Single-wavelength two-photon excitation–stimulated emission depletion (SW2PE-STED) superresolution imaging

Abstract

We developed a new class of two-photon excitation–stimulated emission depletion (2PE-STED) optical microscope. In this work, we show the opportunity to perform superresolved fluorescence imaging, exciting and stimulating the emission of a fluorophore by means of a single wavelength. We show that a widely used red-emitting fluorophore, ATTO647N, can be two-photon excited at a wavelength allowing both 2PE and STED using the very same laser source. This fact opens the possibility to perform 2PE microscopy at four to five times STED-improved resolution, while exploiting the intrinsic advantages of nonlinear excitation.

Two-photon excitation (2PE) fluorescence microscopy (1) is a widely used technique in medicine and biology (2, 3). Its characteristics make it particularly suitable for deep tissue and in vivo imaging applications (4). The use of infrared wavelengths guarantees a high penetration depth in scattering samples and a reduced photodamage in living specimens. Moreover, the quadratic dependence of the excitation allows intrinsical rejection of the out-of-focus fluorescence background and reduces the FWHM of the focal spot by approximately compared to the diffraction limited spot of excitation light. However, 2PE cannot be considered a genuine superresolution technique: The “doubling” in wavelength causes that the size of the diffraction spot of excitation light is twice than the one achievable by single-photon excitation (1PE). In 1994, Hell and Wichmann introduced the stimulated emission depletion (STED) concept as a way to achieve optical superresolution by breaking the diffraction barrier (5). In general, STED imaging is based on the fluorescence switching due to stimulated emission depletion introduced by a second red shifted beam. The fluorescent sample is probed by a focused excitation beam overlaid with a STED beam commonly shaped to a doughnut-like intensity distribution featuring zero intensity in the center. Because of the saturation of the stimulated emission depletion, the fluorescence is switched in the whole focal area except for the small region around the zero of the STED beam. The resulting resolution is theoretically unlimited being governed by the efficiency of the depletion process (6, 7). Recently, STED has been proposed coupled with 2PE microscopy combining the advantages of 2PE with its superresolution ability (8, 9). In general, continuous wave (CW)-STED (10) is preferred for 2PE imaging due to a simplification of the experimental setup at the expenses of an increased cost. Unfortunately, the use of two distinct wavelengths for excitation and depletion requires mostly a special optical filter design making the setup invariable in terms of the choice of the marker dye, and potential light beam distortions have to be treated separately. Working with only just one wavelength for both excitation and STED, one directly simplifies the image formation scheme. In this paper, we propose a method to perform 2PE-STED imaging using a single wavelength (SW) and, consequently, the very same laser source for 2PE and depletion. Such a SW approach is in agreement with early reported spectral evidences for different dyes (11, 12).

Physical Principles and Setup

The key point of the method lies on finding a wavelength that has nonzero cross-section for 2PE and STED processes and avoids 1PE excitation of a given dye. Then one has the chance to find a balance, in terms of power and pulse length of the beams, to achieve a significant fluorescence on-off contrast that enables such SW 2PE-STED superresolution imaging method.

The molecular cross-sections for the two-photon absorption are typically very low, so high light fluxes approximately equal to 10³⁰ s^-1 m^-2 photons are required. In order to reach this requirement, 2PE imaging profits of high-repetition pulsed lasers with high peak power, ultrashort pulses (tens to hundreds of femtoseconds), whereas average power remains fairly low thus avoiding local trapping or heating effects (3). It is worth recalling here that a common organic fluorophore follows the Kasha’s rule as depicted by the simplified Jablonsky diagram in Fig. 1A. When the simultaneous (ca. 10^-17 s) absorption of two photons occurs, the molecule goes to an excited vibrational level of one of the singlet excited states. In our case, because the wavelength used is close to the visible region, we expect that an S₂ singlet state could be involved in the absorption process. In general, a fluorophore very efficiently undergoes internal conversion (IC) from the S₂ state to the lowest vibrational level of the S₁ state in about the same time required to relax from an excited vibrational level of the S₁ state to its zeroth one (i.e., 10^-13 to 10^-11 s). Thus fluorescence and stimulated emission occur between S₁ and a vibrational sublevel of the ground state (green and red arrows, respectively, in Fig. 1A) (13). Because the decays nonradiatively within 0.2–5 ps, a long (few hundreds of picoseconds) STED pulse is more effective in the depletion of S₁ (14, 15). Moreover, a long STED pulse decreases the peak intensity, and thus avoids two-photon excitation from the STED beam. Notably, we use the same wavelength also for nonlinear excitation of the fluorophore, the fact that 2PE pulses are shorter than the IC time avoids a possible competing STED effect along with the excitation process.

Fig. 1.

(A) Simplified Jablonski diagram showing the states involved in 2PE-STED single wavelength. (B) Scheme of the setup. P, polarizer; DL, delay line; VPP, vortex phase plate.

For the above-mentioned reasons we realized a setup (Fig. 1B) where the laser beam of an ultrafast Ti:sapphire IR laser (Chameleon UltraII; Coherent) has been split by a polarizing beam splitter (PBS) into two beam paths whose relative powers can be tuned by a half-wave plate in front of the PBS. Now, the one used for two-photon excitation has to dodge stimulated emission depletion whereas the other, quenching the fluorescence excited by 2PE, has to avoid priming any linear or nonlinear excitation. Therefore, in order to perform superresolved imaging, the two beams have to differ in about three orders of magnitude in pulse widths and of more than two orders of magnitude with respect to the intensity.

For these reasons, we pointed most of the power for the beam path which is devoted to act as the STED beam. We stretched its pulses with 60-cm glass rod and 100-m polarization-maintaining optical fiber, reaching about 250 ps pulse length. On this beam path, before entering the fiber, we placed an optical delay line, made by a retroreflector mounted on a motorized linear stage, in order to temporally overlap the pulses of the two beam patches in the focal area. We chose the final position of the delay line by optimizing the maximum fluorescence depletion obtained in an aqueous dye solution. For the two-photon excitation beam path we kept the time pulses as short as possible (ca. 220 fs, Fig. 1B), monitoring them at the microscope scanning head entrance by an FR-103WS autocorrelator (Femtochrome Research). Two-photon excitation and STED beam pathways are recombined by a PBS before entering the infrared port of the confocal unit of the microscope (Leica TCS STED-CW; Leica Microsystems).

Results

In classical STED microscopy, the wavelength of the STED beam has to be chosen taking into account two competing effects, namely, the efficiency of STED decreases when moving far away from the emission peak and the probability to excite directly the fluorophore increases when moving close to the emission peak. Because in the SW implementation the excitation beam relies on the same wavelength, one has to consider that the cross-section of two-photon excitation has to be maximized for the specific fluorophore being used in order to provide an adequate brightness. Thus, we performed experiments overlaying the 2PE and the STED beams, both Gaussian shaped, in an aqueous solution of ATTO647N (ATTO-TEC).

We evaluated the excitation induced by the STED beam and the efficiency of nonlinear excitation and of the depletion as function of wavelength and power. Fig. 2A reports the wavelength dependencies while keeping the 2PE and STED intensities constant. The black line shows the fluorescence generated by the STED beam without the presence of the 2PE beam and well represents the red tail of the one photon excitation spectra of ATTO647N. The trend of the fluorescence excited by the 2PE beam shows that, between 720 and 740 nm, the dominating effect is one photon event. After that, the trend deviates from the linear excitation graph and shows a relative peak around 840 nm. For comparison we plotted (Fig. 2A, Inset) the linear excitation spectra of ATTO647N (data taken from ATTO-TEC) that interestingly shows a peak around 420 nm matching the 2PE peak. Within the same experiment series, we measured the fluorescence behavior while overlaying the 2PE and the STED beam (green triangles, Fig. 2A). An efficient reduction of fluorescence under STED condition is observed for wavelengths greater than 750 nm.

Fig. 2.

(A) 1PE spectrum by means of the STED beam (black line), 2PE spectrum (red dots), and depletion effect on the nonlinear excited fluorescence (green triangles) of ATTO647N. In the inset, the complete 1-photon excitation spectrum of ATTO647N. (B) Fluorescence intensity versus 2PE laser beam power (red dots), effect of the STED beam on the nonlinear excited fluorescence (green triangle), and ratio between the former and the latter, respectively (blue squares). (C) Depletion of the fluorescence, 2P excited and depleted at 770 nm, as a function of the STED beam power P_STED, corrected for direct excitation of the depletion beam. Measurement of I_sat ∼ 10 MW cm^-2 is comparable with those already published.

Fig. 2B shows the log–log plot of the fluorescence intensity in dependence on the 2PE power while keeping STED power and STED wavelength constant. The graph for the 2PE, in absence of any STED beam (red dots), clearly shows the expected slope for two-photon excitation (slope = 1.7 ± 0.1). Superimposing a STED beam to the same measurement (green triangles, Fig. 2B) shows the limitations of the excitation power value in order to perform high-resolution imaging: For excitation powers greater than 2.5 mW, the fluorescence generated by the STED beam itself is, by way of comparison, so high that STED imaging would not be possible under such conditions. For excitation power values greater than 4 mW, the fluorescence value can be sufficiently reduced by superimposing the STED beam. For higher excitation powers, the depletion efficiency (blue squares, Fig. 2B) becomes almost constant, although at higher power level saturation of the excited state becomes relevant.

We analyzed the depletion efficiency by measuring the fluorescence value depending on the STED beam intensity as reported in Fig. 2C. Here, the 2PE power was kept constant. The curve shows the fluorescence reduction due to the STED effect, representing a key experiment for this approach. In fact, because the fluorescence value results effectively reduced by using the same wavelength for excitation and STED beam, high-resolution imaging with the same scheme should be possible. The efficiency of the stimulated emission process requires quantitative analysis by measuring the saturation intensity I_sat, defined as the intensity needed to deplete half of the excited molecules (6). The measured value of I_sat ∼ 10 MW cm^-2 is in agreement with previously reported results for ATTO647N (16).

Under a typical imaging configuration, the focused excitation beam is superimposed by a donut-shaped STED beam, hereby realized by the use of a vortex phase plate (RPC Photonics) in the STED path (Fig. 1). We then compared SW 2PE-STED imaging mode (Fig. 3C) with classical confocal (Fig. 3B) and 2PE imaging (Fig. 3A) of Potoroo kidney cell line (PTK-2) cells in which microtubules are immunolabeled with ATTO647N. The raw data immediately confirm the benefit of the SW 2PE-STED technique—i.e., the resolution is remarkably enhanced. Measuring the FWHM of several microtubules, we determined that the resolution achieved by the current implementation of SW 2PE-STED is approximately 80 nm. This limit is of experimental origin, mainly due to the maximum power available in our setup. Further optimization, in terms of the optical STED beam pathway, will increase the STED effect, hence the resolution. It is worth noting that substructures within the cellular complex of the cytoskeleton can only be resolved using the superresolution achievable by the SW 2PE STED technique, whereas both confocal and 2PE imaging cannot reveal such structures. In fact the resolution gain hereby attained is three to four times with respect to confocal and up to four to five times with respect to the best 2PE performances.

Fig. 3.

Immunofluorescence 2PE-STED single-wavelength microscopy shown in comparison with confocal and 2PE microscopy (A–C). (A) Confocal (excitation at 633 nm), (B) 2PE (excitation at 770 nm), and (C) 2PE-STED (excitation and depletion at 770 nm) micrographs of microtubules immunostained with the dye ATTO647N. (Scale bars: 5 μm.) The lower panels respectively show the raw data in the corresponding boxed areas and the plots of a line profile (black squares) along the arrows indicated in the zoomed images, together with a multiple peak fit (red). (Scale bars: 1 μm.) We acquired the images with a pixel size of 32 × 32 nm, a pixel dwell time, respectively, of 8 (A), 65 (B), 65 μs (C), and a pinhole size, respectively, of 1 (A), 2 (B), 2 airy unit (C). The powers of 2PE and STED beam were 12 and 168 mW, respectively, measured at the objective back plane.

Discussion and Outlook

We have developed a class of 2PE-STED using the very same wavelength for excitation and depletion. We have demonstrated that this method allows superresolved imaging on biological specimens employing the very common fluorophore ATTO647N, achieving a resolution four to five times better than conventional 2PE microscopy. The single-wavelength approach simplifies the optical scheme in terms of number of dichroic filters and laser sources, allowing an easy coupling to a conventional commercial confocal microscope.

The broadening of the cross-section in the 2PE regime makes the microscope more flexible for multicolor high-resolution imaging, which allows the simultaneous combination of STED microscopy with conventional 2PE and second harmonic generation microscopy. Recently, it has been demonstrated that the STED microscopy can be applied to the imaging of mouse brain (17, 18). Because scattering mainly depends on the wavelength, using the very same wavelength for excitation and depletion, the aberration introduced by the specimen should eventually affect both beams in the same manner. Therefore the SW 2PE-STED microscopy seems a promising technique to better actively control distortions when imaging thick, highly scattering specimens. Although the resolution achieved at certain depths essentially depends on the intensity that can be brought at the focal plane by the STED beam, we believe that this imaging technique will allow envisaging developments toward bioimaging of thick samples at nanoscale resolution.