A Spontaneous Acinar Cell Carcinoma Model for Monitoring Progression of Pancreatic Lesions and Response to Treatment Through Noninvasive Bioluminescence Imaging

Pancreatic cancer is one of the leading causes of cancer deaths in the United States, with a 5-year survival from the time of diagnosis of <5% (1). The aggressiveness, the lack of effective treatments, and late detection of only advanced symptomatic pancreatic cancers due to their anatomic location behind other organs contribute to such poor outcomes (2). Lack of effective treatment for pancreatic tumors prompted development of preclinical models for evaluation of novel therapeutics. Spontaneous mouse tumor models have been used for two decades in preclinical drug development to better understand many critical components of tumor development, such as tumor-stromal interactions, the role of the immune system, metastasis, angiogenesis, and tumor cell proliferation in immunocompetent animals (3–5). Most pancreatic epithelial neoplasms recapitulate to some degree one or more of the normal cell types of the pancreas: ductal, acinar, and endocrine. Most pancreatic neoplasms (>90%) have ductal differentiation, including the most common tumor, infiltrating ductal adenocarcinoma. The small group of “nonductal” pancreatic neoplasms includes endocrine and acinar neoplasms, as well as those with mixed or undetermined differentiation (6). Spontaneous pancreatic tumor models wherein tumor development in the acinar cells of the pancreas is mediated by expression of SV40 T antigens (7, 8) or mutant Kras (9) has been previously reported. These studies increased our understanding of the molecular, physiologic, and histologic aspects of tumorigenesis. However, the use of such models in the evaluation of novel therapeutics is complicated by the variable onset of pancreatic tumor development, as well as a lack of methods to assess tumor burden without sacrificing the animal. As a result, these models usually required relatively large cohorts of animals and extensive invasive endpoint analyses to produce statistically significant drug response data.

Recent advances within the field of bioluminescence imaging have added a new tool for monitoring tumor development noninvasively in preclinical cancer models (10, 11). Bioluminescence imaging allows sensitive and rapid analysis of light emitted from luciferase-labeled cells and can be used to measure tumor burden from otherwise nonvisible deep tissue locations when caliper-based measurements are not possible. Firefly luciferase has been widely used as a bioluminescence reporter for in vitro and in vivo applications. Its activity is ATP and O₂ dependent; thus, only viable cells bioluminesce. This approach allows the researcher to noninvasively measure relative numbers of viable cells at regular intervals following the onset of treatment. Tight correlations have been shown between photon emission and tumor burden, so the growth of primary tumors and spontaneous metastases can be quantified noninvasively, even at deep tissue sites (11, 12). Imaging studies that examine tumor response to therapeutic interventions are also able to achieve statistical significance using far smaller cohorts of experimental animals as compared with “classic” studies wherein animals had to be sacrificed at each measurement point. This can be accomplished because the tumor burden can be determined longitudinally within the same animal, and all the animals can be stratified for equal tumor burden between the groups at the beginning of the study.

Rapamycin was originally used as an antifungal agent and a suppressor of the immune system. Recently, two rapamycin analogs were shown to have clinical antitumor activity. Torisel, which has been approved, and RAD001 (Everolimus), which is currently completing phase III trials. Rapamycin and its analogs inhibit mammalian target of rapamycin (a serine/threonine kinase)–mediated signaling, which regulates cell growth, cell cycle progression, and tumorigenesis (13, 14). The inhibitory effect of rapamycin and its derivatives on in vivo tumor progression has been previously shown in several tumor models, including a lung adenocarcinoma with activated PI3K/Akt signaling (15), an Akt-induced mouse prostatic intraepithelial neoplasia (16), a transgenic ErbB2 mouse mammary tumor model (17), and a TgMISIIR-TAg transgenic mouse model with progression of ovarian carcinoma (18). Therapeutic efficacies of rapamycin shown in these tumor models, including T antigen–mediated ovarian carcinoma, prompted us to evaluate this drug in our EL1-luc/TAg model to illustrate its utility for testing the efficacy of mammalian target of rapamycin/PI3K/Akt pathway inhibitors as potential therapeutics.

Here, we also show for the first time that spontaneous acinar cell carcinoma development can be visualized using bioluminescence imaging. Further application of the EL1-luc/TAg acinar cell carcinoma model should enable efficient testing of the efficacy of therapeutic candidates against acinar cell carcinoma, one of the most untreatable human cancers today.

We used a 216-bp rat elastase 1 (EL1) promoter (19) to construct EL1-luc and EL1-TAg transgenes to confer acinar cell–specific expression of luciferase (Promega Corporation) and SV40 T and t antigens in the EL1-luc/TAg transgenic mouse. The EL1 promoter was PCR amplified from Sprague-Dawley rat genomic DNA and cloned into the pCR4Blunt-TOPO vector (Invitrogen). The EL1 promoter fragment was retrieved from the pCR4Blunt-TOPO vector as an EcoRI fragment and cloned into the SmaI site of the pGL3-Basic vector (Promega Corporation). The EL1 promoter fragment was also cloned as an EcoRI fragment upstream of a 2.7-kb SV40 T and t antigen sequence inside a pBlueScript vector (Stratagene Cloning Systems). The EL1-luc and EL1-TAg transgenes were used to generate transgenic mice by mixing in a 1:1 molar ratio before pronuclear microinjection of oocytes of mouse FVB/N strain (20). Both transgenes cosegregated in subsequent generations of mice. Genotyping of the transgenic mice was done primarily by imaging. The PCR amplification of a 1-kb luciferase fragment with primer pairs (forward, 5′-tggattctaaaacggattaccaggg-3′; reverse, 5′-ccaaaacaacaacggcggc-3′) was also done using mouse tail DNA.

Mouse tissues were removed and homogenized in three volumes of PBS containing a protease inhibitor cocktail (Roche Applied Science) and lysed with passive lysis buffer (Promega). After centrifugation at 14,000 rpm at 4°C for 10 min, the supernatant was collected. Luciferase activity was measured by luminometer (TD-20/20, Turner Designs) using a Luciferase Assay System (Promega). Protein concentration was estimated using the Bradford assay (Sigma-Aldrich).

In vivo imaging of luciferase activity in the pancreas was done using an IVIS imaging system (Caliper Life Sciences, formerly Xenogen Corp.). Mice were anesthetized with isoflurane and injected i.p. with 150 mg/kg of luciferin. The animals were imaged 10 min after luciferin injection. To reduce variability in measured bioluminescence resulting from variable internal placement of the pancreas, mice were imaged in three positions (ventral, left flank, and right flank). Photons emitted from the pancreas region in each position were quantified using LivingImage software, and the sum of these measurements was used as the total bioluminescence signal from the pancreas.

Rapamycin (LC Laboratories) was dissolved in ethanol at a concentration of 30 mg/mL and stored at −20°C. The rapamycin dosing solution was freshly prepared by diluting the rapamycin stock 1:20 with 5% polyethylene glycol–400, 5% Tween 80 solution to make a 100-μL dose (containing 0.15 mg rapamycin) for a 30-g mouse. Rapamycin was administered by i.p. injection three times weekly for the duration of the treatment. Control animals received i.p. injection of the vehicle (100 μL of 5% polyethylene glycol–400, 5% ethanol, and 5% Tween 80).

For histologic analysis, isolated tissues were immersion fixed in 4% formaldehyde (pH 7.4) overnight. The samples were then embedded in paraffin, and sections were cut for H&E staining.

To create a mouse model with spontaneous pancreatic adenocarcinomas and specific labeling of the pancreas tissue with the luciferase reporter, we made two transgene constructs (Materials and Methods) whereby the transcription of SV40 T and t antigens, and firefly luciferase were regulated by the rat EL1 promoter. After coinjection of the transgenes into the FVB/N oocytes, we obtained several transgenic founder lines. One of the founder lines was selected for this study based on specific luciferase expression in the pancreas and tumor latency roughly comparable with previously published work (21).

The EL1-luc/TAg mice were imaged with an IVIS imaging system following i.p. injection of luciferin. We initially imaged our transgenic mice in the ventral position only. However, we observed a high degree of variability in our longitudinal bioluminescent measurements, even in normal non–tumor bearing (young) animals, because of variable internal placement of the pancreas and resulting variation in the tissue depth of the light emitting cells. We found significantly less variability in our measurements when we took the sum of measured bioluminescence from right flank, left flank, and ventral positions.

Imaging analysis showed that bioluminescence was detectable from the region of the pancreas of EL1-luc/TAg mice in all positions, as expected. Male and female mice showed similar intensity in the basal bioluminescence signal (Fig. 1A). The specificity of the signal from the pancreas was further confirmed by comparing the luciferase signal from the homogenized pancreatic tissue to that from other tissues (Fig. 1B). In agreement with the in vivo bioluminescence imaging data, pancreatic tissue showed the highest level of luciferase activity ex vivo. Low levels of luciferase expression were detectable ex vivo in tissue homogenates of kidneys and brains of the EL1-luc/TAg mice.

Imaging of the EL1-luc/TAg mice longitudinally offers a noninvasive measure of acinar cell carcinoma progression (Fig. 2). In younger mice of 8 to 10 weeks of age, the total bioluminescence signal from the pancreas ranges from 2 × 10⁶ to 3 × 10⁶ photons/s. Increase of the bioluminescence signal from the baseline level to the range of 4 × 10⁶ to 6 × 10⁶ photons/s indicates early tumor development. The latency of tumor development varied from 10 to >20 weeks of age in these mice.

Representative mice with acinar cell carcinomas at various developmental stages, as indicated by imaging, were sacrificed for ex vivo imaging and for necropsy analysis. As shown in Fig. 3, a 10-week-old female mouse (Fig. 3A) has a bioluminescence signal emitted from the pancreatic region. Ex vivo imaging at necropsy highlighted that the predominant source of the bioluminescence was indeed the pancreas. No pancreatic tumor was observed in this mouse. In another female mouse of 12 weeks of age, an early stage of acinar cell carcinoma development was detected by imaging (Fig. 3B, black arrows). Necropsy analysis confirmed that small lesions had started to develop in the pancreas (Fig. 3B, ex vivo imaging). These areas were highly vascularized with stronger ex vivo bioluminescence signals. In a 19-week-old female mouse, a significant increase of the bioluminescence signal in the pancreatic and abdominal regions was observed (Fig. 3C, left). Necropsy analysis revealed the presence of a large acinar cell carcinoma and metastatic lesions to the liver and the mesenteric fat tissue (Fig. 3C, right). At late stages of tumor development in EL1-luc/TAg mice, multiple metastatic lesions to the fat tissues of the gastrointestinal tract and the reproductive organs were often detectable (Fig. 3D, white arrows). Metastatic tumors in the fat tissue remained bioluminescent (Fig. 3D, middle and bottom right). However, ∼25% of female transgenic mice had also developed nonglowing ovarian tumors by 21 to 23 weeks of age. In the male mice, the progression of the primary acinar cell carcinoma, as well as the development of metastatic lesions, is similar to what have been observed in the female mice.

We did histologic analysis of pancreas and tumor samples harvested from selected mice. The pancreas from a wild-type FVB/N mouse displaying normal histologic features is shown on Fig. 4A. Figure 4B to F shows pancreata from transgenic mice at various stages of tumor development. A small neoplastic nodule (left) consistent with early stage acinar cell carcinoma is shown on Fig. 4B. The rest of the pancreatic tissue showed dysplastic acini. In a large lesion displayed in Fig. 4C, the presence of pancreatic carcinoma of acinar cell origin is apparent. These carcinoma cells are highly mitotic (black arrows). In addition, necrosis occurred in certain areas of the tumor (red arrows). Figure 4D shows a poorly differentiated pancreatic carcinoma of acinar cell origin. The tumor is highly vascularized (arrows in red color, blood vessels), with increased mitotic activity (black arrows). In some mice, liver metastases were detected. Analysis of the liver metastatic lesions showed that these small carcinomas (arrows) are of acinar cell origin (Fig. 4E). Many of the EL1-luc/TAg transgenic mice showed multiple metastases in the peritoneal cavity, often on the surface of the mesentery membrane. The acinar cell origin of these metastatic lesions has been confirmed by histologic examination (Fig. 4F).

We tested the effect of rapamycin on acinar cell carcinoma development in the EL1-luc/TAg model. Selection of the treatment regimen was based on a previous publication (18). For this study, 24 male EL1-luc/TAg mice were used. All mice were imaged several times at the age of 9 to 10 weeks. Based on the quantification of the bioluminescence signals from the pancreas, the mice were stratified into two groups (rapamycin treated and vehicle treated, with the goal to minimize the differences in the mean light emission between the groups and the differences in light emission between the individual animals within the groups. Treatment started when the mice were 11 weeks of age. Mice in the treatment group were dosed with rapamycin thrice a week at 5 mg/kg via i.p. injections, and mice in the control group were injected with the vehicle alone for 7 weeks. Imaging was done twice a week.

Of the 24 mice used in this study, four mice (three from the rapamycin-treated group and one from the vehicle-treated group) were lost shortly after the study was initiated. The mortality of the rapamycin-treated group could be due to the adverse effects of the drug because ulceration was also observed in these mice. Initial images of a representative mouse from each group are shown in Fig. 5A. An increase of the bioluminescence signal in this vehicle-treated mouse indicated tumor development from 15 weeks of age, which was confirmed by necropsy analysis at the endpoint. The rapamycin-treated mouse showed steady basal level of luciferase signal in the pancreas region, which corresponded with the lack of tumor at necropsy. Quantification of the bioluminescence signal for each mouse is shown on Fig. 5B and C. Although most of the vehicle-treated mice showed elevation of the bioluminescence signal in time, the rapamycin-treated mice mostly showed a steady basal luciferase signal.

We did necropsy analysis at the end of the experiment (Table 1). In the male vehicle group, 8 of the 11 mice (M05, M06, M27, M31, M42, M49, M95, and M21) showed an increase of bioluminescence signal above a 30% to 40% threshold during the course of the study. Upon necropsy analysis, all these mice had developed acinar cell carcinoma in the pancreas. In addition, one mouse that showed no significant increase of the bioluminescence signal, M98, presented early stage acinar cell carcinoma development. In contrast, only two of the nine rapamycin-treated mice (M40 and M52) showed an increase of bioluminescence signal above the 30% to 40% threshold level during the imaging study. Upon necropsy analysis, both mice showed tumor development in the pancreas. Thus, the increase of bioluminescence signal in vehicle- and rapamycin-treated mice correlated well with the presence of tumor development, as confirmed by necropsy. The pancreatic tumor mass in the vehicle-treated group ranged from <0.5 g to >1 g. Metastasis to the mesenteric fat, as indicated by ex vivo imaging, occurred in some of the mice (M06, M49, and M52; Table 1). Three mice from the vehicle-treated group (M42, M49, and M98) and one mouse from the rapamycin-treated group (M52) also had small metastatic tumors in the livers. Some of these metastatic tumors were no-longer glowing, possibly because of the transdifferentiation of the acinar cell carcinoma in the novel host environment.

Statistical analyses of the bioluminescence signal at the endpoint and change of the bioluminescence signal from day 0 to the endpoint (Table 1) using the Wilcoxon rank sum test were done. Both analyses showed a significant difference between the rapamycin-treated mice and the vehicle-treated controls (P < 0.05). In addition, analysis of the total tissue mass [pancreas plus tumor(s)] also showed a significant difference between the rapamycin-treated mice and vehicle-treated mice (P < 0.001).

SV40 T antigen induces tumors in a wide variety of cell types in transgenic mice. Tissue-specific promoters controlling T-antigen expression can be used to limit tumor development to precisely defined tissues (3). Spontaneous pancreatic tumor models mediated by the expression of SV40 T antigen or Kras have previously been developed to recapitulate many key biological features of human pancreatic adenocarcinoma (7–9, 21–23). Despite the fact that SV40 T antigen is not a driver for human tumor development per se, the pathways altered, including p53 and Rb (retinoblastoma protein) tumor suppressor pathways, are relevant to human tumorigenesis (8, 19). Alterations in these tumor suppressors cause pleiotropic effects that are reflected in the deregulation of signaling pathways controlling cell proliferation, survival, adhesion, and migration (24, 25). In the past, the nonvisible nature of tumor development in the pancreas significantly limited the utility of these models for preclinical drug development purposes. Because bioluminescence imaging allows noninvasive assessment of luciferase labeled tumors, we established an EL1-luc/TAg mouse model to circumvent this limitation. We show here that bioluminescence imaging confers a noninvasive means to measure changes in pancreatic tumor burden in such mice. In addition, the animals can be stratified for equal tumor burden between the groups at the beginning of the study.

We used a rat EL1 promoter to drive the expression of SV40 T antigen and luciferase in the acinar cells of EL1-luc/TAg transgenic mice. In this double-transgenic mouse model, the DNA of the two transgenes was mixed before pronuclear microinjection, which usually results in the cointegration of both constructs at a single chromosomal site (26). We confirmed this by showing that the transgenes were coinherited and did not segregate in the F1 or subsequent generations following breeding of the founder mice. We generated a number of transgenic founder lines in this manner and did imaging analysis of all of them. We then selected one founder line that showed specific luciferase expression in the pancreas, as well as a tumor latency comparable with previously published work. In agreement with these previously published results (7, 8, 21), SV40 T antigen expression in the acinar cells triggered acinar cell dysplasia and acinar cell carcinomas in these mice. Expression of luciferase in the acinar cells resulted in specific emission of bioluminescence signals from the pancreas, as indicated by in vivo imaging and ex vivo analysis of various tissues from the EL1-luc/TAg mice. At the time this mouse was made, EL1 was the only well-characterized promoter that mediated exocrine cell–specific expression in the pancreas. The application of other promoters such as p48 and Pdx for mediating ductal cell–specific gene expression have since been published (27, 28). Moreover, conditional mouse models have since been developed that involve the critical players driving the development of human pancreatic cancer such as Kras, p53, and p16, and so comprise mouse models with potentially more tractable genetics than our T antigen–driven model (22, 23). In our model, we have shown for the first time the utility of bioluminescence imaging for imaging spontaneous tumors originating from the exocrine pancreas, and it should serve as a proof-of-concept for the development of future and further improved genetically engineered bioluminescent pancreatic cancer models.

Histologic analysis of the pancreas from our transgenic mice confirmed acinar cell dysplasia and neoplasia, as previously described (8, 21). The presence of acinar cell carcinoma was evident in the tissues analyzed, with a marked increase of mitotic activity in the tumor regions. Metastatic lesions observed in livers and on the surface of mesentery membranes were of acinar cell origin as well, ruling out the possibility that ectopic expression of T/t antigens in these tissues was the cause of tumor development.

Rapamycin and its derivatives have been shown to suppress tumor growth in a number of spontaneous animal tumor models, including an SV40 T antigen–mediated ovarian tumor model (18). Our data showed that rapamycin treatment significantly inhibited incidence of acinar cell carcinoma development in the EL1-luc/TAg transgenic mice and that tumor development could be monitored noninvasively using bioluminescence imaging. Some of the rapamycin-treated mice escaped the effect of this mammalian target of rapamycin inhibitor. It is possible that T antigen–mediated oncogenesis activated many different pathways that contribute to cell proliferation and that certain pathways in the rapamycin-resistant tumor could compensate for the inhibition of the mammalian target of rapamycin pathway resulting from rapamycin treatment. Obviously, identifying these pathways will help design therapeutic strategies that circumvent resistance to mammalian target of rapamycin inhibitors.

When doing bioluminescence imaging, there are several factors that can contribute to variation of the luciferase signal. There factors include the quality of i.p. delivery of luciferin, the difference of Luciferin distribution kinetics, and timing of imaging acquisition. As a result, a difference of 30% to 40% of the luciferase activity could be caused by these variables. When doing data analysis, it is necessary to review all the data longitudinally. Luciferase signal that is consistently above a 30% to 40% threshold should be regarded as an increase of luciferase expression per se. The converse is also true when luciferase signal declines during the study.

Development of acinar cell carcinoma was accompanied by metastases in mesenteric fat associated with the digestive tract and in liver tissues in some mice. Although metastases to the mesenteric fat are detectable with in vivo imaging, we were unable to definitely locate liver metastases because of the fact that not all of the liver metastases were glowing, as revealed by ex vivo imaging. This could be due to the dramatic change of tissue environment surrounding cancer cells, which may have caused transdifferentiation of the acinar cell carcinoma in the liver and inactivation of the EL1 promoter activity, which controls luciferase expression.

We have also examined the female EL1-luc/TAg mice for pancreatic tumor development. Similar to the male mice, development of acinar cell carcinoma and metastases in livers and fat was observed. In addition, we observed nonglowing ovarian tumors in a number of rapamycin-treated female mice that did not have acinar cell carcinoma in the pancreas (data not shown). Histologic analysis also showed that the ovarian tumor originated from the ovary, not from the pancreas. Thus, the development of ovarian tumors slightly complicates the use of female mice for future applied in vivo studies, and so male EL1-luc/TAg mice will be preferred for future drug screening applications.

In summary, we have established a spontaneous tumor model for tracking acinar cell carcinoma development noninvasively and longitudinally with bioluminescence imaging. Treatment of these mice with rapamycin significantly inhibited tumor development, as indicated by bioluminescence imaging and necropsy analysis. This study provides evidence of a positive drug response of acinar cell carcinomas in the male EL1-luc/TAg mice. Further utility of this model will be in preclinical drug development of other agents against acinar cell carcinomas. The clinical complexity of human pancreatic tumorigenesis and the poor prognosis after detection currently makes pancreatic cancer an incurable disease. The establishment of additional animal models that better recapitulate clinical pancreatic tumor development is necessary for the effective preclinical evaluation of novel therapeutics against pancreatic cancer.

No potential conflicts of interest were disclosed.

We thank Drs. Ken Campbell, Richard Coffee, and David Grass in generating transgenic mouse founder lines; and Michael Vigil, Rhonda Gregory, and Margaret Cobos for animal husbandry.