The unrivalled combination of molecular specificity, relatively easy sample preparation and live-cell imaging compatibility makes far-field fluorescence microscopy the most popular imaging modality in cell biology. However, conventional far-field fluorescence microscopy is for many biological applications still hindered by its moderate resolution, limited by the diffraction of light . Indeed, diffraction prohibits to focus a beam more sharply than to a spot of λ/(2NA) in size, where λ is the wavelength and NA the numerical aperture of the lens. Likewise, due to diffraction, the image of a point-like object through an optical system cannot be smaller than λem/(2NA), where λem is the emission wavelength. As a result, if two fluorescent objects are closer to each other than the minimum size of the excitation spot, they are virtually excited simultaneously and since their individual images through the optical system are larger than the distance between them, it is impossible to distinguish their overlapping fluorescence signals in time and space. The resolution limit of conventional fluorescence microscopy, given a wavelength in the visible range (400–700 nm) and a maximum numerical aperture of 1.5, is therefore about hundreds of nanometers.
It is an undeniable fact that sub-diffraction microscopy (or nanoscopy), seeded two decades ago by the introduction of Stimulated Emission Depletion (STED) microscopy , has revolutionized our approach to observing sub-cellular mechanisms. The diffraction limit of resolution has been overcome by taking into account the photo-physical properties of fluorescent markers into the image formation process, in order to control their ability to fluoresce sequentially in time and space . Resulting resolutions in the range of a few tens of nanometers provide insights into biological processes at the cellular and molecular scale that were hitherto unattainable [4–6].
In a STED microscopy, the sequential probing of fluorescent reporters closer than the diffraction limit is achieved by restricting the spatial extent of the volume from which fluorescence occurs by depriving the fluorescent molecules of the outer part of the excitation area of their ability to fluoresce by stimulated-emission. In practice, a so-called STED beam is shaped in a doughnut-like pattern, featuring a central zero-intensity point in the focal plane, and superimposed on the regular diffraction limited excitation spot. If the intensity of the STED beam at the doughnut crest ISTED exceeds the value Is at which half the fluorescence signal is suppressed (switch off) by stimulated emission the spatial extent of the effective fluorescence volume is confined to sub-diffraction dimensions. Scanning the co-aligned excitation and STED beams through the sample yields the final image whose resolution can be tuned by the intensity of the STED light.
Historically, STED microscopy has been first demonstrated using pulsed lasers for both, the excitation and the STED beams (all-pulsed-STED implementation). In that case STED pulses reach the focal plane virtually simultaneously with – or a few picoseconds after – the excitation pulses so as to instantly inhibit fluorescence emission from potentially excited molecules. Moreover, it turned out that the performance of the pulsed modality is optimized when the STED pulses are in the hundred picoseconds range . Although time-alignment and pulse length adjustment are routinely handled in many laboratories, the apparent sophisticated pulse preparation hampered the wider use of the all-pulsed-STED configurations. The implementation of the technique with a CW STED beam (in combination with a pulsed or a CW beam for excitation) greatly simplifies the system [8, 9], since no laser pulse preparation is required. But as a drawback, its potential has been so far limited by the poorer fluorescence on-off switching contrast, thus, by the less efficient spatial confinement of the fluorescence than in the all-pulsed-STED systems. This results in a blurring effect in the final images and in a reduced effective resolution .
Fortunately, a recent investigation of the time-course of the fluorescence and stimulated-emission probabilities across the focal volume has revealed the possibility to further increase the resolution of a CW-STED implementation by selecting in time only photons contributing to the high spatial resolution information [11–12]. We denote this new implementation as gated STED (gSTED). The following section provides an overview of the basic principle of the gCW‑STED implementation and illustrates the strong benefits of this promising approach.
Unlike in the all-pulsed-STED, where the STED pulse action is concentrated in time immediately after the excitation, and usually over a period much shorter than the excited-state lifetime τfl of the fluorophore, in the CW-STED implementation the instantaneous STED intensity and thereby the instantaneous probability of stimulated de-excitation is comparatively low. As a consequence, a non-negligible part of the fluorophores emit fluorescence before having been exposed to much of STED light (Figure 1a). Roughly speaking this is because the suppression of spontaneous emission (fluorescence) does not depend only on the average STED intensity but rather on the amount of STED photons an excited molecule sees (on average) before a possible spontaneous emission. In terms of imaging, this early spontaneous emission results in a residual fluorescence outside the zero-intensity point of the STED pattern and thereby in a blurring effect in the final images. Based on this insight, this problem can be solved by combining CW-STED, pulsed excitation and time-gated detection for recording fluorescence only after a certain time-delay Tg after excitation such that fluorophores under STED illumination have time to see enough STED photons for not contributing to the fluorescence signal anymore (Figure 1a). Selecting photons after a detection delay Tg after excitation enhances the effective fluorescence on-off contrast (Figure 2b) that is critical to attaining subdiffraction resolution.
Fig. 1: Principles of gCW-STED microscopy. (a) Time-evolution of the fluorescence signal in absence (ISTED = 0) and presence (ISTED = 5 × Is) of the STED light with experimental time sequence (upper pannel). (b) Detected fluorescence signal as function of the STED intensity for increasing recorded delay time Tg. In CW-STED implementation time-correlated information is discarded and the whole signal is registered.
Another formulation is that the STED reduces the average time that a fluorophore spends in the excited state, i.e., its effective excited-state lifetime τ is mainly determined by the de-excitation rate of stimulated emission, which increases linearly with the STED intensity. Consequently, in the doughnut-shaped pattern (Figure 2a), the excited-state lifetime of a molecule changes according to its position. In particular, the excited-state lifetime decreases away from the zero‑intensity point, reaching a minimum in the proximity of the maximum STED intensity (Figure 2b). This excited-state lifetime signature can be used to reject photons emitted by short-lived excited state fluorophores, from the periphery, and thus to select only photons emitted by long-lived excited state fluorophores located close to the zero-intensity point.
The improvement of gCW-STED over standard CW-STED is evidenced by comparing their effective point-spread functions (E-PSFs) (Figure 2c), namely the volume from which the fluorescence signal is acquired. A sharpening of the E-PSF is observed with increasing the time delay Tg of the gated detection. Time-gated detection acts like a spatial filter reducing the E-PSF amplitude at the periphery and hence reducing the E-PSF pedestal. Removing the pedestal inherently also reduces the full-width half-maximum (FWHM). This concomitant reduction of FWHM and suppression of pedestal translates immediately in a strong improvement of the imaging local contrast, and thereby an improvement of the effective resolution, without notably expanding the core bandwidth of image frequencies per se.
Fig. 2: (a) Schematic drawing of the gCW-STED setup with pulsed excitation and CW-STED limited Gaussian and doughnut-shaped focal intensity distribution, respectively. Fluorescence light (Green) is detected by the objective lens and imaged onto an avalanche-photon-detector the excitation pulses (Trigger). (b) Calculated lateral intensity distribution for the excitation beam (blue), the effective region in which the dyes are allows to fluoresce (green) with the relative excite-state lifetime (black). The lower panel depicts the two-dimensional counter plot of the effective fluorescence area and the relative excite-state lifetime. (c) Calculated lateral intensity distribution for the gCW-STED effective PSF for increasing time delay Tg of the gated detection. The lower panel depicts the two-dimensional counter plot of the effective PSF. Scale bars 100 nm.
Time-gated detection also rejects "desired" photons, namely those that are emitted during Tg from the doughnut center. Therefore, the increase in the effective resolution has to be pondered against the reduction in signal. A longer acquisition time can compensate for the concomitant decrease in signal-to-background or -noise ratios.
Gated CW-STED microscopy can be realized by offline processing of time-correlated single‑photon counting recordings or, like in the present case, in real time using a fast electronic gate. To test the improvement induced by the alteration reported here, we compared CW-STED and gCW-STED images of fixed PtK2 cells whose vimentin filaments were immunolabeled with three-different dyes (Figure 3), Alexa Fluor 488, Oregon Green and Chromeo 488, respectively. The gCW-STED images are clearly superior in contrast and detail. The lower CW-STED power required to obtain subdiffraction images of the same clarity (200 mW of 592 nm light compared to >600 mW in previous recordings of similar samples) highlights the potential of gCW-STED nanoscopy for live-cell imaging. Motivated by these results, we imaged a living PtK2 cell with keratin filaments tagged with the yellow fluorescent protein Citrine (Figure 4). Structural details of this network could much better be highlighted in the gCW-STED than in the CW‑STED or confocal images. Notably, the reduction of the signal strength due to the time-gated detection and the consequent decrease of the signal-to-noise and -background ratios do not deteriorate drastically the quality of the images.
Fig. 3: gCW-STED nanoscopy on fixed cells. Scanning fluorescence images of vimentin filaments in fixed PtK2 cells labeled with the organic dyes Alexa 488 (a), Oregon Green (b) and Chromeo 488 (c): confocal (left), CW-STED (middle right) and g-STED (middle left) recordings. The insets show a magnified view of the marked areas, renormalized in signal intensity. The right panels depict normalized intensity profiles along the dashed line of the insets demonstrating the distinction of features as small as <60 nm. Excitation: 485 nm, 80 MHz and 10 μW average power; STED: 592 nm, CW and 200 mW average power; gated detection: Tg = 1.5 ns. Scale bars 1 µm. Figure is modified, with permission, from Ref. .
Fig. 4: gCW-STED nanoscopy on living cells. Scanning images of keratin labeled with the fluorescent protein citrine in a living PtK2 cell for confocal (left panel), gCW-STED (middle left panel) and CW-STED (middle right panel) recordings. The insets show a magnified view of the marked areas, renormalized in signal intensity. The right panel depicts normalized intensity profiles along the dashed line of the insets demonstrating the separation of features as small as 60 nm. Excitation: 485 nm, 80 MHz and 11 μW average power; STED: 592 nm, CW and 200 mW average power; gated detection: Tg = 1.5 ns. Scale bars 1 μm. Figure is modified, with permission, from Ref. .
Conclusion and discussion
The fluorescence on-off switching contrast increases with the duration of the STED beam action. Time-gated detection uses this fact to improve the effective resolution of CW-STED microscopy. The improvement is attained without increasing the intensity ISTED; in fact, gating facilitates reducing ISTED in practical imaging. The main practical limitation of the gCW‑STED implementation is the inherent loss of signal and concomitant reduction of both the signal to noise and signal to background ratios. Even so, we have shown that for typical parameters, time-gated detection greatly improves the effective resolution in CW-STED imaging, and helps to reveal finer structure detail in the sample.
Notably, time gated detection alleviates the performance difference between all-pulsed-STED and CW-STED implementations. Moreover, gCW‑STED reduces the instantaneous intensity required to obtain images with a given sub-diffraction resolution: For attaining the same effective resolution, we applied ISTED = 100 MW cm−2, which is ~10 times lower than in typical pulsed systems. In fact, of all STED modalities reported so far, gCW-STED provides the sharpest images with the lowest instantaneous intensity. The intensity demand is further reduced when gCW-STED microscopy is combined with AStEx-STED : Shifting the STED beam wavelength closer to the emission maximum of the fluorophore reduces the intensity required to obtain images with a given sub-diffraction resolution. On the other hand, the fluorescence background stemming from the anti-Stokes excitation (AStEx) of the fluorophores by the STED beam compromises the image local contrast. A synchronous photon counting (lock-in mode) method removes the AStEx background and fully recovers the imaging local contrast.
Finally, time-gated detection allows also combining CW-STED microscopy with fluorescence correlation spectroscopy (