2.1.

Preparing AuMBs and Targeted AuMBs for Flow Cytometry and Cell Attachment

AuNRs with peak absorption at 760 nm (${\mathrm{AuNR}}_{760}$) were modified with cystamine using a previously developed procedure.²⁹ The electrochemical synthesis method can provide high yield and stabilize the dispersibility of AuNRs in solution, which help the next steps of MB fabrication. Briefly, 10 mL of 0.4 M cystamine dihydrochloride and 10 mL of 16 mM nitric acid were added to the AuNR solution. The solution was aged for 30 min before being centrifuged twice with relative centrifugal force of 3420 rcf and 7695 rcf separately for 10 min each to remove excess cystamine. The cystamine modified AuNR usually shows a slight shift of absorption peak and can be restored by dispersing process with deionized water. Hence the absorption peak of cyst-AuNR is hardly changed. The concentration of cystamine-modified AuNRs was adjusted to 30 nM, which is derived from its molar extinction coefficient of the standard solution ($1.534\times {10}^9{\mathrm{M}}^{-1}{\mathrm{cm}}^{-1}$).

Next, the modified AuNRs and 20% human serum albumin (HSA; Octapharma, Vienna, Austria) were mixed at equal volumes of 0.33 mL for each and incubated for 24 h at 4°C. AuMBs were produced by using a digital sonicator (Model 102C, Branson, Danbury, CT) to sonicate the 5 mL diluted solution—which contained 1.32% HSA (w/v), 1.98 nM ${\mathrm{AuNR}}_{760}$, and perfluorocarbon (${\mathrm{C}}_3{\mathrm{F}}_8$) gas—for 2 min.

AuMBs of different sizes were then stratified by varying the centrifuge acceleration, with larger (AuMB1) and smaller (AuMB2) ones being collected after centrifugation with relative centrifugal force of 126 rcf for 2 min and 2754 rcf for 5 min, respectively.³⁰ The freshly prepared AuMBs were examined. The concentration of the solution and particle size distribution were measured by a size-analysis device (Multisizer 3, Beckman Coulter, San Jose, CA) using a 30-µm aperture. Some of the AuMBs were filtered with a 5-µm syringe filter (Sartorius, Goettingen, Germany) and then hardened using 2.5% glutaraldehyde. A transmission electron microscope (TEM, H7100, Hitachi, Tokyo, Japan) was used to investigate the morphology of the hardened AuMBs at an accelerating voltage of 100 kV.

To produce the targeted AuMBs, avidin was added prior to sonication for synthesis. The avidin-incorporated AuMBs was prepared by sonicating the solution, which contains 1.98 nM AuNR, 1.32% HSA, 0.04% (w/v) avidin. The synthesis procedure is similar to the description above for preparation of AuMBs except the addition of avidin. Next, the AuMBs were washed three times to remove the free avidin. The targeting ability was examined by incubating the samples with biotin-FITC (eBioscience, San Diego, CA) for 30 min. The biotin-FITC–avidin-conjugated AuMBs were washed three times to remove the unbound fluorophores. Flow cytometry (FACSAria II, Becton Dickinson, San Jose, CA) was used to examine the targeting efficacy of the produced AuMBs.

On the other hand, biotinylated antimouse VEGFR2 (eBioscience, San Diego, CA) was incubated with avidin-incorporated AuMBs for 30 min. Three washes were performed to remove unbound antibodies. These anti-VEGFR2-avidin-conjugated AuMBs were then used to assess their attachment to the specific receptor. Mouse endothelium bEnd.3 cells purchased from the American Tissue Culture Collection (ATCC, Manassa, VA) were grown in Dulbecco’s modified Eagle’s medium. With 10-min exposure to 3 mL of AuMBs at a concentration of $0.7\times {10}^7$ microbubbles per milliliter, the cells were washed by PBS to remove the unbound ones. For each group, five plates of cells were examined, and the microscopic pictures were taken randomly.

2.2.

Experimental Setup of the Imaging System and LIT

Figure 1 shows the experimental setup of the dual-modality imaging system and direct temperature measurements during LIT. In the imaging experiments, four groups of samples (MB, AuNR, AuMB1, and AuMB2) were separately embedded into polyacrylic acid (PAA) gel at a depth of around 12 mm. The PAA gel was prepared by adding 3 μL of 10% APS solution to per mL of acrylamide solution. The sound velocity in the phantom was $1595\mathrm{m}/\mathrm{s}$ at room temperature.³¹ The samples were irradiated by a Ti:sapphire laser (CF-125, SOLAR TII, Minsk, Belarus) that uses an Nd:YAG laser as the pump (LS-2137 U, LOTIS TII, Minsk, Belarus). The pulse duration of laser is around 15 ns, and the repetition frequency is 10 Hz. This tunable laser can provide excitation at wavelengths ranging from 700 to 980 nm and hence was suitable for our study. A custom 20-MHz fiber-integrated transducer was used for detection, with the PA or US signal being sent to a preamplifier (5077PR, Panametrics, Waltham, MA). The transducer was mounted on a precision translation stage, and B-scan images in $x-z$ plane were obtained by moving the driving motor (HR-8, Nanomotion, Yokneam, Israel) with a step size of 0.02 mm. These images were recorded by a data acquisition card (CompuScope 14200, Gage, Lachine, QC, Canada) in a PC at 100 MHz. All PA images were formed using a laser pulse fluence of around $8\mathrm{mJ}/{\mathrm{cm}}^2$.

Fig. 1

Experimental setups of measurements employing the PA/US agent: (a) dual-modality imaging system and CW laser for LIT, and (b) temperature measurement during LIT.

LIT was performed using an 808-nm CW diode laser (ONSET Electro-optics, Taipei, Taiwan) operating at 980 mW. Figure 1(a) shows the use of PA imaging for monitoring the treatment, while Fig. 1(b) shows how an infrared thermometer (Optris LS, Optris, Berlin, Germany) was used to record the temperature rise of the sample placed in the agarose gel for comparison. In the thermal treatment, the pressure ratio was acquired from PA intensity differences between two images obtained before and during heating and thus assessed the temperature. To avoid sound-velocity variations, the position of the object has to be corrected by comparing two images in advance.

3.1.

Particle Analysis and TEM

The concentration of AuMBs was measured for each production batch. Our protocols using 1.32% (w/v) HSA typically produce $3.5\sim 3.8\times {10}^9\mathrm{MBs}/\mathrm{mL}$. Figs. 2(a) and 2(b) show the size distributions as a function of the differential number and volume, respectively, for AuMB1, AuMB2, and MB. The AuMB1 group contained a greater proportion of MBs with sizes up to several micrometers, while most of the MBs in the AuMB2 group were smaller than 0.7 μm. The figure indicates the presence of only one peak in the differential volume at about 0.6 μm for AuMB2, while there were two peaks (i.e., at 2.5 and 5 μm) for AuMB1. The MBs exhibited the expected size diversity without stratification. There are differences in the volume, which leads to a smaller buoyancy for AuMB2 and hence a higher acceleration being needed during centrifugation. Figure 2(c) is a photograph of three samples of MBs, AuMB1, and AuMB2. TEM images are also shown in Figs. 2(d) and 2(e) for the incorporated nanorods and the hardened AuMB1, respectively. These images directly display the dimensions of the MBs and their contents. The AuNR tends to submerge into the bubbles from our TEM observation, which might be the most stable form in liquid. In addition, the microscopic images suggested that there were several tens of AuNRs per AuMBs, and some of them assembled with HSA, which made it difficult to assess the exact number. In Fig. 2(e), red arrows indicate single AuNRs, which can be seen more clearly in (d), and dashed arrows point to part of the assemblies. The enlarged views were captured to demonstrate locations of AuNR and the contents of AuNR-HSA assemblies. We assume that the dispersed nanorods and the AuNR-HSA assemblies were randomly located in the microbubble to evaluate the number of nanoparticles. For the assemblies, there were five (or more) AuNRs inside [inset of Fig. 2(e)]. For the dispersed nanorods, there were approximately 20 rods observed in the scope. As there were approximately $3.8\times {10}^9$ MBs per mL, the concentration of AuNRs in the solution may ranges from 0.1 to 0.5 nM (30 AuNRs to 80 AuNRs per microbubble).

Fig. 2

Size distributions of MBs (green), AuMB1 (red), and AuMB2 (blue) for the differential number (a) and the differential volume (b). (c) Photograph of MBs, AuMB1, and AuMB2 (from left to right). TEM images of AuNRs in AuMB (d) and AuMB1 (e) where red arrows indicate single AuNRs and dashed ones point to assemblies of AuNR and HSA.

3.2.

PA Characteristics

Figure 3 shows the optical absorption characteristics of dual-modality contrast agents, as assessed by the optical density [Fig. 3(a)] and PA spectroscopy [Figs. 3(b) and 3(c)]. The optical density was measured by double-integrating spheres and a spectrometer (USB4000, Ocean Optics, Largo, FL). Both absorption and scattering effects contribute to the attenuation, with scattering being especially exhibited by a foam-like liquid such as MBs but is less apparent for an AuNR solution. Figure 3(a) indicates that the peak of AuMB2 is consistent with that of ${\mathrm{AuNR}}_{760}$, which implies invariance of the optic absorption of the contrast agent. The increased attenuation at 550 nm is due to HSA. The diluted AuMB2 ($\sim 25\%$) has optical density 0.28 with half-centimeter path length at the wavelength 760 nm as shown. From macroscopic observation, the optic density of 25% AuMB2 had 0.2 higher (per cm) than MBs at the wavelength, which suggested that 0.52 nM AuNR were in the solution.

Fig. 3

(a) Optical density of AuMB2 (black) and ${\mathrm{AuNR}}_{760}$ (gray), (b) Comparison of normalized optical density (dark gray), SNR (light gray), and CNR (black dotted) in PA images of AuMB2, (c) PA spectroscopy of AuMB2.

PA spectroscopy involved varying the excitation wavelength from 720 to 940 nm. Figure 3(b) compares the optical density, signal-to-noise ratio (SNR) and the contrast-to-noise ratio (CNR) of AuMB2. The SNRs and CNRs were derived from the PA images shown in Fig. 3(c). The SNR compares the intensity in the ROI to the background, while CNRs are the SNRs normalized by the background standard deviation to eliminate variations in the laser output. The SNRs were 8.89, 13.55, 11.72, 10.50, 4.23, and 3.22 dB and the corresponding CNRs were 1.531, 2.447, 2.085, 1.932, 0.765, and 0.573 at wavelengths of 720, 760, 800, 860, 900, and 940 nm, respectively. Both the SNR and CNR resemble the values expected for optical density, except for less scattering at longer wavelengths. The dual-modality contrast agent had a peak absorption at 760 nm as expected.

3.3.

Dual-Modality Imaging

US and PA images of the contrast agents are shown in Fig. 4: Figs. 4(a)–4(d), are the US images, Figs. 4(e)–4(h), are the PA images, and Figs. 4(i)–4(l), are the dual-modality images. Four sample types (1% MB, 2 nM AuNR, 1% AuMB1, and 50% AuMB2; diluted by agarose) are shown in the US images, which illustrate that the MBs and AuMB1 exhibited considerably enhanced contrast at the same concentration, whereas AuNR produced little pulse-echo. The higher concentration needed for AuMB2 was due to the particles being much smaller, which implies that they contain less gas. In fact, 10% AuMB2 could provide sufficient enhancement in the experiments (not shown). PA images of four other samples (100% MB, 2 nM AuNR, 75% AuMB1, and 50% AuMB2) are also presented. Note that the concentrations for MB and AuMB1 were higher for effective PA generation. With an excitation wavelength of 760 nm there was virtually no PA signal from the MBs, whereas the AuNR group had high absorption and generated PA signal efficiently. Both AuMBs with encapsulation of ${\mathrm{AuNR}}_{760}$ exhibited strong PA signals. The contrast for AuMB1 was dominant in the upper ROI since the optical scattering impedes the light penetration, while AuMB2 provided more-uniform enhancement. Dual-modality images of the samples (100% MB, 2 nM AuNR, 75% AuMB1, and 50% AuMB2) are displayed as grayscale US images overlaid with PA images in thermal color. The higher concentration of larger-size bubbles leads to acoustic shadowing, which was evident in (i) 100% MB and (k) 75% AuMB1 samples. Because of acoustic shadowing, the inclusion in the US images became less visible than that in the low concentration images [i.e., (a) and (c)]. Hence, the smaller AuMB2 is more suitable as the dual-modality contrast agent than AuMB1, which has larger size and higher concentration.

Fig. 4

Images of contrast agents: (a to d) US images of 1% MB, 2 nm AuNR, 1% AuMB1, and 50% AuMB2; (e to h) PA images of 100% MB, 2 nm AuNR, 75% AuMB1, and 50% AuMB2; and (i to l) fused images of 100% MB, 2 nm AuNR, 75% AuMB1, and 50% AuMB2, where the US and PA data are in grayscale and thermal color, respectively.

3.4.

Targeted Efficacy and LIT

Surface bioconjugation extends the contrast agents to molecular probes. Flow cytometry was used to analyze the binding of the biotin-FITC to the MBs/AuMBs and to the avidin-incorporated MBs/AuMBs. Both of these sample types were mixed with biotin-FITC at an equivalent concentration and then washed as described in Sec. 2.1. Figure 5 shows example histograms of the fluorescence intensity (FI), with and without avidin conjugation, for MBs (as the control group) [Fig. 5(a)], AuMB1 [Fig. 5(b)] and AuMB2 [Fig. 5(c)]. An increased mean FI, as indicated by a shift of the distribution to the right, confirms more ligand binding of biotin-FITC on the MB surface. The increase in FI with FITC was greater for MBs and AuMBs conjugated with avidin. These results indicate that avidin can be incorporated into the shell of MBs and AuMBs during sonication and serve as a direct ligand for biotinylated antibodies. Quantitative analysis shows that the geometric mean FI was more than twofold higher for avidin-incorporated AuMB1 (124.09) than for nontargeted AuMB1 (50.94), and that the mean FI of avidin-incorporated AuMB2 was increased by more than 50% (from 235.01 to 371.80). This demonstrated usefulness of the AuMBs for PA/US dual-modality molecular imaging is expected to lead to therapeutic applications. [Fig. 5(d)] shows a picture of AuMBs attachment (red arrows) to mouse endothelium cells (bEnd.3) after a 10-min incubation and three washes. Cell contour can be observed in the images, and the number of cells and AuMB adhesion are confirmed. The AuMBs can be identified by the image intensity, size, and locations or directly seen in bright-field image. The bEND.3 cell is known for their VEGFR2 expression. With anti-VEGFR2 bound on avidin-incorporated AuMBs, the conjugation with cellular receptor was observed ($0.92\pm 0.19$ targeted AuMBs per cell), whereas MBs without antibodies can be easily washed out ($0.03\pm 0.01$ AuMBs per cell, $p<0.05$). The control group is shown in Fig. 5(e).

Fig. 5

Flow-cytometry FI histograms of MB, AuMB1, and AuMB2 with (black) and without (gray) avidin conjugation: (a) avidin-MB conjugated with biotin-FITC; (b) avidin-AuMB1 conjugated with biotin-FITC; and (c) avidin-AuMB2 conjugated with biotin-FITC. (d) The microscopic picture of anti-VEGFR2-avidin-incoroprated AuMBs and mouse endotheliual cells (bEnd.3). (e) The control group. Arrows point to MBs that attached to the cells. (Original magnification: 200×, bar: 35 μm.)

Finally, we present the results of LIT with the dual-modality contrast agents. Figure 6 shows the temperature rise measured by an infrared thermometer [Fig. 6(a)] and monitored by PA imaging [Figs. 6(b) and 6(c)]. The LIT employed a 980-mW CW laser to heat the samples. For the data shown in Fig. 6(a), an Eppendorf-shaped pit inside the agarose base was sequentially loaded with 1.5 mL of water, MB, 1.8 nM AuNRs, and AuMBs (without dilution), which can directly compare the abilities of the samples as a heat source. The AuNR concentration of 1.8 nM, as the control group, is closer to that of the solution before sonication ($\sim 1.92\mathrm{nM}$), thus it was used to evaluate the efficiency. As the CW laser was switched on for the first 660 sec and then turned off, a temperature rise was clearly evident for AuNR and AuMBs, while the temperature of the water and MBs hardly changed. Although the optical absorbance of AuMB2 suggests that there were fewer nanorods inside compared with the 1.8 nM AuNR sample, it can be observed that the slight scattering of AuMB2 helps accumulate the optical energy. Whereas AuNRs suspended in water had lower increased temperature than expectation may result from the thermal diffusivity by water and energy penetration. Besides, the temperature increase was more than twofold greater for AuMB2 than for AuMB1 under the same conditions. The strong scattering caused by large bubbles impedes penetration of light. On the other hand, the increased temperature in AuMB2 is mostly because the ratio of AuNRs is greater for its smaller size. Note that AuNR adsorbs with HSA before the sonication, and AuMBs are filled most with gas. Besides, in equal volumes of solutions there may be more small-sized AuMB2, which can also be observed in our particle analysis results. Hence compared with AuMB1, AuMB2 is more efficient in photothermal therapy due to more AuNRs included and more uniform illumination for heating.

Fig. 6

Temperature changes in contrast-agent samples during LIT. (a) Temperature rise measured by an infrared thermometer (magenta, water; green, MB; purple, AuNR; red, AuMB1; and blue, AuMB2). (b) (c) Thermal images obtained by PA imaging for a 75% AuMB1 phantom (b) and a 50% AuMB2 phantom (c).

Figure 6 also provides PA-monitored thermal images of 75% AuMB1 [Fig. 6(b)] and 50% AuMB2 [Fig. 6(c)] inside the PAA phantom and immersed in water. Different concentrations of embedded samples were used for image quality to compare the PA intensity variation. After the CW laser was turned on, the average intensity increased by 3.35 and 6.02 dB for AuMB1 and AuMB2, respectively. For an initial temperature of 22.8°C, the results imply that the temperature would rise to around 35°C for AuMB1 and to 49°C for AuMB2 in a water-filled environment. In this case, the temperature increase was higher for AuMB2 than for AuMB1 and slightly exceeded what we observed in our previous experiment. Two possible reasons for this discrepancy are (1) underestimation by the infrared thermometer because it only measures the surface temperature and (2) local overestimation in the PA images due to the smaller specific heat of the metallic nanoparticles. It is evident that AuMB2 can help increase the temperature more with fewer amounts and that AuMBs are capable heat sources. Moreover, it is noteworthy that the experimental thermal images indicate that the temperature increase was concentrated in the ROI (not in the surroundings), which supports that LIT can be enhanced by AuMBs.