissue and Tumors Using Surface-Enhanced Raman Scattering’’. Appl. Spectrosc. 2009. 63(10): 1095-1100. 7. O. Aydin, M. Kahraman, E. Kilic, M. Culha. ‘‘Surface-Enhanced Raman Scattering of Rat Tissues’’. Appl. Spectrosc. 2009. 63(6): 662-668. 8. R. Petry, M. Schmitt, J. Popp. ‘‘Raman Spectroscopy—A Prospective Tool in the Life Sciences’’. ChemPhysChem. 2003. 4(1): 14-30. 9. C. Kendall, N. Stone, N. Shepherd, K. Geboes, B. Warren, R. Bennett, H. Barr. ‘‘Raman Spectroscopy, A Potential Tool for the Objective Identification and Classification of Neoplasia In Barrett’s Oesophagus’’. J. Pathol. 2003. 200(5): 602-609. 10. C. Yu, E. Gestl, K. Eckert, D. Allara, J. Irudayaraj. ‘‘Characterization of Human Breast Epithelial Cells by Confocal Raman Microspectroscopy’’. Cancer Detect. Prev. 2006. 30(6): 515-522. 11. F.M. Lyng, E.O´. Faola´in, J. Conroy, A.D. Meade, P. Knief, B. Duffy, M.B. Hunter, J.M. Byrne, P. Kelehan, H.J. Byrne. ‘‘Vibrational Spectroscopy for Cervical Cancer Pathology, from Biochemical Analysis to Diagnostic Tool’’. Exp. Mol. Pathol. 2007. 82(2): 121-129. 12. S. Feng, R. Chen, J. Lin, J. Pan, G. Chen, Y. Li, M. Cheng, Z. Huang, J. Chen, H. Zeng. ‘‘Nasopharyngeal Cancer Detection Based on Blood Plasma Surface-Enhanced Raman Spectroscopy and Multivariate Analysis’’. Biosens. Bioelectron. 2010. 25(11): 2414-2419. 13. X. Qian, X.-H. Peng, D.O. Ansari, Q. Yin-Goen, G.Z. Chen, D.M. Shin, L. Yang, A.N. Young, M.D. Wang, S. Nie. ‘‘In Vivo Tumor Targeting and Spectroscopic Detection with Surface-Enhanced Raman Nanoparticle Tags’’. Nat. Biotechnol. 2007. 26: 83-90. 14. S. Sigurdsson, P.A. Philipsen, L.K. Hansen, J. Larsen, M. Gniadecka, H.C. Wulf. ‘‘Detection of Skin Cancer by Classification of Raman Spectra’’. IEEE Trans. Biomed. Eng. 2004. 51(10): 17841793. 15. K. Kneipp, M. Moskovits, H. Kneipp. ‘‘Surface-Enhanced Raman Scattering: Physics and Applications’’. Berlin: Springer-Verlag, 2006, vol. 103. 16. K. Kneipp, H. Kneipp, V.B. Kartha, R. Manoharan, G. Deinum, I. Itzkan, R.R. Dasari, M.S. Feld. ‘‘Detection and Identification of A Single DNA Base Molecule Using Surface-Enhanced Raman Scattering (SERS)’’. Phys. Rev. E. 1998. 57: 6281(R)-R6284(R). 17. Y. Wang, H. Wei, B. Li, W. Ren, S. Guo, S. Dong, E. Wang. ‘‘SERS Opens A New Way in Aptasensor for Protein Recognition with High Sensitivity and Selectivity’’. Chem. Commun. 2007. 48: 5220-5222. 18. W.R. Premasiri, D.T. Moir, M.S. Klempner, N. Krieger, G. Jones II, L.D. Ziegler. ‘‘Characterization of the Surface Enhanced Raman Scattering (SERS) of Bacteria’’. J. Phys. Chem. B. 2005. 109(1): 312320. 19. K. Kneipp, A.S. Haka, H. Kneipp, K. Badizadegan, N. Yoshizawa, C. Boone, K.E. Shafer-Peltier, J.T. Motz, R.R. Dasari, M.S. Feld. ‘‘Surface-Enhanced Raman Spectroscopy in Single Living Cells Using Gold Nanoparticles’’. Appl. Spectrosc. 2002. 56(2): 150-154.
20. I.R. Nabiev, H. Morjani, M. Manfait. ‘‘Selective Analysis of Antitumor Drug Interaction with Living Cancer Cells As Probed by Surface-Enhanced Raman Spectroscopy’’. Eur. Biophys. J. 1991. 19(6): 311-316. 21. G. Breuzard, J.F. Angiboust, P. Jeannesson, M. Manfait, J.M. Millot. ‘‘Surface-Enhanced Raman Scattering Reveals Adsorption of Mitoxantrone on Plasma Membrane of Living Cells’’. Biochem. Biophys. Res. Commun. 2004. 320(2): 615-621. 22. C. Eliasson, A. Lore´n, J. Engelbrektsson, M. Josefson, J. Abrahamsson, K. Abrahamsson. ‘‘Surface-Enhanced Raman Scattering Imaging of Single Living Lymphocytes with Multivariate Evaluation’’. Spectrochim. Acta, Part A. 2005. 61(4): 755-760. 23. X. Huang, I.H. El-Sayed, W. Qian, M.A. El-Sayed. ‘‘Cancer Cell Imaging and Photothermal Therapy in the Near-Infrared Region by Using Gold Nanorods’’. J. Am. Chem. Soc. 2006. 128(6): 2115-2120. 24. X. Huang, I.H. El-Sayed, W. Qian, M.A. El-Sayed. ‘‘Cancer Cells Assemble and Align Gold Nanorods Conjugated to Antibodies to Produce Highly Enhanced, Sharp, and Polarized Surface Raman Spectra: A Potential Cancer Diagnostic Marker’’. Nano Lett. 2007. 7(6): 1591-1597. 25. J.-H. Kim, J.-S. Kim, H. Choi, S.-M. Lee, B.-H. Jun, K.-N. Yu, E. Kuk, Y.-K. Kim, D.H. Jeong, M.-H. Cho, Y.-S. Lee. ‘‘Nanoparticle Probes with Surface Enhanced Raman Spectroscopic Tags for Cellular Cancer Targeting’’. Anal. Chem. 2006. 78(19): 6967-6973. 26. K. Kneipp, H. Kneipp, J. Kneipp. ‘‘Surface-Enhanced Raman Scattering in Local Optical Fields of Silver and Gold Nanoaggregates from Single-Molecule Raman Spectroscopy to Ultrasensitive Probing in Live Cells’’. Acc. Chem. Res. 2006. 39(7): 443450. 27. S. Shanmukh, L. Jones, J. Driskell, Y. Zhao, R. Dluhy, R.A. Tripp. ‘‘Rapid and Sensitive Detection of Respiratory Virus Molecular Signatures Using a Silver Nanorod Array SERS Substrate’’. Nano Lett. 2006. 6(11): 2630-2636. 28. T. Vo-Dinh, L.R. Allain, D.L. Stokes. ‘‘Cancer Gene Detection Using Surface-Enhanced Raman Scattering (SERS)’’. J. Raman Spectrosc. 2002. 33(7): 511-516. 29. Z. Wang, A. Bonoiu, M. Samoc, Y. Cui, P.N. Prasad. ‘‘Biological pH Sensing Based on Surface Enhanced Raman Scattering Through a 2-aminothiophenol-Silver Probe’’. Biosens. Bioelectron. 2008. 23(6): 886-891. 30. W. Xie, L. Wang, Y. Zhang, L. Su, A. Shen, J. Tan, J. Hu. ‘‘Nuclear Targeted Nanoprobe for Single Living Cell Detection by SurfaceEnhanced Raman Scattering’’. Bioconjugate Chem. 2009. 20(4): 768773. 31. S. Feng, J. Lin, Z. Huang, G. Chen, W. Chen, Y. Wang, R. Chen, H. Zeng. ‘‘Esophageal Cancer Detection Based on Tissue SurfaceEnhanced Raman Spectroscopy and Multivariate Analysis’’. Appl. Phys. Lett. 2013. 102(4): 043702-043704. 32. W.-H. Chow, L.M. Dong, S.S. Devesa. ‘‘Epidemiology and Risk Factors for Kidney Cancer’’. Nat. Rev. Urol. 2010. 7(5): 245-257. 33. B. Ljungberg, D.C. Hanbury, M.A. Kuczyk, A.S. Merseburger, P.F. Mulders, J.-J. Patard, I.C. Sinescu. ‘‘Renal Cell Carcinoma Guideline’’. Eur. Urol. 2007. 51(6): 1502-1510. 34. W.C. Black, H.G. Welch. ‘‘Advances in Diagnostic Imaging and Overestimations of Disease Prevalence and the Benefits of Therapy’’. N. Engl. J. Med. 1993. 328(17): 1237-1243. 35. S. Keskin, M. Culha. ‘‘Label-Free Detection of Proteins From DriedSuspended Droplets Using Surface Enhanced Raman Scattering’’. Analyst. 2012. 137(11): 2651. 36. M. Li, B. Yuan. ‘‘2D-LDA: A Statistical Linear Discriminant Analysis for Image Matrix’’. Pattern Recognit. Lett. 2005. 26(5): 52-532. 37. I.T. Jolliffe. ‘‘Principal Component Analysis’’. New York: John Wiley & Sons Inc. 2005. 38. P.C. Lee, D. Meisel. ‘‘Adsorption and Surface-Enhanced Raman of Dyes on Silver and Gold Sols’’. J. Phys. Chem. 1982. 86. (17): 33913395. 39. X. Shen, C.-M. Ho, T.-S. Wong. ‘‘Minimal Size of Coffee Ring Structure’’. J. Phys. Chem. B. 2010. 114(16): 5269-5274. 40. H. Hu, R.G. Larson. ‘‘Marangoni Effect Reverses Coffee-Ring Depositions’’. J. Phys. Chem. B. 2006. 110(14): 7090-7094. 41. J. Jiang, K. Bosnick, M. Maillard, L. Brus. ‘‘Single Molecule Raman Spectroscopy at the Junctions of Large Ag Nanocrystals’’. J. Phys. Chem. B. 2003. 107(37): 9964-9972. 42. V. Safonov, V. Shalaev, V. Markel, Y.E. Danilova, N. Lepeshkin, W. Kim, S. Rautian, R. Armstrong. ‘‘Spectral Dependence of Selective
APPLIED SPECTROSCOPY
623
Photomodification in Fractal Aggregates of Colloidal Particles’’. Phys. Rev. Lett. 1998. 80(5): 1102. 43. K. Kneipp, H. Kneipp, P. Corio, S. Brown, K. Shafer, J. Motz, L. Perelman, E. Hanlon, A. Marucci, G. Dresselhaus. ‘‘SurfaceEnhanced and Normal Stokes and Anti-Stokes Raman Spectroscopy of Single-Walled Carbon Nanotubes’’. Phys. Rev. Lett. 2000. 84(15): 3470.
624
Volume 68, Number 6, 2014
44. J. De Gelder, K. De Gussem, P. Vandenabeele, L. Moens. ‘‘Reference Database of Raman Spectra of Biological Molecules’’. J. Raman Spectrosc. 2007. 38(9): 1133-1147. 45. R. Brigelius-Flohe. ‘‘Tissue-Specific Functions of Individual Glutathione Peroxidases’’. Free Radical Biol. Med. 1999. 27(9-10): 951965. 46. Z. Movasaghi, S. Rehman, I.U. Rehman. ‘‘Raman Spectroscopy of Biological Tissues’’. Appl. Spectrosc. Rev. 2007. 42(5): 493-541.
