Multimodality Molecular Imaging-Guided Tumor Border Delineation and Photothermal Therapy Analysis Based on Graphene Oxide-Conjugated Gold Nanoparticles Chelated with Gd

Tumor cell complete extinction is a crucial measure to evaluate antitumor efficacy. The difficulties in defining tumor margins and finding satellite metastases are the reason for tumor recurrence. A synergistic method based on multimodality molecular imaging needs to be developed so as to achieve the complete extinction of the tumor cells. In this study, graphene oxide conjugated with gold nanostars and chelated with Gd through 1,4,7,10-tetraazacyclododecane-N,N′,N,N′-tetraacetic acid (DOTA) (GO-AuNS-DOTA-Gd) were prepared to target HCC-LM3-fLuc cells and used for therapy. For subcutaneous tumor, multimodality molecular imaging including photoacoustic imaging (PAI) and magnetic resonance imaging (MRI) and the related processing techniques were used to monitor the pharmacokinetics process of GO-AuNS-DOTA-Gd in order to determine the optimal time for treatment. For orthotopic tumor, MRI was used to delineate the tumor location and margin in vivo before treatment. Then handheld photoacoustic imaging system was used to determine the tumor location during the surgery and guided the photothermal therapy. The experiment result based on orthotopic tumor demonstrated that this synergistic method could effectively reduce tumor residual and satellite metastases by 85.71% compared with the routine photothermal method without handheld PAI guidance. These results indicate that this multimodality molecular imaging-guided photothermal therapy method is promising with a good prospect in clinical application.


Introduction
Hepatocellular carcinoma (HCC) is the second most common cancer for death rate in the world and claims more than 700000 lives per year worldwide [1]. With advances in surgical techniques (resection, transplantation, and thermal ablation) and imaging techniques, 30% of patients are diagnosed and cured at early stage [2]. However, the management of 70% of patients with more-advanced stages remains challenging [3]. Besides surgical method, photothermal therapy has garnered the attention because of its tumor destruction ability [4]. Photothermal therapy utilizes photosensitizers to convert light to heat, which could destroy the tumor tissues. Cancer cells take up the photosensitizers, resulting in cell death because of photoablation. To avoid nonspecific heating of healthy cells, photosensitizers must be selectively taken up by cancer cells.
Therefore, designing a high absorption photosensitizer is essential to generate as much heat as possible from optical energy for photothermal therapy. In recent years, graphene 2 Contrast Media & Molecular Imaging oxide (GO) garnered the attention as a promising material for biomedical applications as a photoabsorber for photothermal therapy and photoacoustic imaging (PAI) due to its superior optical absorption in the NIR and high photothermal conversion efficiency and nanocarrier feature through the -stacking between drugs and GO nanosheets [5][6][7]. Functionalized GO is widely used as an efficient theranostic agent due to the existence of sp2 domain with hydroxyl, epoxy, and carboxyl groups [8]. We previously reported the ability of PLA microcapsules containing GO and Au nanoparticles as a theranostic system for multimodal imaging-guided cancer therapy [9,10]. However, the micron scale agent cannot easily accumulate at the tumor site through the blood circulation system.
Additionally, the tumor margin must be delineated more accurately so that the treatment can be carried out more precisely to avoid damaging normal tissues. Simultaneously, the photothermal materials should accumulate to its peak value at the tumor site so that the irradiation could be initiated. Therefore, developing an effective imaging technique and analysis method is imperative to detect the tumor margin and monitor the pharmacokinetics process. Molecular imaging, especially optical molecular imaging, has been recently used to monitor the antitumor efficacy and calculate the location of orthotopic tumors [11][12][13]. However, single-modality imaging technique has its limitations in tumor margin delineation and precise photothermal therapy. Multimodality molecular imaging including MRI, PAI, and optical imaging combined with image processing technologies could solve the above problems.
In this study, GO-AuNS-DOTA nanoparticles chelated with Gd (GO-AuNS-DOTA-Gd) were prepared and used in photothermal therapy against HCC. MRI was used to delineate the tumor margin in vivo before treatment; PAI was used to monitor the pharmacokinetics process of GO-AuNS-DOTA-Gd and real-time detect the tumor location during treatment, and bioluminescent imaging was used to evaluate the antitumor therapy. Compared with the current photothermal methods, this synergistic method based on this synergistic method could improve treatment efficacy by 36.36% evaluated by BLI for subcutaneous tumor. For orthotopic tumor, this method could effectively reduce tumor residual and satellite metastases by 85.71% compared with the routine photothermal method without handheld PAI guidance.

Preparation of GO-AuNS-DOTA Nanoparticles Chelated
with Gd. The preparation of GO-AuNS-DOTA nanoparticles chelated with Gd was carried out using the method reported in [14,15]. Briefly, 5 g/mL graphene oxide was mixed with 1200 L of 0.1 M HEPES and then 2.4 L of HAuCl 4 (0.1 M) was added to the mixture. Subsequently, GO-AuNS hybrids were centrifuged at 5000 rpm for 10 min to remove HEPES and any free gold nanostars in the supernatant. The precipitate was then resuspended in water.
To prepare the GO-AuNS-DOTA nanoparticles, GSH (5.0 mg) was added to the above solution with stirring overnight to form the nanoconstruct (GO-AuNS-GSH). After 12 h, the solution was centrifuged at 5000 rpm for 10 min. The precipitate was retained and DI water (1.0 mL) was added to redissolve the precipitate. Meanwhile, DOTA-NHS was dissolved into DI water under stirring at 55 ∘ C, followed by addition of GdCl 3 ⋅6H 2 O in 5 mL water. The molar ratio of Gd to DOTA-NHS was 3 : 1; after chelating for 30 min, the GO-AuS-GSH was added to this solution and stirred for 24 h at room temperature to form the GO-AuNS-DOTA nanoparticles chelated with Gd (GO-AuNS-DOTA-Gd).

Characterization of GO-AuNS-DOTA Nanoparticles.
The size distribution and zeta potential of GO-AuNS-DOTA nanoparticles were measured using a ZetaPlus dynamic light scattering instrument (Brookhaven Instruments) after dispersion of the nanoparticles at an appropriate concentration (200 g/mL). Six replicate samples of each type of vesicles were analyzed at a scatter angle of = 90 ∘ and a wavelength of 660 nm at room temperature.
The morphology of GO-AuNS-DOTA nanoparticles was observed by transmission electron microscopy (TEM, Tec-nai6 G2 at 200 kV). Copper grids with carbon coating were used for sample preparation. First, nanoparticles were dispersed in deionized (DI) water at 0.2 mg/mL concentration. Then, 10 L of nanoparticle suspension was placed on the grid and absorbed for 30 seconds. Finally, the grid was washed with DI water and dried at room temperature in a draught cupboard.
The UV-Vis-NIR absorption spectra of the GO-AuNS-DOTA nanoparticles were acquired with a UV-Vis spectrometer (Varian 4000, USA). The amount of Gd and Au in the nanoparticles was quantified by the inductively coupled plasma (ICP) analysis.

In Vitro Photothermal Heating Experiments.
To examine the temperature elevation efficiency, different concentrations of GO-AuNS-DOTA nanoparticles dispersed in DI water were under the irradiation of an 808 nm laser with an output power of 2 W. The temperature of all samples was monitored and periodically confirmed by a digital thermometer with a thermocouple probe every 10 seconds. DI water was irradiated as a control. The heating curve was determined by plotting the measured temperatures.
To confirm the photothermal stability of GO-AuNS-DOTA nanoparticles, solutions of 50 g/mL GO-AuNS-DOTA nanoparticles were irradiated with an 808 nm NIR laser for 10 min (LASER ON), followed by natural cooling to room temperature without NIR laser irradiation for 30 min (LASER OFF). This cycle was repeated five times and the Contrast Media & Molecular Imaging 3 heating curve of the five repeated cycles was plotted to characterize the photothermal stability of nanoparticles.

Photothermal
Effect of the Nanoparticles against HCC-LM3-fFluc Cells. Photothermal cytotoxicity of GO-AuNS-DOTA was evaluated on HCC-LM3-fFluc cells. For qualitative analysis, HCC-LM3-fFluc cells were seeded onto 6well plate at a density of 2 × 10 5 cells per well and incubated at 37 ∘ C in a humidified atmosphere containing 5% CO 2 overnight. After incubation with different concentrations of GO-AuNS-DOTA nanoparticles for 4 h, the medium containing nanoparticles was replaced with fresh medium, and the cells were exposed to an 808 nm NIR laser with an output power of 2 W for 5 min. After incubation for 2 h, the cells were stained with propidium iodide (PI) to evaluate the photohyperthermic effect of the nanoparticles on cancer cells and the fluorescence images were collected with an inverted fluorescence microscope (Leica, Germany) equipped with a cooled CCD camera. For quantitative evaluation of the phototoxicity, HCC-LM3-fFLuc (1 × 10 4 cells per well) were incubated in 96-well plate and incubated overnight to allow the cells to attach to the surface of the wells. After incubation with GO-AuNS-DOTA for 4 h, the medium containing nanoparticles was replaced with fresh medium and the cells were irradiated with a diode NIR laser centered at 808 nm at an output power of 6 W/cm 2 for different times. After incubation for 4 h, cell viability was determined by Trypan Blue assay in quadruplicate.

Tumor Models.
Balb/c mice were obtained from the Department of Experimental Animals, Peking University Health Science Center, and experiments were performed under protocols approved by the Institutional Animal Care and Use Committee at Peking University. The HCC-LM3-fFLuc subcutaneous tumor-bearing mice were established by injecting 6 × 10 6 cells into the right flanks and used for photothermal therapy when the tumor volume reached about 160 mm 3 . After image acquisition, tumors and outline in mice were segmented according to the gray value difference and the segmented regions of interest (ROI) were visualized using 3D slicer [16]. The maximal radium was deemed to half of maximum value of the distance between two random voxels at the tumor surface, as shown in the following formula:

In Vivo Target Study of GO-AuNS-DOTA-Gd and
where ( , , ), ( , , ) are two random points on the surface of the segmented tumor. The average MRI gray value of the tumor in 18 slices was calculated using Philips DICOM Viewer R3.0 SP3 [17]. The tumor volume was calculated according to the following formula: where is the number of voxels of the tumor ROI in th slice, V 0 is the volume of every voxel, and = 18 is the number of slices. Meanwhile, the average gray value contrast ratio was used to demonstrate the discrimination ability of the probe for tumor and normal tissue, which was calculated using the following formula: where AG is the average gray value of the tumor ROI in the th slice and AR is the average gray value of region around the tumor in the th slice. Here, the region around the tumor was acquired through the following two steps. = 18 is the number of slices. First, the tumor ROI was carried out morphological dilation until the area was twice the tumor ROI, except the background. Then, the total area minus the area of tumor ROI was considered the region around tumor.

4
Contrast Media & Molecular Imaging

In Vivo Positron Emission Computer Tomography (PET)
Using 18 -FDG. PET scans were acquired with a micro-PET GENISYS 4 scanner system (Sofie Biosciences, USA) after the mice received a tail-vein injection of 18 F-FDG (100 Ci). The mouse was scanned 30 minutes after injection with a 10 min static PET scan. During the scan, the mouse was anesthetized using 2% isoflurane. The image was reconstructed using a 3dimensional maximum a priori algorithm and into 9696208 matrices with an isometric voxel size of 0.46 mm 3 .

Photothermal Therapy.
Female Balb/c mice bearing HCC-LM3-fLuc tumors were divided into 8 groups (7 mice per group) when the tumor volume reached about 150 mm 3 . Three groups were treated with 80 L of GO-AuNS (GO: 2 mg/mL, Au: 1.31 mg/mL) via tail veil injection and laser treatment was carried out at 2, 5, and 12 h after injection (named (D2) group, (D5) group, and (D12) group), respectively. One group was only treated with 80 L of GO-AuNS via tail vein but did not receive laser treatment. Two groups were treated with 80 L of GO-AuNS via intratumor injection and one received laser treatment, while the other group did not. Two groups were treated with 80 L of 0.9% saline via tailvein injection and one received laser treatment, while the other group did not. For the laser treatment groups, the tumors were irradiated by an optical fiber coupled with an 808 nm NIR laser (Hi-Tech Optoelectronics Co., Ltd. Beijing, China) at the power density of 0.2 W/cm 2 for 5 min. The temperature around the tumor after irradiation was collected by using an infrared camera.

Orthotopic Tumor Models and Multimodality Molecular
Imaging-Guided Cancer Treatment. The 21 orthotopic tumor models were established by injecting 2 × 10 6 HCC-LM3-fLuc tumor cells into the liver of each BALB/c nude mouse. The mice were randomly separated into 3 groups each consisting of 7 mice. One group (control) was injected i.v. with PBS. The other two groups (treatment) were injected i.v. with GO-AuNS-Gd on the 10th day. In vivo MRI was acquired at preinjection, 1, 2, 3 and 12 h after the tail injection. Photothermal therapy was conducted on all mice of the three groups. The process included the following two steps. First, the abdomen of mice was opened and the liver was exposed to the air. Then handheld photoacoustic imaging scanning was conducted on mice of one treatment group and, according to the PAI which indicated the tumor location, photothermal therapy was conducted using an optical fiber coupled 808 nm NIR laser at the power density of 0.2 W/cm 2 for 6 min.
where BI ex is the average bioluminescent intensity of mice tumors in experiment group. BI control is the average bioluminescent intensity of mice tumors in control group.

Drug Toxicity Assessment and Biodistribution Study.
For subcutaneous tumor experiments, to assess the toxicity of GO-AuNS-DOTA-Gd and photothermal efficacy, six organs including the heart, liver, spleen, lung, kidney, and tumor were harvested after euthanasia on the 21st day after tail injection and stained with hematoxylin and eosin (H&E).
Mouse weight was also acquired on days 0, 2, 5, 8, 11, 15, and 20 after treatment and plotted as a weight graph in OriginPro for each group. For orthotopic tumor experiments, tumors from PBS group and probe group after treatment were harvested and stained with H&E. For biodistribution assay, HCC-LM3-fLuc subcutaneous tumor-bearing mice were sacrificed at 2 h, 5 h, and 12 h after i.v. injection of GO-AuNS-DOTA-Gd (200 L per mice, = 3). Major organs including heart, liver, spleen, lung, kidney, intestines, brain, and tumor were collected, weighed, and then solubilized by aqua regia. The concentrations of Au and Gd in those tissue lysate samples were measured by inductively coupled plasma mass spectrometry (ICP). The levels of GO-AuNS-DOTA-Gd in various organs are presented as the percentage of injected dose per gram of tissue (ID g/g). In addition, the concentrations of Au and Gd in liver tumor and normal liver including above tissues based on the orthotopic tumor model were measured by ICP. The mice were sacrificed at 1 h and 12 h after i.v. injection of GO-AuNS-DOTA-Gd (200 L per mice, = 3).
2.11. Statistical Analysis. Quantitative data are expressed as mean ± SD. Results of two sets were compared using oneway analysis of variance (ANOVA) and a Student -test, and values < 0.05 were considered statistically significant difference.

In Vitro Photothermal Effects of GO-AuNS-DOTA Nanoparticles.
The temperature of nanoparticle solutions showed a rapid increase within 4 min of laser irradiation, and the temperature elevation rose as the nanoparticle concentration increased (Figure 3(a)). In contrast, no apparent temperature change was detected for PBS. As shown in Figure 2(b), the temperature increased 27 ∘ C after the first LASER ON and, in the next four cycles, no significant variation of temperature elevation was observed. The result of in vitro photothermal experiment indicated that cancer cells in three control groups, including GO-AuNS-DOTA group, laser irradiation group, and group without any treatments, maintained their healthy state (Figure 3(c)). In contrast, cells from experimental groups with both GO-AuNS-DOTA and laser irradiation treatment showed a bright red fluorescence in the laser irradiation region, indicating that GO-AuNS-DOTA could mediate the photothermal destruction of HCC-LM3-fFluc cells.
Cell viability using MTT method (Figure 3(d)) demonstrated GO-AuNS-DOTA nanoparticles without laser irradiation exhibited little cytotoxicity on HCC-LM3-fFluc cells. In contrast, GO-AuNS-DOTA nanoparticles followed by laser irradiation (808 nm and 2 W/cm 2 ) induced cell cytotoxicity in a dose-and time-dependent fashion.

In Vivo Target Study of GO-AuNS-DOTA-Gd Based on Subcutaneous Tumor Model and Multimodality Molecular
Imaging. PAI and MRI were performed to investigate targeting ability of GO-AuNS-DOTA-Gd in tumor-bearing mice. Figures 4(a) and 4(c) depict drug concentration within tumor as a function of time before and after tail-vein injection of probe and quantitative analysis of the corresponding imaging using PAI. The signals in cross-sectional images of tumor region of interest (tumor ROI) started to accumulate from 2 h and reached a peak at 5 h with a nearly 2.4-fold enhancement compared with that of preinjection until 24 h in close proximity to that of preinjection.   (Figures 4(f) and 4(g)). In contrast with 18 F-FDG PET image (Figure 4(h)), the MRI image after probe injection allowed us to distinguish tumor margin from the normal tissue. To quantitatively analyze the probe's ability of tumors targeting and enhancement, the tumor was segmented and visualized using 3D slicer (Figure 4(e)). The calculated tumor volume and maximal radium on 5 h exceeded that of preinjection and on 2 h and then the value declined slowly over 5 h (Table 1). Gray contrast ratio of tumor versus peripheral tissue on 5 h was higher than that on the other three time points. The results indicated that probe concentration in the tumor on 5 h reached a peak and presented the best ability to distinguish the tumor from normal tissues. tumor at 5th h after probe injection via tail vein achieved the same antitumor effect as laser irradiation treatment on tumor after intratumor injection ( Figure 5(d)). Compared with the current photothermal methods (for (D12) group, if the therapy was initiated on 12th h postinjection (7)(8)), this synergistic method based on multimodality molecular imaging (for (D5) group) could improve treatment efficacy by 36.36% on 2nd day evaluated by BLI for subcutaneous tumor. Tumor growth for groups (D2) and (D12) was inhibited after treatment until day 4 and then tumor grew slowly close to that of the other groups.

In Vivo Antitumor Efficacy Evaluation Based on Image
To evaluate the toxicity of the probe, body weight of mice was monitored throughout the antitumor study. None of mouse weight in the eight groups declined more than 20% of its original weight ( Figure 5(e)). H&E staining of the heart, liver, spleen, lung, and kidney of the mice showed no distinct damage for all drug treatment groups ( Figure 5

Discussion
Graphene has been used for photothermal therapy and has shown promise in many in vitro and in vivo studies [18,19]. However, photothermal therapy based on graphene and its related detailed imaging analysis, including tumor volume and gray contrast ratio, have been rarely reported [20]. In this study, GO-AuNS-DOTA-Gd was prepared and its physicochemical properties indicated that the GO-AuNS-DOTA could act as an efficient material for photothermal therapy and be used for precisely delineating the tumor margin from normal tissues. GO is believed to have a role in promoting  Au nanostar photothermal transduction efficacy in certain studies which deserved further study [21]. Determining tumor boundary is important and can influence the tumor surgery and treatment [22]. In contrast with 18 F-FDG PET imaging, MRI imaging based on GO-AuNS-DOTA-Gd can accurately distinguish the tumor from normal tissues. Furthermore, MRI and PAI analysis demonstrated that the concentration of GO-AuNS-DOTA-Gd reached its peak in tumor regions at 5th h for subcutaneous tumor and 1st h for orthotopic tumor after tail-vein injection, which was determined as the optimal implementing time of photothermal therapy. For orthotopic tumor, handheld PAI guided photothermal therapy can effectively reduce the tumor residual and ensured the killing of tumor cells.   Molecular imaging has been widely used in tumor research [23][24][25]. However, reports on multimodality molecular imaging-guided tumor margin delineation and tumor treatment with deep tumor feature analysis are scarce. Imaging analysis has gradually showed its importance not only for subcutaneous tumor but also for orthotopic tumor and some groups developed image processing methods to resolve problems encountered in tumor surgery and cancer research. However, there are still some technological difficulties to be resolve, including quick acquisition of massive data and Contrast Media & Molecular Imaging 13 image segmentation accuracy [24,25]. In the future, with the development of multidiscipline research, better solutions will be developed to solve biomedical problems using more mature imaging technologies.

Conclusions
In this study, GO-AuNS-DOTA-Gd was prepared and its physicochemical properties indicated that the probe could act as an efficient photosensitizer for photothermal therapy and be used for precisely delineating the tumor margin from normal tissues based on multimodality molecular imaging. This multimodality imaging strategy is a promising technology for clinical use, especially for the orthotopic tumor, which deserves further development for application in other tumor surgery and treatment.