In stem cell research, there is a high demand on techniques for the noninvasive, marker-free observation of growth, proliferation, differentiation, and stability of living stem cells under physiological conditions. The use of exogenous markers may alter the metabolic balance, influences the cell life, and avoid further therapeutic application. Noninvasive multiphoton microscopes with near-infrared (NIR) femtosecond laser sources1, 2, 3, 4 have been applied to image living single cells and different tissues with a high spatial resolution without any staining. Two-photon autofluorescence is obtained due to intrinsic fluorophores such as nicotinamide adenine (phosphorylated) dinucleotide [NAD(P)H], flavins, porphyrins, elastin, and melanin. In addition, second-harmonic generation (SHG) is induced by certain biomolecule structures such as collagen.5, 6, 7, 8, 9
In addition to the measurement of the fluorescence/SHG intensity by optical sectioning (3-D imaging), fluorescence lifetime imaging (FLIM, 4-D) and spectral imaging (5-D) can be performed. In particular, the arrival times of the fluorescence photons with respect to the excitation of the molecule and the particular location (pixel) can be determined by time-correlated single photon counting (TCSPC) and the use of photomultipliers with as short rise time. When using a photomultiplier tube (PMT) array in combination with a polychromator, the “color” of the emitted photon per pixel can be also determined (spectral imaging).
Multiphoton imaging is very suitable for long term analysis of living cells due to the absence of out of focus photostress. The excitation volume is limited to a subfemtoliter focal volume. NIR femtosecond lasers enable deep light penetration into tissues and provide the possibility of optical sectioning of 3-D biological objects. Nondestructive imaging of living specimens is restricted to a certain optical window determined by the spectral range and the light intensities.10 Intrinsic cellular fluorophores deliver information on the cell structure as well as on cell activities. Authors could distinguish certain cancer cells from normal cells,11, 12 even malignant from benign tumors,13 by optical properties of native fluorescent molecules. Many studies on the intrinsic metabolic state show that innate cellular fluorescence has the potential to discriminate proliferating and nonproliferation cell population,14 and also metabolic changes of self-renewing and differentiating cells.15, 16 The differentiation of human mesenchymal stem cells into an adipogenic pathway (formation of lipid droplets, change of morphology and autofluorescence) has been investigated recently by two-photon microscopy.17
A unique feature of stem cells is the extended self-renewal of the cell population and their differentiation potency. The aim of the present study was to investigate the two-photon excited autofluorescence of human stem cells and the onset of collagen production of differentiated cells by the detection of SHG signals. An interesting question arises as to whether it is possible to select differentiated cells from undifferentiated cells without interfering with the native environment and without the use of external markers.
Here we report on the application of a 5-D two-photon microscope for imaging different human and animal stem cell lines by autofluorescence and SHG in monolayers and spheroids.
Material and Methods
Stem Cell Lines
The stem cell lines human salivary gland stem cells (hSGSC), human pancreatic stem cells (hPSC), rat pancreatic stem cells (rPSC), and human dental pulp stem cells (hDPSC) were used18, 19, 20, 21 for two-photon autofluorescence imaging. FLIM measurements and spectral imaging were performed on hSGSC stem cells. The first three cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (GIBCO, Germany) containing glucose, L-glutamine, and supplemented with 15% fetal calf serum Gold (FCS-Gold) (PAA Laboratories GmbH, Germany), penicillin, and streptomycin. hDPSC were cultured in with 15% FCS, L-ascorbic acid 2-phosphate, L-glutamine, penicillin, and streptomycin. Cultures were maintained at in a 5% humidified atmosphere.
For imaging, cells were transferred (confluence 80 to 90%) into sterile miniaturized cell chambers with two thick glass windows with etched grids (MiniCeM-grid, JenLab GmbH, Jena) and additionally incubated for another day until they were attached to the glass surface.
Pellet Culture and Chondrogenic Differentiation
For chondrogenic differentiation, a pellet culture was used. Approximately 200,000 cells were placed in a polypropylene tube and centrifuged to a pellet for . The flattened pellet at the bottom of the tube was cultured at with 5% in the culture medium for until a spherical form (spheroid) was formed. For the differentiation, the spheroids were cultured in of the chondrogenic medium DMEM (GIBCO, Germany) supplemented with transforming growth factor , dexamethasone, ascorbic acid-phosphate, L-proline, natrium piruvate, 15% FCS-Gold, penicillin, and streptomycin. The medium was replaced every for a period of . For microscopy, the pellets were transferred to dishes with thick glass coverslips.
Instrumentation and Measurements
A laser scanning microscope (LSM 510 Meta NLO, Zeiss, Germany) equipped with a tunable Ti:sapphire laser source (Chameleon, Coherent Inc., Santa Clara, California) was used to study autofluorescence and SHG of stem cells. For autofluorescence imaging, laser beams at 750, 850, and ( laser output) at mean power were used. The beamsplitter HFT KP 650 and the short pass filter KP 685 were employed. The cells were scanned in a field at dwell time. Fluorescence emission was recorded through a Plan-Neofluar 1.3 oil objective. The experimental parameters such as scanning time and the settings for contrast and brightness were identical for all autofluorescence images obtained from different cell lines for comparison purposes.
For SHG detection, the external PMT-GaAsP (H7422P-40, Hamamatsu, Japan) was mounted to the microscope in a forward direction. A SP 610 filter was installed in front of the detector to prevent scattered laser light onto the detector. SHG signals were collected through a bandpass filter at an excitation wavelength of .
3-D two-photon autofluorescence and SHG images were obtained by optical sectioning with -intervals of . Fluorescence lifetime imaging (FLIM) was performed by single-photon counting (TCSPS) using the board SPC-830 (Becker and Hickl GmbH, Germany) and the fast detector PMC-100 with an SP 610 filter.
The instrumental function (IRF) of the optical system was measured using the second-harmonic generated signal from collagen II (Sigma-Aldrich, Germany). SPCImage software was used to analyze the fluorescence lifetime decay curves as well as the calculation of histograms (Becker and Hickl GmbH, Germany). Spectral imaging was performed with the 32 channel PMT array (META, Hamamatsu). The grating in the polychromator provided a resolution of per channel. Typically, the spectral range of was imaged. Figure 1 shows the experimental setup.
Two-Photon Autoflourescence Imaging and Spectral Analysis
The two-photon measurements showed that different stem cell lines have an autofluorescence with different intensity. hDPSC, hPSC, and hSGSC human stem cells exhibited an intense autofluorescence compared to rPSC rat cells. Figure 2 demonstrates the differences in the fluorescence intensity between human and rat pancreatic cells. Also, we noted significant fluorescence intensity variations between cells of the same line (Fig. 3 ). Intrinsic fluorescence from living cells was detected at different NIR excitation wavelengths (750, 850, and ) to separate different native fluorescence molecules. When using , NAD(P)H as well flavins/flavoproteins can be excited by a two-photon process. However, when changing to 850 or even , flavins/flavoproteins (e.g., FAD) only and not NAD(P)H can be efficiently excited.22 Spectral measurements (Fig. 4 ) showed that the detected fluorescence light was intense in the blue/green spectral region. The maximum of the fluorescence was found to be at when excited with light, which is consistent with the emission behavior free and protein bound NADH. The maximum shifted to when using light, which is consistent with flavin emission (Fig. 5 ). We have proven that the presence of the culture medium did not add significant background to the autofluorescence signals from the cells.
Fluorescence Lifetime Imaging Measurement and Analysis
Fluorescence decay curves have been obtained by exciting at 750 and . Figure 6 presents a typical excited FLIM image, a histogram, and a fluorescence decay curve of hSGSC stem cells. The decay curve reflects the autofluorescence of one mitochondrium in a specific cell. 6740 photons were detected for this curve. The biexponential fit with the fitting parameter reveals two components. A fast one with a short lifetime of and an amplitude of , and a second one with and the amplitude .
To get a statistical information on the variance of intracellular decay curves, we analyzed the curves of ten different perinuclear fluorescence regions of the bright cell on the right side within Fig. 6. The data are depicted in Table 1 . Interestingly, only one major short component with an average value of [standard deviation (SD) 0.019, amplitude 70.5%] was found in this particular cell. The long component had an average value of (SD 0.077).
Intracellular FLIM data obtained from ten decay curves within the cytoplasm of the bright cell on the right in Fig. 6.
|τ1 (ns)||τ2 (ns)||τm (ns)||a1 (%)||χ2|
When calculating the mean fluorescence lifetime of the long-lived component within this frame of about ten cells in Fig. 6, we obtained a major value of , as depicted in the histogram. When analyzing ten frames of about 100 cells in total of the same cell line, we obtained a mean value of . In contrast, when changing the excitation wavelength to , an average value of 100 cells of was obtained.
When calculating the mean lifetime , per pixel we obtained an average value of for the image in Fig. 6. Considering 100 cells within ten frames, the mean value was found . The employment of the excitation wavelength resulted in a longer value of .
Interestingly, when analyzing the short fluorescent components within these FLIM images (frames), we obtained up to three maxima in the -histogram (Fig. 7 ). The first one at around reflects the influence of backscattered laser light, which was able to transmit through the beamsplitter and the short pass filter (SP 610). When adding a second short pass filter, this peak nearly disappeared. The other two peaks occurred around and at excitation wavelength and shifted to and , respectively, at . Control measurements were done in culture medium as well as after washing the cells with phosphate buffer saline (PBS). The bands were therefore not background signals from the system such as autofluorescence of optics, etc., but of biological origin. We analyzed these two short components in more detail. For that purpose, we performed a statistical analysis of the histograms (Table 2 ) on ten bright fluorescent cells out of Fig. 6 and 8 . Rectangular regions of interest (ROI) were created as depicted in Fig. 8. Each ROI covered one cell. According to the table, just three cells exhibited only one short fluorescence lifetime of about , whereas seven out of ten cells possessed two short components with lifetimes of 0.2 and , respectively.
Average FLIM data calculated from ten regions of interest covering one single cell in each region. “x” means no component was detected.
|cell||τ1a (ns)||τ1b (ns)||τ2 (ns)||τm (ns)||a1a (%)||a1b (%)||ratio a1a:a2b||χ2|
Second-Harmonic Generation Detection
We monitored the biosynthesis of collagen in 3-D salivary gland and pancreatic stem/progenitor cell cultures by monitoring the occurrence of SHG signals for a long time period up to . The spheroids had dimensions of in diameter. The two-photon autofluorescence and SHG images of spheroids were obtained by optical sectioning in -steps of . The occurrence of the first SHG signal was detected after eight days after the introduction of the stimulating agent [Figs. 9a and 9b ].
Detected two-photon autofluorescence with emission maxima at and , and the long fluorescence lifetimes of 1.8 and , respectively, prove the presence of the biomolecules NAD(P)H and flavins/flavoproteins such as FAD in stem cells and stem cell clusters. The occurrence of some extremely bright cells shows the heterogeneity within a cell population of the same culture dish.
Interestingly, the analysis of the short-lived component of human salivary gland stem cells was found to be even more complex. Some cells exhibited only one short-lived fluorophore with a fluorescence lifetime of about only (excellent biexponential fit with of 1.00, Fig. 6), whereas the majority of cells revealed an additional short-lived component with fluorescence lifetime.
The interpretation of these two short fluorophores is complicated. Free NAD(P)H has a short lifetime around , whereas a variety of flavoproteins have short picosecond lifetimes. The free flavin mononucleotide (FMN) has a typical lifetime of ,23, 24 while flavin adenine dinucleotide (FAD) has a lifetime of .24, 25, 26 When binding to a variety of proteins, the fluorescence lifetime of NAD(P)H in a variety of cells and solutions shifts to higher values of about .23 Chia 27 reported on the presence of three NAD(P)H fluorophores in rat brain tissue. A very short one of for the free (nonbound) coenzyme, a second short one of and a long lifetime of for bound NAD(P)H.
Future work needs to be done for the interpretation of the cellular FLIM data and spectra with regard to the specific cellular metabolism and differentiation value. For that purpose, the particular imaged cell of interest should also be characterized with selective biochemical means, such as antibody staining of the cell as well as the surrounding cells.
Very interestingly, SHG can be used to detect the biosynthesis of collagen as a result of the differentiation process.28, 29 We performed long-term studies of up to on 3-D stem cell spheroids and were able to monitor the expression of the extracellular matrix protein. The possibility of nondestructive marker-free imaging allows the study of the organization and development of the ECM structure, and of feedback mechanisms. Recently it was shown that collagen influences the differentiation of stem cells.30
Femtosecond laser microscopes with their capability of nondestructive two- and three-photon excited autofluorescence and SHG/THG imaging may become novel noninvasive multidimensional tools for marker-free optical stem cell characterization and for on-line monitoring of differentiation of living cells in a 3-D environment, including ECM components.
The authors would like to thank Katja Schenke-Layland [University of California, Los Angeles (UCLA)], and Alexander Ehlers for helpful discussions and comments, Daniel Sauer for technical support, and Erwin Gorjup (Fraunhofer IBMT, Sankt Ingbert, Germany) for stem cell characterization.