The past two decades saw the emergence of two-photon excited fluorescence (TPEF) laser-scanning microscopy as a powerful imaging tool for thick tissue imaging. Owing to its inherent optical sectioning capability, deep penetration of the excitation light in tissues and its capability to excite multiple fluorophores, TPEF laser-scanning microscopy has been used in a wide range of imaging applications. The demand for two-photon imaging of fast biological events prompted the development of scanning techniques that pushed two-photon imaging at video rates including resonant scanning [1,2], line-scanning [3], multifocal scanning [4], polygonal mirror scanning [5], and diffractive optical element-based (DOE) scanning [6,7].

In recent years, novel TPEF imaging techniques that do not involve mechanical scanning of the excitation light have emerged. Temporal focusing of 10 fs infrared laser pulses achieved depth resolved images of DAPI-labeled Drosophila embryos, albeit with out-of-focus fluorescence background [8]. A subsequent study has demonstrated that combining spatial modulation of the excitation light with the temporal focusing technique resulted in better axial confinement, with an axial resolution practically similar to that of two-photon line-scanning microscope [9]. Excitation multiplexing by using a spatio-temporal light modulator (SLM) also enabled scanless two-photon imaging and simultaneous photostimulation of neurons [10]. Recently, lateral wavelength division scanning for TPEF was demonstrated whereby different wavelengths are focused on to the sample at different lateral positions by tuning the excitation wavelength [11] and this has been extended to axial wavelength division scanning using a Fresnel lens which has a large chromatic aberration [12]. Axial scanning has also been achieved by simultaneous spatial and temporal focusing which was controlled by adjusting the group velocity dispersion [13].

In this study, we demonstrate a novel and simple scanless TPEF imaging technique based on selective plane illumination microscopy (SPIM). Light-sheet-based microscopes such as SPIM use a wide-field fluorescence microscope in their detection unit. For excitation, illumination plane (light sheet), coinciding with the sample plane of the microscope (see Fig. 1 inset), is used. This configuration restricts the excitation to the fluorophores in the volume around the plane of illumination and in effect provides instantaneous (as opposed to raster scanning) optical sectioning. This allows for a fast (2D) high resolution imaging of the specimen. Since in SPIM, fluorophores outside the illumination are not excited, photobleaching and phototoxic effects are dramatically reduced. SPIM is also used for imaging relatively large samples which require objective lenses with low NA and low magnifications. Recently, larger samples such as embryos have been image with the development of digital scanning light sheet microscopy (DSLM) [14]. Although SPIM has been well studied and applied to a wide range of imaging applications [15–18], this paper is, to the best of our knowledge, the first demonstration of two-photon selective plane illumination fluorescence microscopy (henceforth, 2p-SPIM). We note here that there has been a previous proposal of incorporating to a confocal theta two-photon microscope a detection system oriented at a right angle to the illumination [19]. We show the relative simplicity of the experimental setup which mainly consists of two cylindrical lenses forming a thin sheet of light, and an objective lens forming the optically sectioned image on to an imaging detector. We present results on the characterization of the light sheet and finally its application to depth-resolved in vivo imaging of cameleon-expressing in the pharynx of Caenorhabditis elegans (C. elegans).

 

Fig. 1 Schematic diagram of the 2p-SPIM experimental setup.

Download Full Size | PDF

The practical setup for 2p-SPIM is shown in Fig. 1. For the excitation source, a typical Kerr lens mode-locked Ti:sapphire laser (MIRA 900f, Coherent, France), with pulses of 160 fs with a repetition rate of 76 MHz was operated at a central wavelength of 860 nm. After attenuation of the excitation beam using a variable neutral density filter wheel, its diameter was increased using a pair of spherical lenses in a telescope configuration. The beam was allowed to pass through a square aperture (4 mm × 4 mm) and through the first cylindrical lens (f = 200mm) that determines the height (δy) of the plane illumination. After passing through a mirror, the beam is focused horizontally to the sample specimen by a second cylindrical lens (f = 20mm) resulting in an effective excitation NA of 0.1. Samples were embedded in ~1.5% agar gel and immersed in a water-filled chamber made of cover slip windows.

For the C. elegans imaging experiments, adult hermaphrodite worms were immobilized using sodium azide-NaN3 (50 mM). To obtain optical section at different depths (z-direction), we attached the specimen to a computer-controlled linear translation stage (M-505.6DG, Physik Instrumente GmbH & Co. KG, Karlsruhe, Germany). An air objective lens (20 × /NA0.40/WD3.90, Nikon) oriented perpendicular to the excitation axis was used to collect the fluorescence from the sample. To block scattered laser excitation light an emission filter (BG39, Schott) was placed before the high-sensitivity imaging detector. Images were recorded on a back-illuminated EMCCD (iXonEM + 897, Andor Technology plc., Belfast, Northern Ireland) with a full frame image size of 512 × 512 pixels and a pixel size of 16 µm.

In order to estimate the achievable resolution of 2p-SPIM, we have calculated the two-photon illumination point-spread function (PSF_ill) in the transversal (δz) and axial (δx) directions (see Fig. 1, inset) [20]. Here, the transversal and the axial PSF_ill determine the thickness and length of the two-photon plane illumination, respectively. Note here that the transversal PSF_ill plays a role in determining the axial resolution of the image (i.e., relative to the detection axis, z) while the axial PSF_ill determines the length of the image field-of-view. In theory, the length of the two-photon illumination length is expected to be by a square root of two shorter than the one-photon illumination length (at the same illumination thickness). Notice that the lateral imaging resolution is determined by the objective lens and the pixel size of the CCD but it is independent of the illumination numerical aperture. Shown in Fig. 2 are the results of the calculations for the axial and transversal PSF_ill measured at the full-width at half-maximum (FWHM) as a function of the illumination numerical aperture of the illumination system (NA_ill) at the two-photon excitation wavelength of 860 nm in air (refractive index is 1.0). Based on Fig. 2, a two-photon PSF_ill with length of ~100 µm and thickness ~3 µm can be achieved using NA_ill = 0.1. Longer PSF_ill (>100µm) can be achieved by using lower NA_ill (<0.1) albeit accompanied by an increase in PSF_ill thickness. On the other hand, higher NA_ill (>0.1) results in narrower PSF_ill but with shorter PSF_ill length. For example, a two-photon illumination of ~1 µm can be achieved using PSF_ill ≈0.3 but with a short illumination plane of ~10 µm. The vertical extent of the PSF_ill (δy) can be conveniently varied by the first cylindrical lens oriented horizontally with a long focal length as shown in the experimental setup (see Fig. 1). Alternatively, the height of the aperture can be varied at the expense of losing excitation intensity.

 

Fig. 2 Calculated FWHM of TPEF axial and transversal PSF as a function of the illumination numerical aperture of cylindrical lens 2 (see Fig. 1) for two-photon excitation wavelength of 860 nm.

Download Full Size | PDF

To gain knowledge on the vertical and horizontal dimensions of the two-photon excitation profile in 2p-SPIM, we recorded an XY image of the fluorescence of Coumarin 540 (Excition, Dayton, OH, USA) in methanol as shown in Fig. 3, A . Here, we adjusted the first cylindrical lens (see Fig. 1) so that we obtain the minimum height in the fluorescence image. We measured the horizontal extent (δx) of the two-photon illumination to be 268±21 µm (see Fig. 3, B) which is more than two-fold larger than the computed axial PSF_ill length of 143 µm for NA_ill = 0.1 (see Fig. 2, red line). This discrepancy may be due to the combined effect of optical aberration introduced by the cylindrical lens and the refractive index mismatch between the different media (i.e. air, water and agar gel). Moreover, an experimental error in the value of the numerical aperture may also be attributed to this large discrepancy. We calculate using the relation between the axial PSF and the NA, that 268±21 µm corresponds to an experimental numerical aperture NA = 0.07±0.01. On the other hand, the vertical extent (δy) measured 21.9±1.0 µm (see Fig. 3, C) which is in good agreement with the minimum achievable height of the light sheet (20.4 µm) considering the focal length of cylindrical lens 1 to be 200 mm corresponding to a numerical aperture (in the horizontal direction) of 0.01.

 

Fig. 3 A: Pseudocolor 2p-SPIM fluorescence image of Coumarin 540 dye in methanol. Depicted also are the horizontal (B) and vertical (C) PSF obtained from cross-sectional intensity profiles (green lines). We measured the two-photon illumination length (δx) and height (δy) as 268 µm and 21.9 µm, respectively. D: Transversal PSF of the two-photon plane illumination obtained from z-section intensity profile of a typical XZ pseudocolor image (D, inset) of a 0.028 µm fluorescent bead excited. An average thickness (δz) of 3.4 ± 0.2 µm (n = 5) were measured. Image dimensions: (A) 590 µm × 150 µm (X × Y), (D, inset) 4 µm × 16 µm (X × Z).

Download Full Size | PDF

To demonstrate the optical sectioning capability of the constructed 2p-SPIM experimental setup, images of fluorescent polymer microspheres (Firefli Fluorescent Green, size: 0.028 µm, Duke Scientific Corporation, Palo Alto, CA, USA) embedded in 2% agar gel were acquired for different relative depths (1 µm steps) of the sample. Shown in Fig. 3, D is a typical z-profile of a microsphere depicting a Gaussian fit with FHWM of 3.4 µm and an XZ image perpendicular to the optical section (inset, image dimension: 4 µm × 16 µm) of the microsphere. The gray level is presented in pseudocolor. This clearly indicates optical sectioning capability of the 2p-SPIM setup with an average two-photon plane illumination thickness of 3.4±0.5 µm obtained from five measurements of microspheres. Although this value is in good agreement with the calculated PSF_ill thickness of 3.2 µm for NA_ill = 0.1, the obtained two-photon plane illumination thickness in fact corresponds to a numerical aperture NA = 0.09±0.01. This is close to the value (NA = 0.07±0.01) obtained when we calculated for the corresponding experimental numerical aperture to two-photon illumination length.

We then proceeded to test the sectioning capabilities of 2p-SPIM in a typical imaging experiment involving a living model organism, the C. elegans. To do that we acquired depth-resolved images of C. elegans pharyngeal muscle expressing the fluorescent protein “cameleon”. This protein is normally used as calcium indicator in imaging calcium transients in intact C. elegans [21]. It is composed of four domains, cyan fluorescent protein (CFP), calmodulin, M13 (a calmodulin binding domain), and yellow fluorescent protein (YFP). In our proof-of-principle experiment, the worms were deeply anesthetized and no calcium transients were expected. Hence, the observed fluorescence signal was attributed only to two-photon excitation of CFP.

Figure 4 shows the observed 2p-SPIM fluorescence signal (Fig. 4, upper) and its overlay on a brightfield image of the anterior body section of C. elegans (Fig. 4, lower). In the fluorescence imaging, we used an average laser excitation power of ~40mW and an acquisition frame rate of 5Hz (image exposure time, 200ms). At this excitation power levels, no evidence of photobleaching could be observed. The overlay image clearly shows that the fluorescence signal originates from the pharynx. Moreover, the fluorescence image illustrates identifiable anatomical structures of the pharynx, including the spherical terminal bulb, the elongated isthmus, the metacarpus and the procorpus (see Fig. 4).

 

Fig. 4 Upper: 2p-SPIM fluorescence image section of cameleon-expressing in C. elegans pharyngeal muscles. Lower: Overlay of the 2p-SPIM fluorescence (cyan) and brightfield image of anterior of C. elegans depicting colocalization between the fluorescence signal and the pharynx. Also indicated are the distinct structures in the fluorescence image associated with the known anatomy of C. elegans. BC: bucal cavity, PC: procorpus, MC: metacarpus, IS: isthmus, TB: terminal bulb. Scale bars: 50 µm.

Download Full Size | PDF

The capability of 2p-SPIM to record optical sections, as demonstrated earlier (Fig. 3) allowed us to reconstruct a three-dimensional image of the cameleon-expressing pharynx in the soil worm by acquiring images at different specimen depth. Shown in Fig. 5 are 2p-SPIM optical sections of the specimen at 4 µm step intervals (Fig. 5, A to J), and the three-dimensional reconstruction of the pharynx (Fig. 5, K and Media 1).

 

Fig. 5 A to J: 2p-SPIM fluorescence optical sections of cameleon in C. elegans pharyngeal muscle at 4 µm depth intervals clearly depicting the optical sectioning capability of 2p-SPIM. K: Movie showing the reconstructed three-dimensional structure of the C. elegans pharynx (Media 1). Scale bars: 50 µm.

Download Full Size | PDF

The results presented in this work illustrate, for the first time, a practical demonstration of 2p-SPIM and its implementation to three-dimensional imaging of C. elegans pharyngeal muscle. One disadvantage of the 2p-SPIM is the reduced field size due to the nonlinear nature of excitation. However, this is compensated by the use of excitation wavelength more or less twice that of one-photon excitation. Furthermore, it is clear that there is a lot of room for technical improvements: the quality of the images which relies on the signal-to-noise ratio (SNR) can be improved and ii), 2p-SPIM imaging can be implemented with faster image frame rates. It is important to note, however, that there is a trade-off between the quality of the image and the image frame rate and is limited by the amount of collected two-photon excited fluorescence photons. As such, both these improvements require increased fluorescence photons which can be achieved by one or both of the following approaches: 1) more efficient detection and; 2) higher excitation power. The use of higher excitation power levels, although technically an easier approach of the two may be accompanied by possible adverse effect of photodamage. Our estimates show that photodamage due to both nonlinear absorption processes [22–24] and sampling heating [25] are negligible in the present 2p-SPIM setup. The total laser power of 40 mW is distributed to a large area in the sample and that the optical power density being is actually very small. In fact, the 40 mW 2p-SPIM excitation light expanded into a vertically elongated beam having a cross-sectional area at the focal plane (relative to the cylindrical lens) of ~3 µm (transversal) × ~100 µm (vertical) ≈300 µm² corresponds to an optical power density or intensity of 0.13 MW⋅cm⁻². Moreover, based on the relation between photodamage, excitation power, and pixel dwell time derived previously [24] together with the recommended “safe” working condition (power = 10 mW, pixel dwell time = 10 ms for similar laser parameters), we computed that for an exposure time of 200 ms (our experimental condition), it is “safe” to use intensities up to 3 MW⋅cm⁻². Thus, the present laser intensity condition is below the limit of nonlinear-induced photodamage and that a 23-fold increase in laser power can still be considered in the “safe” limit. Reports have also confirmed that temperature increase occurs due infrared absorption in the focus of an objective lens both in optical trapping and multiphoton imaging studies. For instance, an increase in temperature of ~3 K after 8-s irradiation of an aqueous sample by 170 mW of 1064-nm laser light focused with NA = 1.3 oil immersion lens [26]. However, at excitation wavelengths typically used for multiphoton imaging (750 nm to 850 nm), a less significant temperature increase of < 250 mK is found even for irradiation time of 10 s using a focused excitation beam (NA = 1.2) of 100 mW [25]. These results suggest that the 2p-SPIM setup using 40 mW focused excitation beam (NA = 0.1) at 800-nm can induce temperature increase in an aqueous sample but only by less than 250 mK, which cannot be considered a limiting factor for in vivo biological imaging.

Whereas this report demonstrated a practical 2p-SPIM setup for the first time, demonstration of its advantages over other imaging techniques is yet to be carried out. In the future, our experiments will be focused on showing the capabilities of 2p-SPIM in terms of fast imaging rates, large specimen imaging, and long-term imaging. It is our view that the potential advantages of 2p-SPIM lies primarily in its use of long-wavelength excitation light. Near infrared (NIR) light scatters less inside turbid media than visible or UV light implying deeper penetration of the excitation light inside tissues and large specimens can potentially be imaged by 2p-SPIM. This exemplifies a possible future application of 2p-SPIM in large-specimen imaging. Moreover, the use of near infrared as excitation light source in 2p-SPIM has its advantages in terms of wavelength-dependent phototoxicity issues. It has been demonstrated that long wavelength illumination has less detrimental effects than visible- or UV-light in mammalian embryos [27]. This sensitivity to short-wavelength light, mainly attributed to radical oxygen species (ROS) production has also been reported in the development of mammalian zygotes [28]. Thus, it is likely that imaging studies on long-term biological processes such as mammalian embryogenesis will benefit from the low phototoxicity of NIR-excited 2p-SPIM.