Two-Photon FLIM of Pollen Grains Reveals Ultra-Fast Decay Component

Using our DCS-120 MP multiphoton FLIM system with ultra-fast hybrid detectors, we find extremely fast fluorescence-decay components in a wide variety of biological material. Here, we report on FLIM of pollen grains from a number of different plants. We found decay components with lifetimes, t₁, from 10 ps to 80 ps, and with amplitudes, a₁, as large as 93 %.

Keywords: FLIM, Ultra-Fast Fluorescence Decay, Pollen, Multiesponential Decay Analysis

Related Applications

TCSPC FLIM

Related Application Notes

Download PDF

Two-Photon FLIM of Pollen Grains Reveals Ultra-Fast Decay Component

Wolfgang Becker, Taravat Saeb-Gilani, Cornelia Junghans, Becker & Hickl GmbH

Abstract: Using our DCS-120 MP multiphoton FLIM system with ultra-fast detectors, we find extremely fast fluorescence-decay components in a wide variety of biological material. Here, we report on FLIM of pollen grains from a number of different plants. We found decay components with lifetimes, t₁, from 10 ps to 80 ps, and with amplitudes, a₁, as large as 93 %.

Experiment Setup

From earlier experiments we know that fluorescence lifetimes of biological material can be in the sub-100-ps range. FLIM of mushroom spores revealed decay components with lifetimes down to less than 10 ps [2], and FLIM of mammalian hair shows components in the range below 20 ps. The fast decay components are not only clearly detectable, they are actually the dominating constituent of the decay functions. Normal one-photon confocal systems with diode-laser excitation then show a short apparent lifetime but do not reliably resolve the decay functions into individual decay components. Our DCS-120 MP multiphoton system, in contrast, is able to show the fast decay components directly. The instrument is based on fast beam scanning, two-photon excitation with a femtosecond fibre laser (Toptica Femto Fibre Pro, 780 nm) and detection by ultra-fast hybrid detectors (bh HPM-100-06) [3, 4, 5]. The laser is coupled free-beam into the scan head, the fluorescence is transferred to the detectors via a non-descanned beam path. The data are recorded by bh SPC-150 NX TCSPC/FLIM modules using bhs multi-dimensional TCSPC technique [6]. Please see [1, 3] for details. The instrument response of the DCS-120 MP system with the HPM-100-06 detectors is about 18 ps, full width at half maximum [4]. This is about 5 times faster than for a typical diode-laser based system with GaAsP hybrid detectors, and 10 times faster than for FLIM systems with conventional PMT detectors. Fast decay components thus become directly visible in the decay curves, without the need of deconvolution from an IRF wider than the decay time.

Sample Preparation and FLIM Procedure

Anthers were collected from fresh flowers and placed on 180 µm thick microscope slides. Pollen released from the anthers were collected on the slides, see Fig. 1. In some cases, especially when we were not able to collect enough pollen, the entire anthers were left on the slides, and embedded in immersion oil. The slides were then placed on the sample stage of an inverted microscope, and FLIM recordings were taken.

Fig. 1: Left: Stamens of Hippeastrum with anthers carrying pollen (from www.wikipedia.org). Right: Anthers on a microscope slide to collect pollen.

For FLIM measurement we used an oil immersion lens with NA = 1.3 numerical aperture, and 40x magnification. The emitted photons were recorded via the non-descanned detection path of the DCS-120 MP system. A Chroma SP 700 short-pass filter was used to reject scattered excitation light, and a 400-nm long pass filter to suppress possible SHG light. Consequently, fluorescence was detected from about 400 nm up to the upper detection limit of the detector, which is about 650 nm.

We admit that, from the point of optics, the configuration of the sample is not entirely correct. For best resolution, the pollen should be embedded in a medium with a refractive index close to that of glass. However, the usual embedding media cannot be used because they emit too much fluorescence. Attempts to image the pollen grains in water or immersion oil failed because the pollen grains were moving in the liquid. The best and only way to run the experiments was to image the bare pollen on the glass from the back of the slide, as described above.

Another problem is that the pollen, especially those of brown or black colour, are likely to absorb at the fundamental wavelength of the excitation laser. They are therefore easily destroyed by the laser. To avoid any sample destruction, we kept the laser power below 4 mW, in some cases even below 2 mW. Due to the low laser power the photon detection rate was no higher than 100,000 to 300,000 counts per second. Consequently, the data acquisition times were relatively long, typically in the range from one to five minutes. The long acquisition time was no problem, however, because our system has excellent timing stability and a detector background count rate of no more than 60 counts per second [6].

FLIM Measurements

We recorded FLIM data from the pollen of different flowers, see Fig. 2. The data were processed by SPCImage NG FLIM data analysis software using a triple-exponential decay model [7]. The results are shown in Fig. 3. Left to right, the figure shows images of the amplitude-weighted mean lifetime, tm, images of the fast decay component, t1, and images of the amplitude of the fast component, a1. A decay curve in a representative spot of the image is shown far right. The order of the images was chosen by the tint of the pollen; pollen with bright colour are displayed on the top, pollen with dark colour at the bottom.

Fig. 2: Flowers from the pollen of which FLIM data were recorded. Left to right: Three different brands of Lilium orientalis, Papaver rhoeas.

As can be seen from the inserts in the decay panels the lifetimes of the fast component, t1, range from 50 ps down to 10 ps. The amplitudes of the fast component range from about 80% to more than 90%. However, different than in earlier results from mushroom spores, there is no correlation of these parameters with the tint of the pollen grains.

Fig. 3: Top to bottom: Pollen of white Lilium orientalis, pink Lilium orientalis, dark-pink Lilium orientalis, and Papaver rhoeas. Left to right: Lifetime images of the mean (amplitude-weighted) lifetime, tm, the lifetime of the fast decay component, t1, and the amplitude of the fast decay component. Decay curves in a selected spot of the image are shown on the right, decay parameters are shown in the inserts. Triple-exponential decay analyses with bh SPCImage NG.

Fig. 4 and Fig. 5 show lifetime images taken from intact anthers of Rhododendron catawbiense and Ranunkulus repens. A tm image is shown on the left, decay curves in selected spots on the right. Interestingly, the fastest decay time, t1, and the largest amplitude, a1, are not found in the pollen but in the inner tissue of the anther. More than that, we found the fast decay component also in other plant tissue, even in lifetime images take from green leaves.

Fig. 4: tm image from the anther of a Rhododendron catawbiense. Fluorescence decay curves in two selected spots shown on the right.

Fig. 5: tm image from the anther of Ranunkulus repens. Fluorescence decay curves in three selected spots shown on the right.

It could be argued that the fast component may not be real but caused by insufficiently blocked laser light or by second-harmonic generation (SHG) in the tissue. However, an additional laser blocking filter in front of the detector neither changed the signal intensity nor the shape of the decay curves. Similarly, an additional long-pass filter with 420 nm cut-off wavelength did not change the decay curves. Therefore leakage of laser light or SHG can be excluded as a source of the fast decay. This is further confirmed by the decay curves in Fig. 5. There are regions with extremely fast decay (curve at the bottom) but also regions where the decay functions look absolutely 'normal' (curve at the top). This supports the assumption that the fast component is real.

We further checked whether the fast decay component is somehow related to excitation with femtosecond pulses. For that purpose, we recorded FLIM data with a 'normal' bh DCS-120 confocal FLIM system. This system uses diode lasers for excitation. The pulse width is on the order of 40 ps so that nonlinear effects can be excluded. A result is shown in Fig. 6.

Fig. 6: Lilium orientale (white), image recorded by one-photon confocal FLIM with diode-laser excitation. Left: tm image. Right: Decay function of 10x10 pixel area around cursor position. Decay parameters shown in insert.

Although the IRF of the system is much wider than that of the MP system the decay function looks unusually steep and fast. Triple-exponential decay analysis indeed turns up a fast decay component of 39 ps with an amplitude of 67%. The results from the multiphoton results are thus confirmed.

Discussion of the Results

With instrumental artefacts being excluded, the question is where the fast fluorescence components are coming from, and why they are preferentially found in plants. The most likely explanation is that plants contain a wide variety of coloured compounds, such as Carotinoids and Flavonoids [8, 9]. Most of these compounds are not known to be strongly fluorescent. However, exactly the low fluorescence quantum yield may be the explanation of the short lifetime: A molecule which absorbs light but has a dominating non-radiative de-activation pathway will not be entirely free of fluorescence. It will just fluoresce with an extremely short lifetime, of approximately the reciprocal rate constant of the non-radiative pathway. If the sources of the fast signals are indeed Carotinoids and Flavonoids they should be detectable also in pigmented tissues of other organisms. Indeed, fast decay components with lifetimes <80 ps are found in the fundus of mammalian eyes [10] and components with lifetimes <20 ps in mammalian hair. The lower picosecond range of FLIM should therefore no longer be neglected as a source of biological information.