An orthotopic xenograft model for high-risk non-muscle invasive bladder cancer in mice: influence of mouse strain, tumor cell count, dwell time and bladder pretreatment

The human BCa cell line UM-UC-3 (ATCC CRL-1749; ATCC, Rockville, MD, USA; bought in 2011) was cultured in minimum essential medium with 10% fetal calf serum and 1% non-essential amino acids (all from Life Technologies, Karlsruhe, Germany). Cells were cultured at 37 °C in a humidified atmosphere containing 5% CO₂. To enable non-invasive visualization of tumor growth, UM-UC-3 cells were transduced with a retroviral pRevCMV-c-Luc vector containing the firefly luciferase gene (LUC+) [16] and a hygromycin B resistance cassette [17]. Packaging of viral particles and transductions were performed as previously described [18]. Afterwards, single UM-UC-3^LUC cells were seeded into 96-well plates, cultured and selected by adding 300 μg/ml hygromycin B to the culture medium. Depending on the growth and apoptosis pattern as well as on the strength of the luminescence signal the UM-UC-3^LUCK1 clone was chosen for experiments.

For monitoring the cells, in vitro luciferase activity was measured regularly with Luciferase Assay System according to the manufacturer’s instructions (Promega, Mannheim, Germany). Furthermore, luciferase activity was measured after pouring different cell counts into a 96-well flat clear bottom black polystyrene TC-treated microplate using the In-Vivo Xtreme imaging system (Bruker BioSpin MRI GmbH, Ettlingen, Germany). In doing so, at least 2 wells were left blank between the measuring points. Five microliters D-luciferin (15 mg/ml in PBS; PerkinElmer, Rodgau, Germany) were added to 200 μl cell solution directly before imaging.

Protein separation and subsequent Western blotting were performed as described previously [19]. Membranes were probed with primary antibodies against EGFR (1:1000; EGF Receptor Antibody #2232; Cell Signaling, Danvers, MA, USA) and β-actin (1:50,000; clone AC-74; Sigma, St. Louis, Missouri, USA); the latter served as a loading control. The secondary polyclonal swine anti-rabbit immunoglobulin HRP-linked antibody (1:1000; P0217; Dako Deutschland GmbH, Hamburg, Germany) as well as the Enhanced Chemiluminescence Kit (GE Healthcare, Freiburg, Germany) were used for visualization.

The following immunocompromised mouse strains were used in the study: athymic NMRI nude (NMRI-Foxn1 ^nu /Foxn1 ^nu ; Charles River Laboratories, Sulzfeld, Germany), BALB/c nude (BALB/cAnNRj-Foxn1 ^nu ; Janvier Labs, Saint-Berthevin Cedex, France) as well as SCID-beige (CB17.Cg-Prkdc ^scid Lyst ^bg-J /Crl; Charles River Laboratories). All three mouse strains lack T cells. In contrast to the other two mouse strains, SCID-beige mice also lack B cells, have impaired natural killer cell activity and are not hairless. General anesthesia was induced with 10% (v/v) and maintained with inhalation of 8% (v/v) desflurane (Suprane; Baxter, Unterschleissheim, Germany) in 30/10% (v/v) oxygen/air. For tumor cell instillation, fourteen weeks old female mice were used. Mouse bladders were catheterized using 24G permanent venous catheters (Becton Dickinson, Heidelberg, Germany) that were coated with petroleum jelly (Bombastus-Werke AG, Freital, Germany). To prevent bladder overexpansion residual urine was removed by massaging the bladder with thumb and trigger finger. UM-UC-3^LUCK1 BCa cells were harvested, resuspended in PBS and vital cell count was determined using the cell counting system CASY model TT (Schaerfe System, Reutlingen, Germany). The desired cell count was adjusted in a total volume of 100 μl and cells were instilled into the urinary bladders. Pretreatment of the bladder wall was performed by incubating either 100 μl of 0.1 mg/ml poly-L-lysine (PLL, Sigma-Aldrich, Steinheim, Germany) for 20 min or 100 μl 0.5% trypsin in 0.2% EDTA (Sigma-Aldrich) for 30 min. Detailed conditions for the different experiments performed are listed in Table 1. General condition of the mice was determined every day and mouse weights twice a week. Necropsy was performed in dependence on luminescence intensity and occurrence of blood in urine as well as at reduced general conditions. Whole bladders were removed for histologic examinations. Additionally, kidneys, livers and lungs of all mice were removed in experiment 6.

Tissues were fixed in 4% buffered formalin, embedded in paraffin and cut in 3 μm sections which were stained with haematoxylin and eosin (H&E) using standard techniques. All slides were reviewed by an experienced pathologist. T stage was assessed according to 7th edition of TNM Classification of Malignant Tumours [20].

Multimodal imaging of tumor growth (BLI, MRI) and functional characteristics (PET) was performed as published elsewhere [21‐24]. In brief, BLI (exposure times 1 s, 10 s, and 60 s) of anesthetized mice in prone position was performed using a dedicated small animal multimodal imaging system (In-Vivo Xtreme) 10–12 min after intraperitoneal injection of 200 μl of D-luciferin (15 mg/ml). In parallel, an X-ray image was taken from the same animals at the same position. MRI of continuously anesthetized mice was performed using a 7 T small animal imaging system BioSpin 70/30 (Bruker). Motion artifacts were reduced using a respiratory gating module (SA Instruments, Stony Brook, NY, USA). T2-weighted image series were acquired using the TRARE sequence with an echo time of 38 ms and a repetition time of 4724.9 ms at a resolution of 0.2 × 0.2 × 0.6 mm and an intersection space of 0.8 mm. PET investigations were performed as a pilot experiment in two SCID beige mice using a dedicated small animal PET/CT system (NanoPET/CT, Mediso, Budapest, Hungary). For targeting of EGFR as molecular characteristic of UM-UC-3^LUCK1 cells engrafted in the bladder the ⁶⁸Ga-radiolabeled EGFR antibody cetuximab (27 MBq; antibody modified with NOTA (1,4,7-triazacyclonane-1,4,7-triacetic acid) as ⁶⁸Ga-chelator) was transurethrally injected. After 30 min incubation and flushing with PBS (0.3 ml) for three times static scan PET acquisition was done at 1 h after administration. Afterwards, transmission CT was acquired. Then the bed-fixed animal was positioned in the MRI system and, in addition, registered T2-weighted image was acquired to get high contrast between the urine with high intensity and the tumor tissue with lower intensity.

The luciferase expressing UM-UC-3^LUCK1 clone was generated to enable non-invasive tumor detection in the mouse bladders. Besides periodic measurement of luciferase activity with the Luciferase Assay System, the in vitro luminescence intensity was quantified after D-luciferin incubation using the In-Vivo Xtreme imaging system. A strong relationship of luminescence intensity and cell count was observed with both measuring methods (Fig. 1a). A representative in vivo measurement series is shown in Fig. 1b. This SCID-beige mouse displayed first luciferase signal 15 days after tumor cell instillation. Tumor growth could be monitored for 10 days with steadily rising luminescence intensity. On day 25 the mouse was sacrificed due to high tumor load as indicated by the signal intensity.

First, the period of time between harvesting the cells and instillation into the mouse bladder was an important factor for optimal tumor growth. Although >96% of UM-UC-3^LUCK1 cells were vital 5 h after incubation in culture media, PBS and urine, respectively, no in vivo tumor growth was achieved in NMRI nude mice when the time span between harvesting the cells and instillation was 2 h or longer.

Based on literature studies, NMRI nude mice were selected for establishing an orthotopic UM-UC-3^LUCK1 BCa model. However, only 22–40% of NMRI nude mice developed a bladder tumor, although, pretreatment of the urinary bladders was performed before instillation of 2.0 × 10⁶ tumor cells for 2 h in two independent experiments (Table 1, No 1 and 2). There was no difference in tumor cell engraftment comparing the bladder pretreatment with trypsin and PLL (Table 1, No 1). The induction of lesions in the mucosa by carefully scratching with the cannula of the permanent venous catheters did not considerably improve tumor cell engraftment after PLL pretreatment (Table 1, No 2). Therefore, PLL pretreatment – without scratching with the cannula – was selected for further experiments. Exemplarily, the development of BLI signal intensities of the four tumor-bearing NMRI nude mice in experiment 1 is shown in Fig. 2.

A switch in the mouse strain to BALB/c nude and SCID-beige mice increased tumor cell engraftment to 70% and 100%, respectively (Table 1, No 3). Since BALB/c nude mice showed first BLI signal late – after 33 days on average – and with high variance, cell count for tumor cell instillation was increased to 5.0 × 10⁶ in the next experiment. In contrast, tumor cell count was decreased to 1.0 × 10⁶ in SCID-beige mice because of the fast tumor growth that is reflected by the short period of signal duration of 8.7 days on average (Table 1, No 3). The signal duration represents the possible treatment period in the evaluation of new therapeutics and should be at least two weeks.

With the adjusted cell counts the mean time until occurrence of first BLI signal remained at days 33.9 ± 18.3 (mean deviation) for BALB/c nude and at days 14.8 ± 2.1 for SCID-beige mice (Table 1, No 4). The differences in BLI signal intensity development for the individual animals of both mouse strains are shown in Fig. 3a and b. Due to this late onset of tumor growth with its high variance in BALB/c nude mice, further optimization was done using SCID-beige mice. To extend the short mean signal duration of 10.4 days in experiment 4, cancer cell count was further decreased.

Instillation of 1.0 × 10⁶ and 0.5 × 10⁶ cancer cells in SCID-beige mice in combination with a decrease in instillation time to 1 h caused a shift in the start of tumor detection to days 25.8 and 22.4 (mean values), respectively (Table 1, No 5). Exemplarily, BLI signal intensities of mice after instillation of 1.0 × 10⁶ UM-UC-3^LUCK1 cells are shown in Fig. 3c. The average signal duration remained below two weeks (Table 1, No 5). Therefore, instillation time of UM-UC-3^LUCK1 BCa cells was further decreased down to 30 min. In this manner, the mean luminescence signal duration extended to 19.6 ± 8.2 days while the mean signal start remained unchanged at 22.4 days (Table 1, No 6).

During catheterization of mice bladders an air bubble was formed in the urinary bladder due to the air that was present in the catheter (Fig. 4a). To analyze if this air bubble influences tumor onset, an alternative instillation method was conducted. In doing so, the catheter itself was filled with tumor cell suspension prior to catheterization of the murine bladders. This prevented the air bubble formation (Fig. 4b). The comparison of both instillation techniques showed no differences in tumor formation in SCID-beige mice (Table 1, No 6). Exemplarily, luminescence intensities of individual animals after instillation of 0.5 × 10⁶ UM-UC-3^LUCK1 cells without the air bubble in the bladder are shown in Fig. 3d.

Selected animals were analyzed by MRI and PET (combined with CT). MRI measurements were carried out every 2 to 4 days to visualize size, location and growth of the tumor. Exemplarily, the MRI and BLI images of a UM-UC-3^LUCK1 tumor in a BALB/c nude mouse are shown (Fig. 5). Both imaging techniques displayed the rapid tumor growth within the 6 days shown. In these MRI images, the orthotopic tumor was easily distinguishable in the bladder. Overall, MRI of the urinary bladder of living mice is challenging because of the movement of the bladder and the intestine. Blurring occured preventing the quantitative evaluation in 8 of the 61 imaging series. In 5 cases no tumor could be detected in MRI despite positive BLI signals.

Western blot analyses proved presence of EGFR protein in UM-UC-3^LUCK1 cells (Fig. 6). The pilot PET experiment with the transurethrally administered ⁶⁸Ga-labelled EGFR antibody cetuximab was carried out on two SCID beige mice (Fig. 7). The retaining activity allowed imaging of the bound antibody both in the tumor and the bladder. The registration of the PET and CT images showed the localization of most remaining activity in the tumor region revealing targeting of EGFR-expressing UM-UC-3^LUCK1 cells.

After HE staining, sections of the UM-UC-3^LUCK1 tumors were examined for staging and grading. Only slices of tumors with association to the urothelium that allowed TNM classification were included in the evaluation. Of the 68 evaluable xenografts 53 (78%) and 11 (16%) displayed tumor stages T1 and Ta, respectively, whereas 4 tumors (6%) already invaded the musculature (pT2a) (Table 2). All muscle invasive tumors were observed in SCID-beige mice. All evaluable tumors were graded as high-grade. Representative histological images are shown in Fig. 8. In 39 cases (57%) a single tumor could be identified in the urinary bladder whereas in 29 cases two or more tumors grew (Table 2). Kidneys, livers and lungs of all 16 mice in experiment 6 were examined histopathologically to evaluate a possible metastasis formation. Two mice with pathological BCa stage Ta and T1, respectively, showed metastasis in the kidneys whereas one of these mice also showed lung metastasis (Fig. 9).