Introduction

All photosynthetic organisms contain thylakoid membranes to convert solar energy into high-energy molecules such as ATP and NADPH which are used to fix carbon in energy-rich molecules that are used for reproduction, growth, and maintenance. The thylakoids contain the membrane-protein complexes called photosystem I (PSI), photosystem II (PSII), cytochrome b6/f, and F-ATPase, which are the major players in oxygenic photosynthesis (Dekker and Boekema 2005; Moore et al. 1998; Nelson and Ben-Shem 2004). Both PSI and PSII contain a reaction center which is surrounded by a large “antenna”, which consists of light-harvesting pigment–protein complexes. The chlorophylls (Chls) and other pigments in the antenna harvest light and transport a large part of the corresponding energy to the reaction center in which charge separation takes place. In most plants and some green algae, the thylakoid membrane is differentiated into grana stacks and stroma lamellae (Fig. 1) (Anderson 1999; Dekker and Boekema 2005; Mustárdy and Garab 2003). Other classes of photosynthetic organisms have their own unique membrane stacking which is considerably different from that of higher plants (Gunning and Schwartz 1999). The dominant antenna species of PSII in higher plants is light-harvesting complex II (LHCII) which is not only important for ‘"harvesting light" (van Amerongen and van Grondelle 2001), but also plays a role in nonphotochemical quenching (Pascal et al. 2005; Ruban et al. 2007), while it is, in addition, essential for grana stacking (Lambrev et al. 2007). PSI contains a large part that sticks out of the membrane and does not fit into the inner stacks of the grana. This leads to a separation of the two photosystems (Fig. 1) (Dekker and Boekema 2005). This separation is thought to allow the regulation of ATP production, by balancing the linear and cyclic electron transport (Berry and Rumberg 1996; Joliot et al. 2004) and to avoid ‘spill-over’.

Fig. 1
figure 1

Schematic model of the thylakoid membrane. The margins are the strongly curved membranes, the end membranes are located at the bottom and the top of the grana stack and the stroma lamella is the non-stacked region

Full size image

In higher plants about ~85% of PSII is located in the grana and about ~15% is present in the stroma lamellae (Fig. 1), while for PSI these numbers are approximately 35 and 65%, respectively (Albertsson and Andreasson 2004). These percentages are not fixed but can differ between plant species while they also depend on growth conditions. However, the relative proportion of stroma lamellae and grana is rather constant (Albertsson and Andreasson 2004). The opposite is true for the number of layers in a single granum. Plants such as Alocasia that are grown in low-light intensities can have more than 50 layers in one granum, which can extend across the whole chloroplast (Goodchild et al. 1972), whereas most other plants have only ~10 till 20 layers. The diameter of the disc layer in the grana is more or less constant across plant species (300–600 nm) (Dekker and Boekema 2005; Mustárdy and Garab 2003). With the decrease of light intensity, the amount of LHCII per PSII increases, while the PSII/PSI ratio decreases. The PSII/PSI reaction centers (RCs) ratio for Alocasia, grown under low-light conditions of 10 μmol photons m−2 s−1 is 1.43 (Chow et al. 1988). In this study, the same low-low light growing conditions are used (see Materials and Methods). The Alocasia plant was used in many chloroplast visualization studies because of its giant grana stacks (Anderson 1999; Chow et al. 1988; Goodchild et al. 1972).

The best noninvasive optical imaging technique for measuring photosynthetic systems in leaves is multiphoton fluorescence microscopy, because it allows imaging up to a depth of 500 μm in living plant tissue (Williams et al. 2001; Zipfel et al. 2003). The leaves of Arabidopsis thaliana and Alocasia wentii are 200 and 300 μm thick, respectively, and in principle, suitable for complete scanning by FLIM with two-photon excitation (TPE) at 860 nm. In contrast, one-photon excitation (OPE) microscopy only allows imaging up to a depth of ~100 μm (Cheong et al. 1990; Williams et al. 2001). Two-photon (nonlinear) microscopy depends on the simultaneous interaction of two photons with a molecule, resulting in a quadratic dependence of light absorption on light intensity as opposed to the linear dependence of one-photon fluorescence microscopy. For pigment molecules such as chlorophylls (Chl) and carotenoids (Car),the two-photon absorption spectra, which are only partly known, are significantly different from their one-photon counterparts, but the emission spectra are in general identical (Xu et al. 1996). For LHCII, the TPE spectrum was measured in the region from 1,000 to 1,600 nm, ‘"corresponding"’ to one-photon wavelengths of 500–800 nm (Walla et al. 2000).

This study combines microscopy with fluorescence lifetime measurements to investigate to which extent it is possible to study the primary steps in photosynthesis in living tissue and to determine at which spatial and time resolution this is possible. The final goal is to study these primary events in vivo under a variety of (stress) conditions.

In this study, the two-photon absorption of 860 nm light is used for excitation. The instrument response function (IRF) of the FLIM setup is 25 ps (van Oort et al. 2008). Because carotenoids and Chl b transfer most of their excitation energy to Chl a in less than 1 ps (Croce et al. 2001, 2003; Eads et al. 1989; Gradinaru et al. 2000; Peterman et al. 1997; van Amerongen and van Grondelle 2001; Visser et al. 1996) only fluorescence from Chl a is observed (Broess et al. 2008). We focus on the detection of fluorescence lifetimes of Chl in PSI and PSII in intact leaves, both under low-light conditions and under conditions in which the PSII reaction centers are closed by DCMU. The shade plant Alocasia wentii is used to make it easier to try to spatially resolve the different photosystems, because it is known to form giant grana stacks (Albertsson and Andreasson 2004; Goodchild et al. 1972). In addition, Arabidopsis thaliana is studied because it is widely used as one of the model organisms in plant sciences.

Materials and methods

Fluorescence lifetime imaging microscopy

Multiphoton imaging was performed on a multiphoton dedicated Biorad Radiance 2100 MP system, coupled to a Nikon TE300 inverted microscope (Borst et al. 2003). A tunable Ti-Sapphire laser (Coherent Mira) was used as an excitation source which was pumped with a 5-Watt (Coherent) Verdi laser, resulting in excitation pulses of ~200 fs at a repetition rate of 76 MHz. In the beam-conditioning unit (BCU), the excitation power was tuned by a pockell cell. The laser beam was collimated in the scanhead and focused by a Nikon 60x water immersion Apochromat objective lens (NA 1.2) into the sample. The fluorescence was detected by non-descanned direct detectors (NDDs), which were coupled to the sideport of the microscope. Using this type of detection, the loss of fluorescence light was reduced, and 3–5 times more signal was obtained compared to internal detectors. The emission light was split into two channels using a dichroic mirror filter wheel. FLIM measurements were performed by directing the fluorescence via a secondary dichroic (770DCXR, Chroma Technology Corp.) into a Hamamatsu R3809U photomultiplier, operated at 3.1 kV. Fluorescence was selected using a dichroic (FF 495—DiO2, Semrock) and 2x a bandpass filter (HQ700/75, Chroma Technology Corp). In the excitation branch, a longpass filter (RG 780 3 mm, Schott) was used for reduction of the excitation light.

The multichannel-plate photomultiplier allows single photon detection at high time-resolution, with an IRF of 25 ps (van Oort et al. 2008, 2009). The output of the detector was coupled to a Becker & Hickl single-photon-counting module (SPC 830) (Becker and Bergmann 2002). The signal from the Hamamatsu triggers the START of the time ramping for the time-correlated single-photon-counting (TCSPC). The pulses from the Ti-Sapphire laser serve as the SYNC signal to stop the time ramping and allowing the timing of the arrival of the fluorescent photons. The time window (ADC) was set to 1,024 channels and typically fluorescence was recorded for 2 min at a photon count rate of approximately 20 kHz. The signal from the PMT is combined with the pixel clock and line predivider signals from the Biorad scanhead to create 2D lifetime images. Fluorescence decay curves were fitted to a sum of N exponentials Σaiexp(−τ/τ i ) (i runs from 1 to N), convoluted with the IRF (Digris et al. 1999, van Oort et al. 2008, 2009), which was determined from the decay of pinacyanol iodide in methanol. From these results, an average lifetime <τ> was also calculated, according to <τ> = Σa i τ i . The number of counts in the peak channels is ~100 in the fluorescence intensities images and traces.

Growth condition and sample preparation

Arabidopsis was grown in growth chambers of 23°C with alternating 16 h of light at a light intensity of 60–100 μE m−2 s−1 and 8 h of darkness. Note that the light intensity is relatively low, which may lead to an altered antenna composition as compared to that of plants grown under high-light conditions. Alocasia was grown at room temperature with alternating 16 h of light at a light intensity of 10–15 μE m−2 s−1 and 8 h of darkness. For closing the reaction centers of PSII in leaves, vacuum infiltration was performed with 0.1 mM DCMU, 20 mM Hepes, 5 mM NaCl, and 5 mM MgCl2 buffer with pH 7.5.

Isolation of chloroplasts: Alocasia wentii leaves were ground in semi-frozen buffer 1 (0.45 M sorbitol, 20 mM Tricine, 10 mM EDTA, 10 mM NaHCO3, and 0.1% BSA, pH 8.4) using a blender for 5-s, and then filtrated through eight layers of cheesecloth and centrifuged at 3,000 × g for 20 s at 4°C. The supernatant was discarded and the pellet washed with buffer 2 (0.3 M Sorbitol, 20 mM Tricine, 5 mM MgCl2, and 2.1 mM EDTA, pH 7.4). The collected resuspended chloroplasts were put on 50% Percoll/50% buffer 3 (0.6 M Sorbitol, 20 mM Tricine, and 5 mM MgCl2, pH 7.6) and centrifuged at 3,500 × g for 10 min at 4°C. The supernatant was disposed, and the pellet was diluted in buffer 2 before measuring.

Results and discussion

It has been demonstrated that FLIM can be a noninvasive tool (Holub et al. 2000; Lukins et al. 2005) for measuring Chl a fluorescence lifetimes in plants and algae which can then be correlated to the response of the photosynthetic apparatus to, for instance, the effect of dehydration. However, measurements so far have only been performed under high-light conditions at the maximum fluorescence level (FM) in which PSII reaction centers are closed and average lifetimes were found to be 1.7 ns ± 0.2 ns (Holub et al. 2000) and 611 ps (Lukins et al. 2005), indeed corresponding to lifetimes of PSII with (partially) closed reaction centers.

With the FLIM setup used in the present study it is possible to measure under low-light conditions. In Fig. 2 two images with 1,024 pixels are presented, showing Alocasia wentii chloroplasts. The fluorescence images illustrated in the figure are intensity-based, whereas the fluorescence kinetics has been obtained for each pixel and has been analyzed with a combination of SPCimage2.3 software (Becker & Hickl) and home-built software using a exponential decay model (Digris et al. 1999; Mullen et al. 2007; Novikov et al. 1999). The fitted chloroplast fluorescence lifetimes and amplitudes averaged over all the pixels of Fig. 2b are as follows: τ 1 = 59.5 ps (44.1%), τ 2 = 205 ps (35.3%) and τ 3 = 588 ps (20.6%). Without further experiments and knowledge, it is not possible to assign the lifetimes to processes in PSI and PSII. The amplitudes are expected to depend strongly on the excitation and detection wavelength. A complicating factor at this stage is the fact that the two-photon absorption spectra of all the relevant pigments are not known. Scanning at different depths in the leaf does not result in different average lifetimes or different amplitudes. Figure 2a and b are intensity-based images with a linear gray scale. Pixels with zero fluorescence counts are dark and pixels with maximum fluorescence are white. The chloroplasts in Fig. 2b appear to be heterogeneous, and small white dots can be observed within the chloroplast. Similar heterogeneity was observed earlier in microscope measurements (Anderson 1999; Spencer and Wildman 1962; van Spronsen et al. 1989), and it is likely that the white spots correspond to the grana stacks. The Chl concentration is higher in the grana and moreover, as they contain mainly PSII, which leads to more fluorescence than PSI because of the longer fluorescence lifetimes.

Fig. 2
figure 2

Room temperature fluorescence intensity-based image (1,024 pixels) with a linear gray scale as measured with FLIM. The chloroplasts in Alacosia wentii leaves are excited with TPE at 860 nm and detected with a bandpass filter centered at 700 nm and with a bandwidth of 75 nm. For each pixel a fluorescence decay trace is measured. The average lifetimes and amplitudes in the 1,024 pixels in this image are: τ 1 59.5 ps (44.1%), τ 2 205 ps (35.3%) and τ 3 588 ps (20.6%)

Full size image

It is known from TPE FLIM measurements on LHCII aggregates (Barzda et al. 2001a) that the pulse repetition rates of more than 1 MHz can lead to the shortening of fluorescence lifetimes of photosynthetic systems because of excitation quenching by Car and Chl triplets (singlet–triplet annihilation). Moreover, singlet–singlet annihilation can occur, also leading to a shortening of the lifetime (Barzda et al. 2001b). Since the number of triplets formed is expected to increase on increasing the number of excitations, the fluorescence lifetimes have been measured as a function of laser intensity. In Fig. 3, three decay traces are presented, obtained at 150, 330, and 1600 μW (laser power measured directly at the sample holder of the setup). The 150 and 330 μW decay traces are identical after normalization at the top, whereas the 1,600 μW trace is substantially faster. It should be noted that the initial number of excitations for TPE scales quadratically with the light intensity, and thus the number of excitations increases by a factor of 4.8 when going from 150 to 330 μW. Therefore, the results clearly demonstrate the absence of singlet–triplet (and singlet-singlet) annihilation at relatively low intensities. Using extremely high powers of 1,600 μW, the RCs are being closed, but the kinetics are faster, which is ascribed to a combination of singlet–singlet and singlet–triplet annihilation.

Fig. 3
figure 3

Room temperature fluorescence decay traces (measured with FLIM) of chloroplasts in Arabidopsis thaliana leaves. The chloroplasts are excited with TPE at 860 nm and detected with a bandpass filter centered at 700 nm with a bandwidth of 75 nm. Identical traces are observed for chloroplasts with laser powers of 150 μW (black squares) and 330 μW (red open circles) and correspond to PSII with open reaction centers. The green open triangles represent a decay trace obtained with a laser power of 1,600 μW which is faster than the black and red decay traces due to annihilation processes. The number of counts in the peak channels are 28, 156, and 2028, respectively

Full size image

The fluorescence decay traces of isolated chloroplasts have also been measured with FLIM and are compared to those of leaf tissue (Fig. 4). The in vivo fluorescence kinetics of chloroplasts are similar to those of the isolated chloroplasts for the first 170-ps part of the trace. There is a small discrepancy in the middle part of the trace, but overall the traces are nearly identical. The chloroplasts were isolated with percoll and are smaller in size (not shown) than the chloroplasts in leaves.

Fig. 4
figure 4

Room temperature fluorescence decay traces (measured with FLIM). The chloroplasts in Alocasia wentii are excited with TPE at 860 nm and detected with a bandpass filter centered at 700 nm with a bandwidth of 75 nm. Round open circles are isolated chloroplasts (in vitro) with an average lifetime of 180 ps. Black squares correspond to chloroplasts in leaves (in vivo) with an average lifetime of 212 ps

Full size image

In order to try to distinguish between PSI and PSII in the microscopic images, the difference in fluorescence lifetimes between the two photosystems has been increased by closing the reaction centers of PSII by vacuum infiltration of Arabidopsis thaliana with 0.1 mM DCMU in 20 mM Hepes, 5 mM NaCl, and 5 mM MgCl2 buffer with pH 7.5. The average lifetime for the leaf infiltrated with DCMU is 1.3 ns (Fig. 5) whereas for "normal" leaves the average lifetime is 290 ps. Both photosystems are separated in space and have substantially different lifetimes in the presence of DCMU (Lukins et al. 2005; Pfündel 1998; Zucchelli et al. 1992) because the average lifetime of PSI with antenna complexes is reported to be ~60 ps (Croce et al. 2000; van Oort et al. 2008) and that of closed PSII is ~1.5 ns (Zucchelli et al. 1992). This is visible in the traces and images of the chloroplasts of Alocasia wentii in Fig. 6. The expectation is that pixels with more grana stacks have a higher intensity compared to pixels with relatively more stroma lamellae (Spencer and Wildman 1962). In Fig. 6a, the fluorescence kinetics of 10 high-intensity pixels (white) are compared with those of 10 low-intensity pixels (grey). The 10 high- and low-intensity pixels have 623 (266,342) and 541 (195,833) counts in the peak (and total number of fluorescence counts), respectively. The global fitting results with linked lifetimes and independent amplitudes are τ 1 = 116 ps (53.3, 59.6%), τ 2 = 1,027 ps (35.1, 29.5%), and τ 3 = 3,957 ps (11.6, 10.9%). The first amplitude for each lifetime refers to the high-intensity pixels and the second amplitude, to the low-intensity pixels. The first lifetime of 116 ps probably reflects a mixture of PSI and open PSII reaction centers (Broess et al. 2006, 2008) because Alocasia grown under low-light conditions have a PSII/PSI RCs ratio of 1.43 (Chow et al. 1988). This corresponds to 59% Chl of PSII and 41% Chl of PSI. If all the PSIIs are closed, one might expect 59% Chl contribution of slow lifetimes and 41% of fast lifetime. The amplitudes of the lifetime of 116 ps for both groups of pixels is more than 41%, so the conclusion should be such that not all the PSII reaction centers are closed by the DCMU. The two slow lifetimes of ~1 and ~4 ns must correspond to closed PSII reaction centers because these lifetimes are absent for open RCs. The 6.3% difference in the amplitude of the slow lifetimes for the high- and low-intensity pixels is probably caused by the fact that the high-intensity pixels comprise more PSII than PSI. This is expected because the grana, where PSII is concentrated, have a higher chlorophyll concentration per pixel than the stroma lamellae. There are two straightforward explanations for the lifetime differences in the pixel groups: (i) The DCMU buffer is not penetrated evenly in every part of the chloroplasts which results in different lifetimes and intensities for each pixel; (ii) In one pixel group, there are more grana than in the other pixel group which will also result in different lifetimes and intensities for each pixel. In Fig. 6b, the intensity of the different pixels seems to have a random distribution in the chloroplast, which is not expected as a result of varying penetration of the DCMU buffer. The differences in lifetimes for the two pixel groups can thus better be explained by pixels with more or less grana. It should be kept in mind that the model that is used here (PSI and PSII fluorescence kinetics are both homogeneous) is oversimplified, for instance, because of the action of the PSII repair cycle and the presence of PSII heterogeneity. In conclusion, it appears to be very difficult to distinguish between regions with more or less grana.

Fig. 5
figure 5

Room temperature fluorescence decay traces (measured with FLIM). The chloroplasts in Arabidopsis thaliana leaves are excited with TPE at 860 nm and are detected with a bandpass filter centered at 700 nm with a bandwidth of 75 nm. Black squares represent a "normal" fluorescence decay trace of chloroplasts in an Arabidopsis leaf with an average lifetime of 290 ps. Round open circles represent a fluorescence decay trace of a vacuum infiltrated leaf with a 0.1 mM DCMU buffer with an average lifetime of 1.3 ns

Full size image
Fig. 6
figure 6

a Room temperature fluorescence decay traces (measured with FLIM) of chloroplasts in Alocasia wentii leaves excited with TPE at 860 nm detected with a bandpass filter centered at 700 nm with a bandwidth of 75 nm. The leaves are vacuum infiltrated with a 0.1 mM DCMU buffer for closing the PSII reaction centers. The black (1) trace with its fit corresponds to the summed fluorescence decay of 10 white (high) pixels from the chloroplast in the intensity-based image in Fig. 6b. The blue (2) trace with its fit corresponds to the summed fluorescence decay of 10 grey (low) pixels from the chloroplast in the intensity-based image in Fig. 6b. b (right) Room temperature fluorescence intensity-based image (measured with FLIM). The chloroplast in Alacosia wentii leaves are excited with TPE at 860 nm and detected with a bandpass filter centered at 700 nm with a bandwidth of 75 nm. The pixel size is 0.26 μm. Fluorescence in each pixel is detected in 4,096 channels with a time resolution of 3 ps per channel. (left) Global fitting with linked lifetimes (τ1, τ2, and τ3) and independent amplitudes for the black trace (1) and blue trace (2) shown in Fig. 6a

Full size image

For Arabidopsis thaliana leaves, it appears to be not at all possible to resolve variations in lifetimes between pixels, which is probably due to the fact that for Arabidopsis thaliana, the grana stacks are smaller than for Alocasia wentii.

Conclusions

  • In vivo measurements on chloroplasts are possible under low-light conditions with TPE FLIM and the obtained fluorescence kinetics are very similar to those observed in in vitro measurements on isolated chloroplasts.

  • While scanning through the leaves of Alocasia wentii and Arabidopsis thaliana that were both grown under low-light conditions, no differences could be observed in the fluorescence kinetics, indicating no variation in the chloroplast composition/organization as a function of depth.

  • The spatial resolution of the FLIM measurements does not allow to observe individual grana stacks in Arabidopsis thaliana chloroplasts, but in the case of chloroplasts of Alocasia wentii variations in the lifetimes are observed, which may be ascribed to variations in the grana density.

  • In the future, the TPE fluorescence lifetime imaging microscope can be used to study individual chloroplasts in leaves under different stress conditions.