Molecular profiling of angiogenesis in hypericin mediated photodynamic therapy

Hypericin fluorescence in the tumor was evaluated using a fluorescence endoscopy system to determine the uptake kinetics in the tumor at 0.5 and 6 h post drug administration. Fluorescence intensity in the tumor was higher at 6 h compared to 0.5 h post HY administration suggesting higher tumor selectivity (Fig. 1A). At 6 h fluorescence in muscle and skin was lower indicating higher clearance from normal tissue. The tumor to normal tissue ratio tabulated from the fluorescence images was greater at 6 h compared to 0.5 h post HY administration, p < 0.01 (Fig. 1B).

Hypericin microfluorescence was detected using a confocal laser scanning microscope to establish the correlation between macro and microfluorescence observed in the tumors at 0.5 and 6 h post drug administration. Microscopic fluorescence of HY in the tumor was significantly higher at 6 h compared to 0.5 h time point (Fig. 2A). Fluorescence increased twofold at 6 h post HY administration compared to 0.5 h, p < 0.01 (Fig. 2B).

Tumors were allowed to grow up to sizes of 6–7 mm in diameter before PDT treatment was carried out, and were measured three times a week. Animals subjected to vascular targeted short DLI PDT exhibited greater tumor response compared to cellular targeted long DLI PDT, p < 0.05 and control, p < 0.01 (Fig. 3). The decrease in growth rate in short DLI PDT group is indicative of significant damage to the tumor vasculature during treatment.

Blood vessel staining was performed to calculate vessel density after HY-PDT treatment. Immunohistochemistry revealed highest endothelium density at 24 h after long DLI PDT (IHC score 4) (Fig. 4C), however at 24 h post short DLI PDT, lesser smudged and diffused staining pattern (IHC score 2) of endothelial cells were noted (Fig. 4B). Moderate staining (IHC score 1) of blood vessels were observed in control, short and long DLI PDT 30 days treated tumors (Fig 4A, D, E).

Blotting of angiogenic proteins, i.e. VEGF, TNF-α and IFN-α was carried out to understand the modulations in protein expression at 24 h and 30 days post PDT treatment. TNF-α was significantly downregulated at 30 days post vascular targeted short DLI PDT compared to all other treatment groups. Expression of VEGF and IFN-α was significantly upregulated in long DLI 24 h group compared to all other groups (p < 0.05) (Fig 5A, B).

Expression of major angiogenic proteins i.e., VEGF, bFGF, IFN-γ and IL-6 were detected in the treatment groups using human angiogenesis antibody arrays. VEGF expression was significantly greater (p < 0.05) in the treatment groups compared to control tumors. bFGF expression was significantly greater in vascular mediated short DLI PDT compared to cellular mediated long DLI PDT groups. IFN-γ expression was most pronounced at 30 days in the long DLI PDT group. Upregulation of inflammatory protein IL-6 was remarkably higher compared to all other angiogenic proteins. At 24 h post PDT significantly high levels of IL-6 were expressed (Fig. 6).

Gene expression of different genes involved in angiogenesis was assessed to understand the role of transcriptional factors before tumor regrowth. Adhesion molecules i.e., cadherin 5 (CDH5) and collagens (COL18A1 and COL4A3) were downregulated 24 h post vascular targeted short DLI PDT and cellular targeted long DLI PDT, however at 30 days post treatment CDH5 and COL18A1 were upregulated with the exception of COL4A3. Intergrin alpha V (ITGAV) and Intergrin beta 3 (ITGB3) were downregulated in short and long DLI PDT at 24 h, however increased mRNA transcription was noticed at 30 days time point with the exception of ITGAV that was downregulated at short DLI PDT. Ephrin A3 (EFNA3) and Ephrin B2 (EFNB2) are ligands of the eph receptor family that plays a major role in tumor remodeling. Ephrin A3 was upregulated at 24 h post PDT, however ephrin B2 was downregulated. Both EFNA3 and EFNB2 were overexpressed 30 days post PDT treatment. Hepatocyte growth factor (HGF) expression was found to be 4-fold greater than epidermal growth factor (EGF) at 24 h post vascular mediated short DLI PDT. HGF acts through the Met-receptor pathway to promote angiogenesis and its transcription was well pronounced in all our PDT treated groups. EGF induced signaling has often been associated with tumor invasion and metastasis. In our experiments, EGF was significantly overexpressed at 24 h in vascular targeting short DLI PDT and was downregulated in all other treatment groups. Increased expression of platelet-derived growth factor alpha (PDGFA) was observed in all the treatment groups. Neuropilin 2 (NRP2), a nonsignaling transmembrane receptor was suppressed only at 30 days post long DLI PDT. The notch ligand Jagged 1 (JAG1) was downregulated in all the treatment groups. TEK tyrosine kinase (TEK) was significantly upregulated at 30 days post short DLI PDT only. Angiogenesis growth factors such as transforming growth factor alpha (TGFA) and vascular endothelial growth factor C (VEGF-C) were downregulated 24 h post PDT but were upregulated by 30 days. Tumor necrosis factor, alpha induced protein 2 (TNFAIP2) was downregulated only at 24 h post long DLI PDT. The function of each gene is provided in Table 1, and fold change and p-values for the genes are provided in Tables 2 and 3.

The photosensitizer HY (Molecular Probes, OR, USA) was dissolved in dimethyl sulfoxide, DMSO (Sigma Aldrich Inc, St Louis Mo, USA) and diluted in phosphate-buffered saline (PBS) and administered intravenously at a dose of 5 mg/kg.

Male balb/c nude mice, 6–8 weeks of age, weighing an average of 24–25 g were obtained from the Animal Resource Centre, Western Australia. To establish a murine xenograft model of human bladder carcinoma, the present study used MGH epithelial bladder cell line. Briefly, cells were cultured as a monolayer in RPMI-1640 medium. Approximately 3.0 × 10⁶ MGH bladder cells suspended in 150 μl of Hanks' balanced salt solution (Gibco, USA) were injected subcutaneously into the lower flanks of the mice. The mice were kept for 10–14 days to allow the tumors to grow till they were palpably greater than 5–6 mm. The tumor volume was calculated using the following formula: volume = (π/6 × d1 × d2 × d3), where d1, d2 and d3 are tumor dimensions in 3 orthogonal directions. All procedures carried out in this study were approved by the Institutional Animal Care and Use Committee (IACUC), SingHealth, Singapore and were conducted in accordance with international standards.

A dose of 5 mg/kg of hypericin was administered to mice through tail vein injections. The mice were anesthetized using 1:1 vol/vol cocktail of ketamine hydrochloride (Trittau, Germany) and valium (David Bull Laboratories, Australia). The skin overlaying the tumor was carefully removed to expose the tumor and imaging was performed at 0.5 and 6 h post hypericin administration. A fluorescence endoscope system (Karl Storz, Tuttlingen, Germany) was used to perform macroscopic fluorescence digital imaging. This system consists of a fluorescence detection unit, an illumination console, a video displaying and recording unit, and a computing system for image acquisition, display and processing. The D-Light system contains an excitation filter (375–440 nm) that allows high transmission in the violet part of the spectrum and blocks the rest of the visible part of the spectrum. The fluorescence can be detected since a large portion of the excitation light is being blocked by an observation filter (cut-off wavelength at 470 nm). An optical filter is located in the eyepiece of the fluorescence endoscope (Karl Storz, Hopkins II 0°, 4 mm) and it reduces blue light excitation that is diffusely back-scattered by the tissue and transmits red fluorescent light. The images thus obtained were analysed using the image analysis software, MicroImage 4.0. Images obtained in blue light needed to be further processed to obtain normalized fluorescence intensity images. The processing involves contrast enhancement of the original image. This is followed by hue extraction where the fluorescence color (red) is extracted. Following hue extraction, the thresholding and segmentation of the area of interest (AOI) was carried out. This gives us the red-black image, from the red-black image we obtained the relative intensity distribution over the tumor and the normal regions. Finally, the normalized intensity image, which is the ratio of the tumor fluorescence to the normal (blue) background, was obtained. The ratio of red intensity over blue intensity from 10 images per group was quantified using the software and the graph was plotted.

After intravenous administration of hypericin the animals were sacrificed at 0.5 and 6 h and tumors were subsequently extracted. The tumors were then snap frozen in liquid nitrogen. Cryosections of 20 μm thickness were obtained using a microtome cryostat (Cryo-Star HM 560 MV, Germany) and the sections were mounted onto slides. A laser confocal fluorescence microscope (Meta LSM 510, Carl Zeiss, Germany) was used to obtain confocal fluorescence images. A 488 nm Argon laser was used to excite the tissue. Fluorescence emissions in the wavelength range of 590–630 nm were spilt by a diachronic filter and detected through a band-pass filter (BP-610 nm, Omega Optical, USA). Each fluorescence image of any field was reconstructed from a sequence of 20 sequential optical images. Voltage gain, PMT voltage and sensitivity (contrast, brightness and filters) were fixed for all fields and slides imaged. Fluorescence data analysis was carried out using an image software package (Kontron KS400, version 3.0, Hallbergmoos, Germany). A total of 10 images, acquired from different tumor samples, were used to quantify fluorescence intensity. Briefly, digital quantification was carried out by using box superimposition of an area of 100 pixels, or contour super-imposition where the contour of the area of interest was outlined and the pixel intensities per unit area determined. This was done to determine the relative hypericin fluorescence between different regions of the same specimen.

Animals were assigned to five different groups: (i) control (untreated tumor), (ii) short DLI PDT 24 h, (iii) long DLI PDT 24 h, (iv) short DLI PDT 30 days and (v) long DLI PDT 30 days, and each group comprised of 10 animals. Tumors from groups (ii) and (iii) were extracted at 24 h post PDT whereas tumors from groups (iv) and (v) were extracted at 30 days post PDT. The PDT treatment groups received intravenous injection of HY followed by irradiation with a light source consisting of broadband halogen light (Zeiss KL1500) fitted with a customized 560–640 nm band-pass filter. A light dosage of 120 J/cm² and 100 mW/cm² was used for PDT treatment. It should be noted that only the DLI was varied, the photosensitizer dose was maintained for all the treated groups.

Tumor tissues from both control and HY-PDT treated animals were used for immunohistochemistry. Immunohistochemical staining was done using the Chemicon Blood vessel staining kit on formalin-fixed, paraffin-embedded control and HY-PDT tumor tissues. Sections were incubated with anti-CD31 (1:100) for 1 h followed by incubation with anti-rabbit secondary antibodies conjugated to horseradish peroxidase-linked labeled polymers. 3,3'-Diaminobenzidine-positive substrate-chromogen was added to sections and incubated for 5 minutes for color development. Finally, the slides were counterstained with Harris's haematoxylin for 5 seconds, dehydrated, and mounted. Primary antibody was replaced with normal mouse serum IgG₁ (1:500) to confirm the specificity and served as negative control. Microvessels stained with CD31 were counted by light microscopy at X200 magnification. Any stained endothelial cell or cell cluster clearly separated from adjacent vessels was considered a single countable microvessel. Slides were examined by two independent observers. IHC scoring was performed based on the prevalence of CD31 staining within the tumor (No staining = 0, < 10% staining = 1, 10–25% = 2, 25–50% = 3, 50–75% = 4 and 75–100% = 5). Images from the slides were captured using image processing software (Kontron KS400, version 3.0, Hallbergmoos, Germany).

For Western blot analysis, 30 μg of protein were resolved on 10–12% polyacrylamide-SDS gels and transferred onto a nitrocellulose membrane as described elsewhere. The membrane was immersed in blocking buffer (5% non-fat dry milk/1% Tween 20 in 20 mM TBS, pH 7.5) for 1.5 h at room temperature and incubated with the appropriate primary antibody i.e., mouse monoclonal anti-TNF-α (1: 50) and IFN-α (1:100) (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and mouse monoclonal anti-human VEGF (1:100) (BD Pharmingen, BD Biosciences, CA, USA) solved in blocking buffer overnight at 4°C, followed by incubation with complementary secondary antibody. Proteins were detected by chemiluminescence (Supersignal, Pierce Technology, Rockford, IL, USA) and autoradiography using Hyperfilm ECL (Amersham Biosciences, UK).

Expression of angiogenesis proteins following HY-PDT were measured using the RayBio^® Human Angiogenesis Antibody Array 1.1 (RayBiotech Inc, USA) in control and PDT subjects (8 control and 8 HY-PDT) using one array per tissue sample. Only expressions of proteins relevant to this study are reported in this article. The components in the kit included protein array membranes, biotin-conjugated anti-cytokines, HRP-conjugated streptavidin and detection buffer. Briefly, the protein array membranes were immersed in blocking buffer, after which the membranes were incubated with tumor lysate for 2 h. After extensive washing the membrane was incubated with primary antibody and after further washing the HRP-conjugated streptavidin was added and incubated for another 2 h. After adding the detection buffer, the membranes were exposed to X-ray film (Hyperfilm ECL, Amersham Biosciences UK) and the signal was detected using film developer (Kodak M35 OMAT Processor). The intensities of the signal were quantified using a densitometer. By comparing the intensities of the signals, the relative expression levels of the angiogenic proteins were analyzed. The results were normalized to an internal positive control provided on each membrane. RayBio Analysis Tool, a program specifically designed for the analysis of antibody arrays, was used to plot the graphs.

Total RNA was extracted from tumor tissue using the commercially available Nucleospin RNA II kit (Macherey-Nagel, Germany). Briefly, the frozen tissue samples were crushed into powder using liquid nitrogen and lysis buffer, and β-mercaptoethanol was added to prepare the lysate. The lysate was then filtered and 70% ethanol was added to adjust RNA binding to the columns. Later DNA digestion was performed and pure RNA was eluted. RNA quality and purity was checked using UV Spectrophotometry and by detecting the ribosomal RNA integrity. Complementary DNA (cDNA) was synthesized from 2 μg DNA-free total RNA before PCR analysis.

The Human Angiogenesis RT² Profiler™ PCR Array (SuperArray, USA) was used to identify the expression of 84 key genes involved in modulating the biological processes of angiogenesis. The array includes growth factors and their receptors, chemokines and cytokines, matrix and adhesion molecules, proteases and their inhibitors as well as transcription factors, involved in angiogenesis. For each PCR evaluation, amplification and detection was performed with the ABI Prism 7000 Sequence Detection Systems (Applied BioSystems, USA). Average ΔC_t value for each gene across triplicate arrays for each treatment group was calculated. ΔΔC_t for each gene across control and experimental group were determined. Finally the fold change was computed for each gene from control and other groups as 2^(-ΔΔC_t).