Nerve damage sustained during surgery remains a major risk to patients, resulting in postsurgical complications such as pain, paralysis, and persistent numbness.1,2 Current techniques to avoid intraoperative nerve injury rely heavily on a surgeon’s skill level and experience to visually avoid these typically small, buried structures. A contrast agent specific for nerve tissue could aid in identification and visualization of nerves during surgery, as a number of fluorescence image-guided surgical systems are either in clinical trial or have FDA approval such as the FLARE™ by Curadel, Fluorobeam by Fluoptics, da Vinci robotic systems by Intuitive Surgical, SPY by Novadaq, Photodynamic Eye by Hamamatsu, and Artemis by Quest Medical.3,4 Imaging agents that can provide tissue-specific fluorescent signal have the potential to deliver a high contrast means of visualizing target tissues using such fluorescence-guided surgical systems. However, currently there are only a handful of fluorescent contrast agents approved for clinical use including fluorescein, aminolevulinic acid-protoporphyin IX, methylene blue, and indocyanine green, none of which have any nerve specificity.5–9
A limited number of small molecule fluorescent contrast agents exist that stain nerve tissue in vivo, with six known classes of small molecule fluorophores that have been shown to have nerve or brain specificity. These include a handful of stilbene derivatives,10 a coumarin analog,11 a library of distyrylbenzene derivatives,12–15 8 styryl pyridinium (FM) fluorophores,16,17 an oxazine fluorophore,2,18 and a tricarbocyanine fluorophore.19 Unfortunately, adipose specificity often accompanies nerve specificity for these small molecule fluorophores,2,13,14 which is largely by necessity since the molecular composition of nerve and adipose tissues is similar,20,21 making specific accumulation in the nerve without adipose accumulation a substantial challenge. This dual tissue accumulation compounds the problem of identifying and visualizing nerve tissue using fluorescence, as nerves are often adjacent to or embedded within adipose tissue. Of the known small molecule nerve-specific fluorophores, oxazines have demonstrated the greatest utility since the lead compound is redshifted and peripheral nerve specificity is high. However, the current lead nerve-specific oxazine compound, oxazine 4, does not have the desired near-infrared (NIR) excitation or emission,2,18 making additional chemical modifications necessary to generate a clinically relevant agent.
Interestingly, the commercially available Nile Red is also an oxazine fluorophore; however, unlike oxazine 4, it has a strong tendency to accumulate in lipid- and adipose-rich environments.22 In addition, in apolar environments such as lipids, Nile Red exhibits a high quantum yield and hypsochromatically shifted absorbance and emission spectra. In contrast, in polar environments such as water, Nile Red exhibits bathochrommatically shifted absorbance and emission spectra with a quantum yield that approaches zero.23 These spectral properties have led to a wide range of applications from evaluation of lipid droplets in tissue to its use in PEGylated nanoparticles for iron detection.24 More recently, Nile Red has been shown to improve nerve-specific contrast of oxazine 4 in a study that used dual fluorophore staining to enhance visualization of nerve tissue during in vivo studies where adipose, nerve, and muscle tissues were present.2 While the adipose specificity of Nile Red was high, it showed some accumulation in nerve tissue after administration, reducing the contrast enhancing benefits of its use as a ratiometric fluorophore to enhance nerve contrast.
Herein, we report the synthesis of eight Nile Red derivatives with varied electron donating and withdrawing groups to try to influence tissue-specific accumulation and generate a truly adipose-specific fluorophore. In vitro, ex vivo, and in vivo staining studies were completed to assess general lipid staining as well as nerve-, adipose-, and muscle-specific accumulation and contrast from each Nile Red derivative was compared to the parent compound. Excitingly, a Nile Red derivative, Nile Red 5 (NR5), was found that showed adipose-specific contrast and decreased nerve accumulation as compared to Nile Red. Notably, NR5 also resulted in improved adipose-to-nerve and adipose-to-muscle contrast compared to Nile Red, especially when longer time points relevant to surgical interventions were considered. Ratiometric imaging with NR5 and oxazine 4 compared to Nile Red and oxazine 4 demonstrated improved nerve contrast using NR5 as compared to Nile Red. This derivative could be used in future work to improve nerve-specific ratiometric imaging with oxazine 4 or other nerve-specific fluorophores for fluorescence image-guided surgery.
Materials and Methods
Chemicals and Solvents
All fluorophores, chemicals, and solvents were purchased from Sigma-Aldrich (St. Louis, Missouri) or Thermo Fisher Scientific (Waltham, Massachusetts) and used without further purification unless otherwise noted. All solvents were American Chemical Society grade unless otherwise noted. Silica gel, 230 to 400 mesh was used for all column chromatography.
Synthetic Confirmation and Fluorophore Characterization
Tandem liquid chromatography mass spectroscopy (LCMS) analysis (Agilent 1260 LC system with a 6224-time of flight MS, Agilent, Santa Clara, California) was performed using a C18 column (, Agilent). High-performance liquid chromatography (HPLC) purification (Agilent 1260 system with a preparative pump) was performed using a C18 column (, Agilent). For both HPLC and LCMS, the mobile phase consisted of HPLC grade acetonitrile and water, treated with 0.1% formic acid. Spectroscopic analysis was performed on a Spectramax M5 spectrometer (Molecular Devices, Sunnyvale, California) using a quartz cuvette with a 1-cm pathlength. Quantum yield was calculated as25 and refers to the unknown sample.
Synthesis of fluorophore base structures
1.02 g (6.16 mmol) of 3-diethylaminophenol () was dissolved in 16 mL of 2 M hydrochloric acid (HCl) and cooled in an ice bath for the duration of the reaction. 466.1 mg (6.76 mmol) of sodium nitrite () was added dropwise, and the reaction was stirred for 2 h to yield the nitroso compound . The nitroso compound was filtered and used as the HCl salt without further purification (Fig. 1). calculated for : ; measured: .
129.6 mg (0.81 mmol) of 1,6 dihydroxynapthalene and 142.9 mg (0.62 mmol) of were dissolved in 3 mL of dimethylformamide (DMF) and heated to 100°C under a blanket of nitrogen. The reaction was allowed to proceed for 4 h and was monitored via thin layer chromatography (TLC) [dichloromethane (DCM):methanol (MeOH): 95:5] (Fig. 1). The solvent was removed under reduced pressure and the product was purified by column chromatography, eluting with chloroform ():methanol (MeOH):triethylamine (TEA), 100:3:2). calculated for : ; measured: .
196.0 mg (1.22 mmol) of 1,7-dihyroxynaphthalene and 209 mg (0.91 mmol) of compound were dissolved in 3 mL of DMF and heated to 100°C under a blanket of nitrogen. The reaction was allowed to proceed for 4 h and was monitored via TLC (DCM:MeOH, 95:5) (Fig. 1). The solvent was removed under reduced pressure and the product was purified by column chromatography, eluting with DCM:MeOH: (100:3). calculated for : ; measured: .
9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-3-yl trifluoromethanesulfonate (5)
88.2 mg (0.25 mmol) of -phenyl-bis(trifluoromethanesulfonimide) () and 32 mg (0.10 mmol) of were added to a scintillation vial and flushed with nitrogen. Anhydrous tetrahydrofuran (THF, 3 mL) followed by (2.86 mmol) of TEA were added to the reaction vessel and it was tightly capped. The reaction was carried out overnight at room temperature. Solvent was removed under reduced pressure and the product was purified by column chromatography, eluting with (100:3) (Fig. 1). calculated for ; measured: .
9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-4-yl trifluoromethanesulfonate (6)
64.8 mg (0.181 mmol) of and 23.8 mg (0.0712 mmol) of were added to a scintillation vial and flushed with nitrogen. Anhydrous THF (3 mL) and (2.04 mmol) of TEA were added to the reaction vessel and it was tightly capped. The reaction was carried out overnight at room temperature. The solvent was removed under reduced pressure and the product was purified by column chromatography, eluting with (100:3) (Fig. 1). calculated for ; measured: .
Sonogashira Nile Red derivative synthesis
4-([9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-2-yl]ethynyl)-2-(methylsulfonyl)benzaldehyde [Nile Red 6 (NR6)]
1.6 mg (0.009 mmol) of palladium(II) chloride (), 1.8 mg (0.0095 mmol) of copper(I) iodide (CuI), and 2.3 mg (0.0088 mmol) of triphenylphosphine () were added to a round-bottom flask. All solvents were sparged with nitrogen prior to use and the reaction was performed under a blanket of nitrogen. 6.1 mg (0.013 mmol) of was dissolved in DMF (2 mL) and added to the flask followed by (0.3803 mmol) of TEA. Trimethylsilylacetylene (TMS acetylene) was cooled in an ice bath and (0.3826 mmol) were added to the reaction mixture. The reaction was run overnight at room temperature. The following morning, (0.156 mmol) of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU), (0.006 mmol) of water, and 11.4 mg (0.043 mmol) of 4-bromo-2-(methylsulfonyl)benzaldehyde were added to the reaction, and it was heated to 70°C for 24 h (Fig. 2). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
2-[(3-aminophenyl)ethynyl]-9-(diethylamino)-5H-benzo[a]phenoxazin-5-one [(Nile Red 7 (NR7)]
0.8 mg (0.0045 mmol) of , 1.9 mg (0.01 mmol) of CuI, and 2.4 mg (0.0092 mmol) of were added to a round-bottom flask. All solvents were sparged with nitrogen prior to use and the reaction was performed under a blanket of nitrogen. 7.8 mg (0.0167 mmol) of were dissolved in DMF (2 mL) and added to the flask followed by (0.3803 mmol) of TEA. TMS acetylene was cooled in an ice bath and (0.3826 mmol) were added to the reaction mixture. The reaction was run overnight at room temperature. The following morning (0.20 mmol) of DBU, (0.007 mmol) of water, and (0.167 mmol) of 3-bromoaniline were added to the reaction, and it was heated to 70°C for 3 h (Fig. 2). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
3-([9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-2-yl]ethynyl)benzaldehyde [Nile Red 8 (NR8)]
1.0 mg (0.0056 mmol) of , 1.2 mg (0.0063 mmol) of CuI, and 2.7 mg (0.0103 mmol) of were added to a round-bottom flask. All solvents were sparged with nitrogen prior to use and the reaction was performed under a blanket of nitrogen. 11.2 mg (0.024 mmol) of were dissolved in DMF (2 mL) and added to the flask followed by (0.2368 mmol) of TEA. TMS acetylene was cooled in an ice bath and (0.2382 mmol) were added to the reaction mixture. The reaction was run overnight at room temperature. The following morning (0.288 mmol) of DBU, (0.0096 mmol) of water, and 5.8 mg (0.025 mmol) of 3-iodobenzaldehyde were added to the reaction, and it was heated to 75°C for 3 h (Fig. 2). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
3-([9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-2-yl]ethynyl)-4-methoxybenzaldehyde [Nile Red 9 (NR9)]
1.4 mg (0.008 mmol) of , 1.7 mg (0.0089 mmol) of CuI, and 2.3 mg (0.0088 mmol) of were added to a round-bottom flask. All solvents were sparged with nitrogen prior to use and the reaction was performed under a blanket of nitrogen. 13 mg (0.028 mmol) of were dissolved in DMF (1 mL) and added to the flask followed by (0.3803 mmol) of TEA. TMS acetylene was cooled in an ice bath and (0.3826 mmol) were added to the reaction mixture. The reaction was run overnight at room temperature. The following morning (0.336 mmol) of DBU, (0.011 mmol) of water, and 8.4 mg (0.039 mmol) of 4-bromo-2-(methylsulfonyl)benzaldehyde were added to the reaction, and it was heated to 70°C for 24 h (Fig. 2). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
Suzuki Nile Red derivative synthesis
3-(4-aminophenyl)-9-(diethylamino)-5H-benzo[a]phenoxazin-5-one [Nile Red 1 (NR1)]
14.5 mg (0.031 mmol) of , 0.3 mg (0.0013 mmol) of palladium (II) acetate , 0.6 mg (0.0023 mmol) of , 5.8 mg (0.1 mmol) of potassium fluoride (KF), and 7.8 mg (0.036 mmol) of 4-aminophenylboronic acid pinacol ester were added to a round-bottom flask and subsequently flushed with nitrogen. 1 mL of DMF was degassed under vacuum with sonication and added to the reaction vessel. The reaction was run at 70°C overnight (Fig. 3). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
4-[9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-3-yl]benzonitrile [Nile Red 5 (NR5)]
14.5 mg (0.031 mmol) of , 0.3 mg (0.0013 mmol) of , 0.6 mg (0.0023 mmol) of , 5.8 mg (0.1 mmol) of KF, and 6.4 mg (0.044 mmol) of 4-cyanophenylboronic acid were added to a round-bottom flask and subsequently flushed with nitrogen. 1 mL of DMF was degassed under vacuum with sonication and added to the reaction vessel. The reaction was run at 70°C overnight (Fig. 3). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
2-(4-aminophenyl)-9-(diethylamino)-5H-benzo[a]phenoxazin-5-one [Nile Red 10 (NR10)]
8.0 mg (0.017 mmol) of , 0.3 mg (0.0013 mmol) of , 0.5 mg (0.0019 mmol) of , 3.1 mg (0.053 mmol) of KF, and 5.0 mg (0.023 mmol) of 4-aminophenylbornic acid pinacol ester were added to a round-bottom flask. The flask was flushed with nitrogen and nitrogen-sparged THF () was added. The reaction was run at 60°C overnight (Fig. 3). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
4-[9-(diethylamino)-5-oxo-5H-benzo[a]phenoxazin-2-yl]benzonitrile [Nile Red 12 (NR12)]
7.4 mg (0.022 mmol) of , 0.2 mg (0.0011) of , 0.7 mg (0.0027) of , and 3.9 mg (0.0265 mmol) of 4-cyanophenylboronic acid were added to a round-bottom flask. The flask was flushed with nitrogen and of nitrogen sparged THF was added. A 2 M solution of potassium carbonate was sparged with nitrogen and was added to the reaction mixture. The reaction was run at 60°C overnight (Fig. 3). The final compound was purified via HPLC with a water:acetonitrile gradient of 90:10 to 10:90 over 20 min. calculated for : ; measured: .
Cell Culture for Nile Red Derivative Staining
The U2OS human osteosarcoma cell line was cultured in Dulbecco’s modified Eagle medium (DMEM, Thermo Fisher Scientific) with 10% fetal bovine serum (FBS, VWR, Radnor, Pennysylvania) and 1% penicillin/streptomycin (Thermo Fisher Scientific) at 37°C in 5% . For live cell imaging studies, FluoroBrite DMEM (Thermo Fisher Scientific) was used in place of DMEM. For imaging studies, cells were grown in 96-well glass bottom plates (Cellvis, Mountain View, California) for 48 h, with a final confluence of . Nile Red and its derivatives were prepared at 1 to 10 mM stock concentration in dimethyl sulfoxide (DMSO) and diluted into media or buffer for cell staining and imaging studies.
Nile Red Derivative Cell Staining and Imaging Studies
Cell staining studies were completed on fixed and live U2OS cells using Nile Red and each synthesized Nile Red derivative described as follows. For fixed cell staining, cells were washed once with phosphate-buffered saline (PBS), then fixed with 4% paraformaldehyde (PFA) for 20 min at room temperature. The PFA was removed and each well was washed three times for 5 min per wash () with PBS. The Nile Red fluorophores were diluted into PBS to a final concentration of , incubated for 5 min with the cells, and images were collected without washing. Live cell staining was completed using two methods. First, cells were washed once with FluoroBrite DMEM. The Nile Red fluorophore solution was prepared in FluoroBrite DMEM by diluting to a final concentration of for staining, where cells were stained for 5 min. The fluorophore-containing media were removed and the cells were fixed for 20 min using 4% PFA. Cells were then washed with FluoroBrite DMEM prior to imaging. Second, cells were washed with FluoroBrite DMEM without FBS once for 5 min. Nile Red fluorophore solutions were prepared in FluoroBrite without FBS at final concentration for cell staining. The cells were stained with the prepared solution for 5 min and imaged in the staining solution. All images were collected at their ideal exposure times to permit visualization of stained intracellular structures.
Nile Red Derivative Ex Vivo Tissue Staining, Imaging, and Fluorescence Quantification
Cryosections of CD-1 mouse sciatic nerve and surrounding muscle and adipose tissues from unrelated experiments were cut from optimal cutting temperature (OCT, Thermo Fisher Scientific) embedded frozen tissues. Cryosectioned tissues were captured on superfrost plus slides (Thermo Fisher Scientific) at thickness. The tissue sections were washed with PBS for 2 min to remove OCT, and then fixed with 2% PFA for 15 min. The tissue sections were then washed with 0.1% Tween 20 in PBS (), followed by incubation with of Nile Red and each Nile Red derivative formulated using 10% DMSO, 5% Kolliphore (Sigma-Aldrich), 63.75% FBS, and 21.25% PBS as previously described.26 Tissue sections were incubated with the formulated fluorophores for 40 min at room temperature, followed by washing with blank formulation not containing fluorophore (). Each tissue section was then washed with PBS (), fixed for 15 min using 2% PFA, followed by a final wash with PBS for 5 min before mounting with glycerol (Thermo Fisher Scientific). Images were collected at an exposure time of 650 ms for all Nile Red and Nile Red derivative-stained sections. Four tissue sections were stained for Nile Red and each Nile Red derivative, facilitating adipose-to-muscle and adipose-to-nerve contrast ratio quantification where tissue intensities were quantified using region of interest analysis with custom written MatLab code (The MathWorks, Natick, Massachusetts). For each tissue type, a representative area was selected, where a threshold-based mask was set to determine tissue from background. The mean tissue intensity was calculated in adipose, nerve, and muscle tissues as the intensity per second and used to calculate the adipose-to-nerve and adipose-to-muscle contrast ratios.
All ex vivo and cellular fluorescence and white light images were collected using a Zeiss AxioObserver inverted fluorescence microscope (Carl Zeiss, Oberkochen, Germany). All cell images were collected at magnification and all tissue section images were collected at magnification. Fluorescence excitation was completed using a Photofluor II broadband light source (Burlington, Vermont) filtered with a bandpass excitation filter. Fluorescence images were collected using a bandpass emission filter and an AxioCam 506 monochrome camera (Zeiss). All filters were purchased from Chroma Technologies (Bellows Falls, Vermont).
In Vivo Imaging Following Direct/Topical Administration in Rodent Nerves
In vivo images were acquired using a previously published, custom-built small animal fluorescence imaging system,27 described briefly here. Fluorescence images were obtained using a Qimaging EXi Blue monochrome camera (Surrey, British Columbia, Canada) with a removable Bayer mask for collection of coregistered color and fluorescence images. A Photofluor II broadband light source was expanded onto the surgical field through a liquid light guide and used unfiltered for white light illumination. Nile Red and its derivatives were excited using a bandpass filter, while fluorescence emission images were collected with a -bandpass filter. Oxazine 4 (Exciton, West Chester, Ohio) was used alone as well as for nerve and adipose costaining and ratiometric imaging studies. Oxazine 4 was excited using a -bandpass filter, while fluorescence emission images were collected through a bandpass filter.
Approval for the use of animals in the study was obtained from the Institutional Animal Care and Use Committee at Oregon Health and Science University. In vivo staining studies were carried out on male CD-1 mice using an optimized direct/topical administration protocol.2 Staining was performed concurrently on no more than two nerve sites (sciatic nerve or brachial plexus) per mouse. Nile Red and all Nile Red derivatives were formulated in a previously published cosolvent formulation (10% DMSO, 5% Kolliphor, 63.75% serum, and 21.25% PBS14) at a final concentration of . Each formulated fluorophore was incubated for 5 min on the nerve site ( nerve sites/fluorophore), followed by nine short flushes with PBS. The nerve site was then incubated with blank cosolvent not containing fluorophore for 5 min, followed again by nine short flushes with PBS.
In vivo costaining was completed with (1) Nile Red and oxazine 4 as well as (2) NR5 and oxazine 4. For these studies, cosolvent-formulated oxazine 4 () was incubated on the nerve site for 5 min followed by nine short flushes with PBS. The cosolvent-formulated Nile Red or NR5 () was then incubated on the nerve site for 5 min and also followed by nine short flushes with PBS. Coregistered (1) Nile Red and oxazine 4 or (2) NR5 and oxazine 4 images were collected from nerve sites per groups (1) and (2). For comparison, oxazine 4 () only in vivo staining using the optimized direct administration protocol was also completed on nerve sites.
Ratiometric Image Calculation and Statistical Analysis
Ratiometric images were calculated and quantified using region of interest analysis on the white light images to identify the nerve, adipose, and muscle tissue types. Due to the known cross talk of oxazine 4 into the Nile Red channel,2 the oxazine 4 image was subtracted from the Nile Red or NR5 image resulting in an image of only adipose. The ratiometric images were then calculated as the oxazine 4 image minus the corrected Nile Red or NR5 image divided by the corrected Nile Red or NR5 image. Oxazine 4 only images were generated from single fluorophore staining and were not ratiometric in nature. Receiver operator characteristic (ROC) curve analysis was performed to assess differences among the ratiometric images, where the area under the curve (AUC) was used as the metric to differentiate between groups (1) and (2) using custom written MatLab code.28 Oxazine 4 only images were also compared to images from groups (1) and (2) to assess contrast improvement using ratiometric imaging. Statistical significance between the three staining conditions was determined using the method described by Hanley and McNeil29 with standard error measurements and correlation coefficients for pixel intensity values in each channel used for -score value calculations.28 A -score indicated that the populations were statistically different with 95% confidence, while a -score indicated that the populations were statistically different with 99.7% confidence.29
Synthesis of the Nile Red Core Derivatives
The syntheses of the Nile Red core structures were completed in accordance with previously described methods (Fig. 1).30 The nitroso compound was prepared by reacting 3-diethylaminophenol with sodium nitrite on ice in HCl for 2 h. The product was filtered and used immediately without further purification. Syntheses of compounds and were completed by condensation of and either 1,6-dihydroxynapthalene or 1,7-dihydroxynapthalene at 100°C. The resulting 2-hydroxy or 3-hydroxy Nile Red derivatives were converted to their triflated versions (compounds and ) via reaction with -phenyl bis(trifluoromethanesulfonimide) at room temperature. Compounds and were readily purified using column chromatography (, 100:3).
Nile Red derivatives NR6, NR7, NR8, and NR9 were prepared using a one pot Sonogashira reaction (Fig. 2).31 Sparging with nitrogen was sufficient to remove dissolved oxygen from the solvent, and the reactions were carried out overnight with a catalytic amount of palladium, , and CuI, in the presence of excess TEA. To induce the addition of the target aryl group, the aryl halide was added with catalytic water and excess DBU. The reaction was heated to 70°C and run for 3 to 24 h.
Nile Red derivatives NR1, NR5, NR10, and NR12 were prepared using a Suzuki coupling reaction (Fig. 3).32,33 Rigorous degassing with sonication and house vacuum prior to use of any solvents yielded the most reliable reactions, as reagents were oxygen sensitive. Boronic acids, KF, the triflated Nile Red compounds, a catalytic amount of , and were added to the reaction vessel under nitrogen. The reactions were heated to 60°C to 70°C and run overnight.
Photophysical Properties of Nile Red Derivatives
The resulting eight Nile Red derivatives all exhibited similar absorption and emission characteristics to one another as well as to Nile Red (, in DMSO). Spectral evaluation in DMSO displayed absorbance maxima from all eight Nile Red derivatives between 555 and 570 nm, while the emission maxima ranged from 635 to 640 nm [Fig. 4(a)]. Nile Red has well-characterized solvochromatic behavior due to a twisted intramolecular charge transfer state formed by the rotation of the diethyl amino group in the excited state. Rotation about this bond only occurs in polar solvents where it is stabilized, resulting in a bathochromic shift.23,34 As expected, all synthesized Nile Red derivatives demonstrated similar solvochromatic behavior to Nile Red (, in toluene) when spectrally evaluated in toluene, with absorbance maxima between 520 and 545 nm and emission maxima ranging from 575 to 590 nm [Fig. 4(b)]. The quantum yield was determined for all synthesized derivatives in DMSO and varied widely across the Nile Red derivatives ranging from 0.04 to 0.56 (Table 1). Notably, derivatives with an electron withdrawing group (NR5, NR6, NR8, NR9, and NR12) were significantly brighter than those with electron donating groups (NR1, NR7, and NR12), where derivatives with electron withdrawing groups had quantum yields , while derivatives with electron donating groups had quantum yields .
Optical properties of Nile Red derivatives in DMSO.
|NR Derivative||Max Abs (nm)||Max Em (nm)||Stokes Shift (nm)||QY|
Nile Red Derivative Cell Staining and Imaging Studies
To assess the effect of the Nile Red derivative structural diversity on lipid accumulation, fixed and live cells were stained with Nile Red and the synthesized derivatives (Fig. 5). In fixed cells, the fluorescence intensity was bright for Nile Red as well as NR5 and NR8, where both derivatives contained electron withdrawing groups and had high quantum yields. Interestingly, quantum yield was not directly correlated to in vitro fluorescence intensity since NR1 had a relatively low quantum yield () but showed brighter in vitro fluorescence than NR12 () as determined by the optimal exposure times used for image collection [Fig. 5(a)]. As expected, lipid droplets were the prominently visible in vitro structure with Nile Red as well as its derivatives.
Fluorescence staining patterns of Nile Red and the synthesized derivatives were also assessed in cells that were stained live, then fixed [Fig. 5(b)]. Fluorescence intensity was substantially decreased, as determined from the long exposure times, which was likely due to the extra washing steps used during staining. NR6 was the only derivative to have an exposure time , suggesting it bound more strongly to subcellular structures than Nile Red or the other Nile Red derivatives. Overall, fluorescent staining of lipid droplet-like structures appeared diminished and more cytoplasmic structures were stained when washing was completed prior to imaging. This was particularly evident for NR6, NR7, and NR10.
Finally, the fluorescence staining patterns of Nile Red and the synthesized derivatives were assessed in live cells without washing [Fig. 5(c)]. Fluorescence intensity was generally brightest in this staining study compared to staining of fixed or live cells that were fixed prior to imaging, as determined by the short exposure times. This increase in fluorescence signal was likely due to the lack of washing in the staining procedure. Nile Red was again the brightest overall, with NR5 and NR8 being the brightest of the synthesized derivatives. Prominent lipid droplet-like structures were observed for Nile Red, NR1, NR5, NR9, and NR12. Despite the low quantum yield of NR1, it was again brighter than other Nile Red derivatives with higher quantum yields such as NR12 and NR6 as determined based on optimal exposure time used for image collection.
Nile Red Derivative Ex Vivo Tissue Staining
A subclass of oxazine derivatives with structural similarity to Nile Red and the synthesized derivatives have demonstrated significant nerve specificity.26 To determine if any of the structural changes to the Nile Red scaffold in the synthesized derivatives affected the staining specificity for adipose, muscle, or nerve, the synthesized derivatives were used to stain mouse tissue sections containing mouse sciatic nerve as well as the surrounding muscle and adipose tissues (Fig. 6). Nile Red showed adipose specificity as did NR5 and NR12, while little fluorescence staining was seen from the other Nile Red derivatives on mouse tissue sections [Fig. 6(a)]. Region of interest fluorescence intensity analysis showed strong nerve and adipose fluorescence staining from Nile Red and NR5, with substantially lower muscle fluorescence staining intensity [Fig. 6(b)]. The adipose-to-nerve ratio for Nile Red and all of its derivatives was similar to autofluorescence adipose-to-nerve ratio, showing minimal contrast increase from fluorophore staining. Notably, adipose-to-muscle ratio showed increased contrast over control autofluorescence adipose-to-muscle ratio for Nile Red and NR5 [Fig. 6(c)].
Nile Red Derivative In Vivo Tissue Staining and Ratiometric Imaging
To further assess the tissue specificity of the synthesized Nile Red derivatives, in vivo studies were completed where formulated Nile Red and its derivatives were applied to the sciatic and brachial plexus nerves using an optimized direct administration protocol,2 which facilitated similar application to the ex vivo frozen tissue section staining. Fluorescence imaging was completed immediately following staining (Fig. 7) and 30 min after completion of staining (Fig. 8) to determine the utility for contrast during a surgical procedure. In vivo imaging immediately following direct administration showed Nile Red and its derivatives localized to the adipose tissue [Fig. 7(a)]. Tissue-specific fluorescence analysis showed the greatest fluorescence intensity in the adipose tissue, followed by the nerve then muscle tissues. Nile Red and NR5 had the highest fluorescence signal compared to the other Nile Red derivatives, with Nile Red substantially higher than NR5 [Fig. 7(b)]. Adipose–to-nerve ratios showed increased contrast over control autofluorescence with similar adipose-to-nerve contrast from Nile Red, NR1, NR5, NR6, NR10, and NR12. Adipose-to-muscle ratios also showed greater contrast compared to control autofluorescence with the greatest adipose to muscle contrast seen in the Nile Red-stained tissues. While lower than Nile Red, adipose-to-muscle contrast was similar for NR1, NR5, NR10, and NR12 [Fig. 7(c)].
Fluorescence imaging 30 min after direct administration showed qualitatively similar nerve, adipose, and muscle staining patterns to those observed immediately following direct administration [Figs. 7(a) and 8(a)]. Tissue-specific fluorescence analysis showed a decrease in adipose, nerve, and muscle fluorescence for Nile Red compared to imaging immediately following direct administration. However, NR5 showed slightly increased adipose-specific fluorescence intensity, while nerve and muscle tissue fluorescence were unchanged [Figs. 7(b) and 8(b)]. Adipose-to-nerve contrast was higher than control autofluorescence for Nile Red and all its derivatives except NR7 30 min postdirect administration. Notably, NR5 had higher adipose-to-nerve contrast compared to Nile Red 30 min postdirect administration, while adipose-to-muscle contrast was greatest for Nile Red [Fig. 8(c)]. NR5 tissue-specific contrast was also assessed out to 120 min after direct administration, where adipose signal was maintained throughout the 120-min imaging time.
To further assess the in vivo adipose- and nerve-specificity differences between Nile Red and NR5, ratiometric imaging was completed using Nile Red or NR5 and oxazine 4 costaining compared to oxazine 4 staining alone [Fig. 9(a)]. ROC curve analysis was completed to quantitatively assess nerve contrast differences between Nile Red and NR5 with oxazine 4 costaining versus oxazine 4 staining alone. AUC calculations from the ROC curve analysis showed similar nerve-to-adipose contrast ratios for costaining using oxazine 4 with NR5 or Nile Red or oxazine 4 staining alone immediately following direct administration [Fig. 9(b)]. Although all AUC values were , the differences in nerve-to-adipose contrast using ratiometric imaging compared to oxazine 4 staining alone were found to be statistically significant ( for NR versus NR5 costaining with oxazine 4, for oxazine 4 only staining versus NR5 costaining with oxazine 4). ROC AUC analysis demonstrated that NR5 and oxazine 4 costaining provided significantly higher nerve-to-adipose contrast than Nile Red and oxazine 4 costaining or oxazine 4 staining alone 30 min after staining was completed ( for NR versus NR5 costaining with oxazine 4, for oxazine 4 only staining versus NR5 costaining with oxazine 4).
ROC AUC analysis of nerve-to-muscle contrast immediately following staining showed slightly improved values for NR5 costaining with oxazine 4 compared to Nile Red and oxazine 4 costaining or oxazine 4 only staining. Calculation of the -score revealed that the difference in AUC for NR5 and oxazine 4 costaining verus either Nile Red and oxazine 4 costaining or oxazine 4 alone were statistically significant ( for NR versus NR5 costaining with oxazine 4, for oxazine 4 only staining versus NR5 costaining with oxazine 4). ROC AUC analysis of the nerve-to-muscle contrast 30 min after staining revealed little change for NR5 and oxazine 4 costaining and improved values for oxazine 4 staining alone, while Nile Red and oxazine 4 costaining show a substantial drop in AUC value. Calculation of the showed that NR5 costaining with oxazine 4 had significantly better nerve-to-muscle contrast compared to Nile Red costaining with oxazine 4 (). However, nerve-to-muscle contrast was not significantly better 30 min after staining with NR5 and oxazine 4 costaining versus oxazine 4 alone ().
Current surgical techniques rely heavily on a surgeons’ ability to identify normal and diseased tissues using visual and tactile cues during surgery, where outcome is strongly influenced by a surgeon’s experience and skill level. The peripheral nervous system is responsible for voluntary and involuntary activities and critically important to bodily physical functions and maintenance of biochemical homeostasis.35 Thus, surgical peripheral nerve damage can be devastating with potentially lifelong, deleterious ramifications for affected patients’ quality of life. Fluorescence-guided surgery provides an opportunity for rapid, high contrast identification of specific tissue types, such as nerves. However, there are currently few clinically approved fluorophores, none of which are nerve specific.7,13 The oxazine fluorophore scaffold presents a unique chemical scaffold for tissue-specific fluorescence as modifications to its backbone structure result in both nerve- and adipose-specific fluorescent contrast.18,22,23 Accurate intraoperative nerve identification using nerve-specific fluorophores is often hampered by surrounding adipose tissue, since the molecular composition of nerve and adipose tissues is similar.2,13 Therefore, small molecule fluorophores that distribute and accumulate in nerve tissue often demonstrate some degree of off target adipose accumulation and contrast. This is particularly problematic for some classes of nerve-specific small molecules including the stilbene derivatives,10 distyrylbenzene derivatives,12–15 and FM fluorophores.16,17 To improve nerve-specific fluorescent contrast for intraoperative guidance, a library of oxazine-based, adipose-specific fluorophores was designed and synthesized to enable ratiometric imaging for enhancement of nerve-specific intraoperative contrast when used as a costain with nerve-specific fluorophores that demonstrate mixed nerve and adipose accumulation.
The Nile Red scaffold was chosen as the backbone fluorophore for library synthesis due to its known lipophilicity- and adipose-highlighting properties. While Nile Red itself does provide adipose-specific fluorescence, it also stains nerve tissue, diminishing nerve-to-adipose tissue contrast by ratiometric imaging [Figs. 7(a)–9(a)]. The Nile Red scaffold was modified (Figs. 1Fig. 2–3) to diminish nerve specificity while maintaining the adipose specificity for improved nerve contrast using a ratiometric imaging approach. While agents with decreased nerve accumulation were generated, their excitation and emission fell outside the ideal NIR range for in vivo imaging (Fig. 4). Similarly, oxazine 4, which was used here as the nerve-specific fluorophore, also has excitation and emission outside of the NIR range as well as minimal adipose accumulation. Thus, in the current studies these agents were used for proof of concept to demonstrate that ratiometric imaging following costaining with nerve- and adipose-specific fluorophores could improve nerve-specific contrast. However, to enable clinical utility, additional synthesis to generate NIR nerve- and adipose-specific small molecules would be beneficial. Of note, synthesis of NIR nerve-specific fluorophores may result in more mixed nerve and adipose accumulation, highlighting the utility of adipose-specific small molecule fluorophores.
The Nile Red derivatives were screened for tissue specificity ex vivo on rodent sciatic nerve sections to assess the utility of ratiometric imaging for enhanced nerve-specific fluorescent contrast. NR5 demonstrated the brightest overall tissue staining compared to the other Nile Red derivatives; however, Nile Red demonstrated the highest overall tissue fluorescence [Figs. 6(a) and 6(b)]. The utility of the Nile Red derivatives for adipose-specific signal was further assessed following direct administration in vivo immediately after as well as 30 min following the completion of staining [Figs. 7(a) and 8(a)]. In vivo direct administration in rodent sciatic and brachial plexus nerves showed strong fluorescence intensity, following Nile Red staining in the adipose, nerve, and muscle tissues immediately following staining [Fig. 7(b)]. Nile Red staining per tissue decreased over the 30-min period after staining, while NR5 showed increased adipose intensity with unchanged nerve and muscle tissue fluorescence 30 min postdirect administration [Fig. 8(b)]. NR5 showed the highest fluorescence intensity in the adipose, nerve, and muscle tissues of the synthesized Nile Red derivatives both immediately poststain as well as at 30 min after staining [Figs. 7(b) and 8(b)]. The adipose-to-nerve and adipose-to-muscle ratios were calculated to further assess tissue-specific contrast. The mean adipose-to-nerve ratios were similar for NR1, NR5, NR10, NR12, and the parent Nile Red compound immediately after staining [Fig. 7(c)]. Adipose-to-nerve ratio decreased 30 min after staining for Nile Red, while it increased for NR5 demonstrating the potential utility of this agent to highlight adipose-tissue specificity in the context of nerves [Figs. 7(c) and 8(c)]. The change in the adipose-to-nerve ratio was largely due to an overall decrease in nerve and adipose fluorescence signal in the Nile Red group and an increase in the adipose signal for the NR5 group improving overall contrast [Figs. 7(b) and 8(b)]. Adipose-to-muscle signal was greatest for Nile Red compared to all Nile Red derivatives immediately poststain as well as up to 30 min following staining [Figs. 7(c) and 8(c)].
Direct administration using Nile Red and oxazine 4 costaining was previously optimized, demonstrating adipose- and nerve-specific staining, respectively; however, Nile Red accumulation in the nerve tissue was visible hindering ratiometric analysis for improved nerve contrast.2 NR5 was selected for ratiometric imaging studies due to its tissue-specific staining and adipose-to-nerve contrast to determine if contrast was improved over Nile Red and oxazine 4 costaining or oxazine 4 staining alone. Ratiometric images using both Nile Red and NR5 to specifically highlight the adipose tissue and oxazine 4 to highlight nerve tissue resulted in high contrast immediately poststain as well as 30 min after staining [Fig. 9(a)]. ROC curve analysis showed the most significant improvement in nerve-to-adipose contrast 30 min poststaining, where the AUC was decreased for Nile Red and oxazine 4 costaining but maintained for NR5 and oxazine 4 costaining [Fig. 9(b)]. Notably, the nerve-to-muscle contrast did not show the same level of improvement as nerve-to-adipose contrast using either Nile Red or NR5 costaining with oxazine 4, which was likely due to the fact that oxazine 4 does not show substantial adipose or muscle accumulation. ROC curve analysis showed similar AUC immediately poststaining, where a statistically significant improvement was seen using NR5 costaining with oxazine 4 versus oxazine 4 alone. However, at 30 min poststaining, the nerve-to-muscle contrast was not statistically different between NR5 and oxazine 4 costaining versus oxazine 4 staining alone. This points to a limitation of the current compounds, where only nerve-to-adipose contrast would be improved during the surgical time frame using the dual staining technique whereas nerve-to-muscle contrast would not benefit. Additional synthetic optimization is ongoing to generate an adipose-specific derivative that can both reach NIR wavelengths and improve nerve-to-muscle contrast, justifying the use of two small molecule fluorophores instead of a single agent for nerve-specific contrast clinically.
In summary, a Nile Red derivative, NR5, was synthesized and characterized for its adipose specificity as a proof of concept small molecule to enable ratiometric imaging for improved nerve-specific contrast. NR5 showed both adipose specific contrast as well as minimal nerve accumulation, differentiating it from the parent compound Nile Red, which demonstrates some nerve accumulation. In vivo costaining with nerve-specific oxazine 4 and NR5 showed improved nerve-to-adipose contrast as compared to the parent Nile Red compound due to this decreased nerve accumulation. This contrast improvement was most apparent 30 min postdirect administration, demonstrating the potential utility for surgical applications. These proof of concept studies show the potential utility of ratiometric imaging to improve contrast in tissues such as nerve, where development of a tissue-specific contrast agent that does not accumulate in other surrounding tissue is a continued challenge.
We would like to thank Sarah Cook and Evan Muschler for experimental assistance. This work was funded by the National Cancer Institute (R01CA188491) and the National Institute for Biomedical Imaging and Bioengineering (R01EB021362).