Anal. Chem. 1995, 67,2575-2579
Neurotransmitter Imaging in Living Cells Based on Native Fluorescence Detection Weihong Tan,t Vladimir Parpura,#Philip 0. Haydon,* and Edward S. Yetang**+
Ames Laboratoty- USDOE and Department of Chemistry and Department of Zoology and Genetics, Laboratory of Cellular Signaling, Iowa State University, Ames, Iowa 5001 1
A UV laser-based optical microscope and CCD detection system with high sensitivity has been developed to image neurotransmitters in living cells. We demonstrate the detection of serotonin that has been taken up into individual living glial cells (astrocytes) based on its native fluorescence. We found that the fluorescence intensity of astrocytes increased by up to 10 times after serotonin uptake. The temporal resolutionof this detection system M serotonin is as fast as 50 ms, and the spatial at resolution is dilli-actionlimited. This UV laser microscope imaging system shows promise for studies of spatialtemporal dynamics of neurotransmitter levels in living neurons and glia. The in vivo monitoring of intracellular molecules is of great biological importance. To better understand developmental biology, cellular differentiation,physiology, and cell biology, a variety of real-time analytical methodologies have been developed.'-1° These approaches have been applied to analyze a large variety of intracellular species ranging from small ions, such as calcium, to large protein molecules. One particularly interesting and widely studied class of molecules is the neurotransmitters, which are responsible for intercellular communication between neurons.5 As in many areas of biology, advances in neuroscience are often directly l i k e d to the development of new techniques for chemical analysis. The release of neurotransmitter has been studied by electrochemical methods, i.e., microelectrodes2 and electrochemical detection coupled with highly efficient separation techniques such as liquid chromatography (HPLC) and capillary electrophoresis (CE) .3 Even though analysis at the single-cell level and subattomole detection limits have been achieved, several problems exist. One of them is the difficulty in achieving simultaneous spatial and temporal Ames Laboratory-USDOE and Department of Chemistry. Department of Zoology and Genetics. (1) Herman, B.; Jacobson, K Optical Microscopy for Biology; Wiley Liss: New York, 1989. (2)Garris, P. A; Collins, L. B.; Jones, S. R.; Wightman, R. M. 1.Neurochem. 1993,61,637-647. (3) Ewing, A; Mesaros, J. M.; Gavin, P. F. Anal. Chem. 1994,66,527A-537A (4) Yeung, E. S. Ace. Chem. Res. 1994,27, 409-414. (5) Parpura, V.; Basarsky, T. A; Liu, F.; Jeftinija, K; Jeftinija, S.; Haydon, P. G. Nature 1994,369,744-747. (6) Tan, W.; Shi, Z.-Y.; Smith, S.; Bimbaum, D.; Kopelman, R Science 1992, +
258, 778-781. (7) Barber, A J. &p. Biol. 1986,121,395-406. (8) Wang, X. F.; Periasamy, A; Herman, B.; Coleman, D. M. Crit. Rev. Anal. Chem. 1992,23, 369-395. (9) Malgaroli, A; Tsien, R W. Nature 1993,357, 134-139. (10)Lakowicz, J. R; Szmacinski, H.; Nowaczyk, K; Johnson, M. L. Cell Calcium 1992,13,131-147.
0003-2700/95/0367-2575$9.00/0 0 1995 American Chemical Society
resolution. Another is the inability to monitor multiple cells or cell compartments in a single experiment. Other inherent problems associated with electrochemical detection schemes are electrical interference from stimulation and the need for reference electrodes. In addition, after the targeted neurotransmitter is released from a specifk location, in order to be detected it has to diffuse toward the microelectrode sitting a certain distance away." This not only limits the temporal resolution but may also lead to uncertainties regarding the position of the release sites. For example, two closely spaced release sites may interfere with each other. In contrast, a fast imaging technique can detect the release of the neurotransmitter at all locations as it occurs. Native fluorescence monitoring provides unique advantages for ultratrace biochemical analysis. There is no need for sample handling or chemical derivatization, and thus no contamination or additional background will be involved. The biological entity will also be undisturbed by an additional reagent. Although sensitive fluorescence detection of biomolecules is typically performed after derivatization with the appropriate fluorogenic reagent, it is known that these analytes are themselves fluorescent. The quantum yields are low and normally would not have been considered to be useful for sensitive detection. However, laserexcited fluorescence provides such low limits of detection COD) l2 that the weak native fluorescence from biomolecules can still be useful. In the largeframe argon ion lasers, one can obtain W output ranging from 275 to 360 nm. The sensitive detection of proteins based on native fluorescence has been successfully d e m ~ n s t r a t e d ~in~ Jcombination ~ with CE. For example, tryp tophan in the 1 nM rangeI3 and typical proteins in the 0.1 nM rangeI4have been detected. Hemoglobin and carbonic anhydrase in individual human erythrocytes have also been determined.15 There are also strong laser lines which are close matches to the absorption maxima of some neurotransmitters. Recently, we demonstrated that native fluorescence detection combined with CE separation at low pH provides high sensitivity (down to nanomolar concentration), high resolution, high speed, and low interference for the analysis of catecholamines.'6 This method has been applied successfullyfor the measurement of the amounts of epinephrine and norepinephrine in individual bovine adrenal medullary cells. Therefore, successful implementation of native fluorescence detection in optical microscopy should have good potential in in vivo studies of neurotransmitters and their release. (11) Schroeder, T. J.; Jankowski, J. A; Kawagoe, K. T.; Wightman, R. M.; Lefrou, C.; Amatore, C. Anal. Chem. 1992,64,3077-3083. (12) Yeung, E. S. Adu. Chromatogr. 1995,35,1-51. (13) Lee, T. T.; Yeung, E. S. 1.Chromatogr, 1992,595,319-325. (14) Lee, T. T.; Lillard, S. J.; Yeung, E. S. Electrophoresis 1993,14,429-438. (15) Lee, T. T.; Yeung, E. S. Anal. Chem. 1992,64,3045-3051. (16) Chang, H. T.; Yeung, E. S. Anal. Chem. 1995,67,1079-1083.
Analytical Chemistry, Vol. 67, No. 15, August 1, 1995 2575
CCD Camera
Materials. All chemicals and materials except those for cell culture were purchased from Sigma Chemical Co. (St. Louis, MO) and used without further puriiication. All buffers were prepared locally Serotonin Detection. A 0.1 mM serotonin solution was used to obtain absorption, excitation, and fluorescence spectra in commercial spectrometers. When studied with the CCD-based W microscope system, serotonin standard solutions were introduced into the detection region of the microscope by micropipets and capillary tubes through homemade narrow channels under hydraulic pressure. The liquid was imaged, and its intensity was compared with a blank buffer solution. Cell Culture. Enriched astrocyte type-1 cultures were prepared from 1to 4-day-old Sprague-Dawleyrat cortices as previously described.5,18Briefly, cortices were freshly dissected, and tissue was enzymatically (papain 20 iu/mL; 1h at 37 "C) and mechanically dissociated. Cells were plated into culture flasks and maintained at 37 "C in a humidified 5%C02-95% air atmosphere. After 7-13 days in culture, cells were shaken twice, first for 1.5 h and then for 18 h on an orbital shaker at 260 rpm. Adherent cells was subsequently detached using trypsin (O.l%),spun at l00g for 10 min, resuspended, and plated onto poly(L-lysine) (1 mg/ mL; MW 100 000)-coated glass cover slips. Cell Imaging and Analysis. Cultured astrocytes5,1sat room temperature were used to test the system. We confirmed that these cells remain viable for hours at 25 "C. Before the astrocytes were put on the microscope stage, the 305nm laser beam was directed toward the center of microscope field of view and focused onto a quartz plate. The laser beam was then blocked. The astrocytes were placed on the microscope stage. The standard illumination system of the microscope was used to align the cells with the objective. Using the same light source, the cells were focused and a CCD image was taken of the area of interest for each dish of cells. The sample dish was then maintained in a fixed position throughout the experiment. Great caution was taken in limiting the exposure of the cells to UV light since excessive exposure can damage living systems. This is confirmed by visual inspection of the morphological features of the cells (granules, shapes, edges of the membrane) as a function of exposure to the laser beam. Serotonin Uptake Experiments. Astrocytes were first imaged for a duration of 0.5 s at 305 nm (about 6 mw) to record the autofluorescence of the cells. Then, a 1 mM serotonin solution was injected into the culture media for time periods of 2-20 min. Subsequently,fresh culture medium was flushed into the chamber to remove serotonin in the bath. A second CCD image and another optical image were then recorded.
yMo A r Ionlaser I
w I
v
Figure 1. Schematic diagram of the CCD-based microscope system for neurotransmitter monitoring by UV-excited native fluorescence: F, optical filters: L, lens: M, mirror; MS, microscope sample stage; MO, microscope body; 0, UV objective; S, sample dish.
EXPERIMENTAL SECTION
Experimental Setup. Our microscope-based system is similar to those used for intracellular studies of biological cell^.^*^**-^^ Figure 1 shows a schematic of the detection setup used in our experiments. A scientific grade CCD camera system (Photometr i c ~Ltd., Tucson, AZ) was connected to a fluorescence microscope (Axioskop, Carl Zeiss, Germany). The CCD camera was mounted on the top entry port of the microscope, facing downward. A 305 nm laser beam from an Art laser (Model 2045, Spectra-Physics, Mountain View, CA) was isolated from the other lines with an external prism. It was focused and directed by mirrors and lenses to the microscope sample stage for excitation. Specifically, a l c m focal length quartz lens (Melles Griot, Irvine, CA) was used underneath the sample to focus the laser beam into the imaged region. Another lens ( 7 k m focal length) was used to collimate the beam prior to focusing. Optical alignment for this backillumination geometry is simpler compared to side-illumination or epi-illumination geometries. The native fluorescence was collected via quartz microscope objectives (Zeiss) with either l o x (NA 0.20) or 40x (NA 0.60) magnification. These objectives transmit in the UV, especially in the range of 300-400 nm, and have very limited autofluorescence. In front of the CCD camera, a combination of band-pass filters was used to selectively collect the desired signal. The CCD system is the same one as used previou~ly.'~It is a thermoelectrically cooled m o m s o n TH7882) 384 x 576 pixel array, each pixel being square with 23-pm edges. The camera head (CH-220, Photometrics TH7883-PM) was cooled to -40 'CC. The camera electronics unit (CE200) contains an analog-to-digital converter (ADC) providing 14 bits of resolution, at a conversion gain that is software controllable. The camera controller (CC200) contains a 68000 processor, image-frame RAM (12 MB), h w a r e ROM, an IEEE-488 interface to a host personal computer (386 compatible),a "mouse" port, and interface circuitry for a video monitor (RS170). The E 1 7 0 subsystem and the mouse provide feedback for the experimenter during both equipment alignment and data acquisition. The exposure time and data acquisition time were controlled by the commercial CCD software provided by Photometrics. Data analysis is performed off-line, using imaging analysis software written in TurboBasic. During data acquisition, all images are stored in the cache memory, which are later transferred to the hard drive for further manipulation. (17) Koutny, L. B.; Yeung, E. S. Anal. Chew. 1993,65, 182-187.
2576 Analytical Cbemistv, Vol. 67, No. 75, August 1, 1995
RESULTS AND DISCUSSION
Some Design Considerations of the Apparatus. Many components of the on-line imaging system, as shown in Figure 1, had to be carefully selected because both absorption and fluorescence occur in the ultraviolet region. Many commercially available CCDs have a low quantum efficiency below 400 nm. The CCD in our system was coated with 0.4-O.&pm-thick MetaChrome 11, a proprietary chromophore that is vacuum deposited on the CCD sensor. With this coating, a quantum efficiency of 20 % is maintained for the entire ultraviolet region. Since deep (18) Levison, S. W.; McCarthy, IC D. In Cultun'ngNerue Cells; Banker, G., G o s h , IC, Eds.; MIT Press: Cambridge, MA 1991; pp 309-336.
1
300
305 nm
0
300
Wavelength (nm)
Figure 2. Excitation (350-nm fluorescence) spectrum of 1 x M serotonin solution (pH 7.4 balanced salt solution) at room temper-
3.50
400
500
450
Wavelength (nm)
Fluorescence spectra of serotonin solution (1 at different excitation wavelengths. Figure 3.
x
M)
ature.
W excitation is employed, most nonquartz lenses and objectives have strong autofluorescence that overlaps with the sample fluorescence, even though they may be designed for 3 4 h m operation, We did not h d it necessary to replace all of the internal optics inside the microscope since the fluorescence from our sample is in the range between 350 and 400 nm (see below). However, it is necessary to use a quartz objective since the laser beam is directed toward it during excitation. Another concern was that even though the optical filter blocks most of the laser beam, there are still a large number of stray photons reaching the CCD detector. We addressed this problem by placing two filters in the emission path. An UG-1 filter passes 305 and 350 nm with efficienciesof about 1%and 40%,respectively. So,when two UG1 filters were used, the efficiencies in passing these two wavelengths of light will be 0.01% and 16%. As expected, we achieved a better signal-to-noiseratio with the two-filter arrangement even though transmission is reduced. There were spatial fluctuations in the background and the sample signals due to nonuniformities in illuminationby the laser beam and uneven sensitivities of the CCD pixel elements. To correct the images for these variations, the CCD imaging system was operated in a mode commonly called flat-fieldmg.l7 This was accomplished by placing an ultraclean quartz slide in the field of view, illuminating it with the laser beam, acquiring a blank image of the transmitted light through partially blocking filters, and storing that digitally prior to the sample imaging process. After sample images were acquired, they were normalized by simply dividing the image of interest by the blank, pixel by pixel, and multiplying the result by a constant factor. These calculations require only a few seconds for our 576 x 384 images. The resulting images can then be displayed immediately and are also stored digitally for later analysis and printing. Excitationand Fluorescence Spectra of Serotonin. Figure 2 shows that serotonin has significant native fluorescence. There are two peaks in its excitation spectrum, one around 250 nm, and the other around 315 nm. When excited at 275 nm or at 305 nm, it has a broad emission band ranging from 315 to 400 nm (Figure 3), peaking at 350 nm. Of all the Ar ion laser lines, the strongest emission is achieved with 305-nm excitation. Illumination at 330 nm also causes fluorescence of serotonin, but the band is red shifted (Figure 3). It appears that there are two unresolved peaks between 370 and 400 nm. This indicates that one might be able to monitor serotonin release or uptake even by conventional visible
optics for the collection of fluorescence signal. Exposure times may be unacceptably long in conventional microscopes, however, due to the low output of standard light sources. Alternatively, the 325-nm HeCd laser line might be used. We have chosen 305 nm excitation for the native fluorescence studies. The molar absorptivity of serotonin at this wavelength is only about onetenth of that at the maximum. However, our laser can produce hundreds of milliwatts at this wavelength and enough light was absorbed to produce adequate signals for detection in our present experiments. We tested the sensitivity of this system for detection of serotonin by building a narrow channel on a microscope slide to contain controlled concentrations of this neurotransmitter. Our UV laser microscope system could readily detect serotonin to 1 x lo-’ M in a 100,um-deep channel. On washing with buffer solution, no residual fluorescence was detected. For a M solution, the detector response is linear to 5@msexposure time, as limited in our case by the electronic shutter. As is well-known, the minimum cycle time of CCD cameras in the shutter-open mode depends on the size of the array imaged (up to 4MHz data rate for the newer CCD systems). The minimum exposure time depends on the signal level, which is a function of laser power, fluorescence efficiency, and concentration. Quantitative Analysis of SerotoninUptake. In cell culture, astrocytes are known to take up serotonin from the external saline and internalize this neurotransmitter. This uptake process had been conflrrned by control experiments with chemical b10ckers.l~ We have used this property to test the ability of our W laserbased microscope to detect neurotransmitter in living cells. Prior to imaging astrocytes, culture medium was exchanged for a standard saline that contained 140 mM NaCl, 2 mM MgClz, 2 mM CaC12,5 mM KCl, and 10 mM HEPES @H 7.4). Typically, one native fluorescence image of a region of the cover slip containing astrocytes was acquired in the presence of the standard saline solution. Subsequently,serotonin was added to the solution for times of 2, 5, 10, or 20 min. Serotonin uptake in astrocytes is known to be rapid under these conditions and at these concentrat i o n ~ .We ~ ~found no difference in the final fluorescence level with these different incubation periods. Then the serotonin was washed out by flushing the cells with saline for at least 2 min, and a second fluorescence image was acquired. Further washing for up to 20 min did not reduce the signal levels in the areas devoid (19) Dave, V.; Kimelberg, H.K 1.Neurosci. 1994,14, 4972-4986.
Analytical Chemistry, Vol. 67, No. 15, August 1, 1995
2577
Table 1. Enhancement oi Serotonin Signal before and after Uptake
cell
dish no. 1
normalized signala before uptake &er uptake 0.038 0.019 0.033 0.021 0.025 0.029 0.015
2 3 A
0.089 0.088 0.18 0.20
0.096 0.13 0.10
enhancement factor 2.4 4.6 5.4 9.6 3.9 4.3 6.7
Normalized signals are calculated as follows. first, the cell (signal) and noncell (background) areas in the images are identified. The intensity of each area is calculated. Then the background intensity is subtracted from the signal intensity. Finally, this intensity difference is divided by the background intensity to obtain the normalized signal.
Histograms of fluorescence intensity in individual CCD pixels for cell samples before (a) and after (b) serotonin uptake. Figure 4.
of cells. For densely cultured cell samples, we did not attempt to gain spatial resolution. Instead, we calculated the native fluorescence enhancement due to serotonin uptake for all the cells present in the field of view. Because of stray light and fluorescence from the buffer solutions and the cover slips, even the noncell areas in the image produce a background signal in the CCD camera. This background must be subhafted from the intensities recorded in all pixels to depict fluorescence originating from the cells. To account for variations in laser intensity and in l i h t collection for each cell sample studied, the backgroundsubtracted intensities are further normalized by dividing each by the background intensity (nonce11 area) in the corresponding image. The results are presented in Table 1. We note that there is native fluorescence from the cells even before serotonin uptake, as one can clearly observe outlines of the cells in the CCD image. Such preuptake images are not due to Raman or Rayleigh scattering, which are estimated to be substantially weaker. This is not surprising considering that natural cell components such as proteins and other aromatic compounds, including serotonin, can absorb and emit at similar wavelength ranges. The present setup is thus limited in selectivity unless multiple excitation wavelengths, dispersed fluorescence, or both are incorporated. It is however entirely appropriate to study serotonin uptake, since a difference signal over time can be recorded regardless of the original fluorescence levels. As shown in Table 1,we have obtained enhancement factors ranging 2578
Analytical Chemktfy, vol. 67, NO. 15, August I , 1995
of astrocytes grown on a quartz cover slip: (a. top) optical image before the introduction of serotonin; (b. bottom) image of increase in UV-excited fluorescence after serotonin uptake. Comparison of the two panels reveals that serotonin uptake ranges from being negligible (A), moderate (B), to substantial (C).In some cases (D), it appears that the serotonin is localized in certain regions of the cell body rather than uniformly spread over the cytosol. Figure 5.
from about 2.5 to 10. The enhancement of cellular fluorescence in all cell samples studied demonstrates that this UV microscope system has the sensitivity to detect changes in the concentration of the serotonin in living cells.
Figure 4 shows a histogram of the frequency of pixel intensities before (a) and after (b) serotonin uptake. Initially almost all the imaged pixels have an intensity between 13 000 and 16 000 (the background has not been subtracted here) before serotonin uptake. For these high-density cultures, almost all pixels correspond to cell areas. So, we did not attempt to separately identjfy the pixels associated with the cell vs noncell areas. The signals presumably reflect native fluorescence from the cell membrane (higher values) and the cytoplasm (low values). After uptake, most of the pixels have a much greater intensity, ranging from 19 000 to 26 000 with a minority retaining intensities of the order of 15 OOO-19 OOO. We conclude that the former pixels correspond to areas where serotonin has been taken up and the latter pixels correspond to noncell areas. S i c e astrocytes were grown at high density on the cover slip, areas of which did not contain cellular material were a small portion of the total area, as indicated by the frequencies of intensities below 19 000 in Figure 4b. The background fluorescence did increase slightly due to incomplete washing from the poly (L-1ysine)coated surface but is readily distinguishable from cell fluorescence. Cell Imaging. In subsequent experiments, we took advantage of the spatial resolution of our system to resolve changes in fluorescence on a pixel-by-pixelbasis withii individually identified cells. To perform this experiment we ensured that the position of the astrocytes would not be changed upon serotonin addition by affixing the dish tightly onto the microscope stage. Figure 5a shows that astrocyte cells are sparsely grown in this region on the cover slip. This conventional optical image of the cells clearly shows that most of the cells are viable, as can be inferred from ‘theirmorphology. A native fluorescence image of the same region was then acquired with 305-nm laser excitation. After serotonin addition and washout, a second fluorescence image was obtained. We digitally subtracted these images from each other to yield a “difference”image to display changes in cellular fluorescence.This is especially important since other cellular species (e.g., proteins) also fluoresce when excited at 305 nm. As shown in Figure 5b, the fluorescence difference image shows certain cells as bright areas with a one-to-one correspondence with the optical image (C). The increase in intensities for each cell is not identical (B vs C). Furthermore, within individual cells, the fluorescence is not uniformly distributed throughout the cellular cytosol but appears to be restricted to a specific cellular compartment 0). Previous studies, using radiolabeled serotonin, have not been able to resolve the specific
cellular region that serotonin is taken into.lg It is also clear that some cells did not take up serotonin (A). It is, however, not clear why these have distinct physiology, since morphologically they appear viable. This observation clearly demonstrates the utility of this newly developed technique, which provides both temporal and spatial resolution of changes in the concentration of the neurotransmitter, serotonin, within living cells. CONCLUSION
We have successfully demonstrated a detection scheme for the in vivo imaging of serotonin in astrocytes grown on cover slips. The combination of W laser excitation, native fluorescence detection, and CCD-based microscopy is well suited for real-time monitoring of neurotransmitters, although time resolution was not required here. The detection methodology has high sensitivity and low LODs and does not require coupling to fluorescence dyes. Individual astrocyte cells can be identified before and after serotonin uptake. We found a large enhancement in the fluorescence of cells due to the transport of serotonin from the bathing saline into individual astrocytes. The spatial resolution achievable for the system is 0.22 pm, and the temporal resolution obtainable is around 50 ms for low4M serotonin. We have simultaneously identified and monitored about 20 astrocytes in one single experiment. The potential applications of our system include the mapping of neurotransmitters in living cells as well as the ability to image the release of neurotransmitters from specific regions of living cells in response to external stimuli. ACKNOWLEDGMENT
The authors thank Christopher Price and Xiandan Lu for technical assistance. W.T. is a U.S. Department of Energy Distinguished Postdoctoral Fellow sponsored by the Office of Science Education and Technical Information and administered by Oak Ridge Institute for Science and Education. The Ames Laboratory is operated for the US. Department of Energy by Iowa State University u der Contract W-7405-Eng-82. This work was supported by the irector of Energy Research, Office of Basic Energy Sciences, Division of Chemical Sciences, the McKnight Foundation, and the National Institutes of Health (NS26650).
a
Received for review February 16, 1995. Accepted May 17, 1995.B AC950172S @Abstractpublished in Advance ACS Abstracts, July 1, 1995.
Analytical Chemistty, Vol. 67, No. 15, August 1, 1995
2579
