Anal. Chem. 2008, 80, 9576–9582
Noninvasive Imaging of Protein Metabolic Labeling in Single Human Cells Using Stable Isotopes and Raman Microscopy Henk-Jan van Manen,† Aufried Lenferink, and Cees Otto* Biophysical Engineering Group, MESA+ Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands We have combined nonresonant Raman microspectroscopy and spectral imaging with stable isotope labeling by amino acids in cell culture (SILAC) to selectively detect the incorporation of deuterium-labeled phenylalanine, tyrosine, and methionine into proteins in intact, single HeLa cells. The C-D stretching vibrational bands in these amino acids are observed in the 2100-2300 cm-1 spectral region that is devoid of vibrational contributions from other, nondeuterated intracellular constituents. We found that incubation with deuterated amino acids for 8 h in cell culture already led to clearly detectable isotoperelated signals in Raman spectra of HeLa cells. As expected, the level of isotope incorporation into proteins increased with incubation time, reaching 55% for deuterated phenylalanine after 28 h. Raman spectral imaging of HeLa cells incubated with deuterium-labeled amino acids showed similar spatial distributions for both isotopelabeled and unlabeled proteins, as evidenced by Raman ratio imaging. The SILAC-Raman methodology presented here combines the strengths of stable isotopic labeling of cells with the nondestructive and quantitative nature of Raman chemical imaging and is likely to become a powerful tool in both cell biology applications and research on tissues or whole organisms. Stable isotopes and radioisotopes such as 2H (deuterium), 13C, N, 18O, 32P, and 35S are valuable tools in molecular and cell biology to selectively trace isotope-labeled proteins, nucleic acids, lipids, or other species among the large and diverse pool of unlabeled cellular biomolecules. Pulse-chase experiments using isotopes have been instrumental in unraveling metabolic pathways in cells and organisms. For example, the discovery of deuterium by Urey in 1932 immediately led to its use as a stable tracer in pioneering intermediary metabolism studies by Schoenheimer and Rittenberg.1 A major advantage of isotope tracers compared to exogenous labels such as fluorescent dyes is the close similarity in chemical properties between the isotope and its naturally more abundant counterpart. Although deleterious effects of isotope 15
* Corresponding author. Phone: +31 53 4893159. Fax: +31 53 4891105. E-mail:
[email protected]. † Present address: Akzo Nobel Chemicals BV, Research, Development and Innovation, Department of Chemicals Analytics and Physics, Molecular Spectroscopy Group, Velperweg 76, 6824 BM Arnhem, The Netherlands. (1) Schoenheimer, R.; Rittenberg, D. Science 1938, 87, 221–226.
9576
Analytical Chemistry, Vol. 80, No. 24, December 15, 2008
labeling on cells or organisms do exist, as evidenced by the fact that higher plants and even the simplest mammals, but not bacteria, resist full deuteration,2 it is nowadays commonly assumed and sometimes reported3,4 that the introduction of at least a certain level of isotopes does not interfere with the biochemical properties of cells or organisms. A relatively recent method for the introduction of stable isotopes into the cellular proteome is stable isotope labeling by amino acids in cell culture,5 SILAC, in which cells are incubated with 2H- and/or 13C-labeled amino acids in tissue culture for 5-6 cell doubling periods in order to fully replace amino acid residues in cellular proteins by their isotope-labeled counterparts. Labeled proteins are subsequently identified by mass spectrometry (MS) performed on the mixture of small peptides obtained by proteolytic digestion of cell lysates. The possibility for relative quantification of proteins by comparing stimulated versus unstimulated cells makes the SILAC-MS combination a powerful tool in cellular proteomics.6,7 The SILAC methodology can also be extended to mammalian systems, as exemplified by a very recent report describing the complete labeling of mice with 13C6-lysine.8 However, the destructive nature of proteolysis and MS has precluded the use of SILAC on intact cells or tissues. Since chemical bonds involving (stable) isotopes (e.g., C-D bonds) can be distinguished by vibrational spectroscopy from bonds lacking isotopes (e.g., C-H bonds), optical techniques such as infrared and Raman spectroscopy might be successfully used to detect isotope-labeled proteins in SILAC experiments. Here, we have combined SILAC with the only technique that is capable to nondestructively detect stable isotopes with submicrometer spatial resolution in single living cells, i.e., Raman microspectroscopy. In addition to quantitative detection of isotopelabeled phenylalanine, tyrosine, and methionine residues in single cells by Raman microspectroscopy, we will show that the spatial distribution of isotope-labeled proteins in cells can be visualized by Raman spectral imaging. To the best of our knowledge, this is the first report of isotope-labeled protein detection in single Katz, J. J.; Crespi, H. L. Science 1966, 151, 1187–1194. Ong, S.-E.; Foster, L. J.; Mann, M. Methods 2003, 29, 124–130. Amanchy, R.; Kalume, D. E.; Pandey, A. Sci. STKE 2005, pl2. Ong, S.-E.; Blagoev, B.; Kratchmarova, I.; Kristensen, D. B.; Steen, H.; Pandey, A.; Mann, M. Mol. Cell. Proteomics 2002, 1, 376–386. (6) Mann, M. Nat. Rev. Mol. Cell Biol. 2006, 7, 952–958. (7) Beynon, R. J.; Pratt, J. M. Mol. Cell. Proteomics 2005, 4, 857–872. (8) Kru ¨ ger, M.; Moser, M.; Ussar, S.; Thievessen, I.; Luber, C. A.; Forner, F.; Schmidt, S.; Zanivan, S.; Fa¨ssler, R.; Mann, M. Cell 2008, 134, 353–364. (2) (3) (4) (5)
10.1021/ac801841y CCC: $40.75 2008 American Chemical Society Published on Web 11/12/2008
eukaryotic cells by vibrational spectroscopy. Previous work involving isotope detection in cells has focused on isotope-labeled lipids visualized by spontaneous Raman microspectroscopy9,10 or coherent anti-Stokes Raman scattering microscopy.11,12 MATERIALS AND METHODS Details about materials are provided in the Supporting Information. SILAC Experiments with Phe-d5 and Tyr-d4. HeLa cells were adhered to CaF2 slides (Crystran Ltd., Poole, U.K.) and synchronized overnight in serum-deficient DMEM culture medium. The cells were subsequently incubated for 8 or 28 h with Phe-d5-containing SILAC medium or for 8 h with Tyr-d4-containing medium at 37 °C in a 5% CO2 atmosphere, washed 3 × with PBS, and fixed in 2% paraformaldehyde in PBS for 1 h at room temperature. For Raman spectroscopy and spectral imaging, the CaF2 slides containing the fixed cells were immersed in PBS. SILAC Experiments with Met-d3. HeLa cells were cultured for 6 days in Met-d3-containing SILAC medium at 37 °C in a 5% CO2 atmosphere and subsequently adhered to CaF2 slides. Cells were washed 3 × with PBS and fixed in 2% paraformaldehyde in PBS for 1 h at room temperature. For Raman spectroscopy, the CaF2 slides containing the fixed cells were immersed in PBS. Raman Microspectroscopy and Imaging. Confocal nonresonant Raman microspectroscopy and spectral imaging experiments were performed on a home-built laser-scanning Raman microspectrometer that was very similar to a previously described setup.13 In brief, the 647.1 nm excitation light from a Kr+ laser (Innova 90-K; Coherent Inc., Santa Clara, CA) was focused through a 63×/1.2 NA water-dipping objective (Zeiss Plan Neofluar; Carl Zeiss MicroImaging GmbH, Go¨ttingen, Germany) onto HeLa cells that had been grown on CaF2 slides. The scattered signal was collected by the same objective, passed through a dichroic beamsplitter (DCLP660; Chroma Technology, Rockingham, VT) and a RazorEdge 647 filter (Semrock Inc., Rochester, NY), focused onto a pinhole (diameter 15 µm) at the entrance of a custommade spectrograph and detected by a 1600 × 200 pixels backilluminated CCD camera (Newton DU-970N-BV; Andor Technology, Belfast, Northern Ireland). The spectrograph/CCD combination provides a spectral bandwidth of 3650 cm-1 (Raman shift from -50 to 3600 cm-1) and an average spectral resolution of 2.3 cm-1/ pixel. For Raman experiments involving the detection of deuteriumlabeled species, the wavenumber axis was calibrated on a daily basis using a 1:1 (v/v) mixture of toluene and toluene-d5. Toluene bands at 521, 785, 1004, 1210, 1604, 2738, 2866, 2921, 2981, and 3054 cm-1 and the toluene-d5 band at 2285 cm-1 were used for calibration.14 All spectra were corrected for the frequencydependent optical detection efficiency of the setup using a tungsten halogen light source (AvaLight-HAL; Avantes BV, Eerbeek, The Netherlands) with a known emission spectrum. Raman (9) Van Manen, H.-J.; Kraan, Y. M.; Roos, D.; Otto, C. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10159–10164. (10) Mattha¨us, C.; Kale, A.; Chernenko, T.; Torchilin, V.; Diem, M. Mol. Pharm. 2008, 5, 287–293. (11) Holtom, G. R.; Thrall, B. D.; Chin, B.-Y.; Wiley, H. S.; Colson, S. D. Traffic 2001, 2, 781–788. (12) Xie, X. S.; Yu, J.; Yang, W. Y. Science 2006, 312, 228–230. (13) Uzunbajakava, N.; Lenferink, A.; Kraan, Y.; Volokhina, E.; Vrensen, G.; Greve, J.; Otto, C. Biophys. J. 2003, 84, 3968–3981. (14) Hitchcock, A. P.; Laposa, J. D. J. Mol. Spectrosc. 1975, 54, 223–230.
spectral imaging experiments were performed by raster-scanning the laser beam over a cell or an intracellular region of interest and accumulating a full Raman spectrum at each pixel. Noise in the resulting 3-D (spatial × spatial × spectral dimension) data matrix was reduced by singular value decomposition.13 Univariate Raman images were constructed by plotting the integrated intensity (after background subtraction) of the vibrational band of interest as a function of position. Hierarchical cluster analysis (HCA) was performed on Raman data matrices to visualize regions in cells with high Raman spectral similarity.9 All data manipulations were performed in routines written in MATLAB 6.5 (The MathWorks Inc., Natick, MA). RESULTS AND DISCUSSION To detect the incorporation of stable isotope-labeled amino acids into cellular proteins in single cells, we first performed confocal nonresonant Raman microspectroscopy on HeLa cells that had been incubated with phenyl-deuterated phenylalanine (Phe-d5) for 8 or 28 h during cell culture and subsequently fixed with paraformaldehyde. We used fixed cells because these samples were also subjected to Raman spectral imaging (vide infra), which is usually performed on fixed cells because of the long acquisition time of ∼18 min/image on our setup. However, we expect that Raman microspectroscopy on live cells incubated with stable isotope-labeled amino acids will provide similar results as experiments on fixed cells. Phenylalanine was chosen as our first SILAC-Raman probe because the ring breathing vibrational mode of the phenyl group in Phe-h5 gives a sharp and strong band around 1000 cm-1 in cellular Raman spectra, so any spectral shift of this diagnostic band upon replacement of the phenyl protons by deuteriums would be easily detectable. Indeed, work by Overman and Thomas15 has shown that substitution of Phe-h5 for Phe-d5 in the bacterial virus fd leads to a shift in the phenyl ring breathing mode from 1002 to 960 cm-1. Replacement of 12C by 13 C in phenylalanine leads to similar spectral shifts in this Raman band.16 Moreover, C-D stretching vibrations in deuterated molecules are usually observed at 2100-2300 cm-1 in Raman spectra, a region that is devoid of contributions from nondeuterated cellular constituents. The absence of Raman background in this intermediate region allows the selective detection of the deuterated species, which we demonstrated before for cellular lipids by Raman microscopy9 and will be shown here for cellular proteins. Another advantage of using phenylalanine in SILAC experiments is that it is an essential amino acid, which ensures that cells only incorporate the added deuterated phenylalanine in their proteome. Figure 1A shows overlaid average Raman spectra (fingerprint region) recorded from the nucleus of HeLa cells incubated with either Phe-h5 or Phe-d5 for 8 and 28 h. The large similarity among these spectra, showing major bands at 725, 781, 825, 850, 1001, 1093, 1125, 1251, 1315, 1448, and 1655 cm-1 that coincide in frequency to reported bands for the HeLa cell nucleus,13,17 indicates that the incubation of cells with Phe-d5 instead of Pheh5 does not affect the overall intracellular chemical composition (15) Overman, S. A.; Thomas, G. J., Jr. Biochemistry 1995, 34, 5440–5451. (16) Huang, W. E.; Stoecker, K.; Griffiths, R.; Newbold, L.; Daims, H.; Whiteley, A. S.; Wagner, M. Environ. Microbiol. 2007, 9, 1878–1889. (17) Mattha¨us, C.; Chernenko, T.; Newmark, J. A.; Warner, C. M.; Diem, M. Biophys. J. 2007, 93, 668–673.
Analytical Chemistry, Vol. 80, No. 24, December 15, 2008
9577
Figure 2. Raman difference spectra of deuterated minus normal amino acids in solution and incorporated into cellular proteins in HeLa cells. (A) Phe-d5 minus Phe-h5. (B) Tyr-d4 minus Tyr-h4. Cells were incubated for 28 h with Phe-d5 in part A and for 8 h with Tyr-d4 in part B. Solution spectra were acquired for 250 s on 0.2 M solutions of amino acids in 1 M HCl. Cell spectra were acquired as described in the caption of Figure 1. Bands marked with asterisks in part B originate from protein, lipid, and nucleic acid vibrations.
Figure 1. Average Raman spectra recorded from the nucleus of 20 different HeLa cells incubated for 8 or 28 h with Phe-h5 or Phe-d5. (A) Fingerprint region (500-1800 cm-1). (B) 850-1150 cm-1 region. (C) 2100-2500 cm-1 region. Spectra were acquired for 60 s while scanning a nuclear region of 5 × 5 µm2 at a rate of 0.2 scan/s. The band at 2329 cm-1 in C is due to N2 in the beamline of the laser.
as reflected by the Raman spectra. Raman spectroscopy therefore corroborates previous statements3,4 about the lack of adverse effects of SILAC on cellular morphology, growth rates, or biological responses, in contrast to the reported deleterious effects of, e.g., 35S radioisotope labeling on cell cycle progression and cell proliferation and survival.18 As clearly evident from Figure 1B, the major difference among the spectra displayed in Figure 1A is the decrease in the 1001 cm-1 band of Phe-h5 and a concomitant increase in a new band at 959 cm-1, which is assigned to the (18) Wu, V. W.; Heikka, D. FASEB J. 2000, 14, 448–454.
9578
Analytical Chemistry, Vol. 80, No. 24, December 15, 2008
ring breathing vibrational mode of Phe-d5. These results were corroborated by principal component analysis (PCA) on the 600-1400 cm-1 spectral region of the Raman spectra, which revealed that changes in the 959 and 1001 cm-1 bands are a major source of variance among the spectra of cells incubated with Pheh5 or Phe-d5 for 8 and 28 h (Figure S-1, Supporting Information). Figure 1C shows that the C-D stretching modes of Phe-d5 appear at 2267 and 2292 cm-1 in cellular spectra. To better visualize differences between Raman spectra of cells incubated with normal or deuterated phenylalanine for 28 h, we recorded Phe-d5 minus Phe-h5 difference spectra of HeLa cells and compared them to difference spectra taken from phenylalanine solutions. As shown in Figure 2A, the solution and cellular difference spectra are strikingly similar, confirming that incubation of cells with Phe-d5 does not lead to spectral differences other than those originating from the deuterated phenyl ring. Whereas in the fingerprint region the peaks and troughs are found at very similar frequency in cells and in solution, the C-D stretching band in cells at 2292 cm-1 has downshifted by ∼7 cm-1 compared to Phe-d5 in solution (2299 cm-1). Similar downshifts are also observed for deuterated tyrosine (Figure 2B) and methyldeuterated methionine (vide infra) in cells vs solution. We
speculate that the apolar side chains of these amino acids are predominantly present in hydrophobic regions within proteins, causing the C-D stretching vibration frequency to be shifted compared to the hydrophilic environment in aqueous solution. In contrast, the frequency of the phenyl ring breathing vibration at 1001 cm-1 in Phe-h5 is known to be conformation-insensitive,19 and Figure 2A shows that this is also the case for the ring breathing band of Phe-d5 at 959 cm-1. In addition to the detection of deuterated phenylalanine incorporation into cells by Raman microspectroscopy, we also performed SILAC experiments using deuterated tyrosine. SILAC experiments using isotope-labeled tyrosine20,21 and other amino acids22,23 have been employed by others to shed light on (the dynamics of) tyrosine phosphorylation, which is a reversible protein post-translational modification that is crucial in cell signaling. Since it has been shown that tyrosine phosphorylation of small peptides leads to detectable spectral changes in Raman spectra,24 we envisage that Raman microspectroscopy on cells that have incorporated deuterated tyrosine in their proteome might offer opportunities for detecting tyrosine phosphorylation in single living cells without the need for immunostaining using antiphosphotyrosine antibodies. Figure 2B shows Tyr-d4 minus Tyr-h4 Raman difference spectra of HeLa cells incubated with Tyr for 8 h and Tyr in aqueous solution. The two difference spectra are again very similar, except the relative intensities of the two bands in the Tyr-h4 Fermi doublet (824 and 843 cm-1 in solution, 827 and 848 cm-1 in cells). In addition to tyrosine signals, there are several bands in the cell difference spectrum (Figure 2B, bottom trace) that originate from proteins or lipids. The fact that such bands appear in this spectrum and not in Figure 2A indicates that tyrosine spectral differences induced by incubating cells for 8 h with Tyr-d4 instead of Tyr-h4 are of similar order of magnitude as the biological variability in Raman spectra and much smaller than the Phe changes induced by incubating cells for 28 h with Phe-d5 instead of Phe-h5. As mentioned above, the C-D stretching vibration in Tyr-d5 has shifted to lower frequency (by ∼5 cm-1) in cell spectra compared to solution spectra. After incubating HeLa cells for 8 or 28 h with Phe-d5 and, with the aid of brightfield microscopy, recording Raman spectra in both the nucleus and cytoplasm of 20 cells, we quantified the incorporation of deuterated phenylalanine into proteins by calculating intensity ratios of the 959 cm-1 (Phe-d5) and 1001 cm-1 (Phe-h5) bands. Figure 3 shows that, as expected, the incorporation of Phed5 increases with incubation time and is very similar for the nucleus and cytoplasm of HeLa cells. The intensity ratios of 0.38 ± 0.06 (average of nucleus and cytoplasm) for 8 h and 1.20 ± 0.22 for 28 h correspond to incorporation levels of 28 ± 3% and 55 ± 4% after 8 and 28 h, respectively. The fact that more than half of the normal phenylalanine residues in proteins have been replaced by (19) Li, T. S.; Chen, Z. G.; Johnson, J. E.; Thomas, G. J., Jr. Biochemistry 1990, 29, 5018–5026. (20) Ibarrola, N.; Molina, H.; Iwahori, A.; Pandey, A. J. Biol. Chem. 2004, 279, 15805–15813. (21) Amanchy, R.; Kalume, D. E.; Iwahori, A.; Zhong, J.; Pandey, A. J. Proteome Res. 2005, 4, 1661–1671. (22) Blagoev, B.; Ong, S.-E.; Kratchmarova, I.; Mann, M. Nat. Biotechnol. 2004, 22, 1139–1145. (23) Kru ¨ ger, M.; Kratchmarova, I.; Blagoev, B.; Tseng, Y.-H.; Kahn, C. R.; Mann, M. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 2451–2456. (24) Xie, Y.; Zhang, D.; Jarori, G. K.; Davisson, V. J.; Ben-Amotz, D. Anal. Biochem. 2004, 332, 116–121.
Figure 3. Quantification of the incorporation of Phe-d5 in HeLa cells. The intensity ratio of Raman bands at 959 cm-1 (Phe-d5) and 1001 cm-1 (Phe-h5) was determined in the nucleus and cytoplasm of cells incubated for 8 h with Phe-h5 and cells incubated for 8 or 28 h with Phe-d5.
deuterated analogues after 28 h of incubation with Phe-d5 is consistent with the estimated HeLa cell doubling time of ∼24 h in tissue culture and an equal distribution of newly synthesized proteins over daughter cells. The ability to quantitatively determine isotope incorporation into proteins in single, intact cells is unique to Raman microspectroscopy and offers possibilities for investigating cell proliferation in physiological and pathophysiological processes such as cell differentiation and tumorigenesis. The similarity between the levels of isotope incorporation in nuclear and cytoplasmic proteins, as shown in Figure 3, suggests that newly synthesized isotope-containing proteins are homogeneously distributed over the cell. We tested this hypothesis by Raman spectral imaging of individual Phe-d5-containing cells at high spatial resolution (step size
