Latin American and Caribbean Consortium of Engineering Institutions

Spectrum-Resolved Lifetime Imaging of Biological Tissues In Vitro and In Vivo using Multiphoton Laser Scanning Microscopy

Department of Mechanical and Materials Engineering
Florida International University
Miami, FL 33174, USA

In this work, a modern biomedical instrument—multiphoton laser scanning microscope (MPLSM)—is used to image biological specimens by inducing fluorescence from stained fluorophores or tagged fluorescence proteins both in vitro and in vivo. By integrating with a time-correlated single-photon counting (TCSPC) detector and a lifetime measurement module, the modified MPLSM system extracts further optical information—fluorescence lifetime—from the fluorescent emission of the specimens. Lifetime images of several biological specimens are presented, showing that the lifetime information extracted from the fluorescence signals can serve as indicators to identify the biological, physical, and chemical microenvironments of the fluorophores, e.g., biochemical constituents, pH value, and temperature. Moreover, lifetime images at various emission wavelengths are collected using a concave optical grating, providing additional optical information from the same biological specimen. The presented biological imaging technique exhibits promising potentials for human diseases diagnosis.

Multiphoton laser scanning microscopy, Spectral-lifetime imaging, Fluorescence imaging

Multiphoton laser scanning microscopy (MPLSM) is the new generation of laser scanning microscopy technique, which is based on nonlinear matter-light interactions to visualize the distribution and dynamics of biological tissues, cells, and sub-cellular organelles. Unlike conventional confocal laser scanning microscopy (CLSM), where a blue or UV excitation laser light with wavelength ~400 nm or shorter is applied to induce fluorescence emission from a biological specimen via single-photon excitation, MPLSM employs a high-intensity pulsed laser with longer wavelength (NIR) to induce multiphoton excitation, where several photons are simultaneously absorbed to excite a fluorophore (Zipfel et al., 2002). One immediate benefits of a longer excitation wavelength is the reduced scattering, as the Rayleigh scattering effect scales as l^-⁴ (Hecht, 2001) . If the wavelength is doubled, the scattering is 16× weaker than that of the original wavelength––over an order of magnitude reduction. This unique feature makes it possible to induce deeper fluorescence sectioning in a highly scattering biological environment, e.g., lipids in cytoplasm, dramatically increasing the sectioning capability of a laser scanning microscope. Meanwhile, since multiphoton excitation occurs exclusively at high-photon-density region, which is the focal point, optical section of the image is intrinsically well-defined; the confocal aperture used in CLSM to block fluorescence emission out of the focal region is no longer required. The confinement of excitation to the small focal area also eliminates most of the undesired interactions with the excitation light, resulting in less phototoxicity and photobleaching for living specimens, which makes MPLSM valuable for studying biological behaviors and functions in vivo (Squirrell et al., 1999) .

Although MPLSM has been used to image and trace labeled proteins, observe thick sections, and visualize three-dimensional structures, due to the strum overlapping of fluorescence spectra in a multi-labeled specimen, e.g., yellow fluorescent proteins (YFP) and green fluorescent proteins (GFP), optical filters and specific computer algorithms are generally required to enhance image discrimination (Dickinson et al., 2001; Lansford et al., 2001) , since a photon emitted in a wavelength region shared by two (or more) labeling proteins has no history of which particular protein it comes from. However, one useful information that can be further extracted from the fluorescence emission is the fluorescence lifetime (Pawley, 1995) , which will shift when the histological stains and the fluorescence proteins are bound to different components of the cells and tissues, i.e., lifetime values can clearly identify the microenvironments of the specimens. The combination of spectral images with lifetime information will maximize the usage of fluorescence signals. Such combined information will be presented in this work and can be significant for biological imaging and medical diagnosis.

The lifetime imaging system in this work combines a multiphoton laser scanning microscope with a fluorescence lifetime measurement system. Fluorescence emission from fluorophores is induced by fs laser pulses through multiphoton excitation, and the lifetimes are determined using a time-correlated single-photon counting (TCSPC) system. To resolve the spectral information of the emission, a concave optical grating is used to disperse the fluorescence emission. For the overall or specific wavelength, the lifetime values of each pixel are then displayed as pseudocolors to form a lifetime image.

After the fluorophores are excited by single- or multiphoton excitation, the excited electron states are unstable and can be occupied only for a very short time period. One of the mechanisms for the excited electrons to relax back to the ground state is fluorescence emission. If one monitors the fluorescent decay with time, an exponential decay of the intensity I can be observed, which takes the form:

where I_f₀ is the peak intensity of fluorescence emission, t is time, and t is the fluorescence or excited electron lifetime. When time t equals the fluorescence lifetime t, 63% of the excited electrons have returned to the ground state. Generally speaking, the lifetime values of each scanned pixel can be determined by single exponential decay curve with the associated lifetime t. However, due to the limited resolution of each scanned pixel, the collected fluorescence emission may be resulted from more than one fluorophore. In such a case, a double-exponential or higher order exponential decay will appear. The lifetime, therefore, can be determined by (Subramaniam et al., 1999) :

where a_i is the i^th lifetime component associated with t_i. Instead of directly monitoring the intensity decay, the current technology also allows lifetime measurement to be conducted in either frequency or temporal domain (Pawley, 1995) . For measurement in vivo in the current work, due to the weaker fluorescence emission and higher sensitivity to signal-to-noise ratio of the detectors, a TCSPC system combined with a fast, high-sensitivity photomultiplier (PMT) is employed to efficiently measure the fluorescence lifetime in the temporal domain for each scanned pixel.

The idea of TCSPC is based on the detection of every single fluorescence photon over a periodical excitation light signal to form a waveform of intensity decay (O'Connor et al., 1984) . The current technology makes the system very efficient, easy to set up, and able to detect weak fluorescence signals. TCSPC system employs a fast PMT with an instrument response < 30 ps (Becker et al., 2001) . As the fs pulsed laser has high repetition rate of 80–120 MHz, a state-of-the-art TCSPC device can achieve count rates in the MHz range and acquisition times down to a few ms (Becker et al., 2001) . A high-resolution lifetime image, e.g., 512 × 512 pixel, can be collected within several tens of seconds.

The experimental configuration for lifetime imaging using a multiphoton laser scanning microscope is shown in Fig.1(a). The specimens are mounted on a Nikon inverted microscope (Diaphot 200) with an oil immersion objective (NA = 1.4). A Coherent DMP-1000 Nd:YLF modelocked fs laser with a nominal wavelength l = 1047 nm, a pulse duration t_p ~ 200 fs, a maximum pulse energy E_p = 7 nJ, and a repetition rate R_p = 120 MHz serves as the light source to excite the fluorophores via either two- or three-photon excitation. The fs laser pulses are delivered to the specimens via a Bio-Rad MRC 600 scanning head. To avoid photodamage, the pulse energy is attenuated by a neutral density optical filter. The induced fluorescence is then collected by a lifetime PMT (Becker & Hickl PML-100) mounted underneath the microscope. The dichroic mirror reflects the IR excitation light to the microscope and scanning head but allows the visible fluorescence to pass through and reach the lifetime PMT (Fig. 1(a)). The signals from both the lifetime PMT and the synchronization fast PIN photodiode located at the laser exit are finally used to determine the fluorescence lifetime through a Becker & Hickl SPC 730 TCSPC system to form a lifetime image of 512 × 512 pixel, each of which has 256 time channels.

To include the other dimension into the images and measurement, i.e., fluorescence wavelength, a second setup was built, which include a concave optical grating locating in front of a 16-channel PMT (Becker & Hickl PML-16), as shown in Fig. 1(b). The grating is able to dissolve the emission spectra, with resolution of ~10 nm. Different wavelength component will be focused on one of the channels. Each channel serve as an independent lifetime PMT to image the specimens associated with the specific wavelength. Calibration of channels in terms of their corresponding wavelength were made by sending two attenuated lasers with given wavelengths.

Several specimens were examined in this work, including fixed slices and living embryos. The dimension of each image is roughly 100 × 100 mm. Collecting time is 40–60 s for fixed slices, while it takes longer for live embryos, about 80 s, due to their relatively weaker fluorescence emission. Further discussion and analysis are addressed as follows.

The fixed specimens under test are slices of brain, liver, uterus, and kidney stained with several conventional chemical dyes to identify cellular histology with high quality and resolution. Four regular fluorescence dyes are used in this work: Thionine, methyl green, Tfl, and acridine orange. Two fixed sliced specimens stained by Thionine are demonstrated in Fig. 2. The Thionine has a maximal excitation wavelength of 596 nm (Gurr, 1971) , thus fluorescence is induced by two-photon excitation. Fluorescence lifetime of each scanned pixel was measured by photon counting PMT and determined using TCSPC system, the values of which are then expressed as different colors. As the fluorophores bind with other molecules in their microenvironments, e.g., lipids and proteins in a cell, the fluorophore exhibits a shift in the fluorescence lifetime. In Fig.2 (a), the lifetime image of a Cynomolgus monkey uterus slice is shown, where the red color represents a shorter lifetime around 400 ps while the blue indicates a relatively longer value of 800 ps. As can been seen, the blood cells around the center exhibit shorter lifetime, i.e., yellow, when compared with their neighboring tissues with green and blue colors. In Fig. 2 (b), Thionine shows even shorter lifetime when bound with brain cells and tissues. Due to the distinct histology of the brain slice, it is very clear to see the resulted variation in lifetime distribution in the image. The lifetime images, therefore, highly improve the quality and contrast compared to the regular confocal and multiphoton images, providing additional information to pathologists and biomedical researchers.

Another two fixed slices stained with methyl green with nominal excitation wavelength of 632 nm (Gurr, 1971) are shown in Fig. 3. Figure 3(a) shows a lifetime image of the medulla of a kidney from a Cynomolgus monkey. From this image, tissue histology can be clearly identified: (1) collecting ducts (cd), which are recognized by their columnar epithelium, (2) collecting tubules (ct), which have a cuboidal epithelium with a diameter that is wider and less regular than the ascending thick limbs, (3) the loop of Henle labeled as (I), (4) the descending thin limbs of the loop of Henle labeled as II, and (5) the vasa recta labeled as III, which are capillaries filled with erythrocytes. In addition to the histological information, shorter lifetimes of nuclei in the collecting duct, tubule and ascending thick limb than that of the surrounding cytoplasm can be found. On the other hand, the cells of descending thin limbs exhibit longer lifetimes than the surrounding structures. Another image is a liver slice stained with methyl green and shown in Fig. 3(b). Though applying the identical fluorophore, a big shift in lifetime distribution is found as compared with the kidney slice in Fig. 3(a), indicating the lifetime of the fluorophore varies and is very sensitive to the biochemical and biophysical environments.

One unique feature of MPLSM is to minimize phototoxicity and improve viability, which makes it ideal to observe living specimens. Moreover, with the availability of extensive sequence data from many organisms, fluorescent protein chimeras can be made as probes for any specific protein and make it possible to identify the location and distribution of the tagged protein within a living organism through MPLSM. Using the identical system, not only for fixed slices, the lifetime images of wild-type C. elegans and drosophila are also taken in vivo in this work.

Two wild-type C. elegans embryos were separately tagged with GFP and YFP to histones, which appear in chromosomes at the cell nuclei. During the scanning process, autofluorescence was observed with the membranes and some other organelles in the embryos due to their endogenous fluorophores. Figure 4(a) shows the obtained lifetime image of early-stage C. elegans embryos tagged with GFP, where a continuous lifetime spectrum is shown with a range of 0.8–2.0 ns. The nuclei in the embryos can be clearly identified, which exhibit a longer lifetime shown by blue color. The lifetime of the autofluorescence shows relatively shorter values with a green color. Interestingly, the gut of C. elegans is also autofluorescent and shows a shortest lifetime, which is displayed as yellow at the bottom of Fig. 4(a).

A second C. elegans embryos were tagged with YFP. Similar to the results of GFP, fluorescence from nuclei shows longer lifetimes around 2.0–2.6 ns (Fig. 4(b)), which agrees very well with the values when tagged with GFP in Fig. 4(a). One interesting finding is a dead embryo at the upper-left region in the figure, which only generates autofluorescence without any lifetime signature of the labeled nuclei.

One significant advantage of MPLSM is its capability to image deeper sectioning–––therefore, a stack of sectioning images can be reconstructed to form a 3-D image. Furthermore, if the images are recorded in a sequence of time, then a 4-D movie can be made (Thomas and White, 1998) . In this work, a 3-D lifetime image is presented using a stack of sectioning images, providing additional information for diagnostics as well as for the identification of cellular structures.

The specimen is kidney stained with Tfl. Nine sectioned lifetime images within a total depth of 50 mm were collected individually, which can be seen in Fig. 5(a). The stacks of images exhibit various lifetime ranges at different sections, e.g., shorter lifetime at the upper slices in Fig. 5(a). This variation may be caused by the residual stress distributed in the specimen, which suggests an additional application of the current technique in biomechanics study of cellular structures.

Note in Fig 5(a), no information was collected for the space between slices due to the finite sectioning depth. However, a user-friendly software, VisBio, developed by Laboratory for Optical and Computational Instrumentation at University of Wisconsin–Madison can be employed to produce 3-D image with information interpolated between the 2-D slices (Eliceiri et al., 2002) . As VisBio allows any arbitrary slice across the stacks of images to be visualized, the cross-section image cut by an inclined slice in Fig. 5(a) can be obtained, which is shown in Fig. 5(b). Comparing with Fig. 5(a), the new 3-D image in Fig. 5(b) provides more information beyond the 2-D slices, with reasonable lifetime distributions observed, i.e., shorter lifetime in the upper sections than the lower ones.

Text Box:

Fig. 6 Spectral multiphoton fluorescence images of a monkey uterus slice stained by AO. (a)–(e) images with wavelength from 501–550 nm; (f)–(j) images with wavelength from 611–660 nm.

In addition to lifetime imaging, the information of fluorescence wavelength can be reported in situ to form spectral images. One example of spectral images of Cynomolgus monkey uterus slice stained by AO is shown in Fig. 6. Figure 6(a)–(e) present the shorter-wavelength images from 501–550 nm (10-nm wavelength difference in between), while Fig. 6(f)–(j) demonstrate the longer-wavelength counterparts from 611–660 nm. As can be seen, the images successfully discriminate cell nuclei amongst membranes of the surrounding tissues by longer wavelength. The presented wavelength resolved image not only demonstrate its capability to identify specific cell organelles from the multiphoton fluorescence imaging, but also present promising potentials for disease diagnostics such as human cancers.

Not limited to spectral images, the presented technique is also able to measure the spectral-resolved lifetime values in situ to generate lifetime images. Figure 7 demonstrates the associated lifetime images at two specific wavelength regimes for the identical uterus specimen shown in Fig. 2, where the lifetime values are presented as pseudocolors. Figure 3(a) is the image for l = 531–540 nm, which shows good agreement with Fig. 6(d). However, the cell structures can be more clearly identified by the colors, i.e., the lifetimes, in Fig. 7(a). As can be seen, although the same fluorophore is used, different organisms present different fluorescence lifetimes and the red color (shorter lifetime) clearly labels the nuclei of cells in the center. As the cell nuclei have shorter lifetime values, the orange color of the neighboring tissues represents potential locations of nuclei that appear at longer-wavelength images. At the other extreme, the lifetime image at longer wavelength of l = 621–630 nm is shown in Fig. 7(b), where the yellow color identifies the nuclei of the surrounding tissues with good agreement to Fig. 6(h). Based on all the lifetime measurement conducted in this work, an interesting observation is that the fluorescence lifetime tents to increase with longer wavelength components, indicating a further research topic on the fundamental mechanisms for wavelength-dependent lifetime.

The authors gratefully acknowledge the support for this work by Prof. J. G. White and Mr. K. W. Eliceiri at UW-Madison. Also, the authors appreciate Dr. Al Kutchera for specimen preparation and the valuable comments from Dr. Jayne Squirrell, Dr. Fern Finger, and Prof. Nirmala Ramanujam.

Dickinson, M.E., Bearman, G., Tille, S., Lansford, R., and Fraser, S.E. (2001). "Multi-spectral imaging and linear unmixing add a whole new dimension to laser scanning fluorescence microscopy," Biotechniques, Vol. 31, No. 6, pp. 1272.

Gurr, E. (1971). Synthetic Dyes in Biology, Medicine and Chemistry, Academic Press, New York.

Eliceiri, K.W., Rueden, C., Mohler, W.A., Hibbard, W.L., and White, J.G. (2002). "Analysis of multidimensional biological image data," Biotechniques, Vol. 33, No. 6, pp. 1268.

Hecht, E. (2001). Optics, 4^th edition, Addison-Wesley Publishing Co., New York.

Lansford, R., Bearman, G., and Fraser, S.E. (2001). "Resolution of multiple green fluorescent protein color variants and dyes using two-photon microscopy and imaging spectroscopy," Journal of Biomedical Optics, Vol. 6, No.3, pp. 311-318.

O'Connor, D.V. and Phillips D. (1984). Time-Correlated Single Photon Counting, Academic Press, New York.

Pawley, J.B. ed. (1995). Handbook of Biological Confocal Microscopy, Kluwer Academic Publishers, New York.

Squirrell, J.M., Wokosin, D.L., White, J.G., and Bavister, B.D. (1999). "Long-term two-photon fluorescence imaging of mammalian embryos without compromising viability," Nature Biotechnology, Vol. 17, No. 8, pp. 763-767.

Subramaniam, V., Volkmer, A., Striker, G., Birch, D.J.S., and Jovin, T.M. (1999). "Comparison of one and two-photon excited fluorescence lifetime and anisotropy of green fluorescent protein (GFP) and its mutants," Biophysical Journal, Vol. 76, No. 1, pp. A446.

Thomas, C.F. and White, J.G. (1998). "Four-dimensional imaging: the exploration of space and time," Trends in Biotechnology, Vol. 16, No. 4, pp. 175-182.

Zipfel, W.R., Williams, R.M., and Webb, W.W. (2002). “Nonlinear magic: multiphoton microscopy in the biosciences”. Nature Biotechnology, Vol. 21, No. 11, pp. 1369-1377.

† Assistant Professor. Author to whom correspondence should be addressed.

E-mail: dfan@fiu.edu. Tel: (305) 348–4027, Fax: (305) 348–1932