Identification of Tumor Cells through Spectroscopic Profiling of the

pubs.acs.org/JPCL

Identification of Tumor Cells through Spectroscopic Profiling of the Cellular Surface Chemistry €rn M. Reinhard* Bo Yan and Bjo Department of Chemistry and The Photonics Center, Boston University, Boston, Massachusetts 02215

ABSTRACT A key challenge for molecular cancer diagnostics is the identification of appropriate biomarkers and detection modalities. One target of interest is the cell surface, which is involved in important steps of cancerogenesis. Here, we explore the feasibility of a label-free spectroscopic profiling of the cell surface chemistry for the detection of tumor cells. Vibrational spectra of the cellular surfaces of tumor and nontumor breast and prostate cell lines were recorded on silver nanoparticle substrates using surface-enhanced Raman spectroscopy (SERS). The quantitative analysis of the spectra revealed characteristic differences, especially in the spectral range between 600 and 900 cm-1. The detection of tumor-cell-specific differences in the recorded SERS spectra indicates the possibility of improving the precision of current cancer detection and staging approaches through a spectroscopic profiling of cell surface cancer biomarkers. SECTION Biophysical Chemistry

C

spectroscopy relies on the amplification of Raman cross sections through noble metal nanostructures, which effectively enhance the incident electromagnetic field. The rational of our experimental approach is that due to the distance dependence of the signal enhancement (only molecules within ∼10 nm of the metal surface contribute to the signal14,15), SERS enables one to selectively probe the composition of the surface of a cell resting on a nanostructured metal substrate. Different from conventional Raman spectroscopy that provides molecular information of the entire cell,16-21 SERS interrogates selectively the nanoparticle-cell interface. SERS could therefore potentially enable a spectroscopic identification of tumor cells based on cell-surfacespecific molecular features. To test this hypothesis, we performed SERS of tumor and nontumor breast (MCF7; MCF10A) and prostate (PC3; RWPE1) cell lines on Ag nanoparticle agglomerate substrates. A dilute solution of washed cells that were detached from culture plates by nonenzymatic ethylenediamine tetraaceticacid (EDTA) solution was pipetted on the substrate and dried. Figure 1 shows a representative optical image and scanning electron microscope (SEM) images of prostate cells on a SERS substrate. After a cluster of >4 cells was localized on the SERS chip in the optical microscope, a SERS spectrum was recorded. Figure 2A1 and B1 shows SERS spectra of tumor (red) and nontumor (blue) breast and prostate cells. Each spectrum is an average of 36 individual measurements of three independent experiments. A detailed list of the individual bands is given in Figure S1

onventional histopathological cancer diagnostics has recently been complemented by advanced imaging techniques to locate tumors, such as X-ray-computed tomography (CT), magnetic resonance imaging (MRI), or positron emission tomography (PET). CT and MRI recognize changes in the tissue morphology resulting from an abnormal cell growth, whereas PET detects areas of increased cellular activity indicative of malignant tumors.1 While CT, MRI, and PET have led to significant advances in cancer diagnostics, research, and therapy, the false rate in the diagnosis of some cancers such as prostate cancer2 remains high. This is especially true for early stages of the disease. To further improve diagnostic reliability and detection sensitivity, ideally to a level that precancerous lesions can be detected, novel chemical imaging and sensing modalities are required that enable a diagnosis based on quantifiable and cancer-specific molecular signatures.3,4 A key challenge is to identify appropriate molecular cancer biomarkers and to develop the corresponding sensors that allow the detection of these biomarkers with high sensitivity.5 One promising target for cancer diagnostics is the cell surface. Changes in the composition of the lipid bilayer,6 cell surface glycoproteins,7 and glycolipids8 and the morphology9 of the cell surface are all indicated to play important roles in carcinogenesis. To improve cell-surface-based molecular cancer diagnostics, entirely new approaches are sought that provide quick insight into the composition of the plasma membrane and the glycoprotein coating (glycocalyx) of intact cells without the need for complicated sample preparation. Many of the above requirements could be met by an optical spectroscopy that provides molecular-specific information about the surfaces of entire cells. The method of choice for this application is surfaceenhanced Raman spectroscopy (SERS).10-13 This vibrational

r 2010 American Chemical Society

Received Date: April 8, 2010 Accepted Date: April 28, 2010 Published on Web Date: May 03, 2010

1595

DOI: 10.1021/jz100459m |J. Phys. Chem. Lett. 2010, 1, 1595–1598

pubs.acs.org/JPCL

(Supporting Information). In Figure 2B1, we also included averaged background SERS spectra (36 measurements) from the supernatant of the prostate cells acquired under identical conditions. Although the background shares some spectral features with the cell surface due to cellular debris in the solution, the signal intensities from the cells are much higher, confirming that the recorded signals are cell-specific. Since the major building blocks of the investigated cell surfaces are identical, the recorded cell surface spectra in

Figure 2A1 and B1 share many common spectral features. Subtle differences can arise, for instance, from differences in the relative contribution of the individual cell surface components. To verify and quantify these differences, we analyzed the recorded spectra with principal component analysis (PCA). This is a mathematical procedure in which a set of correlated variables (in this case, the spectra) is reduced to a smaller set of uncorrelated variables (principal components (PCs)) which display as much variation in the data as possible.22 The PC values (the so-called scores) for the individual measurements in a data set provide a quantitative metric to detect differences between complex spectra. Figure 2A2 and B2 shows score plots in the three-dimensional PC spaces which achieve the best separations of the recorded data for the respective cell line pairs. For breast and prostate cell lines, PCA leads to a clustering of tumor (TC) and nontumor (NTC) cells in different, nonoverlapping regions of the PC space. We also performed PCA for the whole data set without distinguishing between the different cell types and still obtained a clear separation of TCs and NTCs (Figure S2, Supporting Information). These observations indicate general cell-type independent spectral differences between TCs and NTCs that enable a robust and reliable separation despite fluctuation in the SERS enhancement provided by the random SERS substrates used. To identify the spectral features that are most relevant for the spectral differentiation of TCs and NTCs, we analyzed the loading spectra of those PCs along which the clusters separate most clearly. The relevant PCs are PC3 for the breast and PC4 for the prostate cells. The loading spectra of these PCs are plotted in Figure 2A3 and B3. In both cases, the 722 cm-1 band dominates and has the highest absolute contribution to

Figure 1. (A) The field enhancement generated in the vicinity of Ag nanoparticles enables recording of vibrational spectra of cell surfaces using surface-enhanced Raman spectroscopy (SERS). (B) Optical dark-field backscattering images of PC3 cells on a Ag nanoparticles substrate. SEM images of individual cells on a Ag nanoparticles substrate from the top view (C) and sideview (D). Scale bars correspond to 5 μm in (B) and 2 μm in (C) and (D).

Figure 2. SERS Spectra (A1 and B1), PC plots (A2 and B2), and loading spectra (A3 and B3) of tumor (red) and nontumor (blue) breast (A) and prostate (B) cell lines. The dashed lines in B1 show the background SERS signal from the cell supernatant. The SERS spectra in A1 and B1 are the averages of 36 measurements of 3 independent experiments, and the PC plots in A2 and B2 contain the individual measurements.


1596


pubs.acs.org/JPCL

that the cell surface chemistry of the investigated TCs and NTCs contains sufficient differences for a reliable spectroscopic differentiation using SERS. The intensity ratio (R) of the marker bands at 722 and 655 cm-1 was sufficient to differentiate between TCs and NTCs. We only investigated two different pairs of tumor and nontumor in vitro cell lines; our work therefore motivates further systematic studies using a larger set of in vitro cell lines and cancer tumor tissue to verify if this approach is generally applicable. Remote SERS sensing through optical fibers has been demonstrated, and endoscopic in vivo SERS imaging modalities are currently under development,27-30 potentially paving the way to a SERSbased molecular cancer diagnostics in vivo.

Figure 3. Intensity ratio R of the 722 and 655 cm-1 bands.

the PC. This finding indicates that cancer-specific changes in the cell surface chemistries, which are independent of the cell source (breast or prostate), cause the intensity differences in the 722 cm-1 band. Adetailed analysis reveals that changes in the spectral intensity of the 722 cm-1 band are anticorrelated with changes in the 655 cm-1 band, which is another prominent feature in all recorded spectra. In fact, the evaluation of the band intensity ratio R = I722/I655 is sufficient to differentiate between the TCs and NTCs, as shown in Figure 3. At least for the cell lines investigated in this work, R represents a metric that is easy to quantify and sufficient to discriminate between TCs and NTCs. The R threshold values to discriminate between tumor and normal cell lines were R=0.74 for breast and R=1.12 for prostate cell pairs. An important question is the molecular origin of the 722 and 655 cm-1 bands. We cannot exclude that during the sample preparation some DNA, which has a characteristic band close to 722 cm-1,23 is released. However, the DNA band is significantly shifted and can be differentiated in our experimental setup (Figure S3, Supporting Information). Furthermore, control experiments with plasma membranes from lysed cells also show a strong 722 cm-1 band (Figure S4, Supporting Information). This observation indicates that the 722 cm-1 band arises from molecular species located at the cell surface, and we assign this band to the C-N bond of quaternary ammonium groups in phosphatidylcholines and sphingomyelins,24 which are the main components of the membrane lipid bilayer. Consistent with this interpretation, the 953 cm-1 band of choline is also stronger for NTCs than that for TCs.24 The 655 cm-1 band which is stronger in the TC spectra can be assigned to the C-S stretch mode in cysteines25 and is also commonly observed in the Raman spectrum of sialic acids.26 Malignant tumor cells commonly overexpress cell surface proteins which are often glycosylated. Considering the strong distance dependence of the SERS effect, a higher density of glycoproteins on the surface of TCs, where they effectively act as spacers between the lipid bilayer and the Ag nanoparticles, can account for the relative intensities of the 722 and 655 cm-1 bands in the SERS spectra of TCs and NTCs. This model is consistent with recent mechanical studies of the cell surface using AFM by Iyer et al.,9 which revealed that cancer cells exhibit thicker brush-type structures on their surface than their healthy counterparts. Independent of the exact molecular nature of the observed spectral differences, our in vitro SERS studies demonstrate


EXPERIMENTAL METHODS Cell Culturing and Sample Preparation. Tumor cell lines MCF7 (breast cell) and PC3 (prostate cell) and nontumor cell lines MCF10A (breast cell) and RWPE1 (prostate cell) were cultured in 25 cm2 cell culture flasks following the supplier's instructions (ATCC). The cells were detached from the flask surface with nonenzymatic EDTA solution at 37 °C for 5 min. Then, the suspended cells were washed with Hanks' buffer by centrifugation and resuspension for three times. In a final centrifugation step, the cell suspension was concentrated to a volume of approximately 20 μL; 10 μL of this solution was pipetted onto a SERS chip and dried for 10 min. For background measurements, the respective supernatant from the final 20 μL solution was deposited onto the substrate. SERS chips were prepared through drop-coating of a 40 nm Ag colloid (1 1012 particles/mL) on a silicon wafer and used immediately. SERS Measurements. All SERS measurements were performed on an upright microscope (Olympus BX51WI) equipped with a 300 mm focal length imaging spectrometer (Andor Shamrock 303i) and a back-illuminated CCD camera (Andor Idus) optimized for the near-infrared (peak quantum efficiency >90% at 785 nm). A 600 lines/mm grating with a blazing wavelength of 750 nm was used. The excitation laser was a 785 nm diode laser (Innovative Photonic Solutions). After passing through a 785 nm laser line filter (Semrock, LL01-785-25), the laser light was injected into the objective using a dichroic (Semrock, LPD-785RU) and focused into the sample plane by a 40 air objective (numerical aperture (NA) = 0.65). The laser power at the sample was 21.5 mW (corresponding to an irradiance of 0.004 mW/μm2). Light scattered off the sample was collected by the same objective and filtered by the dichroic and an additional 803 nm long pass filter (Semrock, LP02-785-RS). The active area for recording SERS spectra was limited by a slit in the entrance port of the spectrometer to 2 μm 78 μm. Typically 4-5 cells filled this region. Ten individual acquisitions with a 2 s integration time were accumulated for each spectrum. Data Analysis. Data processing was performed using home-written Matlab (version 2007B) analysis codes. First, the spectra were baseline-corrected by subtraction of a linear fit to the baseline of the spectrum and normalized by dividing the spectrum through the peak intensity of the most intense band. Principal component analysis (PCA) was

1597


pubs.acs.org/JPCL

performed using the PCA implementation in the Matlab statistical toolbox. (15)

SUPPORTING INFORMATION AVAILABLE Figures S1-S4. This material is available free of charge via the Internet at http:// pubs.acs.org.

(16)

AUTHOR INFORMATION Corresponding Author: *To whom correspondence should be addressed. E-mail: bmr@ bu.edu.

(17)

ACKNOWLEDGMENT This work was financially supported by the National Science Foundation through Grant CBET-0853798. (18)

REFERENCES (1) (2)

(3)

(4) (5) (6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(19)

Weissleder, R. Scaling Down Imaging Molecular Mapping of Cancer in Mice. Nat. Rev. Cancer 2002, 2, 1–8. Kilpelainen, T. P.; Tammela, T. L. J.; Maattanen, L.; Kujala, P.; Stenman, U.-H.; Ala-Opas, M.; Murtola, T. J.; Auvinen, A. False-Positive Screening Results in the Finnish Prostate Cancer Screening Trial. Br. J. Cancer 2010, 102, 469–474. Kumar, S.; Richards-Kortum, R. Optical Molecular Imaging Agents for Cancer Diagnostics and Therapeutics. Nanomedicine 2006, 1, 23–30. Weissleder, R.; Pittet, M. J. Imaging in the Era of Molecular Oncology. Nature 2008, 452, 580–589. Sawyers, C. L. The Cancer Biomarker Problem. Nature 2008, 452, 548–552. Schlager, S. I.; Ohanian, S. H. Tumor Cell Lipid Composition and Sensitivity to Humoral Immune Killin. I. Modification of Cellular Lipid and Fatty Acid Content by Metabolic Inhibitors and Hormones. J. Immunol. 1980, 124, 626–634. Dube, D. H.; Bertozzi, C. R. Glycans in Cancer and Inflammation Potential for the Therapeutics and Diagnostics. Nat. Rev. Drug Discovery 2005, 4, 477–488. Hakomori, S. Tumor Malignancy Defined by Aberrant Glycosylation and Sphingo(glyco) Lipid Metabolism. Cancer Res. 1996, 56, 5309–5318. Iyer, S.; Gaikwad, R. M.; Subba-Rao, V.; Woodworth, C. D.; Sokolov, I. Atomic Force Microscopy Detects Differences in the Surface Brush of Normal and Cancerous Cells. Nat. Nanotechnol. 2009, 4, 389–393. Jeanmaire, D. L.; van Duyne., R. P. Surface Raman Spectroelectrochemistry Part I. Heterocyclic, Aromatic, and Aliphatic Amines Adsorbed on the Anodized Silver Electrode. J. Electroanal. Chem. 1977, 84, 1–20. Kneipp, K.; Kneipp, H.; Itzkan, I.; Dasari, R. R.; Feld, M. S. Ultrasensitive Chemical Analysis by Raman Spectroscopy. Chem. Rev. 1999, 99, 2957–2976. Qian, X.; Peng, X.-H.; Ansari, D. O.; Yin-Goen, Q.; Chen, G. Z.; Shin, D. M.; Yang, L.; Young, A. N.; Wang, M. D.; Nie, S. In Vivo Tumor Targeting and Spectroscopic Detection with SurfaceEnhanced Raman Nanoparticle Tags. Nat. Biotechnol. 2008, 26, 83–90. Huang, X.; El-Sayed, I. H.; Qian, W.; El-Sayed, M. A. 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, 1591–1597. Lal, S.; Grady, N. K.; Goodrich, G. P.; Halas, N. J. Profiling the Near Field of a Plasmonic Nanoparticle with


(20)

(21)

(22) (23)

(24)

(25)

(26)

(27) (28)

(29)

(30)

1598

Raman-Based Molecular Rulers. Nano Lett. 2006, 6, 2338–2343. Kennedy, B. J.; Spaeth, S.; Dickey, M.; Carron, K. T. Determination of the Distance Dependence and Experimental Effects for Modified SERS Substrates Based on Self-Assembled Monolayers Formed Using Alkanethiols. J. Phys. Chem. B 1999, 103, 3640–3646. Puppels, G. J.; Mul, F. D.; Otto, C.; Greve, J.; Robert-Nicoud, M.; Arndt-Jovin, D. J.; Jovin, T. M. Studying Single Living Cells and Chromosomes by Confocal Raman Microspectroscopy. Nature 1990, 347, 301–303. Chan, J. W.; Taylor, D. S.; Zwerdling, T.; Lane, S. M.; Ihara, K.; Huser, T. Micro-Raman Spectroscopy Detects Individual Neoplastic and Normal Hematopoietic Cells. Biophys. J. 2006, 90, 648–656. Mahadevan-Jansen, A.; Richards-Kortum, R. R. Raman Spectroscopy for the Detection of Cancers and Precancers. J. Biomed. Opt. 1996, 1, 31–70. Zeng, H.; McWilliams, A.; Lam, S. Optical Spectroscopy and Imaging for Early Lung Cancer Detection: A Review. Photodiagnosis Photodyn. Ther. 2004, 1, 111–122. Utzinger, U.; Heintzelman, D. L.; Mahadevan-Jansen, A.; Malpica, A.; Follen, M.; Richards-Kortum, R. Near-Infrared Raman Spectroscopy for In Vivo Detection of Cervical Precancers. Appl. Spectrosc. 2001, 55, 955–959. Haka, A. S.; Shafer-Peltier, K. E.; Fitzmaurice, M.; Crowe, J.; Dasari, R. R.; Feld, M. S. Diagnosing Breast Cancer by Using Raman Spectroscopy. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 12371–12376. Jolliffe, I. T. Principal Component Analysis, 2nd ed.; Springer: Berlin, Germany, 2002. Barhoumi, A.; Zhang, D.; Tam, F.; Halas, N. J. SurfaceEnhanced Raman Spectroscopy of DNA. J. Am. Chem. Soc. 2008, 130, 5523–5529. Akutsu, H. Direct Determination by Raman Scattering of the conformation of the Choline Group in Phospholipid Bilayers. Biochemistry 1981, 20, 7359–7366. Tuma, R.; Russel, M.; Rosendahl, M.; Thomas, G. Solution Conformation of the Extracellular Domain of the Human Tumor Necrosis Factor Receptor Probed by Raman and UVResonance Raman Spectroscopy: Structural Effects of an Engineered PEG Linker. Biochemistry 1995, 34, 15150–15156. Oleinikiv, V.; Kryukov, E.; Kovner, M.; Ermishov, M.; Tuzikov, A.; Shiyan, S.; Bovin, N.; Nablev, I. Sialylation Sensitive Bands in the Raman Spectra of Oligosaccharides and Glycoproteins. J. Mol. Struct. 1999, 4, 475–480. Stokes, D. L.; Vo-Dinh, T. Development of An Integrated Single-Fiber SERS Sensor. Sens. Actuators, B 2000, 69, 28–36. Scaffidi, J. P.; Gregas, M. K.; Seewaldt, V.; Vo-Dinh, T. SERSBased Plasmonic Nanobiosensing in Single Living Cells. Anal. Bional. Chem 2009, 393, 1135–1141. Guieu, V.; Garrigue, P.; Lagugne-Labarthet, F.; Servant, L; Sojic, N.; Talaga, D. Remote Surface Enhanced Raman Spectroscopy Imaging Via A Nanostructured Optical Fiber Bundle. Opt. Express 2009, 17, 24030–24035. Stoddart, P. R.; White, D. J. Optical Fiber SERS Sensors. Anal. Bional. Chem 2009, 394, 1761–1774.


Identification of Tumor Cells through Spectroscopic Profiling of the

Recommend Documents