1. Introduction

There is a dire need for a reliable, non-invasive imaging tool to detect, diagnose and characterize cancer – one of the leading causes of death in the United States. The early detection of cancer is absolutely necessary for effective therapeutic outcome and is a primary indicator for long term survival [1]. Moreover, demarcating tumor boundaries with high specificity is required to direct therapeutic interventions to tumor location and cause less or no damage to the surrounding healthy tissue. Here, we demonstrate the optoacoustic imaging technique [2] using molecularly targeted plasmonic gold nanoparticles [3] to detect highly proliferative cancerous cells.

Optoacoustic imaging is a technique relying on the illumination of tissue by nanosecond pulsed laser light and the subsequent emission of acoustic waves from tissue under thermal stress confinement. Previous studies have shown that gold nanoparticles can be used as contrast agents in optoacoustic imaging because of their strong optical absorption and scattering properties [4]. The gold nanoparticles can also be used as therapeutic agents in photothermal therapy [5]. The extinction spectra of the gold nanoparticles can be modified by varying their shape and size. The gold nanoparticles can be tuned to resonate in the NIR region [6] as light has higher penetration depth in the tissue at these wavelengths.

Epithelial cancer cells tend to overexpress epithelial growth factor receptor (EGFR) [7], causing the specifically targeted nanoparticles to cluster on the cell surface. This clustering leads to plasmon resonance coupling between nanoparticles and a red shift in the plasmon resonance frequency of the gold nanoparticle assembly [3, 8, 9]. The red-shift provides the opportunity to differentiate cancer cells from surrounding benign cells by using a combination of labeling with gold nanoparticles and multi-wavelength illumination.

In this work, tissue phantoms prepared with human keratinocyte cell line were used to demonstrate the application of molecular targeted gold nanoparticles in optoacoustic imaging. The 532 nm and 680 nm pulsed laser illuminations were chosen for the optoacoustic experiments due to overlap with the absorbance spectra of the isolated and aggregated gold nanoparticles [3, 8, 9]. Our study shows that optoacoustic imaging can differentiate between cancer cells labeled with the molecular targeted gold nanoparticles and the cells mixed with non-specific gold nanoparticles, even when the concentration of isolated non-targeted particles is much higher. These results provide a basis for development of highly sensitive approaches for detection of asymptomatic epithelial pre-cancers in vivo.

2. Materials and methods

2.1 Preparation of tissue phantoms

Specifically for the optoacoustic imaging experiments, three tissue phantoms consisting of human epithelial carcinoma cells (A431 keratinocyte) were used: (1) the control tissue phantom with no gold nanoparticles; (2) the targeted tissue phantom labeled with EGFR targeted gold nanoparticles; and (3) the non-targeted tissue phantom with nanoparticles coated only with a polyethylene glycol-thiol (mPEG-SH) layer which has no molecular specificity.

Gold particles (50 nm diameter) were synthesized via citrate reduction of chloroauric acid (HAuCl₄) under reflux. Anti-EGFR monoclonal antibody (clone 225, Sigma) was conjugated with gold nanoparticles using a protocol described elsewhere [10]. The protocol resulted in nanoparticles functionalized with antibodies via a bifunctional PEG linker, and any remaining bare gold was then passivated with mPEG-SH. PEGylated particles without antibodies were obtained by mixing the colloid suspension with mPEG-SH in deionized water.

The A431 cells were purchased from American Type Culture Collection (ATCC) and cultured in DMEM supplemented with 5% fetal bovine serum at 37°C in a 5% CO₂ environment. Cells were harvested and resuspended in DMEM at a concentration of 2∙10⁶ cells/mL and divided into three 450 μl aliquots. The aliquot for the targeted sample was mixed with an equal volume of the anti-EGFR gold bioconjugate solution and allowed to interact for 45 minutes at room temperature. The other two aliquots were not exposed to the nanoparticles. The three cell suspensions were then spun down at 200 g and resuspended separately using 250 μL aliquots of a buffered collagen solution (2.1 mg/mL, pH 7.4), resulting in a final cell concentration of 3.6∙10⁶ cells/ml. To determine the amount of nanoparticles attached to cells in the targeted sample, the optical density of the solution of gold bioconjugates at the concentration used for labeling was compared to the optical density of the supernatant obtained after the labeled cells were spun down. The UV-Vis measurements showed ca. 260,000 particles per cell corresponding to approximately 25% of the total number of receptors per cell [11]. The targeted sample contained approximately 1∙10¹² gold nanoparticles/mL. The buffered collagen solution for the non-targeted sample contained approximately 4∙10¹² PEGylated nanoparticles/mL. The control tissue sample had no gold nanoparticles. The cell/collagen solutions (200 μL) were pipetted into separate stacked spacers (0.5 mm silicone isolators) in Petri dishes for optical characterization and optoacoustic imaging. The cell/collagen solution in the Petri dish was allowed to gel in a 37°C incubator for 1 hour. This procedure resulted in phantoms with randomly distributed cells in a three-dimensional collagen matrix [12]. The 3D arrangement of cells gives an opportunity to study imaging approaches having depth resolution. The phantoms were then covered with 50 μL of media and stored in an incubator for several hours prior to imaging.

A blocking assay was done to verify the molecular specificity of the anti-EGFR particles. A431 cells were harvested and exposed to 1 mg/mL C225 antibody or nonspecific (anti-goat IgG) antibody in PBS for 15 minutes. Furthermore, separate aliquots of A431 cells (ca. 10⁶ EGFR receptors per cell [11]) and MDA-MB-435 cells (which do not express EGFR) were exposed only to PBS for use as positive and negative controls. Targeted particles were added to all cell suspensions in a 1:1 volume ratio and mixed for 20 minutes. The cells were then spun down and the optical density (O.D.) of the supernatant obtained from each sample was compared to the O.D. of the labeling solution to determine the particle uptake.

2.2 Optical imaging of tissue phantoms

The tissue phantoms were characterized using a 75W Xenon light source and Leica DM 6000 upright microscope in epi-illuminated darkfield mode. Images were collected through a 20x, 0.5 NA darkfield objective and detected using a Q-Imaging Retiga EXi ultra-sensitive 12-bit CCD camera.

The absorbance spectra were collected with a PARISS hyperspectral imaging device (Lightform, Inc.) in transmitted brightfield mode and a halogen light source. The hyperspectral device was coupled to the Leica microscope and was used to measure the absorbance spectra at each pixel in the image. A single vertical section of the sample image was projected onto a prism through a 25 μm slit. The prism spectrally dispersed the one-dimensional image onto a two-dimensional Q-imaging Retiga EXi CCD detector. The sample was translocated laterally via a piezoelectric stage and the imaging process was repeated to construct the three-dimensional hyperspectral data cube. The spatial resolution of hyperspectral image was 1.25 μm and the spectral resolution was 1 nm. A blank slide containing 1x PBS was used to acquire the illumination lamp spectra. Transmitted brightfield spectral data cubes were then acquired from 20 μm by 300 μm areas of each sample and normalized to the illumination lamp spectra.

2.3 Ultrasonic and optoacoustic imaging of tissue phantoms

A block diagram of the experimental setup for ultrasound and optoacoustic imaging is shown in Fig. 1. A microprocessor unit with a custom built LabVIEW application controlled all modules of the imaging system including the ultrasound pulser/receiver, pulsed lasers, data acquisition unit, and all motion axes needed for imaging via 3-D mechanical scanning. A 48 MHz single element focused ultrasound transducer (focal depth = 5.5 mm, f# = 1.4) was used to obtain both ultrasonic and optoacoustic images of the tissue phantoms. The tissue phantom was attached to a 3-D positioning stage and placed in the focal region of the transducer. The Petri dish was filled with 1x PBS solution to maintain the appropriate pH in the medium surrounding the tissue phantoms.

In optoacoustic imaging, either a Q-switched Nd:YAG laser (532 nm wavelength, 5 ns pulses, 20 Hz pulse repetition frequency) or a tunable OPO laser operating at 680 nm wavelength and capable of producing 7 ns pulses at 10 Hz pulse repetition frequency were used. The ultrasound and optoacoustic images were obtained by mechanically scanning the tissue samples over the desired region with 12 μm lateral steps. At each step, the pulsed laser light irradiated the sample and the optoacoustic response from the sample was captured using 8-bit, 500 MHz digitizer. The signal recording was initiated by the trigger signal from the laser source. The same trigger signal, delayed by several microseconds, was sent to the pulser/receiver to initiate the pulse-echo ultrasound imaging. An acquired A-line, therefore, contained the optoacoustic signal followed by the conventional ultrasound signal. During the offline processing, digital bandpass (20–70 MHz) filtering was employed to reduce noise in the signals. Finally, the signals were processed to produce spatially co-registered 2-D optoacoustic and ultrasound images.

 

Fig. 1. Block diagram of the combined ultrasound and optoacoustic imaging system.

Download Full Size | PDF

3. Results

Blocking the cells with an excess of C225 antibody resulted in a 26x decrease in labeling efficiency as compared to the positive control while the cells exposed to nonspecific IgG showed no decrease in labeling efficiency. The MDA-MB-435 cells do not express EGFR and showed no particle uptake. These results were also confirmed using darkfield optical imaging and indicate molecular specific binding of the bioconjugated gold nanoparticles to membrane-bound EGF receptors.

The absorbance spectra of the control, targeted and non-targeted tissue phantoms are shown in Fig. 2. The control phantom has low absorbance in wavelength range of 450–800 nm. The non-targeted phantom has an absorbance peak at 520 nm, which is in excellent agreement with the absorbance spectrum of a suspension of isolated gold nanoparticles. The targeted phantom has the peak red-shifted and broadened due to EGFR-mediated aggregation of gold nanoparticles [3, 8, 9]. The absorbance spectra were used as a guideline to gauge the difference in the optical properties of the targeted and non-targeted tissue phantoms.

The darkfield images of the control, targeted and non-targeted tissue phantoms are presented in Figs. 3(a), 3(b) and 3(c). The control phantom in Fig. 3(a) does not contain any gold nanoparticles and hence the cells appear bluish white due to their intrinsic light scattering properties. The targeted phantom (Fig. 3(b)) shows orange colored cells caused by the plasmon-resonance scattering of anti-EGFR conjugated gold nanoparticles which interact with EGFR molecules on the cytoplasmic membrane of A431 cells. The non-targeted tissue phantom (Fig. 3(c)) has gold particles in suspension surrounding the cells. These isolated gold particles are associated with the greenish haze in the background surrounding the unlabeled A431 cells which appear bluish in the image.

 

Fig. 2. Absorbance spectra of control, targeted and non-targeted tissue samples normalized to the illumination lamp spectrum.

Download Full Size | PDF

The ultrasonic images of the three tissue phantoms are presented in Figs. 3(d), 3(e) and 3(f). As expected, these images do not reveal any information regarding the isolated or clustered state of the gold nanoparticles due to insufficient acoustic contrast. The optoacoustic images of the tissue phantoms shown in Figs. 3(g), 3(h) and 3(i) were obtained with 532 nm laser irradiation and images presented in Figs. 3(j), 3(k) and 3(l) were obtained using 680 nm laser irradiation. All optoacoustic images are displayed using the same dynamic range. The optoacoustic images of the control phantom do not show any signals at both 532 nm (Fig. 3(g)) and 680 nm (Fig. 3(j)), indicating that the tissue absorbs less light at these wavelengths as compared to the other two tissue phantoms. The faint signal in the lower region of the images in Figs. 3(g) and 3(j) was due to absorption of the laser by the plastic bottom of the Petri dish holding the phantom.

 

Fig. 3. Darkfield, ultrasound and optoacoustic images (λ = 532 nm and 680 nm) of control, targeted and non-targeted tissue phantoms. The darkfield images measure 440 μm by 340 μm field of view. The ultrasound and optoacoustic images measure 2 mm by 1.67 mm.

Download Full Size | PDF

At 532 nm laser irradiation, the optoacoustic image of the non-targeted phantom (Fig. 3(i)) indicates higher optical absorbance than the targeted phantom (Fig. 3(h)). Due to the high absorbance of the non-targeted phantom at 532 nm, the laser fluence decreases exponentially with depth. Depth dependent compensation was applied to optoacoustic signals in Fig. 3(i) to compensate for the signal loss due to decrease in the laser light fluence. At 680 nm illumination, very little optoacoustic response was obtained in the non-targeted phantom (Fig. 3(l)), unlike the targeted tissue phantom that produced signal from the entire thickness of approximately 1 mm (Fig. 3(k)). Hence in congruence with the hyperspectral analysis (Fig. 2), a relatively low overall bulk extinction coefficient was observed in the non-targeted phantom as compared to targeted phantom at 680 nm. Thus, specific targeting of gold nanoparticles to EGFR molecules that are overexpressed in certain types of cancer results in significant increase in the optoacoustic signal in the red optical region. We attribute the increase in optoacoustic signal to EGFR mediated assembly of gold nanoparticles on the cytoplasmic membrane of the cancerous cells; this leads to plasmon resonance coupling between adjacent gold particles and changes in their absorbance spectra which are shown in Fig. 2.

4. Discussion and conclusions

The results of our study indicate that optoacoustic imaging at 532 nm could identify the distribution of the gold nanoparticles in tissue. Furthermore, a large difference between EGFR targeted and non-targeted (isolated) gold nanoparticles was observed in the optoacoustic images at 680 nm laser irradiation. The contrast mechanism in optoacoustic imaging is based upon the difference in optical properties of the tissue constituents, and hence optoacoustic imaging could be used to differentiate between cancerous and normal cells using molecularly targeted gold nanoparticle contrast agents, as was previously demonstrated with purely optical techniques [3, 8, 9]. Also, compared to optical imaging, the penetration depth of optoacoustic imaging can be on the order of centimeters if near-infrared laser light is used. Moreover, optoacoustic imaging is a non-ionizing method and it does not have the safety concerns associated with some other radiological imaging modalities such as CT or PET. The combined ultrasound and optoacoustic imaging has many applications ranging from cancer detection and therapy monitoring to tissue engineering [2, 13, 14].

Gold nanoparticles have excellent biocompatibility [15] and the conjugation protocols to attach proteins to gold nanoparticles are also well developed [16–18]. The optoacoustic imaging with gold nanoparticles demonstrated in this report can be potentially extended to a combined diagnostic imaging and therapy approach. Photothermal therapy with plasmonic nanoparticles was previously demonstrated for both pulsed [5] and continuous light sources [19]. Based on the information obtained with optoacoustic imaging, pulsed or continuous wave photothermal therapy could be performed to induce localized tumor necrosis, potentially even using the same light source as was used in optoacoustic imaging. Moreover, ultrasound based strain imaging can be performed together with optoacoustic imaging and phototherapy at no additional cost to monitor tumor necrosis over time [14].

In conclusion, the optoacoustic imaging technique could detect tumors by selectively targeting cancerous cells that overexpress EGFR. In our studies, optoacoustic imaging was performed at 532 nm and 680 nm on three tissue phantoms prepared using A431 skin cancer cells targeted with anti-EGFR gold bioconjugates. The results of our study demonstrate that using molecular targeted gold nanoparticles and optoacoustic imaging, specific molecular differentiation and highly sensitive and selective detection of cancer could be achieved. Further studies are required to evaluate this molecular specific imaging technique in vivo and its potential in combination with phototherapy.