Background: Multifunctional two-photon laser scanning microscopy provides attractive advantages over conventional two-photon laser scanning microscopy. For the first time, simultaneous measurement of the second harmonic generation (SHG) signals in the forward and backward directions and two photon excitation fluorescence were achieved from the deep shade plant Selaginella erythropus.
Results: These measurements show that the S. erythropus leaves produce high SHG signals in both directions and the SHG signals strongly depend on the laser's status of polarization and the orientation of the dipole moment in the molecules that interact with the laser light. The novelty of this work is (1) uncovering the unusual structure of S. erythropus leaves, including diverse chloroplasts, various cell types and micromophology, which are consistent with observations from general electron microscopy; and (2) using the multifunctional two-photon laser scanning microscopy by combining three platforms of laser scanning microscopy, fluorescence microscopy, harmonic generation microscopy and polarizing microscopy for detecting the SHG signals in the forward and backward directions, as well as two photon excitation fluorescence.
Conclusions: With the multifunctional two-photon laser scanning microscopy, one can use noninvasive SHG imaging to reveal the true architecture of the sample, without photodamage or photobleaching, by utilizing the fact that the SHG is known to leave no energy deposition on the interacting matter because of the SHG virtual energy conservation characteristic.
Nonlinear optical effects, such as two-photon (Denk et al., 1990) and three-photon (Wokosin et al., 1996; Maiti et al., 1997; Schrader et al., 1997; Tuer et al., 2008) fluorescence, significantly improve depth resolution and reduce the background noise. Nonlinear optical techniques have been used to develop a new generation of optical microscopes with novel capabilities. These new capabilities include the ability to use near-infrared light to induce absorption and enhance fluorescence from fluorophores that absorb in the ultraviolet region. Other capabilities of nonlinear microscopes include improving spatial and temporal resolution without the use of pinholes or slits for spatial filtering, improving signal strength for deeper penetration into thick and highly scattering tissue and confining photobleaching to the focal volume (Denk et al., 1990). The invention of nonlinear laser microscopy has opened new opportunities for noninvasive examination of the structure and functioning of living cells and tissues (Denk et al., 1990).
Second harmonic imaging microscopy (SHIM) is based on a nonlinear optical effect known as SHG (Barzda et al., 2004; Barzda et al., 2005; Greenhalgh et al., 2006). SHIM has been established as a viable microscope imaging contrast mechanism for visualization of cell and tissue structure and function. SHIM using SHG as a probe is shown to produce high-resolution images of transparent biological specimens (Campagnola & Loew, 2003). A second harmonic microscope obtains contrasts from variations in a specimen's ability to generate second harmonic light from the incident light whereas a conventional optical microscope obtains its contrast by detecting variations in optical density, path length or refractive index of the specimen. SHG requires intense laser light to pass through a material with a noncentrosymmetric molecular structure (Reshak et al., 2009). Second harmonic light emerging from SHG material is exactly half the wavelength (frequency doubled) of the light entering the material (Reshak et al., 2009). The alternative technique, two-photon-excited fluorescence (TPEF) is also a two-photon process. TPEF involves some energy loss during relaxation from the excited state, whereas SHG is energy conserving.
Advances in the developments of SHIM have provided researchers with novel means by which noninvasive visualization of nonbiological and biological specimens can be achieved with high penetration and high spatial resolution, and is known to leave no energy deposition on the interacting matter because of SHIM's virtual energy conservation characteristic (Gao et al., 2006). That is, the emitted SHG photon energy is the same as the total absorbed excitation photon energy. The inhomogeneity inherent to most biological specimen, and in particular, to the internal structure of various cells, leads to high quality SHG images without any preconditioning such as labelling or staining that might induce undesirable effects in the living cell (Reshak, 2009). Historically, resolution in fluorescence optical microscopy has been limited by the Rayleigh criterion. The Rayleigh criterion states that two images are just resolved when the principal maximum (of the Fraunhofer diffraction pattern) of one image coincides with the first minimum of the other (Born & Wolf, 1980). Techniques with better resolution than the Rayleigh criterion have recently been established, among which is harmonic excitation light microscopy (Frohn et al., 2000).
The novelty of this work is the application of these techniques to reveal the structure of plant tissue. In particular, for the first time the deep shade plant Selaginella erythropus is investigated by means of the multifunctional-two-photon laser scanning microscopy (MF-2PLSM), which the first author established by combining three platforms of laser scanning microscopy: fluorescence microscopy, harmonic generation microscopy and polarizing microscopy. MF-2PLSM provides attractive advantages over conventional fluorescence microscopy for revealing the true architecture of the samples that can not produce autofluorescence without labelling or staining, which might induce undesirable effects in the living cell. Reconstruction of complementary images by eliminating the angle dependence of images, when using linear polarized laser, helps maximize the SHG signals and hence improves the brightness and the sharpness of the features in SHG images of samples. This technique will provide biologists and medical researchers another useful visualization tool for exploring the nature of living cells.
The study organism, S. erythropus, is an unusual plant growing in the low light understory of tropical rain forests. A giant chloroplast, termed a bizonoplast, was first discovered in this plant (Sheue et al., 2007). The bizonoplast is characterized by unique dimorphic ultrastructure differentiating the chloroplast into upper and lower zones. However, the leaves (viz. microphyll) of S. erythropus also contain typical chloroplasts. Novel patterns of silica bodies on leaf surface of this plant have also been observed (Sheue et al., 2006). Baseline studies of the leaf structure of this plant from general electron microscopy contrast with MF-2PLSM, revealing the advantages of these new nonlinear techniques to better understand this deep shade plant noninvasively.
Material and methods
Laser sources and imaging system
The schematic of the MF-2PLSM is shown in Figure 1. This MF-2PLSM consists of an inverted i-mic 2 microscope (Till-Photonics, Grafelfing, Germany), equipped with Ti:sapphire femtosecond laser with a tuning range of 690–820 nm. The laser is a Tsunami 3941-M3B pumped by a Millennia-V, 5W solid-state pump laser (Spectra-Physics, Mountain View, CA, USA). The Tsunami laser was used to generate linearly polarized pulses at 810 nm, 20 mW and 100 fs pulse width at frequency of 80 MHz, for fluorescence excitation and SHG. Thus, in general to maximize the signal (fluorescence emission and SHG), short pulses should be used and average laser power should be kept low to prevent heating of the sample as well as unwanted one-photon absorption and to reduce the risk of highly nonlinear photodamage (Denk et al., 1995). A beam expander was used to fill the back aperture of the objective and λ/2 plate was used to control or maximize the status of the laser's polarization at the sample. The excitation light was directed onto a pair of galvanometer XY scanners (Yanus; Till-Photonics). The scanned excitation light was focused onto the sample through the microscope objective to scan the sample in the x–y direction at the focal plane. The stage of the microscope is driven by a computer controlled motor to take the sample to different z positions following each x–y scan. The scanning mirrors are metal coated (silver) with a good thermal resistance (Diaspro, 2001). Further components from the set-up in Figure 1 are: the dichroic mirrors [DM-1: Q565LP for TPEF (for the materials which produce autofluorescence) (Fig. 1b]; emission filter EF-1: red glass 665 nm; DM-2: Omega 475DCLP for SHG (Fig. 1c); interference blue emission filter EF-2 (405 nm; Fig. 1d); photomultipliers: Hamamatsu R6357; objective 1: Olympus uplanApo/IR 60×/1.20 water immersion; objective 2: Zeiss 40×/1.2W korr or Olympus uplanFLN 10×/0.3). The Laser power was maintained to be 20 mW at the Tsunami aperture and 5 mW at the sample. Additional infrared beam block filters BF (HQ700SP-2p 58398) were placed in front of each photomultiplier to ensure that illumination light was effectively suppressed and only TPEF or SHG signals were recorded. For SHG imaging, optical filtering is achieved with an interference filter centred on the expected SHG frequency (Fig. 1d) configurations of the photomultipliers were identical for both SHG and TPEF imaging. This set-up will enable the simultaneous measurement of SHG in the forward and backward directions as well as TPEF (Barzda et al., 2004). The signals from the photomultipliers are reconstructed by a computer into images. Images were obtained in stacks stepping along the z-axis with 0.5 μm steps. Preliminary imaging of the sample has been performed with a scan rate of 0.25 s−1 (512 × 512 pixels) and signal-to-noise ratio is about 20 dB. The lateral resolution is about 270 nm and the axial resolution is 973 nm using Olympus uplanApo/IR 60×/1.20 water immersion objective. The microscopy is controlled via a standard high-end Pentium-4 PC and linked to the electronic control system via an ultrafast interface.
The material used here is S. erythropus, a deep shade plant native to South America. This plant was originally collected from Singapore Botanic Gardens and grown in the laboratory in a deep shade environment. General electron microscopy was applied to semithin sections of a leaf prepared by standard TEM methods (Sheue et al., 2007). In addition, a tabletop microscope (TM3000, Hitachi, Japan) gave leaf surface images. This plant was moved to a dark location for two weeks before the investigation of MF-2PLSM to eliminate starch grains from the chloroplasts. To apply MF-2PLSM, a leaf was detached with watchmakers forceps from a darkened part of the plant. The leaf was mounted between two cover slips in water and the edges of the smaller cover slip were sealed to the lower larger cover slip by means of nail varnish. The paired cover slips were placed on the stage of a Till-Photonics microscope and illuminated with a Titanium sapphire laser at 810 nm (linearly polarized laser), 5 mW and 100 fs pulse width. The objectives were aligned relative to one another and focused on the sample. A set of images was captured.
Selaginella erythropus is anisophyllous, with two rows of small dorsal leaves and two rows of large ventral leaves on each branch of its stem (Fig. 2a). The arrangement of these leaves is prominently dorsiventral, with a stem located between dorsal leaves on the top and ventral leaves beneath (Fig. 2b). Because the axis (viz. stem) is in the middle and the leaf structure of both types is basically the same relative to the vertical direction, here we apply the terms ‘ventral side’ to the lower surfaces and ‘dorsal side’ to the upper surfaces of both types of leaf rather than the common terms ‘abaxial and adaxial sides’, which with this unique foliar arrangement are not helpful. The dorsal sides of both dorsal and ventral leaves are green (Fig. 2b). The ventral side of the dorsal leaf, which cannot be easily viewed from either the dorsal or ventral surface of the shoot, is green except for a red margin, whereas the ventral side of the ventral leaf is deep red. Silica bodies appear as conical protrusions from the epidermal cell walls on both surfaces of dorsal and ventral leaves as previously reported by Sheue et al. (2006). Silica bodies forming a single row on a single ventral epidermal cell of ventral leaves are the most evident silica body pattern (Fig. 2c). Stomata are aggregated in a band along the vein on the dorsal side of dorsal leaves and the ventral side of ventral leaves (Fig. 2d).
Dorsal and ventral leaves of S. erythropus are six cells thick in the vein region, with leaf thickness gradually reduced to two layers (the upper and lower epidermis) towards the margin (Fig. 2d). The outer tangential cell wall of ventral epidermal cells is very thick with multiple layered ultrastructure (Figs 2d and e). Chloroplasts are found in dorsal epidermal cells, mesophylls and ventral epidermis, including guard cells in leaves of S. erythropus. However, the size and number of chloroplasts vary between these tissues (Table 1). Bizonoplasts, giant cup-shaped unique chloroplasts with dimorphic ultrastructural organization in a single chloroplast, are located in dorsal epidermal cells: the upper zone is occupied by numerous layers of two to four stacked thylakoid membranes, whereas the lower zone contains both unstacked stromal thylakoids and thylakoid lamellae stacked in normal grana structures oriented in different directions (Fig. 2e). The chloroplasts in other tissues (such as mesophyll, ventral epidermis) are normal chloroplasts and are smaller. These observations viewed by LM, SEM and TEM provide a substantial basis of comparison for the results from SHG signals.
Table 1. Chloroplast diversity in microphyll of Selaginella erythropus.
Chloroplast no. per cell, shape and type
Average length (μm; N= 10)
Dorsal epidermal cell
1, cup, bizonoplast
26.7 ± 3.5
4∼6, disc, normal chloroplast
8.1 ± 1.5
Ventral epidermal cell
4∼9, beadlike, normal chloroplast
7.0 ± 1.0
4, oval, normal chloroplast
8.4 ± 1.1
0∼13, disk, normal chloroplast
5.3 ± 0.6
Figure 3 shows the backward direction SHG signal of the ventral side of the dorsal leaf before (Fig. 3a) and after (Fig. 3b) maximizing the polarization of the linearly polarized laser in the orientation of the dipole moment in the molecules. The orientation of the laser's polarization is illustrated by double arrows in Figure 3. These images were collected using the objective Olympus uplanFLN 10×/0.3. Figure 3(a) shows the weak signal of the SHG before the orientation of the laser's polarization parallel to the orientation of the dipole moment in the molecules; then after slowly changing the polarization's direction of the laser beam, the SHG signals significantly increased to reach the maximum value, as it is illustrated by Figure 3(b).
Figures 4 and 5 show the simultaneously acquired forward and backward SHG images of the dorsal and ventral surfaces of a dorsal leaf. The forward SHG signal was collected with a Zeiss 40×/1.2 water immersion objective and the backward SHG signal was collected using the objective Olympus uplanApo/IR 60×/1.2 water immersion. These figures of the forward (Figs 4a, c and 5a, c, e) and backward images (Figs 4b, d and 5b, d, f) are almost identical except that the backward images usually have slightly higher contrast. Figure 4 shows the images of the dorsal epidermal cells with a stomata band along the middle part. The area of this stomata band is slightly curved, leading to different focal planes under a microscope. The outlines of the dorsal epidermal cells, stomata and guard cells surrounding stomata can be recognized easily with SHG signals. The bizonoplasts in dorsal epidermal cells are revealed as much bigger than the chloroplasts in the mesophyll cells and guard cells (Figs 4a and b). Compared to a single giant chloroplast in a dorsal epidermal cell, there are three to five chloroplasts per mesophyll cell and four chloroplasts per guard cell (confirmed by confocal scanning light microscopy, data not shown). Scanning to a deeper position of these dorsal epidermal cells reveals numerous vacuole-like vesicles in each cell (Figs 4c and d). Whether these signals are derived from vacuoles or other organelles needs further investigation.
SHG images from the ventral epidermis show that its outer tangential cell walls have very strong signals with multiple layered dark curve patterns (Figs 5a and b). Compared to isodiametric dorsal epidermal cells, ventral epidermal cells are much elongated with smaller bead-like chloroplasts arranged as chains (Figs 5c and d). There are three to five disc-shaped chloroplasts in a mesophyll cell, with median size (Figs 5e and f).
Because the simultaneously acquired forward and backward SHG images are very similar (see supplementary figures), here we show only the backward SHG images of the dorsal and ventral surfaces of the ventral leaf (Fig. 6). These images were collected using the objective Olympus uplanApo/IR 60×/1.2 with water immersion. The isodiametric dorsal epidermal cells (Fig. 6a) and oblong mesophyll cells (Fig. 6b) shown in the top view in a ventral leaf are similar to those observed in a dorsal leaf. Silica bodies on the ventral side of a ventral leaf also emit strong signals (Figure 6d) matching the results observed by SEM in Figure 2(c). The smallest chloroplasts in the leaves of S. erythropus were observed in the trichomes of the leaf margin near the basal part (Figs 6c and e), note that trichomes along the leaf margin have smaller chloroplasts (arrows) see Figure 6(c). Ventral leaves (Figs 6e and f) and dorsal leaves (Figs 4 and 5) are similar in the patterns of size and arrangements of chloroplasts in dorsal epidermal cells, mesophyll cells and ventral epidermal cells.
This study reveals high SHG signals in the leaves of S. erythropus originating from micromophology, cell walls, cell contents and chloroplasts. Various categories of size and number of chloroplasts can be recognized from the leaves of S. erythropus (Table 1), in strong agreement with the observations of Sheue et al. (2007). These diverse chloroplasts include (1) large cuplike chloroplasts, bizonoplasts, in the dorsal epidermal cells; (2) disk-shaped chloroplasts in the mesophyll; (3) elongated or beadlike chloroplasts arranged as a chain in the elongated, ventral epidermal cells; (4) trichome chloroplasts; and (5) stomatal chloroplasts. In terms of ultrastructure, only the bizonoplasts have dimorphic ultrastructure: the upper zone is occupied by numerous layers of two to four stacked thylakoid membranes whereas the lower zone contains both unstacked stromal thylakoids and thylakoid lamellae stacked in normal grana. The other types of chloroplasts in the leaves of S. erythropus are typical chloroplasts with grana and stoma thylakoid membranes mingled together. These features of chloroplasts observed from S. erythropus are consistent with previous findings that many shade plants have large chloroplasts with numerous thylakoids per granum (Nasrulhaq-Boyce & Duckett, 1991; Sarafis, 1998). The SHG signal is not as effective as TEM in differentiating the upper zone and lower zone of a bizonoplast, but it provides strong signals with information on arrangement, shape and size of the five types of chloroplasts. The chloroplasts exhibit strong birefringence with large local variations, most likely originating from grana, and the stacked regions of the thylakoid membranes (Garab et al., 2005). The birefringence is important in fulfilling phase-matching conditions (Boyd, 1992; Reshak et al., 2008; Reshak, 2009). The birefringence is the difference between the extraordinary and ordinary refraction indices. Generally, materials show high birefringence (a considerable anisotropy in the linear optical susceptibility) that favours an important quantity in second-order susceptibility (determining SHG) because of better fulfilling of phase-matching conditions, determined by birefringence (Reshak et al., 2008). SHG is very efficiently generated in chloroplasts (Chu et al., 2001). Chloroplasts in celery showed a signal in the SHG image, which did not colocalize with the autofluorescence of the chlorophyll. Crystalline starch in starch grains is typically organized with the crystallites in a radial fashion, yielding a characteristic cross image in polarized light (Clowes & Juniper, 1968). This in turn means that SHG image will be orientation dependent (Cox et al., 2004). The significant SHG seen in biological materials arises from low local symmetry and the large nonlinear coefficient typical for biological molecules and structures (Lukins et al., 2003; Helmchen & Denk, 2005). The chloroplasts containing starch grains (Chu et al., 2001), which are strong sources of SHG signals. In this measurement, the plant was kept in the dark for approximately 2 weeks to eliminate the starch. From above, one can conclude that the origin of the high amount of SHG signals which comes from S. erythropus leaves is attributed to the unusually large chloroplasts (bizonoplasts) and various categories of size and number of chloroplasts with numerous and thick unusual thylakoid membranes, which are very strong sources of SHG signals. In this study, the chloroplasts in trichomes are the smallest chloroplasts with relatively weaker SHG signals than the other chloroplasts in this plant. This result is consistent with the observation of trichome chloroplasts examined by TEM (data not shown). The chloroplasts in trichomes have limited and poorly developed thylakoid membranes. In addition to the abundant SHG signals derived from chloroplasts, some cell contents, silica bodies and cell walls also displayed strong SHG signals. However, we do not know which structure causes the curve pattern of SHG signals around ventral epidermal cells in Figures 5(a) and (b). Further study is needed.
As SHG was established by earlier works (Stoller et al., 2002; Lukins et al., 2003; Reshak 2009; Reshak et al., 2009), the SHG signal strongly depends on the laser's status of polarization and the orientation of the dipole moment in the molecules that interact with the laser beam. It is therefore advantageous to control the laser's status of polarization, to maximize the SHG signals.
Our results support the contention that the collecting efficiency of SHG signals is highly dependent on the numerical aperture of the objective (Han et al., 2005; Cox et al., 2004; Reshak et al., 2009). Higher values of numerical aperture with immersion medium allow increasingly oblique rays to enter the objective front lens, by capturing higher order of diffraction rays from the samples, producing a more highly resolved image (Reshak et al., 2009). The strength of the SHG signals significantly depends on the numerical aperture and the immersion medium of the objective. Also, it is strongly dependent on the polarization direction of the laser beam. The sample will produce a strong SH signal when the polarization direction of the linearly polarized laser is parallel to the orientation of the dipole moment in the molecules.
SH imaging is especially helpful for biological studies of living samples. Acquiring fluorescence images with conventional microscopy leads to photobleaching and photodamage, whereas the SH imaging process does not. Because the SHG does not use an absorptive process, the intense laser field induces a nonlinear polarization in the molecules resulting in the production of coherent waves, twice the incident frequency. Moreover the SHG image results from a few femtoseconds, and is energy conserving process. This is another advantage of the SH imaging when one needs to work with sensitive samples. Thereby, one can investigate the true architecture of the sensitive samples.
This is the first time the deep shade plant S. erythropus has been investigated by means of the MF-2PLSM established by combining three platforms of laser scanning microscopy. The MF-2PLSM offers several advantages for uncovering the true architecture of the sample and enables simultaneous measurement of the SHG signals in the forward and backward directions. The leaves of S. erythropus produce very strong SHG signals that are attributed to various categories of size and number of chloroplasts with numerous thylakoid membranes. Moreover, the leaves are multilayered providing another reason for the strong SHG signals, which accumulate from these layers. Cell wall, cell content and big silica bodies also provide signals. This measurement provides noninvasive, effective and informative images similar to paradermal sections of the leaf but without the disadvantages of photobleaching and photodamage.
In summary, the SHG signals strongly depends on two objects: the first object is the microscope – the laser's status of polarization and the numerical aperture of the objective; and the second object is the biological materials – the structure of the materials whether if it is homogenous or not, or centrosymmetric or non-centrosymmetric, and the orientation of the dipole moment in the molecules that interact with the laser beam. This new emerging microscopy shows high potential for the study of living samples in biological and medical research.
We would like to thank Prof. V. Sarafis and Singapore Botanic Gardens for providing the plants, Prof. P. Chesson for correcting the English and two anonymous referees for valuable comments on the manuscript. This work was supported from the program RDI of the Czech Republic, the project CENAKVA (No. CZ.1.05/2.1.00/01.0024), grant No. 152/2010/Z of the Grant Agency of the University of South Bohemia. The School of Material Engineering, Malaysia University of Perlis, P.O Box 77, d/a Pejabat Pos Besar, 01007 Kangar, Perlis, Malaysia.