Probing Intrinsic and Extrinsic Components in Single Osteosarcoma

Apr 18, 2007 - Select a Journal or Book, Acc. Chem. Res. .... Lucas A. Lane , Ximei Qian , and Shuming Nie .... Bioconjugate Chemistry 0 (proofing), ...
0 downloads 0 Views 645KB Size
Download PDF
Anal. Chem. 2007, 79, 3646-3653

Probing Intrinsic and Extrinsic Components in Single Osteosarcoma Cells by Near-Infrared Surface-Enhanced Raman Scattering Hong-Wu Tang,*,† Xuebin B. Yang,‡ Jennifer Kirkham,‡ and D. Alastair Smith§

College of Chemistry and Molecular Science, Wuhan University, Wuhan 430072, P. R. China, and Department of Oral Biology and School of Physics and Astronomy, University of Leeds, Leeds LS2 9JT, UK

We report on the capabilities of near-infrared surfaceenhanced Raman scattering (SERS) using gold nanoparticles to obtain detailed chemical information with high spatial resolution from within single cancer cells, living or fixed. Colloidal gold particles, 60 nm in size, were introduced into live human osteosarcoma cells by endocytosis by adding them to the growth medium. Rapid SERS mapping of cells indicated that not only could rich vibrational spectra be obtained from intrinsic cellular constituents both in the cytoplasm and nucleus and but also the distribution of extrinsic molecules introduced into the cells, in this case, rhodamine 6G could be characterized, suggesting that the intracellular distribution of chemotherapeutic agents could potentially be measured by this technique. We show that the SERS signal intensity from the cellular components increases and more spectral detail is acquired from dried cells when compared with hydrated cells in buffer. The data also show spectral fluctuations, mainly in intensity but also in peak position, which are dependent upon the intensity of the excitation light and are probably due to diffusion of molecules on the surface of the gold nanoparticles. A detailed understanding of the origins of these effects is still not complete, but the ability to acquire very sensitive SERS inside living cancer cells indicates the potential of this technique as a useful tool in biomedicine. Raman spectroscopy can provide a high degree of information about the chemical structure of substances, which makes this technique potentially a very promising tool in biomedical analysis.1,2 Within the past decade, Raman spectroscopy has been successfully applied to living cells,3-12 which has opened up the * Corresponding author. Fax: 0086-27-68754067. E-mail: [email protected]. † Wuhan University. ‡ Department of Oral Biology, University of Leeds. § School of Physics and Astronomy, University of Leeds. (1) Spiro, T. G., Ed. Biological Applications of Raman Spectroscopy; John Wiley & Sons: New York, 1987. (2) Smith, E.; Dent, G. Modern Raman Spectroscopy-A Practical Approach; John Wiley & Sons: Chichester, 2005. (3) Puppels, G. J.; Mul, F. F. M. D.; Otto, C.; Greve, J.; RobertNicoud, M.; ArndtJovin, D. J.; Jovin, T. Nature 1990, 347, 301-303. (4) Otto, C. de Grauw, C. J.; Duindam, J. J.; Sijtsema, N. M.; Greve, J. J. Raman Spectrosc. 1997, 28, 143-150. (5) Sijtsema, N. M.; Otto, C.; Segers-Nolten, G. M. J.; Verhoeven, A. J.; Greve, J. Biophys. J. 1998, 74, 3250-3255.

3646 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

possibility of developing diagnostic tools, for example. However, Raman scattering has an extremely small cross section, typically ∼10-30-10-25 cm2/molecule, with the larger values occurring only under favorable resonance Raman conditions. Cellular components are present at very low absolute amounts, requiring that the excitation laser intensity be limited to avoid cell damage, and as a result, the Raman signal observed from a single cell is extremely weak. Consequently, the data collection times for Raman spectroscopy of a living cell using a confocal Raman system can be several minutes per point, leading to very long data acquisition times if whole cells are to be mapped. However, surface-enhanced Raman spectroscopy (SERS) makes it possible to enhance the scattering cross section by several orders of magnitude and has even enabled single-molecule detection.13-16 In SERS, Raman signals are enhanced when the target molecules are attached to metallic nanostructures and Raman scattering takes place in the high local optical fields of these structures.1,13-16 In addition, the metal clusters quench any fluorescence of closely associated molecules, which can help to reduce the background noise in these measurements.17,18 The possibility of large Raman signals, coupled with the fact that the spectral information in SERS only comes from the immediate vicinity of the nanostructures, has raised considerable interest in the possible use of SERS as a tool for targeted biochemical studies within living biological systems. In the near(6) Feofanov, A. V.; Grichine, A. I.; Shitova, L. A.; Karmakova, T. A.; Yakubovskaya, R. A.; Egret-Charlier, M.; Vigny, P. Biophys. J. 2000, 78, 499-512. (7) Sijtsema, N. M.; Tibbbe, A. G. J.; Segers-Nolten, J.; Verhoeven, A. J.; Weening, R. S.; Greve, J.; Otto, C. Biophys. J. 2000, 78, 2606-2613. (8) van Manen, H.-J.; Uzunbajakava, N.; van Bruggen, R.; Roos, D.; Otto. C. J. Am. Chem. Soc. 2003, 125, 12112-12113. (9) Uzunbajakava, N.; Lenferink, A.; Kraan, Y.; Volokhina, E.; Vrensen, G.; Greve, J.; Otto. C. Biophys. J. 2003, 84, 3968-3981. (10) Chan, J. W.; Taylor, D. S.; Zwerdling, T.; Lane, S. M.; Ihara, K.; Huser, T. Biophys. J. 2006, 90, 648-656. (11) Krafft, C.; Knetschke, T.; Siegner, A.; Funk, R. H. W.; Salzer, R. Vib. Spectrosc. 2003, 32, 75-83. (12) Krafft, C.; Knetschke, T.; Funk, R. H. W.; Salzer, R. Anal. Chem. 2006, 78, 4424-4429. (13) Nie, S.; Emory, S. R. Science 1997, 275, 1102-1106. (14) Kneipp, K.; Wang, Y.; Dasari, R. R.; Feld, M. S. Appl. Spectrosc. 1995, 49, 780-784. (15) Kneipp, K.; Kneipp, H.; Itzkan, I.; Dasari, R. R.; Feld, M. S. Chem. Rev. 1999, 99, 2957-2975. (16) Kneipp, K.; Kneipp, H.; Kneipp, J. Acc. Chem. Res. 2006, 39, 443-450. (17) Dulkeith, E.; Morteani, A. C.; Niedereichholz, T.; Klar, T. A.; Feldmann, J. Phys. Rev. Lett. 2002, 89, 203002. (18) Moskovits, M. Rev. Mod. Phys. 1985, 57, 783-826. 10.1021/ac062362g CCC: $37.00

© 2007 American Chemical Society Published on Web 04/18/2007

infrared (NIR) region, both silver and gold aggregates are suitable for SERS. Gold nanoparticles have been a more attractive choice for use with cells because of their favorable physical and chemical properties and biocompatibility.19,20 Kneipp et al.20 reported the first use of colloidal gold nanoparticles for the detection of biological molecules such as nucleic acids and proteins inside cells. Delivery of nanoparticles into the living cells, as well as routing of the particles or targeting of cellular compartments, can be achieved in various ways depending not only on the nature of the experiment but also on the type of cell line used and particle physicochemical properties such as size, shape, and surface functionalization.21-26 Much of the drive for this work has come from the desire to probe local biochemical properties in cells, but additionally, in recent years, a few studies have been made into identifying extrinsic components in single cells with SERS. For example, the distribution of different antitumor drugs such as doxorubicin and mitoxantrone and, their interaction with the cell nuclei and cytoplasm in living K562 cells, was studied by Manfait and co-workers.27,28 Rhodamine 6G (R6G), despite not being a chemotherapeutic drug, has the same pathway and intracellular destination as many anthracycline chemotherapeutic drugs such as doxorubicin and daunorubicin,29 and it is therefore an easily available model compound for such studies. Eliasson et al.30 succeeded in separating the rhodamine SERS signal from the intracellular components within single living lymphocytes. Kneipp et al.31 recently proposed a novel in vivo optical probe based on the SERS signal of the dye indocyanine green on gold nanoparticles, which has been shown to provide molecular structural information from the local biological environment around the particles. The potential of the technique as a diagnostic probe of biomolecular structure, distribution, and chemistry is intriguing, but there is still a great deal to be learned about in vivo SERS measurements. In this study, a NIR-SERS technique using gold nanoparticles is employed to detect intracellular components in a single highly malignant osteosarcoma cell and to identify R6G molecules that have been introduced into the cell. A comparison is made between the SERS spectra from live cells in culture medium and air-dried (19) Kneipp, K.; Kneipp, H.; Manoharan, R.; Hanlon, E.; Itzkan, I.; Dasari, R. R.; Feld, M. S. Appl. Spectrosc. 1998, 52, 1493-1497. (20) Kneipp, K.; Haka, A. S.; Kneipp, H.; Badizadegan, K.; Yoshizawa, N.; Boone, C.; Shafer-Peltier, K. E.; Motz, J. T.; Dasari, R. R.; Feld, M. S. Appl. Spectrosc. 2002, 56, 150-154. (21) Limbach, L. K.; Li, Y.; Grass, R. N.; Brunner, T. J.; Hintermann, M. A.; Muller, M.; Gunther, D.; Stark, W. J. Environ. Sci. Technol. 2005, 39, 9370-9376. (22) Rejman, J.; Oberle, V.; Zuhorn, I. S.; Hoekstra, D. Biochem. J. 2004, 377, 159-169. (23) Arlein, W. J.; Shearer, J. D.; Caldwell, M. D. Am. J. Physiol. 1998, 44, R1041R1048. (24) Tkachenko, A. G.; Xie, H.; Liu, Y. L.; Coleman, D.; Ryan, J.; Glomm, W. R.; Shipton, M. K.; Franzen, S.; Feldheim, D. L. Bioconjugate Chem. 2004, 15, 482-490. (25) Feldherr, C. M.; Kallenbach, E.; Schultz, N. J. Cell Biol. 1984, 99, 22162222. (26) Feldherr, C. M.; Akin, D.; Cohen, R. J. J. Cell Sci. 2001, 114, 4621-4627. (27) Beljebbar, A.; Sockalingum, G. D.; Morjani, H.; Manfait, M. Proce. SPIE 1999, 3608, 175-184. (28) Nabiev, I. R.; Morjani, H.; Manfait, M. Eur. Biophys. J. 1991, 19, 311-316. (29) Loetchutinat, C.; Saengkhae, C.; Marbeuf-Gueye, C.; Garnier-Suillerot, A. Eur. J. Biochem. 2003, 270, 476-485. (30) Eliasson, C.; Loren, A.; Engelbrektsson, J.; Josefson, M.; Abrahamsson, J.; Abrahamsson, K. Spectrochim. Acta, Part A 2005, 61, 755-760. (31) Kneipp, J.; Kneipp, H.; Rice, W. L.; Kneipp, K. Anal. Chem. 2005, 77, 23812385.

cells, and the complex behavior of SERS spectra as a function of time in this system is discussed. EXPERIMENTAL SECTION Cell Culture. Human osteosarcoma cells (G292 cell line) were seeded on sterilized calcium fluoride windows (Crystran) in Dulbecco’s modified Eagle’s medium, supplemented with 100 IU/ mL penicillin/streptomycin and 10% fetal calf serum and incubated at 37 °C and 5% CO2. The CaF2 windows were coated with 0.01% poly(L-lysine) (MW 150 000-300 000; Sigma-Aldrich) for cellular adhesion. A 200-µL colloidal gold solution (Aqueous solution contains gold as single colloidal particles ∼60 nm in diameter and HAuCl4 at 0.01% concentration; Agar.) and, when appropriate, rhodamine 6G (5 × 10-6 M, Sigma-Aldrich) were added to the cultivation chambers and incubated at 37 °C in 5% CO2 for 20 h. Immediately before the measurements, the culture medium was removed and the living cells were either placed in phosphatebuffered saline (PBS) at pH 7.4 for immediate Raman measurements or fixed in a 10% phosphate-buffered formalin solution (Sigma-Aldrich) for 10 min for later use. Addition of the colloidal gold particles did not appear to adversely affect cell viability. Cells grew normally after addition of the colloidal gold when compared with the controls, with no evidence of cell detachment from the substrate. Raman Measurements. Raman spectra were obtained using a Renishaw Raman microspectrometer (Wotton-under-Edge, Gloucestershire, UK) with a 785-nm diode laser at 3 mW with a 1-µm spot size. The objectives used were either a 50× (NA ) 0.55, long working distance) for dried cells or a 100× (numerical aperture, NA ) 1.20, water immersion) for living cells in fluid. Backscattered light was collected through the same objective onto a charge-coupled device detector. A collection time of 1 s yielded good-quality SERS spectra with a spectral resolution of ∼3 cm-1. When Raman mapping was performed, raster scans over single cells were carried out with a computer-controlled x, y stage in 1-µm steps at a laser spot size of ∼1 × 10-8 cm2. Excitation intensities in experiments of living cells were between 1.2 × 104 (0.12 mW laser output power) and 3 × 105 W/cm2 (3 mW laser output power). Comparisons of formalin fixed and dried cells used excitation intensities of up to 1.2 × 106 W/cm2 (12 mW laser output power). Confocal Fluorescence Microscopy. G292 osteosarcoma cells from the same cell line were incubated with gold colloids for 20 h as described above and then fixed in a 10% phosphatebuffered formalin solution for 10 min before permeabilized using 0.1% Triton X in PBS for 15 min at room temperature. The cells were incubated in 5 µM TO-PRO-3 (Molecular Probes, Invitrogen) for 10 min at room temperature to label cell nuclei and visualized under a Leica scanning confocal microscope (Leica TCS SPZ). The scanning confocal system was coupled to an epifluorescence microscope equipped with a 63× magnification water immersion objective of NA ) 0.90. Experiments were performed with a He/ Ne laser at 633 nm for nuclei fluorescence images and with an argon laser at 488 nm for colloidal particle backscattered light images. Each image comprised 512 × 512 pixels of 0.1 × 0.1 µm2 each. In order to locate gold colloid particles within the cells, merged images were produced using the fluorescence and scattered light data. Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3647

Figure 1. Merged confocal microscope image made up of a reflected light image that highlights the gold nanoparticles (green) within the cells and a fluorescence emission image that highlights the cell nuclei that have been stained with TO-PRO-3. The image corresponds to 512 × 512 pixels each of 0.1 × 0.1 µm2 size.

RESULTS AND DISCUSSION SERS Mapping of Cellular Components. Gold nanoparticles are known to enter living cells by endocytosis. This process involves one of three pathways:24 (i) cell entry followed by exocytosis; (ii) cell entry and transport up to the nuclear membrane; (iii) cell entry and nuclear translocation. In order to determine the distribution of nanoparticles in the human osteosarcoma cells used in this study, laser scanning confocal microscopy was used. Backscattered light images showed that the gold particles were often clustered together and distributed nonuniformly inside the cells. Figure 1 shows the particle distribution (in green) within a typical sample. In Figure 1, the backscattered light image is merged with a fluorescence image of the cells stained to highlight the nuclei (in blue). Figure 1 clearly shows that most of the clusters of nanoparticles are in the cytoplasm, and among this population, a significant proportion are found accumulated at the nuclear membrane. Only a small population of the nanoparticles have been translocated inside cell nuclei, which is perhaps not surprising given the size of the particles (60 nm) relative to the nuclear pore size. 3D analysis of confocal images (data not shown) demonstrated that the nanoparticles were definitely within the nucleus itself and not simply associated with the surface. Having established the typical distribution of the nanoparticles within the cells, it was possible to work with unstained, living cells for SERS measurements. A typical optical transmission image of a living osteosacoma cell incubated with gold colloids obtained using the Renishaw system using a 50× long working distance objective is shown in Figure 2A, and a backscattered light image at higher magnification (100×) more clearly shows the distribution of the gold nanoparticle clusters within the cell (Figure 2B). A SERS map of 41 × 51 pixels (2091 mapping points) with 1-s collection time/point was obtained. At each point in the image, a SERS spectrum is obtained and SERS images can be constructed 3648 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

using a Raman band of choice. For example, Figure 2C shows a SERS image of a cell constructed using the SERS intensity between 900 and 1000 cm-1, which indicates the presence of phenylalanine following subtraction of the background signal taken as the intensity at 1800 cm-1. The spectra from two specific points corresponding to the cytoplasm and nucleus, points 908 and 1049, respectively, are shown in Figure 2D. Although the collection time is only 1 s, the spectra have very good signal-to-noise ratio, suggesting the possibility of short collection times for a SERS map of this size, perhaps as fast as 30 min, during which a living cell is stable. Using unenhanced Raman spectroscopy, the mapping process may last a whole day during which time it is very difficult to keep the cells viable. As can be seen from Figure 2D, most of the Raman shifts, relative intensities, and line widths are in agreement with those features which appear in unenhanced Raman measurements.3 Table 1 shows the SERS band assignments for the spectra in Figure 2D. The spectrum at point 908 (cytoplasm) shows more contributions typical of proteins, whereas characteristic Raman bands of DNA are more clearly visible in the spectrum from point 1049. The Raman lines of 986, 1027, 1067, and 1106 cm-1 are assigned to vibrations of the DNA backbone, which confirms that some gold nanoparticles are able to enter the nucleus and form SERS-active clusters despite their size. Clearly, if this approach is to be a useful intracellular biomolecular probe, then one would expect to see a difference between the spectra acquired in the nucleus and cytoplasm, but how homogeneous are the spectra from within one region? Figure 3 shows the unprocessed SERS spectra collected from three adjacent points, 1048, 1049 and 1050, all of which are in the nucleus. Interestingly, the spectra from these three points differ significantly, but all comprise bands largely attributable to DNA. Like spectrum 1049, spectrum 1048 exhibits information about DNA as would be expected: the SERS line at 1066 cm-1 is evidence of the DNA backbone, the band at 654 cm-1 is assigned to guanine, the peak 800 cm-1 is related to deoxyribomononucleotides, the band at 1346 cm-1 is assigned to adenine, the bands at 1387 and 1519 cm-1 are assigned to adenine and thymine, and the band at 1146 cm-1 is related to deoxyribose phosphate. The other peaks that appear in spectrum 1048 are assigned to proteins (613 cm-1 to phenylalanine, 949 cm-1 to protein C-C skeletal, 1129 and 1237 cm-1 to amide III, and 1182 cm-1 to phenylalanine or tyrosine ring vibration). Spectrum 1050 only shows five obvious peaks, and except for the band at 1458 cm-1, which is assigned to C-H2 deformation of proteins, all the other bands are assigned to DNA: the 1240-cm-1 band and the strong band at 1392 cm-1 are due to vibration of adenine and thymine, and the 1496- and 1585-cm-1 peaks are assigned to adenine and guanine. The nonuniformity of nanoparticle size, shape, and distribution within the cell gives rise to the significant variations in the SERS spectra over the entire cell region although the probed volume is fixed (pL scale in this study). It is known that SERS takes place most strongly in the local fields of metallic nanostructures, and it can be a selective effect showing those molecules or molecular groups that are in the very close vicinity of the colloidal gold clusters. Moreover, the sizes of gold clusters play an important role in SERS. Among the colloidal gold particles of different sizes, 60-nm size particles show maximum enhancement factors, and it

Figure 2. Optical microscope image and Raman maps showing the gold nanoparticle distribution within a single living cell in PBS and typical SERS spectra at two different points within the cell. (A) Optical transmission image using a 50× long-working-distance objective. (B) Backscattered light image using a 100× water immersion objective. (C) SERS map at 990-1000 cm-1 with background (intensity at 1800 cm-1) subtraction showing the distribution of phenylalanine. (D) Unprocessed SERS spectra collected from points 908 and 1049 corresponding to points in the cytoplasm and nucleus, respectively.

increases by 11 orders of magnitude when aggregation of the nanoparticles occurs and when near-infrared excitation is applied.19 Spectrum 1048 is typical of data collected directly from the position of a gold nanoparticle cluster within a single cell with the surface-enhanced Raman lines appearing on a broad background. Kneipp et al.20 ascribe this to ‘‘surface-enhanced” fluorescence signal. In this study, with near-infrared, 785-nm excitation, the fluorescence from the single cell itself is certainly very weak and it might be quenched by the colloidal gold. Figure 3 indicates that the closer the distance between the measuring point and the aggregated gold nanoparticles, the stronger the background might be. In this sense, the background should emit directly from the gold clusters. The fluorescence background should mainly take place in enhanced local fields because it is closely dependent upon the colloidal gold distribution within the cell. Therefore, the lateral distribution of the broad fluorescence signal should monitor the distribution of the field enhancement. As shown in Figure 2C, the SERS map at 1800 cm-1 clearly shows the distribution of gold colloids and the field enhancement within

the cell, because no peaks appear around this position throughout our experiment. The distribution of detectable components can be shown more precisely when the signals are normalized to the average fluorescence signal in the measuring frequency range and mapping using peak area with baseline subtraction is applied. For example, Figure 2C shows the distribution of phenylalanine within the cell, demonstrating that phenylalanine distributes mainly in the cytoplasm. However, spectral information is dependent upon proximity to nanoparticles, and in the absence of homogeneity of nanoparticles within the cell, no firm conclusions in respect of the distribution of phenylalanine per se can be drawn. Comparison of SERS Spectra of Live Cells in Buffer and Fixed, Air-Dried Cells. In general, living cells are stable for only a few hours at room temperature when they are undergoing spectroscopic measurements. Therefore, fixed cells may be a more attractive option if only intracellular chemical information is needed rather than information about cell processes. However, the effects of fixatives on SERS spectroscopy of cells are largely Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3649

Table 1. SERS Peak Positions and Tentative Assignments of Major Vibrational Bands in the Spectra in Figure 2D mapping point

peak position (cm-1)

908

649 682 702 825 877 913 945 995 1131 1216 1278 1339 1374 1456 1517 1549 1585

1049

673 706 751 934 986 1027 1067 1106 1212 1241 1299 1385 1448 1498 1515 1564 1585

protein assignments CsC twisting of Tyr G unassigned Tyr Trp CsC stretch of Pro ring protein CsC skeletal symmetric ring breathing mode of Phe protein C-N stretch protein (amide III band) protein (amide III band) protein C-H deformation A, T Protein C-H2, C-H3 bending A Trp G, A G unassigned

T ring vib CsC stretching mode of proline and valine C-H in-plane bending mode of Phe

protein (amide III band) protein (amide III band) protein (amide III band)

DNA backbone: C-O stretching DNA backbone: C-O stretching DNA/RNA, PO2 symmetric stretching DNA backbone, PO2 symmetric stretching A, T A, C A, T

protein (CH2 deformation) G, A A, T Trp

Figure 3. Comparison of unprocessed 1-s SERS spectra collected from three adjacent 1-µm points 1048, 1049, and 1050.

unknown, and we therefore made an effort to compare the results of formalin-fixed, air-dried cells and those of live cells in buffer. Fixing tissues and cells in formalin is a routine procedure for specimen preservation for histopathological examination. Formalin is known to fix samples by crosslinking the amino groups at the proteins’ N-terminii or the NH3+ side chains. In this study, the effect of formalin fixation on SERS measurement is evaluated. No spectral contamination from formalin was observed in the spectra. This is clear from the absence of the very strong formalin 3650

nucleic acid assignments

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

G, A

bands at 1041 and 1492 cm-1(spectra not shown). Raman bands close to 1492 cm-1 sometimes appear in both living and fixed cells, but they should not be ascribed to formalin necessarily. For instance, the SERS peaks at ∼1496 cm-1 in Figure 3 may be assigned to adenine and guanine.32 In the present study, we found that the SERS spectra of the osteosarcoma cells change dramatically when dried in the air at room temperature. Figure 4 shows the SERS spectra from the same spot inside a single cell that has been incubated with gold colloidal solution. Figure 4A shows the spectrum of the cell while live in the growth medium, and Figure 4B shows the spectrum from the same point in the same cell after fixation and air-drying. There are two significant differences between the two traces: (a) the SERS intensity in the spectrum from the dried cell is ∼5 times greater than that of the cell in medium, and (b) there are more Raman bands in the spectrum of the dried cell. Figure 4A exhibits only ∼20 SERS peaks, whereas Figure 4B shows a spectrum with good signal-to-noise ratio in more than 30 bands. In (A), the band at 1067 cm-1 arises from the DNA backbone, and those bands around 1138 cm-1 are very likely due to a superposition of deoxyribose phosphate and the protein C-N stretching mode. The bands at 1349 and 1433 cm-1 may be assigned to guanine and adenine. All the other peaks in (A) may be ascribed to protein or amino acid vibrations. The marker band for phenylalanine is very clear at 996 cm-1, and those at 628 and 1538 cm-1 are due to (32) Ke, W. Z.; Zhou, D. F.; Wu, J. Z.; Ji, K. Appl. Spectrosc. 2005, 59, 418-423.

Figure 4. Comparison of SERS spectra obtained with a 50× long working distance objective from the same spot inside an osteosarcoma cell incubated with gold colloidal solution at 785-nm excitation when the cell is (A) living in PBS and (B) after it has been fixed and air-dried.

phenylalanine and tryptophan, respectively. The band at 1207 cm-1 could arise from tryptophan, phenylalanine, or both and that at 737 cm-1 also arises from tryptophan. The bands at 647 and 825 cm-1 are ascribed to the C-C twisting and out-of-plane ring breathing modes of tyrosine, respectively, and the strong band at 1459 cm-1 is due to the CH2 bending mode of proteins. The 908-cm-1 band is very likely due to C-C stretching mode of proteins (R-helix conformation). SERS lines at 1233 and 1294 cm-1 are ascribed to the protein amide III vibrations. Additional bands can be seen in (B). There are more SERS bands arising from amino acids such as phenylalanine (624, 997, 1021 cm-1), tyrosine (636, 846, 1615 cm-1), tryptophan (741, 875, 1211, 1569, 1615 cm-1), proline (846, 941 cm-1) and the nucleotides guanine (675, 1482, 1527 cm-1), adenine (1371, 1482, 1527), cytosine (656 cm-1), and thymidine (1371 cm-1). The lines at 1076, 1107, and 1165 cm-1 are due to strong DNA signals, and those at 1256 and 1284 are ascribed to the amide III band of proteins. The band at 1448 cm-1 is from protein CH2 bending. It is worth noting that the bands over 1600 cm-1 become clear only after the cell is dried. For example, the peak at 1615 cm-1 is due to the CdC stretching mode of either tryptophan or tyrosine. Moreover, the peak at 1669 cm-1 is ascribed to protein amide I, which is not exhibited in the SERS spectra of cells in medium. While the cell is drying, it shrinks and atomic force microscopy (data not shown) reveals that the volume of the dried cell is only one-third of that in medium. Therefore, the increase in the density of the cellular materials in the proximity of the metal nanoparticles is probably responsible for both the increase of the SERS intensity and the number of bands observed. The increase in the number of metal particles per unit volume will also be responsible for the increase in Raman signal intensity. Our results show that the SERS intensity may increase by a factor of 3-12 times in different parts of the cells. Identification of R6G Signals from a Single Osteosarcoma Cell. Of particular interest in this area is whether it is possible to use Raman spectroscopy to follow the uptake of extrinsic molecules, such as drugs, into a cell. A suitable model compound for such a study is the dye molecule R6G because it has very similar

structure to some cancer therapeutic agents and R6G has a strong SERS signal at the excitation wavelength used in this study. Figure 5 shows a comparison of the SERS spectrum of R6G in a gold colloidal solution (A) and the unprocessed SERS spectrum from a single living osteosarcoma cell that was incubated with R6G and the gold nanoparticles (B). In Figure 5A, vibrational information due to the xanthene ring deformation can be seen at 610 and 771 cm-1, as well as intense bands at about 1125, 1180, 1308, 1360, 1507, 1597, and 1647 cm-1, which arise from the totally symmetric modes of in-plane C-C stretching vibrations. A comparison with Figure 5B indicates that the SERS features of both intrinsic cellular components and the extrinsically applied R6G molecules that have been incorporated into the cell can be observed. Only when the R6G concentration in the incubation solutions is increased above 10-7 M can its signal be unequivocally separated from the signal of the intrinsic cellular components. However, cell death resulted when R6G concentrations higher than ∼10-5 were used, and there is therefore a narrow range of R6G concentrations that can be used in the incubation solution. The relative contributions of R6G and intrinsic cell components to the overall spectrum in Figure 5B depend on the coadsorption of both kinds of molecules. The capability to extract qualitative vibrational information on the intrinsic biomolecules is illustrated in trace C (Figure 5), which displays a spectrum indicative of the intrinsic cellular components obtained after subtraction of the R6G spectrum (spectrum A) from spectrum B. In trace B, SERS bands at around 1308, 1360, 1507, and 1647 cm-1 might originate from both R6G molecules and cellular components.33 However, trace C demonstrates that they should be assigned to the R6G. Interestingly, although a living or fixed cell in PBS exhibits an obvious R6G signal after incubation with R6G and gold nanoparticles, we have shown that the R6G signal disappears from the spectrum of the cell when it is fixed and then completely dried. For example, Figure 6 shows the SERS spectra from the same spot inside a cell incubated with the R6G-gold nanoparticles while the cell is air-drying at room temperature and then reimmersed in buffer after it is completely dried. Spectra A-C in Figure 6 show that the eight SERS bands of the R6G molecules that can be seen when the cell is hydrated in buffer reduce to only five bands after 5 min of air drying, and as shown in (C), after the cell is completely dried, no R6G bands can be detected. However, when the dried cell is put back in buffer, the R6G spectral information at the same spot recovers very quickly (D). As was discussed above, as the cell dries, the increasing density of the cellular material leads to an increase in the strength of the SERS signal from these components. A photochemical degradation of the R6G dye during the drying process is not the cause of the fall in SERS signal of R6G since it can be recovered easily upon rehydration of the cells. It would therefore appear that the cellular components preferentially bind to the metal nanoparticles during drying and the R6G SERS signal becomes overwhelmed and cannot be observed in the fixed and fully dried cells. Blinking of SERS Spectra from Inside a Cell. During the past decade, several researchers have observed fluctuations (blinking) in the scattering signals including on/off behavior when (33) Chowdhury, M. H.; Gant, V. A.; Trache, A.; Baldwin, A.; Meininger, G. A.; Cote, G. L. J. Biomed. Opt. 2006, 11, 024004.

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3651

Figure 5. Typical SERS spectra measured in a single living osteosarcoma cell incubated with both rhodaminee 6G and gold nanoparticles obtained using a 100× objective and 785-nm excitation. (A) The SERS spectrum of 10-7 M R6G in gold colloidal solution. (B) The unprocessed SERS spectrum from within the cytoplasm of a single living cell incubated with 5 × 10-6 M R6G solution and gold nanoparticles. (C) The difference spectrum B minus A displays only cell component Raman lines. In trace B, R6G bands are marked with a red asterisk, and contributions from intracellular components are marked with a green asterisk.

Figure 6. SERS spectra from the same spot inside a cell incubated with the R6G-gold nanoprobe at 785-nm excitation while the cell is air-drying at room temperature and rerinsed in medium. (A) The cell is in PBS. (B) The cell is air-dried for 5 min. (C) The cell is air-dried for 10 min. (D) The cell is rerinsed in PBS for 1 min. Only R6G bands are marked on all the traces.

probing single molecules and single nanoparticles by SERS.13,34-38 In this study, blinking of SERS spectra from nanoparticles inside single cells was also observed. In a series of SERS spectra taken (34) Otto, A. J. Raman Spectrosc. 2002, 33, 593-598. (35) Weiss, A.; Haran, G. H. J. Phys. Chem. B 2001, 105, 12348-12354. (36) Lukatsky, D. B.; Haran, G.; Safran, S. A. Phys. Rev. E 2003, 67, 062402. (37) Jiang, J.; Bosnick, K.; Maillard, M.; Brus, L. E. J. Phys. Chem. B 2003, 107, 9964-9972. (38) Andersen, P. C.; Jacobson, M. L.; Rowlen, K. L. J. Phys. Chem. B 2004, 108, 2148-2153.

3652

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

at 1-s intervals from a fixed point within a cell where gold nanoparticles are present, the intensity of the observed Raman bands varies considerably, some bands appearing and disappearing during the total measurement time. For example, the SERS peak at 904 cm-1 appears abruptly in spectrum 3, and the two peaks at 791 and 1088 cm- show quite correlated variation in intensity. Both of the peaks at 904 and 1088 cm-1 can be caused either by DNA or by protein (904 cm-1 could be ascribed to DNA backbone or protein C-C skeletal modes, and 1088 cm-1 is either the signal from O-P-O DNA backbone or caused by protein C-N stretching). The band at 791 cm-1 is ascribed to cytosine, and all the other bands in these spectra are protein SERS bands. The blinking behavior of the SERS signal was dependent on the excitation light intensity. The five spectra shown in Figure 7 were obtained with a laser power of 1.2 mW, and when the excitation power was increased from to 12 mW, the blinking behavior became more marked (data not shown). Movement of the cell or hydration effects can be ruled out as the cause of this blinking behavior because the cells were completely immobilized on the substrate and because the blinking effects were observed both in hydrated cells in buffer and in completely air-dried cells. In single-molecule detection studies using SERS in solution, such blinking behavior is readily observed and is characterized by a variable number of peaks, with seemingly random shifts and fluctuating relative intensities. In a recent study, Nie et al.39 suggested that the most probable mechanism that could result in the abrupt on/off SERS signal fluctuations was changes in the optical properties of the metal nanoparticles and suggested that (39) Emory, S. R.; Jensen, R. A.; Wenda, T.; Han, M.; Nie, S. Faraday Discuss. 2006, 132, 249-259.

in changing occupancy of active sites, which gives rise to the variations in intensity and small peak shifts.

Figure 7. Time-resolved surface-enhanced Raman spectra of cellular components within a single osteosarcoma cell in buffer recorded at 1-s intervals. The Raman signals abruptly change in intensity in all five spectra, and small spectral shifts are also observed. The laser excitation power was 1.2 mW.

the observed fluctuations are caused by thermally activated diffusion of individual molecules on the particle surface, coupled with photoinduced electron transfer and structural relaxation of surface active sites or atomic-scale roughness features. Almost all of the previous studies report blinking behavior of single molecules adsorbed on silver nanoparticles. Here we see that the SERS signal from gold nanoparticles exhibits similar fluctuations even when they are inside a cell with a large amount of cellular components adsorbed on them. Surface diffusion of adsorbed cellular molecules, enhanced by heating of the system by the laser excitation, appears the most plausible explanation for the signal fluctuations. Even though there will be a large number of molecules bound to the nanoparticle surface inside the cell, diffusion of molecules adsorbed to the particle surfaces will result

CONCLUSIONS In summary, we have demonstrated the potential of NIR-SERS using gold nanoparticles to probe the chemical composition inside single cancer cells and to identify and localize extrinsic molecules that could be therapeutic agents. Most of the gold nanoparticles that enter the living cell reside in the cytoplasm and around the nucleus, but a small number of gold nanoparticles appear to enter the cell nucleus, allowing a detailed SERS characterization of cellular components throughout. The study has shown that the SERS signals from fixed and dried cells provide much more intense and rich SERS spectra than those from hydrated cells (whether fixed or live) in buffer. This is most probably because of the reduction in cell volume upon drying and the consequent increase in cellular components that tightly interact with the metal particles. Interestingly, the SERS spectra of intracellular rhodamine 6G molecules do not show such an increase in signal intensity while the cell is dried. The reason for this is not clear, but it may be associated with preferential adsorption of intrinsic intracellular biomolecules to the metal nanoparticles. Our SERS data also show fluctuations (blinking) in the scattering signals from the cell, and this behavior is dependent on the excitation light intensity. The origin of these fluctuations, also observed in other SERS studies, is not entirely clear but may be due to the surface diffusion of molecules on the nanoparticles, which brings them in and out of contact with active sites on the metal surface. ACKNOWLEDGMENT The authors are grateful to Mrs. Jackie Hudson for technical assistance in use of the laser scanning confocal microscope. H.W.T. thanks the National Natural Science Foundation of China for financial support (Grant 20475041). Received for review December 14, 2006. Accepted March 12, 2007. AC062362G

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007

3653