NanoFN10: A High-Contrast Turn-On Fluorescence Nanoprobe for Multiphoton Singlet Oxygen Imaging

An “off-on” fluorescent nanoprobe for near-infrared multiphoton imaging of singlet oxygen has been developed. The nanoprobe comprises a naphthoxazole fluorescent unit and a singlet-oxygen-sensitive furan derivative attached to the surface of mesoporous silica nanoparticles. In solution, the fluorescence of the nanoprobe increases upon reaction with singlet oxygen both under one- and multiphoton excitation, with fluorescence enhancements up to 180-fold. The nanoprobe can be readily internalized by macrophage cells and is capable of imaging intracellular singlet oxygen under multiphoton excitation.


Introduction
Singlet molecular oxygen (O 2 ( 1 ∆ g ), hereafter 1 O 2 ), the first electronic excited state of the oxygen molecule [1,2], is a highly reactive oxygen species (ROS) capable of oxidizing electron-rich cellular components such as proteins, nucleic acids, and lipid structures [3]. 1 O 2 plays a key role in a wide variety of chemical and biological processes, e.g., in organic synthesis [4], photodynamic therapy of cancer [5], the eradication of pathogenic microorganisms [6], cell signaling [7], or plant defense [8].
Notwithstanding its relevance, only a few techniques are available for monitoring 1 O 2 in biological media [2]. The preferred one is the optical detection of its near-infrared phosphorescence (NIR) at 1270 nm, which is selective, non-invasive, and robust [9,10]. Nonetheless, the detection of NIR phosphorescence requires sophisticated equipment available only in a few specialized laboratories [11]. In addition, the 1 O 2 phosphorescence quantum yield is exceedingly low in water (<1 × 10 −6 , meaning that only one in a million 1 O 2 molecules decays by emitting a photon) and even lower in biological media due to the action of 1 O 2 quenchers. Hence, the detection of this reduced number of photons poses severe technical challenges. Not surprisingly, alternative, simpler, yet indirect detection methods have been developed based on probes that react selectively with 1 O 2 [12][13][14].
Among them, fluorescent probes are attractive tools for the detection of 1 O 2 in biological media due to their high sensitivity and fast response, together with the widespread adoption of fluorescence microscopy and imaging techniques [15][16][17][18]. The most popular fluorescence probes use anthracenes to trap 1 O 2 , e.g., anthracene dipropionic or dimalonic acids (ADPA and ADMA, respectively) [19] or Singlet Oxygen Sensor Green (SOSG) [20]. Anthracenes react with 1 O 2 , forming endoperoxides and losing their native fluorescence as a result. ADPA and ADMA are thus turn-off probes. In contrast, SOSG, composed of an anthracene bound to a fluorescein-like fluorophore, is a turn-on probe whose fluorescence (due to fluorescein) is quenched by the intact anthracene moiety in its native state. Oxidation of anthracene by 1 O 2 leads to fluorescence enhancements up to 10-fold [15]. A negative effect of the presence of anthracenes in both probes is their non-negligible ability to self-photosensitize the production of 1 O 2 , which may lead to false positives if appropriate control experiments are not employed [21]. In addition, molecular probes readily form complexes with proteins, which inhibit their internalization by cells and lead to strong fluorescence enhancements. This decreases the contrast between the native and oxidized forms, hampering the ability to detect 1 O 2 [22,23]. We have recently shown that conjugation of ADPA and SOSG to nanoparticles minimizes binding to proteins, thereby allowing the detection of 1 O 2 in bacteria [24] and in eukaryotic cells [25]. Moreover, we have recently described a new class of 1 O 2 fluorescent molecular probes, the furyl naphthoxazoles (FNs), which achieve fluorescence enhancements up to 300-fold owing to a change in absorption and fluorescence spectrum upon reaction with 1 O 2 [26]. In this work, we have combined the two approaches to develop a furyl-naphthoxazole nanoprobe (termed NanoFN10) that brings together the high contrast of FN probes and the advantages of the nano approach to improve the detection of 1 O 2 in biological media.
NMR spectra were recorded at room temperature on a Varian Mercury 400 ( 1 H at 400 MHz and 13 C at 100.6 MHz; Palo Alto, CA, USA) spectrometer. The chemical shifts (δ) are reported in parts per million (ppm) relative to the solvent signal, and coupling constants (J) are reported in Hertz (Hz). The following abbreviations are used to define the multiplicities in 1 H NMR spectra: s = singlet, d = doublet, t = triplet, q = quartet, dt = doublet of triplets, ddd = doublet of doublet of doublets, m = multiplet, or combinations of these descriptive names.
Infrared spectra were recorded in a Nicolet Magna 560 FTIR (Madison, WI, USA) spectrophotometer supported on a potassium bromide disk. Values are reported in wave numbers (cm −1 ).
Mass spectrometry (MS) for reported compounds was conducted on an Agilent Technologies 5975 mass spectrometer (Santa Clara, CA, USA) operating in electron ionization (EI) mode at 70-eV and 4-kV accelerating potential.

Grafting of Rose Bengal onto Mesoporous Silica Nanoparticles (MSNP-RB)
One mL of the (MeO) 3 Si-RB solution was added to 12 mL of MSNP solution (33 mg/mL). The solution was left stirring at room temperature for 24 h. Afterward, the MSNP-RB was recovered by centrifugation (15,000× g; 20 min). The MSNP was washed with EtOH. This procedure was repeated several times until no color was observed in the supernatant. The final MSNP-RB nanoparticles were stored suspended in absolute EtOH (10 mg/mL).

Grafting of the Sylilated PEG-Amino Linker onto Mesoporous Silica Nanoparticles
Two hundred mg of (EtO) 3 Si-L-NH 2 (430 µmols) were added to 12 mL of MSNP or MSNP-RB suspension (33 mg/mL) and left stirring at room temperature for 24 h. Afterward, the nanoparticles were recovered by centrifugation (15,000× g; 20 min). The nanoparticles were washed twice with ethanol and once with ACN. The final MSNP-L-NH 2 or RB-MSNP-L-NH 2 nanoparticles were stored suspended in ACN (10 mg/mL).

Synthesis of NanoFN10
Three hundred and fifty mg (2.3 mmol) of EDC, 350 mg of NHS (3.3 mmol), and 300 mg (2.5 mmol) of DMAP were directly dissolved into 12 mL of a suspension of MSNP-L-NH-CO-CH 2 -CH 2 -COOH or RB-MSNP-L-NH-CO-CH 2 -CH 2 -COOH nanoparticles (ACN, 33 mg/mL). The suspension was left stirring at room temperature for 2 h. Then, 100 mg of FN10 was added and left stirring for an additional 72 h. Afterward, the nanoparticles were recovered by centrifugation (15,000× g; 20 min) and washed twice with ACN and twice with MeOH. The final nanoFN10 and nanoFN10RB nanoparticles were stored suspended in MeOH (5 mg/mL).

Determination of the Hydrodynamic Size, ζ-Potential, and Organic Elemental Analysis of the Synthesized Nanoparticles
Hydrodynamic size and ζ-potential of the synthesized (MSNPs) were measured using a Nano-ZS Zetasizer (Malvern Instruments Ltd., Worcestershire, UK). For hydrodynamic size, a diluted aliquot in EtOH was measured. For ζ-potential examination, a diluted aliquot in milli-Q water was measured.
Time-resolved near-infrared phosphorescence experiments were carried out using a customized Fluotime 200 fluorescence lifetime system (PicoQuant GmbH, Berlin, Germany). Briefly, a diode-pumped Nd:YAG laser (FTSS355-Q, Crystal Laser, Berlin, Germany) was used for excitation at 355 or 532 nm (10 kHz, 0.5 or 1.2 µJ per pulse, respectively). A 1064-nm rugate notch filter (Edmund Optics Ltd., Barrington, NJ, USA) was placed at the exit port of the laser to remove any residual component of its fundamental emission in the NIR region. The luminescence exiting from the side of the sample was filtered by one long-pass filter of 1000 nm to remove any scattered laser radiation and isolate the NIR emission. Additional narrow bandpass filters at 1270, 1220, or 1110 nm were used to select a particular near-infrared region. A TE-cooled Hamamatsu NIR sensitive photomultiplier tube assembly (H9170-45, Hamamatsu Photonics, Hamamatsu City, Japan) was used as the detector. Photon counting was achieved with a multichannel scaler (PicoQuant's NanoHarp 250; PicoQuant GmbH, Berlin, Germany). into thin quartz cuvettes and positioned to focus on the air-solvent interface. Focusing at this interface provides a negative control for non-specific reflections and background noise. FN10 was characterized by using excitation and emission scans acquired using the Leica LAS AF software suite.

Two-Photon Spectroscopy
Excitation scans were performed using a fixed emission detection window (330-550 nm) and variable multiphoton excitation (700-1040 nm in 10 nm intervals). Average intensity was measured in regions of interest (ROIs) positioned in sample and air regions of each assay image. FN10 was photooxidized by adding MSNP-RB and irradiating with green light. After the irradiation, MSNP-RB was removed by centrifugation (15,000× g; 10 min).

Cell Culture
N13 microglia adherent cells were grown to 60% confluence in a Labtek 8 wellchambered slide in standard culture medium (RPMI media with 10% Fetal Bovine Serum, 1% Pen/Strep 10,000 units/mL, 1% Gentamicin) and under standard conditions (37 • C and 5% CO 2 in a humidified incubator) and incubated nanoFN10RB for 24 h. Cells were washed three times with fresh culture medium to remove nanoFN10RB in the remaining medium before transfer to the microscope.

Microscopy Analysis
Multiphoton experiments employed a Leica SP5 MP scanning confocal/multiphoton system using a Leica DM6000 inverted microscope and HCX IRAPO L x25.0 NA0.95 WATER objective lens with Spectra Physics MaiTai HP tunable IR-pulsed and 561-nm CW lasers as light sources. N13 microglia cells within the chambered slide were maintained under optimal conditions (37 • C and 5% CO 2 in a semi-closed, humidified microscope incubator) throughout the experiment.
Experiments were performed using a sequential time-lapse mode at a single focal plane. Samples were alternately measured for fluorescence using the multiphoton laser tuned to 720 nm (1% power with 3 accumulations) and a 413-532-nm detection window, and then activated by scanning the sample a total of 50 times using the 561-nm laser with the transmission set to between 0% and 8% maximum power. Each complete cycle of detection and activation was completed in 27 s and immediately followed by the next cycle. Conditions were chosen to allow photooxidized molecules to be detected while minimizing the photobleaching effect of the multiphoton laser and maximizing the 1 O 2 generation. All photographs were processed and analyzed using the Fiji J software (Adobe Systems, San Jose, CA, USA) [28,29].

Synthesis and Characterization of the Molecular Probe FN10
The molecular probe FN10 is a modification of the furyl naphthoxazoles reported in our previous communication [26], in which the hydrogen in position C-5 of the furan has been replaced by a hydroxymethyl group. This improves the furan's reactivity towards 1 O 2 and allows its conjugation to nanoparticles via a Steglich esterification [30,31]. FN10 has been synthesized by a one-pot, two-step reaction, in which the conjugation of methylnaphthoxazole anion to furaldehyde is followed by a dehydration step ( Figure 1A). The spectroscopic characterization is shown in the supplementary information ( Figure S1). our previous communication [26], in which the hydrogen in position C-5 of the furan has been replaced by a hydroxymethyl group. This improves the furan's reactivity towards 1 O2 and allows its conjugation to nanoparticles via a Steglich esterification [30,31]. FN10 has been synthesized by a one-pot, two-step reaction, in which the conjugation of methylnaphthoxazole anion to furaldehyde is followed by a dehydration step ( Figure 1A). The spectroscopic characterization is shown in the supplementary information ( Figure  S1). The absorption spectrum of FN10 in methanol shows a maximum at 363 nm and a second peak at 290 nm, while the fluorescence spectrum shows a maximum at 439 nm ( Figure 1B). The fluorescence quantum yield (ΦF) ranges from 0.05 in acetonitrile to 0.005 in methanol ( Figure S2), indicating efficient quenching of the naphthoxazole singlet excited state by the furan in the protic solvent. The photophysical properties of FN10 in methanol are collected in Table 1.  The absorption spectrum of FN10 in methanol shows a maximum at 363 nm and a second peak at 290 nm, while the fluorescence spectrum shows a maximum at 439 nm ( Figure 1B). The fluorescence quantum yield (Φ F ) ranges from 0.05 in acetonitrile to 0.005 in methanol ( Figure S2), indicating efficient quenching of the naphthoxazole singlet excited state by the furan in the protic solvent. The photophysical properties of FN10 in methanol are collected in Table 1.

Reactivity of FN10 with Singlet Oxygen
The reactivity of FN10 towards 1 O 2 was studied by generating it using new methylene blue photosensitization and monitoring the concomitant changes in the absorption and emission spectra of the probe (Figure 2). In the presence of 1 O 2 , the absorbance of the 363-nm FN10 band decreased strongly ( Figure 2B), and an intense fluorescence emission appeared with a maximum at 400 nm ( Figure 2C). HPLC analysis of the photoirradiated samples revealed the formation of a single photoproduct, FN10-ox ( Figure S3). The observation of isosbestic points in the absorption spectra is consistent with this finding. The structure of FN10-ox was determined by MS, IR, and 1 H/ 13 C-NMR spectroscopies ( Figure S4) and is shown in Figure 2A. Furan oxidation eliminates the fluorescence quencher and shortens the conjugation path in the probe, leading to changes in the absorption and emission spectra that enable the selective excitation and detection of FN10-ox fluorescence ( Figure 3). The excitation spectra of the photooxidized samples show a maximum at 335 nm ( Figure S5). Exciting the samples at this wavelength leads to a remarkable fluorescence enhancement of 180-fold when observed at 400 nm (inset of Figure 2C). emission spectra of the probe (Figure 2). In the presence of 1 O2, the absorbance of the 363nm FN10 band decreased strongly ( Figure 2B), and an intense fluorescence emission appeared with a maximum at 400 nm ( Figure 2C). HPLC analysis of the photoirradiated samples revealed the formation of a single photoproduct, FN10-ox ( Figure S3). The observation of isosbestic points in the absorption spectra is consistent with this finding. The structure of FN10-ox was determined by MS, IR, and 1 H/ 13 C-NMR spectroscopies ( Figure S4) and is shown in Figure 2A. Furan oxidation eliminates the fluorescence quencher and shortens the conjugation path in the probe, leading to changes in the absorption and emission spectra that enable the selective excitation and detection of FN10ox fluorescence (Figure 3). The excitation spectra of the photooxidized samples show a maximum at 335 nm ( Figure S5). Exciting the samples at this wavelength leads to a remarkable fluorescence enhancement of 180-fold when observed at 400 nm (inset of Figure 2C).   emission spectra of the probe (Figure 2). In the presence of 1 O2, the absorbance of the 363nm FN10 band decreased strongly ( Figure 2B), and an intense fluorescence emission appeared with a maximum at 400 nm ( Figure 2C). HPLC analysis of the photoirradiated samples revealed the formation of a single photoproduct, FN10-ox ( Figure S3). The observation of isosbestic points in the absorption spectra is consistent with this finding. The structure of FN10-ox was determined by MS, IR, and 1 H/ 13 C-NMR spectroscopies ( Figure S4) and is shown in Figure 2A. Furan oxidation eliminates the fluorescence quencher and shortens the conjugation path in the probe, leading to changes in the absorption and emission spectra that enable the selective excitation and detection of FN10ox fluorescence (Figure 3). The excitation spectra of the photooxidized samples show a maximum at 335 nm ( Figure S5). Exciting the samples at this wavelength leads to a remarkable fluorescence enhancement of 180-fold when observed at 400 nm (inset of Figure 2C).   The rate constant for overall (physical and reactive) 1 O 2 quenching by FN10 was determined by measuring the 1 O 2 decay rate as a function of FN10 concentration in time-resolved 1 O 2 phosphorescence assays ( Figure S6A) [32]. The value found, 3.0 × 10 7 M −1 s −1 (Figure S6B), is 32-fold larger than for unsubstituted furyl naphthoxazoles [26]. The reactive rate constant was determined by comparing the rate of photooxidation of FN10 with that of dimethyl anthracene [32] and is 1.7 × 10 7 M −1 s −1 (Figure S6C), comparable to that of the most reactive furans [33]. The reactivity of FN10 towards other ROS was examined, and it was found that it does not react with superoxide or hydrogen peroxide ( Figure S7). Finally, FN10 shows a negligible quantum yield of 1 O 2 self-sensitization (Φ ∆ = 0.003 in methanol and acetonitrile, Figure S8); for comparison, SOSG has Φ ∆ = 0.03 at 355 nm and 0.009 at 532 nm in methanol [21]. Taken together, the above results suggest that FN10 could be an excellent fluorescent probe for 1 O 2 detection.

Synthesis and Characterization of the Nanoprobe NanoFN10
Given the excellent properties of FN10, we aimed to synthesize a nanosensor (nanoFN10) where FN10 would be covalently linked to the surface of mesoporous silica nanoparticles through a 2.3-nm PEG linker ( Figure S9). Previous results from our laboratory show that linkers of this length preclude the complexation of the probe with proteins while still preserving its reactivity towards 1 O 2 (Scheme 1) [24,25]. In the first step, 10 eq of the diamino PEG linker was reacted with 1 eq of (3-isocyanate propyl)-triethoxysilane to yield a monosubstituted aminosilylated-PEG linker, which was then grafted onto the surface of 160-nm mesoporous silica nanoparticles. Since the amino moieties cannot react directly with the hydroxyl group of FN10, they were converted to carboxylic moieties via reaction with succinic anhydride. Finally, the FN10 fluorescent probe was conjugated to the functionalized nanoparticles by Steglich esterification. The bare MSNPs and the final nanoFN10 probe were characterized by their hydrodynamic diameter (150 to 200 nm) and ζ-potential (−21 to −23 mV). Covalent conjugation of FN10 did not significantly modify the size and the ζ-potential of the silica nanoparticles.
Once the nanoFN10 was in hand, the particles were suspended in aerated methanol for optical and photophysical characterization. The emission and excitation spectra of nanoFN10 were essentially identical to those of its molecular counterpart, although 10-nm blue-shifted probably due to the esterification of the hydroxymethyl group ( Figure 4A,B). Irradiation of the suspension with red light in the presence of added new methylene blue led to the expected growth of a blue-shifted, bright fluorescence band, as in the case of FN10. A fluorescence enhancement of 20-fold was observed at 390 nm ( Figure 4C,D).   Having established that the nanoprobe enhances its fluorescence upon reaction with 1 O 2 , we prepared a new nanosystem by simultaneously grafting the sylilated diamino PEG linker and a sylilated 1 O 2 photosensitizer (rose bengal, RB) to its surface, thereby enabling orthogonal attachment of FN10 and RB (Scheme 1). This ensured the generation of 1 O 2 in close vicinity to the sensor in the cellular experiments. The bifunctional nanoprobe (termed nanoFN10RB) reproduced the behavior of nanoFN10 ( Figure S10).

Imaging of Singlet Oxygen in Cells
We have successfully developed a novel turn-on singlet oxygen-specific fluorescent probe that shows a remarkable contrast between the native and oxidized forms. We have attached it to the surface of mesoporous silica nanoparticles and combined it with a photosensitizer to ensure co-localization with the singlet oxygen source. However, an excitation wavelength of 335 nm is required to achieve the maximum fluorescence enhancement upon reaction with 1 O 2 , which is not suitable for studies in biological media. Notwithstanding, the probe could be valuable for multiphoton imaging microscopy. Multiphoton excitation would confer additional advantages for bioimaging such as longer excitation wavelength, deeper tissue penetration, lower fluorescence background, reduced photodamage in living systems, and better three-dimensional spatial resolution [34,35]. In light of a recent report on a naphthalimide-anthracene fluorescent probe that showed multiphoton absorption properties [36], we investigated whether the fluorescence of FN10 could be amenable to multiphoton excitation.
As shown in Figure 5A, multiphoton excitation was able to elicit the fluorescence of FN10 in microscopy imaging experiments using a simple solution with an air bubble as a zero-contrast reference image. Figure 5B shows the one-photon absorption and the two-photon fluorescence excitation spectra of FN10, the latter derived from spectroscopic analysis of the microscopy images. The maximum of the two-photon excitation spectrum is found at exactly twice the maximum of the one-photon absorption (and excitation spectra), confirming that FN10 is amenable to two-photon excitation.
showed multiphoton absorption properties [36], we investigated whether the fluorescence of FN10 could be amenable to multiphoton excitation.
As shown in Figure 5A, multiphoton excitation was able to elicit the fluorescence of FN10 in microscopy imaging experiments using a simple solution with an air bubble as a zero-contrast reference image. Figure 5B shows the one-photon absorption and the twophoton fluorescence excitation spectra of FN10, the latter derived from spectroscopic analysis of the microscopy images. The maximum of the two-photon excitation spectrum is found at exactly twice the maximum of the one-photon absorption (and excitation spectra), confirming that FN10 is amenable to two-photon excitation.  Upon reaction with 1 O 2 , a 3-fold increase in fluorescence was observed ( Figure 5C,D). The contrast was much smaller than under one-photon excitation because the excitation wavelength for FN10-ox had to be set at 720 nm rather than 670 nm (=2 × 335 nm) due to equipment limitations. Despite the lower fluorescence enhancement, we attempted to image 1 O 2 in N13 microglia adherent cells.
In the first series of experiments, the N13 microglia cells were incubated with the nanoFN10RB probe for 24 h, and images were taken under two-photon excitation at λ exc 720 nm ( Figure S11). As shown in Figure S12, the two-photon fluorescence colocalizes with the one-photon fluorescence of RB (λ exc 561 nm), confirming its assignment to FN10.
In a second series of experiments, a group of cells preincubated with nanoFN10RB were irradiated with a CW green laser (λ exc 561 nm) at different laser powers (0%, 4%, and 8%), and biphotonic fluorescence images (λ exc 720 nm) were recorded every 30 s ( Figure 6A). At 0% green laser power, when no 1 O 2 was produced, the only effect observed was the photobleaching of the nanoprobe upon prolonged two-photon excitation. However, when 1 O 2 was generated by green light irradiation of RB, the initial photobleaching was followed by a fluorescence enhancement phase ( Figure 6). The duration and extension of the photobleaching and fluorescence-enhancement phases depend on the laser power. At the maximum power (8%), a fluorescence enhancement of 80% could be observed. The time delay between the photobleaching and the fluorescence enhancement is attributed to the period required to eliminate the endogenous 1 O 2 quenchers that compete with the probe oxidation, as has been previously reported for bacteria during a PDT treatment [24]. These results confirm the ability of nanoFN10RB to generate and detect 1 O 2 inside N13 cells by NIR two-photon microscopy.
the photobleaching of the nanoprobe upon prolonged two-photon excitation. However, when 1 O2 was generated by green light irradiation of RB, the initial photobleaching was followed by a fluorescence enhancement phase ( Figure 6). The duration and extension of the photobleaching and fluorescence-enhancement phases depend on the laser power. At the maximum power (8%), a fluorescence enhancement of 80% could be observed. The time delay between the photobleaching and the fluorescence enhancement is attributed to the period required to eliminate the endogenous 1 O2 quenchers that compete with the probe oxidation, as has been previously reported for bacteria during a PDT treatment [24]. These results confirm the ability of nanoFN10RB to generate and detect 1 O2 inside N13 cells by NIR two-photon microscopy. Figure 6. (A): Fluorescence images of N13 microglia cells incubated with nanoFN10RB. The images in the left column are before photoirradiation, and those in the right column were taken after 700-s photoirradiation using different 561-nm power laser intensities. The length of the scale bar is 10 μm. (B): NanoFN10RB fluorescence enhancement kinetics as a function of the 561-nm laser power intensity (0%, 4%, and 8%; black, blue, and green lines, respectively). Open bars represent the mean SD for 20 independent NP systems where only the mean pixel intensity of the images is considered.

Conclusions
We have developed a novel naphthoxazole-furan turn-on fluorescent probe, FN10, which is selective and highly reactive towards 1 O2; it shows enhancement factors up to 180 and produces very little 1 O2 by photosensitization, and its fluorescence can be excited by multiphoton processes. FN10 was successfully attached to MSNPs to protect it from interaction with proteins and other biological components. NanoFN10RB is readily

Conclusions
We have developed a novel naphthoxazole-furan turn-on fluorescent probe, FN10, which is selective and highly reactive towards 1 O 2 ; it shows enhancement factors up to 180 and produces very little 1 O 2 by photosensitization, and its fluorescence can be excited by multiphoton processes. FN10 was successfully attached to MSNPs to protect it from interaction with proteins and other biological components. NanoFN10RB is readily internalized by N13 microglia cells and reacts with intracellularly generated 1 O 2 , enabling two-photon fluorescence imaging of this reactive oxygen species. Work is currently in progress to further enhance the contrast between the native and oxidized nanoprobes.