Photodynamic therapy (PDT) induces tumor destruction through a photochemical reaction involving a photosensitizer, light of a specific wavelength matching the absorption of the photosensitizer and molecular oxygen.1 Singlet oxygen, a product of this photochemical reaction, causes oxidative damage to target cells and tissues and is the primary reactive oxygen species responsible for the biological effects of PDT.2 Although direct tumor cyotoxicity and immune responses are involved as well, damage to the tumor vasculature has been shown to contribute significantly to the overall PDT effect of most photosensitizers.3
Verteporfin (the lipid-formulation of benzoporphyrin derivative monoacid ring A) is a photosensitizer that is currently approved for the treatment of age-related macular degeneration (AMD).4 We have shown previously that the dynamic distribution of verteporfin is predominantly intravascular at 15 min after intravenous injection and becomes mainly extravascular at 3 hr after injection.5 Based on this pharmacokinetic property, preferential tumor vascular targeting can be achieved by illumination at 15 min after verteporfin administration. We have been exploring this passive vascular targeting principle for the treatment of prostate tumors. Intravital fluorescence microscopy studies in the MatLyLu rat prostate tumor model has demonstrated that vascular-targeting PDT with verteporfin induces vascular permeability increase and thrombus formation, which ends in vascular shutdown and tumor necrosis.6 These results indicate that vascular-targeting PDT using verteporfin can be used for the management of localized prostate cancer.
Because vascular damage is the dominant effect of PDT, especially in the case of vascular-targeting PDT, it is important to study in detail how photosensitization modifies vascular functions. Most studies on PDT-induced tumor vascular changes have been done on excised tumor specimens after sacrificing the animals. Although they have been valuable in revealing microscopic details, such studies are only able to provide snap-shot information on each individual animal. To obtain longitudinal information in a single animal, noninvasive imaging techniques are necessary to examine vessel functional changes after PDT. Imaging modalities such as laser Doppler perfusion imaging,7, 8 diffuse correlation spectroscopy,9 laser speckle imaging,10, 11 optical coherence tomography12 and ultrasonography13 have all been shown to be useful techniques for monitoring tumor blood flow dynamics noninvasively after PDT. Moreover, noninvasive imaging using contrast agents allows one to follow perfusion changes and also provides real-time information regarding vascular permeability. For instance, angiography with fluorescent dyes such as fluorescein or indocyanine green is routinely used to examine vessel leakage and occlusion in AMD patients treated with PDT.14 Changes in tumor perfusion and vascular permeability after PDT have also been studied with contrast-enhanced MRI.15, 16
Because of its high sensitivity and versatility, in vivo fluorescence imaging is able to provide both macroscopic and microscopic longitudinal data in individual animals, which cannot be obtained in other ways.17–19 In this study, we used an in vivo whole-body fluorescence imaging system to monitor vascular response and tumor cell survival in an EGFP-expressing prostate tumor model following treatment with verteporfin-PDT. Moreover, we compared the in vivo tumor imaging results with the ex vivo fluorescence microscopy of frozen tumor sections. Our results indicate that the vascular response to vascular-targeting PDT is clearly different between tumor interior vessels and peripheral blood vessels.
Material and methods
Production and titer of lentivirus
Lentiviral production was performed as previously described.20 Briefly, we cotransfected pWPT-EGFP and third-generation packaging vectors into 293FT cells (Invitrogen Life Technologies) and collected culture supernatants after 48 and 72 hr of incubation in a 37°C and 5% CO2 incubator. We recovered virus by ultracentrifugation (1.5 hr at 25,000 rpm) in a Beckman SW28 rotor and resuspended the virus pellet in 25 μl of Opt-MEM media (Invitrogen Life Technologies). Viral titers were determined by infecting 293FT cells with serial dilutions of concentrated lentivirus followed by flow cytometry analysis 48 hr later. Typical viral preparations yielded 5 × 108 transducing units/ml.
Tumor cells and lentiviral transduction
R3327-MatLyLu rat prostate cancer cells were maintained in the RPMI-1640 medium with glutamine (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (HyClone, Logan, UT) and 100 units/ml penicillin–streptomycin (Mediatech) at 37°C in a 5% CO2 incubator. For lentiviral transduction, the MatLyLu cells were infected with a multiplicity of infection of 50 and allowed to incubate overnight. Polybrene (8 μg/ml, Sigma) was used to facilitate lentiviral transduction. Supernatant was then removed after infection and replaced with complete RPMI-1640 growth medium. EGFP-transduced MatLyLu cells were examined with a fluorescence microscope at 48 hr after transduction. EGFP-MatLyLu cells were harvested, serial diluted and seeded in a 96-well plate with cell density of 1 cell per well. After incubation for 7 days at 37°C and 5% CO2 atmosphere, the clone exhibiting the highest EGFP fluorescence intensity was selected and expanded for subsequent experiments.
Animals and tumor models
Male NCr athymic nude mice (4–5 weeks old, National Cancer Institute, Frederick, MD) were used throughout the study. Tumors were induced by subcutaneous injection of about 1 × 105 EGFP-MatLyLu tumor cells in the thigh region of mice. Tumors were used for experiments when they reached a size of 5–7 mm in diameter. All animal procedures were carried out according to a protocol approved by the Institutional Animal Care and Use Committee (IACUC).
Verteporfin (benzoporphyrin derivative (BPD) in a lipid-formulation) was obtained from QLT (Vancouver, Canada) as a gift. A stock saline solution of verteporfin was reconstituted according to the manufacturer's instructions and stored at 4°C in the dark.
A diode laser system (High Power Devices, North Brunswick, NJ) at 690-nm wavelength was used for the irradiation of EGFP-MatLyLu tumors. The laser was coupled to an optical fiber with 600 μm core diameter and expanded to generate an 11-mm diameter illumination spot through a collimator. Animals were anesthetized with injection (i.p.) of a mixture of ketamine (90 mg/kg) and xylazine (9 mg/kg) and tumors were exposed to light with an irradiance of 50 mW/cm2. Light intensity was measured with an optical power meter (Thorlabs, North Newton, NJ). Verteporfin was injected (i.v.) 15 min prior to light irradiation at a dose of 0.25 mg/kg.
Noninvasive tumor fluorescence imaging and image analysis
Tumor-bearing animals were i.v. injected with 20 mg/kg albumin labeled with tetramethylrhodamine isothiocyanate (TRITC-albumin, Sigma) immediately after PDT. EGFP-MatLyLu tumors were imaged with a noninvasive whole body fluorescence imaging system for the EGFP and TRITC signal before and at various times after treatment. The setup of this home-built broad beam imaging system has been described in detail in our previous paper.21 Briefly, the system includes a filtered white light source for excitation and a SensiCamQE high performance digital CCD camera (The Cook Corp, Auburn Hills, MI) to capture fluorescence emission passing through an emission filter. We used a 470/20 nm excitation filter and a 520/20 nm emission filter for imaging tumor EGFP fluorescence and a 535/20 nm excitation filter and 590-nm long-pass emission filter for imaging the TRITC fluorescence. Camera settings were kept constant for the control and PDT-treated animals throughout the imaging process. Animals were anesthetized by inhalation of 1.5% isofluorane and imaged first for EGFP and then TRITC fluorescence without moving the animals. The EGFP and TRITC images were pseudocolored and superimposed to generate composite images.
A 2.5-mm diameter region of interest (ROI) was centered over tumor or tumor-adjacent normal tissue areas, and the average EGFP and TRITC fluorescence intensities in the ROI were quantified with NIH ImageJ software. The fluorescence intensity in tumor or tumor-adjacent tissues after PDT was normalized to its own pretreatment value in each animal, and the data from different animals in each group were pooled to generate response curves. To determine the TRITC-albumin distribution in relation to tumor EGFP fluorescence, a straight line was drawn through the tumor tissue on composite images and the corresponding green (EGFP) and red (TRITC) intensities were measured along the line.
Tumor tissue fluorescence microscopy
Tumor-bearing animals were i.v. injected with 20 mg/kg Hoechst (Sigma) as a vascular perfusion marker at different time points after treatment. Animals were euthanized within 1 min after injection and tumor tissues were excised and snap-frozen in isopentane precooled with liquid nitrogen. Frozen tumor sections with thickness of 10 μm were cut and examined under a Leica DMI6000B fluorescence microscope with appropriate filter sets for Hoechst (excitation: 360/40 nm; emission: 470/40 nm) and TRITC (excitation: 546/12 nm; emission: 600/40 nm).
Tumor volume measurement and tumor histology
Three-dimensional tumor sizes were measured regularly after treatment by caliper, and the tumor volume was calculated using the formula π/6 × tumor length × tumor width × tumor height. Animals were euthanized at various time points after treatment. Tumor tissues were excised and fixed in 4% formalin solution. Fixed tumor tissues were dehydrated and then embedded in paraffin. Tissue sections with thickness of 5 μm were cut and stained with H&E.
Students' 2-tailed t-test was used to calculate statistical differences between 2 groups and the significance was accepted at p < 0.05. Statistical analysis was carried out using GraphPad software (GraphPad, San Diego, CA).
The extravasation of TRITC-albumin, as indicated by the increase in TRITC fluorescence, was imaged noninvasively with a whole-body fluorescence imaging system. Figure 1 shows the TRITC fluorescence images (red) merged with tumor EGFP fluorescence images (green) at different time points after vascular-targeting PDT with verteporfin. PDT caused an overall increase in the TRITC fluorescence and this was more pronounced in the peritumor area. PDT-induced TRITC-albumin extravasation appeared to be dose dependent because the 50 J/cm2 light dose PDT caused a greater increase in the TRITC fluorescence compared to the 25 J/cm2 light dose treatment.
The average TRITC fluorescence intensity in tumor and tumor-adjacent normal tissue ROIs was quantified with NIH ImageJ software. It was observed that the average TRITC fluorescence in tumor areas was about 20% lower than tumor-adjacent normal tissue areas presumably because higher blood volume in tumor tissues causes more TRITC fluorescence quenching than in normal tissues.19 Both 25 and 50 J/cm2 PDT treatments significantly increased the TRITC fluorescence intensity in tumor (Fig. 2a, p < 0.05) and tumor-adjacent tissues (Fig. 2b, p < 0.05). Fluorescence intensity increase started from 1-hr post-PDT treatments and reached a plateau at about 4 hr thereafter while untreated control tumors exhibited little change in fluorescence intensity over the same period of time. In both tumor and tumor-adjacent tissues, PDT with 50 J/cm2 light dose induced a greater increase in the TRITC fluorescence intensity than the 25 J/cm2 light dose (p < 0.01). The 25 J/cm2 light dose PDT caused a similar increase (maximally about 1.5-fold increase) in the TRITC fluorescence intensity in both tumor and tumor-adjacent tissues (p > 0.05). The 50 J/cm2 PDT caused significantly higher TRITC fluorescence increase in tumor-adjacent tissues (about 3-fold increase at peak) compared to tumor tissues (about 2-fold increase at peak, p < 0.05).
Changes in the average EGFP fluorescence intensity in tumor tissues were also quantified and are shown in Figure 2c. Both 25 and 50 J/cm2 PDT treatments caused a significant decrease in tumor EGFP fluorescence at 1 hr after treatment (p < 0.05). After the initial decrease, there was no further decrease in tumor EGFP fluorescence intensity. Control tumors showed little change in the EGFP fluorescence during this 5-hr period.
Analysis of TRITC and corresponding EGFP intensity profiles indicated that the TRITC fluorescence intensity in tumor peripheral area was higher than in tumor interior area at 4 hr after injection of TRITC-albumin (Fig. 3). However, an opposite pattern was found in tumor EGFP intensity profiles with the higher intensity values detected in the tumor center. Both 25 and 50 J/cm2 PDT treatments caused an overall increase in the TRITC intensity and decrease in tumor EGFP intensity. The increase in the TRITC intensity was found to be higher in the tumor periphery than in the tumor center.
To verify the whole-body fluorescence imaging results, we euthanized animals at 1, 4 and 24 hr after 50 J/cm2 PDT treatment and excised tumor tissues for fluorescence microscopy. Hoechst dye was i.v. injected shortly before euthanizing animals to highlight functional blood vessels. As shown in Figure 4, tumor staining of Hoechst dye decreased significantly after vascular-targeting PDT with 50 J/cm2 light dose compared to the control tumor, indicating a decrease in functional blood vessels. Moreover, functional blood vessels were mainly detected at the tumor periphery after PDT. In agreement with the macroscopic in vivo tumor imaging results, fluorescence microscopy also demonstrated a significant increase in the TRITC fluorescence intensity after PDT, especially in the tumor periphery.
Tumor response to vascular-targeting PDT was monitored noninvasively by whole body fluorescence imaging. The EGFP-MatLyLu tumors were imaged for EGFP fluorescence before and after treatments. Representative tumor EGFP fluorescence images are shown in Figure 5. Control tumors grew rapidly and exhibited central necrosis when tumor reached about 8–10 mm in diameter. Dead EGFP-MatLyLu tumor cells were unable to produce EGFP, causing dead tumor tissues to appear as dark areas in the EGFP fluorescence images. PDT with 25 J/cm2 light dose induced a partial tumor necrosis, but this PDT condition failed to inhibit tumor growth (Fig. 5). In fact, tumor growth after this PDT treatment was even more rapid than control tumors and average tumor volume was nearly twice that of control tumors at 9 days after treatment (Fig. 6, p < 0.01). In contrast, the 50 J/cm2 PDT effectively inhibited prostate tumor growth as indicated by a substantial decrease in EGFP fluorescence (Fig. 5) and average tumor volume (Fig. 6, p < 0.01 compared to the control tumor) after treatment. EGFP fluorescence was barely detectable at 2 days after PDT. But small EGFP fluorescent spots, indicating the existence of viable tumor cells, were often found at tumor edges several days after treatment and gradually grew in size which led to tumor recurrence. As shown in Figure 7, some viable tumor cells were clearly detected in tumor periphery at 48 hr after 50 J/cm2 PDT.
A whole-body animal fluorescence imaging system was used in this study to visualize noninvasively tumor response following PDT targeting of tumor blood vessels in an EGFP-expressing MatLyLu prostate tumor model. TRITC-albumin was used as a macromolecular probe to image tumor vascular barrier function (vascular permeability). The increase in the TRITC fluorescence intensity, caused by enhanced extravasation from blood vessels, is an indicator of vascular barrier disruption. Albumin has a plasma half-life of more than 24 hr and it was used to follow vascular permeability changes up to several hours after treatment.22
We found in the present study that vascular-targeting PDT increased vascular permeability in a dose-dependent manner, which is in agreement with our previous study and indicates that tumor vasculature is a primary target of PDT with verteporfin.6 Importantly, our results demonstrate that the enhanced TRITC-albumin tumor uptake as a result of PDT-induced permeability increase was not homogeneous in tumor tissues. Both in vivo and ex vivo tumor imaging studies indicate that increase in TRITC-albumin extravasation was significantly higher in the peripheral tumor area than in the interior tumor area. Because the accumulation of a circulating molecule in tumor tissues is dependent upon the existence of functional blood vessels, the enhancement of TRITC-albumin accumulation in the tumor periphery is likely related to the predominant localization of functional blood vessels in peripheral tumor areas after vascular-targeting PDT. As shown in Figure 4, PDT was remarkably effective in inducing interior tumor blood vessel shutdown while some peripheral vessels were still functional up to 24 hr after PDT. Early closure of central tumor vessels limited the enhancement of TRITC-albumin in the tumor interior, whereas prolonged perfusion of some peripheral tumor vessels allowed more TRITC-albumin to continuously extravasate in the tumor periphery. We and others have previously reported that peripheral tumor vessels tend to maintain perfusion function after vascular-targeting PDT.23–25 Our present results further demonstrate that continuous functioning of peripheral blood vessels, which had been permeabilized by PDT, led to preferential accumulation of circulating molecules in the tumor periphery.
The existence of functional blood vessels in the tumor periphery was associated with peripheral tumor cell survival after PDT. As shown in Figure 7, H&E staining indicated a rim of viable tumor cells in the tumor periphery at 48 hr after PDT in spite of extensive tumor necrosis. In vivo imaging of tumor EGFP fluorescence demonstrated that the survival of these peripheral tumor cells resulted in peripheral tumor recurrence (Fig. 5). Here we used EGFP as an indicator of tumor cell viability with the assumption that dead tumor cells are not able to synthesize EGFP and emit EGFP fluorescence. However, because EGFP has a half-life of more than 3 hr,26 monitoring EGFP fluorescence shortly after treatment might not accurately report tumor cell viability. Sufficient time is needed for the degradation of EGFP synthesized before treatment in order to use EGFP fluorescence to report cell viability. The observed decrease in EGFP fluorescence shortly after PDT in the present study was likely due to the oxidative degradation of EGFP during PDT rather than a real decrease in tumor cell viability. This was supported by the fact that there was little further decrease in EGFP fluorescence intensity over the following 5 hr period after PDT (Fig. 2c).
It is still not clear why peripheral tumor blood vessels react differently from interior blood vessels to vascular-targeting PDT. Understanding the mechanism behind this disparity in vascular response will help find ways to enhance the therapeutic effects of vascular-targeting PDT. Differences in vascular structure and function between tumor peripheral and interior blood vessels caused by morbid tumor pathobiology possibly contribute to such variations in vascular response. It is known that tumor tissues have higher tissue interstitial pressure than normal tissues because of leaky tumor blood vessels and poor lymphatic system function.27, 28 High tumor interstitial pressure is able to compress tumor vessels and lead to vessel collapse. Vessel compression and collapse are more severe in the tumor interior where tumor interstitial pressure is higher.29, 30 PDT has been shown to further increase tumor interstitial pressure as a result of enhancing vascular permeability.31, 32 Such an increase in tumor interstitial pressure will likely impose a greater compression on tumor blood vessels and cause vascular shutdown, especially in tumor interior areas. Moreover, we recently found that, compared to the interior tumor vessels, peripheral tumor blood vessels were generally larger and exhibited vascular lumen as well as more coverage of vascular pericytes and basement membrane.33 Less mechanic compression together with more vessel supporting structures might make peripheral tumor vessels more resistant than the interior vessels to vessel closure induced by vascular-targeting PDT.
The survival of peripheral tumor cells as a consequence of disparity in vascular response between peripheral and interior blood vessels represents a therapeutic challenge for the vascular-targeting PDT. Several strategies can be adopted to eliminate or at least minimize surviving tumor cells at the tumor periphery. First of all, we could increase the PDT dose to determine whether a higher dose of vascular-targeting PDT will lead to the shutdown of both interior and peripheral tumor blood vessels, resulting in an increased tumor cure. Secondly, as combination therapies have been routinely used in cancer treatments, one approach of enhancing photodynamic vascular targeting effectiveness is to combine it with other cancer therapies. Combination therapies can be designed based on different targeting principles. Targeting both tumor vascular and cellular compartments by combining vascular-targeting PDT with a cancer cell-targeted therapy could be a promising strategy because the increased vascular permeability induced by PDT has been shown to enhance drug delivery.6, 34, 35 Our present study further demonstrates that the enhancement of drug accumulation mainly occurred at the tumor periphery where tumor cell survival tends to occur after vascular-targeting PDT. Therefore, combining vascular-targeting PDT with other anticancer drug therapies will allow more anticancer agents to be preferentially deposited in the peripheral tumor area to kill tumor cells that otherwise might survive after PDT treatment.
In summary, we utilized in vivo animal fluorescence imaging combined with standard ex vivo tissue fluorescence microscopy to examine changes in vascular function and tumor cell viability after vascular-targeting PDT. Our results indicate that, although PDT causes an overall increase in vascular permeability, peripheral tumor blood vessels are somehow able to maintain perfusion function whereas interior blood vessels are shutdown shortly after PDT. Such a disparity in vascular response is conducive to peripheral tumor cell survival and also explains the preferential accumulation of circulating molecules in the tumor periphery. We are currently investigating the mechanisms underlying this response disparity and exploring therapeutic strategies that minimize the survival of peripheral tumor cells.
The authors gratefully acknowledge Dr. Tayyaba Hasan of the Wellman Center for Photomedicine for helpful discussions and QLT Inc. for providing verteporfin.