Fluorescence Lifetime Imaging by Multi-Dimensional TCSPC Adds New Dimensions to Biomedical Imaging
LIM by time-correlated single-photon counting (TCSPC) is based on scanning the sample with a high-repetition-rate pulsed laser beam, detecting single photons of the fluorescence signal, and building up a photon distribution over the times of the photons in the laser pulse period and the coordinates of the laser beam in the scanning area in the moment of the photon detection. The technique favourably combines the advantages of time-correlated photon counting and laser scanning microscopy: It delivers a near-ideal photon efficiency, minimum sample exposure, high time resolution, resolution of multi-exponential decay profiles, and suppression of out-of focus and laterally scattered light.
Imaging techniques based on fluorescence detection have found broad application in life sciences because they are extremely sensitive and able to deliver information about biochemical interactions on the molecular scale. The intensity of the fluorescence depends on the fluorescence quantum efficiency and the concentration of the fluorophore. Images of the fluorescence intensity thus show where in a sample the fluorophore is located, i. e. show the spatial structure of the specimen. The fluorescence spectrum is characteristic of the fluorophore. Images containing spectral information thus allow the fluorophores in the individual pixels of an image to be identified. The fluorescence lifetime or, more exactly, the fluorescence decay function, depends on the type of the fluorophore but not on its concentration. It depends, however, also on the molecular environment of the fluorophore. Fluorescence decay functions and, consequently, fluorescence lifetime images, therefore contain information on the molecular environment of the fluorophore molecules or fluorescently labelled biomolecules [1–3].
There is a number of different techniques to detect the fluorescence lifetime, and to combine fluorescence lifetime detection with imaging. Different principles differ in their photon efficiency, i. e. in the number of photons required for a given lifetime accuracy [4, 5], the acquisition time required to record these photons, the photon flux they can be used at, their time resolution, their ability to resolve the parameters of multi-exponential decay functions, their multi-wavelength capability, optical sectioning capability, and compatibility with different imaging and microscopy techniques . Here, we will focus on FLIM by time-correlated single-photon counting and laser scanning, a technique that delivers an outstanding combination of photon efficiency, temporal resolution, and spatial resolution .
Principle of TCSPC FLIM
Classic TCSPC excites a sample with a pulsed laser, detects single photons of the fluorescence light, and builds up a photon distribution (or a histogram) of the photon density over the time from the laser pulse to the photon detection. The technique is known since 1961 , a comprehensive overview can be found in . The limitation of the classic technique is that it is intrinsically one-dimensional. It does not directly deliver images, and it cannot be combined with fast scanning, as it is used in modern laser scanning microscopes .
The problem of classic TCSPC has been solved by a multi-dimensional TCSPC introduced by Becker & Hickl in 1993. Here, the recording process builds up a photon distribution not only over the time after the excitation pulse but also over other parameters, such as the position of the laser beam in a scan area in the moment of the photon detection, the wavelength of the photons, or the time from the start of an experiment [6, 10, 11]. The application of multi-dimensional TCSPC to laser scanning microscopy is illustrated in Fig. 1.
The scan head of the microscope scans the sample with the focused beam of a high-frequency pulsed laser. For every detected photon, the TCSPC device determines the time, t, in the laser pulse period, and the location of the laser spot, x, y, in the scan area. From these parameters, a photon distribution over the spatial coordinates, x, y, and the times of the photons, t, after the laser pulses is built up. The recording process is continued over a large number of frames, until the desired signal-to-noise ratio of the photon distribution has been achieved. Please see  for more technical details.
Among all electronic FLIM techniques, multidimensional TCSPC delivers the highest time resolution. It also delivers the best lifetime accuracy, or photon efficiency, for a given number of photons detected from the sample [4, 5]. TCSPC FLIM has a number of other features important to lifetime imaging of biological systems: It is able to resolve complex decay profiles, and it is tolerant to dynamic changes in the fluorescence decay parameters during the acquisition [5, 11]. Moreover, TCSPC FLIM is perfectly compatible with confocal and multiphoton [9, 12] laser scanning microscopes. It has no problems with the fast scan rate used in these systems: The recording process is just continued over as many frames of the scan as necessary to obtain the desired signal-to-noise ratio. Moreover, TCSPC FLIM takes advantage of the optical sectioning capability of confocal or multiphoton scanning: The data are obtained from accurately defined lateral positions and from an accurately defined plane within the sample, without contamination by lateral scattering and out-of focus fluorescence .
Fig. 2 gives an example of the data quality that can be reached by TCSPC FLIM. The image has 2048 x 2048 spatial pixels, and the fluorescence decay functions in the individual pixels are resolved into 256 time channels.
FLIM Applications in Biology
Measurement of molecular environment parameters
There is a wide range of fluorophores, also called "sensors’ or "probes", that change their fluorescence lifetime depending on the local molecular environment . Commonly known ones are sensors for ion concentrations, such as Oregon Green Bapta for Ca 2+ or MQAE for Cl-. Other sensors change their lifetimes with the pH, or on binding to RND and DNA. There are also effects of local viscosity, aggregation of the fluorophore, and electron transfer [1–3, 6, 14]. An example for Ca 2+ measurement is shown in Fig. 3.
The advantage of FLIM over intensity-based techniques in all these cases is that the results are independent of the (variable) concentration of the fluorophore and of absorption variations within the sample.
The most widespread application of TCSPC FLIM is the measurement of protein interaction and protein folding by Fцrster resonance energy transfer (FRET) [15, 16]. The proteins are labelled with two dyes of different absorption and emission spectra. The emission band of the first dye, the donor, overlaps the absorption band of a second one, the acceptor. If the distance between donor and acceptor is smaller than a few nanometers energy can transfer directly from the donor to the acceptor. The result is a decrease in the fluorescence decay time of the donor. The intensity of the energy transfer, i. e. the decrease in the decay time, is an indicator of the distance between the donor and the acceptor.
The use of FLIM for FRET experiments has the obvious benefit that the FRET intensity is obtained from a single lifetime image of the donor. Acceptor images are not needed, thus donor bleedthrough into the acceptor channel and directly excited acceptor fluorescence have no influence on FLIM-FRET measurements. The only reference value needed is the donor lifetime in absence of the acceptor [17–20]. Even this can often be derived from a double-exponential analysis of the donor decay function .
An example of a FLIM FRET result is shown in Fig. 4. The data were obtained from a cell expressing a GFP fusion protein. A Cy3-labelled antibody was used as an acceptor for FRET.
An enormous amount of FLIM-FRET papers has been published in the last few years, most of them using TCSPC FLIM. Please see  for a review of the literature and for further details.
Biological tissue contains a wide range of endogenous fluorophores. The lifetimes depend on local environment parameters, such oxygen saturation, binding to proteins, and importantly, on the metabolic state of the tissue [14, 21–23]. The FLIM data therefore contain direct biological information . Additional information about the constitution of the tissue can be obtained from second-harmonic generation (SHG) signals [24, 25]. An important point is that autofluorescence imaging does not use exogenous labels. It can therefore be directly used in clinical applications.
Autofluorescence images of biological tissue can be surprisingly rich in detail, see Fig. 5. The images show a pig skin sample exited by two-photon excitation at 800 nm. The left image shows the wavelength channel below 480 nm. This channel contains both fluorescence and SHG signals. The SHG fraction of the signal has been extracted from the FLIM data and displayed by colour. The right image is from the channel >480 nm. It contains only fluorescence, the colour corresponds to the amplitude-weighted mean lifetime of a double-exponential decay model.
Clinical FLIM Applications
Multiphoton tomography of human skin uses laser scanning by a focused femtosecond laser beam, two-photon excitation, and non-descanned detection of the fluorescence signals [14, 26]. The technique goes back to the work of Gratton, Kцnig, Masters, So and Tromberg who showed that in-vivo two-photon autofluorescence imaging of cells and, especially, human skin, is possible without impairing the viability [25, 27–29]. Instruments for clinical application of the technique have been developed by Jenlab GmbH, Jena, Germany . Because the technique is based on fast scanning and pulsed excitation it favourably combines with TCSPC FLIM. Fig. 6 shows the stratum granulosum of a human patient recorded with a Jenlab "Dermainspect" system and a Becker & Hickl SPC-152 TCSPC FLIM system.
Ophthalmic FLIM uses a combination of an ophthalmic scanner with one or two ps diode lasers and a TCSPC FLIM system. For technical details please see [6, 11]. Ophthalmic FLIM is currently in the state of clinical trials [31, 32]. Two typical results are shown in Fig. 7. The images were scanned by a FLIO lifetime imaging laser ophthalmoscope of Heidelberg Engineering, Heidelberg, Germany. The detection part has two spectral channels detecting from 490 nm to 560 nm and 560 to 700 nm. The signals are detected by Becker & Hickl HPM-100 hybrid detectors , and recorded by Becker & Hickl SPC-150 TCSPC FLIM modules. The images shown are from the wavelength channel from 560 nm to 700 nm.
TCSPC-FLIM of other organs can certainly be obtained by scanning through endocsopes. The optical principle has been demonstrated with good results . The problem is currently the lack of clinically approved endoscopes with high numerical aperture and low intrinsic fluorescence.
In the last 10 years, FLIM techniques have made impressive progress. Conventional PMTs have been replaced with single-photon avalanche photodiodes and hybrid detectors. These detectors have considerably higher detection efficiency than conventional PMTs. Hybrid detectors also deliver cleaner signals . Thus, they not only detect more photons, they allow the FLIM system to achieve a higher lifetime accuracy for a given number of photons per pixel. In the last few years, speed and memory size of computers has increased by more than an order of magnitude. 64-bit operating systems and 64-bit instrument software have increased the available memory size. As a result, FLIM data can be recorded at mega-pixel resolution [11, 34], as shown in Fig. 8.
Increased efficiency helps avoid artefacts induced by photobleaching, photodamage or photo-induced metabolic changes in the samples. In combination with the large memory space of 64-bit Windows systems, TCSPC FLIM can be extended with additional parameters of the photons or the experiment. One example is multi-wavelength FLIM [5, 12]. A spectrum of the fluorescence light is spread over an array of detector channels. For every photon, the time in the laser pulse period, the channel number in the detector array, and the position, x, and y, of the laser spot in the scan area are determined. These pieces of information are used to build up a photon distribution over the arrival times of the photons in the fluorescence decay, the wavelength, and the coordinates of the image. The result is that several images (usually 16) of different wavelength are recorded simultaneously in a single TCSPC channel. A result is shown in Fig. 8. Please see also [22, 35, 36].
Another way to add additional dimensions to the FLIM photon distribution is "Mosaic FLIM". Mosaic FLIM records the data of subsequent FLIM recordings into subsequent elements of a large data array. The technique has originally been developed to record spatial mosaic data by sample stepping . It can, however, be used also to record Z stacks of FLIM data and to record fast time series [6, 11]. An example is shown in Fig. 9. The mosaic has 64 elements, each recorded with an acquisition time of one second. The sample was a moss leaf, the microscope a bh DCS-120 confocal FLIM system. The time runs from lower left to upper right. The decrease in the fluorescence lifetime by the non-photochemical chlorophyll transient is clearly visible.
By using periodic stimulation of the sample, the technique is able to resolve changes in the Ca 2+ concentration in live neurons at a resolution of 40ms [6, 11]. An even faster technique, called Fluorescence Lifetime-Transient Scanning (FLITS), is based on TCSPC and line scanning. FLITS has been shown to record dynamic fluorescence-lifetime effects at a resolution of about 1 ms [6, 11, 38].
There is currently an increasing interest in performing FLIM at near-infrared wavelengths. In the NIR, the emission of exogenous fluorophores can be detected without contamination from autofluorescence. Moreover, the fluorescence decay signature of NIR fluorophores is important to diffuse optical imaging techniques [5, 6, 11]. An example of NIR FLIM with a Zeiss LSM 710 NLO laser scanning microscope is shown in Fig. 10. Technical details of NIR FLIM are described in  and .
TCSPC FLIM can been combined with STED (stimulated emission-depletion) microscopy [16, 32] and with NSOM (near-field scanning optical microscopy) [8, 35] to obtain fluorescence lifetime images with optical super-resolution. An example of a STED-FLIM recording is shown in Fig. 11.
TCSPC FLIM is able to simultaneously record fluorescence (FLIM) and phosphorescence lifetime images (PLIM). The technique is based on on-off modulating a high-frequency pulsed laser, and assigning two times to the individual photons. One is the time from the previous excitation pulse, the other a time from the modulation pulse [11, 42].
An example is shown in Fig. 12. It shows yeast cells stained with tris (2,2’-bipyridyl) dichlororuthenium (II) hexahydrate. On the ps time scale, autofluorescence from NADH and FAD is detected. The phosphorescence of the ruthenium dye is detected on the microsecond time scale. The fluorescence lifetime image is shown on the left, the phosphorescence lifetime image on the right.
Instruments based on a combination of confocal or two-photon laser-scanning microscopes with TCSPC are, in principle, able to record also fluorescence correlation (FCS) data and single-molecule FRET data [6, 11]. An example of FCS recording is shown in Fig. 13.
The combination of multi-dimensional TCSPC and laser scanning microscopy records fluorescence lifetime images at near-ideal photon efficiency and excellent temporal and spatial resolution. The technique can be extended to record multi-wavelength images, lateral mosaics and Z-stacks of FLIM images, and fast time series showing dynamic changes in the fluorescence behaviour of a sample. Moreover, fluorescence lifetime images can be recorded simultaneously with phosphorescence lifetime images. TCSPC FLIM can also be combined with STED, resulting in FLIM images with optical super-resolution. Typical FLIM applications are mapping of local molecular environment parameters, protein interaction experiments by FRET techniques, and autofluorescence imaging. Clinical applications are at the stage of clinical trials.
 Статья и перевод на английский язык предоставлены компанией ЗАО "НТК"АЗИМУТ ФОТОНИКС"