Development of a metastatic murine colon cancer model

Abstract

Background

It has now become clear that the complex interplay of cancer and the immune responses against it plays a critical role in the tumor microenvironment during cancer progression. As new targets for cancer treatment are being discovered and investigated, murine models used for preclinical studies need to include intact immune responses to provide a closer correlation with human cancer. We have recently developed a modified syngeneic orthotopic murine colon cancer model that mimics human colon cancer progression with consistent results.

Materials and methods

Tumors were created using the murine colon adenocarcinoma cell line, CT26, modified to overexpress the firefly luciferase gene (CT26-luc1), which allowed real time in vivo monitoring of tumor burden when the substrate, D-luciferin, was injected intraperitoneally using the In Vivo Imaging System (IVIS). Mice are Balb/C (Harlan), syngeneic with the CT26-luc1 cells. Cells are injected submucosally, suspended in matrigel, into the cecum wall under direct visualization.

Results

The model has demonstrated consistent implantation in the cecum. In vivo bioluminescence allowed real time monitoring of total tumor burden. Perioperative preparation had a significant impact on reproducibility of the model. Finally, total tumor burden quantified with bioluminescence enabled estimation of lymph node metastasis ex vivo.

Conclusions

This method maintains an intact immune response and closely approximates the clinical tumor microenvironment. It is expected to provide an invaluable murine metastatic colon cancer model particularly in preclinical studies for drug development targeting those mechanisms.

1. Introduction

Colorectal cancer (CRC) is the third most common cause of death by cancer in both sexes in the US, and over 50,000 are expected to succumb to CRC in 2013 [1]. With best care, the five-year survival rate of all cases of CRC is 65%. It is clear that new therapies are still needed to improve survival in advanced disease. Though some promising therapies have been developed in recent years, the vast majority of compounds validated by animal models fail to show benefit in clinical trials. For drugs tested from 1993–2002, only 26% of oncology drugs tested in phase I trials resulted in FDA approval for treatment [2]. In recent years, a number of voices have brought awareness to the growing concern that our preclinical animal models are inappropriate for assessing the utility of novel therapies in actual human patients [3–5]. Criticisms point out the poor predictive value of many pre-clinical trials, the lack of relevance of in vitro data, and the differences between mouse models and human immunology and cancer biology.

Despite these criticisms, murine models for human cancer are still the mainstay of pre-clinical evaluation in drug discovery and emerging technologies. Thus, models that more closely replicate the biology and progression of human cancer are in urgent need. Xenograft models that implant human colon cancer cells into immune-deficient nude mice have been utilized since the 1960s [6], and remain the most commonly used model for drug development. Xenograft models are useful for assessing drug effectiveness against human cancer cells in an animal setting; however, as targeted therapies emerge, the immune-deficient nude mouse model is no longer an ideal model for novel therapies that take advantage of the immunologic characteristics of cancer. Even the latest patient-derived xenograft models that maintain the heterogeneity of the tumor by transplanting a part of a tumor from a patient, argued by many to be the best mimic of a human tumor, is not free from this limitation since the tumor can only be implanted onto immune deficient nude mice [7].

Although syngeneic models that implant murine cells into immune-intact normal mice are limited to using relatively less studied murine colon adenocarcinoma cell lines, these models are capable of demonstrating the complex interaction between the immune system and the tumor microenvironment in cancer progression. Subcutaneous tumors were for many years the mainstay of murine tumor models, with direct measurement of tumor size utilized to study effects of drugs and targeted therapies. We have recently reported that even these genetically homogenous cell lines have markedly different gene expression depending on the location of the tumor in the mouse, with subcutaneous tumors having a different gene expression profile from orthotopically implanted tumors, and models of metastasis, demonstrating the importance of orthotopic models in studying the effects of the tumor microenvironment on cancer growth and progression [8–10].

Real time monitoring of cancer progression in vivo is now possible through genetic overexpression of reporter genes, such as green fluorescent protein or luciferase, within the implanted cell lines. Methods utilizing green fluorescent protein were the first to be developed [11]; however, the strength of the signal is known to dissipate when it travels through the body wall. Syngeneic cancer cell lines engineered to express firefly luciferase have been utilized successfully in our laboratory to track breast cancer progression and metastasis to the lymph nodes and lungs [10,12], and are optimally suited for following intraabdominal tumors and subsequent metastasis. Here we demonstrate our newly established syngeneic colon cancer cell implantation method, and the utility of our luciferase positive model for monitoring cancer progression.

2. Material and Methods

2.1. Cell culture

CT26-luc1 cells generated from CT26 (American Type Culture Collection, Rockville, MD, USA, a murine colon adenocarcinoma cell line derived from BALB/c colon, have been engineered to express firefly luciferase through a retrovirus vector-mediated process by our collaborators at the National Cancer Research Institute in Tokyo, Japan [13]. CT26-luc1 cells were cultured in RPMI Medium 1640 with 10% FBS. Prior to implantation, CT26-luc1 cells were cultured in a 37°C humidified incubator with 5% CO2 and grown to 80% confluence using RPMI with 10% FBS. Cells were re-suspended in PBS and mixed with Matrigel Basement Membrane Matrix (BD Bioscience) at a ratio of 1:9 for a final concentration of 50,000/10 μL prior to implantation.

2.2. Animals

All animal studies were conducted in the Animal Research Core Facility at VCU School of Medicine in accordance with institutional guidelines. Surgical procedures were approved by the VCU Institutional Animal Care and Use Committee (IACUC), accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). Female BALB/c mice (8–10 weeks, weight 20–25g; Harlan) were anesthetized with continuous vaporized isoflurane for general anesthesia, and mice were given analgesia (buprenorphine SR) for at least 72 hours post-operatively, and closely monitored throughout the perioperative period. Any animals appearing to be in significant distress or showing physical signs indicating unlikely survival for an additional 24 hours were euthanized as a humane endpoint.

2.3. In vivo bioluminescence

D-Luciferin (0.2 mL of 15 mg/mL stock; PerkinElmer) was injected intraperitoneally into mice previously implanted with CT26-luc1 cells at indicated times. Bioluminescence was detected and measured using In Vivo Imaging System (IVIS; Caliper-PerkinElmer). Living Image Software (Xenogen, PerkinElmer) was used to quantify the photons/sec emitted by the cells. Bioluminescence was measured and quantified at 5-minute intervals over 30 minutes using a subject height of 1.5 cm, medium binning and an exposure time of 0.5 seconds to 1 minute. The peak number of photons/sec calculated over this time frame then determined bioluminescence.

3. Results

3.1. Development of our metastatic syngeneic orthotopic colon cancer model

The surgical technique to generate a metastatic syngeneic orthotopic colon cancer model was established (Figure 1). CT26-luc1 cells were suspended in a Matrigel Basement Membrane Matrix. Matrigel suspension enables an even distribution of cells throughout the inoculum, and its gel-like consistency when warmed by body temperature prevents the cells from leaking out into the peritoneal cavity and causing carcinomatosis. It also provides an initial scaffold for tumor growth. The abdomen is prepped with povidone iodine after fur removal, and draped using aseptic technique. Laparotomy is obtained through midline incision. The cecum is eviscerated, then draped in separate sterile gauze, and moistened with PBS. After cecotomy, Matrigel suspended syngeneic CT26-luc1 colon adenocarcinoma cells are injected into the submucosal layer of the cecal wall using a 1 mL syringe with a 28 gauge needle (50,000 cells in 10 μL). We found that cecotomy allows identification and inoculation of the cells to the appropriate layer of the thin murine cecal wall more easily than surface injection. The cecotomy was closed with 5-0 silk sutures in interrupted Lembert fashion. The cecum is then again moistened with PBS and gently returned to the left upper quadrant, taking care to avoid malrotation of the viscera. The fascial edges are approximated with 5-0 silk sutures and the skin closed primarily with continuous sutures.

Figure 1. Our metastatic murine colon cancer model has a simple, reproducible surgical technique.

(A) Fur was clipped on syngeneic Balb/c mice, animal was anesthetized with isoflurane and sterilely prepped and draped. (B) A laparotomy incision was made, and the cecum was grasped and eviscerated. (C) The tip of the cecum was sharply removed, making a cecotomy. (D) CT26-luc1 (5 × 10⁵) cells in matrigel suspension were injected through the lumen submucosally. (E) The cecotomy was then closed, followed by (F) two layer closure of the abdominal wall. Animals were monitored closely in perioperative period, with Buprenorphine SR used for analgesia.

3.2. Bioluminescence allows monitoring of chronological tumor burden growth

We have generated a CT26 murine colon adenocarcinoma cell line with stable overexpression of firefly luciferase (CT26-luc1 cells). When D-Luciferin is intraperitoneally injected, it emits a photon signal that can be detected using IVIS and Living Image Software. Note the similar tumor burden between the two animals that were followed chronologically (Figure 2A). This technology allows quantification of the amount of live CT26-luc1 cells (total tumor burden), which allows development of tumor to be assessed ex vivo. Figure 2B demonstrates the growth curve of the CT26-luc1 tumor burden after implantation using our method in the same animals (n=10). In a series of experiments, the majority of animals, 23/26, developed cecal tumor using this technique (Figure 2C). Failure to develop tumor may be because the inoculation was not localized in the appropriate layer of the cecal wall.

Figure 2. Bioluminescence allows monitoring of chronological tumor burden growth.

(A) Balb/c mice underwent surgical implantation of CT26-luc1 cells as described in Figure 1. IVIS was utilized at 3–4 day intervals to follow cancer progression. Representative photographs of two mice are shown over time to demonstrate tumor growth. (B) Photon counts allow quantification of tumor burden and in our model demonstrate rapid and consistent rise over time (representative experiment shown, N=10, Mean+/−SEM). (C) In a series of experiments comprising 26 mice, the majority (23) animals proceed to develop tumor.

3.3 Consistent and reproducible results depend on meticulous technique and preoperative preparation

Even among the animals that underwent the same new method, injection of CT26-luc1 cells suspended in matrigel into the cecum, postoperative morbidity and mortality differed depending on the preoperative preparation used. We found that our early experiments had low survival with 63% at 20 days (n=37). Necropsy of these animals revealed severe bowel dilatation secondary to extensive adhesions and obstruction with ischemia. After these initial experiments, the perioperative preparation was re-examined and modified to include clipping of the fur on the abdominal wall and frequent moistening of the exposed bowel during the operation. After these modifications the survival at 20 days was improved to 89% (N=47) (Figure 3C). Log-rank analysis of the overall Kaplan-Meier survival modeling demonstrates a statistically significant difference (p=0.007).

Figure 3. Consistent and reproducible results depend on meticulous technique and preoperative preparation.

(A) Early attempts at orthotopic implantation with CT26 were plagued by high perioperative morbidity, affecting the reproducibility of the model. A representative necropsy photograph is shown. (B) Morbidity was attributed to bowel obstruction, as is evident by the severely distended bowel with dense adhesions and necrotic bowel associated with internal hernia, noted in this representative image taken at necropsy (C) Animal survival significantly improved after refinement of method. The thick line depicts survival before revision, and the thin line depicts after revision of the procedure, including thorough hair clipping and frequent moistening of the bowel with sterile PBS solution during operative implantation of the cells.

3.4 In vivo bioluminescence allows non-invasive estimation of metastasis

In order to investigate whether IVIS measurements can assess metastasis in the model, mice were implanted with 5×10⁵ CT26-luc1 cells using our newly established method. Out of the 13 animals that survived until approximately 1 month after the injection, 10 had enlarged tumors in the cecum (Figure 4A). Immediately prior to sacrifice, animals were injected intraperitoneally with D-luciferin to enable ex vivo assessment of bioluminescence. As shown in Figure 4B and 4C, some removed mesenteric lymph nodes demonstrated metastasis detected by bioluminescence. IVIS had been performed on the animals within 5 days prior to sacrifice to investigate whether the total IVIS measurement can predict metastasis. Significant difference was noted in the last whole body IVIS photon counts in those animals with grossly positive nodes and nodes positive on ex vivo IVIS (7/13) compared to animals without positive nodes (6/13) (Figure 4C). All animals with a count greater than 3,000,000 photon/second had positive mesenteric lymph nodes at time of sacrifice.

Figure 4. In vivo bioluminescence allows non-invasive estimation of metastasis.

(A) Representative necropsy picture demonstrates primary tumor in the cecum (circled yellow) without significant bowel dilatation to suggest bowel obstruction. No carcinomatosis was observed in this case. (B) Ex vivo IVIS, where Luciferin was administered in vivo and the organs were imaged at fixed timing extracorporealy, was performed to verify metastasis 23 days post implantation. Primary tumor in the cecum as well as metastases to the regional (LN) as well as to the mesenteric lymph nodes (MLN) are indicated with yellow arrows. (C, D) H&E staining of the mesenteric lymph nodes confirmed metastasis by CT26-luc1 cancer cells in both low (C) and high (D) magnifications. (E) In vivo bioluminescence of the live animal prior to euthanasia was found to correlate with the presence of mesenteric lymph node metastases. Photon counts in animals with lymph node metastasis were notably higher, reflecting increasing tumor burden, with a threshold of 3,000,000 photons/second for 100% metastasis.

4. Discussion

The past decade of cancer biology research has resulted in a growing understanding of the importance of the tumor microenvironment in cancer progression. A PubMed search for the keyword “tumor microenvironment” reveals a fifteen-fold increase in publication counts since 2004, with 94% of studies published within the past ten years. It is increasingly recognized that cancer cells do not manifest disease independently, but rather form collaborative interactions with conscripted and corrupted resident and recruited normal inflammatory cells that allow the cancer to progress through tumor growth, invasion, and metastasis. The contributions of T-cells in particular to cancer progression have been well established. Despite the known importance of recruited immune cells and the role these cells play in the tumor microenvironment, xenograft models in immunologically deficient nude mice, where T-cell functions are eliminated, remain commonly used animal models for cancer drug development.

Xenograft models that implant human colon cancer cells into immunedeficient nude mice take one of several typical approaches: injection of human colon cancer cells into the cecum or colon subserosally, or surgical excision of a tumor grown subcutaneously in another animal with replacement onto the cecum of the study animal. Tumors can also be grown subcutaneously, or hepatic metastatic models can be created directly through injection into the ileocecal vein or the portal vein [14,15]. In addition, tumors have been taken directly from human patients and assayed for metastatic potential utilizing these models [16].

A popular approach pioneered by Hoffman at University of California San Diego involves the patented MetaMouse models [11]. For colon cancer, this involves subcutaneous growth of the cancer cells overexpressing GFP or RFP and suture fixation of a small piece of the tumor to the colon wall in a separate animal that can be assessed utilizing whole body imaging for fluorescence. The published literature utilizing this model is predominantly in nude mice with human colorectal cancer established cell lines or histologically intact tumor [17,18].

Xenograft models such as those described are useful for assessing effectiveness against human tumors in an animal setting; however, as targeted therapies emerge, the immunologically deficient nude mouse model is no longer an ideal model for therapies that take advantage of the immunologic characteristics of cancer. Syngeneic models have an advantage over xenograft models in that they are immunologically intact and can demonstrate the interdependence of cancer and the immune system [19]. Techniques developed by Hoffman et al, while allowing the metastasis of a primary tumor, have a tumor that is located extraluminally, making their spread markedly different from that seen in the clinical setting where the tumor originates intraluminally then invades extraluminally while undergoing lymphatic and hematologic spread. Though orthotopically transplanted colon tumors do have hepatic metastasis, they do not regularly undergo lymphatic metastasis, therefore it is not an optimal setting to enable the study of drugs aimed to affect or prevent early metastasis. Furthermore, due to the extraluminal location of the tumor, these models are reported to have a high rate of carcinomatosis [20]. Though our model is not without carcinomatosis, it occurs uncommonly around 10% of the time. It is clear that each existent model is likely optimal for a particular set of clinically relevant research questions and that comparison of effects in multiple in vitro and in vivo models is likely ideal for predicting the utility of an emerging compound. There is no perfect model that can ultimately equal the value of actual clinical results; however, it is best to realize the relative weaknesses of the various models when choosing one to utilize.

Luciferase transgenes have emerged over the past decade as a useful method allowing in vivo imaging of tumors and metastasis through bioluminescence. This allows for non-invasive serial monitoring of tumor growth and responses to treatments or experimental variables in real time in living animals [8–10,12]. Though use of syngeneic or xenograft transgenic tumor cells for subcutaneous or orthotopic implantation is well established, recently, there have been concerns of potential activation of cellular adaptive immunity and sensitization against the viral agents used to transduce the luciferase transgene or to the luciferase enzyme itself. Though some authors have shown that high levels of expression of firefly luciferase result in an immune response [21,22], there is evidence from multiple studies that if luciferase expression starts low and increases gradually over time, as in scenarios resulting from implantation of luciferase expressing cancer cells, the immune response is minimal [22–24]. It is thought that there is immune tolerance to firefly luciferase at smaller levels of expression, and that it is only at sustained expression levels equivalent to 10⁹ photon/sec/cm² that an immune response is elicited, with other studies directly demonstrating that there is no effective alteration in the immune response from luciferase used in the context of tumor imaging [25,26], making it an ideal reporter to use in our model. In our model, luciferase expression as quantified using IVIS directly correlated with tumor burden and was able to be used to set a predictive threshold for lymph node metastasis prior to euthanasia. In addition to the possible effect of luciferase transgene in inciting an immune response, there is a known phenomenon of “quenching” of the luciferase signal, which is an artifact due to less delivery of Luciferin to the tumor as the tumor grows and vascularity to the interior part of the tumor becomes compromised. This phenomenon is frequently observed in subcutaneously or intra-mammary implanted models where the size of the tumor and bioluminescence signal become non-linear, however, this was not observed in this model most likely because the tumor metastasized to the lymph nodes before quenching of the signal became an issue. Additionally, the extensive tumor size that cause this phenomenon is likely anatomically prohibited by the consequence of bowel obstruction.

In our laboratory we are committed to using and developing advanced cancer model systems with a high degree of verisimilitude to the pathology of disease progression seen in human cancer patients. To that end we have worked on refining our methodology in murine colon cancer modeling to achieve a realistic approach to metastatic colorectal cancer modeling. During development of this model, we employed several different surgical techniques in an attempt to make a reproducible metastasis model. Initial attempts simply used injection of cells through the serosa of the colon; however, in our experience, this led to early carcinomatosis prior to lymphatic metastasis as the inoculum leaked through the needle tract and seeded the peritoneum. We employed transplantation of subcutaneously grown tumors to the surface of the cecum, but again experienced a higher likelihood of carcinomatosis without lymphatic spread. A third method utilized a novel technique in which the cecum was invaginated, creating a pouch into which matrigel cell suspensions were injected. This method yielded large tumors, but many of the animals became moribund prior to development of metastasis due to colonic obstruction.

Our current approach entails the use of a mildly immunogenic syngeneic murine colon cancer cell line CT26, originally developed in 1975, with a high metastatic potential [27], which has been altered to express luciferase. Tumors are created through orthotopic injection of the cell line prepared in a matrigel suspension into the wall of the cecum, allowing invasion through the colon wall and lymphatic spread. Use of matrigel is imperative in this method. Attempts without matrigel, with cells in a PBS suspension, in our experience led to excessive leakage with low take and/or carcinomatosis. Implantation in the long, easily mobilized murine cecum is also critical in our approach. The cecum of a healthy adult mouse is approximately 2.5–3.0 cm in length. Implantation in the cecum allows the tumor to grow large enough for metastasis to occur prior to colonic obstruction from the tumor. In studies utilizing implantation in other colorectal locations, with early obstruction has been a barrier to studying metastatsis. A recent rectal cancer model using CT26 was able to demonstrate an excellent take rate of tumor and frequent spontaneous metastasis; however, because the lower rectum was utilized, the pattern of spread is through the systemic venous circulation, limiting the model to exemplifying low lying rectal cancer without broader applicability to colon cancer [28].

Recent advances in cancer biology have resulted in a new and growing emphasis on the role of the immune system in forming a tumor microenvironment that nurtures and promotes cancer progression. Potential applications for the model include experiments with targeted therapies and conventional drugs that affect the tumor microenvironment in a neoadjuvant setting, as well as studies into the role of the tumor microenvironment and the immune system in facilitating or potentially thwarting colon cancer progression. The tumor microenvironment is especially important in the context of metastatic disease, where significant need still exists for new treatments. As drugs that attempt to modulate these factors of cancer progression are developed and evaluated preclinically, it is clear that advanced model systems such as that presented herein will have a large role to play.

5. Conclusions

We report a new murine model for metastatic colon cancer. The important combination of luciferase imaging allowing prediction of metastasis prior to euthanasia, matrigel implantation to reduce peritoneal dissemination, and cecum orthotopic implantation to decrease likelihood of bowel obstruction prior to metastasis is key to our approach, and through relatively simple refinements in technique, the mortality associated with the implantation procedure is minimized. This model replicates the invasion and metastasis of colon cancer closely to that of the human patient and has a role to play in the development of new cancer therapeutics and new understandings of colon cancer metastasis.