We report a method to increase the efficiency of detecting nonlinear fluorescence signals in saturated excitation (SAX) microscopy. With this method, we compare fluorescence signals obtained under different degrees of saturated excitation to extract the nonlinear fluorescent signal induced by saturated excitation. Compared to conventional SAX microscopy using the harmonic demodulation technique, the detection efficiency of the fluorescence signal can be increased up to 8 and 32 times in imaging using the second-order and the third-order nonlinear fluorescence signals, respectively. We combined this approach with pulsed excitation, which is effective to reduce photobleaching effects, and achieved super-resolution imaging using third-order nonlinear fluorescence signals induced by saturated excitation of an organic dye. The resolution improvement was confirmed in the observations of fluorescent beads, actin-filaments in HeLa cells, and a spine in mouse brain tissue.

Optical microscopy is a technique suitable for biological investigation because it enables visualization analysis inside specimens with advantageous features such as non-contact, low damage, and high precision of the submicron order. Classically, the spatial resolution of optical microscopy is limited by the wave nature of light. However, the recent developments in super-resolution microscopy, such as photoactivated localization microscopy (PALM),1 fluorescence photoactivation localization microscopy (FPALM),2 stochastic optical reconstruction microscopy (STORM),3 stimulated emission depletion (STED) microscopy,4 reversible saturable optical fluorescence transitions (RESOLFT) microscopy,5 and structured illumination microscopy (SIM),6 have successfully diminished the diffraction barrier of the spatial resolution.

In our previous study, we have proposed and demonstrated saturated excitation (SAX) microscopy as a super-resolution technique.7 In SAX microscopy, the spatial resolution is enhanced in three dimensions by detecting the nonlinear fluorescence signal induced by saturation, which is localized within the focal spot. SAX microscopy can be realized by a simple modification of a conventional confocal microscope through the addition of a laser-intensity modulator and a lock-in amplifier. As often demonstrated in two-photon excitation microscopy, the use of nonlinear fluorescence is also advantageous for volumetric imaging of thick samples, which is not an easy task for many super-resolution techniques. By using SAX microscopy, three-dimensional imaging of cell clusters has been demonstrated.8 The saturation of the excitation process has also been combined with point spread function (PSF) engineering to effectively enhance the high spatial frequency components in fluorescence images.9–11 

Here, we report the improvement of signal-to-noise ratio (SNR) in SAX microscopy by applying a differential excitation technique. In the first demonstration of SAX microscopy, the harmonic demodulation technique was utilized to extract the nonlinear fluorescence response during the laser scanning process, which is useful to avoid post-processing of image data but required the installation of a light modulator and a lock-in amplifier. Later, as a simpler method, Humpolíčková et al. presented a method to extract the nonlinear signals from fluorescence measurement with differential excitation, where two or more fluorescence images are obtained at different excitation intensities in order to calculate nonlinear signals induced by saturated excitation.12 Differential-excitation SAX (dSAX) microscopy was also demonstrated with two-photon excitation microscopy11 and with a commercial laser scanning microscope.13 In this report, we demonstrate that the technique is also effective to improve the SNR in SAX microscopy. Since the SNR of fluorescence detection is one of the factors that determine the spatial resolution in SAX microscopy, the improvement of the SNR is effective to improve the spatial resolution in practice. In the present study, we combined pulsed excitation with the differential excitation technique and successfully extracted the 3rd-order nonlinear fluorescence responses under saturated excitation of fluorescence. The pulsed excitation used here enables the extraction of the higher-order nonlinear response by reducing the photobleaching effect, which normally becomes severe at high intensities. We examined the technique experimentally and demonstrate its effectiveness in the observation of biological cells and tissues and achieved fluorescence imaging with the 3rd-order nonlinear fluorescence signals induced by saturated excitation of fluorescent dyes.

Since the principle of dSAX microscopy has been introduced in single- and two-photon excitation in previous papers,11–13 we only briefly introduce it here. Figure 1(a) shows the relationship between fluorescence and excitation intensities, in which saturation effect is observed at high excitation intensities. The improvement of spatial resolution in SAX microscopy relies on the saturation effect, which enables the extraction of fluorescence signals from a volume smaller than the laser focus by exploiting the nonlinear fluorescence response. In our first report on SAX microscopy, the extraction of nonlinear fluorescence responses was accomplished by employing a harmonic demodulation technique.7 On the other hand, in the differential excitation techniques, the fluorescence signal that is nonlinearly proportional to the excitation intensity is extracted by comparing fluorescence signals under different degrees of excitation saturation. The measured fluorescence IFL will be reduced from the linear fluorescence response that would be obtained in the absence of saturation IL, in particular at the center of the PSF, as shown in Fig. 1(b). The nonlinear component INL can therefore be extracted by subtracting the measured fluorescence signal from the estimated linear signal that would have been obtained in the absence of saturation. Figure 1(c) shows the profile of the PSF given by the subtraction of IFL from IL.

FIG. 1.

Principle of SAX microscopy with differential excitation. (a) Relationship between fluorescence and excitation intensity. (b) Comparison of linear and saturated fluorescence distribution in the laser focus. (c) Extracting the nonlinear term provides a narrower fluorescence spot in the focus.

FIG. 1.

Principle of SAX microscopy with differential excitation. (a) Relationship between fluorescence and excitation intensity. (b) Comparison of linear and saturated fluorescence distribution in the laser focus. (c) Extracting the nonlinear term provides a narrower fluorescence spot in the focus.

Close modal

Under high excitation intensity, the saturated fluorescence signal can be expressed as a sum of the linear component and non-linear components in the signal as shown in Eq. (1),

(1)

IFL, an, and Iex represent the fluorescence intensity, the coefficient of each fluorescence component, and the excitation intensity, respectively. IL and INL represent the linear and nonlinear terms in the equation. The red plot in Fig. 2(a) shows the relationship between fluorescence and excitation intensity calculated with Rhodamine 6G as a sample.14,15 In Fig. 2(a), the nonlinear term, INL, appears as a deviation from the response without saturation. In SAX microscopy, the nonlinear term is extracted from the fluorescence signal detected in each sample position to construct fluorescence images with improved spatial resolution.

FIG. 2.

(a) Relationships between fluorescence signal and excitation intensities calculated using Rhodamine 6G as a sample. Point-spread-functions (PSFs) in detecting (b) linear, (c) 2nd- and (d) 3rd-order nonlinear signals with a confocal setup. Line profiles along the x-axis from the center of the PSF are shown in (e).

FIG. 2.

(a) Relationships between fluorescence signal and excitation intensities calculated using Rhodamine 6G as a sample. Point-spread-functions (PSFs) in detecting (b) linear, (c) 2nd- and (d) 3rd-order nonlinear signals with a confocal setup. Line profiles along the x-axis from the center of the PSF are shown in (e).

Close modal

With a relatively low degree of saturation, we can omit the higher order nonlinear-terms in Eq. (1), resulting in an equation containing only the linear and the 2nd order-nonlinear term, INL(2). Under this condition, the nonlinear term can be simply calculated by using fluorescence signals measured at two different excitation intensities as shown in Eq. (2),

(2)

where Ifl1 and Ifl2 are the measured fluorescence signals at the excitation intensities of Iex1 and Iex2, respectively, which provide different saturation degrees in the excitation. The green plot in Fig. 2(a) shows the 2nd-order nonlinear term calculated by Eq. (2) at different Iex2 while setting Iex1 set at 0.01 kW/cm2. The calculation results show a quadratic relation to the excitation intensity. We can also see the deviation of the estimated nonlinear signal from the quadratic plot. This is due to the rise of higher-order nonlinear terms.

With a higher degree of saturation, the higher-order nonlinear terms can be extracted in a similar way. For example, fluorescence signals obtained with three different excitation intensities (Iex1, Iex2, and Iex3) can be used to obtain the 3rd-order nonlinear component INL(3) by the following equation:

(3)

where Ifl1, Ifl2, and Ifl3 are the fluorescence signals obtained by the excitation intensities of Iex1, Iex2, and Iex3, respectively. The blue plot in Fig. 2(a) shows the relationship between the nonlinear signals extracted by fluorescence signals obtained with three different excitation intensities. In the calculation, Iex1 and Iex2 were set to 0.01 and 1 kW/cm2, respectively. The calculated fluorescence response is proportional to the cube of the excitation intensity, confirming that fluorescence measurement with three different excitation intensities can extract the third-order nonlinear fluorescence signal.

Next, the excitation PSF of SAX microscopy is calculated by estimating the fluorescence signal extracted by the above differential excitation technique. The effective PSF is obtained by multiplying the excitation and detection PSF that is determined by the detection optics. Figures 2(b)2(d) show the effective PSFs in conventional confocal (linear) and SAX microscopy with confocal detection (2nd- and 3rd-order nonlinear) calculated for observations using a water immersion objective lens with an NA of 1.2 and an excitation wavelength of 532 nm. We assumed that the wavelengths of excitation and emission are the same. The relation between excitation and fluorescence intensities was estimated using the calculation used for Fig. 2(a). Excitation intensities were 0.01, 1, and 10 kW/cm2 at the center of the excitation spot for Figs. 2(b)–2(d), respectively. Figure 2(e) shows the profile in the lateral direction of the PSFs. The comparison of the images and the profiles confirms that the size of the PSF becomes smaller with an increasing order of nonlinearity of the extracted signals.

Compared to SAX microscopy using harmonic demodulation, the differential excitation technique can detect a larger amount of nonlinear signals that contribute to the resolution improvement. In SAX microscopy with harmonic demodulation, the excitation intensity is modulated at a frequency of f, while the 2nd and 3rd-order nonlinear components are extracted by demodulating the fluorescence signal at the harmonic frequencies of 2f and 3f, respectively. However, as shown in Eq. (4), the nonlinear signals are distributed throughout the DC, fundamental, and harmonic frequencies. Therefore, the signal that can be extracted by the harmonic demodulation is smaller than the value that is actually contained in the fluorescence signal. Equations (4) and (5) show the model for the 2nd and 3rd order nonlinear terms (INL_HD2,INL_HD3) in the harmonic demodulation SAX technique, where only the third or fourth terms of Eqs. (4) and (5), which correspond to 1/8 and 1/32 of the total signals, are used for reconstructing the fluorescence image,

(4)
(5)

On the other hand, exploiting differential excitation as described here, with the above mathematical procedure, can extract the nonlinear components in the signal without such a loss. Therefore, the 2nd- and 3rd-order nonlinear signals in dSAX microscopy can be increased by factors of 8 and 32, respectively. Defining the SNR as the ratio of the signal and the shot noise, the SNR is improved by factors of √8 and √32 for detecting the 2nd- and 3rd-order nonlinear signals under the shot-noise limited condition. Since the effective optical-transfer-function (OTF) of laser scanning microscopy typically has strong low-pass characteristics, the increase in SNR is beneficial to improve the spatial resolution in practical conditions. This benefit is also available in saturated two-photon excitation microscopy.11 

Another benefit of using differential excitation is that we can freely shape the temporal profile of excitation intensity to effectively induce the saturated excitation and reduce photobleaching. In Sec. IV, we describe a technique to induce saturated excitation effectively by using pulsed laser light. This benefit is also available in saturated two-photon excitation, where 16 times increase in the 4th-order nonlinear signal can be expected with the differential excitation technique. Since the increase in excitation intensity may significantly damage samples, the increase in the detection signal is beneficial in applying saturated two-photon excitation microscopy for observation of biological samples.

We also found the difference in the fluorescence response curve between the two SAX implementations. The nonlinear signals obtained by the differential excitation technique show weaker saturation at high excitation intensity, compared with the previous report using harmonic demodulation.7 This property of the response curve allows us to increase the SNR while keeping high spatial resolution, which is difficult with the harmonic demodulation technique because the increase in excitation intensity eventually results in the decrease in the spatial resolution due to the saturation of demodulated signals.

We examined the use of pulsed laser light for fluorescence excitation in order to induce saturated excitation effectively. Since a shorter pulse has higher peak intensity with the same average intensity, the use of pulsed laser light reduces the total amount of light exposure to achieve saturated excitation. This allows us to reduce photobleaching, which typically makes it difficult to use a higher excitation intensity to achieve a higher spatial resolution in SAX microscopy.

We experimentally measured the relations between the average excitation intensity and the fluorescence signal with and without pulsed excitation. A fluorescence solution of Rhodamine 6G (Wako) with a concentration of 100 μM was used as a sample. A conventional confocal microscopy setup equipped with a CW diode laser with a wavelength of 532 nm (Samba TM, Cobolt) for a light source and a photomultiplier tube (H7422-40, Hamamatsu Photonics K.K.) were used for signal detection. Square pulses of laser light were generated by using an acousto-optic modulator (AOM, AOM-402AF1, IntraAction) and a function generator. A water-immersion objective lens with an NA of 1.2 (UPLSAPO 60XW, Olympus) was used for excitation and collection of fluorescence.

Figure 3(a) shows the pulse shapes generated by the AOM. In Fig. 3(b), the fluorescence signal measured from the four different excitation pulses is plotted. As expected, the square pulse with shorter width effectively induces saturation of the fluorescent molecules. This result indicates that the pulsed excitation is beneficial in inducing saturated excitation at lower light intensities. We also experimentally and theoretically confirmed that the pulsed excitation used in this experiment does not cause significant nonlinear photobleaching, as shown in Figs. S1 and S2.

FIG. 3.

(a) The shapes of pulsed laser light generated by the AOM. (b) The relationships between the fluorescence signal and averaged excitation power with different excitation pulse shapes. The exposure time for each measurement was 500 ms.

FIG. 3.

(a) The shapes of pulsed laser light generated by the AOM. (b) The relationships between the fluorescence signal and averaged excitation power with different excitation pulse shapes. The exposure time for each measurement was 500 ms.

Close modal

These results suggest that pulsed excitation effectively suppresses photobleaching effects by reduction in the light exposure, which allows us to extract a higher-order nonlinear fluorescence response with a higher SNR. The benefit in the reduction of photobleaching can be obtained easily in dSAX microscopy. On the other hand, the use of pulsed excitation might bring complexity in harmonic-demodulation SAX microscopy.

The use of pulsed excitation can be equivalent to shortening the pixel dwell time in laser scanning microscopy. However, the use of a shorter pixel dwell time decreases the total amount of the fluorescence signal, which may cause difficulty in keeping the high SNR required to extract the nonlinear signals in SAX microscopy. On the other hand, repetitive pulsed excitation allows control over the signal amount by changing the pixel dwell time, which is useful in practical conditions since one can independently tune the efficiency of saturation and the signal amount.

We experimentally confirmed that the differential excitation technique can extract the nonlinear fluorescence signal induced by pulsed excitation. The optical setup used was the same one used in the measurement of the saturation efficiency. In Fig. 4, the plot in red indicates the measured fluorescence signal. The plots in green and blue indicate, respectively, the 2nd- and 3rd-order nonlinear fluorescence signal extracted from the measured signals by using Eqs. (2) and (3). To obtain the 2nd-order nonlinear signal, the fluorescence signal obtained at 0.9 kW/cm2 was used to estimate the linear response. The fluorescence signals obtained at 0.9 and 14 kW/cm2 were used to estimate the linear and the 2nd-order nonlinear responses to extract the 3rd-order nonlinear components. As indicated by the dotted lines in Fig. 4, the extracted nonlinear signals are proportional to the square and cubic of the excitation, which confirms that the differential excitation technique can extract nonlinear signal components induced by saturation.

FIG. 4.

Relationship between the fluorescence signal and excitation intensity measured with a solution of Rhodamine 6G. The plot in red shows the measured fluorescence signals. The plots in green and blue show the 2nd- and 3rd-order nonlinear fluorescence signals extracted from the measured signals, respectively.

FIG. 4.

Relationship between the fluorescence signal and excitation intensity measured with a solution of Rhodamine 6G. The plot in red shows the measured fluorescence signals. The plots in green and blue show the 2nd- and 3rd-order nonlinear fluorescence signals extracted from the measured signals, respectively.

Close modal

We observed polystyrene fluorescent beads to confirm the improvement of the spatial resolution and the SNR in SAX microscopy with differential excitation. Fluorescent beads with a diameter of 100 nm were dispersed on a coverslip. After being dried, the beads were embedded in an antifade reagent (S36937, Thermo Fisher Scientific). Fluorescence images were obtained by a typical confocal microscopy setup equipped with an AOM, and the sample was scanned by a piezoelectric stage. The sample was excited at three different excitation intensities, which are 69 W/cm2, 190 W/cm2, and 540 W/cm2. To reduce the photobleaching effect, we used pulsed exposure with a pulse width and a period of 20 µs and 100 µs, respectively. An oil-immersion objective lens (UPlanSApo, NA1.4, ×100, Olympus) was used for excitation and collection of fluorescence. The pixel size and dwell time were 26 nm and 1 ms, respectively, for all images. The pinhole size was 0.5 Airy unit.

Figures 5(a)–5(c) show the fluorescence images of the beads reconstructed by the linear, 2nd-, and 3rd-order nonlinear signals, respectively. The intensity profile of the bead images indicated by the arrowheads in Figs. 5(a)–5(c) is shown in Figs. 5(d)–5(f). The profiles shown are representative of those measured from 10 different isolated beads in the image. From all 10 profiles, the averaged FWHM values of linear, 2nd-, and 3rd-order nonlinear images were 165 ± 6.9, 158 ± 8.5, and 145 ± 11.3 nm, respectively. This comparison confirms that the nonlinear signals extracted by the differential excitation can improve the spatial resolution of confocal fluorescence microscopy.

FIG. 5.

Fluorescence images of fluorescent beads with a diameter of 100 nm reconstructed using the (a) linear, (b) 2nd-, and (c) 3rd-order nonlinear signals. The intensity profiles of the beads indicated by the arrowheads in (a)–(c) are shown in (d)–(f). No image processing was applied except for the extraction of nonlinear signals.

FIG. 5.

Fluorescence images of fluorescent beads with a diameter of 100 nm reconstructed using the (a) linear, (b) 2nd-, and (c) 3rd-order nonlinear signals. The intensity profiles of the beads indicated by the arrowheads in (a)–(c) are shown in (d)–(f). No image processing was applied except for the extraction of nonlinear signals.

Close modal

We also experimentally compared the SNR in SAX microscopy between differential excitation and harmonic demodulation. Fluorescent beads with a diameter of 200 nm were observed by the two detection modes. For SAX imaging with harmonic demodulation, we used the same optical setup shown in our previous report for excitation modulation,7 where the excitation intensity was modulated at 20 kHz, and the fluorescence signal was obtained through a lock-in amplifier. To compare under the same degree of saturated excitation, the excitation intensity was modulated at 20 kHz also for the differential excitation mode because the saturation efficiency is dependent on the modulation frequency.14 Fluorescence images were acquired using an oil immersion objective (UPlanSApo, NA1.4, ×100, Olympus) through a pinhole with a size of 0.5 Airy unit.

Figures 6(a) and 6(b) show the fluorescence images of the beads obtained by the harmonic demodulation and the differential excitation techniques, respectively. The second-order nonlinear signals were used to reconstruct the images. The comparison of the images shows that the SNR is higher in the image obtained by the differential excitation. The improvement of SNR was confirmed by the spatial frequency distribution of the images, as shown in Fig. 6(c), which was obtained by 2D Fourier transform of the images and averaging the values at the same frequencies for all directions. In Fig. 6(c), we can confirm that the noise level, which appears at the high frequency, is about 3 times smaller in the differential excitation mode, indicating approximately 3-fold improvement of the SNR. Assuming that the noise floor consists of the photon-shot noise, which is given as the square root of the signal, this improvement of SNR corresponds to about 8 times increases in the signal amount as expected from Eq. (4). The total image acquisition time in dSAX imaging is twice longer than that of SAX imaging with harmonic demodulation. However, the nonlinear signal is not included in the images with the lower excitation intensity in dSAX imaging. Therefore, the above experimental condition matches that for the theoretical investigation. Allowing twice longer exposure time in harmonic-demodulation SAX microscopy provides an increase in SNR by a factor of √2, which is still significantly lower than the improvement obtained by dSAX microscopy. For the above comparison, we confirmed that there is no significant noise at the modulation and demodulation frequencies, shown in supplementary material Fig. S3. Additionally, the fluctuation in laser intensity over 1 s with a measurement window of 1 ms was measured to be 0.024%, which is given by dividing the mean power by the standard deviation. For comparison, the fluctuations in fluorescence signals were estimated to be 0.11% or larger, considering the photon shot noise in the bead measurements performed for Fig. 6(b). Overall, these values are not expected to significantly affect the measurements.

FIG. 6.

Fluorescence images of fluorescence beads obtained by SAX microscopy with (a) harmonic demodulation and (b) differential excitation. The fluorescence image in (a) was obtained with an averaged excitation intensity of 0.22 kW/cm2. The fluorescence image in (b) was reconstructed from two images obtained with the averaged excitation intensities of 76 W/cm2 and 0.22 kW/cm2. The pixel size and dwell time were 39 nm and 1 ms for both measurements. (c) The distribution of the spatial frequency obtained by Fourier transform of (a) and (b). The dark lines are the results of moving average on the frequency axis with a window of 10 data points.

FIG. 6.

Fluorescence images of fluorescence beads obtained by SAX microscopy with (a) harmonic demodulation and (b) differential excitation. The fluorescence image in (a) was obtained with an averaged excitation intensity of 0.22 kW/cm2. The fluorescence image in (b) was reconstructed from two images obtained with the averaged excitation intensities of 76 W/cm2 and 0.22 kW/cm2. The pixel size and dwell time were 39 nm and 1 ms for both measurements. (c) The distribution of the spatial frequency obtained by Fourier transform of (a) and (b). The dark lines are the results of moving average on the frequency axis with a window of 10 data points.

Close modal

Improvement of the spatial resolution was also confirmed in biological specimens. Actin-filaments in fixed HeLa cells were observed using dSAX microscopy. HeLa cells were incubated on a cover slip placed in a plastic-bottom dish for 24 h with Dulbecco’s Modified Eagle’s Medium (DMEM). After incubation, the culture medium was removed and cells were rinsed with phosphate buffered saline (PBS). Cells were fixed with 3.7% formaldehyde for 3 min at room temperature and permeabilized by surfactant (Triton X-100, MP Biomedicals) for dye penetration through the plasma membrane. After the removal of the surfactant, cells were immersed in a solution of ATTO Rho6G phalloidin with a concentration of 5% for 40 min to stain filamentous structures of actin. The stained cells were mounted between a coverslip and slide glass using the antifade reagent (ProLong Gold Antifade Reagent, Life Technologies), following washing out the fluorescent dyes. An oil immersion objective lens (UPlanSApo, NA1.4, ×100, Olympus) was used in this observation. The pinhole size was 0.5 Airy unit.

We observed the sample with three different excitation intensities to compare the fluorescence images reconstructed using the linear, 2nd-, and 3rd-order nonlinear signals as shown in Figs. 7(a)–7(c), respectively. The images in Figs. 7(d)–7(f) are the magnified images of the actin-filaments in a region indicated by the white rectangle in Fig. 7(a). Filamentous structures in the cell are more distinctly separated in the 2nd- and 3rd-order nonlinear images. This comparison clearly shows the improvement of spatial resolution in SAX microscopy with higher-order nonlinearity. Some fibers disappeared in the nonlinear images, which is presumably due to the improvement of axial resolution. For the same reason, the image contrast is improved in the nonlinear images by suppressing fluorescence signals from out-of-focus positions.

FIG. 7.

Fluorescence images of actin filaments in fixed HeLa cells. (a)–(c) show the images constructed by using linear, 2nd-, and 3rd-order nonlinear signals, respectively. The excitation wavelength was 532 nm. The pixel size and dwell time were 50 nm and 1 ms, respectively. Three images were obtained with average excitation intensities of 0.35, 1.3, and 3.8 kW/cm2 through an electric low-pass filter of 5 kHz for removing high-frequency noise. The excitation pulse width and period were 10 µs and 100 µs, respectively. (d)–(f) are the enlarged views of (a)–(c), respectively, indicated by the dotted square in (a). (g)–(i) show the intensity profile between the arrows in (d)–(f), respectively.

FIG. 7.

Fluorescence images of actin filaments in fixed HeLa cells. (a)–(c) show the images constructed by using linear, 2nd-, and 3rd-order nonlinear signals, respectively. The excitation wavelength was 532 nm. The pixel size and dwell time were 50 nm and 1 ms, respectively. Three images were obtained with average excitation intensities of 0.35, 1.3, and 3.8 kW/cm2 through an electric low-pass filter of 5 kHz for removing high-frequency noise. The excitation pulse width and period were 10 µs and 100 µs, respectively. (d)–(f) are the enlarged views of (a)–(c), respectively, indicated by the dotted square in (a). (g)–(i) show the intensity profile between the arrows in (d)–(f), respectively.

Close modal

In order to examine the capability of dSAX microscopy for the imaging of biological tissues, we observed mouse brain tissue. Adult male Arc-dVenus reporter mice, which express the destabilized fluorescence protein dVenus driven by the Arc gene promoter on the C57BL/6 background,16 were anesthetized and then perfused intracardially with 4% paraformaldehyde in phosphate-buffered saline. Brains were removed, fixed overnight at 4 °C in 4% PFA, and sliced at a thickness of 20 μm in phosphate-buffered saline on a Leica vibratome (VT 1000S). After quenching Venus fluorescence with 30% H2O2 for 30 min, the sections were incubated with 2% bovine serum albumin in Tris-buffered saline containing 0.1% Triton X-100 for 1 h at room temperature, labeled with a rabbit anti-Green fluorescent protein polyclonal primary antibody (1:1000, MBL, Nagoya, Japan) and visualized with Rhodamine 6G-conjugated goat anti-rabbit IgG secondary antibody (1:1000, Active Motif, Carlsbad, CA). The animal experiments were approved by the Animal Care and Use Committee of the Graduate School of Pharmaceutical Sciences, Osaka University. All efforts were made to minimize the number of animals used. An oil immersion objective (UPlanSApo, NA1.4, ×100, Olympus) was used for excitation and collection of fluorescence. The pinhole size was 0.5 Airy unit.

The fluorescence images of a mouse brain neuron that were reconstructed with linear, 2nd-, and 3rd-order nonlinear signals are shown in Figs. 8(a)–8(c), respectively. For the reconstruction of the x-y images, we corrected the position error among the acquired three images by using the translation function of StackReg in ImageJ without interpolation. The line profile between the arrowheads in Figs. 8(a)–8(c) is shown in Figs. 8(d)–8(f), respectively. The comparison of the image and line profiles also confirms the improvement of the spatial resolution by dSAX microscopy. We also obtained the x-z images of the same sample. As shown in Figs. 8(g)–8(i), the spatial resolutions in both the lateral and axial directions are improved by extracting higher order nonlinear signals. The line profiles in Figs. 8(j)–8(l) show the significant improvement of the axial resolution in dSAX microscopy.

FIG. 8.

Fluorescence images of a fixed spine in a mouse brain tissue. (a)–(c) show linear, 2nd-, and 3rd-order nonlinear images, respectively, reconstructed from three images obtained with the excitation intensities of 0.35, 1.3, and 3.7 kW/cm2. The pixel size and dwell time were 39 nm and 1 ms, respectively. The excitation pulse width and period were 10 µs and 100 µs, respectively. (d)–(f) are the intensity profiles indicated by two arrowheads in (a)–(c), respectively. Images (g)–(i) and the profiles (j)–(l) are obtained from the same sample but in a different scanning plane (x-z plane). Three images obtained with the excitation intensities of 0.37, 1.4, and 4.2 kW/cm2, respectively, for reconstructing (g)–(i). The pixel size and dwell time of 78 nm and 1 ms for the measurement of the x-z images, respectively.

FIG. 8.

Fluorescence images of a fixed spine in a mouse brain tissue. (a)–(c) show linear, 2nd-, and 3rd-order nonlinear images, respectively, reconstructed from three images obtained with the excitation intensities of 0.35, 1.3, and 3.7 kW/cm2. The pixel size and dwell time were 39 nm and 1 ms, respectively. The excitation pulse width and period were 10 µs and 100 µs, respectively. (d)–(f) are the intensity profiles indicated by two arrowheads in (a)–(c), respectively. Images (g)–(i) and the profiles (j)–(l) are obtained from the same sample but in a different scanning plane (x-z plane). Three images obtained with the excitation intensities of 0.37, 1.4, and 4.2 kW/cm2, respectively, for reconstructing (g)–(i). The pixel size and dwell time of 78 nm and 1 ms for the measurement of the x-z images, respectively.

Close modal

In the above experiments for imaging biological samples, we successfully achieved 3rd order nonlinear imaging using an organic dye for the first time. Thanks to the signal increase by differential excitation and reduced photobleaching by pulsed excitation, a SNR sufficient to extract the 3rd order nonlinear signals was realized. Compared with the linear images, the 3rd order nonlinear images show clear improvement in the spatial resolution for both the lateral and axial directions. This simple approach can be easily implemented in a conventional confocal setup and widely used in the fluorescence imaging of various types of biological samples.

In this paper, we reported the improvement of the SNR in fluorescence imaging with SAX microscopy by using differential excitation. In principle, the differential excitation improves the signal amount by factors of 8 and 32 in the detection of 2nd- and 3rd-order nonlinear signals compared to SAX microscopy using harmonic demodulation. The use of pulsed excitation, which allows strong saturation of fluorescence while using moderate laser power, can reduce photobleaching and achieved fluorescence imaging using 3rd-order nonlinear fluorescence signals. The benefit of the differential excitation technique is also available in two-photon excitation imaging, where a 16-fold improvement in the signal amount can be expected in detecting 4th-order nonlinear signals in saturated two-photon excitation microscopy.

One of the issues we faced with using the pulsed excitation is that the SNR of fluorescence detection without saturation was low due to the low exposure. In the demonstration above, the SNR at a lower excitation power significantly affects the SNR of the extracted nonlinear signal as discussed in Ref. 13. This issue can be solved by using different pulse widths for inducing different saturation degrees of fluorescence excitation. Changing the pixel dwell time may not be useful since it would cause the difference in the scanning position depending on the stability of the galvanometer mirror control. Other issues in the differential excitation technique are the drift of the sample position and the fluctuation of laser power during acquisition of multiple images. These may cause an error in extracting the nonlinear signal. This problem can be minimized by changing the laser scanning scheme, such as changing the excitation intensity at every point or line of scanning.

Photobleaching during image acquisition can result in a spatial resolution lower than the theoretical value. With photobleaching, the estimated linear signal at the high excitation intensity can be too large, which causes contamination of linear and nonlinear signals in the extracted signal. In such cases, it is necessary to use a corrective weighting on the estimated linear signal. This can be done by choosing a coefficient so that no negative signal appears in the extracted signal. Note that there can in practice exist negative signals due to the signal fluctuation, but one can choose the coefficient such that negative signals do not exceed the fluctuations.

It is also possible to combine the differential excitation and the harmonic demodulation techniques. It has been demonstrated that even higher nonlinear terms can be extracted by using fluorescence signals demodulated at different harmonic frequencies.17 The technique can be applied to extract linear and lower-order nonlinear signals, which can also solve the issue of the sample drift.

Since the differential excitation technique is simple and provides an improved SNR in SAX microscopy, the technique is also beneficial to SAX microscopy using coherent anti-Stokes Raman scattering18 and scattering from plasmonic nanoparticles.19 It can also be combined with image scanning techniques (ISM),20,21 which might be an optimal implementation of the SAX technique in confocal fluorescence microscopy since ISM provides theoretical resolution with improved SNR compared to conventional confocal microscopy. Since the differential excitation technique can be applied to any scanning microscopy that utilizes saturable optical effects, it can also be applied for super-resolution imaging in photoacoustic, photothermal, and harmonic generation microscopy.

See supplementary material for the investigations on photobleaching and laser power fluctuation.

This work was supported by JST CREST Grant No. JPMJCR15N3, Japan. Arc-dVenus reporter mice were kindly gifted by Dr. Shun Yamaguchi and Ms. Megumi Eguchi (Gifu University).

1.
E.
Betzig
,
G. H.
Patterson
,
R.
Sougrat
,
O. W.
Lindwasser
,
S.
Olenych
,
J. S.
Bonifacino
,
M. W.
Davidson
,
J.
Lippincott-Schwartz
, and
H. F.
Hess
, “
Imaging intracellular fluorescent proteins at nanometer resolution
,”
Science
313
(
5793
),
1642
1645
(
2006
).
2.
S. T.
Hess
,
T. P. K.
Girirajan
, and
M. D.
Mason
, “
Ultra-high resolution imaging by fluorescence photoactivation localization microscopy
,”
Biophys. J.
91
(
11
),
4258
4272
(
2006
).
3.
M. J.
Rust
,
M.
Bates
, and
X.
Zhuang
, “
Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM)
,”
Nat. Methods
3
(
10
),
793
796
(
2006
).
4.
S. W.
Hell
and
J.
Wichmann
, “
Breaking the diffraction resolution limit by stimulated emission: Stimulated-emission-depletion fluorescence microscopy
,”
Opt. Lett.
19
(
11
),
780
782
(
1994
).
5.
M.
Hofmann
,
C.
Eggeling
,
S.
Jakobs
, and
S. W.
Hell
, “
Breaking the diffraction barrier in fluorescence microscopy at low light intensities by using reversibly photoswitchable proteins
,”
Proc. Natl. Acad. Sci. U. S. A.
102
(
49
),
17565
17569
(
2005
).
6.
M. G.
Gustafsson
, “
Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy
,”
J. Microsc.
198
(
2
),
82
87
(
2000
).
7.
K.
Fujita
,
M.
Kobayashi
,
S.
Kawano
,
M.
Yamanaka
, and
S.
Kawata
, “
High-resolution confocal microscopy by saturated excitation of fluorescence
,”
Phys. Rev. Lett.
99
(
22
),
228105
(
2007
).
8.
M.
Yamanaka
,
Y.
Yonemaru
,
S.
Kawano
,
K.
Uegaki
,
N. I.
Smith
,
S.
Kawata
, and
K.
Fujita
, “
Saturated excitation microscopy for sub-diffraction-limited imaging of cell clusters
,”
J. Biomed. Opt.
18
(
12
),
126002
(
2013
).
9.
G.
Zhao
,
C.
Kuang
,
Z.
Ding
, and
X.
Liu
, “
Resolution enhancement of saturated fluorescence emission difference microscopy
,”
Opt. Express
24
(
20
),
23596
(
2016
).
10.
G.
Zhao
,
M. M.
Kabir
,
K. C.
Toussaint
,
C.
Kuang
,
C.
Zheng
,
Z.
Yu
, and
X.
Liu
, “
Saturated absorption competition microscopy
,”
Optica
4
(
6
),
633
636
(
2017
).
11.
R.
Oketani
,
A.
Doi
,
N. I.
Smith
,
Y.
Nawa
,
S.
Kawata
, and
K.
Fujita
, “
Saturated two-photon excitation fluorescence microscopy with core-ring illumination
,”
Opt. Lett.
42
(
3
),
571
575
(
2017
).
12.
J.
Humpolíčková
,
A.
Benda
, and
J.
Enderlein
, “
Optical saturation as a versatile tool to enhance resolution in confocal microscopy
,”
Biophys. J.
97
(
9
),
2623
2629
(
2009
).
13.
Y.
Zhang
,
P. D.
Nallathamby
,
G. D.
Vigil
,
A. A.
Khan
,
D. E.
Mason
,
J. D.
Boerckel
,
R. K.
Roeder
, and
S. S.
Howard
, “
Super-resolution fluorescence microscopy by stepwise optical saturation
,”
Biomed. Opt. Express
9
(
4
),
1613
(
2018
).
14.
C.
Eggeling
,
A.
Volkmer
, and
C. A.
Seidel
, “
Molecular photobleaching kinetics of Rhodamine 6G by one- and two-photon induced confocal fluorescence microscopy
,”
ChemPhysChem
6
(
5
),
791
804
(
2005
).
15.
Y.
Yonemaru
,
M.
Yamanaka
,
N. I.
Smith
,
S.
Kawata
, and
K.
Fujita
, “
Saturated excitation microscopy with optimized excitation modulation
,”
ChemPhysChem
15
(
4
),
743
749
(
2014
).
16.
M.
Eguchi
and
S.
Yamaguchi
, “
In vivo and in vitro visualization of gene expression dynamics over extensive areas of the brain
,”
NeuroImage
44
(
4
),
1274
1283
(
2009
).
17.
A. D.
Nguyen
,
F.
Duport
,
A.
Bouwens
,
F.
Vanholsbeeck
,
D.
Egrise
,
G. V.
Simaeys
,
P.
Emplit
,
S.
Goldman
, and
S.-P.
Gorza
, “
3D super-resolved in vitro multiphoton microscopy by saturation of excitation
,”
Opt. Express
23
(
17
),
22667
22675
(
2015
).
18.
Y.
Yonemaru
,
A. F.
Palonpon
,
S.
Kawano
,
N. I.
Smith
,
S.
Kawata
, and
K.
Fujita
, “
Super-spatial- and -spectral-resolution in vibrational imaging via saturated coherent anti-Stokes Raman scattering
,”
Phys. Rev. Appl.
4
(
1
),
014010
(
2015
).
19.
S.-W.
Chu
,
T.-Y.
Su
,
R.
Oketani
,
Y.-T.
Huang
,
H.-Y.
Wu
,
Y.
Yonemaru
,
M.
Yamanaka
,
H.
Lee
,
G.-Y.
Zhuo
,
M.-Y.
Lee
,
S.
Kawata
, and
K.
Fujita
, “
Measurement of a saturated emission of optical radiation from gold nanoparticles: Application to an ultrahigh resolution microscope
,”
Phys. Rev. Lett.
112
(
1
),
017402
(
2014
).
20.
C.
Sheppard
, “
Super-resolution in confocal imaging
,”
Optik
80
(
2
),
53
54
(
1988
).
21.
C. B.
Müller
and
J.
Enderlein
, “
Image scanning microscopy
,”
Phys. Rev. Lett.
104
(
19
),
198101
(
2010
).

Supplementary Material