Three-photon tissue imaging using moxifloxacin

Moxifloxacin is an antibiotic used in clinics and has recently been used as a clinically compatible cell-labeling agent for two-photon (2P) imaging. Although 2P imaging with moxifloxacin labeling visualized cells inside tissues using enhanced fluorescence, the imaging depth was quite limited because of the relatively short excitation wavelength (<800 nm) used. In this study, the feasibility of three-photon (3P) excitation of moxifloxacin using a longer excitation wavelength and moxifloxacin-based 3P imaging were tested to increase the imaging depth. Moxifloxacin fluorescence via 3P excitation was detected at a >1000 nm excitation wavelength. After obtaining the excitation and emission spectra of moxifloxacin, moxifloxacin-based 3P imaging was applied to ex vivo mouse bladder and ex vivo mouse small intestine tissues and compared with moxifloxacin-based 2P imaging by switching the excitation wavelength of a Ti:sapphire oscillator between near 1030 and 780 nm. Both moxifloxacin-based 2P and 3P imaging visualized cellular structures in the tissues via moxifloxacin labeling, but the image contrast was better with 3P imaging than with 2P imaging at the same imaging depths. The imaging speed and imaging depth of moxifloxacin-based 3P imaging using a Ti:sapphire oscillator were limited by insufficient excitation power. Therefore, we constructed a new system for moxifloxacin-based 3P imaging using a high-energy Yb fiber laser at 1030 nm and used it for in vivo deep tissue imaging of a mouse small intestine. Moxifloxacin-based 3P imaging could be useful for clinical applications with enhanced imaging depth.

for a given fluorophore and the better 3D excitation confinement in scattering tissues allows 3PM to achieve deeper imaging than 2PM. However, 3P excitation is much less efficient than 2P excitation and requires excitation sources with a higher pulse energy. Chloroplast 13 , alveoli of mouse lung 14 , and RFP-labeled neurons 15 have undergone 3P imaging.
In this study, we experimentally tested 3P excitation of moxifloxacin at an excitation wavelength of >1000 nm for deep tissue imaging with moxifloxacin labeling. Once moxifloxacin fluorescence was confirmed at this excitation wavelength range, its excitation and emission spectra were obtained. Moxifloxacin-based 3P imaging was performed on ex vivo mouse bladder and ex vivo small intestine tissues and compared with moxifloxacin-based 2P imaging by switching the excitation wavelength of a Ti:sapphire tunable oscillator. The performance of moxifloxacin-based 3P imaging was analyzed in terms of the image contrast. Lastly, we developed a new system for moxifloxacin-based 3P imaging that uses a high-power fiber laser at 1030 nm and demonstrated its ability to perform in vivo tissue imaging.

Results
Characterization of moxifloxacin fluorescence at >1000 nm excitation wavelength. The excitation spectrum, the variation in fluorescence intensity with excitation power, and the emission spectrum of moxifloxacin are shown in Fig. 1a-b, c and d, respectively. The wavelength used for the excitation spectrum of moxifloxacin ranged from 800 to 1050 nm. The excitation spectrum decreased rapidly as the wavelength increased from 800 to 960 nm, because this wavelength range is far from the 2P excitation peak of moxifloxacin (<700 nm) 16 . For excitation wavelengths >960 nm, the excitation spectrum slowly increased as the wavelength increased (Fig. 1b). The increase in the moxifloxacin excitation spectrum at approximately 1000 nm is indicative of an excitation process different from that of 2P excitation, probably 3P excitation. The fluorescence intensity as a function of excitation power was measured to verify this probability (Fig. 1c). The excitation wavelength for this measurement was 1030 nm because there are commercially available light sources at this wavelength, e.g., Yb fiber femtosecond lasers. The fluorescence intensity of moxifloxacin increased with excitation power, and its slope was estimated to be 2.96 in log-log plot. Because the measured slope was close to 3, we confirmed that moxifloxacin excitation at 1030 nm was 3P excitation. The emission spectrum of moxifloxacin was also measured at an excitation wavelength of 1030 nm (Fig. 1d). It was quite broad, with its peak at approximately 510 nm. The 3P active cross section of moxifloxacin was 0.2 × 10 −83 cm 6 /(s/photon) 2 , at the excitation wavelength of 1000 nm, approximately 0.8 times that of DAPI solution 17 at the same concentration.
Comparison of moxifloxacin-based 3P and 2P imaging of ex vivo mouse bladder. After characterizing moxifloxacin 3P fluorescence, moxifloxacin-based 3P imaging was tested in biological tissues and compared with moxifloxacin-based 2P imaging. First, moxifloxacin-based 2P and 3P imaging was conducted in ex vivo mouse bladder specimens by changing the excitation wavelength of the Ti:sapphire oscillator between 780 and 1000 nm, respectively. 2P and 3P images of the mouse bladder at several depths from the luminal side are presented in Fig. 2. For the 3P imaging, the excitation wavelength was 1000 nm instead of 1030 nm because SHG in the bladder was strong and moxifloxacin fluorescence needed to be distinguished from the SHG. The bladder is a muscular sac that stores urine generated by the kidney. Because various bladder disorders initially occur in the lumen, the luminal side underwent moxifloxacin-based imaging. Both 2P and 3P imaging with moxifloxacin labeling showed various cellular structures with depth, including hexagon-shaped umbrella cells in the uroepithelium, sparsely distributed cellular structures in the lamina propria, and a thick muscle layer below the lamina propria. The umbrella cells ( Fig. 2a and d) and the cells in the lamina propria ( Fig. 2b and e) were clearly visible via both moxifloxacin-based 2P and 3P imaging because of the cell labeling with moxifloxacin. Blood vessels (white arrowheads in Fig. 2b and e) in the lamina propria were visible via moxifloxacin labeling of vascular endothelial cells. Bundles of muscle cells and other small cells (white arrows in Fig. 2c and f) between the bundles were visible in the deep muscle layer.
Both the 2P and 3P images showed almost identical cellular structures via moxifloxacin labeling, and the anatomy captured by these imaging methods was consistent with histology 18 . Although both moxifloxacin-based 2P and 3P imaging visualized cellular structures clearly at relatively shallow depths, down to ~125 µm ( Fig. 2g and j), the contrast of the moxifloxacin-based 2P images at deeper depths was less than that of the corresponding 3P images. At 225 µm from the surface, the moxifloxacin-based 2P image showed low contrast with a high background ( Fig. 2h) and individual muscle cells were not well resolved. On the other hand, the moxifloxacin-based 3P image at the same depth showed distinct detailed muscle structures (Fig. 2k). At 275 µm from the surface, there was no contrast in the moxifloxacin-based 2P image ( Fig. 2i). At the same depth, individual muscle cells could be seen in the moxifloxacin-based 3P image (Fig. 2l). Comparison of moxifloxacin-based 2P and 3P images of the mouse bladder specimen showed clearly that moxifloxacin-based 3P imaging can obtain deeper images by using a longer excitation wavelength and by better confinement of 3D excitation in the scattering tissue.
To analyze the image contrast, we plotted the intensity profiles ( Fig. 2m-o) of the yellow dashed lines crossing the muscle cells in the moxifloxacin-based 2P and 3P images (insets in Fig. 2g-l). At 125 µm below the surface, the intensity profiles for both sets of images were high in the muscle cells and noticeably low in the regions between the muscle cells ( Fig. 2m). At 225 µm below the surface, the intensity profile for the moxifloxacin-based 2P image showed a small peak in the muscle cells, with a slightly higher intensity than the surrounding area, while that for the moxifloxacin-based 3P image had a large peak in the muscle cells, with a noticeably stronger fluorescence intensity than the surrounding area ( Fig. 2n). At 275 µm below the surface, the intensity profile of the moxifloxacin-based 2P image did not show distinguishable intensity variation between the muscle cells and surrounding area, while that of the moxifloxacin-based 3P image had noticeably high intensities in the muscle cells.
The image contrast of moxifloxacin-based 2P and 3P imaging was also analyzed quantitatively by calculating the coefficient of variation (CV), which is a measure of relative variability in the intensity image and calculated as the ratio of the standard deviation to the mean. Figure 2p shows the changes in the CV at several depths in muscle. The CV values for the moxifloxacin-based 2P images of the mouse bladder specimen monotonically decreased as the depth increased, which was consistent with the decrease in the image contrast. Moxifloxacin-based 3P images of the mouse bladder specimen had higher CV values than the corresponding 2P images at all depths in the muscle layer. However, the CV values of the 3P images decreased somewhat at depths >200 µm from the surface. At the maximum depth of 275 µm, the CV value of the 3P image was still quite high compared to that of the 2P image. This indicates that moxifloxacin-based 3P imaging still had good contrast at this depth.
Comparison of moxifloxacin-based 3P and 2P imaging of ex vivo mouse small intestine. In addition to the mouse bladder specimen, moxifloxacin-based 3P imaging was conducted on ex vivo mouse small intestine specimens and compared with moxifloxacin-based 2P imaging. The small intestine, part of the gastrointestinal tract, absorbs nutrition from digested food and maintains gut microbiota. Intravital microscopy has been used in structural and functional studies of the small intestine but deep-tissue imaging methods would be more beneficial.
Moxifloxacin-based 2P and 3P imaging of the mouse small intestine specimen was conducted by switching the excitation wavelength of the Ti:sapphire oscillator between 780 and 1030 nm. Figure 3 shows moxifloxacin-based 2P and 3P images obtained at different depths from the serosa. Both moxifloxacin-based 2P and 3P imaging showed various cellular structures at different depths, including the myenteric plexus, muscle, and intestinal glands. The myenteric plexus, located between muscle layers, is part of the neuronal network in the gastrointestinal tract and is involved in the contractile and refractive motion used for digestion. Some cells in the myenteric plexus ( Fig. 3a and d) and spindle-shaped cells in the muscle layer below the myenteric plexus ( Fig. 3b and e) were clearly visible in both 2P and 3P images. Intestinal glands are depressed tubular glands in the lining of the epithelium that covers the lumen of the small and large intestines and contain Paneth cells and stem cells at the base and enterocytes and goblet cells in the lining. Images from both 2P and 3P imaging showed epithelial cells along the lining of the gland at the relatively shallow depth of 60 µm ( Fig. 3c and f). However, at the depth of 80 µm, the epithelial cells in the moxifloxacin-based 2P image were barely distinguishable because of reduced contrast (Fig. 3g), and at larger depths, the individual epithelial cells could not be imaged at all (Fig. 3h). Only some cells in the lamina propria (white arrowhead in Fig. 3i) were visible down to 120 μm. On the other hand, image contrast was maintained in moxifloxacin-based 3P images and individual cells in deep regions of the glands could be seen ( Fig. 3j-l).
To analyze the image contrast, we plotted the intensity profiles ( Fig. 3m-o) of the yellow dashed lines crossing the membrane of the epithelial cells in the intestinal gland, seen in the insets of Fig. 3g-l. At 80 μm from the surface, the epithelial cell membrane showed a higher fluorescence signal intensity than the surrounding area in both moxifloxacin-based 2P and 3P images (Fig. 3m). At larger depths, there was no difference between the signal intensities of the cell membrane and the surrounding area in the moxifloxacin-based 2P images. However, moxifloxacin-based 3P images still had relatively high signal levels in the cell membrane ( Fig. 3n and o).
We also calculated the CV values of the moxifloxacin-based 2P and 3P images of the mouse small intestine specimen; the change in the CV with depth is shown in Fig. 3p. The CV values of the moxifloxacin based 2P and 3P images rapidly decreased as the depth increased, probably because of the complex structure of the small intestine. The CV values of the 2P images were the lower than those of the corresponding 3P images at the same depths, which is consistent with the image contrast comparison. At the maximum depth of 120 µm from the surface, the CV value of the 2P image approached the minimum value of 0.025. However, the CV value of the corresponding 3P image was still high, indicating relatively good image contrast at 120 µm. This quantitative analysis of the image contrast showed that larger imaging depths could be reached in moxifloxacin-labeled tissues by using 3P excitation instead of 2P excitation.

Moxifloxacin-based 3P imaging of in vivo mouse small intestine. Moxifloxacin-based 3P and 2P
imaging techniques used on ex vivo biological tissues were compared by switching the excitation wavelength. Moxifloxacin-based 3P imaging of in vivo biological tissues was difficult with current multiphoton microscopy systems because the imaging takes a long time, especially at large depths (655 s/frame at a depth of 120 μm from the serosa surface of the mouse small intestine). The long imaging time is due to the energy limitation of the Ti:sapphire oscillators at a >1000 nm excitation wavelength. The excitation energy needed for moxifloxacin-based 3P imaging is approximately a few tenths of a nanojoule per pulse at large depths; however, the maximum excitation energy from a current Ti:sapphire oscillators are approximately 0.5 nJ at a >1000 nm excitation wavelength. Because the maximum excitation energy was used for moxifloxacin-based 3P imaging at the tissue surface, there was no extra energy to compensate for the loss of excitation energy as the depth increased. Therefore, the very long acquisition time was required to obtain a sufficient fluorescence signal at large depths, making in vivo moxifloxacin-based 3P imaging difficult. For in vivo moxifloxacin-based 3P imaging, we developed another 3P imaging system that uses a high-pulse-energy Yb fiber laser as the excitation source. This Yb fiber laser had a pulse energy of 10 µJ at 1034 nm and a pulse repetition rate of 102.9 kHz. It performed deep tissue imaging by compensating for the energy loss with increasing depth, although it had a very low pulse repetition rate for 3P imaging compared to that of the Ti:sapphire oscillator running at 80 MHz. The image acquisition rate was 66 s/ frame, approximately 20 times slower than Ti:sapphire oscillator-based 3P imaging at the tissue surface due to the low repetition rate. However, the same acquisition rate could be maintained for all imaging depths and was much faster than that of Ti:sapphire oscillator-based 3P imaging at the large depths.
In vivo imaging of the mouse small intestine in a gas-anesthetized mouse was conducted with the use of a small intestine holder; the details are discussed in the Methods section. Figure 4 presents moxifloxacin-based 3P images of the in vivo mouse small intestine at several depths from the surface of the serosa. In vivo moxifloxacin-based 3P imaging visualized the cellular structures of the small intestine, similar to the images acquired in the ex vivo study. Muscle layers oriented in alternating directions were visible at the shallow depths of 25 and 50 µm ( Fig. 4a  and b), and epithelial cells in tubular intestinal glands and surrounding cells in the lamina propria were visible at larger depths ( Fig. 4c and d). Cells with bright fluorescent signals were imaged at a depth of 150 μm (Fig. 4e), and membranes of epithelial cells in intestinal glands were seen at a depth of 175 μm (Fig. 4f). Cells were resolved up to a depth of 250 μm (Fig. 4g and h).

Discussion
The use of moxifloxacin, an FDA-approved antibiotic, as a cell-labeling agent for 2P imaging had been demonstrated in a previous study 8 . However, the moxifloxacin-based 2P imaging depth is shallow due to the relatively short excitation wavelength of <800 nm used. In this study, 3P excitation of moxifloxacin with a >1000 nm excitation wavelength was experimentally verified by measuring the relationship between fluorescence intensity and excitation power. 3P imaging with moxifloxacin labeling was tested in ex vivo mouse bladder and ex vivo mouse small intestine using a Ti:sapphire laser at near 1030 nm excitation wavelength, and compared with moxifloxacin-based 2P imaging using a 780 nm excitation wavelength. Although both 3P and 2P imaging visualized cellular structures via moxifloxacin labeling, moxifloxacin-based 3P imaging using longer excitation wavelengths resolved small cellular features with better contrast than moxifloxacin-based 2P imaging. In this comparison study, moxifloxacin-based 3P imaging using the Ti:sapphire oscillators could not reach the maximum imaging depth because of the low excitation energy of the lasers at longer wavelengths and the very long acquisition time. Therefore, we constructed a new 3P imaging system that uses a high-energy Yb fiber laser and used it to demonstrate in vivo deep tissue imaging with moxifloxacin labeling. 3P excitation typically requires higher excitation photon density and excitation sources with higher pulse energy than does 2P excitation, because the probability of 3P excitation is much lower than that of 2P excitation. This also applies to moxifloxacin-based multiphoton imaging. Moxifloxacin-based 3P imaging needs to use a laser with high pulse energy, in the range of a few tenths of a nanojoule. However, Ti:sapphire oscillators were used in this study so moxifloxacin-based 3P and 2P imaging could be compared by switching the excitation wavelength of the oscillators. In addition, 3P images were acquired by increasing the pixel dwell time of the laser scanning instead of increasing the excitation energy because the current lasers have limited energy at wavelengths >1000 nm. The maximum powers of the Ti:sapphire oscillators, i.e., 50 mW at 1000 nm and 31.5 mW at 1030 nm, were used for moxifloxacin-based 3P tissue imaging at all depths. These powers corresponded to approximately 0.5 nJ/pulse. Therefore, the pixel dwell time needed to increase rapidly as the depth increased to acquire enough fluorescence signals. The image acquisition rates for the bladder sample were 1.25 s/frame at the surface and 320 s/frame at 225 μm below the surface. Although the imaging speed at large depths was too slow for practical applications, our experiment was performed to characterize moxifloxacin-based 3P imaging and compare it with corresponding moxifloxacin-based 2P imaging.
3P excitation of moxifloxacin at >1000 nm would allow the use of commercial Yb fiber lasers, which have an 1030 nm wavelength and high pulse energies up to tenths of a microjoule per pulse, as the excitation source. Because the pulse energies of these lasers are high enough for 3P excitation of moxifloxacin, the lasers could solve the problems associated with using the Ti:sapphire oscillator for 3P imaging of biological tissues, especially at large imaging depths. We used a commercial Yb fiber amplifier for in vivo 3P imaging with moxifloxacin labeling of the mouse small intestine. However, the imaging speed of the in vivo study was not satisfactory because the Yb laser has a very low repetition rate of 102.9 kHz, 800 times lower than that of the Ti:sapphire oscillator, which runs at 80 MHz. An optimal laser for moxifloxacin-based 3P imaging would have a repetition rate of 1-2 MHz 19 and a pulse energy of a few hundred nanojoules; such Yb fiber lasers are commercially available. Because Yb fiber lasers are currently used in high-precision ophthalmic surgeries such as laser-assisted in situ keratomileusis (LASIK) and cataract surgeries, moxifloxacin-based 3P imaging could be easily adapted to use similar lasers in the clinic.
Fluorescence enhancement in moxifloxacin-based 3P imaging was analyzed by conducting tissue imaging with and without the use of moxifloxacin at near 1030 nm excitation wavelength. Tissue images obtained with and without the use of moxifloxacin are shown in Supplementary Figs 1 and 2. For the imaging of the ex vivo mouse bladder, the autofluorescence intensities at 1000 nm excitation wavelength were negligible compared to the intensities of moxifloxacin 3P fluorescence (Supplementary Fig. 1). For the imaging of the ex vivo mouse small intestine, moxifloxacin 3P fluorescence was approximately 7.7 times higher than the intrinsic signals. Autofluorescent molecules excitable at 1030 nm might exist in the small intestine.
The imaging resolutions of moxifloxacin-based 3P imaging using 1030 nm excitation wavelength were similar to those of moxifloxacin-based 2P imaging using 780 nm excitation wavelength. The Rayleigh length of 3P excitation is smaller than that of 2P excitation, if the same excitation wavelength is used 20 . However, 3P excitation usually requires excitation wavelengths 1.5 times longer than those for 2P excitation for the same fluorophores. In this study, the excitation wavelength for moxifloxacin-based 3P imaging was approximately 1.3 times longer than that for the corresponding 2P imaging. Therefore, the resolutions of moxifloxacin-based 3P and 2P images were approximately the same.
The imaging depth of multiphoton microscopy is limited by the scattering of both excitation and emission photons. Although moxifloxacin-based 3P imaging reduced the effect of excitation photon scattering compared to moxifloxacin-based 2P imaging by using a long excitation wavelength, the effect of emission photon scattering remained the same. The relatively short emission wavelength of moxifloxacin is not optimal for deep tissue imaging 21,22 . However, the effect of emission photon scattering is less significant than the effect of excitation photon scattering because the non-descanned photon detector used in 2P and 3P imaging can collect scattered emission photons 23 .
The emission spectra of moxifloxacin with 3P excitation at a 1030 nm excitation wavelength and 2P excitation at a 780 nm excitation wavelength were slightly different. The emission spectrum obtained with the 1030 nm excitation wavelength had a blue-shifted emission peak when compared to the emission spectrum obtained with the 780 nm excitation wavelength. Although an emission spectrum is usually constant, independent of either the excitation wavelength or the excitation process, according to Kasha's rule, we measured the variation in the emission spectrum with these two excitation wavelengths. This variation might result from the dependence of the emission spectrum on the different excitation wavelengths or the different nonlinear excitation processes. Because 780 nm in 2P excitation and 1030 nm in 3P excitation roughly correspond to 390 and 340 nm in one-photon (1P) excitation, different excitation wavelengths might be the cause of the variation in the emission spectrum. The phenomenon of emission spectrum variation with variation in excitation wavelength is not uncommon and has been called the "red-edge effect" 24 . For testing this phenomenon, the emission spectrum of moxifloxacin was obtained at the 2P excitation wavelengths of 700 and 780 nm, where 700 nm in 2P excitation roughly corresponds to 350 nm in 1P excitation. The emission spectrum of moxifloxacin at 700 nm 2P excitation was shifted from that at 780 nm 2P excitation and closely matched that at 1030 nm 3P excitation. To check the variation phenomenon at 1P excitation, the emission spectrum of moxifloxacin was obtained at various 1P excitation wavelengths and found to vary with the excitation wavelength. Therefore, the different emission spectra of moxifloxacin, between 780 nm in 2P excitation and 1030 nm in 3P excitation, might be due to the variation of the emission spectrum with excitation wavelength, i.e., the red-edge effect.
In this study, we demonstrated 3P imaging of biological tissues using moxifloxacin as a cell-labeling agent. Moxifloxacin fluoresced at an excitation wavelength >1000 nm as a 3P excitation process, and its excitation and ScIentIfIc RepoRTS | (2018) 8:9415 | DOI:10.1038/s41598-018-27371-8 emission spectra were obtained. Moxifloxacin-based 3P imaging showed better contrast and larger imaging depth than did moxifloxacin-based 2P imaging. Moxifloxacin-based 3P imaging could be useful for deep tissue imaging in both preclinical and clinical applications.

Methods
Multiphoton microscopy systems. A commercial and two custom-built multiphoton microscopy systems were used in this study. The commercial system (SP-5, Leica, Germany) was used to characterize moxifloxacin fluorescence at a >1000 nm excitation wavelength and to perform ex vivo 2P and 3P imaging of a mouse bladder. The excitation source was a Ti:sapphire femtosecond laser (Chameleon Vision, Coherent Inc., USA) with a tunable wavelength from 680 to 1080 nm, an 80 MHz repetition rate, and a 140 fs pulse width. A 40x objective lens (water immersion NA 1.1 LD C-Apochromat Corr M27, Zeiss, Germany) was used. The emitted light could be spectrally resolved using a dispersion prism and a sliding slit in the system, a setup used to obtain the moxifloxacin emission spectra. For imaging, the emitted light was spectrally collected at four detection channels by a combination of dichroic mirrors and filters. The custom-built multiphoton microscopy system was equipped with another Ti:sapphire femtosecond laser (Chameleon Ultra II, Coherent Inc., USA), and used for 2P and 3P imaging of ex vivo mouse small intestines. The same 40x objective lens used in the commercial system was used in the custom system. The emitted light could be separated at up to three channels using a combination of dichroic mirrors and filters. Photomultiplier tubes (H7421-40P, Hamamatsu, Japan) in photon-counting mode were used to detect the emission.
Multiphoton microscopy systems based on Ti:sapphire femtosecond lasers were used for ex vivo moxifloxacin-based 2P and 3P imaging of biological tissues by switching the excitation wavelength between 780 and near 1030 nm. These Ti:sapphire femtosecond laser-based multiphoton systems were not appropriate for in vivo moxifloxacin-based 3P imaging because the image acquisition time was too long at large depths, a result of insufficient pulse energy for efficient 3P excitation of moxifloxacin. To demonstrate in vivo moxifloxacin-based 3P imaging, we constructed a new 3P imaging system with a high-energy Yb fiber laser (Uranus 2500-1030-0800-PM, PolarOnyx, USA) that had a wavelength of 1034 nm, a pulse energy of ~10 µJ, and a repetition rate of 102.9 kHz. The high pulse energy, controlled electronically, allowed in vivo deep tissue imaging based on efficient 3P excitation of moxifloxacin. The laser beam went through a standard multiphoton microscopy system that included a beam expander (AC254-050-B and AC254-150-B, Thorlabs, USA), an x-y galvanometer scanner (GVS012/M, Thorlabs, USA), a scan lens (AC254-050-B, Thorlabs, USA) and tube lens combination (AC508-200-B, Thorlabs, USA), a dichroic mirror (T860LPXR, Chroma, USA), and an objective lens, which was the same as that used in the other multiphoton microscopy systems. A dichroic mirror (T560LPXR, Chroma, USA) and an emission filter (ET850SP, Chroma, USA) spectrally separated emitted light into two channels. Two photomultiplier tubes (R928P, Hamamatsu) detected the emitted light in photon-counting mode.
Characterization of moxifloxacin fluorescence at a >1000 nm excitation wavelength. Ophthalmic solution (Vigamox, Alcon, Fort Worth, TX, USA) containing 5.45 mg/ml of moxifloxacin hydrochloride (pH 6.8) was used in all the experiments. The ophthalmic solution (50 μl) was placed on a microscope well glass, covered with a microscope cover glass, and sealed with nail polish. The excitation and emission spectra were measured using the commercial multiphoton microscopy system. The excitation spectrum was measured by changing the wavelength from 800 to 1050 nm in 10 nm steps and collecting the light emitted from the ophthalmic solution sample. The emission spectrum was measured from 405 to 664 nm using a 7 nm spectral step size. Once moxifloxacin fluorescence at the >1000 nm excitation wavelength was confirmed, the multiphoton excitation process was analyzed by measuring the relationship between the fluorescence intensity and the excitation power. The fluorescence intensity was measured while varying the excitation power from 8 to 18 mW at a 1030 nm excitation wavelength. The results were displayed in a log-log plot, the slope of which was estimated by least-squares curve fitting. The fluorescence intensity of moxifloxacin at a 1000 nm excitation wavelength was compared with that of DAPI (#62248, Thermo Fisher Scientific, USA), a cell nucleus-labeling dye with known multiphoton fluorescence properties 17 . Both moxifloxacin and DAPI solutions were diluted to 2.86 mM with phosphate-buffered saline.
2P and 3P imaging of ex vivo mouse bladder and small intestine. The SKH1-HrHr mice (5-6 weeks old) used in this study were kept under specific-pathogen-free conditions at the POSTECH Biotech Center. All procedures were performed in accordance with the relevant guidelines and regulations, and all experimental procedures performed on the animals were approved by the POSTECH Institutional Animal Care and Use Committee (approval number POSTECH-2015-0030-R2). For the ex vivo specimens, mice were sacrificed and the bladder or the small intestine were excised. Approximately 2.5 μl/mm 2 of moxifloxacin ophthalmic solution was administered to the tissue, the maximum amount that the tissue could hold. The time needed for the moxifloxacin to penetrate into the tissues, i.e., the incubation time, was 20 min 8 , after which 2P and 3P imaging with moxifloxacin labeling was performed within 90 min post administration to avoid a decrease in the moxifloxacin concentration in the tissue via diffusion.
Two different excitation wavelengths, 780 and 1000 nm, were used for moxifloxacin 2P and 3P imaging of the mouse bladder, respectively. The excitation wavelength used for moxifloxacin-based 3P imaging was 1000 nm instead of 1030 nm because the bladder yielded a strong SHG signal due to collagen and current emission filters and dichroic mirrors cannot distinguish SHG from moxifloxacin fluorescence at a 1030 nm excitation wavelength. Two detection channels (Ch1: 430-480 nm and Ch2: 520-550 nm) were used and the SHG signal from collagen was omitted. The images comprised 256 × 256 pixels and the imaging field of view was 387.5 μm × 387.5 μm. The image acquisition rate for moxifloxacin-based 2P imaging was 1.25 s/frame at all imaging depths, and that for moxifloxacin-based 3P imaging varied with depth from 1.25 s/frame on the surface to 320 s/frame at 275 µm. The excitation power for 2P imaging increased with depth from 2 to 12 mW, and that for 3P imaging was 50 mW, which was the maximum power of the Ti:sapphire oscillator at the 1000 nm.
Moxifloxacin-based 2P and 3P imaging of the mouse small intestine was performed using excitation wavelengths of 780 and 1030 nm, respectively. Moxifloxacin fluorescence was collected in one channel (Ch1: 400-560 nm). The images comprised 256 × 256 pixels and the imaging field of view was 150 μm × 150 μm. The image acquisition rate for moxifloxacin-based 2P imaging was 3.3 s/frame at all imaging depths, and that for moxifloxacin-based 3P imaging varied with depth from 3.3 s/frame on the surface to 655 s/frame at 120 μm. The excitation power for 2P imaging increased with depth from 6 to 65 mW, and that for 3P imaging was 31.5 mW, which was the maximum power of the Ti:sapphire oscillator at 1030 nm.
3P imaging of in vivo mouse small intestine. For in vivo imaging, the mouse was anesthetized with a gas mixture of 1.5%/vol isoflurane (Terrell, Pirimal Critical Care, Inc., USA) and medical grade oxygen administered via face mask. The abdomen was cut open and a portion of the small intestine was gently pulled out and hung on a custom-built intestine holder (Live Cell Instrument, Korea). The intestine holder consisted of a grooved holder that kept the pulled small intestine in position. In addition, the intestine holder had a magnetically attached cover that gently pressed the small intestine. A fluid channel supplied saline to the holder groove to keep the small intestine moist. After the small intestine was administered moxifloxacin ophthalmic solution (2.5 μl/mm 2 ), it was incubated for 20 min. A temperature controller kept the mouse body and pulled small intestine at 37 °C. After incubation, we performed in vivo moxifloxacin-based 3P imaging of the mouse small intestine using our new custom system based on the high-energy Yb fiber laser with an excitation wavelength of 1034 nm. Moxifloxacin fluorescence was collected at Ch1 (400-560 nm). The images comprised 256 × 256 pixels and the imaging field of view was 300 μm × 300 μm. The image acquisition time was 66 s/frame at all imaging depths, and the excitation power increased with depth from 5.2 to 8 mW.

Contrast analysis of 2P and 3P tissue images.
To compare moxifloxacin-based 2P and 3P imaging of biological tissue, we analyzed the variation of image contrast with depth. The image contrast was quantified by calculating the CV, which is a measure of the relative variability in the intensity of the image and is calculated as the ratio of the standard deviation to the mean. Before calculating the CV, the 2P and 3P images were preprocessed. The background of the original images was nonuniform. After removing the noise, the nonuniformity was corrected by multiplying a gain matrix the same size as the image. The gain matrix was obtained by local adaptive intensity analysis. Valid pixels containing cellular structures were selected by thresholding. The CV of an image was calculated using the valid pixels of the background-normalized 2P and 3P images. All the image-processing procedures were conducted using Matlab (R2017a, MathWorks, USA). Data availability. The data generated during the current study are available from the corresponding author on reasonable request.