Epidermal Growth Factor Receptor (EGFR)-targeted Photoimmunotherapy (PIT) for the Treatment of EGFR-expressing Bladder Cancer

Bladder cancer is the fourth most common cancer in men and the 12th most common cancer in women. There will be an estimated 76,960 new cases and 16,390 deaths attributed to bladder cancer in 2016 (ranking eighth in all cancer-related deaths in men; ref. 1). In addition to its prevalence, bladder cancer is the most expensive malignancy to treat from diagnosis to death (2). Unfortunately, the standard of care for localized bladder cancer treatment has undergone only incremental improvement over the past several decades. Approximately 70% of cases are non-muscle–invasive bladder cancer (NMIBC) at presentation and are treated by transurethral resection of bladder tumor (TURBT) followed by intravesical treatment with BCG (Bacillus Calmette-Guerin) or mitomycin C. However, in the setting of high-grade disease, these therapies can become ineffective over time in up to two-thirds of patients (3), and disease progression to muscle-invasive bladder cancer (MIBC) can occur. MIBC is aggressive and only 50% of patients will survive five years despite undergoing radical cystectomy (4). Clearly, there is a large unmet need in therapeutic options for NMIBC that recurs or progresses.

A potential therapeutic target is the EGFR. EGFR is overexpressed in up to 74% of bladder cancer tissue specimens (5) but has a relatively low expression in normal urothelium (6). In addition, EGFR is an independent predictor of decreased survival and stage progression in bladder cancer (5). The Cancer Genome Atlas (TCGA) project revealed EGFR amplification in up to 11% of MIBC with predominant urothelial cell carcinoma histology (UCC; ref. 7). EGFR is localized to the basal layer of urothelial cells in normal urothelium but is present in both the luminal and basal layers of urothelial cells in bladder cancer (8). This amplification and luminal localization of EGFR in urothelial tumors make intravesical therapy a potential treatment option in bladder cancer. EGFR is especially amplified in squamous cell carcinomas (SCC) of the bladder. Only 2%–5% of all bladder cancer cases are pure squamous cell carcinoma (SCC) of the bladder (9); however, up to 60% of UCC contain histologic features of squamous differentiation (10). Previous IHC observations demonstrate that about 67% of bladder cancers with squamous differentiation express EGFR (11) and about 90% of pure SCCs of bladder express EGFR (12, 13). Therefore, EGFR appears to be a viable target in all pure SCC, most UCC with squamous differentiation, and some UCC without squamous differentiation.

In this study, we targeted EGFR-expressing bladder cancer cells using the anti-EGFR antibody panitumumab conjugated with IR700, a photosensitizer (PS)/photoabsorber (PA). IR700 is a highly hydrophilic dye activated by near infrared radiation (NIR) of approximately 689 nm. By itself, it has no therapeutic effect as its hydrophilic nature prevents it from interacting with cell membranes, thereby decreasing toxicity. However, the panitumumab–IR700 conjugate selectively binds to EGFR-expressing cells and, when activated with NIR, selectively destroys only cells bound to the conjugate. Unlike traditional photosensitizers, panitumumab-IR700 is more selective with fewer potential side effects. This novel use of a mAb–PA conjugate has been termed “photoimmunotherapy” (PIT; ref. 14). We detail our preclinical study of the efficacy and mechanism of action of PIT in bladder cancer cell lines using the panitumumab–IR700 immunoconjugate as a novel and selective therapeutic strategy for bladder cancer.

The water-soluble phthalocyanine dye IRDye 700DX NHS ester was purchased from LI-COR Biosciences. A fully human mAb against human EGFR (hEGFR), panitumumab, was procured from Amgen. Rat mAb against hEGFR conjugated with phycoerithrin (PE), PE-conjugated Rat IgG2a, and kappa mAb (isotype control) were obtained from Abcam. All other chemicals were reagent grade.

Panitumumab was conjugated with IRDye 700Dx NHS ester according to a previously published protocol (14). The conjugated antibody was purified using Sephadex G50 column (PD-10 from GE Healthcare). The antibody concentration was determined by measuring the absorption at 280 nm (8453 Value System; Agilent Technologies). Similarly, the amount of IR700 conjugated was determined by absorption at 689 nm. The synthesis of panitumumab-IR700 was controlled such that each antibody molecule bound on average three to four IR700 molecules. The purity of panitumumab-IR700 was assessed using SDS-PAGE. The fluorescent band of panitumumab-IR700 was measured using Odyssey Infrared Imager (LI-COR) at 700 nm.

The bladder cancer cell lines TCCSUP (HTB-5), 5637 (HTB-9), RT4 (HTB-2), T24 (HTB-4), ScaBER (HTB-3), HT1197 (CRL-1473), HT1376 (CRL-1472), UMUC-3 (CRL-1749), SW780 (CRL-2169), epidermoid carcinoma cell line A431 (CRL-1555) and breast cancer cell line MDA-MB-453 (HTB-131) were obtained from ATCC. RT112 was obtained from DSMZ. Metastatic lines of T24 and UMUC-3, T24T, FL3, SLT3 and Lul-2 were gifted by Dr. Michael Nickerson (Division of Cancer Epidemiology & Genetics, NCI, Bethesda, MD). MGH-U3 was obtained from Massachusetts General Hospital. The bladder cancer cell line UMUC-5 was a kind gift of Dr. David McConkey (University of Texas MD Anderson Cancer Center, Houston, TX). UOBL103 is a human SCC cell line established within our laboratory. The normal urothelial cell line, UPS 54, was a kind gift from Dr. Toby Chai (Yale University, New Haven, CT). All the cells except MDA-MB-453 and UPS54 were grown in minimum essential media (MEM; Life Technologies) supplemented with 10% FBS, 1% penicillin/streptomycin, and 1% GlutaMAX (Life Technologies). For UPS54, this medium was supplemented with insulin (1 U/mL) and amphotericin B (1.25 μg/mL). MDA-MB-453 was grown in RPMI-1640 (Life Technologies) instead. All cells required humidified incubator at 37°C with 5% CO₂ for growth. The cell lines were authenticated using short tandem repeat (STR) analysis by Protein Expression Laboratory Core facility of Frederick National Laboratory for Cancer Research (FNLCR). The STR multiplex assay amplifies both 15 tetranucleotide repeat loci and the Amelogenin gender determination marker in a single PCR amplification. Large number of freezes were prepared from each authenticated cell line and once thawed, the cells were used only for 5–8 passages to maintain the integrity of cells.

The presence of EGFR on the surface of bladder cancer cell lines was determined by flow cytometry. Single-cell suspension of these cells (∼1 × 10⁶ cells/tube) was incubated in the presence of PE-tagged rat mAb to hEGFR (Abcam) or PE-tagged Rat IgG2a, kappa mAb (isotype control; Abcam) for 30 minutes at 4°C. Unbound antibody was then washed off and fluorescence was measured on FACSCalibur flow cytometer (BD Biosciences) and data was analyzed on software FlowJo (Treestar Inc.). A431 and MDA-MB-453 were used as positive and negative controls, respectively, for the expression of EGFR.

The bladder cancer cells were plated in 35-mm dishes or 96-well plates for 24 hours. The medium was replaced by fresh, phenol-free media containing no drug, panitumumab, panitumumab-IR700, or IR700 for 1 hour at 37°C. The cells were then irradiated with 4–100 J/cm² NIR (670–710 nm). Unless otherwise stated, most of the following assays were carried out 20–30 minutes after NIR irradiation.

The cytotoxic effect of panitumumab-IR700–based PIT was tested on UMUC-5 and 5637 cells. The PIT-treated cells (panitumumab-IR700 10 μg/mL + 4 J/cm² NIR for UMUC-5 and 32 J/cm² for 5637) were trypsinized and washed with PBS. One microliter/tube of LIVE/DEAD reagent (Life Technologies) was added to cell suspension. After the incubation at 18–25°C for 30 minutes, cells were analyzed on a flow cytometer.

About 20,000 cells/well were seeded in a 96-well plate and incubated for 24 hours, followed by addition of increasing concentrations of panitumumab/panitumumab-IR700. After incubation at 37°C for 1 hour, the cells were exposed to NIR and kept in dark at 37°C for 24 hours. Twenty microliters of MTS reagent were added to each well and plates were kept again in dark for an additional 2–3 hours. The optical density was measured at 490 nm. The half maximal inhibitory concentration (IC₅₀) values were calculated using GraphPad Prism version 6.01 (GraphPad) software.

Single-cell suspensions (1 × 10⁶ cells/tube) were prepared from PIT-treated cells and incubated with FITC Annexin V (BioLegend) and FxCycle Violet (Molecular Probes, Life Technologies) solutions for 15 minutes in dark at room temperature. The type of cell death was evaluated on a flow cytometer using appropriate gates and quadrants.

About 10, 000 cells/well of UMUC-5, 5637, and UMUC-3 cell lines were incubated overnight in white-walled, clear-bottom 96-well plates. The apoptosis inducer, Staurosporine (1 μmol/L; SelleckChem; PubChem CID – 44259), was incubated with the cells for 3–4 hours in the presence and absence of the caspase inhibitor Z-VAD-FMK (20 μmol/L; Promega Corporation; PubChem CID – 5497174). Experimental wells were treated with 10 μg/mL of panitumumab/panitumumab-IR700 or an equivalent concentration of IR700 with or without Z-VAD-FMK (20 μmol/L) for 1 hour followed by irradiation with an appropriate amount of NIR (UMUC-5, 4 J/cm²; 5637, 32 J/cm²; and UMUC-3, 64 J/cm²). Approximately 20 to 30 minutes post-NIR treatment, 100 μL of Caspase-Glo 3/7 reagent was added to each well. Plates were incubated at room temperature for 30 minutes and luminescence was measured using EnSpire multimode plate reader (Perkin Elmer).

PIT-treated cells were trypsinized and fixed overnight in 2.5% glutaraldehyde in 0.1 mol/L cacodylate buffer, pH 7.4. This was followed by secondary fixation in 1% osmium tetroxide in 0.1 mol/L cacodylate buffer, pH 7.4, dehydration in increasing strength of ethanol, and finally infiltration and embedding in resin. Images for cellular ultra-structure were obtained by thin section transmission electron microscopy (TEM).

About 25,000 cells/well were incubated overnight in black-walled, clear-bottom 96-well plates. The cells were washed in assay buffer and stained with 20 μmol/L DCFDA prepared in assay buffer for 45 minutes at 37°C in the dark. DCFDA was washed off using assay buffer and cells were incubated with panitumumab/panitumumab-IR700/IR700 in the presence or absence of water-soluble antioxidant Trolox (1 mmol/L). The fluorescent signal pre- and post-NIR exposure was measured on VICTOR³V Multi-label counter (Perkin Elmer) using fluorescein filter (Ex. 485 nm/Em. 535 nm).

SOSG reagent (Molecular Probes, Life Technologies) was used to detect the presence of singlet oxygen in PIT-treated cells. The assay protocol is similar to the DCFDA assay above except a known inhibitor of singlet oxygen generation, NaN₃ (10 mmol/L), was used instead of Trolox.

Three million UMUC-5 cells resuspended in 50% Matrigel or 3 million UMUC-3 cells were injected subcutaneously on the right thigh of female Athymic Nu/Nu mice (Charles River Laboratory, Frederick National Laboratory) of 6–8 weeks of age. All in vivo procedures were conducted in compliance with the Guide for Care and Use of Laboratory Animal Resources (1996), U.S. National Research Council, and approved by the local Animal Care and Use Committee. Volume of each tumor was determined using an external caliper. The tumor volumes were calculated using the following formula: tumor volume = tumor length × tumor width² × 0.5. When the tumor volumes reached approximately 50 mm³, the mice were randomized in two groups of 10 mice each for the following treatment—control group: 120 μg panitumumab-IR700 injected intravenously without NIR; and experimental group: 120 μg panitumumab-IR700 injected intravenously followed by 2 doses of NIR, 100 J/cm² and 50 J/cm², 24 and 48 hours after panitumumab-IR700 injection, respectively. The mice were anesthetized using 3% isoflurane and were kept asleep throughout the period of NIR exposure. After treatment, the mice were monitored periodically and tumors were measured twice per week. When the UMUC-5 tumors reached 500–1,000 mm³, the mice were euthanized using CO₂. The tumors were dissected out, weighed, and fixed immediately in 10% buffered formalin. Mice bearing rapidly growing UMUC-3 tumors were euthanized when tumor size reached 20 mm in any dimension.

Paraffin blocks were prepared from fixed tumors. Four-micron sections were cut on slides from these blocks and stained for hematoxylin and eosin (H&E) to detect the percentage of necrotic cells in tumors treated with/without PIT. These tumor sections were also analyzed for presence of hEGFR using anti-hEGFR antibody (Cell Signaling Technology no. 4267). In short, slides were deparaffinized and rehydrated to distilled water. Antigen retrieval was performed with 1 mmol/L EDTA for 15 minutes at 95°C. Following a normal goat serum block, sections were incubated with the 1:100 diluted primary antibody overnight. Sections were rinsed and incubated with biotinylated goat anti-rabbit IgG, followed by ABC Elite reagent. 3,3′ Diaminobenzidine (DAB) was used for detection. Slides were counterstained with hematoxylin and covered with coverslip.

Commercially available tissue array blocks BL2081 and BL 806 (Biomax) were similarly stained with anti-EGFR antibody to detect the expression of EGFR in bladder tumors of various stages, grades, and histologies as well as adjacent normal bladder tissues. There were a total of 288 samples with 232 tumors, 8 normal samples, and 48 normal adjacent tumor samples. Of the malignant samples, 86 (36%) were Ta/T1, 108 (47%) were T2, and 38 (16%) were T3. The majority of the malignant samples (197, 85%) were pure urothelial carcinoma, with 7 adenocarcinomas, 10 mucinous adenocarcinomas, and 18 squamous tumors. Each sample was graded by a single pathologist (D. Haines) for staining as follows: 0 = <10% of cells positive; 1 = 10%–24%; 2 = 25%–49%; 3 = 50%–74%; 4 = 75%–100%. For the purpose of evaluating the expression of EGFR in tissue microarrays, staining was considered negative for grade 0 and positive for grade 1–4.

Data are expressed as mean ± SEM from experimental triplicates. Statistical analysis was carried out by the statistical software, GraphPad Prism. Student t test with Mann–Whitney analysis was used to compare the treatment and control effects. χ² was performed for categorical variables. P < 0.05 was used as an indicator of statistically significant difference.

Of the 232 malignant samples, 69 (30%) had <10% staining (grade 0), and 163 (70%) had grade 1 or higher staining. Of those with positive staining, 43 (19%) had grade 1 staining, 11 (5%) had grade 2 staining, 36 (16%) had grade 3 staining, and 73 (31%) had grade 4 staining (Fig. 1A). There was no relationship between EGFR staining and T stage in our analysis (P = 0.9). There was a significantly higher rate of any staining for squamous tumors (17/18, 94%) versus nonsquamous tumors (147/214, 69%), P = 0.04 (Fig. 1B).

The expression of EGFR on the cell surface of various bladder cancer cell lines was examined using flow cytometry without fixation or permeabilization of the cells. The experiment was carried out at 4°C to limit internalization of EGFR. The breast cancer cell line MDA-MB-453, a HER-2 (ErbB2)-positive cell line, was used as a negative control cell line for EGFR (Supplementary Fig. S1A), while, the epidermoid carcinoma cell line, A431, expressing about 2 million EGF-binding EGFRs, was used as a positive control cell line (Supplementary Fig. S1B) (15). The normal urothelial cell line (UPS 54) was used to establish baseline urothelial EGFR expression (Supplementary Fig. S1C). As seen from Fig. 1C, UMUC-5 and ScaBER, the cell lines derived from squamous cell carcinoma (SCC) of the bladder, have very high expression of EGFR (approaching the levels seen in A431 cells; Supplementary Fig. S1D and S1E). ScaBER and UMUC-5 are basal-like cell lines (16) and consequently highly express EGFR. On the other hand, “non-basal-like” cell lines, such as T24, TCCSUP, and RT4 (Supplementary Fig. S1F–S1H; and metastatic derivatives of T24, T24T, FL3 and SLT3; ref. 17) have comparatively lower expression of cell surface EGFR (Fig. 1C). However, all of the bladder cancer cell lines that we have evaluated, with the exception of RT4, still have significant EGFR expression as compared with the negative control MDA-MB-453 cell line and the normal urothelial cells (UPS 54).

Conjugation of panitumumab with IRDye 700Dx NHS ester resulted in approximately three to four IR700 molecules conjugated to each antibody molecule. IR700-conjugated panitumumab did not show any aggregation based on both Coomassie Brilliant Blue–stained gels as well as infrared images of SDS-PAGE gels (Supplementary Fig. S2A). The conjugation of IRDye 700Dx to panitumumab could theoretically lead to the loss of EGFR binding, but this was ruled out in a flow cytometry experiment of UMUC-5 cells incubated with panitumumab-IR700. The median fluorescent intensity of panitumumab-IR700 binding is 605 times more than that of cells alone (Supplementary Fig. S2B) indicating no reduction in binding as a consequence of conjugation.

The amount of NIR light needed to induce discernible morphologic changes on light microscopy differed for individual cell lines based on surface EGFR expression. After incubation with panitumumab-IR700 (10 μg/mL) for 1 hour, individual 35-mm dishes of UMUC-5, 5637, and UMUC-3 were irradiated with increasing amounts of NIR (0, 4, 20, 32, 64, and 100 J/cm²). About 1 hour later, light microscopy images were collected from these plates followed by addition of CellTiter-Glo to detect the viability of cells. Clear morphologic changes were observed in UMUC-5 at 4 J/cm² (Supplementary Fig. S3A), in 5637 at 32 J/cm² (Supplementary Fig. S3C) and in UMUC-3 at 64 J/cm² (Supplementary Fig. S3E). These changes include loss of stellate morphology with cells becoming more round and turgid. There is a concomitant loss of viability in these cells lines at the above mentioned NIR amounts as seen from Supplementary Fig. S3B, S3D, and S3F, respectively.

The cytotoxic effect of panitumumab-IR700–based PIT on UMUC-5 cells was examined using LIVE/DEAD reagent. This reagent specifically stains dead cells that are subsequently analyzed by flow cytometry. After incubation with panitumumab-IR700 for one hour, UMUC-5 cells were exposed to 4 J/cm² of NIR. Within 30 minutes post-NIR, more than 60% cells of the panitumumab-IR700–treated cells were stained with LIVE/DEAD reagent, indicating rapid cell death (Fig. 2A). In a blocking condition to confirm specificity of panitumumab-IR700 for EGFR, cells were incubated initially with excess panitumumab and then with panitumumab-IR700. Subsequent NIR resulted in only 20% cell death (similar to the cytotoxicity seen with panitumumab alone), thus demonstrating the specificity of panitumumab-IR700–based PIT for EGFR. In addition, UMUC-5 cells treated only with 4 J/cm² of NIR, panitumumab-IR700 with no NIR, and panitumumab with NIR demonstrated little to no cytotoxicity. Similarly, when 5637 cells were treated with the same conditions followed by irradiation with 32 J/cm² NIR (Supplementary Fig. S4). Cell death (41%) was only observed in the cells treated with panitumumab-IR700 + 32 J/cm² of NIR. No cell death was seen in any other conditions. These data demonstrate that both panitumumab-IR700 and its excitation by NIR are required for cell death. Panitumumab, even when incubated with cells for 96 hours, demonstrated little to no effect on the survival of the bladder cancer lines tested (UMUC-5, TCCSUP, 5637 and RT4; Supplementary Fig. S5A). Similarly, IR700 dye alone did not induce phototoxicity in UMUC-5 cells, except at a very high concentration of 1 μmol/L and at a very high NIR dose of 100 J/cm² (Supplementary Fig. S5B). Our preliminary work reveals that most experiments require a dose of 10 μg/mL of panitumumab-IR700, which generally contains approximately 100 nmol/L of IR700, associated with the mAb. This dose of dye demonstrated no overt toxicity at either 4 J/cm² or 100 J/cm².

The IC₅₀ of panitumumab-IR700 NIR-PIT for UMUC-5 cells was calculated by incubating the cells with increasing concentrations of panitumumab-IR700 followed by irradiation with 4 J/cm² of NIR. The IC₅₀ for UMUC-5 cells under these conditions was 4.7 nmol/L for panitumumab-IR700 (Fig. 2B, i). Other bladder cancer cell lines expressing less EGFR compared with UMUC-5, such as TCCSUP, 5637 and T24, did not show any cell death at 4 J/cm² of NIR. The minimum amount of NIR required for demonstrating panitumumab-IR700–based phototoxicity was 64 J/cm² for TCCSUP, T24 and UMUC-3 and 32 J/cm² for 5637 (Supplementary Fig. S3). The IC₅₀ of panitumumab-IR700 for 5637 at 32 J/cm² is 5.8 nmol/L, TCCSUP cells at 64 J/cm² is 4 nmol/L, and for UMUC-3 at 64 J/cm² is 108 nmol/L (Fig. 2B, ii, iii, & iv, respectively) indicating there is an inverse relation between the surface expression of EGFR and the amount of NIR required to achieve the nanomolar range IC₅₀s (Fig. 2C). As a control, normal urothelial cells, UPS54 were treated with 10 μg/mL of panitumumab-IR700 and 100 J/cm² of NIR. These cells do not show any cell death 6 hours post-NIR (Supplementary Fig. S6).

UMUC-5 cells treated with panitumumab-IR700 were irradiated with NIR of 4 J/cm². Twenty minutes post-NIR, a single-cell suspension of these cells was treated with FITC Annexin V and FxCycle Violet and then analyzed by flow cytometry. UMUC-5 cells treated with the apoptosis inducer, staurosporine (1 μmol/L), for 3 hours were used as a positive control for adjusting the voltages for FSC, SSC, FITC, and Pacific Blue filters (Supplementary Fig. S7). As seen from Fig. 3A, about 50% of the panitumumab-IR700–treated cells localize to the top right quadrant (high Annexin V/high FxCycle Violet staining), representing cells in the late apoptosis/early necrosis stage of cell death. Only 35% of cells remained in the bottom left quadrant that are live cells with low Annexin V and low FxCycle violet. No treatment, panitumumab-IR700 without NIR, and panitumumab or IR700 dye with or without NIR result in some inadvertent cell death due to the use of trypsin for preparation of single-cell suspension. Unfortunately, other methods of single-cell preparation, such as accutase or EDTA, are too mild for these cells to prepare single-cell suspension, whereas cell scraping resulted in far more incidental cell death. However, most of the cells treated with these conditions remain in the bottom left (low Annexin V FITC/low FxCycle) quadrant. This again confirms the previous experiments demonstrating cell death only in the condition of panitumumab-IR700 activated by NIR. Of note, cell death in PIT is rapid as Annexin V/FxCycle staining at 60 minutes reveals very few events on forward scatter and side scatter plots of flow cytometry due to rapid cell death (Supplementary Fig. S8). Such rapid cell death suggests necrosis.

To rule out the induction of apoptosis as the primary mechanism of cell death in PIT-treated cells, the Caspase-Glo 3/7 assay was performed with UMUC-5, 5637, and UMUC-3 cells. This assay detects cleaved caspase-3 and -7 and is a direct measure of the end-products of apoptosis. As seen from Fig. 3B (ii), none of the cells treated with panitumumab-IR700 and cytotoxic NIR doses demonstrated any caspase-3/7 activation over and above untreated controls or non-NIR–treated controls (Fig. 3B, i)). The known apoptosis inducer staurosporine (1 μmol/L) was used as a positive control. This treatment resulted in the production of significantly elevated levels of cleaved caspase-3/7 which were inhibited by the cell-permeable caspase-specific inhibitor Z-VAD-FMK. The addition of Z-VAD-FMK did not alter the levels of cleaved caspase-3/7 in PIT-treated (panitumumab-IR700 + NIR) cells.

To establish necrotic cell death definitively, TEM was performed. TEM is considered the “gold standard” for determining the mechanism of cell death (18). As seen in Fig. 3C, the panitumumab-IR700–based PIT-treated cells on the right (Fig. 3C, ii) are considerably larger compared with normal cells (Fig. 3C(i)). The plasma membrane is mostly disintegrated in these cells. The integrity of the nuclear membrane is compromised; nucleoplasm and cytoplasm appear to be depleted of all material. Although some mitochondria are surprisingly normal, most other organelles appear to be swollen. These are the classic findings of late necrotic cells, thus confirming that panitumumab-IR700-NIR-PIT causes necrotic cell death.

Reactive oxygen species (ROS) were measured using a cell-permeable reagent called 2′,7′-dichlorofluorescein diacetate (DCFDA). DCFDA is a fluorogenic dye that gets oxidized to 2′,7′-dichlorofluorescin, a highly fluorescent moiety detected by fluorescent spectroscopy, in the presence of ROS. As seen from Fig. 4A, both panitumumab-IR700 and IR700 dye-treated UMUC-5 cells previously incubated with DCFDA demonstrate a 3- to 4-fold increase in fluorescence just 5 minutes post-NIR activation, suggesting the production of a large excess of ROS. Trolox, a water-soluble, ROS-specific scavenger, was able to inhibit this ROS generation. Moreover, ROS were specifically generated only when IR700 was present and NIR was delivered. As a result, panitumumab-IR700 and IR700-treated cells in the absence of NIR displayed baseline levels of ROS production similar to untreated or panitumumab-treated UMUC-5 cells with or without NIR (Supplementary Fig. S9A and S9B).

Singlet oxygen species (SOS) were measured using Singlet Oxygen Sensor Green (SOSG) reagent. In the presence of SOS, this reagent emits green fluorescence similar to fluorescin. As shown in Fig. 4B, UMUC-5 cells preincubated with SOSG, and treated with either panitumumab-IR700 or IR700 produce 4-fold more SOS in the presence of NIR compared with baseline levels (non-NIR conditions or untreated or panitumumab only treated cells). Again, this SOS production happens rapidly in the first 5 minutes post-NIR treatment and can be quenched to baseline levels in the presence of the SOS-specific scavenger, sodium azide (NaN₃).

When cell survival assay was carried out in the presence of trolox and sodium azide (specific scavengers of ROS and SOS, respectively), only sodium azide was able to rescue UMUC-5 cells from panitumumab-IR700–based PIT. Therefore, although both ROS and SOS are produced during panitumumab-IR700–based PIT, SOS likely contributes more significantly toward cell death than ROS. Furthermore, although IR700 dye alone produces both ROS and SOS, the production in the absence of targeting the cell membrane of EGFR-expressing cancer cells does not cause any cell death (Fig. 4C).

About 3 million UMUC-5 cells resuspended in 50% Matrigel in 1× PBS were injected subcutaneously in the right thigh of female athymic Nu/Nu mice to generate UMUC-5 xenografts. Seven days postinjection, xenograft tumors of size approximately 50–100 mm³ were observed. The H&E staining of these tumors indicated squamous morphology retained in UMUC-5 tumors (Fig. 5A, i)). Moreover, staining of hEGFR indicated UMUC-5 tumors maintained high hEGFR expression in xenografts similar to cells in vitro (Fig. 5A(ii)). To elucidate the effect of panitumumab-IR700 on xenografts, athymic nude mice bearing UMUC-5 tumors were injected with 120 μg of panitumumab-IR700. One group of mice (n = 10) was treated with 100 J/cm² and 50 J/cm² of NIR 24 and 48 hours post- panitumumab-IR700 treatment, respectively. The NIR treatment was given focally in the area of subcutaneous tumor by placing the light source just above the tumor. The other group of mice (n = 10) was not treated with any NIR. As seen from Fig. 5B, the growth of NIR-treated tumors is attenuated in comparison with non-NIR–treated tumors. In fact, three of ten NIR-treated tumors regressed completely and therefore could not be measured by external calipers during the period of experiment (Supplementary Fig. S10B). On the other hand, non-NIR–treated tumors did not slow down their growth as seen in Supplementary Fig. S10A, where tumor volumes of individual mice are plotted as a function of time. The tumor weights measured at the end of the experiment showed that non-NIR–treated tumors were significantly larger (median weight – 0.5 g) than NIR-treated tumors (median weight – 0.2 g; two-tailed P test, Mann–Whitney analysis, P = 0.0077; Fig. 5C). On the other hand, when the same experiment was repeated with UMUC-3, a cell line with considerably lower EGFR surface expression, there was no difference in the growth of tumors irradiated or nonirradiated with NIR (Supplementary Fig. S11). These data highlight both the importance of expression of the targeted cell surface antigen as well as the specificity of this treatment.

We demonstrate a novel therapy in bladder cancer cell lines called molecular targeted PIT (14). In this method, a humanized mAb against a specific target is conjugated to a photoabsorbing dye and this conjugate is then activated by NIR after sufficient time to allow cell binding. IR700 functions as a PA instead of a traditional photosensitizer because, in its nonconjugated state, it does not result in generalized cytotoxicity when activated by NIR. In our study, EGFR was chosen as the target given its high rate of amplification in bladder cancer. However, this therapy does not require addiction to the EGFR pathway; instead, it merely requires relatively high surface expression of EGFR in tumor cells as compared with normal cells. The lack of addiction to the EGFR pathway (due to downstream mutations and alternate signaling pathways) and the inability to reproduce ADCC in the absence of an immune system are likely reasons for the relative inefficacy of panitumumab alone in our studies (19–21). We demonstrate successful conjugation of IR700 to panitumumab and the resulting conjugate induces cell death in EGFR-expressing cell lines at low doses of NIR. For instance, these light doses do not produce significant thermal effects on the treated surface. Furthermore, cell death only occurs in cells that express EGFR on their surfaces and only in the presence of NIR and the panitumumab–IR700 conjugate.

We also demonstrated that cell death occurs by necrosis. First, rapid cell death within 60 minutes of NIR suggests necrosis. Second, the FITC Annexin V – FxCycle violet assay demonstrates localization of treated cells to the late apoptosis/early necrosis quadrant at 20 minutes. At 20 minutes, this is more consistent with early necrosis. At the same time there was no generation of cleaved caspase-3/7. Finally, TEM shows preservation of mitochondria but considerable disruption of the plasma membrane. Most lipophilic photosensitizers associated with PDT (e.g., Photofrin) localize to the mitochondria and induce apoptosis through mitochondrial disruption, release of cytochrome c, and activation of the intrinsic pathway of apoptosis (22). Because of their lipophilicity, they tend to enter both normal and neoplastic cells leading to collateral damage. Conversely, photosensitizers localized to the plasma membrane are likely to cause necrosis. The IR700 dye is hydrophilic and therefore does not freely enter cells. However, the antibody-conjugated IR700 binds to specific cell surface receptors and when exposed to NIR, rapid cell damage ensues. The exact mechanism of action of PIT is still uncertain. We demonstrate that NIR-PIT produces ROS and SOS suggesting that oxidative damage may cause degradation of membrane lipids leading to rupture and necrosis, although we did not detect any oxidative changes on major unsaturated lipid molecules after NIR-PIT so far. IR700 by itself produces a large amount of SOS but does not cause cell death likely because the molecule is sufficiently far away from the cell membrane and SOS have an extremely transient lifespan of <0.04 microseconds in biologic systems and a very narrow radius of activity at <0.02 micrometers (23). However, panitumumab-IR700 bound to EGFR and activated by NIR releases SOS at the plasma membrane to induce localized necrosis. Finally, singlet oxygen also has a direct cytotoxic effect on local tumor cells and vasculature and can attract dendritic cells and neutrophils thereby initiating an acute inflammatory response (24). Free radical scavengers reduced the effect of NIR-PIT, although this has not been universally true (25, 26).

The bladder is a well-suited organ for light therapy given the easy accessibility via cystoscopy and the need to treat the entire organ due to the multifocal nature of bladder cancer. Kelly and colleagues first demonstrated photodynamic destruction of bladder cancers implanted in mice using a hematoporphyrin derivative (HPD) in 1975 (27). Subsequently, they conducted the first human clinical trial showing preferential localization of HPD in malignant and premalignant urothelium with tumor destruction upon illumination of these areas (28). However, HPD fluorescence was also present in normal urothelium. There is no difference in light penetration between benign and malignant bladder tissue; therefore, toxicity from therapy was a major limitation (29). In addition, HPD was given intravenously that can result in systemic accumulation and cutaneous phototoxicity for many weeks. Despite the introduction of newer PS, such as 5-aminolevulinic acid (5-ALA), Photofrin I, Photofrin II, and hexaminolevulinate (HAL), toxicities such as irritative urinary symptoms, skin photosensitivity, and bladder contracture still occurred in clinical trials. Furthermore, late responses to PDT in these trials were quite variable ranging from 11% to 64% (30). Although an initial report with a novel PS called Radachlorin is promising with a recurrence-free rate of 64.4% at 2 years with minimal toxicity (30), a previous report with a similar chlorine-containing compound reported a case of vesicoenteric fistula as a complication (31). Therefore, most prior phototherapies have failed due to their side effects.

EGFR has been previously targeted using the EGFR-humanized chimeric mAb C225 (cetuximab) conjugated to a benzoporphyrin derivative (verteporfin) successfully in cancer cell lines (32). However, the biodistribution of this conjugate has made it difficult to treat bladder cancers selectively while avoiding generalized phototoxicity. In our current study, we also used an anti-EGFR humanized mAb, panitumumab, but applied it to bladder cancer cell lines. What is different about the current work is the use of a phthalocyanine dye called IR700 instead of a porphyrin derivative. Unlike 5-ALA or the hematoporphyrin derivatives, IR700 is completely water soluble; hence, free, unconjugated dye gets rapidly excreted in urine without accumulation within the body resulting in no photosensitizing effect (Supplementary Fig. S5B and S5C; ref. 14). Compared with conventional photosensitizers, it has a greater than 5-fold higher extraction coefficient that allows rapid elimination from the blood/plasma (14). Furthermore, it is not toxic to normal tissues or phototoxic upon NIR and so it is theoretically safer than traditional photosensitizers. The panitumumab-IR700 conjugate differs from PDT in that internalization of the conjugate is not required to induce cell death. Finally, as IR700 uses a higher wavelength of light, it can penetrate tissue more deeply than the wavelengths of light needed to activate traditional PS. For example, light at 693 nm penetrates 40% deeper than light at 633 nm (P < 0.002; ref. 29). NIR at 693 nm has a potential depth of penetration of 1–2 cm. As the average bladder wall thickness is 3.35 mm, we believe that this may be a viable therapy for bladder cancer in the future (33).

In this preclinical article, we evaluated the effect of PIT in subcutaneous xenografts. Although the intravesical approach is ideal for future use in bladder cancer therapy, preclinical intravesical therapy is not as straight forward. Although we attempted direct inoculation of UMUC-5 and other cell lines intravesically in nude mice to establish orthotopic intravesical human tumors, the tumor formation rate was <10% and in animals where tumor developed, it spontaneously regressed in certain cases. Therefore, we did not feel this was a reliable model to pursue and hence the work in this article focuses on subcutaneous tumors. The best published approaches of inducing orthotopic tumors involve either a carcinogen model or implantation of isogenic MB49 cells in C57/BL6 mice (34). However, both of these approaches may lead to mouse-EGFR (mEGFR)-bearing tumors and we cannot test the panitumumab–IR700 conjugate in such models because panitumumab does not cross-react with mEGFR (35). In the future, we will embark on orthotopic models to directly evaluate intravesical therapy with PIT but this initial work used subcutaneous xenografts to prove that human tumors can be treated.

The prevalence of EGFR expression in bladder cancer cell lines makes it a viable therapy in at least basal-like tumors. Molecular stratification reveals that the basal phenotype, which is enriched for EGFR activation/amplification, can be found in almost 25% of bladder tumors (16). Furthermore, squamous differentiation is extremely common in UCC and increases with T stage and can be as high as 60%–80% in T3 tumors (36). Although SCC of the bladder is often locally advanced, it does not metastasize as often as urothelial cancer, potentially allowing for localized therapy as well (12). Perhaps our strategy may be applicable to any bladder tumor with a basal phenotype and now this can be assessed using BASE47, a gene set predictor, that can distinguish between basal and luminal molecular subtypes of bladder cancer (37). Moreover, this strategy can be extended to other cell surface proteins overexpressed or aberrantly expressed on bladder cancer cells, such as FGFR-3, ErbB-2, Nectin-4, Muc-1, and CEA to name a few. UMUC-3, a low EGFR-expressing cell line, did not respond well to panitumumab-IR700–mediated PIT both in vitro and in xenografts. To treat such a cell line or cancer, the use of other cell surface antigens or a cocktail of several different mAbs conjugated to a PA like IR700 will need to be employed. This is the goal of a future project. However, other targets and combinations are available. For example, a urothelial cancer cell line with the lowest level of EGFR surface expression, RT4, actually has the highest ErbB-2 expression and our preliminary data demonstrate that it may be amenable to an anti-ErbB-2-IR700 approach (M.R. Siddiqui; unpublished observations).

In summary, we describe a proof-of-concept study of molecular targeted PIT in bladder cancer cells using a panitumumab—IR700 conjugate. Given the high amplification rate of EGFR in most bladder cancers, it may provide a selective and novel therapy for non-muscle –invasive bladder cancer. We are actively conducting orthotopic murine models with this therapy and hope that our efforts translate into a clinically viable strategy for bladder cancer patients in the future.

No potential conflicts of interest were disclosed.

Conception and design: R. Railkar, Q.Q. Li, S. Vourganti, P.L. Choyke, H. Kobayashi, P.K. Agarwal

Development of methodology: R. Railkar, Q.Q. Li, T. Sanford, S. Vourganti, S.J. Brancato, P.L. Choyke, H. Kobayashi, P.K. Agarwal

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Railkar, L.S. Krane, Q.Q. Li, M.R. Siddiqui, D. Haines, S. Vourganti, S.J. Brancato, P.K. Agarwal

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Railkar, L.S. Krane, Q.Q. Li, T. Sanford, D. Haines, S. Vourganti, S.J. Brancato, P.L. Choyke, P.K. Agarwal

Writing, review, and/or revision of the manuscript: R. Railkar, Q.Q. Li, T. Sanford, D. Haines, S. Vourganti, S.J. Brancato, P.L. Choyke, H. Kobayashi, P.K. Agarwal

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Railkar, D. Haines, P.L. Choyke, P.K. Agarwal

Study supervision: P.L. Choyke, H. Kobayashi, P.K. Agarwal

We thank Ms. Catherine Wells for maintenance of the animal facility and technical support with animal work. Special thanks to Ms. Donna Butcher from Pathology/Histotechnology Laboratory, NCI, for staining of TMAs. We thank Dr. Ulrich Baxa from Electron Microscopy Core Facility of NCI for help with TEM.

This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research (NIH intramural grant number ZIABC011458-02; to P.K. Agarwal).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.