2.1.

Materials

Indocyanine green (ICG) and ninhydrin were purchased from Fluka, and 1,3-diphenylisobenzofuran (DPIBF), phenyltrimethoxysilane (PTMS), methyltrimethoxysilane (MTMS), aminopropyltrimethoxysilane (APTMS), ammonium hydroxide, nitric acid, and 2-iminothiolane hydrochloride (2-IT) were obtained from Aldrich. Sulfosuccinimidyl 4-[ $N$ -maleimidomethyl]cyclohexane-1-carboxylate (Sulfo-SMCC) was purchased from Pierce. PBS buffer was self made. Other materials used are agarose (GPG/LE, American Bioanalytical) and intralipid (Lyposin 10%, Abbott laboratories). All dyes and chemicals were used without further purification.

2.2.

ICG PEBBLE Synthesis

ICG-embedded ormosil nanoparticles were synthesized based on a sol-gel process combined with microemulsion polymerization.^{24, 25, 30} $80\mu \mathrm{L}$ of MTMS was added to $31\mathrm{mL}$ of water containing $34\mu \mathrm{L}$ of nitric acid. The mixture was heated to $60°\mathrm{C}$ in a water bath and stirred at 1,000 RPM for $30\min$ on a hot plate stirrer (Signature 575, VWR International, West Chester, Pennsylvania). The reaction mixture was maintained under constant heating and stirring for 2 more hours after quick addition of $6\mathrm{mL}$ of preheated 30% ammonium hydroxide. Then, the reaction mixture was cooled to 35°C or less. $1\mathrm{mL}$ of solution containing $6\mathrm{mg}$ of ICG and $100\mu \mathrm{L}$ of MTMS was added to the mixture. The reaction was run for another $1\mathrm{h}$ , and $10\mu \mathrm{L}$ of APTMS was added to give surface amine functionality for conjugation with antibody. After $1\mathrm{h}$ , the product was filtered with a Magna nylon membrane filter (pore size $0.1\mu \mathrm{m}$ ) with air suction. The products were resuspended in water and filtrated with an Amicon filtration system five times to remove impurities such as nonreacted monomers, dyes, and other chemicals. The resultant nanoparticles were recovered by additional suction filtration and dried under nitrogen atmosphere. Drying is necessary in order to avoid deactivation of ICG dyes in liquid environment. It also allowed us to obtain accurate particle number density information by weighing the dried powder. Morphology of the ICG ormosil PEBBLEs was characterized in the dried phase with scanning electron microscopy (XL30 FEG SEM, Philips) and in the wet phase with dynamic light scattering particle sizing (Nicomp 380/ZLS, Particle Sizing System).

The mechanism of nanoparticle formation can be described as follows: Hydrophobic monomer PTMS forms droplets in aqueous solution by fast stirring. The MTMS molecules inside the droplets are hydrolyzed in an acidic environment, from the surface of the droplets inward, resulting in homogeneous droplets with sub micrometer size. Addition of excess ammonium hydroxide highly basifies the solution, and thus, formation and growth of nanoparticles is triggered by a condensation reaction and generation of a three-dimensional (3-D) $\text{-}\mathrm{Si}\text{-}\mathrm{O}\text{-}\mathrm{Si}\text{-}$ network. To avoid deactivation of dye by heat, the reaction mixture needs to be cooled to 40°C or lower before addition of ICG. Addition of the second monomer, MTMS, helps to entrap ICG dye molecules in the network structure by making the nanoparticles more hydrophilic and forming a denser network. APTMS is used to functionalize the surface of the nanoparticle by forming primary amine groups for further surface modifications.

2.3.

PEBBLE Absorption and Dye Loading Efficiency

To estimate ICG dye loading capacity in nanoparticles, we prepared a set of calibration samples consisting of a known concentration of free ICG dye and blank ormosil nanoparticles. Blank particles are added to the calibration samples to match their light scattering coefficient to that of the ICG-PEBBLEs. Extinction spectra of several calibration samples with various dye concentrations were measured and compared to the extinction spectrum of the ICG-PEBBLE sample. Extinction values at $790\mathrm{nm}$ were plotted and fitted with a linear dependence on dye concentration.

2.4.

Aqueous Stability Measurement

Solutions of $0.1\mathrm{mg}∕\mathrm{mL}$ ICG ormosil PEBBLEs in water and $1\mu \mathrm{M}$ free ICG dye in water were prepared. They were covered with aluminum foil and stored at room temperature for 5 days. Absorption spectra of both PEBBLE and free dye solutions were measured daily, after ultrasonication for several minutes to remove aggregation.

2.5.

Photosensitization

The photodynamic efficacy of ICG dye has been investigated recently.^{16, 17, 18} To evaluate singlet oxygen generation from ICG-embedded ormosil PEBBLEs, a singlet oxygen-sensitive chemical probe, $1,3$ -diphenylisobenzofuran (DPIBF) was used. DPIBF can rapidly quench singlet oxygen, reducing its fluorescence emission near $450\mathrm{nm}$ . Because of the hydrophobicity of DPIBF, the photodynamic effect was measured in an $\mathrm{EtOH}$ -water $(50:50)$ mixture. A $0.1\mathrm{mg}∕\mathrm{mL}$ suspension of ICG-embedded ormosil PEBBLEs in an $\mathrm{EtOH}$ -water mixture and a $1\mu \mathrm{M}$ (concentration was arbitrarily chosen) solution of free ICG dye in an $\mathrm{EtOH}$ -water mixture were prepared. $0.2\mathrm{mL}$ of $10\mu \mathrm{M}$ DPIBF in $\mathrm{EtOH}$ -water solution was mixed with $2\mathrm{mL}$ of either ICG PEBBLE or free dye solution in a 4-mL capacity fluorescence quartz cuvette. The free ICG dye–or PEBBLE-DPIBF mixture was illuminated by a Ti-Sapphire laser at 800-nm wavelength with 1.4-mm beamwidth and 700-mW beam power (light fluence $1.36\mathrm{kJ}∕\mathrm{cm}2$ ), in continuous wave (CW) mode, for $30\mathrm{s}$ with shaking. Fluorescence spectra of DPIBF were measured by a fluorometer in the 420 to $620\mathrm{nm}$ range, with a 412-nm excitation wavelength, before and after laser illumination.

2.6.

PEBBLE-Antibody Conjugation

Specific binding of ICG PEBBLEs to cancer cells is achieved by conjugating HER-2 antibody (neu Ab-17, Labvision Corp.) onto its surface. HER-2 antibody is known to have a good affinity to breast cancer and prostate cancer cells.³¹ PEBBLE-antibody conjugation was performed following three steps: attaching a cross-linker, sulfo-SMCC, on the PEBBLE surface; thiolization of the antibody; and cross-linking mediated with sulfo-SMCC between PEBBLE and the thiolized antibody. First, $2\mathrm{mL}$ of ICG-embedded amine-functionalized ormosil PEBBLE $\left(12\mathrm{mg}\right)$ solution was added into $2\mathrm{mL}$ of PBS buffer ( $10\mathrm{mM}$ , pH 7.5) in a 20-mL scintillation vial. The solution was purged with argon and then treated with $3\mathrm{mg}$ of cross-linking agent, sulfo-SMCC (heterobifunctional, amine reactive on one side and thiol reactive on the other side). The glass vial was sealed with parafilm and covered with aluminum foil. The mixture was stirred in a 37°C water bath for $1\mathrm{h}$ . The reaction mixture was washed in a 10-mL Amicon ( $100\mathrm{kDa}$ filter membrane) with sodium phosphate buffer ( $100\mathrm{mM}$ , pH 7.2) two times. The final volume of solution was $4\mathrm{mL}$ .

Second, to add a free thiol group (-SH) on the antibody available for conjugation with an SMCC attached ICG PEBBLE, the antibody was treated with 2-iminothiolane hydrochloride (2-IT). $0.1\mathrm{mL}$ of a $1\mathrm{mg}∕\mathrm{mL}$ antibody $(6.7\bullet {10}^{-7}\mathrm{mmol})$ solution was mixed with $2\mathrm{mL}$ of 2-IT in PBS buffer solution containing the same number of 2-IT molecules. The vial was sealed with parafilm, and the mixture was stirred at room temperature for $2\mathrm{h}$ . Finally, $4\mathrm{mL}$ of SMCC attached ICG PEBBLE solution and $2\mathrm{mL}$ of thiolized antibody solution were mixed together. This mixture was purged with argon for 1 to $2\min$ and stirred overnight sealed with parafilm. Then, the mixture was washed in a 10-mL Amicon filter with sodium phosphate buffer (pH 7.2) two times and with DI water three times. The washed solution was carefully transferred back to a scintillation vial and lyophilized for 3 days. The prepared antibody-conjugated ICG-ormosil PEBBLE was stored in a freezer until future use.

2.7.

Cell Targeting Efficiency of Antibody-Conjugated ICG PEBBLE

We have tested the binding efficiency of antibodies to the PEBBLE surface and their binding to target cells. LNCaP cells, a cell line extracted from a lymph node metastasis of prostate cancer, were cultured in a cell chamber with two compartments (CultureSlides, BD $\mathrm{Falcon}™$ , BD $\mathrm{BioCoat}™$ ). Cells were attached to the bottom of the chamber and their surface density was less than 50%. $0.2\mathrm{mg}∕\mathrm{mL}$ of antibody-conjugated ICG-ormosil PEBBLE and bare ICG-ormosil PEBBLE (the same batch) in PBS buffer solution were prepared and filtered with $5\text{-}\mu \mathrm{m}$ syringe filters. They were adjusted to give similar absorption intensity at $800\mathrm{nm}$ by adding buffer solution. $1\mathrm{mL}$ of antibody-conjugated PEBBLE solution was added into one compartment, and $1\mathrm{mL}$ of unconjugated PEBBLE solution was added into the other compartment. The cell solutions were incubated for $1\mathrm{h}$ at room temperature, and the PEBBLEs were removed from the cell chamber by washing 2 times with PBS buffer.

The efficiency of antibody targeting was confirmed by measuring the photoacoustic signal from cells treated with antibody-conjugated PEBBLE and bare PEBBLE (control). A scanning photoacoustic microscope was used to produce high-resolution images of the cells in the two compartments. The setup is shown in Fig. 1 . A doubled YAG pulsed laser (Continuum, I-20) pumps an optical parametric oscillator (OPO; Continuum, Surelite OPO Plus) to produce 5-ns pulses with a tunable wavelength in the range of $690\mathrm{nm}$ to $980\mathrm{nm}$ . The output pulse energy was set to $8\mathrm{mJ}$ , and the beam diameter was expanded to $20\mathrm{mm}$ . The beam was directed to illuminate the cell chamber bottom. A single element, focused, high-frequency ultrasound transducer ( ${\mathrm{LiNbO}}_3$ $50\mathrm{MHz}$ , focal $\mathrm{length}=4\mathrm{mm}$ , f/1.5) detected the photoacoustic signal. The transducer focal plane was aligned with the chamber bottom surface. The transducer was mounted on a motorized XYZ translation stage to accurately scan the total volume of the cell chamber. The transducer signal is amplified (Panametrics, Model 5910PR), digitized by a digital oscilloscope (LeCroy, WaveSurfer 432), and transferred to a computer for storage. The computer was also used to control scanning.

Fig. 1

A high-resolution photoacoustic scanner consisting of a doubled YAG laser, an OPO, a beam expander (BX), a sample holder (CC), a high-frequency $\left(50\mathrm{MHz}\right)$ ultrasound transducer (UT), a motorized mechanical 2-D stage (SC), a signal amplifier (AM), and a digitizing oscilloscope (OS).

2.8.

Photoacoustic Imaging Setup

The scanning photoacoustic microscope described earlier is designed for imaging of tissue culture specimens. Given its limited depth of field ( $60\mu \mathrm{m}$ for the low f# transducer used) and long data acquisition time (limited by mechanical scanning), this instrument is not appropriate for clinical applications. To explore the potential of PEBBLEs for clinical applications, a second photoacoustic imaging system with lower spatial resolution but appropriate for clinical applications was developed. The system uses a clinical ultrasound scanner (Ultrasonix, Sonix RP). A software development kit, available from the scanner manufacturer, allows full control over the ultrasound beam former operation and signal acquisition hardware using $\mathrm{C}∕\mathrm{C}++$ programming language. We have utilized this feature to customize the scanner for a dual-mode operation where both B-mode ultrasound and photoacoustic imaging modes are implemented. Figure 2 shows the schematic of the dual-mode imaging system.

Fig. 2

Schematic of a photoacoustic imaging setup based on a clinical ultrasound scanner and a doubled YAG pulsed laser pumping an OPO. The beam is expanded (BX) before illuminating the phantom (PH). An ultrasound array probe (UA) detects the laser-generated acoustic waves. The ultrasound scanner (US) acquires signals in synchronization with the laser pulses. The scanner employs a beam-forming algorithm to form the photoacoustic image.

In the photoacoustic mode, the ultrasound scanner receives the acoustic emission (generated by the pulsed laser) using a linear array ultrasound probe (Ultrasonix L-14-5/38, 128 elements, bandwidth 5 to $14\mathrm{MHz}$ ). Laser pulses and ultrasound signal acquisition are synchronized using a common clock source (Agilent function generator) and a field programmable gate array (ezFPGA-C6-8 Module, Dallas Logic, Plano, Texas). The system is programmed to acquire the ultrasound signal on a single array element every laser pulse. A complete data set is acquired every 128 laser pulses $\left(6.4\mathrm{s}\right)$ . To reduce noise, 16 data sets are averaged before processing. A beam-forming algorithm, employing coherent delay-and-sum and implemented using MATLAB (Mathworks, Inc.) code, was applied to reconstruct the photoacoustic image.³²

The imaging system is more practical than the scanning photoacoustic microscope in three respects: electronic scanning for high frame rate; electronic focusing, providing uniform resolution; and the ability to change probes for different medical applications. We have used it to image phantoms containing small gelled ICG PEBBLE solutions embedded in optically diffusive gel (agarose 2%, intralipid 5%). The phantom was illuminated from the side while immersed in a water tank. The ultrasound transducer was positioned above the phantom.

3.1.

ICG PEBBLE Characterization

ICG-embedded ormosil PEBBLEs were synthesized based on sol-gel chemistry and yielded $70\mathrm{mg}$ of nearly monodispersed nanoparticles with 100-nm diam. An SEM image of the particles (dried phase) and particle size distribution obtained from a dynamic light scattering (DLS) particle sizer (wet phase) are shown in Fig. 3 . Only negligible difference in the mean particle size was observed between the dried phase and the wet phase, implying negligible swelling of particle matrix.

Fig. 3

(a) SEM image of ICG ormosil PEBBLE particles, and (b) particle size distribution obtained from dynamic light scattering particle sizing. (Image was taken from PEBBLE before antibody conjugation.)

3.2.

PEBBLE Absorption and Dye Loading Efficiency

Dye loading in a PEBBLE particle was tested by comparing its optical absorption with that of free ICG dye at different concentrations [see Fig. 4 (left)]. Blank ormosil particles were added to the free ICG solutions to match the scattering coefficient to that of ICG PEBBLEs. Linear curve fitting for peak absorption (at $\lambda =790\mathrm{nm}$ ) was evaluated to estimate the dye concentration in the PEBBLE particle [Fig. 4 (right)].

Fig. 4

(a) Dye loading in a PEBBLE particle was tested by comparing optical absorption of PEBBLE particles with optical absorption of free ICG dye at different concentrations. (b) Linear curve fitting for peak absorption (at $\lambda =790\mathrm{nm}$ ) was evaluated to estimate the dye concentration in the PEBBLE particle. (c) Dye leaching was examined by filtration to verify that the absorption was from ICG molecules encapsulated in the PEBBLE particle.

The estimated dye concentration in the PEBBLE sample in $0.1\mathrm{mg}∕\mathrm{mL}$ ICG ormosil PEBBLE solution is $3.6\mu \mathrm{M}$ . Assuming that the density of the ormosil nanoparticle is about $2.0\mathrm{g}∕{\mathrm{cm}}^3$ (Refs. 33, 34) and considering the average particle size of $100\mathrm{nm}$ , the particle number density of ${10}^{11}$ particle/mL can be computed using the weight of particles in the solution, particle volume, and density. This dye concentration corresponds to 23,000 ICG molecules per 1 ormosil nanoparticle. High loading capacity is advantageous in cell-targeting applications. Because the number of cell-membrane receptors per area is limited, high loading per particle provides a mechanism to enhance the signal from each targeted cell.

The characteristics of the ICG PEBBLE’s absorption spectrum still remain the same as that of the ICG free dye, although a minor change was observed in the reflecting interference from the ormosil matrix. According to Landsman ³⁵ and Zhou, ³⁶ at concentrations above $1\mathrm{mM}$ , ICG can forms, dimer or higher aggregated forms, resulting in a change of absorption peak shape represented by the rise of the dimer peak around $700\mathrm{nm}$ and the decline of the monomer peak at $790\mathrm{nm}$ . The shape and location of ICG PEBBLE’s absorption spectrum indicates that the dye molecules are mostly present in monomer form in spite of their high packaging density in nanoparticles. In order to verify that the observed absorption spectrum corresponds to ICG molecules encapsulated in nanoparticles and is not affected by free dye in the solution due to dye leaching, the following test was performed: The PEBBLE solution was filtrated by suction using a Whatman Anodisc filtration membrane with $0.02\text{-}\mu \mathrm{m}$ pore diameter, and absorption of the filtrate was measured. Only negligible dye leaching was observed, indicating that ICG molecules were present in monomer form in the nanoparticle. The enhanced stability of ICG in the nanoparticle is attributed to the interaction of the dye molecules with the ormosil matrix favoring monomer form by isolating individual dye molecules and preventing dye-dye interaction.

3.3.

Aqueous Stability Measurement

The enhanced stability of ICG nanoparticles is a major advantage over free ICG dye in clinical applications. Several environmental conditions are known to affect ICG and reduce its optical absorption. Free ICG dye is unstable in aqueous media and sensitive to pH.^{26, 37} This instability can limit its application as an optical contrast agent in medicine. Saxena ²⁶ have reported that ICG-doped nanoparticles based on PLGA (poly latic-glycolic acid) biodegradable polymer could improve stability of ICG under the conditions described earlier. To test the stability of ICG ormosil PEBBLEs in aqueous media, we have monitored their optical absorption in a water solution for 5 days. The absorption intensity of ICG ormosil PEBBLE decreased by less than 10% over 5 days, while free ICG dye showed above 70% decrease, indicating severe deactivation in water [Fig. 5 ]. This result shows that encapsulation of ICG in ormosil nanoparticles improves ICG stability in aqueous media by a large factor.

Fig. 5

Relative optical absorption of free ICG dye (diamonds) and ICG PEBBLE (triangles) measured over 5 days. Stabilization due to PEBBLE encapsulation is evident.

3.4.

Photosensitization

The photodynamic efficacy of ICG ormosil PEBBLEs was evaluated by monitoring the change of fluorescence intensity of a DPIBF chemical probe. DPIBF is a well-known singlet oxygen quencher with high sensitivity³⁸ and loses its fluorescence by reaction with singlet oxygen. As shown in Fig. 6 , for laser illumination at $800\mathrm{nm}$ over $30\mathrm{s}$ , fluorescence emission of DPIBF decreased with either free ICG dye or ICG ormosil PEBBLE present. In contrast, no such decrease was observed from controls with DPIBF but without free ICG dye or ICG PEBBLEs, indicating that the decrease of DPIBF fluorescence intensity was induced by ICG. The measurement confirms the potential of ICG as a photosensitizer for PDT. The irradiated light fluence of our experiment was about $1.36\mathrm{kJ}∕{\mathrm{cm}}^2$ , and the PDT dose can be roughly estimated to be $2\times {10}^{19}$ photons absorbed by ICG/mL, which is relatively higher than typical levels of clinical PDT. The system will be optimized for more reasonable conditions in clinical applications and the PDT efficacy of ICG ormosil PEBBLE will be investigated in more detail in our future work.

Fig. 6

(a) Fluorescence intensity of DPIBF with free ICG dye; (b) DPIBF with ICG-ormosil PEBBLEs; and (c) pure DPIBF. In each case, intensity was measured before (solid line) and after (dashed line) laser illumination at a wavelength of $800\mathrm{nm}$ for $30\mathrm{s}$ . Reduced fluorescence intensity indicates singlet oxygen generation.

3.5.

Cell Targeting Efficiency of Antibody-Conjugated PEBBLEs

ICG PEBBLEs were conjugated with HER-2 antibody to test targeting of prostate cancer cells (LNCaP cell line). Cells were incubated with conjugated particles and with bare particles (control) to compare specific and nonspecific binding of PEBBLEs to cells. In both cases, we used a high-resolution photoacoustic scan to detect and image a single cell layer. The scanned image of cells targeted by conjugated particles [Fig. 7 (top)] has a mean signal energy $22\mathrm{dB}$ higher than that of the control cells sample [Fig. 7(bottom)], for an illumination wavelength of $800\mathrm{nm}$ . This result indicates that antibody molecules were successfully conjugated to the PEBBLEs surfaces and maintained their targeting ability. The experiment also demonstrates the potential of photoacoustic imaging for single cell detection using ICG ormosil PEBBLE cell tagging. The photoacoustic signal from cells in the control sample was higher than the noise level, indicating nonspecific binding or insufficient washing to remove all PEBBLEs. The specificity of targeting moiety will be further investigated in our future work.

Fig. 7

Photoacoustic images of LNCaP cells incubated with antibody-conjugated PEBBLEs (top), and LNCaP cells incubated with unconjugated PEBBLEs (bottom). In both cases, cells were incubated $1\mathrm{h}$ and then washed twice. Signal gray scale is in dB.

3.6.

Photoacoustic Phantom Imaging

We have studied the detection limits of the photoacoustic imaging setup by imaging phantom objects with inclusions of PEBBLE nanoparticles in varying concentrations. A phantom with three cylindrical inclusions, each $1.5\mathrm{mm}$ in diameter, with PEBBLE concentrations of $1\mathrm{mg}∕\mathrm{mL}$ $({10}^{12}\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ , $0.1\mathrm{mg}∕\mathrm{mL}$ $({10}^{11}\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ , and $0.01\mathrm{mg}∕\mathrm{mL}$ $({10}^{10}\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ were constructed. A photoacoustic image was acquired using a laser pulse energy of $12\mathrm{mJ}$ , a beamwidth of $25\mathrm{mm}$ (light fluence is $2.6\mathrm{mJ}∕{\mathrm{cm}}^2$ ), and an illumination wavelength of $800\mathrm{nm}$ . The photoacoustic image of the phantom is shown in Fig. 8 . The result showed that the PA signal intensity is clearly correlated with the PEBBLE concentration in the inclusions. The instrumental detection limit of ICG PEBBLEs for the current conditions is estimated to be about $0.01\mathrm{mg}∕\mathrm{mL}$ $({10}^{10}\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ , which is equivalent to $0.36\mu \mathrm{M}$ of free ICG dye based on particle density calculation and the calibration curve in Sec. 3.2. At this concentration, the absorption coefficient is $0.1{\mathrm{cm}}^{-1}$ , which is lower than the absorption coefficient of typical tissue. For example, a recent study on the optical properties of the human prostate gland shows a mean absorption coefficient of $0.23{\mathrm{cm}}^{-1}$ with intersubject and intrasubject (different locations within the prostate gland) variations on the order of $0.2{\mathrm{cm}}^{-1}$ (Ref. 39). In this case, the minimal concentration required to provide sufficient contrast (estimated as 4 times the mean tissue background) is about ${10}^{11}\mathrm{PEBBLEs}∕{\mathrm{cm}}^3$ .

Fig. 8

Photoacoustic image of a gel phantom with three, $1.5\mathrm{mm}$ diameter, cylinders of different PEBBLE concentrations: $1\bullet {10}^{10}$ $(\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ (left), $1\bullet {10}^{11}$ $(\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ (middle), and $1\bullet {10}^{12}$ $(\mathrm{PEBBLEs}∕{\mathrm{cm}}^3)$ (right). The dynamic range of the gray scale is $40\mathrm{dB}$ .