Noninvasive Detection of Bladder Cancer in an Orthotopic Murine Model With Green Fluorescence Protein Cytology

0022-5347/03/1703-0975/0 THE JOURNAL OF UROLOGY® Copyright © 2003 by AMERICAN UROLOGICAL ASSOCIATION

Vol. 170, 975–978, September 2003 Printed in U.S.A.

DOI: 10.1097/01.ju.0000073209.65128.c1

NONINVASIVE DETECTION OF BLADDER CANCER IN AN ORTHOTOPIC MURINE MODEL WITH GREEN FLUORESCENCE PROTEIN CYTOLOGY MOTOYSOHI TANAKA, JASON R. GEE, JORGE DE LA CERDA, CHARLES J. ROSSER, JAIN-HUA ZHOU, WILLIAM F. BENEDICT AND H. BARTON GROSSMAN*, † From the Departments of Urology and Genitourinary Medical Oncology, University of Texas M. D. Anderson Cancer Center, Houston, Texas

ABSTRACT

Purpose: Orthotopic models of bladder cancer mimic the normal microenvironment and provide an opportunity to study new therapies for superficial bladder cancer. The use of green fluorescent protein (GFP) transduced cells provides a sensitive way of monitoring this disease. We investigated whether examining voided urine for GFP expressing cells would indicate the presence of GFP producing tumors in an orthotopic bladder tumor model in nude mice. Materials and Methods: The human bladder cancer cell lines KU-7, UM-UC-3 and UM-UC-14 were used. GFP transductants were generated after transfection with pEGFP-N3, followed by G418 selection. After the cells were inoculated in an orthotopic model of superficial bladder cancer voided urine was collected on slides weekly for 3 weeks and observed for GFP expressing cells by fluorescence microscopy. Bladder tumor imaging for GFP was performed in surgically exposed bladders to determine the tumor incidence. Results: KU-7 GFP cells produced tumors in all 16 mice on whole bladder GFP imaging. UM-UC-3 and UM-UC-14 GFP cells produced tumors in 8 of 12 (67%) and 18 of 25 (72%) mice, respectively. The rate of GFP positive cells in spontaneously voided urine varied by cell line and increased with time but it was generally less than the rate of detection by whole bladder GFP imaging. All mice with GFP expressing cells in the urine had GFP expressing bladder tumors. Conclusions: Examining urine for GFP expressing cells is less sensitive than imaging surgically exposed bladders but it is 100% specific. KEY WORDS: bladder; bladder neoplasms; mice, nude; urine; fluorescent protein tracing

In 2002 more than 55,000 new cases of bladder cancer were detected in the United States.1 Superficial bladder cancers comprise 80% of these neoplasms and they are associated with a high rate of tumor recurrence despite treatment consisting of transurethral resection and intravesical chemotherapy or immunotherapy.2 Tumor progression within 5 years occurs in up to 30% to 40% of patients with high risk superficial disease and long-term followup demonstrates that a third die of cancer.3 The prognosis for patients with advanced tumors is poor even for those who receive aggressive multimodal therapy with only 20% to 40% with advanced bladder cancer surviving 5 years. Novel treatments targeted at aggressive superficial disease are needed to prevent progression and improve the prognosis for patients with bladder cancer. Intravesical chemotherapy and immunotherapy are widely used in the treatment of bladder cancer. Intravesical gene therapy with the p53 tumor suppressor gene has been tested in a phase I bladder cancer clinical trial at our institution.4 However, effective gene delivery has been difficult. Another study of intravesical therapy with adenoviral p53 demonstrated that transduction efficiency is modest.5 An animal

model of bladder cancer resembling human disease would facilitate the development of novel methods of therapy. Ectopic subcutaneous implantation of cancer cells in athymic mice is often used to test new treatment modalities. However, cells implanted ectopically are not in their usual microenvironment and they behave differently than where they are implanted orthotopically.6 Additional experimental advances have resulted from the use of green fluorescent protein (GFP) transduced cells. Yang et al reported that GFP expressing human prostate cancer cells that were orthotopically implanted metastasized to bone and could be detected by whole body imaging for GFP.7 Although several orthotopic animal models of bladder cancer have been reported, the tumor incidences for such models vary and may not be reproducible.8 –10 We have previously established a reproducible intravesical model of human bladder cancer in athymic mice.11 Using 2 different human bladder cancer cell lines we established models of superficial bladder cancer and carcinoma in situ. We have also shown that GFP transduced human bladder cancer cells produce tumors that can be detected in vivo by exposing the bladder and imaging for GFP.12 We now report a noninvasive technique for detecting GFP transduced tumors in this nude mouse model.

Accepted for publication March 28, 2003. Supported by National Cancer Institute Grants CA91846 and CA16672, and Bladder SPORE, the Retina Research Foundation and Tobacco Settlement, as appropriated by the Texas State Legislature, and American Foundation for Urologic Disease. MATERIALS AND METHODS * Requests for reprints: Department of Urology, Box 446, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Bladder cancer cell lines and GFP transductants. The huHouston, Texas 77030. man bladder cancer cell lines KU-7,13 UM-UC-314 and UM† Financial interest and/or other relationship with Anthra Phar15 maceuticals, UroCor, Astra Zeneca, Paladin Laboratories, Pharma- UC-14 were maintained in Dulbecco’s modified Eagle’s mecia and PhotoCure. dium containing 10% fetal calf serum. The GFP expression 975

976

ORTHOTOPIC MODEL OF HUMAN BLADDER CANCER Tumor incidence Cell Line

No. Mice/No. Evaluable*

No. With Tumor (%)

KU-7 GFP 18/16 16 (100) UM-UC-3 GFP 12/12 8 (67) UM-UC-14 GFP 28/25 18 (72) * Five animals died within a few days of the procedure.

vector pEGFP-N3 (Clontech, Palo Alto, California) was used. GFP expressing cells (KU-7 GFP, UM-UC-3 GFP and UMUC-14 GFP) were generated by G418 selection after transfection with pEGFP-N3, as described previously.12 Transfection of the pEGFP-N3 vector into KU-7, UM-UC-3 and UMUC-14 human bladder cancer cells was performed using Fugene (Roche, Indianapolis, Indianapolis). At 48 hours after transfection the cells were subcultured into selective culture medium containing 500 mg/ml G418. GFP expressing cells were isolated and cloned. Selected clones had a high level of GFP expression, were consistently detectable by fluorescence and had a growth rate similar to that of parental cells. Orthotopic implantation of GFP transductant cells. An orthotopic model of bladder cancer was used, as previously described.11 Briefly, female athymic nude mice were anesthetized by intraperitoneal injection of sodium pentobarbital. A 24 gauge catheter was inserted into the bladder transurethrally and 100 ␮l 0.2% trypsin in 0.02% ethylenediaminetetraacetic acid were infused and retained in the bladder for 30 minutes. After trypsinization the bladder was catheterized and washed with phosphate buffered saline. Subsequently a 100 ␮l suspension of serum-free medium containing 1 ⫻ 107 KU-7 GFP, UM-UC-3 GFP or UM-UC-14 GFP cells was instilled into the bladder. A purse-string suture was placed around the urethra. Three hours later the suture was removed and the bladder was evacuated by spontaneous voiding. Fluorescence analysis of voided urine. To determine whether voided urine would be useful for monitoring GFP expressing bladder tumor formation we analyzed spontaneously voided urine from mice whose bladders were instilled with GFP expressing cells. Voided urine was obtained weekly for 3 weeks. A mouse was hand held and spontaneously voided urine was collected onto a glass slide, which was promptly examined for GFP expressing cells using a fluorescence microscope. A slide containing a single fluorescing cell was scored as positive.

Incidence of bladder tumor. Whole body imaging of GFP was performed to analyze the tumor incidence with a light source that fluoresces GFP (Lightools Research, Encinitas, California) or with a stereo microscope (Leica, Bannockburn, Illinois) equipped with a spot camera.12 Fluorescence was determined visually. Two mice were sacrificed weekly and the bladders were exposed and observed for GFP signals, as described. Tumor incidence was defined as the identification of GFP expressing tumors in surgically exposed bladders. Three weeks following the intravesical administration of cancer cells the tumor incidence was assessed by GFP fluorescence and all mice were sacrificed. Following fixation with 10% buffered formalin the bladders were embedded in paraffin, serially sectioned and stained with hematoxylin and eosin. RESULTS

Tumor incidence in the orthotopic model. We previously reported that KU-7 tumors instilled intravesically formed tumors with a 100% tumor incidence.11 In the current study GFP expressing bladder tumors could occasionally be visualized through the skin by illumination with blue light. However, the sensitivity of this technique was limited. Therefore, tumor assessment was determined in surgically exposed bladders (see table). KU-7 GFP cells were successfully instilled in 16 mice. In all of these mice GFP expressing tumors developed. UM-UC-3 GFP cells were successfully instilled into 12 mice and UM-UC-14 GFP cells were successfully instilled into 25. As determined by visualization of fluorescence in surgically exposed bladders, the tumor incidence was 67% for UM-UC-3 GFP and 72% for UM-UC-14 GFP (fig. 1). These data were reproducible and determined in 3 independent experiments. Fluorescence analysis of urine. We analyzed urine for GFP expressing cells to determine whether this noninvasive technique could detect bladder tumors. Urine was obtained weekly for 3 weeks. Although the incidence of GFP expressing tumor cells in urine samples taken 1 week after intravesical inoculation was 10 of 16 KU-7 GFP (63%) instilled mice and zero in UM-UC-3 GFP and UM-UC-14 GFP instilled mice, the rate of GFP expressing cells in urine at week 3 increased to 100%, 44% and 47% in KU-7 GFP, UM-UC-3 GFP and UM-UC-14 GFP tumor bearing mice, respectively. Figure 2 shows the incidence of fluorescing cells in the urine

FIG. 1. GFP expressing bladder tumor images of KU-7 GFP, UM-UC-3 GFP and UM-UC-14 GFP bladder tumors 1 to 3 weeks after orthotopic tumor cell inoculation. Reduced from ⫻4.

ORTHOTOPIC MODEL OF HUMAN BLADDER CANCER

977

produce tumors of the same size as KU-7 GFP tumors at 10 days. DISCUSSION

FIG. 2. Incidence of animals with fluorescing cells in voided urine compared with direct visualization of fluorescing tumors.

as a percent of the incidence of fluorescing tumors detected by direct observation of surgically exposed bladders. None of the mice without tumors had GFP fluorescing cells in the urine. All mice with GFP positive cells in the urine had GFP expressing bladder tumors. Figure 3 shows examples of GFP positive cells in voided urine. KU-7 GFP cells produce 100% bladder tumor growth and these tumors are detectable in surgically exposed bladders at 5 to 7 days.12 UM-UC-3 GFP and UM-UC-14 GFP grow at a slower rate. Direct observation of fluorescing tumors in the surgically exposed bladder was more sensitive than evaluating fluorescing cells in voided urine. Importantly all mice with fluorescing cells in the urine had bladder tumors on GFP imaging of surgically exposed bladders. Histological analysis of bladder tumors. Four weeks after intravesical implantation of KU-7 GFP, UM-UC-3 GFP and UM-UC-14 GFP cells all mice were sacrificed and the bladder was removed. Large GFP expressing tumors were confirmed microscopically. However, small GFP expressing bladder tumors could not always be detected by conventional histological examination. It most likely occurred because small tumors were missed when cutting the paraffin blocks. KU-7 GFP tumors grew and produced large tumors 10 days after tumor cell inoculation. On the other hand, UM-UC-3 GFP and UM-UC-14 GFP tumors required more than 3 weeks to

Animal orthotopic models of bladder cancer provide experimental conditions relevant to the study of new treatments for superficial bladder cancer. A syngeneic orthotopic model of murine bladder cancer has yielded a high rate of tumor formation10 but this model may not be suitable for exploring gene therapy for human bladder cancer. Direct injection of tumor cells into the bladder wall is an effective orthotopic model of human bladder cancer.16 However, this model is more reflective of muscle invasive bladder cancer. In this study intravesical instillation of the GFP transduced human bladder cancer cell lines KU-7 GFP, UM-UC-3 GFP and UM-UC-14 GFP produced GFP detectable tumors in 100%, 67% and 72% of mice, respectively. Torti et al quantified GFP expression in a murine model of superficial bladder cancer using syngeneic GFP transduced bladder cancer cells.17 While it was quantitative, this assay required sacrificing the mice and excising the bladder. We recently reported a nonlethal method for detecting GFP fluorescing tumors in mice.12 A modest limitation to this model is that it requires periodic surgical exposure of the bladder to monitor tumor growth. However, a significant strength of this approach is that it can reliably monitor the individual tumor response to therapy. As reported previously, in this study imaging surgically exposed bladders provided a sensitive technique for monitoring tumor growth. We then examined whether a noninvasive technique could be used to identify the presence of GFP expressing bladder tumors. GFP expressing cells could be detected in spontaneously voided urine. While this technique was less sensitive than imaging surgically exposed bladders for GFP expressing tumors, all animals that voided GFP expressing cells had tumors in the bladder. In addition, the technique was able to identify the presence of all KU-7 GFP tumors. This technique is a noninvasive, highly specific method of documenting that tumors have developed and it can be used to select animals for therapeutic studies. CONCLUSIONS

The instillation of human GFP expressing bladder cancer cell lines provides an excellent model for studying new treatments for superficial bladder cancer. The rate of tumor formation varies according to the cell line used. Direct visual-

FIG. 3. Urine cytology in animals bearing KU-7 GFP (A), UM-UC-3 GFP (B) and UM-UC-14 GFP (C) tumors. Top row, fluorescence photomicrograph. Bottom row, phase contrast photomicrograph. Reduced from ⫻200.

978

ORTHOTOPIC MODEL OF HUMAN BLADDER CANCER

ization of surgically exposed bladders provides the most sensitive method of detecting these GFP expressing tumors. However, examining voided urine for GFP expressing cells provides a less sensitive but 100% specific noninvasive method for determining that tumor is present in the bladder. The human bladder cancer cell line KU-7 was provided by Dr. M. Tachibana, Keio University, Japan. REFERENCES

1. Jemal, A., Thomas, A., Murray, T. and Thun, M.: Cancer statistics, 2002. CA Cancer J Clin, 52: 23, 2002 2. Grossman, H. B.: Superficial bladder cancer: decreasing the risk of recurrence. Oncology, 10: 1617, 1996 3. Cookson, M. S., Herr, H. W., Zhang, Z.-F., Soloway, S., Sogani, P. C. and Fair, W. R.: The treated natural history of high risk superficial bladder cancer: 15-year outcome. J Urol, 158: 62, 1997 4. Pagliaro, L. C.: Gene therapy for bladder cancer. World J Urol, 18: 148, 2000 5. Kuball, J., Wen, S. F., Leissner, J., Atkins, D., Meinhardt, P., Quijano, E. et al: Successful adenovirus-mediated wild-type p53 gene transfer in patients with bladder cancer by intravesical vector instillation. J Clin Oncol, 20: 957, 2002 6. Nakajima, M., Morikawa, K., Fabra, A., Bucana, C. D. and Fidler, I. J.: Influence of organ environment on extracellular matrix degradative activity and metastasis of human colon carcinoma cells. J Natl Cancer Inst, 82: 1890, 1990 7. Yang, M., Jiang, P., Sun, F. X., Hasegawa, S., Baranov, E., Chishima, T. et al: A fluorescent orthotopic bone metastasis model of human prostate cancer. Cancer Res, 59: 781, 1999 8. Huland, E., Arndt, R. and Huland, H.: Human transitional cell carcinoma in the NMRI-nu/nu mouse bladder: a new animal model for the in vivo use of monoclonal antibodies and cyto-

toxic agents. Cancer Res, 46: 2488, 1986 9. Ahlering, T. E., Dubeau, L. and Jones, P. A.: A new in vivo model to study invasion and metastasis of human bladder carcinoma. Cancer Res, 47: 6660, 1987 10. Gunther, J. H., Jurczok, A., Wulf, T., Brandau, S., Deinert, I., Jocham, D. et al: Optimizing syngeneic orthotopic murine bladder cancer (MB49). Cancer Res, 59: 2834, 1999 11. Watanabe, T., Shinohara, N., Sazawa, A., Harabayashi, T., Ogiso, Y., Koyanagi, T. et al: An improved intravesical model using human bladder cancer cell lines to optimize gene and other therapies. Cancer Gene Ther, 7: 1575, 2000 12. Zhou, J. H., Rosser, C. J., Tanaka, M., Yang, M., Baranov, E., Hoffman, R. M. et al: Visualizing superficial human bladder cancer cell growth in vivo by green fluorescent protein expression. Cancer Gene Ther, 9: 681, 2002 13. Tachibana, M., Miyakawa, A., Uchida, A., Murai, M., Eguchi, K., Nakamura, K. et al: Granulocyte colony-stimulating factor receptor expression on human transitional cell carcinoma of the bladder. Br J Cancer, 75: 1489, 1997 14. Grossman, H. B., Wedemeyer, G., Ren, L., Wilson, G. N. and Cox, B.: Improved growth of human urothelial carcinoma cell cultures. J Urol, 136: 953, 1986 15. Liebert, M., Wedemeyer, G., Stein, J. A., Washington, R. W., Jr., Van Waes, C., Carey, T. E. et al: The monoclonal antibody BQ16 identifies the alpha 6 beta 4 integrin on bladder cancer. Hybridoma, 12: 67, 1993 16. Dinney, C. P. N., Fishbeck, R., Singh, R. K., Eve, B., Pathak, S., Brown, N. et al: Isolation and characterization of metastatic variants from human transitional cell carcinoma passaged by orthotopic implantation in athymic nude mice. J Urol, 154: 1532, 1995 17. Torti, S. V., Golden-Fleet, M., Willingham, M. C., Ma, R., Cline, M., Sakimoto, Y. et al: Use of green fluorescent protein to measure tumor growth in an implanted bladder tumor model. J Urol, 167: 724, 2002