1. Introduction

Fluorescence microscopy is widely used in medical and biological research. However, the spatial resolution of fluorescence microscopes is limited to ca. 200 nm because of the wave nature of light. Several methods, such as atomic force microscopy and electron microscopy, are used to capture images of biological nanostructures. In this decade, several novel techniques to overcome the limitations of the spatial resolution of optical microscopy have been proposed; these techniques are known as super-resolution microscopies [1, 2]. Some of these, such as structured illumination microscopy [3], photoactivated localization microscopy [4], stochastic optical reconstruction microscopy [5] and stimulated emission depletion (STED) microscopy [6, 7] are now commercially available. However, these microscopic techniques are only suitable for thin specimens.

On the other hand, a two-photon excitation laser scanning fluorescence microscope (TPLSM) is widely used because it has several advantages [8]. First, excitation is spatially localized at the focus of the excitation laser light used in TPLSM due to non-linear dependence on photon density, which allows for acquiring optically sectioned fluorescence images. Second, the excitation induced by near-infrared (NIR) laser lights provides other advantages of superior penetration depth and is less invasive for in vivo imaging of living specimens [9]. However, in principle, in comparison with conventional confocal laser scanning microscopy, the spatial resolution with TPLSM tends to be inferior because the wavelength of the excitation laser light used is longer. Recently, the spatial resolution of TPLSM was improved by applying STED microscopy and enabled to visualize nanostructures of <100 nm [10, 11]. This two-photon excitation STED (TP-STED) microscopy is also applicable for deeper regions of biological tissues at depths of 50–100 μm from the surface [12, 13]. This STED is induced by a donut-shaped high-power laser light (STED light) that is superimposed with a two-photon excitation laser light. The STED light wavelength is usually set at the longer wavelength range of the emission spectra of fluorophores. By using optical filters to separate the stimulated emission light, a diffraction unlimited central spot can be selected.

Due to a donut-shaped STED beam, STED microscopy often adopts an optical vortex with a helical phase distribution of 2nπ (n is an integer, called as topological charge) per round [6, 7, 10–13]. An optical vortex can be generated using either a vortex phase plate (VPP) or a spatial light modulator (SLM). A VPP changes the phase distribution by locally varying its thickness, so that one VPP can be applied for only one specific wavelength. On the other hand, a SLM modulates the length of the optical path through which the beam passes locally by changing the alignment of liquid crystal molecules in the SLM. This modulation can be compatible over a wide wavelength region by tuning the applied voltage to the SLM. Although the modulation pattern depends on the geometrical pattern of the SLM [14], an optical vortex with a stepwise phase distribution is available.

Recently, we utilized transmissive liquid crystal devices (tLCDs) that enabled to modify laser beam optical properties, such as phase, polarization, and intensity, by tuning the applied voltage at a high conversion efficiency. Compared with reflective liquid crystal devices, wherein installation makes the optical system larger, tLCDs can be installed compactly by only placing them in the optical path. In our previous studies, higher-order radially polarized beams generated by tLCDs showed that lateral resolution improved both in confocal laser scanning microscopy and TPLSM [15, 16]. These tLCDs can be used for applications over a wide wavelength region by simply changing the applied voltage.

For this study, we constructed a new system for TP-STED microscopy by adding a STED laser light source into a conventional TPLSM system. The STED light beam was modulated by tLCDs to create an optical vortex at the focus. Here, we demonstrate this system's effectiveness for phase shift compensations, its applicability for multiple wavelengths, and use in super-resolution microscopy.

2. Materials and methods

2.1 Optical setup

A schematic of our optical setup is shown in Fig. 1(a). A two-photon excitation light (wavelength of 900 nm) was generated using a femtosecond mode-locked Titanium Sapphire laser (Tsunami, Spectra Physics) that operated at a repetition rate 80 MHz. The light source for STED was an optically pumped semiconductor continuous-wave (CW) laser (Genesis CX 577-3000, Coherent) that provided 3 W at a maximum of 577 nm. Spatial mode cleaning of the STED light was accomplished using a spatial filter. The directions of linear polarization for both lights were oriented to liquid crystal molecules in tLCDs by passing through half-wave plates. Two types of light, which were modulated by tLCDs as noted below, were merged at the first dichroic mirror (RDM680, Olympus) and introduced into a Galvano mirror scanner and an upright microscope (FV-1000 and BX-61WI, Olympus). Both lights were reflected by the second dichroic mirror (custom-made filter combining a 577 nm notch with a 880-nm-edge short pass filter; Asahi Spectra) and were focused on a specimen by using a water immersion objective lens with a numerical aperture of 1.2 (UPlanSAPO 60XW, Olympus). Fluorescent light was collected by the objective lens, passed through the second dichroic mirror and emission filters, and was finally detected with a photomultiplier tube.

 

Fig. 1 (a) Schematic of our TP-STED system using tLCDs. DM: dichroic mirror; F: filter; GM: galvano mirrors; tLCD: transmissive liquid crystal device; -P: plain cell; −24: 24-divided cell; M: mirror; OL: objective lens; PMT: photomultiplier tube; TL: tube lenses; λ/2: half-wave plate. (b) Theoretical phase distribution of an optical vortex generated by tLCD-24.

Download Full Size | PDF

We used two types of tLCDs. One was a plain cell tLCD (tLCD-P) with homogeneously aligned liquid crystal molecules functioning as an applied-voltage-dependent variable wave plate. The other was a 24-divided tLCD (tLCD-24) that enabled to produce a spiral-phase distribution with 24 steps around the center of the beam. Periodic square waves were applied to drive the tLCD-24 and tLCD-Ps using AFG3022B (Tektronix) and NI USB-6351 (National Instruments). The two-photon excitation light was modulated to a circular polarization at the position of the objective lens by one tLCD-P. As shown by the phase distribution in Fig. 1(b), by using the tLCD-24, the STED light was converted to an optical vortex for which the topological charge was 1. This optical vortex was passed through a tLCD-P installed in the optical path and tuned its polarization at the position of the objective lens.

To demonstrate the optical vortex of another wavelength using the same tLCDs, a diode-pumped solid-state laser (SDL-473-005T, Shanghai Dream Laser Technology) was used that provided 6 mW at a maximum of 473 nm. A 473 nm light was introduced into the same optical path of the 577 nm STED light, except for a spatial filter and second dichroic mirror. In this case, a second dichroic mirror was replaced with a 505-nm-long pass filter (DM505, Olympus).

2.2 Sample preparation and analysis

Fluorescent beads were mixed with a 1% agarose gel (agarose L, Nippon Gene). Fluorescent green beads with a diameter of 100 nm (100 nm green beads; Thermo Fisher Scientific) were used to evaluate the spatial resolution of our STED-TP microscope. To estimate the spatial resolutions, the full-width at half maximum (FWHM) of fluorescence intensity profiles of tiny beads were determined using the same methods as in our previous studies [15, 16]. Yellow-green beads with a diameter of 170 nm (170 nm yellow-green beads; Thermo Fisher Scientific) and orange beads with a diameter of 170 nm (170 nm orange beads; Thermo Fisher Scientific) were used to check for the donut-shaped focal pattern of the optical vortex at wavelengths of 473 nm and 577 nm, respectively. To compare these focal patterns, fluorescence intensity profiles across the intensity center along the x-axis were fitted to theoretical curves, and the ratios of central intensity to maximum intensity (I_center/I_max) were estimated. To compare the focal planes between the two-photon excitation light and the STED light, 100 nm green beads and 170 nm orange beads were placed on the same cover slip.

COS-7 cells were cultured in Dulbecco’s modified Eagle’s medium (Wako, Osaka, Japan) supplemented with 10% fetal bovine serum and penicillin/streptomycin (Thermo Fisher Scientific) at 37°C in a 5% CO₂ atmosphere. For immunostaining, COS-7 cells, which had been cultured on a cover slip, were fixed with formaldehyde and permeabilized with detergent. These cells were stained with an anti-α-tubulin antibody (clone DM1A; Cell Signaling) and an ATTO 425-conjugated anti-mouse IgG antibody (Rockland Immunochemicals).

3. Results and discussion

For STED microscopy, the focus of the optical vortex should have a steep central dark spot, which indicates that a higher spatial resolution can be achieved as the intensity at the center of the dark spot is reduced [6, 7]. Several numerical calculations have shown that the point spread function (PSF) of an optical vortex depends on the polarization of a focused laser beam [17, 18]. To focus an optical vortex beam, the handedness of circular polarization was chosen so as to create a fine dark spot at the focus [17]. The intensity distribution at the focus of a circularly polarized optical vortex beam produced by the tLCD-24 was numerically calculated based on a vector diffraction theory [19]. Figure 2(a) shows the calculated focal pattern and the intensity profile across the center. As shown in Fig. 2(a), the donut-shaped pattern was obtained at the focus, indicating that the effect of the discrete phase variation with 24 steps in the tLCD-24 was ignorable to form the ideal donut-shaped pattern as the STED light. Therefore, we first estimated the PSFs of various optical vortexes created using our setup by acquiring fluorescent images of a tiny fluorescent bead directly excited by STED lights.

 

Fig. 2 (a) Calculated intensity distribution in the focal plane for focusing of the circularly polarized 577 nm optical vortex created by tLCD-24. (b)–(d) Fluorescence images of a fluorescent bead directly excited by optical vortexes. The lower panels show the fluorescence intensity profiles across the intensity center along the x-axis. (b) Image of a 170 nm orange bead excited by 577 nm light with circular polarization at the objective lens. (c) Image of a 170 nm orange bead excited by 577 nm light with circular polarization at tLCDs. (d) Image of a 170 nm yellow-green bead excited by 473 nm light with circular polarization at the objective lens.

Download Full Size | PDF

For the case of a circularly polarized optical vortex at the position of the tLCDs, a difference in phase shifts between s- and p-polarized lights generated in the optical path resulted in elliptical polarization (aspect ratio of ca. 8) at the position of the objective lens. By tuning the voltage applied to a tLCD-P [see Fig. 1(a)], we generated circularly polarized light at the position of the objective lens as shown in Fig. 2(b). The intensity distribution of the generated circularly polarized optical vortex at the focus was almost identical with the calculated ideal focal pattern shown in Fig. 2(a). In contrast, a slightly distorted donut shape was obtained when the applied voltage was tuned so as to produce a circularly polarized optical vortex at the position of the tLCDs [Fig. 2(b)]. The ratios of central intensity to maximum intensity (I_center/I_max) were estimated for the images in Fig. 2(a) and 2(b). The I_center/I_max values were 0.16 and 0.34 for the images in Figs. 2(a) and 2(b), respectively. Thus, the phase shift caused by the microscope optics was compensated for by adjusting the voltage applied to the tLCD-P.

tLCDs can modify optical retardations by tuning the applied voltages and are applicable for beams over a wide wavelength region from visible to NIR [15, 16]. Thus, we confirmed that optical vortexes of different wavelengths were provided by the same set of tLCDs. Figure 2(c) shows an image of a 170 nm yellow-green bead that was directly excited by a 473 nm optical vortex. Although its PSF was rather distorted, the same set of tLCDs produced an optical vortex by applying a different suitable voltage to the tLCDs. This distortion might have been caused by some optical aberrations including astigmatism of the 473 nm laser beam. Astigmatism is known to perturb the hollow focal pattern of an optical vortex [20, 21].

The focal planes of the two-photon excitation light and STED light used in our system were verified. Figure 3 shows an orthogonal view of the xyz-fluorescent images of two different types of fluorescent beads that were placed on the same cover slip. Because several previous studies on STED employed circularly polarized lights for excitation, we adopted another tLCD-P to generate a laser beam that was circularly polarized at the position of the objective lens. Each of their focal planes became nearly identical by simply introducing these laser beams in parallel into the corresponding laser inputs of the microscope.

 

Fig. 3 Merged fluorescence image of 100 nm green beads (green) and 170 nm orange beads (magenta) placed on the same cover slip.

Download Full Size | PDF

Next, to evaluate the spatial resolution of our TP-STED system, fluorescent images of 100 nm green beads were acquired both with a conventional TPLSM system and with our TP-STED microscope. The two-photon excitation laser power at the output of the microscope objective was 1.3 mW for both TPLSM and TP-STED. The STED laser power at the same position was 89 mW. The pixel size of both images was 2.8 × 2.8 nm, and the pixel dwell time was 12.5 μsec. Two representative TPLSM and TP-STED images in Fig. 4 (a) were averages of 4 and 8 acquired images, respectively. After applying the STED laser light, the fluorescent bead image which had exhibited an isotropic shape in the focal plane became elliptical as the size itself became smaller [Fig. 4(a)]. The FWHM values along the major and minor axes of the resultant PSF estimated in the averaged fluorescence intensity profiles in our TP-STED system were reduced by 46% (from 322 nm to 173 nm) and 17% (from 318 nm to 265 nm) compared to that of the conventional TPLSM system. To estimate more accurate PSFs and the spatial resolution of the microscopy, measurements of fluorescent images of smaller-sized beads would be required. However, the sensitivity of our conventional TPLSM system was insufficient to visualize clearly such a smaller fluorescent bead due to a weak fluorescence.

 

Fig. 4 Comparisons of TPLSM and TP-STED images. (a) A 100 nm green fluorescent bead. The lower panels show the averaged fluorescence intensity profiles of n beads across the intensity center along the red dashed lines in the fluorescent images. Inlet length value indicates the full width at half maximum. (b) Microtubule networks in fixed COS-7 cells after immunostaining with antibodies conjugated with the fluorescent dye ATTO 425. The lower panels show the fluorescence intensity profiles across the red dashed lines in the fluorescent images.

Download Full Size | PDF

Finally, we demonstrated that our method was applicable for biological specimens by observing microtubules in fixed COS-7 cells that were immunostained with fluorescent-dye-labeled antibodies. ATTO 425 was chosen as the fluorescent dye because the wavelength of STED light did not overlap with its absorption band. The excitation laser power at the output of the microscope objective was 3.4 mW for both imaging. The STED laser power at the same position was 72 mW. The pixel size of both images was 41.4 × 41.4 nm, and the pixel dwell time was 10 μsec. The two images in Fig. 4(b) were averaged ones from each 10 images. According to them, TP-STED visualized the fine network structures of microtubules more clearly.

We have demonstrated that adding tLCDs and a CW yellow laser for STED improved the spatial resolution of TPLSM; however, the spatial resolution with our present system was estimated to be >100 nm, unlike those in previous studies [10–13]. This was probably because the optical vortex for STED was slightly distorted, which resulted in a finite intensity at the center of the focus [Fig. 2(a)] due to aberrations, including astigmatism, by the objective lens. This might be resolved by introducing other types of SLMs to cancel various types of aberrations [21, 22]. Another reason might be that we used a CW laser for STED. The STED efficiency was reported to be superior with pulse-type STED microscopy [12, 13].

In this study, we developed a new TP-STED system that had the following useful properties. First, by adjusting the voltage applied to a tLCD-P, we compensated for various phase shifts in the optical path to acquire a fine donut-shaped PSF and confirmed that this optical vortex, which had a 24-divided helical phase distribution of 2π created by a tLCD-24 [see Fig. 1(b)], was sufficiently effective for STED microscopy. Second, our TP-STED microscopy system was sable to be employed at different wavelengths of the STED laser by only tuning the voltages applied to tLCDs. Finally, by modifying a conventional TPLSM system by only adding tLCDs and a depletion laser light source, we developed a TP-STED system. This methodology achieves super-resolution microscopy more readily. Another feature of TPLSM is that various fluorescent dyes can be excited simultaneously using a single laser [8, 23]. Thus, combined with a wavelength tunable light source for STED, this methodology might provide for super-resolution microscopy to visualize multiple components in living specimens by using a variety of fluorophores.