Combination of Patch Clamp and Raman Spectroscopy for Single-Cell

Dec 8, 2010 - In this contribution we present the combination of patch clamp with Raman spectroscopy for a label-free quantitative detection of intrac...
0 downloads 0 Views 2MB Size
Anal. Chem. 2011, 83, 344–350

Combination of Patch Clamp and Raman Spectroscopy for Single-Cell Analysis Ute Neugebauer,† Stefan H. Heinemann,‡ Michael Schmitt,§ and Ju¨rgen Popp*,†,§ Institute of Photonic Technology, Albert-Einstein-Strasse 9, 07745 Jena, Germany, Center for Molecular Biomedicine, Department of Biophysics, Friedrich Schiller University of Jena and University Hospital Jena, Hans-Kno¨ll-Strasse 2, 07745 Jena, Germany, and Institute of Physical Chemistry, Friedrich Schiller University of Jena, Helmholtzweg 4, 07743 Jena, Germany In this contribution we present the combination of patch clamp with Raman spectroscopy for a label-free quantitative detection of intracellular components. Patch clamp is used to gain controlled access to the cytosol and internalize water-soluble compounds into the cell. The presence and concentration of these substances inside the living mammalian cell are probed by means of Raman spectroscopy in a label-free manner. A proof of principle was given using the carotinoid crocin as a sample compound that does not show specific interaction with the cell. When the intracellular crocin concentration as determined from the Raman spectra was monitored, the kinetics of internalization/diffusion into the cell could be characterized by a single-exponential function. Furthermore, the technique was successfully applied to observe differences in the internalization of free and protein-bound heme into the living cell. Although the peptide-capped microperoxidase MP-11 did not show specific interactions, free heme accumulated in the cell by binding to cellular components. Cells are the fundamental units of life with many vital physiological roles. A vast variety of molecules are involved in the proper functioning of cellular processes by interacting with each other. The detailed understanding of cellular processes on a molecular level is one of the key questions in biomedical research. Better insight into specific interactions of biomolecules with cellular components can help to treat or, even better, prevent disease. Often reporter or marker molecules, drugs, or therapeutics are introduced into a cell to probe the cell environment or to observe specific interactions with cellular constituents. The controlled internalization is by no means straightforward. Although endocytosis might suffer from biodegradation in the lysosome, or it might require modification of the substances of interest with a cell-penetrating peptide (e.g., from the TAT family),1 cell permeabilization in most cases will destroy the cell integrity. The whole-cell configuration of the patch-clamp technique proved to * To whom correspondence should be addressed. Phone: +49 (0) 3641 206300. Fax: +49 (0) 3641 206-399. E-mail: [email protected]. † Institute of Photonic Technology. ‡ Department of Biophysics, Friedrich Schiller University of Jena. § Institute of Physical Chemistry, Friedrich Schiller University of Jena. (1) Wender, P. A.; Mitchell, D. J.; Pattabiraman, K.; Pelkey, E. T.; Steinman, L.; Rothbard, J. B. Proc. Natl. Acad. Sci. U.S.A. 2000, 97 (24), 13003– 13008.

344

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

be an ideal tool for controlling the concentration of important molecules within the cytosol by diffusional exchange between the patch pipet solution and the cytosol of the cell.2 With the use of this approach, it is possible to deliver substances directly into the cytoplasm of the cell and detect them, e.g., by optical spectroscopy. Fluorescence imaging is commonly used to monitor the distribution of a fluorophor in the cell and its interaction with cellular components.2 However, many drugs and signaling molecules do not show strong autofluorescence and, therefore, cannot be investigated with fluorescence imaging. Raman spectroscopy as a label-free, nondestructive spectroscopic technique could be a powerful alternative. It provides highly specific and fingerprintlike chemical and structural information for both qualitative and quantitative analysis.3 The spatial distribution of biomolecules in living samples as well as their local environment can be probed via vibrational fingerprint pattern of the molecule.4,5 However, thus far the power of a combined patch-clamp/Raman approach to study cellular processes has not been exploited. Nair et al. reported about a combined patch-clamp/SERS (surface-enhanced Raman spectroscopy) approach on isolated artificial bilayers to quantify the permeability of rhodamine-6G across hemichannels.6 In this contribution we describe the novel combination of the patch-clamp technique with Raman spectroscopy for the investigation of living cells. This approach was first tested with crocin, a biological compound that does not interact with the cellular components. The quantitative loading of the cell with crocin, i.e., the loading kinetics, was studied. In a next step, the application was extended to heme, a molecule of exceptional biological relevance. Heme plays a critical role in many biological processes and was recently found to be also involved in cellular signal transduction.7,8 Degradation of heme results many catabolic byproducts, which exhibit physiological functions as well.9 However, excess intracellular heme is highly toxic to cells and its (2) Park, M. K.; Tepikin, A. V.; Petersen, O. H. Pfluegers Archiv. 2002, 444 (3), 305–316. (3) Schmitt, M.; Popp, J. J. Raman Spectrosc. 2006, 37 (1-3), 20–28. (4) Walter, A.; Erdmann, S.; Bocklitz, T.; Jung, E. M.; Vogler, N.; Akimov, D.; Dietzek, B.; Rosch, P.; Kothe, E. Analyst 2010, 135 (5), 908–917. (5) Frosch, T.; Konearevic, S.; Zedler, L.; Schmitt, M.; Schenzel, K.; Becker, K.; Popp, J. J. Phys. Chem. B 2007, 111 (37), 11047–11056. (6) Nair, C. M.; Sabna, C.; Murty, K.; Ramanan, S. V. Pramana 2005, 65 (4), 653–661. (7) Horrigan, F. T.; Heinemann, S. H.; Hoshi, T. J. Gen. Physiol. 2005, 126 (1), 7–21. (8) Tang, X. D.; Xu, R.; Reynolds, M. F.; Garcia, M. L.; Heinemann, S. H.; Hoshi, T. Nature 2003, 425 (6957), 531–535. 10.1021/ac1024667  2011 American Chemical Society Published on Web 12/08/2010

concentration has to be tightly kept at a low level.10 Analytical methods to detect and quantify heme inside living cells are still missing. Here, results on cell-target interactions by loading high concentrations of free and protein-bound heme into living mammalian cells by means of a combined patch-clamp/Raman approach are presented. METHODS Cell Culture. HEK 293 cells (CAMR, PortonDown, Salisbury, U.K.) were grown in Dulbecco’s modified Eagle’s medium mixed 1:1 with Ham’s F12 medium, 5 mM glucose, and supplemented with 10% fetal calf serum and 2.5 mM glutamine. Cells were maintained at 37 °C in a humidified atmosphere with 5% CO2. At around 80% confluence cells were trypsinized, diluted with culture medium, and plated onto 35 mm dishes with a coverslip on the bottom. Spectroscopic measurements and patch-clamp experiments were performed 1-3 days after plating. Buffers and Solutions. The patch pipettes were loaded with a buffer containing 35 mM sodium chloride (NaCl), 105 mM cesium fluoride (CsF), 10 mM glycol-bis(2-aminoethylether)N,N,N′,N′-tetraacetic acid (EGTA), 10 mM 4-(2-hydroxyethyl)-1piperazineethanesulfonic acid (HEPES) and adjusted to pH 8 with cesium hydroxide (CsOH). During the patch-clamp experiments the cells were kept in a bath solution containing 150 mM NaCl, 2 mM KCl, 1.5 mM CaCl2, 1 mM MgCl2, 10 mM HEPES (pH 7.4 with NaOH). Crocin (C44H64O24, FW 976.96 g/mol) was obtained from Sigma-Aldrich and dissolved in the pipet buffer to yield final concentrations between 0.025 and 1.7 mM. Hemin (C34H32ClFeN4O4, FW 651.96 g/mol) was obtained from Fluka and dissolved in the pipet buffer containing 30 mM NaOH to give a stock solution of 1 mM hemin. This stock solution was aliquoted and kept in the dark and cold (-80 °C). For each measurement a fresh aliquot was used and diluted with the pipet buffer to give a diluted stock concentration of 200 µM. For Raman measurement concentrations between 16 and 160 µM were used. Microperoxidase MP-11 (C84H116FeN20O21S2, FW 1861 g/mol) was obtained from Sigma, stored as 2 mM stock solution in water. Pipette buffer was used to prepare solutions for the Raman measurements (concentration range: 2.5-82 µM). Electrophysiological Measurements. Whole-cell voltageclamp experiments were performed as described previously.11,12 Patch pipettes of 1-3 MΩ resistance were fabricated from borosilicate glass (Kimble Glass, Vineland, NJ, U.S.A.), coated with dental wax to reduce the electrical capacitance, and fire-polished. A patch-clamp amplifier EPC9 was operated by PatchMaster software (both HEKA Elektronik, Lambrecht, Germany). All experiments were performed at 19-21 °C. The patch pipettes were loaded with pipet buffer containing various crocin concentrations in the range of 0.05-1.7 mM. Access resistance in the whole-cell configuration was measured with the auto-Cslow function of (9) Wagner, K. R.; Sharp, F. R.; Ardizzone, T. D.; Lu, A. G.; Clark, J. F. J. Cereb. Blood Flow Metab. 2003, 23 (6), 629–652. (10) Hou, S. W.; Reynolds, M. F.; Horrigan, F. T.; Heinemann, S. H.; Hoshi, T. Acc. Chem. Res. 2006, 39 (12), 918–924. (11) Chen, H.; Gordon, D.; Heinemann, S. H. Pfluegers Archiv. 2000, 429, 423– 432. (12) Hamill, O. P.; Marty, A.; Neher, E.; Sakmann, B.; Sigworth, F. Pfluegers Archiv. 1981, 391, 85–100.

Figure 1. Combination of patch clamp and Raman spectroscopy. Transmission image of a HEK 293 cell contacted by a patch-clamp pipet. (a) The asterisk indicates the position of the Raman laser focus. Schematics of the recording configuration. (b) A mammalian cell on a glass slide is in contact with a patch-clamp pipet, filled with analyte. In the whole-cell configuration the analyte can diffuse into the cell. The physical access to the cell interior is characterized by the electrical access resistance, Ra. The cell size can be estimated by measuring the total electrical membrane capacitance, Cm. Note that the schematic is not drawn to scale; the cell is only about 20 µm in diameter, and the pipet volume is about 40 µL, i.e., about 10 000 times the cell volume.

PatchMaster, and it ranged between 2.8 and 3.9 MΩ; the cell capacitancesa direct measure of the cell surface areasranged between 5 pF (HEK cells of about 10 µm diameter) and 20 pF (about 20 µm diameter). The holding potential was kept at -60 mV. Raman Measurements. Raman measurements were performed with a HR LabRam Raman spectrometer (Horiba JobinYvon) with focal length of 800 mm and equipped with a 300 lines/ mm grating. The spectrometer was coupled to an inverted microscope (Olympus IX71). The exciting Raman laser was focused to the cell on a spot away from the patch-clamp pipet tip using a 40× dry objective (Olympus, NA 0.75). The Raman scattered light was collected in 180° backscattering and recorded with a liquid nitrogen cooled CCD camera (1024 × 512 pixels). For the crocin experiments the Raman spectra were excited with 532 nm laser light (SUWTECH LDC-1500, Shanghai, China) giving 0.8 mW in the object plane. To selectively probe the hemederived vibrations a laser wavelength in resonance with the Soret absorption band of the porphyrin ring at 407 nm was used (krypton ion laser, Coherent). In order to minimize the application of energy the laser power was kept at 380 µW in the object plane. RESULTS AND DISCUSSION Proof of Principle: Cell Dialysis with Crocin. The experimental setup used to combine the advantages of the patch clamp with the noninvasive Raman spectroscopic detection is illustrated in Figure 1. Mammalian cells are grown on 170 µm glass slides. The patch pipet, filled with the substance of interest that should be loaded into the cell, approaches the cell from above to form a tight seal. A whole-cell configuration is established by destroying Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

345

Figure 2. Raman spectra of crocin at various concentrations measured in aqueous solution. The inset shows the chemical structure of crocin (a). Integrated Raman intensity vs crocin concentration using the CdC stretching vibrational band of crocin at 1540 cm-1 (b) and the C-C stretching band at 1167 cm-1 (c). The red lines are the results of linear fits.

the membrane patch underneath the pipet tip by brief application of negative pressure. Starting from the time of membrane rupture, the cell interior is dialyzed with the pipet solution. Since the pipet volume is much greater than the cell volume, it is expected that the analyte concentration inside the cell ultimately will reach the analyte concentration inside the patch pipetsproviding the absence of specific interactions of the analyte with immobile cellular components. By means of electronic control via a patch-clamp amplifier the electrical capacitance of the cell (Cm) and the access resistance (Ra), i.e., the electrical resistance between the pipet and cell interior, are measured continuously. The Raman laser is focused within the cell from below by means of an inverted microscope. In our experiments a fixed Raman measuring point well-separated from the area where the patch pipet made contact to the cell was chosen (as indicated in Figure 1a). As a proof of principle we aimed at quantitatively detecting crocin inside living HEK 293 cells via (resonance) Raman spectroscopy. Crocin was chosen as a sample compound because this carotinoid is biocompatible, should not interfere with the cell’s metabolism, and is not endogenous to mammalian cells, so no intrinsic background had to be taken into account. Because of its many hydroxyl groups (structure inset Figure 2a) it is watersoluble and therefore should distribute well inside the cytosol. Crocin shows a prominent π-π* transition (S0-S2 electronic transition) around 450 nm. The Raman excitation wavelength of 532 nm is just at the long-wavelength edge of that transition, allowing for utilizing the resonance Raman enhancing effect for increased Raman sensitivity. The corresponding resonance Raman spectra, depicted in Figure 2a for various concentrations, are dominated by the CdC stretching vibration around 1540 cm-1 and the C-C stretching vibration around 1167 cm-1, while the 346

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

C-Me rocking vibration shows up with minor intensity at 1214 cm-1. Other less intense bands can be found at 1286 and 1022 cm-1. Given constant experimental conditions such as laser power and integration time, the Raman intensity IR of those bands scales linearly with the crocin concentration c: IR ) γc

(1)

where the constant γ depends on the Raman scattering efficiency σ (which itself depends primarily on the molecular species, environment, and excitation wavelength), the interrogated volume V, and the instrumental throughput.13 The constant γ was experimentally determined by recording Raman spectra of aqueous crocin solutions of known concentration (Figure 2a) and determining the Raman intensity. Parts b and c of Figure 2 show the calibration plots of the two most intense crocin bandssthe CdC stretching band around 1540 cm-1 and the C-C stretching band around 1167 cm-1sindicating the linearity of these measurements. In a next step, the same crocin concentrations as used to construct the calibration plot in Figure 2 were backfilled into the patch pipet. The patch pipet was approached to the cell, and a tight seal (>1 GΩ) was formed. Before rupturing the cellular membrane underneath the patch pipet tip, a reference Raman spectrum of the undisturbed and unfilled cell was recorded. The establishment of a whole-cell configuration, i.e., providing access of the pipet analyte to the cellular cytosol, was taken as time zero. In intervals of 20-60 s Raman spectra were subsequently recorded (13) Shaver, J. M. Chemometrics for Raman Spectroscopy. In Handbook of Raman Spectroscopysfrom the Research Laboratory to the Process Line; Lewis, I. R., Edwards, H. G. M., Eds.; Marcel Dekker: New York, Basel, 2001.

Figure 3. Raman spectra of crocin recorded inside a cell in the whole-cell configuration (a). The indicated crocin concentrations were loaded into the patch pipet. Integrated Raman intensity vs nominal crocin concentration using the CdC stretching vibrational band of crocin at 1540 cm-1 (b) and the C-C stretching band at 1167 cm-1 (c). The error bars show the standard deviation from repeated measurements. The red lines are the linear fits obtained from the in vitro calibration plots shown in Figure 2. Increase of the crocin signal for the CdC stretching vibrational band of crocin at 1540 cm-1 (d) and the C-C stretching band at 1167 cm-1 (e) as a function of time after establishing the whole-cell configuration. The red curves are the exponential fits to the data starting about 300 s after patch rupture. Crocin concentration in the pipet was 0.807 mM, Cm ) 19.4 pF, Ra ) 2.8 MΩ.

from inside the cell, and the increase of the crocin signal was monitored. Depending on the cell size and the access resistance, the increase of the Raman intensity of the crocin bands leveled off after about 10-25 min. The cell was allowed to reach equilibrium for another 20 min before a last Raman spectrum was recorded. The integrated intensities of the crocin peaks (CdC and C-C stretch at 1540 and 1167 cm-1, respectively) were determined from the last Raman spectrum obtained from the cell loaded with crocin (Figure 3a). The corresponding concentration dependences are shown in Figure 3, parts b and c, with the calibration plot obtained from the in vitro measurements (Figure 2b,c). A very good agreement is noticeable between the in vitro and in vivo data. Raman measurements were repeated three times for each concentration in two independent series. Thus, the results illustrated in Figure 3 clearly show that substances loaded from a patch pipet into living cells can be quantitatively detected inside the cell by means of Raman spectroscopy. The time courses of the crocin Raman signals inside the cell are shown in Figure 3, parts d and e. As described by Pusch and Neher for the loading kinetics of various substances,14 our Ramandetected signals are well fit with a monoexponential function: IR(t) ) I0 + (I∞ - I0)(1 - exp(-t/τ) (14) Pusch, M.; Neher, E. Pfluegers Archiv. 1988, 411, 204–211.

(2)

where t is the time after establishing the whole-cell configuration, IR(t) is the Raman intensity at time t, I0 is the Raman intensity at the beginning of the experiment, I∞ is the intensity after achieving the equilibrium. The latter should reflect the data obtained from the in vitro calibration experiments, and τ is the time constant of cell loading. Right after patch rupture the analyte has to diffuse into the focus volume, and hence, there is a lag between patch rupture and exponential increase in crocin-mediated Raman signal. Therefore, for estimating the loading kinetics, the initial data points have been omitted in Figure 3, parts d and e. At a concentration of 0.807 mM, the maximally achieved Raman signals (I∞) were 712 ± 42 and 462 ± 36 for the CdC band at 1540 cm-1 and for the C-C band at 1176 cm-1, respectively. The I∞ values are within the experimental error in good agreement with the values expected from the in vitro calibration plot (see Figure 2) for the applied concentration of 0.807 mM (691 ± 24 and 400 ± 20, respectively). Using both bands and various crocin concentrations a mean loading time constant of 990 ± 180 s (mean ± SEM, n ) 5) was determined. The loading time constant depends besides the diffusion coefficient of the substance itself also on the access resistance Ra, the volume, and geometry of the cell. In order to facilitate comparison of the loading time constants from various experiments Pusch and Neher defined a normalized diffusion rate k, which is the ratio of the measured access resistance Ra Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

347

Figure 4. Chemical structures of MP-11 (a) and hemin (b). UV-vis absorption spectra of hemin and MP-11 (c). The vertical line marks the resonance Raman excitation wavelength of 407 nm. Resonance Raman spectra of hemin (bottom) and MP-11 (top) excited at 407 nm (d). Concentration dependence of integrated resonance Raman intensity for MP-11 (e) and hemin (f) with superimposed linear fits.

and the loading time constant obtained from the fit.14 For crocin diffusing into HEK 293 cells in the whole-cell patch-clamp configuration an averaged normalized diffusion rate of (0.44 ± 0.11) 104 Ωs-1 was measured (n ) 5). This value comes close to normalized diffusion rates of other substances with similar molecular weight; phallacidin with a molecular mass of 1070 g/mol k was (1.97 ± 1.50) 104 Ωs-1,14 crocin has a molecular mass of 977 g/mol. The apparently slower loading kinetics observed in Raman experiments is most likely due to a strong bleaching of crocin as laser powers had to be considerably greater than used for, e.g., fluorescence measurements. Application: Comparison of Free and Peptide-Bound Hemin. Free heme was found to be of physiological and pathophysiological interest in cells. Given the activation of some transcription factors by about 100 nM heme, a free concentration in that range is expected, but thus far quantitative assessments are lacking. Therefore, we started to examine how heme can be detected inside cells using Raman spectroscopy. With the chosen experimental setup (407 nm excitation wavelength, 380 µW at the sample, inverted microscope, 40× air objective) the convenient detectable hemin concentration are in the range of a few micromolar (down to 5 µM) when using short accumulation times (1-10 s). For that purpose we loaded unphysiological high hemin (oxidized form of heme) concentrations into HEK 293 cells via the patch-clamp pipet. In order to assess how hemin might interact with cellular components, we compared these results with measurements using peptide-bound hemin (microperoxidase MP11). Parts a and b of Figure 4 show the chemical structure of the hemin peptide MP-11 and hemin, respectively. The protoporphyrin ring is identical in both compounds. While hemin has two vinyl groups on adjacent pyrrol rings, these groups are used in MP-11 to covalently bind two cysteine residues via a thioether linkage. 348

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

The two cysteines are part of a peptide fragment corresponding to the 11 amino acids 11-21 found in cytochrome c from equine heart (Val-Gln-Lys-Cys-Ala-Gln-Cys-His-Thr-Val-Glu). The imidazole side chain of His18 is acting as a fifth axial ligand for the central iron ion, whereas in free hemin this position is occupied by a chloride ion. The UV-vis absorption spectra of both free hemin and the peptide-bound hemin (Figure 4c) show a strong Soret absorption band around 400 nm which is due to the π-π* transition (S0 f S2) of the delocalized π-system of the porphyrin ring system. The broadened peak and lower extinction coefficient of free hemin as compared to the sharp Soret band in MP-11 can be explained with the formation of hemin dimers in slightly basic aqueous solutions.15,16 A second absorption band of lower intensity can be found around 530 nm in MP-11 and at 608 nm in hemin. These bands arise from another π-π* transition (S0 f S1) and are called Q-bands. In order to utilize resonance Raman enhancement a Raman excitation wavelength of 407 nm in resonance with the Soret band was chosen. The corresponding resonance Raman spectra are shown in Figure 4d displaying in particular totally symmetric vibrations due to a classical A-type enhancement mechanism. These enhanced vibrations are the A1g symmetric stretching bands ν10, ν2, ν3, and ν4, which can be found at 1631, 1575, 1492, and 1374 cm-1, respectively. The most prominent band at 1374 cm-1 (ν4, ν(pyrrol half-ring)) is indicative for iron in an oxidation state of +3, as is expected in aqueous solutions for both hemin moieties. The free hemin exists as a high-spin complex, which can be seen at the wavenumber values of the ν(CRCm), ν(CβCβ), and ν(CRCm)symm (15) Kuzelova, K.; Mrhalova, M.; Hrkal, Z. Biochim. Biophys. Acta 1997, 1336 (3), 497–501. (16) Brown, S. B.; Dean, T. C.; Jones, P. Biochem. J. 1970, 117 (4), 733–739.

Figure 5. Time course of ν4 vibration intensity inside HEK 293 cells while loading them with 20 µM MP-11 (a) and 66 µM hemin (c). Integration time was 1 s in both cases. Cm ) 22.9 pF, Ra ) 2.0 MΩ for MP-11 (a) and Cm ) 10.2 pF, Ra ) 4.2 MΩ for hemin (c). Concentration dependence of the integrated Raman intensity using ν4 of the porphyrin ring at 1374 cm-1 for MP-11 (b) and for hemin (d). The asterisks denote experimental in vivo values; the red lines are the linear fits derived from the in vitro calibration (Figure 4). Superposition of resonance Raman spectra for hemin (66 µM, green) and MP-11 (20 µM, brown) under in vitro conditions and 50 min after loading into an HEK 293 cell (in vivo) (e). Excitation at 407 nm, 10 s integration time.

vibrations showing up at 1631, 1574, and 1492 cm-1, respectively.17 Instead of a chloride the fifth ligand in MP-11 is the strong-field ligand imidazole which causes the central iron ion to change in the low-spin state which manifests itself by the wavenumber values of the ν(CRCm), ν(CβCβ), and ν(CRCm)symm vibrations which are now found at 1642 (overlaid with the water band), 1591, and 1506 cm-1.18 As can be seen in Figure 4d also the relative intensity patterns vary between free hemin and the protein-bound hemin. Whereas in hemin the three vibrations at 1574, 1492, and 1374 cm-1 are of comparable intensity, the vibration at 1374 cm-1 clearly dominates the resonance Raman spectrum of MP-11, where the band at 1506 cm-1 is of relatively low intensity. The overall Raman scattering cross section is larger for the low-spin complex MP-11, because of structural changes due to the different spin state affecting the Soret band position.19 While the Raman excitation wavelength at 407 nm is in resonance with the Soret band maximum in MP-11, it hits only the long-wavelength trailing edge of the hemin Soret absorption band. For both hemin compounds an in vitro concentration series was recorded. The integrated Raman intensities of the ν4 vibrational band centered at 1374 cm-1 were used to construct the linear concentration calibration curves shown in Figure 4, parts e and f, respectively. (17) Wood, B. R.; Langford, S. J.; Cooke, B. M.; Lim, J.; Glenister, F. K.; Duriska, M.; Unthank, J. K.; McNaughton, D. J. Am. Chem. Soc. 2004, 126 (30), 9233–9239. (18) Spiro, T. G.; Burke, J. M. J. Am. Chem. Soc. 1976, 98 (18), 5482–5489. (19) Spiro, T. G.; Strekas, T. C. J. Am. Chem. Soc. 1974, 96 (2), 338–345.

A representative concentration of each hemin compound was used to load the cells via the patch-clamp whole-cell configuration, and the Raman signals from inside the cells were recorded as described previously for the crocin experiments. MP-11 showed similar loading kinetics as crocin (Figure 5a): after some time (5-10 min) of no or only minor increase, the MP-11 Raman bands started to increase monoexponentially and leveling off after 30-50 min, depending on the size of the cell. The final concentration inside the cell corresponded to the concentration that was used to fill the patch pipet. This can be seen in Figure 5b, where the integrated Raman intensity of the ν4 vibration (around 1374 cm-1) from inside the cell lies exactly on the calibration plot constructed from aqueous solution data. A different behavior is observed for hemin. After the lag time in the whole-cell configuration the hemin Raman bands increased indicating loading of the cell. However, even after 50 min no saturation of the signal was observed (Figure 5c). At that time the integrated Raman intensity of the ν4 band of hemin inside the cell was already much higher than expected from the applied hemin concentration in the patch pipet. (Figure 5d). Longer measuring times in order to reach a constant hemin signal inside the cell were not feasible because the tight seal was lost together with dramatic drop of the access conductivity. The high, accumulated hemin concentrations inside the cell are most likely explained by hemin binding to cellular components and therefore, by clearing free hemin from the cytosol. Binding of free hemin to cellular components, mainly membrane compoAnalytical Chemistry, Vol. 83, No. 1, January 1, 2011

349

nents, has already been reported in the literature.10,20,21 These hemin-cell interactions might also be responsible for the loss of the whole-cell configuration after long measuring episodes. The power of Raman spectroscopy can be utilized to assign molecular changes in the hemin environment occurring while it is binding to cellular components. While the resonance Raman spectrum of MP-11 does not change upon internalization into the cell, clear changes are visible in the hemin resonance Raman spectrum while loading into the cell (Figure 5e). The intensity ratio of the Raman bands resembles now the ones of proteinbound, low-spin hemin and not anymore that of free, high-spin hemin. The resonance Raman spectrum is now dominated by the pyrrol half-ring stretch at 1374 cm-1. The spin-state marker bands ν2 and ν3 at 1585 and 1495 cm-1 are shifted to slightly higher wavenumber, without reaching the wavenumber values representing the low-spin state of the central iron. This is probably because there is still a significant amount of free, high-spin iron in the cytosol, which is fed from the patch pipet. CONCLUSIONS We successfully demonstrated the combination of the patchclamp technique and Raman spectroscopy to study cellular processes in living cells. It was shown that substances, even if they cannot easily penetrate the cellular membrane, can be loaded into the cell in a quantitatively defined manner and detected based on molecular vibrations by means of Raman spectroscopy. Intensity ratios and position of the vibrational bands give further valuable insights into the interactions of the internalized substance. The combination of patch clamp with Raman was shown exemplarily in a proof-of-principle experiment for the carotinoid crocin. The increase of this internalized substance with time in(20) Li, S. D.; Su, Y. D.; Li, M.; Zou, C. G. Acta Biochim. Biophys. Sin. 2006, 38 (1), 63–69. (21) Shviro, Y.; Shaklai, N. Biochem. Pharmacol. 1987, 36 (22), 3801–3807.

350

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

side the cell up to the final concentration could be followed by recording Raman spectra. A first real application example was shown internalizing, detecting, and quantifying free hemin and a peptide-bound hemin derivative inside living cells in a concentration rage of a few micromolar. As Raman spectroscopy uses the molecular vibration of the molecule of interest as intrinsic marker, no additional labels, such as fluorescence tags, are necessary. By analyzing the wavenumber positions and relative intensities of the Raman bands it could be shown that free hemin binds to cellular components, whereas peptide-bound hemin does not show specific interactions. The combined patch-clamp resonance Raman spectroscopy technique can be expanded to many other substances enabling new insights in the label-free, living single-cell analysis. Possible further examples include drug-target interaction or the effect of special metabolites. In this experiment Raman spectra were only recorded from one spot within the cell. However, implementation of a moveable xy-stage will allow spatial Raman mapping and, therefore, easy localization of substances of interest. Furthermore, extension to nonlinear Raman methods, such as coherent anti-Stokes Raman scattering (CARS) microscopy allowing one to monitor online processes, is conceivable. ACKNOWLEDGMENT Financial support by TMWFK PE114-1 is acknowledged. Thanks are given to Guido Gessner and Nirakar Sahoo for fruitful discussions and Angela Rossner for valuable help in the cell culture lab.

Received for review September 17, 2010. Accepted November 17, 2010. AC1024667